Description
Financial repression is any of the measures that governments employ to channel funds to themselves, that, in a deregulated market, would go elsewhere. Financial repression can be particularly effective at liquidating debt.
ABSTRACT
Title of dissertation: A STUDY OF DEBT AND INFLATION
DURING A PERIOD OF FINANCIAL REPRESSION
María Belén Sbrancia, Doctor of Philosophy, 2012
Dissertation directed by: Carmen M. Reinhart
Peterson Institute for International Economics
Carlos A. Végh
Department of Economics
The large accumulation of public debt which took place as a consequence
of the recent ?nancial crisis, poses the question of how governments are going to
reduce their debts. In this dissertation I look at prior episodes where most advanced
economies were highly indebted to understand the possible courses of action that
may be available to governments in the future.
Chapter 1, written with Carmen M. Reinhart, documents the role played by ?-
nancial repression and in?ation in reducing government debt after the end of World
War II. This mechanism represents a more subtle and gradual form of debt re-
structuring. Evidence on the evolution and distribution of several measures of real
interest rates show that during the period 1945-1980 interest rates were markedly
more negative than what they were both in the period before World War II and
after 1980. When looking at the real interest rate on the portfolio of domestic debt,
real interest rates were negative half of the time for the 12 countries in the sample.
A negative real interest rate on the portfolio of domestic debt represents a saving
on interest payments for the government. In this case, savings on interest payments
averaged 2 to 3 percent of GDP. The chapter also documents the series of defaults,
restructurings and conversions that took place after the end World War I and the
Great Depression.
Chapter 2 builds on the results of the ?rst chapter. It provides a conceptual
framework to understand the di?erent ways through which in?ation can a?ect the
value of government debt, and also how to think of ?nancial repression as a restruc-
turing mechanism. Several econometric exercises are performed to disentangle the
relative contribution of ?nancial repression and unanticipated in?ation. The results
point to ?nancial repression combined with in?ation (i.e., a nonmarket interest rate
combined with in?ation) as a relatively more important factor explaining the debt
reduction in the post-World War II period. Finally, the question of how investors
were a?ected during this period is explored. It is shown that the equity premium
almost doubled during this period. This result suggests that the presence of ?nan-
cial repression, during the period 1945-1980, could help explain to some extent the
equity premium puzzle.
A STUDY OF DEBT AND INFLATION DURING A PERIOD OF
FINANCIAL REPRESSION
by
María Belén Sbrancia
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial ful?llment
of the requirements for the degree of
Doctor of Philosophy
2012
Advisory Committee:
Dr. Carmen M. Reinhart, Co-Chair
Professor Carlos Végh, Co-Chair
Professor Anton Korinek
Professor Pablo D’Erasmo
Professor Phillip L. Swagel
c Copyright by
María Belén Sbrancia
2012
Dedication
A mis abuelos Olga, Elidia, Atilio y Santín porque gracias a sus esfuerzos y
dedicación fue posible que hoy yo estuviera aquí. A mis papás, Gladys y Guido, por
todo el amor y apoyo que me han brindado en estos 27 años. A Tim, el amor de mi
vida, por ser mi compañero.
ii
Acknowledgments
I would like to thank my main advisors, Carmen Reinhart and Carlos Végh,
for their guidance, support and criticism along the way which have helped me com-
plete this dissertation. They are both outstanding academics and excellent people
who made a big di?erence, both professionally and personally, in my experience at
Maryland. I owe to Carmen my passion for empirical work, she has been a devoted
mentor who has inspired me in so many di?erent ways. She supported early on
when I started to work on my dissertation and encouraged me to pursue this topic.
I feel extremely lucky to have had the opportunity to learn so much from her. The
?rst chapter of this dissertation is written jointly with her. I don’t think there many
advisors like Carlos who is always available to talk, and who excels at the art of
being rigorous and supportive at the same time. He has taught me the importance
of keeping things simple. As he would often say "you don’t need an AK47 to kill a
?y."
I am also grateful to Anton Korinek for his comments, help and sometimes
hard questions which have helped me to judge my work from a di?erent angle and
become a better researcher. I would also like to thank all the other professors at
Maryland who have provided comments and suggestions along the way.
I owe immense gratitude to Adrian Guissarri who introduced me to the fasci-
nating ?eld of Economics, and encouraged me to pursue a doctoral degree.
I would like to extend my thanks to the administrative sta? at the Economics
Department, specially Vickie Fletcher, for their dedication, time and help during this
iii
time. I am also grateful to the sta? at the Library of the Congress who provided
me with invaluable help while I put together the database for this dissertation.
Thanks to my parents. I wouldn’t be here if it wasn’t for them. Thanks for all
your support, love, and for helping me pursue my dreams. I will always be grateful to
my dad for teaching me to persevere, work hard, and be an honest person. And very
special thanks to my mom, one of the strongest persons that I know, for listening to
me, for teaching me to be optimistic, and for always believing in me. To Nazareno
for constantly reminding me that "it’s not such a big deal" and to relax. To María
José for distracting me with fashion and music talks, and cheering me during these
years.
My deepest thanks go to my husband and best friend, Tim, for his uncon-
ditional love and support. This journey would have been a lot harder without his
jokes and company.
And last but not least thanks to all my good friends (Armando, Carla, Car-
olina, Cesar, Fernando, Florencia, Francisca, Giorgo, Jose?na, Laura, Lisa, Mariana,
Nieves, Pablo, Paula, Rong, Tiziana, Ximena) with whom I have shared so many
special and fun moments throughout the years.
iv
Table of Contents
List of Tables vii
List of Figures viii
1 Introduction 1
2 The Liquidation of Government Debt (with Carmen M. Reinhart) 5
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Default, Restructuring and Conversions: Highlights from 1920s-1950s 12
2.2.1 Global debt surges and their resolution . . . . . . . . . . . . . 13
2.2.2 Default, restructurings and forcible conversions in the 1930s . 14
2.3 Financial Repression: policies and evidence from real interest rates . . 19
2.3.1 Selected ?nancial regulation measures during the "era of ?-
nancial repression" . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.2 Real Interest Rates . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 The Liquidation of Government Debt: Conceptual and Data Issues . 23
2.4.1 Benchmark basic estimates of the "liquidation e?ect" . . . . . 25
2.4.2 An alternative measure of the liquidation e?ect based on total
returns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.4.3 The role of in?ation and currency depreciation . . . . . . . . 28
2.5 The Liquidation of Government Debt: Empirical Estimates . . . . . . 29
2.5.1 Incidence and magnitude of the "liquidation tax" . . . . . . . 30
2.5.2 Estimates of the Liquidation E?ect . . . . . . . . . . . . . . . 32
2.6 In?ation and Debt Reduction . . . . . . . . . . . . . . . . . . . . . . 34
2.7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3 Debt and In?ation during a Period of Financial Repression 61
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.2 Conceptual Framework . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.3 Empirical Measures and Data . . . . . . . . . . . . . . . . . . . . . . 77
3.3.1 Empirical Measures . . . . . . . . . . . . . . . . . . . . . . . . 77
3.3.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.4 The Liquidation E?ect from the Perspective of the Government . . . 81
3.4.1 In?ation and Real Interest Rates on Debt Portfolios . . . . . . 82
3.4.2 Incidence and Magnitude of the Liquidation E?ect . . . . . . . 84
3.4.3 Liquidation Revenues for the Government . . . . . . . . . . . 87
3.4.4 The Role of In?ation Expectations and Financial Repression . 89
3.4.5 Are the Results Biased by Bonds Issued Before 1945? . . . . . 95
3.4.6 Comparison to In?ation Tax . . . . . . . . . . . . . . . . . . . 98
3.4.7 E?ect on the Stock of Debt . . . . . . . . . . . . . . . . . . . 101
3.4.8 Understanding What A?ects the Probability of a Liquidation
Year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
3.5 How did Investors Fare During this Period? . . . . . . . . . . . . . . 108
v
3.5.1 Stocks, Bills and Bonds during the Sample Period . . . . . . . 109
3.5.2 A Longer Historical Perspective . . . . . . . . . . . . . . . . . 111
3.5.3 The Equity Premium Puzzle Revisited . . . . . . . . . . . . . 113
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
A Evidence of the Presence of Financial Repression 134
A.1 United States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
A.2 United Kingdom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
A.3 Japan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
A.4 France . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
A.5 Italy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
A.6 Australia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
A.7 India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
B Appendix Tables and Literature Review 157
C Data Appendix and Glossary 163
C.1 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
C.2 Variable De?nition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
C.3 Data Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Bibliography 168
vi
List of Tables
2.1 Episodes of Domestic Debt Conversions, Default or Restructuring,1920s-
1950s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Selected Measures Associated with Financial Repression . . . . . . . 45
2.3 Incidence and Magnitude of the Liquidation of Public Debt: Selected
Countries, 1945-1980 . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.4 Incidence of Liquidation Years for Di?erent Real Interest Rate Thresh-
olds: Selected Countries, 1945-1980 . . . . . . . . . . . . . . . . . . . 58
2.5 Government Revenues (interest cost savings) from the "Liquidation
E?ect:" per year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.6 Debt Liquidation through Financial Repression: Selected Countries,
1945-1955 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.7 In?ation Performance during Major Domestic Public Debt Reduction
Episodes: 28 Countries, 1790-2009 . . . . . . . . . . . . . . . . . . . 60
3.1 Countries in the Sample and Sample Periods . . . . . . . . . . . . . . 123
3.2 Summary of In?ation and Real Interest Rate (in percent) . . . . . . . 123
3.3 Incidence of Liquidation Years . . . . . . . . . . . . . . . . . . . . . . 124
3.4 Liquidation Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.5 Liquidation Revenues . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
3.6 Comparison Liquidation Revenues as percentage of: . . . . . . . . . . 125
3.7 Comparison of Median In?ation between LE Years and 1930-2010 . . 126
3.8 Regressions for In?ation Expectations . . . . . . . . . . . . . . . . . . 127
3.9 In?ation Surprises and Liquidation Years . . . . . . . . . . . . . . . . 128
3.10 Comparison between Ex Ante and Ex Post Yield to Maturity for
securities issued within sample period (in percent) . . . . . . . . . . . 128
3.11 Comparison Liquidation E?ect Revenues and In?ation Tax (as %GDP)129
3.12 Stock of Debt under di?erent paths for the in?ation rate . . . . . . . 130
3.13 Panel Estimation Results- Dependent Variable: CIR . . . . . . . . . . 130
3.14 Comparison of Real Returns Bills, Bonds and Equity 1945-1980 (in
percent) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
3.15 Percentage of the time stocks outperform bonds for di?erent holding
periods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.16 Real Returns on Bonds and Bills during 1900-2000 . . . . . . . . . . 132
3.17 Real Returns on Stocks during 1900-2000 . . . . . . . . . . . . . . . . 133
B.1 Real Interest Rates during Financial Repression and Post-Liberalization157
B.2 Measuring "Taxes" from Financial Repression: Selected Papers . . . . 160
C.1 Country Speci?c Data Sources . . . . . . . . . . . . . . . . . . . . . . 166
C.2 India: Composition of Domestic Debt (as percentage of total domestic debt)167
C.3 US: Composition of Domestic Debt (as percentage of total domestic debt) . 167
vii
List of Figures
2.1 Surges in Central Government Public Debts and their Resolution:
Advanced Economies and Emerging Markets, 1900-2011 . . . . . . . . 39
2.2 Average Ex-post Real Rate on Treasury Bills: Advanced Economies
and Emerging Markets, 1945-2009 (3-year moving averages, in percent) 40
2.3 Average Ex-post Real Discount Rate: Advanced Economies and Emerg-
ing Markets, 1945-2009 (3-year moving averages, in percent) . . . . . 41
2.4 Average Ex-post Real Rate on Deposits: Advanced Economies and
Emerging Markets, 1945-2009 (3-year moving averages, in percent) . . 42
2.5 Real Interest Rates Frequency Distributions: Advanced Economies,
1945-2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.6 Real Deposit Interest Rates Frequency Distributions: United King-
dom, 1880-2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.1 Debt-to-GDP, Real Growth Rate and Distribution of Real Returns . . 119
3.2 Frequency Distributions of Nominal and Ex Post Real Yield to Ma-
turity (YTM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
3.3 Maturity Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.4 Rolling 30-year Equity Premium . . . . . . . . . . . . . . . . . . . . . 122
viii
Chapter 1
Introduction
As a result of the recent ?nancial crisis, the public debt-to-GDP ratios of most
advanced economies have reached historically high levels. A similar phenomena was
observed after the end of World War I and the Great Depression, and again after
the end of World War II.
1
What is novel about this episode is that the debt ratios
of the private sector (both household and corporate) are also at historically high
levels. As of the end of 2009, the total debt-to-GDP ratio of Japan stood at 477
percent, 466 percent in the UK, and almost 300 percent in the US.
2
A high level of private indebtedness could worsen even further the public ?-
nances of several countries. In the past, private debts before a crisis have usually
become public debt afterwards. Reinhart (2012) points at the cases of Ireland were
public debt was around 25 percent of GDP in 2007 and has since then increased to
107 percent as of the end of 2011; and also to the case of the US where the federal
debt increased by 25 percentage points as a result of the transfer of the debts of
Fannie Mae and Freddie Mac. An added element of concern is that there is increas-
ing evidence that high debt ratios can negatively a?ect real growth once they reach
a certain threshold (See for instance, Reinhart and Rogo?, 2010; Checherita and
1
For instance, the debt ratio for advanced economies, calculated as a simple average of the
debt-to-GDP ratios of 19 advanced countries, reached 79 percent in 1922, 73 percent in 1932, 100
percent in 1946, and 70 percent in 2009.
2
Total debt includes: government, non?nancial businesses, households, and ?nancial
institutions.http://www.gfmag.com/tools/global-database/economic-data/10403-total-debt-to-
gdp.html#axzz1rMxbYikW
1
Rother, 2010; Kuman and Woo, 2010), making it less likely that countries will be
able to grow their way out of debt (at least in the short term).
Revisiting similar episodes in the past can be useful for understanding the
possible courses of action currently available to governments. This dissertation looks
at how governments have reduced high debt burdens in the past, with particular
emphasis to the post-World War II period. World War I and Great Depressions debts
were reduced primarily via default, restructurings and conversions. There were also
some cases of hyperin?ation such as Germany and Hungary. Following World War
II there were some defaults on external debt (i.e., Italy and Japan) but the majority
of the debts -which were denominated in local currency- were reduced through a
combination of ?nancial repression and in?ation. This mechanism represents a more
subtle and gradual form of debt restructuring or taxation.
The term "?nancial repression" is used to refer to a situation characterized by:
i) numerous policies and regulations which introduce frictions in ?nancial markets,
and ii) large participation of nonmarket players. The policies that will be partic-
ularly relevant are those which create captive investors for government debt, and
hence allow the government to issue debt at a rate below what the market would
charge absent any restrictions. Being able to issue debt at a below market interest
rate represents a saving in interest payments for the government. When combined
with an in?ation rate above the nominal interest rate, this leads to negative real
interest rates that e?ectively reduce government debt. An important observation is
that this mechanism does not require unanticipated in?ation in order to be e?ective.
The main database was constructed from primary sources, and replicates the
2
domestic debt portfolios of 12 countries. These portfolios include detailed infor-
mation on each security outstanding every year, and covers for most countries the
period 1945-1980. The countries in the sample are: Argentina, Australia, Belgium,
France, India, Ireland, Italy, Japan, South Africa, Sweden, the United Kingdom, and
the United States.These countries exhibit di?erent levels of economic development,
di?erent degrees of involvement in World War II, di?erent in?ationary and debt
histories. The diversity across countries is important to understand how general the
results are.
Real interest rates were signi?cantly lower during the ?nancial repression pe-
riod (1945-1980) than both in the periods before and after. This is true regardless
of the measure of interest rate that is used (whether central bank discount, treasury
bills, deposit, or lending rates and whether for advanced or emerging markets). For
the advanced economies, real ex-post interest rates were negative in about half of the
years of the ?nancial repression era compared with less than 15 percent of the time
since the early 1980s. When looking at the portfolio of domestic debt, real interest
rates were also negative half of the time between 1945-1980. This high incidence of
negative real interest rates implied large savings for the government, which averaged
between 2 to 3 percent of GDP per year.
The dissertation is structured in two chapters. Chapter 2 starts by document-
ing the large incidence of defaults following the World War I and Great Depression.
Evidence on the evolution and distribution of several di?erent measures of real in-
terest rates for advanced and emerging economies is also provided. The chapter also
documents the importance of ?nancial repression as a restructuring mechanism for
3
government debt after World War II. It contains ?rst pass estimates which show
that the e?ects are quantitatively important. Chapter 3 complements the results
of the previous chapter in three important ways. First, it presents a conceptual
framework which is useful to understand the di?erent channels through which in?a-
tion can reduce government debt, and also how to think of ?nancial repression as
a restructuring mechanism. Second, through di?erent empirical exercises, it shows
that ?nancial repression in combination with in?ation is the most important channel
through which debt was reduced. Finally, the paper looks at the key features of this
period of ?nancial repression and how the returns of other assets were a?ected. The
main ?nding of that section is that the abnormally low return on government bonds
is also re?ected in the equity premium, which almost doubled in the US during the
period 1945-1980.
Since the beginning of the 1980s the world has moved toward a more liber-
alized environment, which raises the question of how relevant the ?ndings of this
dissertation may be to explain the debt reduction that will need to take place in
the future. However, since the outbreak of the crisis there have been several regula-
tory changes which resemble those in place during the period 1945-1980 (Reinhart,
Kirkeegard, and Sbrancia; 2011). An element to keep in mind is that many of the
policies in place during the post-World War II period were not designed to reduce
government debt. On the contrary, they were designed to preserve and guarantee
the stability of the ?nancial system (i.e., ceilings on deposit rates). There were
however "unintended" consequences of these policies, which facilitated a relatively
quick debt reduction.
4
Chapter 2
The Liquidation of Government Debt (with Carmen M. Reinhart)
2.1 Introduction
Some people will think the 2
3
4
nonmarketable bond is a trick issue. We
want to meet that head on. It is. It is an attempt to lock up as much
as possible of these longer-term issues.
Assistant Secretary of the Treasury William McChesney Martin Jr.
1
The decade that preceded the outbreak of the subprime crisis in the summer
of 2007 produced a record surge in private debt in many advanced economies, in-
cluding the United States. The period prior to the 2001 burst of the "tech bubble"
was associated with a marked rise in the leverage of non?nancial corporate business;
in the years 2001-2007, debts of the ?nancial industry and households reached un-
precedented heights.
2
The decade following the crisis may yet mark a record surge
in public debt during peacetime, at least for the advanced economies. It is not
surprising that debt reduction, of one form or another, is a topic that is receiving
1
FOMC minutes, March 1â??2, 1951, remarks on the 1951 conversion of short-term marketable
US Treasury debts for 29-year nonmarketable bonds. Martin subsequently became chairman of
the Board of Governors, 1951â??70.
2
The surge in private debt is manifest in both the gross external debt ?gures of the private
sector (see Lane and Milesi-Ferretti, 2010, for careful and extensive historical documentation since
1970 and Reinharthttp://www.carmenreinhart.com/ for a splicing of their data with the latest
IMF/World Bank ?gures) and domestic bank credit (as documented in Reinhart, 2010). Relative
to GDP, these debt measures reached unprecented heights during 2007-2010 in many advanced
economies.
5
substantial attention in academic and policy circles alike.
3
Throughout history, debt/GDP ratios have been reduced by (i) economic
growth; (ii) substantive ?scal adjustment/austerity plans; (iii) explicit default or
restructuring of private and/or public debt; (iv) a sudden surprise burst in in?ation;
and (v) a steady dosage of ?nancial repression that is accompanied by an equally
steady dosage of in?ation. (Financial repression is de?ned in Box 1) It is critical
to clarify that options (iv) and (v) are viable only for domestic-currency debts.
Since these debt-reduction channels are not necessarily mutually exclusive, histori-
cal episodes of debt-reduction have owed to a combination of more than one of these
channels.
4
Hoping that substantial public and private debt overhangs are resolved by
growth may be uplifting, but it is not particularly practical from a policy stand-
point. The evidence, at any rate, is not particularly encouraging, as high levels
of public debt appear to be associated with lower growth.
5
The e?ectiveness of
?scal adjustment/austerity in reducing debt -and particularly, their growth conse-
quences (which are the subject of some considerable debate)- is beyond the scope
of this paper. Reinhart and Rogo? (2009 and 2011) analyze the incidence of ex-
plicit default or debt restructuring (or forcible debt conversions) among advanced
3
Among recent studies, see for example, Alesina and Ardagna (2009), IMF (2010), Lilico,
Holmes and Sameen (2009) on debt reduction via ?scal adjustment and Sturzenegger and Zettle-
meyer (2006), Reinhart and Rogo? (2009) and sources cited therein on debt reduction through
default and restructuring.
4
For instance, in analyzing external debt reduction episodes in emerging markets, Reinhart,
Rogo?, and Savastano (2003) suggest that default and debt/restructuring played a leading role
in most of the episodes they identify. However, in numerous cases the debt restructurings (often
under the umbrella of IMF programs) were accompanied by debt repayments associated with some
degree of ?scal adjustment.
5
See Checherita and Rother (2010), Kumar and Woo (2010), and Reinhart and Rogo? (2010).
6
economies (through and including World War II episodes) and emerging markets as
well as hyperin?ation as debt reduction mechanisms.
The aim of this paper is to document the more subtle and gradual form of debt
restructuring or "taxation" that has occurred via ?nancial repression (as de?ned in
Box 1). We show that such repression helped reduce lofty mountains of public debt
in many of the advanced economies in the decades following World War II and
subsequently in emerging markets, where ?nancial liberalization is of more recent
vintage.
6
We ?nd that ?nancial repression in combination with in?ation played an
important role in reducing debts. In?ation need not take market participants en-
tirely by surprise and, in e?ect, it need not be very high (by historic standards). In
e?ect, ?nancial repression via controlled interest rates, directed credit, and persis-
tent, positive in?ation rates is still an e?ective way of reducing domestic government
debts in the world’s second largest economy–China.
7
Prior to the 2007 crisis, it was deemed unlikely that advanced economies could
experience ?nancial meltdowns of a severity to match those of the pre-World War II
era; the prospect of a sovereign default in wealthy economies was similarly unthink-
able.
8
Repeating that pattern, the ongoing discussion of how public debts have
6
In a recent paper, Aizenman and Marion (2010) stress the important role played by in?ation
in reducing U.S. World War II debts and develop a framework to highlight how the government
may be tempted to follow that route in the near future. However, the critical role played by
?nancial repression (regulation) in keeping nominal interest rates low and producing negative real
interest rates was not part of their analysis.
7
Bai et. al. (1999), for example, present a framework that provides a general rationale for
?nancial repression as an implicit taxation of savings. They argue that when e?ective income-
tax rates are very uneven, as common in developing countries, raising some government revenue
through mild ?nancial repression can be more e?cient than collecting income tax only.
8
The literature and public discussion surrounding "the great moderation" attests to this benign
view of the state of the macroeconomy in the advanced economies. See, for example, McConnell
and Perez-Quiros (2000).
7
been reduced in the past has focused on the role played by ?scal adjustment. It
thus appears that it has also been collectively "forgotten" that the widespread sys-
tem of ?nancial repression that prevailed for several decades (1945-1980s) worldwide
played an instrumental role in reducing or "liquidating" the massive stocks of debt
accumulated during World War II in many of the advanced countries, United States
inclusive.
9
We document this phenomenon.
The next section discusses how previous "debt-overhang" episodes have been
resolved since 1900. There is a brief sketch of the numerous defaults, restructurings,
conversions (forcible and "voluntary") that dealt with the debts of World War I
and the Great Depression. This narrative, which follows Reinhart and Rogo? (2009
and 2011), primarily serves to highlight the substantially di?erent route taken after
World War II to deal with the legacy of high war debts. Section III provides a short
description of the types of ?nancial sector policies that facilitated the liquidation
of public debt. Hence, our analysis focuses importantly on regulations a?ecting
interest rates (with the explicit intent on keeping these low) and on policies creating
"captive" domestic audiences that would hold public debts (in part achieved through
capital controls, directed lending, and an enhanced role for nomarketable public
debts).
We also focus on the evolution of real interest rates during the era of ?nancial
repression (1945-1980s). We show that real interest rates were signi?cantly lower
during 1945-1980 than in the freer capital markets before World War II and after ?-
9
For the political economy of this point see the analysis presented in Alesina, Grilli, and
Milesi Ferretti (1993). They present a framework and stylized evidence to support that strong
governments coupled with weak central banks may impose capital controls so as to enable them to
raise more seigniorage and keep interest rates arti?cially low-facilitating domestic debt reduction.
8
nancial liberalization. This is the case irrespective of the interest rate used–whether
central bank discount, treasury bills, deposit, or lending rates and whether for ad-
vanced or emerging markets. For the advanced economies, real ex-post interest rates
were negative in about half of the years of the ?nancial repression era compared with
less than 15 percent of the time since the early 1980s.
In Section IV, we provide a basic conceptual framework for calculating the "?-
nancial repression tax," or more speci?cally, the annual "liquidation rate" of govern-
ment debt. Alternative measures are also discussed. These exercises use a detailed
data base on a country’s public debt pro?le (coupon rates, maturities, composition,
etc.) from 1945 to 1980 constructed by Sbrancia (2011). This "synthetic" public
debt portfolio re?ects the actual shares of debts across the di?erent spectra of ma-
turities as well as the shares of marketable versus nonmarketable debt (the latter
involving both securitized debt as well as direct bank loans).
Section V presents the central ?ndings of the paper, which are estimates of the
annual "liquidation tax" as well as the incidence of liquidation years for ten coun-
tries (Argentina, Australia, Belgium, India, Ireland, Italy, South Africa, Sweden,
the United Kingdom, and the United States). For the United States and the United
Kingdom, the annual liquidation of debt via negative real interest rates amounted
to 2 to 3 percent of GDP on average per year. Such annual de?cit reduction quickly
accumulates (even without any compounding) to a 20-30 percent of GDP debt re-
duction in the course of a decade. For other countries that, recorded higher in?ation
rates the liquidation e?ect was even larger. As to the incidence of liquidation years,
Argentina sets the record with negative real rates recorded in all years but one from
9
1945 to 1980.
Section VI examines the question of whether in?ation rates were systematically
higher during periods of debt reduction in the context of a broader 28-country sample
that spans both the heyday of ?nancial repression and the periods before and after.
We describe the algorithm used to identify the largest debt reduction episodes on
a country-by-country basis and, show that in 21 of the 28 countries in?ation was
higher during the larger debt reduction periods.
Finally, we discuss some of the implications of our analysis for the current debt
overhang and highlight areas for further research. Two appendices to this paper: (i)
compare our methodology to other approaches in the literature that have been used
to measure the extent of ?nancial repression or calculate the ?nancial repression
tax; (ii) provide country-speci?c details on the behavior of real interest rates across
regimes; and (iii) describe the coverage and extensive sources for the data compiled
for this study.
Box 1: Financial Repression De?ned
The pillars of "Financial repression"
The term ?nancial repression was introduced in the literature by the works
of Edward Shaw (1973) and Ronald McKinnon (1973). Subsequently, the term
became a way of describing emerging market ?nancial systems prior to the
widespread ?nancial liberalization that began in the 1980 (see Agenor and Mon-
tiel, 2008, for an excellent discussion of the role of in?ation and Giovannini
10
and de Melo, 1993 and Easterly, 1989 for country-speci?c estimates). However,
as we document in this paper, ?nancial repression was also the norm for ad-
vanced economies during the post-World War II period and in varying degrees
up through the 1980s. We describe here some of its main features.
(i) Explicit or indirect caps or ceilings on interest rates, particularly
(but not exclusively) those on government debts. These interest rate ceilings
could be e?ected through various means including: (a) explicit government reg-
ulation (for instance, Regulation Q in the United States prohibited banks from
paying interest on demand deposits and capped interest rates on saving deposits);
(b) ceilings on banks’ lending rates, which were a direct subsidy to the govern-
ment in cases where it borrowed directly from the banks (via loans rather than
securitized debt); and (c) interest rate cap in the context of ?xed coupon rate
nonmarketable debt or (d) maintained through central bank interest rate targets
(often at the directive of the Treasury or Ministry of Finance when central bank
independence was limited or nonexistent). Allan Meltzer’s (2003) monumental
history of the Federal Reserve (Volume I) documents the US experience in this
regard; Alex Cukierman’s (1992) classic on central bank independence provides
a broader international context.
(ii) Creation and maintenance of a captive domestic audience that
facilitated directed credit to the government. This was achieved through multi-
ple layers of regulations from very blunt to more subtle measures. (a) Capital
account restrictions and exchange controls orchestrated a "forced home bias" in
11
the portfolio of ?nancial institutions and individuals under the Bretton Woods
arrangements. (b) High reserve requirements (usually non-remunerated) as a
tax levy on banks (see Brock, 1989, for an insightful international comparison).
Among more subtle measures, (c) "prudential" regulatory measures requiring
that institutions (almost exclusively domestic ones) hold government debts in
their portfolios (pension funds have historically been a primary target). (d)
Transaction taxes on equities (see Campbell and Froot, 1994) also act to direct
investors toward government (and other) types of debt instruments. And (e)
prohibitions on gold transactions.
(iii) Other common measures associated with ?nancial repression aside
from the ones discussed above are, (a) direct ownership (e.g., in China or India)
of banks or extensive management of banks and other ?nancial institutions (e.g.,
in Japan) and (b) restricting entry into the ?nancial industry and directing credit
to certain industries (see Beim and Calomiris, 2000).
2.2 Default, Restructuring and Conversions: Highlights from 1920s-
1950s
Peaks and troughs in public debt/GDP are seldom synchronized across many
countries’ historical paths. There are, however, a few historical episodes where global
(or nearly global) developments, be it a war or a severe ?nancial and economic
crisis, produce a synchronized surge in public debt, such as the one recorded for
12
advanced economies since 2008. Using the Reinhart and Rogo? (2011) database for
70 countries, Figure 2.1 provides central government debt/GDP for the advanced
and emerging economies subgroups since 1900. It is a simple arithmetic average
that does not assign weight according to country size.
2.2.1 Global debt surges and their resolution
An examination of these two series identi?es a total of ?ve peaks in world
indebtedness. Three episodes (World War I, World War II, and the Second Great
Contraction, 2008-present) are almost exclusively advanced economy debt peaks;
one is unique to emerging markets (1980s debt crisis followed by the transition
economies’ collapses); and the Great Depression of the 1930s is common to both
groups. World War I and Depression debts were importantly resolved by widespread
default and explicit restructurings or predominantly forcible conversions of domestic
and external debts in both the now-advanced economies, and the emerging markets.
Notorious hyperin?ation in Germany, Hungary and other parts of Europe violently
liquidated domestic-currency debts. Table 2.1 and the associated discussion provide
a chronology of these debt resolution episodes. As Reinhart and Rogo? (2009 and
2011) document, debt reduction via default or restructuring has historically been
associated with substantial declines in output in the run-up to as well as during the
credit event and in its immediate aftermath.
The World War II debt overhang was importantly liquidated via the combina-
tion of ?nancial repression and in?ation, as we shall document. This was possible
13
because debts were predominantly domestic and denominated in domestic curren-
cies. The robust post-war growth also contributed importantly to debt reduction
in a way that was a marked contrast to the 1930s, when the combined e?ects of
de?ation and output collapses worked to worsen the debt/GDP balance in the way
stressed by Irving Fisher (1933).
The resolution of the emerging market debt crisis involved a combination of
default or restructuring of external debts, explicit default, or ?nancial repression
on domestic debt. In several episodes, notably in Latin America, hyperin?ation in
the mid-to-late 1980s and early 1990s completed the job of signi?cantly liquidating
(at least for a brief interlude) the remaining stock of domestic currency debt (even
when such debts were indexed, as was the case of Brazil).
10
2.2.2 Default, restructurings and forcible conversions in the 1930s
Table 2.1 lists the known "domestic credit events" of the Depression. Default
on or restructuring of external debt (see the extensive notes to the table) also of-
ten accompanied the restructuring or default of the domestic debt. All the Allied
governments, with the exception of Finland, defaulted on (and remained in default
through 1939 and never repaid) their World War I debts to the United States as
economic conditions deteriorated worldwide during the 1930s.
11
Thus, the high debts of World War I and the subsequent debts associated with
the Depression of the 1930s were resolved primarily through default and restructur-
10
Backward-looking indexation schemes are not particularly e?ective in hyperin?ationary con-
ditions.
11
Finland, being under threat of Soviet invasion at the time, maintained payments on its debts
to the United States so as to maintain the best possible relationship.
14
ing. Neither economic growth nor in?ation contributed much. In e?ect, for all 21
now-advanced economies, the median annual in?ation rate for 1930-1939 was barely
above zero (0.4 percent).
12
Real interest rates remained high through signi?cant
stretches of the decade.
It is important to stress that during the period after World War I the gold
standard was still in place in many countries, which meant that monetary policy was
subordinated to keep a given gold parity. In those cases, in?ation was not a policy
variable available to policymakers in the same way that it was after the adoption of
?at currencies.
12
See Reinhart and Reinhart (2010).
15
Table 2.1: Episodes of Domestic Debt Conversions, Default or Restructuring,1920s-
1950s
Country Dates Commentary
Australia 1931/1932 The Debt Conversion Agreement Act in 1931/32 which
appears to have done something similar to the later NZ
induced conversion. See New Zealand entry.
1
Bolivia 1927 Arrears of interest lasted until at least 1940.
Canada (Al-
berta)
April 1935 The only province to default - which lasted for about 10
years.
China 1932 First of several "consolidations", monthly cost of domes-
tic service was cut in half. Interest rates were reduced
to 6 percent (from over 9 percent)-amortization periods
were about doubled in length.
France 1932 Various redeemable bonds with coupons between 5 and 7
percent, converted into a 4.5 percent bond with maturity
in 75 years.
Greece 1932 Interest on domestic debt was reduced by 75 percent
since 1932; Domestic debt was about 1/4 of total public
debt.
Italy November 6th,
1926
Issuance of Littorio. There were 20.4 billion lire subject
to conversion, of which 15.2 billion were "Buoni Ordi-
nari" (short term securities).
Italy February 3rd,
1934
5 percent Littorio (see entry above) converted into 3.5
percent Redimibile
Mexico 1930s Service on external debt was suspended in 1928. Dur-
ing the 1930s, interest payments included "arrears of
expenditure and civil and military pensions."
16
Table 2.1 – Continued from previous page
Country Dates Commentary
New Zealand 1933 In March 1933 the New Zealand Debt Conversion Act
was passed providing for voluntary conversion of internal
debt amounting to 113 million pounds to a basis of 4 per
cent for ordinary debt and 3 per cent for tax-free debt.
Holders had the option of dissenting but interest in the
dissented portion was made subject to an interest tax of
33.3 per cent.
1
Peru 1931 After suspending service on external debt on May 29,
Peru made "partial interest payments" on domestic
debt.
Romania February 1933 Redemption of domestic and foreign debt is suspended
(except for three loans).
Spain October 1936-
April 1939
Interest payments on external debt were suspended, ar-
rears on domestic debt service.
United States 1933 Abrogation of the gold clause. In e?ect, the U.S. refused
to pay Panama the annuity in gold due to Panama ac-
cording to a 1903 treaty. The dispute was settled in 1936
when the US paid the agreed amount in gold balboas.
United King-
dom
1932 Most of the outstanding WWI debt was consolidated
into a 3.5 percent perpetual annuity. This domestic debt
conversion was apparently voluntary. However, some of
the WWI debts to the United States were issued under
domestic (UK) law (and therefore classi?ed as domestic
debt) and these were defaulted on following the end of
the Hoover 1931 moratorium.
17
Table 2.1 – Continued from previous page
Country Dates Commentary
Uruguay November 1, 1932
- February, 1937
After suspending redemption of external debt on Jan-
uary 20, redemptions on domestic debt were equally sus-
pended.
Austria December 1945 Restoration of schilling (150 limit per person). Remain-
der placed in blocked accounts. In December 1947, large
amounts of previously blocked schillings invalidated and
rendered worthless. Temporary blockage of 50 percent
of deposits.
Germany June 20, 1948 Monetary reform limiting 40 Deutschemark per person.
Partial cancellation and blocking of all accounts.
Japan March 2, 1946-
1952
After in?ation, exchange of all bank notes for new is-
sue (1 to 1) limited to 100 yen per person. Remaining
balances were deposited in blocked accounts.
Russia 1947 The monetary reform subjected privately held currency
to a 90 percent reduction.
April 10 1957 Repudiation of domestic debt (about 253 billion rubles
at the time).
Sources: Reinhart and Rogo? (2011) and the authors.
1
See Schedvin (1970) and Prichard (1970), for accounts of the Australian and New Zealand
conversions, respectively, during the Depression. Michael Reddell kindly alerted us to these
episodes and references. Alex Pollock pointed out the relevance of widespread restrictions
on gold holdings in the United States and elsewhere during the ?nancial repression era.
Notes: We have made signi?cant further progress in sorting out the defaults on World War
I debts to the United States, notably by European countries. In all cases these episodes are
classi?ed as a default on external debts. However, in some case â??such as the UK–some of
the WWI debts to the US were also issued under the domestic law and, as such, would also
qualify as a domestic default. The external defaults on June 15, 1934 included: Austria,
Belgium, Czechoslovakia, Estonia, France, Greece, Hungary, Italy, Latvia, Poland, United
Kingdom. Only Finland made payments. See New York Times, June 15, 1934.
18
2.3 Financial Repression: policies and evidence from real interest rates
2.3.1 Selected ?nancial regulation measures during the "era of ?nancial
repression"
One salient characteristic of ?nancial repression is its pervasive lack of trans-
parency. The reams of regulations applying to domestic and cross-border ?nancial
transactions and directives cannot be summarized by a brief description. Table 2.2
makes this clear by providing a broad sense of the kinds of regulations on inter-
est rates and cross-border and foreign exchange transactions and how long these
lasted since the end of World War II in 1945. A common element across countries
"?nancial architecture" not brought out in Table 2.2 is that domestic government
debt played a dominant role in domestic institutions’ asset holdings–notably that
of pension funds. High reserve requirements, relative to the current practice in ad-
vanced economies and many emerging markets, were also a common way of taxing
the banks not captured in our minimalist description. The interested reader is re-
ferred to Brock (1989) and Agenor and Montiel (2008), who focus on the role of
reserve requirements and their link to in?ation (see also Appendix Table B.2 and
accompanying discussion.)
19
2.3.2 Real Interest Rates
One of the main goals of ?nancial repression is to keep nominal interest rates
lower than would otherwise prevail. This e?ect, other things equal, reduces the
governments’ interest expenses for a given stock of debt and contributes to de?cit
reduction. However, when ?nancial repression produces negative real interest rates,
this also reduces or liquidates existing debts. It is a transfer from creditors (savers)
to borrowers (in the historical episode under study here–the government).
The ?nancial repression tax has some interesting political-economy properties.
Unlike income, consumption, or sales taxes, the "repression" tax rate (or rates) are
determined by ?nancial regulations and in?ation performance that are opaque to
the highly politicized realm of ?scal measures. Given that de?cit reduction usually
involves highly unpopular expenditure reductions and (or) tax increases of one form
or another, the relatively "stealthier" ?nancial repression tax may be a more politi-
cally palatable alternative to authorities faced with the need to reduce outstanding
debts. As discussed in Obstfeld and Taylor (2004) and others, liberal capital- market
regulations (the accompanying market-determined interest rates) and international
capital mobility reached their heyday prior to World War I under the umbrella of
the gold standard. World War I and the suspension of convertibility and interna-
tional gold shipments it brought, and, more generally, a variety of restrictions on
cross-border transactions were the ?rst blows to the globalization of capital. Global
capital markets recovered partially during the roaring twenties, but the Great De-
pression, followed by World War II, put the ?nal nails in the co?n of laissez faire
20
banking. It was in this environment that the Bretton Woods arrangement of ?xed
exchange rates and tightly controlled domestic and international capital markets
was conceived.
13
In that context, and taking into account the major economic dis-
locations, scarcities, etc. which prevailed at the closure of the second great war, we
witness a combination of very low nominal interest rates and in?ationary spurts of
varying degrees across the advanced economies. The obvious result were real inter-
est rates -whether on treasury bills (Figure 2.2), central bank discount rates (Figure
2.3), deposits (Figure 2.4), or loans (not shown)- that were markedly negative during
1945-1946.
For the next 35 years or so, real interest rates in both advanced and emerging
economies would remain consistently lower than the eras of freer capital mobility
before and after the ?nancial repression era. In e?ect, real interest rates (Figures
2.2-2.4) were on average negative.
14
Binding interest rate ceilings on deposits (which
kept real ex post deposit rates even more negative than real ex-post rates on treasury
bills, as shown in Figures 2.2 and 2.4) "induced" domestic savers to hold government
bonds. What delayed the emergence of leakages in the search for higher yields
(apart from prevailing capital controls) was that the incidence of negative returns
on government bonds and on deposits was (more or less) a universal phenomenon
at this time.
15
The frequency distributions of real rates for the period of ?nancial
13
In a framework where there are both tax collection costs and a large stock of domestic govern-
ment, Aizenman and Guidotti, (1994) show how a government can resort to capital controls (which
lower domestic interest rates relative to foreign interest rates) to reduce the costs of servicing the
domestic debt.
14
Note that real interest rates were lower in a high-economic-growth period of 1945 to 1980
than in the lower growth period 1981-2009; this is exactly the opposite of the prediction of a basic
growth model and therefore indicative of signi?cant impediments to ?nancial trade
15
A comparison of the return on government bonds to that of equity during this period and its
21
repression (1945-1980) and the years following ?nancial liberalization (roughly 1981-
2009 for the advanced economies) shown in the three panels of Figure 2.5, highlight
the universality of lower real interest rates prior to the 1980s and the high incidence
of negative real interest rates.
Such negative (or low) real interest rates were consistently and substantially
below the real rate of growth of GDP, this is consistent with the observation of
Elmendorf and Mankiw (1999) when they state "An important factor behind the
dramatic drop (in US public debt) between 1945 and 1975 is that the growth rate of
GNP exceeded the interest rate on government debt for most of that period." They
fail to explain why this con?guration should persist over three decades in so many
countries.
Real interest rates on deposits were negative in about 60 percent of the obser-
vations. In e?ect, real ex-post deposit rates were below one percent about 83 percent
of the time. Appendix Table B.1, which shows for each country average real interest
rates during the ?nancial repression period (the dates vary, as highlighted in Table
B.1, depending on when interest rates were liberalized) and thereafter, substantiates
our claims that low and negative real interest rates (by historical standards) were
the norm across countries with very di?erent levels of economic development.
The preceding analysis sets the general tone of what to expect, in terms of
real rates of return on a portfolio of government debt, during the era of ?nancial
repression. For the United States, for example, Homer and Sylla (1963) describe
connection to "the equity premium puzzle" can be found in Sbrancia (2011).
22
1946-1981 as the second (and longest) bear bond market in US history.
16
To reit-
erate the point that the low real interest rates of the ?nancial repression era were
exceptionally low in relation to not only the post-liberalization period but also the
more liberal ?nancial environment of pre-World War II, Figure 2.6 plots the fre-
quency distribution of real interest rates on deposits for the United Kingdom over
three subperiods, 1880-1939,
17
1945-1980, and 1981-2010.
The preceding analysis of real interest rates despite being qualitatively sugges-
tive falls short of providing estimates of the magnitude of the debt-servicing savings
and outright debt liquidation that accrued to governments during this extended pe-
riod. To ?ll in that gap the next section outlines the methodological approach we
follow to quantify the ?nancial repression tax, while Section 5 presents the main
results.
2.4 The Liquidation of Government Debt: Conceptual and Data Issues
This section discusses the data and methodology we develop to arrive at esti-
mates of how much debt was liquidated via a combination of low nominal interest
rates and higher in?ation rates, or what we term "the liquidation e?ect."
18
Data requirements. Reliable estimates of the liquidation e?ect require con-
siderable data, most of which are not readily available from even the most com-
prehensive electronic databases. Indeed, most of the data used in these exercises
16
They identify 1899-1920 as the ?rst US bear bond market.
17
Excluding the WWI period.
18
Table B.2 and its accompanying discussion also examines other approaches to quantifying the
?nancial repression tax.
23
come from a broad variety of historical government publications, many which are
quite obscure, as detailed in the Data Appendix. The calculation of the "liquidation
e?ect" is a clear illustration of a case where the devil lies in the details, as the struc-
ture of government debt varies enormously across countries and within countries
over time. Di?erences in coupon rates, maturity, distribution of marketable and
nonmarketable debt, and securitized debt versus loans from ?nancial institutions
importantly shape the overall cost of debt ?nancing for the government. There is
no "single" government interest rate (such as a 3-month T-bill or a 10-year bond)
that is appropriate to apply to a hybrid debt stock. The starting point to come up
with a measure that re?ects the true cost of debt ?nancing is a reconstruction of
the government’s debt pro?le over time.
Sample. We employ two samples in our empirical analysis. We use the
database from Sbrancia (2011) of the government’s debt pro?les for 10 countries
(Argentina, Australia, Belgium, India, Ireland, Italy, South Africa, Sweden, the
United Kingdom, and the United States). These were constructed from primary
sources over the period 1945-1990 where possible or over shorter intervals (deter-
mined by data availability) for a subset of the sample. For the benchmark or basic
calculations (described below), this involves data on a detailed composition of debt,
including maturity, coupon rate, and outstanding amounts by instrument. For a
more comprehensive measure, which takes into account capital gains or losses of
holding government debt, bond price data are also required. In all cases, we also
use o?cial estimates of consumer price in?ation, which at various points in history
24
may signi?cantly understate the true in?ation rates.
19
Data on Nominal GDP and
government tax revenues are used to express the estimates of the liquidation e?ect
as ratios that are comparable across time and countries.
For our broader analysis of the behavior of in?ation during major debt reduc-
tion episodes, which has far less demanding data requirements (domestic public debt
outstanding/GDP and in?ation rates) our sample broadens to 28 countries from all
regions for 1790-2010 (or subsamples therein). The countries and their respective
coverage are listed in Appendix C.
2.4.1 Benchmark basic estimates of the "liquidation e?ect"
The debt portfolio. We construct a "synthetic portfolio"
20
for the government’s
total debt stock at the beginning of the year (?scal or calendar, as noted). This
portfolio re?ects the actual shares of debts across the di?erent spectra of maturities
as well as the shares of marketable versus nonmarketable debt.
Interest rate on the portfolio. The "aggregate" nominal interest rate for a
particular year is the coupon rate on a particular type of debt instrument weighted
by that instrument’s share in the total stock of debt.
21
We then aggregate across
19
This is primarily due to the existence of price controls which were mainly imposed during
WWII and remained for several years after the end of the con?ict. See Friedman and Schwartz
(1982) for estimates of the actual price level in the US and UK, and Wiles (1952) for post-World
War II United Kingdom.
20
The term "synthetic" is used in the sense that a hypothetical investor holds the total portfolio
of government debt at the beginning of the period, which is de?ned as either the beginning of the
calendar year or the ?scal year, depending on how the debt data is reported by the particular
country. Country speci?cs are detailed in the data appendix. The weights in this hypothetical
portfolio are given by the actual shares of each component of debt in the total domestic debt of
the government.
21
Giovannini and de Melo (1993) state "the choice of a "representative" interest rate on domestic
liabilities an almost impossible task and because there are no reliable breakdowns of domestic and
foreign liabilities by type of loan and interest rate charged." This is precisely the almost impossible
25
all debt instruments. The real rate of interest,
r
t
=
i
t?1
??
t
1?
t
(2.1)
is calculated on an ex-post basis using CPI in?ation for the corresponding
one-year period. It is a before-tax real rate of return (excluding capital gains or
losses).
22
A de?nition of debt "liquidation years." Our benchmark calculations de?ne
a liquidation year, as one in which the real rate of interest (as de?ned above) is
negative (below zero). This is a conservative de?nition of liquidation year; a more
comprehensive de?nition would include periods where the real interest rate on gov-
ernment debt was below a "market" real rate.
23
It is possible that if the equilibrium
interest rate was negative, using this de?nition would be overestimating the actual
e?ect.
Savings to the government during liquidation years. This concept
captures the savings (in interest costs) to the government from having a negative real
interest rate on government debt. (As noted it is a lower bound on saving of interest
costs, if the benchmark used assumed, for example a positive real rate of, say, two or
three percent.) These savings can be thought of as having "a revenue-equivalent" for
the government, which like regular budgetary revenues can be expressed as a share
task we undertake here. Their alternative methodology is described in appendix Table B.2
22
Some of the observations on in?ation are su?ciently high to make the more familiar linear
version of the Fisher equation a poor approximation.
23
However, determining what such a market rate would be in periods of pervasive ?nancial
repression requires assumptions about whether real interest rates during that period would have
comparable to the real interest that prevailed in period when market were liberalized and prices
were market determined.
26
of GDP or as a share of recorded tax revenues to provide standard measures of the
"liquidation e?ect" across countries and over time. The saving (or "revenue") to
the government or the "liquidation e?ect" or the "?nancial repression tax" is
the real (negative) interest rate times the "tax base," which is the stock of domestic
government debt outstanding.
2.4.2 An alternative measure of the liquidation e?ect based on total
returns
Thus far, our measure of the liquidation a?ect has been con?ned to savings
to the government by way of annual interest costs. However, capital losses (if bond
prices fall) may also contribute importantly to the calculus of debt liquidation over
time. This is the case because the market value of the debt will actually be lower than
its face value. The market value of government debt obviously matters for investors’
wealth but also measures the true capitalized value of future coupon and interest
payments. Moreover, a government (or its central bank) buying back existing debt
could directly and immediately lower the par value of existing obligations. Once
we take into account potential price changes, the total nominal return or holding
period return (HPR) for each instrument is given by:
HPR
t
=
(P
t
?P
t?1
) + C
t
P
t?1
(2.2)
where and are the prices of the bond at time and respectively, and is the annual
interest payment (i.e., the nominal coupon rate).
27
We use this total return measure as a supplement rather than as our core or
benchmark "liquidation measure" (despite the fact that it incorporates more infor-
mation on the performance of the bond portfolio).
24
Bond price data are available
only for a subset of the securities that constitute the government portfolio and, more
generally, consistent time series price data are more di?cult to get for some of the
countries in our sample. It is also worth noting that while price movements for
di?erent bonds are generally in the same direction during a particular year, there
are signi?cant di?erences in the magnitudes of the price changes. This cross-bond
variation in price performance makes it di?cult to infer the price of nonmarketable
debt (for which there are no price data altogether), as well as marketable bonds
for which there is no price data. As before, we de?ne "liquidation years" as those
periods in which the real return of the portfolio is negative.
2.4.3 The role of in?ation and currency depreciation
The idea of governments using in?ation to liquidate debt is hardly a new one
since the widespread adoption of ?at currency, as discussed earlier. It is obvious that
for any given nominal interest rate a higher in?ation rate reduces the real interest
rate on the debt, thus increasing the odds that real interest rates become negative
and the year is classi?ed as a "liquidation year." Furthermore, it is also evident that
for any year that is classi?ed as a liquidation year the higher the in?ation rate (for
a given coupon rate) the higher the saving to the government.
24
See for example, Calvoâ??s (1989) framework which highlights the role of in?ation in debt
liquidation even in the presence of short-term debt.
28
2.5 The Liquidation of Government Debt: Empirical Estimates
This section presents estimates of the "liquidation e?ect" for ten advanced
and emerging economies for most of the post-World War II period. Our main interest
lies in the period prior to the process of ?nancial liberalization that took hold during
the 1980s-that is, the era of ?nancial repression. However, as noted, this three-plus
decade-long stretch is by no means uniform. The decade immediately following
World War II was characterized by a very high public debt overhang-legacy of the
war, a higher incidence of in?ation, and often multiple currency practices (with
huge black market exchange rate premiums) in many advanced economies.
25
The
next decade (1960s) was the heyday of the Bretton Woods system with heavily
regulated domestic and foreign exchange markets and more stable in?ation rates
in the advanced economies (as well as more moderate public debt levels). The
1970s was quite distinct from the prior decades, as leakages in ?nancial regulations
proliferated, the ?xed exchange rate arrangements under Bretton Woods among the
advanced economies broke down, and in?ation began to resurface in the wake of
the global oil shock and accommodative monetary policies in the United States and
elsewhere. To this end, we also provide estimates of the liquidation of government
debt for relevant subperiods.
25
See De Vries (1969), Horse?eld (1969), Reinhart and Rogo? (2002).
29
2.5.1 Incidence and magnitude of the "liquidation tax"
Table 2.3 provides information on a country-by-country basis for the period
under study; the incidence of debt liquidation years (as de?ned in the preceding
section); the listing of the liquidation years; the average (negative) real interest rate
during the liquidation years; and the minimum real interest rate recorded (and the
year in which that minimum was reached). Given its notorious high and chronic in-
?ation history coupled with heavy-handed domestic ?nancial regulation and capital
controls during 1944-1974, it is not surprising that Argentina tops the list. Almost
all the years (92 percent) were recorded as liquidation years, as the Argentine real
ex-post interest rates were negative in every single year during 1944-1980 except
for 1953 (a just de?ationary year). For India, that share was 53 percent (slightly
more than one half of the 1949-1980 observations recorded negative real interest
rates). Before reaching the conclusion that this debt liquidation through ?nancial
repression was predominantly an emerging market phenomenon, it is worth noting
that for the United Kingdom the share of liquidation years was about 58 percent
during 1945-1980. For the United States, the world’s ?nancial center, half of the
years during that same period Treasury debt had negative real interest rates.
As to the magnitudes of the ?nancial repression tax (Table 2.3), real inter-
est rates were most negative for Argentina (reaching a minimum of -72.3 percent in
1976). The share of domestic government debt in Argentina (and other Latin Amer-
ican countries) in total (domestic plus external) public debt was substantial during
1900-1950s; it is not surprising that in light of these real rates the domestic debt
30
market all but disappeared and capital ?ight marched upwards (capital controls
notwithstanding). By the late 1970s Argentina and many other chronic in?ation
countries were predominantly relying on external debt.
26
Italian real interest rates
right after World War II were as negative as 28 percent (in 1947). For the Unites
States real rates were on average -7 percent during 1945-1947 (on average the US
had -3.5 percent real rates during the liquidation years).
There are two distinct patterns in the ten-country sample evident from an
inspection of the timing of the incidence and magnitude of the negative real rates.
The ?rst of these is the cases where the negative real rates (?nancial repression
tax) were most pronounced in the years following World War II (as war debts were
importantly in?ated away). This pattern is most evident in Australia, the United
Kingdom and the United States, although negative real rates re-emerge following
the breakdown of Bretton Woods in 1974-1975. Then there are the cases where
there is a more persistent or chronic reliance on ?nancial repression throughout the
sample as a way of funding government de?cits and/or eroding existing government
debts. The cases of Argentina and India in the emerging markets and Belgium and
Sweden in the advanced economies stand out in this regard.
The preceding analysis, as noted, adopts a very narrow, conservative calcula-
tion of both the incidence of the "liquidation e?ect" or the ?nancial repression tax.
Much of the literature on growth, as well as standard calibration exercises involving
subjective rates of time preference assume benchmark real interest rates of three
percent per annum and even higher. Thus, a threshold that only examines periods
26
See Reinhart and Rogo? (2011)’s forgotten history of domestic debt.
31
where real interest rates were actually negative is bound to underestimate the inci-
dence of "abnormally low" real interest rates during the era of ?nancial repression
(approximately taken to be 1945-1980). To assess the incidence of more broadly
de?ned low real interest rates, Table 3.3 presents for the 10 core countries the share
of years where real returns on a portfolio of government debt (as de?ned earlier)
were below zero (as in Table 2.3), one, two, and three percent, respectively.
27
In the era of ?nancial repression that we examine here, real ex post interest
rates on government debt reached three percent in only two years in the United
States; in e?ect in nearly 60 percent of the years real interest rates were below one
percent. The incidence of "abnormally low" real interest rates is comparable for
the United Kingdom and Australia-both countries had sharp and relatively rapid
declines in public debt to GDP following World War II.
28
Even in countries with
substantial economic and ?nancial volatility during this period (including Ireland,
and Italy), real interest rates on government debt above three percent were relatively
rare (accounting for no more than 20 percent of the observations).
2.5.2 Estimates of the Liquidation E?ect
Having documented the high incidence of "liquidation years" (even by conser-
vative estimates), we now calculate the magnitude of the savings to the government
(?nancial repression tax or liquidation e?ect). These estimates take "the tax rate"
27
An alternative strategy would be to use a growth model to calibrate the relationship between
the real interest rate and output growth for the counterfactual of free markets. That, however,
would make the results model speci?c.
28
"Abnormally low" by the historical standards which include periods of liberalized ?nancial
markets before and after 1945-1980; see Homer and Sylla’s (2005) classic book for a comprehensive
and insightful history of interest rates.
32
(the negative real interest rate) and multiply it by the "tax base" or the stock of
debt. Table 2.5 reports these estimates for each country.
The magnitudes are in all cases non-trivial, irrespective of whether we use the
benchmark measure that is exclusively based on interest rate (coupon yields) or
the alternative measure that includes capital gains (or losses) for the cases where
the bond price data is available. For the United States and the United Kingdom
the annual liquidation of debt via negative real interest rates amounted on average
to 2 and 3 percent of GDP a year. Obviously, annual de?cit reduction of 2 to 3
percent of GDP quickly accumulates (even without any compounding) to a 20 to
30 percent of GDP debt reduction in the course of a decade. Interestingly (but
not entirely surprising), the average annual magnitude of the liquidation e?ect for
Argentina is about the same as that of the UK, despite the fact that the average real
interest rate averaged about -3.5 percent for the UK and -21 percent for Argentina
during liquidation years in the 1945-1980 repression era. Just as money holdings
secularly shrink during periods of high and chronic in?ation, so does the domestic
debt market.
29
Argentina’s "tax base" (domestic public debt) shrank steadily during
this period; at the end of World War II nearly all public debt was domestic and by
the early 1980s domestic debt accounted for less than
1
2
of total public debt. Without
the means to liquidate external debts, Argentina defaulted on its external obligations
in 1982.
Countries like Ireland, India, Sweden and South Africa that did not experience
a massive public debt build-up during World War II recorded more modest annual
29
These issues are examined in Reinhart and Rogo? (2011).
33
savings (but still substantive) during the heyday of ?nancial repression.
30
2.6 In?ation and Debt Reduction
We have argued that in?ation is most e?ective in liquidating government debts
(or debts in general), when interest rates are not able to respond to the rise in in?a-
tion and in in?ation expectations.
31
This disconnect between nominal interest rates
and in?ation can occur if: (i) the setting is one where interest rates are either admin-
istered or predetermined (via ?nancial repression, as described); (ii) all government
debts are ?xed-rate and long maturities and the government has no new ?nancing
needs (even if there is no ?nancial repression the long maturities avoid rising interest
costs that would otherwise prevail if short maturity debts needed to be rolled over);
and (iii) all (or nearly all) debt is liquidated in one "surprise" in?ation spike. Our
attention thus far has been con?ned to the ?rst on that list, the ?nancial repres-
sion environment. The second scenario, where governments only have long-term,
?xed-rate debt outstanding and have no new ?nancing needs (de?cits) remain to be
identi?ed (however, we have a sense such episodes are relatively rare). This leaves
the third case where debts are swiftly liquidated via an in?ation spike (or perhaps
more appropriately surge). To attempt to identify potential episodes of the latter,
we conduct a simple exercise.
32
30
It is important to note that while ?nancial repression wound down in most of the advanced
economies in the sample by the mid 1980s, it has persisted in varying degrees in India through the
present (with its system of state-owned banks and widespread capital controls) and in Argentina
(except for the years of the "Convertibility Plan," April 1991-December 2001).
31
That is, the coe?cient in the Fisher equation is less than one.
32
See Chapter 3, Section 4.4 for an exercise where in?ation expectations are estimated.
34
The exercise begins by identifying debt-reduction episodes and then focusing
on the largest of these. Any decline in debt/GDP over a three year window classi?es
as a debt-reduction episode. For this pool of debt-reduction episodes, we construct
their frequency distribution (for each country) and focus on the lower (ten percent)
tail of the distribution to identify the "largest" three-year debt reduction episodes.
This algorithm biases our selection of episodes toward the more sudden (or abrupt)
ones (even if these are later reversed) which might a priori be attributable to some
combination of a booming economy, a substantive ?scal austerity plan, or a burst
in in?ation/liquidation, or explicit default or restructuring. A milder but steady
debt reduction process that lasts over many years would be identi?ed as a series of
episodes-but if the decline in debt over any particular three-year window is modest
it may not be large enough to fall in the lower ten percent of all the observations.
This exercise helps ?ag episodes where in?ation is likely to have played a signif-
icant role in public debt reduction but does not provide estimates of how much debt
was liquidated (as in the preceding analysis). Because we only require information
on domestic public debt/GDP and in?ation, we expand our coverage to 28 countries
predominantly (but not exclusively) over 1900-2009. Thus, we are not exclusively
focusing on the period of ?nancial repression but examining more broadly the role
of in?ation and debt reduction in the countries’ histories. Table 8 lists the largest
debt reduction episodes by country, the last year of the 3-year episode is shown for
each country; the year that appears in italics represents the largest single-episode
of debt reduction. The next two columns of the table are devoted to the average
and median in?ation performance during the debt reduction episodes listed in the
35
second column in comparison to the in?ation performance (average and median) for
the full sample (the coverage, which varies by country, is shown in Table A.3). In 22
of 28 countries, in?ation is signi?cantly higher in the episodes of debt reduction than
for the full sample. In the extreme cases, it is the wholesale liquidation of domestic
debt, such as during the German hyperin?ation of the early 1920s and the long-
lasting Brazilian and Argentine hyperin?ations of the early 1990s. Even without
these extreme cases, the in?ation di?erentials between the debt reduction episodes
and the full sample are suggestive of the use of in?ation (intentionally or because it
became unmanageable) to reduce (or liquidate) government debts even in periods
outside the era of heavy ?nancial repressions. The evidence is only suggestive of
this interpretation, as no explicit causal pattern is tested.
2.7 Concluding Remarks
The substantial tax on ?nancial savings imposed by the ?nancial repression
that characterized 1945-1980 was a major factor explaining the relatively rapid re-
duction of public debt in a number of the advanced economies. This fact has been
largely overlooked in the literature and discussion on debt reduction. The UK’s his-
tory o?ers a pertinent illustration. Following the Napoleonic Wars, the UK’s public
debt was a staggering 260 percent of GDP; it took over 40 years to bring it down to
about 100 percent (a massive reduction in an era of price stability and high capital
mobility anchored by the gold standard). Following World War II, the UK’s public
36
debt ratio was reduced by a comparable amount in 20 years.
33
The ?nancial repression route taken at the creation of the Bretton Woods
system was facilitated by initial conditions after the war, which had left a legacy of
pervasive domestic and ?nancial restrictions. Indeed, even before the outbreak of
World War II, the pendulum had begun to swing away from laissez-faire ?nancial
markets toward heavier-handed regulation in response to the widespread ?nancial
crises of 1929-1931. But one cannot help thinking that part of the design principle of
the Bretton Woods system was to make it easier to work down massive debt burdens.
The legacy of ?nancial crisis made it easier to package those policies as prudential.
To deal with the current debt overhang, similar policies to those documented here
may re-emerge in the guise of prudential regulation rather than under the politically
incorrect label of ?nancial repression. Moreover, the process where debts are being
"placed" at below market interest rates in pension funds and other more captive
domestic ?nancial institutions is already under way in several countries in Europe.
There are many bankrupt (or nearly so) pension plans at the state level in the
United States that bear scrutiny (in addition to the substantive unfunded liabilities
at the federal level).
Markets for government bonds are increasingly populated by nonmarket play-
ers, notably central banks of the United States, Europe and many of the largest
emerging markets, calling into question what the information content of bond prices
33
Peak debt/GDP was 260.6 in 1819 and 237.9 percent in 1947. Real GDP growth was about
the same during the two debt reduction periods (1819-1859) and (1947-1967), averaging about 2.5
percent per annum (the comparison is not exact as continuous GDP data begins in 1830). As such,
higher growth cannot obviously account for the by far faster debt reduction following World War
II.
37
are relatively to their underlying risk pro?le. This decoupling between interest rates
and risk is a common feature of ?nancially repressed systems. With public and
private external debts at record highs, many advanced economies are increasingly
looking inward for public debt placements. While to state that initial conditions on
the extent of global integration are vastly di?erent at the outset of Bretton Woods
in 1946 and today is an understatement, the direction of regulatory changes have
many common features. The incentives to reduce the debt overhang are more com-
pelling today than about half a century ago. After World War II, the overhang
was limited to public debt (as the private sector had painfully deleveraged through
the 1930s and the war); at present, the debt overhang many advanced economies
face encompasses (in varying degrees) households, ?rms, ?nancial institutions and
governments.
38
Figure 2.1: Surges in Central Government Public Debts and their Resolution: Ad-
vanced Economies and Emerging Markets, 1900-2011
0
20
40
60
80
100
120
1901 1911 1921 1931 1941 1951 1961 1971 1981 1991 2001 2011
WWI and Depression debts
(advanced economies:
default, restructuring and
conversions--a few
hyperinflations)
Advanced
economies
Emerging
Markets
Great depression
debts
(emerging markets-default)
WWII debts:
(Axis countries: default and
financial repression/inflation
Allies: financial
repression/inflation)
1980s Debt Crisis
(emerging
markets:default,
restructuring, financial
repression/inflation and
several hyperinflations)
Second Great
Contraction
(advanced
economies)
Sources: Reinhart (2010), Reinhart and Rogo? (2009 and 2011), sources cited
therein and the authors.
Notes: Listed in parentheses below each debt-surge episode are the main mech-
anisms for debt resolution besides ?scal austerity programs which were not imple-
mented in any discernible synchronous pattern across countries in any given episode.
Speci?c default/restructuring years by country are provided in the Reinhart-Rogo?
database and a richer level of detail for 1920s-1950s (including various conversions
are listed in Table 1). The "typical" forms of ?nancial repression measures are
discussed in Box 1 and greater detail for the core countries are provided in Table 2.
39
Figure 2.2: Average Ex-post Real Rate on Treasury Bills: Advanced Economies and
Emerging Markets, 1945-2009 (3-year moving averages, in percent)
1945-1980 1981-2009
-1.6 2.8
-1.2 2.6
Average Real Treasury Bill Rate
Advanced economies
Emerging markets
-15.0
-10.0
-5.0
0.0
5.0
10.0
1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
`
Advanced economies
(3-year moving average)
Emerging Markets
(3-year moving average)
Financial Repression Era
Sources: International Financial Statistics, International Monetary Fund, various
sources listed in the Data Appendix, and authors’ calculations.
Notes:The advanced economy aggregate comprises: Australia, Belgium, Canada,
Finland, France, Germany, Greece, Ireland, Italy, Japan, New Zealand, Sweden,
the United States, and the United Kingdom. The emerging market group consists
of: Argentina, Brazil, Chile, Colombia, Egypt, India, Korea, Malaysia, Mexico,
Philippines, South Africa, Turkey and Venezuela. The average is unweighted and
the country coverage is somewhat spotty prior for emerging markets to 1960
40
Figure 2.3: Average Ex-post Real Discount Rate: Advanced Economies and Emerg-
ing Markets, 1945-2009 (3-year moving averages, in percent)
1945-1980 1981-2009
-1.1 2.7
-5.3 3.8
Average Real Discount Rate
Advanced economies
Emerging markets
-20.0
-15.0
-10.0
-5.0
0.0
5.0
10.0
15.0
20.0
1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
`
Advanced economies
(3-year moving average)
Emerging Markets
(3-year moving average)
Financial Repression Era
Sources: International Financial Statistics, International Monetary Fund, various
sources listed in the Data Appendix, and authors’ calculations.
Notes:The advanced economy aggregate comprises: Australia, Belgium, Canada,
Finland, France, Germany, Greece, Ireland, Italy, Japan, New Zealand, Sweden,
the United States, and the United Kingdom. The emerging market group consists
of: Argentina, Brazil, Chile, Colombia, Egypt, India, Korea, Malaysia, Mexico,
Philippines, South Africa, Turkey and Venezuela. The average is unweighted and
the country coverage is somewhat spotty prior for emerging markets to 1960
41
Figure 2.4: Average Ex-post Real Rate on Deposits: Advanced Economies and
Emerging Markets, 1945-2009 (3-year moving averages, in percent)
1945-1980 1981-2009
-1.94 1.35
-4.01 2.85
Average Real Interest Rate on Deposits
Advanced economies
Emerging markets
-15.0
-10.0
-5.0
0.0
5.0
10.0
1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
`
Advanced economies
(3-year moving average)
Emerging Markets
(3-year moving average)
Financial Repression Era
Sources: International Financial Statistics, International Monetary Fund, various
sources listed in the Data Appendix, and authors’ calculations.
Notes:The advanced economy aggregate comprises: Australia, Belgium, Canada,
Finland, France, Germany, Greece, Ireland, Italy, Japan, New Zealand, Sweden,
the United States, and the United Kingdom. The emerging market group consists
of: Argentina, Brazil, Chile, Colombia, Egypt, India, Korea, Malaysia, Mexico,
Philippines, South Africa, Turkey and Venezuela. The average is unweighted and
the country coverage is somewhat spotty prior for emerging markets to 1960
42
Figure 2.5: Real Interest Rates Frequency Distributions: Advanced Economies,
1945-2009
Treasury bill rate
Discount rate
Deposit rate
1945-1980 1981-2009
0 46.9 10.5
1 percent 61.6 25.2
2 percent 78.6 36.2
3 percent 88.6 55.0
Real Interest rate on T-bills
Share of obsevations at or below:
-0.2
1.8
3.8
5.8
7.8
9.8
11.8
13.8
15.8
17.8
19.8
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11
1945-1980 1981-2009
1945-1980 1981-2009
0 41.9 11.6
1 percent 54.6 23.5
2 percent 69.4 37.0
3 percent 82.1 54.9
Real discount rate
Share of obsevations at or below:
-0.2
1.8
3.8
5.8
7.8
9.8
11.8
13.8
15.8
17.8
19.8
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9
1945-1980
1981-2009
1945-1980 1981-2009
0 58.8 24.6
1 percent 82.7 58.0
2 percent 94.7 85.4
3 percent 98.4 96.6
Real Interest rate on deposits
Share of obsevations at or below:
-0.2
4.8
9.8
14.8
19.8
24.8
29.8
34.8
39.8
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
1945-1980
1981-2009
43
Figure 2.6: Real Deposit Interest Rates Frequency Distributions: United Kingdom,
1880-2010
0.0
0.1
0.2
0.3
0.4
0.5
0.6
-4 -2 0 2 4 6 8
1945-1980
1981-2010
1880-1939
Sources: International Financial Statistics, International Monetary Fund, various
sources listed in the Data Appendix, and authors’ calculations.
44
Table 2.2: Selected Measures Associated with Financial Repression
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Argentina 1977-82, 1987, and 1991-2001, Ini-
tial liberalization in 1977 was re-
versed in 1982. Alfonsin govern-
ment undertook steps to deregulate
the ?nancial sector in October 1987,
some interest rates being freed at
that time. The Convertibility Plan
-March 1991-2001, subsequently re-
versed.
1977-82 and 1991-2001. Between 1976
and 1978 multiple rate system was uni-
?ed, foreign loans were permitted at
market exchange rates, and all forex
transactions were permitted up to US$
20,000 by September 1978. Controls
on in?ows and out?ows loosened over
1977-82. Liberalization measures were
reversed in 1982. Capital and exchange
controls eliminated in 1991 and rein-
stated on December 2001.
Australia 1980, Deposit rate controls lifted in
1980. Most loan rate ceilings abol-
ished in 1985. A deposit subsidy
program for savings banks started in
1986 and ended in 1987.
1983, capital and exchange controls
tightened in the late 1970’s, after the
move to indirect monetary policy in-
creased capital in?ows. Capital account
liberalized in 1983.
45
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Brazil 1976-79 and 1989 onwards, interest
rate ceilings removed in 1976, but
reimposed in 1979. Deposit rates
fully liberalized in 1989. Some loan
rates freed in 1988. Priority sec-
tors continue to borrow at subsidized
rates. Separate regulation on inter-
est rate ceilings exists for the micro-
?nance sector
1984, System of comprehensive foreign
exchange controls abolished in 1984. In
the 1980’s most controls restricted out-
?ows. In the 1990’s controls on in?ows
were strengthened and those on out-
?ows loosened and (once again) in 2010.
Canada 1967, with the revision of the Bank
Act in 1967, interest rates ceilings
were abolished. Further liberalizing
measures were adopted in 1980 (al-
lowing foreign banks entry into the
Canadian market) and 1986.
1970, mostly liberal regime.
Chile 1974 but deepens after 1984, com-
mercial bank rates liberalized in
1974. Some controls reimposed in
1982. Deposit rates fully market de-
termined since 1985. Most loan rates
are market determined since 1984.
1979, capital controls gradually eased
since 1979. Foreign portfolio and di-
rect investment is subject to a one year
minimum holding period. During the
1990s, foreign borrowing is subject to a
30% reserve requirement.
46
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Colombia 1980, most deposit rates at com-
mercial banks are market determined
since 1980; all after 1990. Loan rates
at commercial banks are market de-
termined since the mid-70’s. Re-
maining controls lifted by 1994 in all
but a few sectors. Some usury ceil-
ings remain.
1991, capital transactions liberalized in
1991. Exchange controls were also re-
duced. Large capital in?ows in the
early 90’s led to the reimposition of re-
serve requirements on foreign loans in
1993.
Egypt 1991, interest rates liberalized.
Heavy "moral suasion" on banks
remains.
1991, Decontrol and uni?cation of the
foreign exchange system. Portfolio
and direct investment controls partially
lifted in the 90’s.
47
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Finland 1982, gradual liberalization 1982-91.
Average lending rate permitted to
?uctuate within limits around the
Bank of Finland base rate or the av-
erage deposit rate in 1986. Later in
the year regulations on lending rates
abolished. In 1987, credit guide-
lines discontinued, the Bank of Fin-
land began open market operations
in bank CD’s and HELIBOR market
rates were introduced. In 1988, ?oat-
ing rates allowed on all loans.
1982.Gradual liberalization 1982-91.
Foreign banks allowed to establish sub-
sidiaries in 1982. In 1984, domestic
banks allowed to lend abroad and in-
vest in foreign securities. In 1987, re-
strictions on long-term foreign borrow-
ing on corporations lifted. In 1989, re-
maining regulations on foreign currency
loans were abolished, except for house-
holds. Short-term capital movements
liberalized in 1991. In the same year,
households were allowed to raise foreign
currency denominated loans.
France 1984, interest rates (except on sub-
sidized loans) freed in 1984. Sub-
sidized loans now available to all
banks, are subject to uniform inter-
est ceiling.
1986, in the wake of the dollar crisis
controls on in/out?ows tightened. The
extensive control system established by
1974, remains in place to early 80’s.
Some restrictions lifted in 1983-85. In-
?ows were largely liberalized over 1986-
88. Liberalization completed in 1990.
48
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Germany 1980, interest rates freely market de-
termined from the 70’s to today. In
the year indicated, further liberaliza-
tions were undertaken.
1974. Mostly liberal regime in the late
60’s, Germany experiments with con-
trols between 1970-73. Starting 1974,
controls gradually lifted, and largely
eliminated by 1981.
India 1992. Complex system of regulated
interest rates simpli?ed in 1992. In-
terest rate controls on D’s and com-
mercial paper eliminated in 1993 and
the gold market is liberalized. The
minimum lending rate on credit over
200,000 Rs eliminated in 1994. In-
terest rates on term deposits of over
two years liberalized in 1995.
1991. Regulations on portfolio and di-
rect investment ?ows eased in 1991.
The exchange rate was uni?ed in
1993/94. Out?ows remained restricted,
and controls remained on private o?-
shore borrowing.
Italy 1983. Maximum rates on deposits
and minimum rates on loans set
by Italian Banker’s Association un-
til 1974. Floor prices on government
bonds eliminated in 1992.
1985. Continuous operation of ex-
change controls in the 70’s. Fragile BoP
delays opening in early 80’s. Start-
ing in 1985, restrictions are gradually
lifted. All remaining foreign exchange
and capital controls eliminated by May
1990.
49
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Japan 1979. Interest rate deregulation
started in 1979. Gradual decontrol
of rates as money markets grow and
deepen after 85. Interest rates on
most ?xed-term deposits eliminates
by 1993. Non time deposits rates
freed in 1994. Lending rates mar-
ket determined in the 90’s (though
they started in 1979, both exter-
nal and domestic liberalizations were
very gradual and cautious).
1979. Controls on in?ows eased after
1979. Controls on out?ows eased in
the mid-80s. Forex restrictions eased in
1980. Remaining restrictions on cross
border transactions removed in 1995.
Korea 1991. Liberalizing measures adopted
in the early 80’s aimed at privati-
zation and greater managerial lee-
way to commercial banks. Signif-
icant interest rate liberalization in
four phases. Signi?cant interest rate
liberalization in four phases in the
90’s: 1991, 1993-94 and 1997. Most
interest rate deregulated by 1995, ex-
cept demand deposits and govern-
ment supported lending.
1991. Current account gradually liber-
alized between 1985-87, and article VIII
accepted in 1988. Capital account grad-
ually liberalized, starting in 1991, usu-
ally following domestic liberalization.
Restrictions on FDI and portfolio in-
vestment loosened in the early 90’s. Be-
ginning with out?ows, in?ows to secu-
rity markets allowed cautiously only in
the mid 90’s. Complete liberalization
planned for 2000.
50
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Malaysia 1978-1985 and 1987 onwards. Ini-
tially liberalized in 1978. Controls
were reimposed in the mid-80’s (es-
pecially 1985-87) and abandoned in
1991.
1987. Measures for freer in/out?ows of
funds taken in 1973. Further ease of
controls in 1987. Some capital controls
reimposed in 1994. Liberalization of the
capital account was more modest, and
followed that of the current account.
Mexico 1977, deepens after 1988.Time de-
posits with ?exible interest rates
below a ceiling permitted in 1977.
Deposit rates liberalized in 1988-
89. Loan rates have been liberalized
since 1988-89 except at development
banks.
1985. Historically exchange regime
much less restrictive than trade regime.
Further gradual easing between mid-
1985 to 1991. 1972 Law gave gov-
ernment discretion over the sectors in
which foreign direct investment was
permitted. Ambiguous restrictions on
fdi rationalized in 1989. Portfolio ?ows
were further decontrolled in 1989.
New
Zealand
1984. Interest rate ceilings removed
in 1976 and reimposed in 1981. All
interest rate controls removed in the
summer of 1984.
1984. All controls on inward and out-
ward Forex transactions removed in
1984. Controls on outward investment
lifted in 1984. Restrictions on foreign
companies’ access to domestic ?nancial
markets removed in 1984.
51
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Philippines 1981. Interest rate controls mostly
phased out between 1981-85. Some
controls reintroduced during the ?-
nancial crisis of 1981-87. Cartel-like
interest rate ?xing remains preva-
lent.
1981. Foreign exchange and invest-
ment controlled by the government in
the 70’s. After the 1983 debt crisis the
peso was ?oated but with very limited
interbank forex trading. O?-?oor trad-
ing introduced in 1992. Between 1992-
95 restrictions on all current and most
capital account transactions were elim-
inated. Outward investment limited to
$6 mill/person/year
South
Africa
1980. Interest rate controls re-
moved in 1980. South Africa Reserve
Bank relies entirely on indirect in-
struments. Primary, Secondary and
Interbank markets active and highly
developed. Stock Exchange modern
with high volume of transactions.
1983. Partially liberalized regime. Ex-
change controls on non-residents abol-
ished in 1983. Limits still apply on pur-
chases of forex for capital and current
transactions by residents. Inward in-
vestment unrestricted, outward is sub-
ject to approval if outside Common
Monetary Area. Several types of ?nan-
cial transactions subject to approval for
monitoring and prudential purposes.
52
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Sweden 1980. Gradual liberalization in the
early 80’s. Ceilings on deposit rates
abolished in 1978. In 1980, controls
on lending rates for insurance compa-
nies were removed, as well as a tax on
bank issues of certi?cate of deposits.
Ceilings on bank loan rates were re-
moved in 1985.
1980. Gradual liberalization between
1980-90. Foreigners allowed to hold
Swedish shares in 1980. Forex controls
on stock transactions relaxed in 1986-
88, and residents allowed to buy for-
eign shares in 1988-89. In 1989 foreign-
ers were allowed to buy interest bear-
ing assets and remaining forex controls
were removed. Foreign banks were al-
lowed subsidiaries in 1986, and opera-
tion through branch o?ces in 1990.
53
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Thailand 1989. Removal of ceilings on in-
terest rates begins in 1989. Ceil-
ing on all time deposits abolished
by 1990. Ceilings on saving de-
posits rates lifted in 1992. Ceilings
on ?nance companies borrowing and
lending rates abolished in 1992.
1991. Liberalized capital movements
and exchange restrictions in successive
waves between 1982-92. Article VIII
accepted and current account liberal-
ization in 1990, capital account liber-
alization starting in 1991. Aggressive
policy to attract in?ows, but out?ows
freed more gradually. Restrictions on
export of capital remain. The reserve
requirement on short-term foreign bor-
rowing in 7%. Currency controls intro-
duced in May-June 1997. These con-
trols restricted foreign access to baht in
domestic markets and from the sale of
Thai equities. Thailand relaxed limits
on foreign ownership of domestic ?nan-
cial institutions in October of 1997.
54
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Turkey 1980-82 and 1987 onwards. Liberal-
ization initiated in 1980 but reversed
by 1982. Interest rates partially
deregulated again in 1987, when
banks were allowed to ?x rates sub-
ject to ceilings determined by the
Central Bank. Ceilings were later re-
moved and deposit rates e?ectively
deregulated. Gold market liberalized
in 1993.
1989. Partial external liberalization in
the early 80’s, when restrictions on in-
?ows and out?ows are maintained ex-
cept for a limited set of agents whose
transactions are still subject to con-
trols. Restrictions on capital move-
ments ?nally lifted after August 1989.
United
Kingdom
1981. The gold market, closed in
early World War II, reopened only in
1954. The Bank of England stopped
publishing the Minimum Lending
Rate in 1981. In 1986, the govern-
ment withdrew its guidance on mort-
gage lending.
1979. July 79: all restrictions on
outward FDI abolished, and outward
portfolio investment liberalized. Oct
1979: Exchange Control Act of 1947
suspended, and all remaining barriers
to inward and outward ?ows of capital
removed.
55
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
United
States
1982. 1951-Treasury accord/debt
conversion swapped marketable
short term debt for nonmarketable
29-year bond. Regulation Q sus-
pended and S&Ls deregulated in
1982. In 1933, President Franklin
D. Roosevelt prohibits private hold-
ings of all gold coins, bullion, and
certi?cates. On December 31, 1974,
Americans are permitted to own
gold, other than just jewelry.
1974. In 1961 Americans are forbid-
den to own gold abroad as well as at
home. A broad array of controls were
abolished in 1974.
Venezuela 1991-94 and 1996 onwards. Interest
rate ceilings removed in 1991, reim-
posed in 1994, and removed again in
1996. Some interest rate ceilings ap-
ply only to institutions and individu-
als not regulated by banking author-
ities (including NGOs).
1989-94 and 1996 onwards. FDI regime
largely liberalized over 1989-90. Ex-
change controls on current and capi-
tal transactions imposed in 1994. The
system of comprehensive forex controls
was abandoned in April 1996. Controls
are reintroduced in 2003.
Sources: Reinhart and Reinhart (2011) and sources cited therein. See
also FOMC minutes, March 1-2, 1951 for US debt conversion particulars,http://www.micro?nancegateway.org/p/site/m/template.rc/1.26.9055/ on current ceilings
and related practices applied to micro?nance, and National Mining Association (2006) on
measures pertaining to gold.
Notes: Liberalization year(s) in italics.
56
Table 2.3: Incidence and Magnitude of the Liquidation of Public Debt: Selected
Countries, 1945-1980
Country Period Share of Liquidation Negative Real Interest Rate
Liquidation
Years
Years Liquidation Years
Average Min(Year)
(1) (2) (3) (4) (5) (6)
Argentina 1942-
1980
92.3 1944-1952,1954-
1980
21.4 72.3 (1976)
Australia 1945-
1980
52.8 1947-1953, 1956-
1957, 1971-1980
4.6 15.1 (1952)
Belgium
1
1945-
1974
48 1945-
1948,1951,1963,1969-
1974
4.2 9.6 (1974)
India 1949-
1980
53 1949,1951,1957,1959-
1960,1964-
1968,1970,1972-
1975,1977,1980
5.4 17.4 (1974)
Ireland 1960-
1990
65.4 1961-1965, 1967-
1977, 1979-1983
3.4 12.7 (1975)
Italy
2
1946-
1980
48.6 1946-1947,1950-
1951,1962-1964,
1970, 1972-1980
6 27.63
(1947)
3
South
Africa
1945-
1980
47.2 1947-1949, 1951-
1953, 1955,
1958, 1972-1980
3 6.8 (1975)
Sweden 1945-
1990
47.8 1947-1948,
1951-1952, 1956-
1958,1960,1962,1965-
1966,1970-
1978,1980-1981
2.6 11.9 (1951)
United
Kingdom
1945-
1980
58.3 1948-
1953,1955-1956,
1958,1962,1965,
1969, 1971-1977,
1979-1980
3.5 10.9 (1975)
United
States
1945-
1980
50 1945-1948,1950-
1951,1956-
1957,1968-
1970,1973-
1975,1977-1980
3.5 13.7 (1946)
Notes: Share of liquidation years is de?ned as the number of years during which the
real interest rate on the portfolio is negative divided by the total number of years
as noted in column (2). The real interest rate is calculated as de?ned in equation (1).
1
No data available for 1964-1968
2
The average and minimum real interest rate during liquidation years were calculated
over the period 1945-1970 to exclude war years
3
In 1944, the negative real return was 82.3 percent and in 1945 it was 46.6 percent.
57
Table 2.4: Incidence of Liquidation Years for Di?erent Real Interest Rate Thresh-
olds: Selected Countries, 1945-1980
Country Period
Share of Years with Real Interest Rate below:
0 percent 1 percent 2 percent 3 percent
Argentina 1942-1980 92.3 92.3 94.9 94.9
Australia 1945-1980 52.3 63.9 83.3 94.4
Belgium
1
1945-1974 48.0 65.4 72 80
India 1949-1980 53.0 62.5 71.9 78.1
Ireland 1960-1990 65.4 74.2 77.4 80.6
Italy
2
1946-1980 48.6 62.9 65.7 82.9
South Africa 1945-1980 47.2 61.1 77.8 97.2
Sweden 1945-1990 47.8 52.2 69.6 82.6
United Kingdom 1945-1980 58.3 72.2 86.1 97.2
United States 1945-1980 50.0 55.6 86.1 94.4
Notes: Share of liquidation years is de?ned as the number of years during which the
real interest rate on the portfolio is negative divided by the total number of years
as noted in column (2). The real interest rate is calculated as de?ned in equation (2.1).
1
No data available for 1964-1968
2
The average and minimum real interest rate during liquidation years were calculated
over the period 1945-1970 to exclude war years.
Table 2.5: Government Revenues (interest cost savings) from the "Liquidation Ef-
fect:" per year
Country Period
Benchmark Measure Alternative Measure
%
GDP
% Tax
Rev-
enues
%
GDP
% Tax
Rev-
enues
Argentina 1942-1980 3.1 38.3 3.1 39
Australia 1945-1980 3.3 12.9 n.a. n.a.
Belgium 1945-1974 2.5 18.6 3.5 23.9
India 1949-1980 1.5 27.2 1.5 27.2
Ireland 1960-1990 1.8 7.9 n.a. n.a.
Italy 1946-1980 1.6 24.5 1.6 26.5
South Africa 1945-1980 1.3 8 n.a. n.a.
Sweden 1945-1990 0.8 4.4 1.3 4.4
United Kingdom 1945-1980 3.0 18.8 3.1 19.6
United States 1945-1980 2.3 13.4 2.7 15.9
Sources: See data appendix and sources cited therein and authors’ calculations.
58
Table 2.6: Debt Liquidation through Financial Repression: Selected Countries,
1945-1955
Country
Public debt/GDP Annual average: 1946-1955
1945 1955 (actual) 1955 without "?nancial repression in?ation
repression revenue"/GDP
savings" (est.)
4
Australia 143.8 66.3 195.7 7.4 8.6
Belgium
1
112.6 63.3 130.1 5.7 3.4
Italy
2
66.9 38.1 120.2 13.3 9.6
Sweden 52 29.6 72.6 5.2 3.8
United Kingdom
3
215.6 138.2 233.8 2.6 3.9
United States 116 66.2 143.8 5.6 4.1
1
The debt-to-GDP ratio corresponds to 1946
2
Italy was in default on its external debt 1940-1946
3
The savings from ?nancial repression are a lower bound, as we use the "o?cial" consumer
price index for this period in the calculations and in?ation is estimated to have been substantially
higher than the o?cial ?gure (see for example Friedman and Schwartz, 1982).
4
The simple cumulative annual savings without compounding.
Notes: The peaks in debt/GDP were: Italy 129.0 in 1943; United Kingdom 237.9 in 1947;
United States 121.3 in 1946. An alternative interpretation of the ?nancial repression revenue
is simply as savings in interest service on the debt.
Sources: See data appendix B and sources cited therein and authors’ calculations; for
debt/GDP see Reinhart (2010) and Reinhart and Rogo? (2011b).
59
Table 2.7: In?ation Performance during Major Domestic Public Debt Reduction
Episodes: 28 Countries, 1790-2009
Country
Major Debt Reduction Episodes* Full Sample
Dates
In?ation In?ation
Average Median Average Median
Argentina 1900-
1902,1990,2006-
2007
479.8 8.2 82.5 8.6
Australia 1948,1949-1953 10.3 9.3 3.0 2.5
Belgium 1925-28, 1949 10.7 12.8 2.0 1.9
Brazil 1990-1992,1995-
1996
898.2 980.2 111.3 11.3
Canada 1948,1949-1952 7.3 5.3 3.2 2.5
Chile 1993-1997, 2004-
2007
7.7 6.1 17.7 5.5
Colombia 2008, 2009 8.5 6.3 12.6 10.8
Egypt 2008 12.0 8.6 11.7 9.9
Finland 1946-1949 34.5 24.9 10.4 3.9
France 1924, 1926-1927,
1938
11.1 12.6 6.4 2.7
Germany 1922, 1923 5555049529.6 1764.7 231460401.3 2.3
Greece 1925-1927 23.7 12.8 8.0 5.1
India 1958, 1996, 2006 7.1 6.2 6.6 6.2
Ireland 1972, 1982, 1998 9.8 8.6 5.9 3.7
Italy 1945, 1946-1948 106.7 44.3 10.6 2.6
Japan 1898, 1912-1913 7.6 6.7 3.6 2.6
Korea 1986 2.5 2.5 6.3 4.6
Malaysia 1995 8.4 8.8 6.9 5.4
Mexico 1991, 1992, 1993 18.9 20.0 13.3 5.6
New Zealand 1935-1937, 1950-
1952
4.9 5.3 4.2 2.8
Phillipines 1998, 2007-2008 7.2 7.7 7.7 6.2
South Africa 1935, 1952, 1981,
2001-2002
7.0 6.6 5.8 4.9
Sweden 1948, 1952, 1989,
2001-2003, 2009
4.7 3.2 4.4 3.2
Thailand 1989-1990 4.4 4.6 4.8 3.8
Turkey 1943, 2006-2008 23.2 9.2 25.3 9.7
UK 1836, 1846, 1854,
1936, 1940, 1948-
1950,1951-1954
4.7 3.7 2.7 1.8
US 1794-1796, 1881-
1882, 1948-1952,
1953, 1957, 1966
4.0 2.6 1.6 1.7
Venezuela 1989, 1997-1998,
2006-2007
41.6 29.5 11.4 5.8
* Consists of the 10% largest reductions for each country.
Notes: shown in italics the year of the major reduction
60
Chapter 3
Debt and In?ation during a Period of Financial Repression
3.1 Introduction
In many countries, the current ratio of government debt to GDP is at histor-
ically high levels. The 2008 ?nancial crisis has created a situation where several
countries are at risk of defaulting on their debt, and many more are struggling with
the economic and political changes needed to reduce their debt to more sustain-
able levels. Broadly speaking, the di?erent alternatives available to governments to
reduce their debt burdens are: (i) growth; (ii) ?scal adjustment (i.e., increases in
taxes and reductions in government spending); (iii) outright default or restructur-
ing; and (iv) in?ation (via in?ation surprises or a combination of ?nancial repression
and in?ation). While there is some empirical evidence on the consequences of de-
fault (Reinhart and Rogo?, 2009; Sturzenegger and Zettlelmeyer, 2006; Borensztein
and Panizza, 2006) and on the potential for ?scal austerity/restructuring (Alesina
and Ardagna, 2010; Perotti, 2011), there is limited empirical evidence on whether
in?ation can reduce the real value of debt.
Revisiting similar episodes in the past can be useful for understanding the
possible courses of action currently available to governments. In this paper, I explore
the relationship between in?ation and debt in the years after the Second World War
61
in 12 countries with very di?erent economic characteristics.
1
Many governments
had high debt levels at the end of the Second World War, which then declined over
the following decades. I construct a detailed database of government debt portfolios
spanning three or more decades for each country and analyze the role of in?ation
in reducing real debt, examine the circumstances under which this occurs, and the
implications of this for both the government and investors. The primary focus of
the paper is to understand whether this is an empirically important phenomenon,
rather than the desirability or optimality of using in?ation to erode the real value
of debt.
To understand the magnitude and nature of the e?ects, it is important to
identify the channels through which in?ation can have an e?ect on debt. The pre-
vious literature on how in?ation may reduce debt is largely theoretical, and focused
on the role of unanticipated in?ation in reducing government debt. Several papers
have looked at the time inconsistency problem present when debt is denominated in
nominal terms, where governments may be tempted to use unanticipated in?ation
to reduce the real value of their debt (Barro and Gordon, 1983; Grossman and Van
Huyck, 1984, 1985; Grossman, 1987, 1988). In this paper, I identify three channels
through which in?ation can have an e?ect on debt: unanticipated in?ation, in?a-
tion in combination with ?nancial repression, and via changes in the market value
of debt.
The term "?nancial repression" is used to refer to a situation characterized by:
1
The countries in the sample are: Argentina, Australia, Belgium, France, India, Ireland, Italy,
Japan, South Africa, Sweden, the United Kingdom, and the United States.
62
i) numerous policies and regulations which introduce frictions in ?nancial markets,
and ii) large participation of nonmarket players. The list of policies is large; some
examples of ?nancial repression are: ceilings on interest rates, directed lending, cap-
ital controls.
2
The policies that will be particularly relevant for this paper are those
which create captive investors for government debt, and hence allow the government
to issue debt at a rate below what the market would charge absent any restrictions.
Being able to issue debt at a below market interest rate represents a saving in in-
terest payments for the government. When combined with an in?ation rate above
the nominal interest rate, this leads to negative real interest rates that e?ectively
reduce government debt. This mechanism can be present even when in?ation is fully
anticipated.
There are several channels through which in?ation can a?ect debt: unantici-
pated in?ation, ?nancial repression combined with in?ation, and via changes in the
market value of debt. I develop a conceptual framework in order to understand how
those channels can be measured separately and together. While it is not possible
to directly observe in?ation expectations and quantify the e?ect of ?nancial repres-
sion on market interest rates, I show that the net e?ects of these channels must be
large whenever real interest rates are negative; that is, when real interest payments
are negative (which can be thought of as a revenue for the government). My pri-
mary empirical strategy is to focus on years where this occurs, which are labeled
liquidation years.
3
2
For a more detailed de?nition, see Appendix ?? and Reinhart and Sbrancia (2011).
3
By focusing on years with negative real interest rates, the estimates may be overestimating
the actual e?ect if for instance the "natural" interest rate in the economy was negative. On the
other hand, the estimates may be underestimating the actual e?ect whenever the "natural" interest
63
On average, real interest rates on the overall portfolios of domestic government
debt were negative in half of the years in the sample.
4
The predominant pattern that
emerges across countries is a high incidence of liquidation years in the period imme-
diately after the end of WWII, and again during the 1970s. In the United States,
50 percent of the years between 1945 and 1980 were liquidation years. Estimates
of the implicit revenues for the governments in the 12 countries, which are labeled
liquidation revenues, average between two and three percent of GDP in liquidation
years.
5
I show that the e?ects are similar whether I measure debt at face value or
at market value, which suggests that changes in the market value of debt are not
responsible for these large e?ects. I conduct several exercises to disentangle the rel-
ative contributions of the other two channels, unanticipated in?ation and ?nancial
repression. First, I estimate in?ation expectations to see whether in?ation surprises
can account for the liquidation years. Across the 12 countries, only 15 percent of
liquidation years are ones where there is an in?ation surprise. This suggests that,
for the period under consideration, ?nancial repression (combined with in?ation) is
important for explaining the high incidence of liquidation years. In a second ex-
ercise, I test whether the overall results are biased by bonds issued prior to 1945.
The thinking behind doing this is that, if the overall results are primarily due to
unanticipated in?ation, then newly- issued bonds should re?ect higher in?ation ex-
rate is positive.
4
Historically most of the domestic debt has been denominated in local currency. Notable
exceptions are Mexican Tesobonos in the 1980’s and Brazil dollar-denominated bonds a decade
after.
5
These revenues estimates do not includes the revenues from seigniorage. A comparison to
seigniorage revenues is performed in section 3.4.
64
pectations. For bonds issued after 1945, I ?nd negative real interest rates were as
common as in the full sample of bonds. This failure of markets to respond provides
further evidence that there was ?nancial repression during this period.
To put the magnitude of the liquidation revenues into perspective, I compare
them to in?ation tax revenues.
6
The liquidation e?ect revenues are consistently
larger than in?ation tax revenues at the beginning of the sample period, when debt
levels were high. In some countries, such as the United Kingdom, the liquidation
e?ect revenues are larger than in?ation tax revenues throughout the whole sample
period. This ?nding may help to explain why Reinhart and Rogo? (2009) ?nd that
the debt stock was signi?cantly larger than the money stock in many episodes of
high and hyperin?ation, as the gains from in?ating away debt may be larger than
in?ation tax revenues during these periods.
I also gain a greater understanding of the role of in?ation by examining the
cumulative e?ect of in?ation on the stock of debt under plausible alternative in?a-
tion paths. I ?nd that these cumulative e?ects are large. For example, if annual
in?ation rate had remained constant at two percent throughout the 1945-1980 pe-
riod, the debt-to-GDP ratio in 1980 would have been 40 percentage points higher
in the United States, 167 percentage points higher in the United Kingdom, and 81
percentage points higher in Australia.
An important ?nding of the paper is that in?ation need not be particularly
high in order to obtain a sizable reduction of the debt. Except in Argentina and Italy,
6
The term "in?ation tax” refers to the component of seigniorage that would be collected in
steady state.
65
median in?ation during liquidation years is below 10 percent. Average in?ation is
four percentage points higher than median in?ation rates over the 1930-2010 period.
A multivariate analysis is conducted to understand how country characteris-
tics and di?erent factors a?ect the incidence of the liquidation e?ect. I ?nd that
two variables strongly and positively correlated with liquidation years are interest
payments (as a proportion of GDP) and the size of the de?cit relative to GDP. The
signi?cance of these ?scal variables points at the presence of important links be-
tween ?scal and monetary policies. I ?nd that neither the share of short term debt
nor central bank independence have have strong relationships with the incidence of
liquidation years.
The last section of the paper focuses on the implication for investors. If Trea-
sury bonds are used as a benchmark in the valuation of other assets, then it is
important to understand how the presence of ?nancial repression a?ected ?nancial
markets more broadly.
7
I examine returns on bonds and other ?nancial assets for
more than 100 years, and ?nd that the returns on bonds were low during the 1945-
1980 period relative to other periods. Furthermore, for all of the countries in the
sample the return from investing in stocks during the 1945-1980 period was much
higher than the return from investing in government bonds. These di?erences do
not appear to be explained by stock return volatility or risk. The equity premium
is shown to be relatively high during this period. For instance, the average equity
premium - calculated as the excess return of stocks on T-Bills over rolling 30-year
7
For instance, Treasury bonds serve as a benchmark in the valuation of corporate bonds
(Crabbe and Fabbozi, 2002).
66
periods - in the US averaged 8.3 percent over 1945-1980, compared to an equity
premium of 4.4 percent in the other years between 1870 and 2010. These patterns
are consistent with the presence of ?nancial repression keeping returns on bonds
arti?cially low over the 1945-1980 period.
The paper contributes to several literatures. First, it extends the results of
Reinhart and Sbrancia (2011) who establish the importance of ?nancial repression
as a restructuring mechanism for government debt. In their paper, the authors
provide ?rst pass estimates which show that the e?ects are quantitatively important.
This paper complements the results of Reinhart and Sbrancia (2011) in three main
ways. First, I examine the di?erent channels through which in?ation can reduce
government debt and consider how to think of ?nancial repression as a restructuring
mechanism. Second, through di?erent empirical exercises, I show that ?nancial
repression in combination with in?ation is the most important channel through
which debt was reduced. Finally, the paper looks at the key features of this period
of ?nancial repression and how the returns of other assets were a?ected.
A related literature looks at the implications of in?ation on a government’s re-
ported ?scal position, and particularly how in?ation a?ects measures of ?scal de?cit
(Siegel,1979; Tanzi et al., 1987; Persson et al.,1996). The results in this paper
provide empirical support for the theoretical literature which has argued that gov-
ernment debt could not be reduced systematically by unanticipated in?ation (Calvo
and Guidotti, 1993). The results also point to ?nancial repression as an important
source of revenue, which could be important for rationalizing why countries seem
to set in?ation rates above what would be optimal if they were just maximizing
67
seigniorage revenues (Calvo and Leiderman, 1992).
The paper also adds to the literature on ?nancial repression, which has pri-
marily focused on the economic growth implications in emerging economies (Mc
Kinnon, 1973; Shaw, 1973; King and Levine, 1993; Fry, 1997). I show that ?nan-
cial repression can be important in advanced economies, and may generate revenues
for governments by reducing their debt burdens.
8
This complements work by Gio-
vannini and de Melo (1991), who measure the revenues from ?nancial repression
by comparing the interest paid by the government on its external and domestic
debt, and Aizenman and Guidotti’s (1990) theoretical work analyzing under what
conditions it may be optimal to impose capital controls.
Finally, the paper contributes to growing e?orts to use historical documents
and government reports to understand key issues in international ?nance. Long time
series on public debt were uncommon until the publication of Reinhart and Rogo?
(2008, 2009). I construct a database on the domestic debt portfolios of 12 countries
for three or more decades after the end of WWII. The database contains a detailed
description of the di?erent instruments that constitute the stock of debt in a given
year together with their coupon rates, maturity date, outstanding amount and in
some cases prices at di?erent points in time.
The next section presents the conceptual framework. Section 3.3 describes the
data and the empirical measures constructed. In Section 3.4, I present the results
for the liquidation e?ect from the perspective of the government. In Section 3.5,
8
For evidence on the presence of ?nancial repression in these countries, please see the Appendix
?? in this paper and Reinhart and Sbrancia (2011).
68
a broader analysis of ?nancial markets during the period of ?nancial repression is
provided, together with its implications for investors. Section 3.6 concludes.
3.2 Conceptual Framework
The ?rst step is to understand the channels through which in?ation can a?ect
the value of debt. When discussing how debts may be "in?ated away," researchers
usually think of a higher than expected in?ation rate eroding the real value of the
debt. However, ?nancial repression (by which governments issue debt at below mar-
ket interest rates) is an additional and important channel through which in?ation
can reduce government debt. The primary objective of this section is to separate, at
least conceptually, the contribution of unanticipated in?ation and ?nancial repres-
sion in the liquidation of government debt.
The consolidated budget constraint for the government is obtained by com-
bining the budget constraints of the ?scal and monetary authorities. This budget
constraint makes explicit the tight linkage that exists between monetary and ?scal
policy. In real terms the consolidated budget constraint is given by:
g
t
+
1 + i
t?1
1 + ?
t
b
t?1
= ?
t
+ b
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.1)
On the left side are the expenditures in a given year: government spending
(g
t
) and the real interest payments on the real stock of debt, which depends on
the nominal interest rate set in the previous period (i
t?1
), the in?ation rate in the
69
current period (?
t
), and the debt from the previous period (b
t?1
).
9
The real interest
rate paid on the stock of debt issued in the previous period is an ex post real interest
rate, since it is determined by the realized rate of in?ation. The right side contains
the sources of income: revenues (?
t
), newly issued real debt (b
t
), and the seigniorage
revenues from printing money, where h
t
is the real monetary base.
10
While in?ation
a?ects seigniorage revenues as well as other items of the budget constraint, I ignore
those e?ects as the focus on the paper is on sources of revenue which have a direct
impact on a government’s real debt payments.
11
These e?ects are compared to
seigniorage revenues in Section 3.4.
The budget constraint can be re-written in terms of the ex post real interest
rate (r
P
t
) as follows:
g
t
+ (1 + r
P
t
)b
t?1
= ?
t
+ b
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.2)
Two additional de?nitions of interest rates are required to capture the role of
unanticipated in?ation and ?nancial repression. The ?rst one is the ex ante real
interest rate. This is the interest rate that is expected to be earned in period t, as
9
Expressing the budget in terms of a one-period bond simpli?es the notation without changing
the implications that would be derived from explicitly considering a richer maturity structure.
10
Seigniorage is the change in the nominal monetary base relative to the previous period, and
divided by the current price level. It arises from two sources as shown below:
H
t
?H
t?1
P
t
= (h
t
?h
t?1
) +
?
t
1 +?
t
h
t
The ?rst component of seigniorage comes from changes in the real stock of monetary base. The
second comes from a depreciation in the outstanding stock of real balances, and is sometimes
referred to as in?ation tax. In steady state, only the second component will be positive.
11
See Persson, Persson and Svensson (1996) for a study on the overall ?scal gains from an
increase in the in?ation rate in Sweden.
70
of period t ?1. It is determined by the nominal interest rate i
t?1
and the expected
in?ation rate ?
e
t
.
The second interest rate de?nition identi?es the e?ect of ?nancial repression.
The free market interest rate (i
F
t?1
) is the interest rate that would be observed in
the absence of ?nancial frictions. If the government issues debt at a below-market
interest rate, then i
F
t?1
> i
t?1
.
The three relevant interest rates are:
1 + r
P
t
=
1 + i
t?1
1 + ?
t
Ex post real interest rate (3.3)
1 + r
A
t
=
1 + i
t?1
1 + ?
e
t
Ex ante real interest rate (3.4)
1 + r
F
t
=
1 + i
F
t?1
1 + ?
e
t
Ex ante free market real interest rate (3.5)
These terms can be incorporated into the government budget constraint. After
some algebraic manipulations, the following equation is obtained:
12
g
t
+(1+r
F
t
)b
t?1
? (1 + r
A
t
)
?
t
??
e
t
1 + ?
t
b
t?1
. ¸¸ .
Unanticipated
In?ation E?ect (A)
?
i
F
t?1
?i
t?1
1 + ?
e
t
b
t?1
. ¸¸ .
Financial
Repression E?ect (B)
= ?
t
+b
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.6)
The "unanticipated in?ation e?ect" is the di?erence between realized and ex-
pected in?ation multiplied by the real cost of previous period stock of debt, while
12
The term,
1+it?1+t?1
1+?
e
t
where
t?1
= i
F
t?1
?i
t?1
, is added and subtracted from the left-hand
side of equation (3.1).
71
the "?nancial repression e?ect" is the di?erence between the free market and actual
nominal interest rate multiplied by the real stock of debt from the previous period.
To better understand this equation, note that if there were no ?nancial frictions
that would cause i
t?1
to be di?erent from i
F
t?1
, and if actual in?ation was equal to
expected in?ation, then the last two terms on the left side would be equal to zero.
In this case, (1 + r
F
t
) would be both the ex ante and ex post real interest rate,
and there would be no savings in interest payments for the government from either
source.
Whenever the actual in?ation rate is above the expected in?ation rate, the
unanticipated in?ation e?ect will be positive and the government will save on interest
payments by the amount given by this term.
13
The opposite is true when expected
in?ation is higher than the actual in?ation rate. The ?nancial repression e?ect will
be positive and represent savings for the government when the nominal interest rate
does not re?ect the true cost of borrowing for the government, so that the actual
nominal interest rate is below the free market interest rate.
Both e?ects can be present at the same time. In this case, ?nancial repression
has an indirect e?ect on the size of the unanticipated in?ation e?ect. This indirect
e?ect comes from the fact that the ex ante real interest rate (r
A
t
) will be lower than
what it would be in the absence of frictions. In other words, for a given
?t??
e
t
1+?t
, the
savings from unanticipated in?ation will be lower in the presence of ?nancial repres-
sion. This interaction is potentially important when modeling in?ation expectations
13
When the only friction is the di?erence between actual and expected in?ation, it follows that
r
F
t
= r
A
t
72
and in?ation surprises, although it will not be separately estimated in this paper.
The Consolidated Budget Constraint at Market Value
Up to this point debt has been expressed at face value. Expressing the consol-
idated budget constraint with debt at market value allows the identi?cation of an
additional e?ect that comes from changes in the market value of debt. This will be
important for understanding returns for investors, and the response of the market in
the presence of the e?ects. In a well-functioning ?nancial system, changes in in?a-
tion expectations are going to a?ect expected returns and should be re?ected in the
price of government bonds. The (real) market value consolidated budget constraint
is given by:
g
t
+ i
t?1
B
t?1
P
t
+
P
B
t
B
t?1
P
t
= ?
t
+
P
B
t
B
t
P
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.7)
Where P
B
t
stands for the price of the debt at time t, and B
t
is the nominal
amount outstanding of debt at time t. The real market value of debt is de?ned as
ˆ
b
t
=
P
B
t
Bt
Pt
. Using these de?nitions, equation (3.7) can be re-written as:
g
t
+
i
t?1
1 + ?
t
ˆ
b
t?1
P
B
t?1
+
ˆ
b
t?1
1 + ?
t
+
P
B
t
?P
B
t?1
P
B
t?1
ˆ
b
t?1
1 + ?
t
= ?
t
+
ˆ
b
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.8)
Note that this equation is similar to equation (3.1), apart from the extra term
that captures changes in the market value of debt, where
P
B
t
?P
B
t?1
P
B
t?1
is the rate of
change in the market value of debt.
14
Following some algebraic manipulations, a
14
Changes in the market price of debt a?ect only the principal and not the interest payment
73
version of equation (3.6) is obtained:
g
t
+ i
F
t?1
b
t?1
? i
t?1
?
t
??
e
t
1 + ?
t
b
t?1
. ¸¸ .
Unanticipated
In?ation E?ect
?
i
F
t?1
?i
t?1
1 + ?
e
t
b
t?1
. ¸¸ .
Financial
Repression E?ect
?
P
B
t?1
?P
B
t
P
B
t?1
ˆ
b
t?1
1 + ?
t
. ¸¸ .
Valuation E?ect
=
?
t
+
_
ˆ
b
t
?
ˆ
b
t?1
1 + ?
t
_
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.9)
When using debt at market values it is possible to distinguish between three
e?ects: a valuation e?ect, the unanticipated in?ation e?ect, and the ?nancial repres-
sion e?ect. The unanticipated in?ation and ?nancial repression e?ects are identical
to the corresponding terms in equation (3.6), and can be interpreted in the same
way. The valuation e?ect has an easy interpretation: when the prices of the govern-
ment bonds go down there is an implicit capital gain for the government due to a
lower value of its liabilities. On the other hand, when prices are increasing there is
an implicit capital loss due to an increase in the value of the government liabilities.
This does not represent a change in the cash payments the government makes, but
a change in the market value of its debt.
Measurement Issues
Equation (3.6) identi?es the di?erent elements required to estimate the sources
of interest payment savings for the government at face value. Similarly, equation
term
_
ˆ
bt?1
P
B
t?1
= b
t?1
_
.
74
(3.9) identi?es the elements required to estimate the sources of interest payment
savings for the government at market value. A central challenge is that, in both
cases, it is not possible to directly observe in?ation expectations and free market
interest rates.
The ?rst approach to dealing with this is to focus on instances when the
net e?ect of in?ation expectations and ?nancial repression is so large that one or
both must be present. When real interest payments are negative, they constitute
a revenue rather than an expenditure for the government. In equation (3.2), when
debt is at face value, this will be the case when r
P
t
< 0. In these years, the sum of
the unanticipated in?ation e?ect and the ?nancial repression e?ect is large enough
to outweigh the free market interest payments, which is given by the second term
on the left hand side of the equation. Given that government debt is liquidated in
any year where real interest payments are negative, those years will be de?ned as
liquidation years.
15
I will refer to the e?ect of in?ation on debt as the liquidation
e?ect and the revenues for the government from this source as liquidation revenues,
which will be calculated in any single year as:
Liquidation Revenues = Negative Real Interest Rate×
Outstanding Stock of Domestic Debt
15
Reinhart and Rogo? (2009) refer to the use of in?ation to erode the value of government
debts as default via in?ation. I use the term "liquidation e?ect” to allow for the possibility that
the in?ation was not caused by a deliberate government action.
75
This de?nition provides a lower bound for the e?ect of in?ation on debt, since
the combined e?ect of unanticipated in?ation and ?nancial repression needs to out-
weigh free market interest payments.
16
It is possible however that the estimates
may be overestimating the actual e?ect. This would be the case for instance if the
"equilibrium interest rate" was negative.
17
There is also a possible source of mea-
surement error arising from the fact that ?nancial repression may have had an e?ect
on the size of the debt stock (the tax base). The relevant question is what would
have been the size of the debt stock absent ?nancial repression.
The measure captures the net e?ect of in?ation expectations and ?nancial
repression on domestic debt, when these are su?ciently large to satisfy the de?nition
(assuming a positive natural interest rate). In order to gain some understanding of
the relative contribution of each, in?ation expectations are estimated. This is done
in Section 3.4 and provides a range for the contribution of in?ation expectations.
By considering the remainder of the total e?ect as due to ?nancial repression, this
approach also provides some understanding of the importance of ?nancial repression.
Similar approaches are used when it comes measuring debt at market value. As
shown in equation (3.8), real interest payments will be negative when
i
t?1
+(P
B
t
?P
B
t?1
)
P
B
t?1
(1+?t)
<
0. The only di?erence is the additional term that re?ects changes in the market value
of government debt. This component turns out to be small, so that the face and
market value measures produce similar results. To the extent that the term re?ects
changes in expectations for future in?ation, this term provides additional informa-
16
Additional assumptions are used in Section 3.4 to understand the likely frequency of instances
where these e?ects are positive but not large enough to satisfy the de?nition adopted here.
17
Available estimates for the US and other countries show a positive and above 1% equilibrium
interest rate from 1960 (Laubach and Williams, 2003; Benati and Vitale, 2007).
76
tion about the role of in?ation expectations for the countries and periods under
study.
18
3.3 Empirical Measures and Data
This section presents the empirical measures constructed to measure the e?ect
of in?ation on government debt, and the data used to calculate those measures. De-
tailed information on the overall portfolio is necessary to obtain accurate estimates
because there is no single interest rate that is going to re?ect the ?nancing cost of
the government.
3.3.1 Empirical Measures
Two measures of nominal interest rates are constructed, one corresponding to
when debt is expressed at face value and the other when debt is at market value. The
face value measure is the Contractual Interest Rate (CIR), which is the coupon rate
at which the bond was issued. From the perspective of the government it represents
the annual interest cost of each security. The CIR is consistent with the accounting
method used by the government.
The market value measure is the Holding Period Return (HPR). The HPR is
the nominal return for a security bought at the beginning of a year and sold at the
18
This term could also re?ect changes in the general economic environment. Siegel (1979)
decomposes the changes in the real market value of debt coming from changes in the general price
level and changes in the price of government bonds. He ?nds that bond price changes are more
volatile but account for a small fraction of the total changes in the real market value of debt. An
additional point made by Siegel is that changes in the general environment will a?ect both the
liabilities and assets side of a government’s balance sheet, which may leave the wealth position of
the government largely unchanged.
77
end, where a year corresponds to the ?scal year of each country.
19
The HPR re?ects
changes in the market interest rate and is the proper way to measure the (before
tax) return on government bonds for investors.
HPR
t
=
C
t
+
_
P
B
t
?P
B
t?1
_
P
B
t?1
(3.10)
The Holding Period Return comes from two sources: the annual interest pay-
ments and any capital gains (losses) coming from increases (decreases) in the price
of the bond. Apart from minor di?erences in notation, this expression is the same
than the one obtained in the conceptual framework.
Both the CPR and HPR are nominal measures. Their real counterparts are
obtained using in?ation data from each country’s Consumer Price Index and that
corresponds to the annual in?ation rate during the ?scal year. Since the rate of
in?ation is sometimes large, the following formula is used to obtain the real return
for security i:
r
i
t
=
x
i
t
??
t
1 + ?
t
(3.11)
Where x
i
t
is the nominal return (either CIR, HPR) for security i at time t.
The last step consists of calculating the real interest rate for the whole portfolio
of government securities. This is done by calculating the weighted average of the real
interest rates of each security, where the weights represent the amount outstanding
19
For the US data as of end of December is used, given that monthly data was available, due
to changes in the ?scal year during the sample period.
78
of that security relative to the total outstanding of all securities. In the case of HPR
the total amount outstanding corresponds to the sum of the amounts outstanding
of the securities for which the measure was calculated. This is done, in order to
have the weights adding up to 1.
Portfolio Real Return (R
t
) =
Nt
t=1
r
i
t
Outstanding Amount
i
Total Outstanding Amount
(3.12)
where N
t
equals the total amount of securities at each point in time.
A liquidation year takes place whenever the real interest rate on the overall
portfolio is negative. This de?nition is a lower bound for the actual e?ect in the
case of the CIR measure since only cases where the combined e?ect of unanticipated
in?ation and ?nancial repression is large enough to make the real interest rate neg-
ative are considered. In the case of the HPR measure there exists the theoretical
possibility that it may be capturing changes in the general economic environment
and not the e?ects studied in this paper. As discussed in the previous section, em-
pirical evidence suggests this should account at most for a small fraction of the total
cases.
Finally, by saving in interest payments there is an implicit revenue for the
government in years of liquidation e?ect which can be calculated as:
Liquidation Revenues = Negative Real Interest Rate×Outstanding Stock of Domestic Debt
79
3.3.2 Data
A database was constructed from primary sources and includes 12 advanced
and developing countries. The sources are usually publications by the ?scal authority
or the central bank of each country; they are listed in Appendix ?? . The stock of
domestic debt in each year consists of the full list of securities outstanding at the
end of that year. For each security, I collected data on the outstanding amount,
maturity date, and coupon rate. Additional information was collected, depending
on its availability, including the price at which the securities were issued, and the
share of marketable and non-marketable debt.
There are 12 countries in the sample: Argentina, Australia, Belgium, France,
India, Ireland, Italy, Japan, South Africa, Sweden, United Kingdom and the United
States. The data on securities prices required to calculate the Holding Period Return
were available for Argentina, Belgium, India, Italy, Sweden, United Kingdom and
the United States. The sample period in each country generally covers 1945 to 1980.
The exceptions are India, which only has data after its independence in 1949, and
Belgium, and France, where data is unavailable for 11, and 8 years respectively.
Data is available for Ireland, Japan and Sweden that extends beyond 1980.
20
Table
3.1 lists the sample periods covered for all 12 countries.
Nine of these countries are advanced economies; the three developing countries
are Argentina, India and South Africa. These countries had quite di?erent economic
outcomes during the sample period. They also di?er in the degree to which they
20
More recent data is also available for other countries in the sample, but has not yet been
compiled.
80
were involved in World War II, and the challenges they faced at the end of it. There
are also large di?erences in their debt over time. Table 3.1 also shows the change in
the debt-to-GDP ratio from 1945 to 1980 for each country. Many of these countries
experienced large reductions in their debt over this period. The UK has the largest
decline in the debt ratio, from 210 to 41 percentage points. In the United States,
the ratio decreased from 118 to 33 percentage points. While debt ratios generally
declined over time, there are several countries in which the debt ratio remained
fairly constant or even increased. The diversity of the types and experiences of
these countries will be important to understand how general the results are, as well
as, how they vary depending on the countries’ economic characteristics..
It is important that the database covers all of a country outstanding securities,
as the composition of debt varies over time. For example, in the United States,
Treasury Bills constituted 6.5 percent of the total domestic debt in 1946 and 25.1
percent of the total in 1976, while non-marketable securities accounted for 22.7
percent in 1946, 16.7 percent in 1966 and 35.4 percent in 1976. As another example,
the share of marketable rupee loans in India went from 59 percent in 1950 to 39
percent in 1970. The composition of the debt portfolio of India and the US at
di?erent dates is shown in Table C.3 of Appendix ??.
3.4 The Liquidation E?ect from the Perspective of the Government
In this section, I begin by describing the in?ation in the 12 countries and real
interest rates on their debt portfolios over the sample period. The incidence and
81
magnitude of the liquidation e?ect is then outlined, together with a description of
the associated revenues for the government.
Several additional exercises are presented, which are aimed at understanding
what mechanisms are driving the aggregate patterns. First, in?ation expectations
are estimated to separate the relative contribution coming from unanticipated in-
?ation and ?nancial repression. Next, I focus on bonds issued within the sample
period in order to understand whether returns on new bond issues responded to
low real returns. This furthers our understanding of the role of ?nancial repression
during this period, as poor returns on new bond issues suggests a captive audience
for those bonds. The magnitude of the revenues from the liquidation e?ect are then
put in perspective by comparing it to those from the in?ation tax, and by looking
at the cumulative e?ect it had on the stock of debt. Finally, a multivariate analysis
is used to identify which country characteristics are particularly relevant to explain
the incidence of the liquidation e?ect.
3.4.1 In?ation and Real Interest Rates on Debt Portfolios
Before getting into an examination of the liquidation e?ect, it is important to
understand the in?ation rates and the distribution of real interest rates during the
period under study. The ?rst three columns of Table 3.2 shows several statistics for
the in?ation rates in the di?erent countries. The average in?ation rate in nine of the
countries is in single digits during this period, while the median in?ation is in single
digits in all of the countries except Argentina. In the United States the average
in?ation rate was 4.6 percent and the median in?ation rate was 3.2 percent, while in
82
the United Kingdom the average in?ation was 6.3 percent and median in?ation 4.2
percent. High in?ation rates in France, Italy, and Japan in the years immediately
after WWII declined signi?cantly in the 1950s.
Table 3.2 also shows the arithmetic mean, the median and the standard de-
viation for the real interest rate on the debt of each country. Columns (4) to (6)
show the results when the Contractual Interest Rate is used, and columns (7) to (9)
show the results when the Holding Period Return is used to measure real returns.
When the CIR is used, the average real interest rate is negative in all countries
except Sweden. The real interest rate is negative in seven of the eight countries
where the HPR can be calculated, with Belgium the only exception. Median real
interest rates are also low: for example, using the CIR, six of the 12 countries have
a median real return that is negative and the median real return is never larger than
1.2 percent. Using the CIR, the median real interest rate was -0.6 percent in the
United Kingdom and 0.3 percent in the United States. In the whole sample, the
fraction of the observations where the real interest rate is above three percent is,
on average, 11 percent; it is 5.6 percent in the United States and 2.8 percent in the
United Kingdom.
Two countries have particularly poor real returns. Argentina has the largest
negative mean and median real interest rate: the median real interest was -14.1
according to CIR and -11.0 according to HPR. France has the next largest average
and median negative real interest rates, which may be explained by two main facts.
First, the missing years of 1953-1958 and 1960-1963 perhaps contributes to that
pattern. In most countries, the real interest rate is above the average and median
83
values in those years. Second, even if France did not lose WWII, it had been
occupied by German forces during the War, which left the country and the economy
in a delicate situation. The average in?ation rate was 40 percent between 1945 and
1950, while the (weighted) average interest rate on the debt was 2.7 percent during
the same period.
It is clear that the results that follow are not going to be driven by a small
number of years in each country. The incidence of negative real interest rates during
the period 1945-80 was consistently high across the countries, and the distributions
of real returns were skewed towards negative values.
21
3.4.2 Incidence and Magnitude of the Liquidation E?ect
Liquidation years are years where the real interest rate on the debt portfolio
is negative. Table 3.3 shows the incidence of liquidation years for each country
according to the CIR measure. Column (1) contains the share of liquidation years
for the full sample period of each country, and the following columns show the share
of liquidation years for di?erent subperiods. The average share of liquidation years
is 56 percent for the full sample period. Excluding Argentina, which is an outlier in
the sample, it is 53 percent. Liquidation years comprise 50 percent of the years in
the United States and 58 percent of the years in the United Kingdom.
When looking at changes in the share of liquidation years across subperiods,
two patterns can be identi?ed. The most common one is a high incidence of liquida-
21
Reinhart and Sbrancia (2011) show the distributions of real deposit rates, discount rates and
T-Bill rates in a larger groups of advanced and emerging economies. In all cases the distributions
for the period 1945-1980 are to the left from those before 1945 and after 1980.
84
tion years immediately after the end of WWII, a lower incidence between 1957-1968,
and a higher incidence again in the 1970s. This is the case for eight countries: Aus-
tralia, Belgium, France, Italy, Japan, South Africa, the UK and the US. In all of
these countries, there is a higher incidence during the period 1969-1980 than during
1945-1956. These are typically countries where the debt ratios were high at the end
of WWII. The low incidence period of 1957-1968 coincides with the golden era of
Bretton Woods, while the high incidence during the 1970s occurs at a time when a
surge in the price of commodities led to an increase in the in?ation rates of most
countries.
The second pattern occurs in the case of Argentina, India, Ireland and Sweden,
who exhibit a reasonably constant incidence of liquidation years across the subpe-
riods. Argentina is the most extreme case, with almost every single year satisfying
the de?nition of liquidation year. In India, there is a lower incidence in the ?rst
subperiod and the subsequent increase in the share of liquidation years in the other
two subperiods. While the debt ratio in India did not vary much during the period
under study, this is explained by the average in?ation rate during 1949-1956 (0.2
percent) being signi?cantly lower than in the other subperiods (6.6 percent during
1957-1968 and 7.0 percent during 1969-1980). In the case of Ireland, the lower share
of liquidation years during the 1980s can be attributed to a higher average nominal
interest rate and a lower in?ation rate. In Sweden, the average in?ation rate during
1969-1980 was 8.4 percent, twice as large the average in?ation for the two previous
subperiods. This, combined with the fact that the nominal interest rate did not
increase by the same proportion, explains the higher incidence of liquidation years
85
during 1969-1980 relative to 1945-1968.
An interesting observation is that the incidence of liquidation years diminishes
in the countries for which the sample period extends beyond 1980, when most of
the controls were lifted and the era ?nancial liberalization began. In Ireland, the
incidence for 1969-1980 was 92 percent, whereas for 1981-1990 it was 30 percent.
It went from 83 percent to 10 percent over the same subperiods in Sweden, while
there is only one liquidation year in Japan after 1980. The period after 1980 is a
period of greater ?nancial liberalization, and the reduced incidence of negative real
returns seems to be associated with higher nominal interest rates and lower in?ation
rates. To illustrate, in Sweden the average nominal interest rate was 6.2 percent and
average in?ation 8.4 percent during 1969-1980, whereas the average nominal interest
rate was 10.9 percent and average in?ation 7.6 percent during 1981-1990. Even if
in?ation did not go down by much on average the nominal interest rate increased
signi?cantly.
Table 3.4 shows the magnitude of the liquidation e?ect in these liquidation
years. For any given year, the liquidation e?ect corresponds to the absolute value of
a negative real interest rate. Column (1) shows the average liquidation e?ect for the
full sample period, while the next ?ve columns show the average liquidation e?ects
for the same subperiods as in Table 3. Excluding Argentina and Japan, which have
large average liquidation e?ects of 21.4 and 13.2 percent respectively, the average
liquidation rate was 4.6 percent. The subperiods results show a similar pattern
to that observed in the previous table, namely that liquidation rates were higher
both after WWII and during the 1970’s. Column (1) of the table shows the largest
86
negative real interest rate for each country, together with the year in which it took
place. For roughly half of the countries, the minimum real interest rate took place
in the years immediately after the end of WWII, and for the other half it occurred
in the early 1970s. The minimum real interest rate in the United States was -13.7
percent in 1946, while in the UK the minimum was -10.9 percent in 1975.
3.4.3 Liquidation Revenues for the Government
The implicit revenues for the government are calculated as a percentage of
GDP and presented in Table 3.5. The way to interpret these estimates is that, if
the average liquidation revenues were two percent, it means that the government’s
de?cit was, on average, two percentage points of GDP lower during liquidation years
and led to savings on interest payments equivalent to two percent of GDP. Revenues
are determined by the magnitude of the liquidation e?ect and the size of the stock
of debt.
The average liquidation e?ect revenues during the entire sample period gener-
ally lies between 1.5 and 3.8 percent of GDP, with Sweden (0.8 percent) and Japan
(5.9 percent) the only countries outside of this range. Table 5 also shows average
revenues within subperiods. Initially high debt-to-GDP ratios means revenues were
highest in the period immediately after WWII. In all countries, the average revenue
is higher for 1945-1956 than for the full sample period. In the United States, average
revenues relative to GDP were 4.3 percent during 1945-1956 and 2.3 percent for the
full period, with both the liquidation rate and the debt ratio highest in that ?rst
subperiod. In the case of Australia, revenues during 1945-1956 (6.7 percent) were
87
twice as high as the average for the full sample (3.3 percent).Debt-to-GDP ratios
were relatively low in India, Ireland, South Africa and Sweden, which is re?ected in
lower average revenues. In India, where both the debt-to-GDP ratio and the liqui-
dation e?ect rate were constant over the full sample, the revenues were reasonably
constant across subperiods as well. Italy also exhibits relatively low revenues, the
large reduction in the debt ratio was between 1942-1947 which explains the high
value in the ?rst subperiod.
It is helpful to also compare liquidation e?ect revenues to tax revenues in those
years. Table 3.6 presents a comparison of the liquidation revenues relative to both
GDP and tax revenues for the Holding Period Return measure as well as the CIR
measure. Tax revenues do not include those from the in?ation tax. Results for
the CIR and HPR measures are similar, which suggests that the valuation e?ect
is generally small. The revenues from the liquidation e?ect can be sizable when
expressed in terms of tax revenues, as they average 20 percent of tax revenues. A
comparison to the revenues from the in?ation tax is presented later in the section.
An important ?nding of this paper is that the in?ation rate does not need to be
very high. Table 3.7 compares the median in?ation rate -due to the presence of very
high in?ation in some of the countries- during liquidation years and during 1930-
2010. The reason to start in 1930 is to focus on the period after the gold standard
when there was a change in the way monetary policy was conducted. Two thirds
of the countries in the sample have median in?ation below 10 percent both during
liquidation years and during 1930-2010. The di?erence is as low as 1.8 percentage
points for South Africa. In the US, median in?ation during liquidation years was
88
6.0 percent, in contrast to 3.0 percent during 1930-2010. In the UK median in?ation
was 8.3 percent, 4.5 percentage points higher than the historical median.
Higher in?ation rates led in some cases to a faster liquidation of the debt,
whereas in other cases this did not happen. In Argentina, in?ation rates were high
during most of the period (53.4 percent on average) but no large reduction in the
debt ratio is observed, possibly because the government kept running de?cits which
forced it to keep borrowing. In contrast, Italy and Japan had very high in?ation
(around 500 percent) at the end of WWII. This spike in in?ation in Italy reduced
the debt ratio from 118 percent in 1942 to 21 percent in 1947.
22
In the Japanese
case there is no good data for GDP during the war but di?erent estimates put the
debt ratio over 110 percent, in 1951 the debt-to-GDP ratio had reached 10 percent.
In both cases the sample was started in 1946 after the end of WWII, in the Italian
case the spike in in?ation was prior to that which is re?ected in a lower incidence
and liquidation rate as well as revenues.
23
If one were to include 1942-1945 to the
Italian sample the revenues numbers for the ?rst subperiod would be very similar
to those of Japan.
3.4.4 The Role of In?ation Expectations and Financial Repression
The measures presented so far do not distinguish between the relative contri-
butions of in?ation surprises and ?nancial repression. The goal of this section is to
estimate in?ation expectations, in orde to identify the relative contribution of each
22
Average real growth was -4 percent between 1942 and 1947
23
The year 1945 was considered as the end of the war despite the fact that Italy surrender in
September of 1943.
89
factor.
24
The empirical strategy to estimate in?ation expectations follows Fama (1975)
and Mishkin (1981), who were interested in testing for market e?ciency. An advan-
tage of this method is that it allows standard errors to be obtained. The analysis
is centered on the Fisher equation, in which the nominal interest rate at time t is
equal to the real interest rate plus the rate of in?ation that is expected between t?1
and t:
i
t
= r
A
t
+ ?
e
t
(3.13)
Where i
t
represents the nominal interest rate between t ? 1 and t, r
A
t
is the
real interest rate expected to be earned between t ?1 and t, and ?
e
t
is the in?ation
rate expected by the market between t ?1 and t.
Contrary to the ex ante real interest rate, which is determined by the expected
rate of in?ation, the ex post real interest rate is determined by the actual in?ation
between t ?1 and t:
24
There are several ways in which in?ation expectations have been estimated. One is to estimate
the anticipated component of in?ation (or money supply) with an ARMA or ARIMA process and
take the residuals as the unanticipated component (Barro, 1978). Other techniques include using
a Kalman ?lter (Burmeister, Wall, and Hamilton, 1986) or indexed bond yields to recover the
in?ation expectations (Deacon and Derry, 1994). The latter one is not a possible alternative in
this case because indexed bonds were issued after the end of the sample period. For example,
they were ?rst introduced in the United Kingdom in 1981 and in the United States in 1997.
While there are surveys on in?ation expectations, such as the Livingston Survey and the Thomson
Reuters/University of Michigan Survey in the United States, such surveys are not widely available
during this period.
90
r
P
t
= i
t
??
t
(3.14)
= r
A
t
?(?
t
??
e
t
) (3.15)
Under the assumption of rational expectations in the bond market, the forecast
error of in?ation should be uncorrelated with information available at t ?1, which
implies that:
E (?
t
??
e
t
|?
t?1
) = 0 (3.16)
Where ?
t?1
denotes the information set at t ? 1. That is, given all of the
available information at t?1, on average the di?erence between actual and expected
in?ation is equal to zero.
Using variables X
t?1
that are part of the information set ?
t?1
, the following
equation can be written:
r
A
t
= X
t?1
? + u
t
(3.17)
The error term u
t
is also determined at t ?1 and is assumed to have a mean
of zero, constant variance, and to be serially uncorrelated. Equation (3.17) can be
substituted into (3.14) to obtain:
r
P
t
= X
t?1
? + u
t
?
t
(3.18)
91
where
t
= ?
e
t
??
t
.
Contrary to (3.17), equation (3.18) can be estimated. Mishkin shows that the
OLS estimate of ? from (3.17) and (3.18) are equal in expectation. The variance-
covariance matrix derived from equation (3.18) will be larger than the one resulting
from (3.17), and the estimates of ? from equation (3.18) will be less precise.
The variables included in X
t?1
are: in?ation rate, money growth rate, real
GDP growth, and trend variables. These variables are the variables usually used in
this literature because of their high correlation with the ex ante real interest rate.
The estimates are used to obtain the following series for the ex-ante real interest
rate:
ˆ
r
A
t
= X
t?1
ˆ
?
r
P (3.19)
Combining this with equation (3.13), estimates for in?ation expectations can
be obtained:
ˆ
?
e
t
= i
t
?
ˆ
r
A
t
= i
t
?X
t?1
ˆ
?
r
P (3.20)
The exact standard errors for the estimates cannot be obtained, as the variance
of the within-sample error depends on the relative size of the variance of u
t
and the
variance of
t
. However, lower and upper bounds for the errors can be obtained.
I estimated this equation as follows. The ?rst step was to run the regression
with one explanatory variable at a time, varying the number of lags of that variable
and selecting the lag structure with the highest adjusted R
2
. Of the regressions
92
with the di?erent explanatory variables, the regression with the highest adjusted
R
2
was chosen. In the second step, a second variable was added to this regression
and the process repeated to choose the variable and lag structure that results in the
highest adjusted R
2
. This is done for the subsequent variables until adjusted R
2
is maximized.
25
This iterative process determined the regression to be estimated,
which was tested for serial correlation in the errors and for heteroscedasticity.
Table 3.8 contains the results of the regressions for each country when CIR
measures are used.
26
The explanatory variables are denoted as follows: in?ation
rate (INFL), growth of money supply (GM1), real growth in GDP (GROWTH),
and a trend variable (TREND). The estimated equation varies country by country,
re?ecting di?erent sets of variables that provided the best ?t. All of the regres-
sions include lagged values of in?ation and/or money growth; the ?rst lag of both
measures is generally negative and statistically signi?cant. Lagged values of growth
are included in several countries, although the magnitude of the coe?cients is never
particularly large. These country-speci?c speci?cations generally explain around
half of the variation in in?ation.
These regressions provide a basis on which to bound in?ation expectations.
27
When people form their expectations about future in?ation, they are likely to have
a distribution for future in?ation rather than a point estimate. For this reason, an
in?ation surprise year is said to take place whenever the actual in?ation rate is two
25
Darrat (1985) applies a similar method to calculate in?ation expectations but with an autore-
gressive process.
26
Similar results are obtained using the HPR, when it is available.
27
As a robustness exercise, in?ation expectations were also estimated by ?tting an autoregressive
process. This generates similar estimates of in?ation expectations. Also, for the United States,
these estimates were found to be similar to the Livingston survey on in?ation expectations.
93
standard deviations above the estimated expected rate of in?ation.
Table 3.9 shows the share of in?ation surprises relative to the total number
of years in the sample for each country, as well as the overlap between in?ation
surprises years and liquidation years. The results are presented using both the
lower and upper bound estimates for the standard errors. The average share of
in?ation surprises each county has is 8 percent using upper bound standard errors
and 17 percent using the lower bound. The frequency of in?ation surprise years in
liquidation years is 15 to 28 percent, depending on which estimate for the standard
errors is used. In either case, they constitute a minority of cases, which suggests
that in?ation surprises are not the primary cause of liquidation years.
For most countries, in?ation surprises are concentrated immediately after the
end of World War II and during the 1970s. It is worth noting that, after the end of
World War I, most countries experienced low in?ation rates as they tried to return
to the gold standard. This led many people to expect low in?ation rates after
World War II, and many economists thought that the biggest challenge after the
war would be slow growth and high unemployment (Studenski and Krooss, 1963).
What actually happened is that average in?ation rate in the decade after WWII
was 7 percentage points higher than the average in?ation rate in the decade after
WWI. The other period with high incidence of in?ation surprise years, in the 1970s,
corresponds to a period of oil shocks and a surge in the price of commodities.
The main conclusion from this exercise is that ?nancial repression appears to
have been more important than in?ation surprises in reducing government debt.
94
3.4.5 Are the Results Biased by Bonds Issued Before 1945?
In?ation rates were lower before WWII which, absent any restrictions, should
be re?ected in lower nominal interest rates. If a large proportion of the bonds in a
country’s portfolio were issued prior to 1945, then the results may be biased by the
returns on these bonds.
The richness of the database allows me to look at the issuance patterns over
time, to get a sense of what fraction of debt is issued before 1945 and the importance
of this issue. In most countries, many new securities were issued every period and
securities issued after 1945 quickly account for most of the debt. The exception is
Japan, where there were not many instruments issued between the end of WWII
and 1965. The maturity structure also means that most of the debt matures within
the sample period; as perpetual securities, such as the consols in the UK, account
for small fractions of the overall portfolios.
It may still be the case that the liquidation e?ects are not common for debt
issued after 1945. To check this, I analyze the performance of bonds issued after
1945 in Australia, India and Ireland, which are countries for which the necessary
data were available. The Yield to Maturity (YTM) is used to study the performance
of the bonds. The YTM is a measure of the per-period return an investor expects
to receive by holding a security until its maturity date. It is sometimes referred to
as the internal rate of return. For a given bond, the YTM at time t is given by the
following equation:
95
P
t
=
T
t=1
C
t
(1 + r)
t
+
Principal
(1 + r)
T
(3.21)
Where C
t
is the coupon payment, P
t
is the price of the security at time t, and
r is the YTM. The YTM is the rate of discount at which the present value of the
promised future cash ?ows is equal to the price of the security.
As mentioned before, the YTM should re?ect the investor’s in?ation expecta-
tions at the time the bond was issued. Investors are likely to have in mind a possible
distribution for future in?ation. While this is not directly observable, two extreme
assumptions about the real returns and in?ation expected by investors can provide
lower and upper bounds on the nominal YTM. First, to calculate a lower bound,
suppose investors expected a zero real return and so expected an in?ation rate equal
to the YTM for each period. Alternatively, suppose investors expected a real return
equal to the nominal YTM, so that they expected an in?ation rate equal to zero.
This can be thought of an upper bound, and these extremes can be used to bound
the ex ante real YTM as of time of issuance.
The ?nal step in calculating the real return consists of expressing the cash ?ows
of each security for each year until maturity in real terms. The year of issuance for
each security is used to de?ate the cash?ows. Having obtained these real cash ?ows,
and using equation (3.21), I can calculate the YTM that would deliver that stream
of cash?ows given the price at which the security was issued. This is the ex post
real YTM as of time of issuance, and provides a measure of real returns within the
sample period based on expectations at time of issuance.
96
Australian data is available for the period 1945-1968, Indian data for 1960-
1978, and Irish data for 1965-1975. The data for each security issued within these
years consists of the price and date of issuance, maturity date, and coupon rate. All
of the securities mature before 2010.
Figure 3.2 shows, at the time securities were issued, the frequency distribution
for both the ex ante real YTM and ex post real YTM. When it is assumed that
the expected in?ation rate is equal to the YTM in each period, there is a mass
concentrated at zero which is shown by the vertical solid line in each panel. It
follows then that, even without knowing what in?ation expectations investors had,
the actual distribution for the ex ante real YTM should be somewhere between the
two solid black lines. As the ?gures for the three countries show, the distribution for
the ex post real return is always to the left of the ex ante, which suggests negative
real returns were common. In the case of India and Ireland, the overlap between the
two distributions is close to zero. The share of observations with negative ex post
real YTM is 54 percent in Australia, 85 percent in Ireland, and 94 percent in India.
Summary information on the nominal and ex post real YTM appears in Table
3.10. For each country, I show the average, maximum and minimum for the nominal
YTM at issuance and the ex post real YTM as of time of issuance. On average, the
ex post real YTM was negative in all countries.
The ?ndings of this exercise show that the results presented in the previous
subsection are not biased by securities that were issued before 1945, and that neg-
ative real returns were common for securities issued within the sample period. The
results provide further evidence that the presence of ?nancial repression is an im-
97
portant factor for explaining the incidence and magnitude of the liquidation e?ect.
3.4.6 Comparison to In?ation Tax
In?ation has usually been considered as a tax on real cash balances (Friedman,
1971). It is important to compare the revenues from the liquidation e?ect to those
from the in?ation tax, in order to understand the incentives of governments to
use in?ation as a broader source of revenue. Calvo and Leiderman (1992) showed
evidence that in some situations the observed in?ation rate is above the rate at
which seigniorage is maximized. The authors argued that countries were sometimes
setting in?ation rates in the ine?cient side of the La?er Curve. However, if debt
liquidation revenues are an additional source of revenue generated by the in?ation
rate, then it may not be that the observed in?ation rates were too high but that the
tax bases were larger than previously thought. Support for this possibility comes
from Reinhart and Rogo? (2008, 2009), who noted during several episodes of high
in?ation that the debt-to-GDP ratio was much higher than the ratio of the monetary
stock to GDP.
The in?ation tax can be collected every period, which is not generally the case
for the "liquidation e?ect" tax. For this reason, instead of comparing year-on-year
revenues from each source, a comparison is conducted over subperiods. The revenues
from the in?ation tax are calculated as:
In?ation Tax =
?
t
1 + ?
t
? Money Supply
98
Where ?
t
is the in?ation rate. This is the component of seigniorage that would
be collected in steady state (see footnote 10). The monetary aggregate used is M1,
which is the most liquid monetary aggregate. This is the measure used by Rodriguez
(1994) and Easterly et al. (1995) for instance. If the monetary base was used instead
of M1, then the estimates for the in?ation tax would be lower because the monetary
base is a fraction of M1.
After calculating the revenues from each source, the sample is split into subpe-
riods of at least 10 years so that each country (except Ireland) has three subperiods
that cover the 1945-1980 period. Using the o?cial CPI, the revenues for each year
are expressed in constant terms. The base year is the ?rst year in each subperiod.
To illustrate, total revenues for the subperiod 1945-1956 are expressed in 1945 dol-
lars. The revenues are then added up and expressed in terms of the GDP of the
base year.
The results of such comparison are displayed in Table 3.11, with countries
grouped according to the main patterns that can be observed. The ?rst group of
countries has liquidation revenues that are consistently higher or similar to those
from the in?ation tax across the subperiods. These countries are Belgium, India,
Ireland, Sweden and the United Kingdom. In Belgium and the UK, the liquidation
e?ect dominated the in?ation tax both in the decade after 1945 and in the 1970s.
For India, Ireland and Sweden, the revenues from each source are broadly similar in
all subperiods.
The second group of countries have liquidation revenues which are relatively
large in the ?rst subperiod (normally 1945-1956), then have relatively larger in?ation
99
tax revenues in later periods. Countries in this group are Australia, France, Japan,
and the United States. Their liquidation e?ect revenues are higher than those from
the in?ation tax in the ?rst subperiod, when debt was very large. In the subsequent
subperiods, when the debt stock had been reduced and the incidence and magnitude
of the liquidation e?ect declined, the in?ation tax revenues were relatively higher.
In Australia and the United States the revenues from the liquidation e?ect increase
again during 1969-1980 but do not surpass those from the in?ation tax. The case of
Japan
28
highlights the point that in some episodes of high in?ation the revenue from
liquidating debt were higher than those from the in?ation tax. For the period 1946-
1956 the revenues from liquidation e?ect were 73.6 percent whereas the revenues
from in?ation tax were 15.3 percent.
In the third group of countries, which are Argentina, Italy, and South Africa,
the revenues from the in?ation tax are higher than those from the liquidation e?ect
in all subperiods. Debt in Italy had been reduced before the start of the ?rst
subperiod, whereas the money supply remained higher for a longer period of time.
In South Africa, the debt ratio was low and in?ation averaged a relatively low 3.5
percent between 1945 and 1972.
Despite the fact that the liquidation e?ect may not be collected every year,
during periods of high incidence of the liquidation e?ect the revenues obtained from
this source are usually higher than the revenues from in?ation tax. Liquidation rev-
enues were relatively high when the debt stocks were large. Looking at the revenues
from both sources together, this suggests that the total tax revenues generated by
28
Similar results are found (though not reported) for Italy for the period 1942-1947.
100
in?ation can be signi?cantly higher than previously thought. The e?ective tax base,
at least during this period of high incidence of negative real interest rates, should
be thought of as the stock of domestic debt in addition to the money stock.
3.4.7 E?ect on the Stock of Debt
To this point, I have focused on the e?ect of in?ation on the government debt
on a year-on-year basis. There is also a cumulative e?ect on the stock of debt, as by
paying lower interest on its debt the government has a lower de?cit which a?ects its
needs for new debt issuance and future interest payments. One way to capture the
magnitude of this cumulative e?ect is by assessing what each country’s debt-to-GDP
ratio would have under plausible alternative in?ation paths. As it will be shown, a
small di?erence in the average in?ation rate can have large e?ects on the stock of
debt.
To capture the e?ect on the stock of debt, the following equation of motion
for the government debt is used:
B
t
P
t
rGDP
t
= (1 + i
t?1
??
t?1
?g
t?1
)
B
t?1
P
t?1
rGDP
t?1
+
def
t?1
P
t
rGDP
t
(3.22)
Where B
t
is the stock of domestic debt, P
t
is the implicit price level, rGDP
t
is the real GDP, i
t
is the nominal interest rate, ?
t
is the net in?ation rate, g
t
is the
net real growth rate and def
t
is the primary de?cit.
A series for the primary de?cit is generated using the estimated values for the
101
nominal interest rate, together with observed values for the real stock of debt, real
growth and in?ation. I assume that the primary de?cit remains unchanged under
the di?erent in?ation scenarios. Debt ratios are then obtained for di?erent in?ation
paths.
Three alternative paths for the in?ation rate are assumed: (i) the in?ation rate
is equal to the country’s median in?ation between 1930 and 2010, (ii) the in?ation
rate is equal to the (weighted) average nominal interest rate of the corresponding
year, and (iii) the in?ation rate is equal to 2 percent.
29
These di?erent alternatives
help to understand the e?ect that in?ation can have in shaping debt dynamics.
Under the ?rst scenario, I assume that the experience in each country during the
sample period is comparable to its in?ation rate during a longer period of time.
With the second scenario, I compare the actual debt dynamics to a situation where
real interest rates are zero in every period. By assuming an in?ation rate of 2
percent, which is a common in?ation target nowadays in many countries, in the
third scenario it is possible to explore how the debt ratios in these countries would
have evolved if that in?ation target had been in place.
Table 3.12 shows the results of the analysis. Column (1) shows the debt-to-
GDP ratios at the start of the sample period (normally 1945), while Column (2)
shows the debt-to- GDP ratios at the end of the sample period (normally 1980).
Columns (3) to (5) show what the debt-to-GDP ratio would have been at the end
of the sample period under the three alternative in?ation paths.
The exercise highlights the cumulative e?ect of in?ation on the dynamics of
29
Under (i) and (iii), the in?ation rate remains constant over time.
102
the debt- to-GDP ratio. Compounding means that relatively small di?erences in
the in?ation rate has large long-term e?ects on debt-to-GDP ratio. For example,
the United States had an actual debt-to-GDP ratio in 1980 of 32 percent. Under
the scenario where the in?ation rate is equal to the median in?ation rate over 1930-
2010, which is 3.0 percent, the ratio would have been 51 percent in 1980; if the
in?ation rate had been a constant 2 percent the ratio would have been 72 percent
in 1980. The cumulative liquidation e?ect implied a reduction of 40 percentage
points, relative to a scenario where the annual in?ation rate had been 2 percent.
For the same scenario of a constant 2 percent in?ation rate, the di?erence between
the observed and estimated debt-to-GDP ratios in the United Kingdom would have
been of 167 percentage points, leaving the debt ratio at 212 percent. In Argentina,
where actual in?ation was signi?cantly higher than the alternatives proposed, the
estimated debt-to-GDP ratios are implausible high. This is also the case for Italy
and Japan if the sample is started in the ?nal years of WWII.
In the case of Japan the estimated ratios are shown until 1980 to make the
results comparable to those of other countries. For the period 1947-1980 the esti-
mated ratios are signi?cantly higher than the actual debt ratio which should not be
surprising given that median in?ation was 6 percent during those years. The ap-
proach can be also used to show what the debt dynamics in Japan would have been
if in?ation had been higher during the period 1981-2008. The in?ation rates during
this period were markedly lower than the in?ation rates in the previous forty years.
Median in?ation between 1981-2008 was 0.7 percent. If in?ation had remained at 2
percent, conducting the same exercise as before for 1980-2008, the debt ratio would
103
have been 115 percent in 2008.
30
The actual debt-to-GDP ratio in 2008 was 167
percent.
3.4.8 Understanding What A?ects the Probability of a Liquidation Year
In order to understand under which circumstances negative real interest rates
are most likely to occur, it is important to understand how di?erent factors a?ect
the return on the portfolio of government debt. For instance, does the share of
short-term debt in the portfolio increase the probability of a liquidation year? I
run a panel-data model to understand such factors. The results should be taken as
identifying conditional correlations rather than causal relationships.
The estimated model is given by:
r
CIR
it
= ?
i
+ ?
t
+X’
it
? + u
it
i = 1, ..., N
u
it
= ?
i
u
i,t?1
+
it
t = 1, ..., T
Where i is the country identi?er and t the time period, N is the total number
of countries and T is the number of time periods. The dependent variable is the
ex post real interest rate on the portfolio given by the CIR measure. The speci?-
cation includes country ?xed e?ects (?
i
) to control for constant di?erences between
countries, and time ?xed e?ects (?
t
) to control for shocks common to all countries.
30
The estimated debt ratio is 87 percent under the scenario of in?ation equal to historical
median in?ation and 70 percent under the scenario of in?ation equal to average nominal interest
rate.
104
The matrix X includes a constant and a number of explanatory variables, which
are detailed in the next paragraph. The error structure allows for correlation across
countries and for autocorrelation over time with country speci?c autocorrelation
coe?cients (?
i
). The estimator is the pooled least-square estimator with an AR(1)
for the errors.
31
The explanatory variables included in the regression are: the ratio of interest
payments to GDP (intgdp); the ratio of de?cit to GDP (defgdp); the ratio of tax
revenues to GDP (trgdp); the share of short term debt (st); and an indicator for
central bank independence (cbind). Higher interest payments could increase a gov-
ernment’s reliance on other sources of ?nancing and make in?ation more attractive.
A higher de?cit and lower tax revenues could have a similar e?ect. One would expect
that a higher share of short-term debt would be associated with a lower probability
of a liquidation year, because the government will need to re?nance its debt sooner
in the market. The idea behind Central Bank independence is that an independent
Central Bank may be less willing to ?nance government de?cits with in?ation, which
could lead to a higher real interest rate. The sources for these variables are detailed
in the Data Appendix.
Table 3.13 shows the results from estimating this equation. The variable with
the greatest explanatory power is the interest payments-to-GDP (intgdp) variable,
which is shown in Column (1). When either de?cit-to-GDP or tax revenues-to-GDP
are added, the estimated coe?cients have the expected signs in each case, but tax
31
The results are robust to other estimators such as Feasible Generalized Least Squares, or
allowing for other types of autocorrelation, such as a common AR(1) parameter.
105
revenues are not statistically signi?cant. This is the case both under di?erent error
structures and di?erent estimations. The results for both regressions are shown in
Columns (2) and (3) of Table 3.13. The negative coe?cient on the interest payments
variable means that increases in the variable are associated with lower real interest
rates. A higher value of the de?cit variable, a less negative de?cit or higher surplus,
have a positive conditional correlation with real interest payments.
32
In all of the speci?cations, the share of short-term debt is not statistically
signi?cant at the ?ve percent level. This could appear at ?rst as a surprising re-
sult given the argument that, as the government attempts to in?ate away its debt,
investors would seek protection from in?ation through shorter maturities and index-
ation (Blanchard et al., 1985; Spaventa, 1986). The result provides further evidence
that there were restrictions on the degree to which ?nancial markets could respond.
Further evidence of this lack of market response is provided in Figure 3.3, which
shows the maturity structure for debt in India, Japan, and the United States. The
dark gray area is short term debt maturing in less than a year and the lighter gray
area is debt with a maturity period longer than one year.
33
In all three countries,
the share of short-term debt represents less than 50 percent of the debt throughout
the sample period. In the United States, the share of short-term debt was highest
during the 1950s and early 1960s, and then declines to around 20 percent of to-
tal debt after 1965. In Japan, the share of short-term debt initially increases and
peaks around 1960, and then steadily declines to around 10 percent of the total debt
32
The de?cit variable is de?ned as the di?erence between revenues and spending and expressed
as a share of GDP.
33
The shares of debt remain fairly constant even when medium term debt -debt maturing in 1
to 5 years- is included.
106
by 1980. While India exhibits the largest share of short-term debt throughout the
whole sample period, after an increase in the ?rst decade the fraction of short-term
debt is reasonably stable. The fact that despite the large incidence of negative real
interest rate, governments were able to issue debts at a fairly unchanged interest
rate and still at long maturities can be interpreted as further evidence of ?nancial
repression during the period 1945-1980.
The last variable added, Column (5), is Central Bank Independence (cbind)
which is not statistically signi?cant at the ?ve percent level. One possible explana-
tion is that there are not many observations after 1980 when central banks became
more independent. An alternative explanation could be related to Alesina and Sum-
mer’s (1993) ?nding that, although a higher degree of central bank independence
is associated to a lower in?ation rate, there is no correlation between central bank
independence and real interest rates levels or variability.
A government’s ?scal position is an important determinant of the likelihood of
a liquidation year. After controlling for several factors, including the share of short-
term debt, worsening government ?nances are associated with lower real interest
rates on government debt. Montiel (2003) argued that ?nancial repression has a
?scal origin, in the sense that the inability of a government to collect revenues from
traditional sources forces it to seek other sources of revenues. In the case of ?nancial
repression, this constitutes an implicit tax on the ?nancial sector. The results of this
exercise suggest that there is a high conditional correlation between ?scal variables
and negative real interest rates, but the exercise cannot determine whether these
are causal relationships.
107
3.5 How did Investors Fare During this Period?
In this section, I compare the relative performance of bills, bonds and stocks.
In order to understand investors’ options and whether ?nancial repression policies
were key to the low returns on government bonds over the 1945-1980 period, I
compare the returns on bonds to other ?nancial assets during this period as well
as in the decades before 1945 and after 1980. The returns on bonds during the
1945-1980 were poor relative to the return on stocks, but higher than that on other
assets during the same period.
This suggests governments were able to generate demand for their securities,
and provides further evidence that the presence of ?nancial repression during the
sample period a?ected the return on government bonds. During World War II,
countries imposed controls on the economy in order to obtain the physical and
monetary resources necessary to ?ght the war.
34
The objective of governments was
the same across countries: to direct funds towards government bonds while keeping
the interest rate on their securities at low levels. Although the War o?cially ended
in 1945, it took several decades to dismantle the restrictions that had been imposed
during the war. Direct evidence on policies implemented during the sample period
can be found in Appendix sec:historyFR.
34
Oosterlinck (2010) discusses the situation in the French stock market during the war; and also
refers to the situation in other countries such as Belgium, the UK and the US.
108
3.5.1 Stocks, Bills and Bonds during the Sample Period
Table 3.14 shows the real returns using the portfolio of government debt us-
ing the Contractual Interest Rate and Holding Period Return measures, together
with real returns on bills (short-term debt) and the stock market. For each of the
variables, I present the geometric return, the arithmetic return and the standard
deviation of the real returns. The geometric (or compound) return is the most
appropriate measure to compare investments over a longer period of time (Bodie,
Kane and Marcus, 2009), but I show the results are also similar when the arithmetic
return is used. This information is provided for every country except Argentina and
Ireland, as they had no stock market data that included dividends.
The most accurate measure on government bonds is the HPR, since it incor-
porates capital gains and losses. The CIR is also presented, however, because the
HPR is not available for all of the countries. The results are generally similar, al-
though the average real returns measured by CIR are one percentage point lower
than those measured by HPR. This suggests that capital gains partially compensated
investors for low interest payments. As a reminder, the measures for the return on
government bonds are for the portfolio of government securities, and hence contain
securities of various maturities. The return for short term government securities,
bills, is presented separately as well.
The stock market returns were calculated using the most comprehensive index
available for each country. In all cases, the indexes include both dividends and
price changes. Details on the indexes used and their sources are included in ??.
109
The results for Japan and Sweden correspond to the period 1945-1980, so that the
experience in these countries can be compared to that of the other countries in the
sample.
In all countries, the average real return from the stock market is positive
whereas the real return from government bonds is negative (except for Belgium
under the HPR measure). This is true for both the short term and overall portfolio
of government securities. The fact that the geometric mean is negative implies that
someone who invested at the beginning of the period recovered less than the initial
investment at the end of the sample period. For an investor who invested $1,000 in
US government bonds in 1945, he would have received $824 in 1980 (measured in
1945 dollars). An investor in the United Kingdom would have recovered 58 percent
of their initial investment during the same sample period.
The standard deviation for the portfolio of debt is lower than the standard
deviation for stocks, which implies that there is a trade-o? between risk and return..
A simple way to look at this trade-o? is to compute the percentage of the time that
stocks outperformed bonds for holding periods of di?erent lengths. In Table 3.15, I
present these percentages for holding periods of one, two, ?ve, ten, 20 and 30 years
for both the CIR and HPR measures. The way to read the table is as follows: in
Australia, stocks perform better than bonds in 71 percent of all two year holding
periods, 96 percent of 10 year holding periods and 100 percent of the 20 and 30 year
holding periods. In all of the countries and under both measures, stocks tend to
outperform bonds regardless of the holding period. In around 60 percent of the one
year holding periods, stocks perform better than bonds. This suggests that, even
110
over short investment horizons, stocks outperform bonds and make it unlikely that
risk could explain the di?erences in returns between bonds and stocks.
The calculations reported in this table are before-tax returns. In order to be
able to calculate after-tax returns, one would need historical information on the
tax codes of the di?erent countries. Siegel (2008) calculated historical after-tax
returns for the United States, and ?nds that stocks to have a tax advantage relative
to bonds. This occurs because relatively most of the return on stocks come from
capital gains rather than dividend payments, and capital gain taxes can be deferred
until the assets are sold so that the investment grows at the before-tax rate. The
di?culty of extending this to other countries is that, while dividend and income
taxes are usually higher than capital gain taxes, government securities often receive
special tax treatment. This special treatment for government bonds makes it hard
to know which group of assets had a tax advantage in other countries.
35
3.5.2 A Longer Historical Perspective
Dimson, Marsh, and Stauton (2002) and Siegel (2008) provide evidence that
investing in the stock market is the best alternative for investors with a long time
horizon, at least in advanced economies. They do this by comparing the returns on
equity, bonds and bills in samples that span more than 100 years.
The results presented by these authors are useful to put the returns during the
1945-1980 period into historical perspective. Siegel (2008), who focuses on the US
35
For instance, in the US municipal bonds are tax-exempt. In the UK private investors in Gilts
are not liable for capital taxes.
111
economy, reports that the annual real return on long-term government bonds was 4.8
percent over 1802-1870, 3.7 percent over 1871-1925, and 2.4 percent between 1926-
2006. If the last subperiod is further split into smaller subperiods, it becomes clear
that the low real returns are driven by the negative average real returns between
1945-1980. This suggests that the 1945-1980 period was distinctive in terms of the
poor real returns on US government bonds.
Dimson et al. (2002) reported real returns on bonds, bills and stocks for all of
the countries in my sample except Argentina and India. Tables 3.16 and 3.17 show
these returns over 1900 to 2000, as well as over 1900-1939, 1940-1979 and 1980-
2000. The returns on bonds for both 1900-1939 and 1980-2000 periods were higher
than those during the 1940-1979 period in all of the countries except Belgium. In
Australia, the annual real return on bonds from 1900-2000 was 1.1 percent while the
real return between 1940-1980 was -2.8 percent. In the United States, the equivalent
numbers are 1.6 percent and -1.8 percent respectively.
Bills are short-term securities, which should quickly re?ect changes in in?ation
expectations and market interest rates in a well-functioning market. The fact that
the average real return on bills was negative in all countries during 1940-1979 o?ers
supportive evidence of the presence of ?nancial repression. Average real returns on
bills are rarely negative in earlier and later periods.. Across the ten countries for
which this information is available, the average real return on bills was 1.0 percent
between 1900-1939, -3.6 percent between 1940-1979, and 3.5 percent between 1980-
2000.
In contrast to the performance of government bonds and bills, the real return in
112
the stock market was positive in all subperiods in all countries (with the exception of
Italy between 1940-1979). There is no common pattern for stock returns across the
1900-1939 and 1940-1979 subperiods, while the return for the 1980-2000 subperiod
was markedly higher. This last observation will be important next, when the equity
premium is re-examined.
3.5.3 The Equity Premium Puzzle Revisited
The equity premium is the excess return on equities over a risk-free asset, such
as US T-bills, and is a key variable in many asset pricing models. Mehra and Prescott
published a paper titled The Equity Premium: A Puzzle?" in 1985, where they found
that the historical equity premium was far higher than the premium that could
be rationalize by a standard neoclassical model. Speci?cally, the observed equity
premium over a 100 year period was more than 6 percent, whereas the premium
predicted by the model was 1.4 percent. An additional ?nding at odds with the
empirical evidence was that the risk free rate predicted by the model was 13.2
percent, whereas the average risk free rate observed in the data was less than 1
percent.
A large literature has attempted to explain this puzzle. Mehra (2006) summa-
rizes the di?erent explanations that have been provided, dividing the explanations
into those that are risk-based and those that are not. Under the ?rst category,
Mehra groups explanations that have been successful at obtaining a risk free rate
in line with the empirical evidence but have failed to explain the equity premium.
113
Examples of this are models which propose alternative preferences (e.g., habit for-
mation), di?erent probability distributions (e.g., adding a disaster state), and be-
havioral models where agents are not fully rational. Mehra concludes that: "The
di?culty that several model classes have collectively had in explaining the equity
premium as a compensation for bearing risk leads us to conclude that perhaps it is
not a "risk premium" but rather due to other factors." Papers that explore explana-
tions that are not risk- ased use models which take into account market frictions or
allow for incomplete markets, by adding characteristics like borrowing constraints,
transaction costs and taxes.
36
The particularly low bond returns during the 1945-1980 period may account
for some of the equity premium. To examine this, I calculate the equity premium
for rolling 30-year periods over time spans in the United States and the United
Kingdom.
37
These are plotted in Figure 3.4, with the US results shown in Panel
A and the UK results in Panel B. The shaded areas correspond to the period of
?nancial repression (1945-1980). The data for the UK is available for the 1800-2010
period and for the US it is available for the 1870-2010 period, so that the ?rst 30-year
period is 1829 and 1899 respectively.
The ?rst observation is that the equity premium has varied signi?cantly over
time, and it has even been negative at times in the United Kingdom. The second,
more important observation is that the peak in the equity premium in both countries
coincides with the period of ?nancial repression. Excluding 1945-1980, the average
36
In addition to Mehra (2006), Kocherlakota (1996), Cochrane (1997) and Campbell (1999,
2001) also o?er surveys of the literature on the equity premium.
37
Each observation is the geometric mean for the equity premium over the preceding 30 years
114
equity premium for the period 1900-2010 is 2.1 percent in the UK and 4.4 percent
in the US. This compares with the average equity premia for the 1945-1980 period
of 6.7 percent in the UK and 8.3 percent in the US.
McGrattan and Prescott (2003) argue that one should take into account the
role of taxes, diversi?cation costs and regulatory constraints as determinants of
the equity premium. Although some of their results are encouraging, further work
should be done to try to measure the e?ect of ?nancial regulations, including taxes
and other restrictions, in shaping the equity premium.
The observations of this section have potentially important implications for
understanding the equity premium and the equity premium puzzle. First, a success-
ful model of the equity premium should take into account the e?ect of regulations
and institutional background. Trying to explain the average equity premium with-
out acknowledging its large variations over time may not be particularly helpful to
model prospective risk. In addition, if one believes that to some extent it is possible
a return to a more tightly regulated ?nancial system, then one should expect to
observe a larger equity premium.
3.6 Conclusion
As a result of the recent ?nancial crisis, the public debt ratios of advanced
economies have increased to levels not seen since World War II. This raises questions
as to how governments will reduce their debt burden. Studying how countries coped
with similar situations in the past can shed light on the e?ectiveness and implications
115
of the di?erent alternatives. In this paper, I provide empirical evidence that in?ation
reduced post-World War II debts. I show the conditions under which in?ation was
e?ective and the implications for both governments and investors.
In combination with ?nancial repression, in?ation was an e?ective mechanism
to reduce large debt burdens. Negative real interest rates were common and large in
magnitude across the 12 countries under study. On average, the real interest rates
on the portfolio of domestic government securities were negative in half of the obser-
vations in my sample. Implicit government revenues averaged two to three percent
of GDP. When the sample is broken into smaller subperiods, it becomes apparent
that there was a high incidence of the e?ect both in the years after the end of World
War II and during the 1970s. In France, Italy and Japan, the countries which ex-
perienced the highest in?ation rates after the war, the revenues as a proportion of
GDP averaged 12 to 17 percent between 1945 and1956.
A conceptual framework was developed to identify the channels through which
in?ation can have an e?ect in reducing government debt, and to show how to think
of ?nancial repression as a restructuring mechanism. Several exercises were con-
ducted to separate their relative contribution. The results consistently point to the
importance of ?nancial repression, rather than unanticipated in?ation, in explain-
ing the high incidence of negative real interest rates. For instance, when in?ation
expectations are measured I ?nd that in?ation surprises occur in only 15 percent of
liquidation years.
Of the various exercises focused on governments, two are worth emphasizing.
First, when compared to other sources of government revenue, liquidation revenues
116
are large. On average, liquidation revenues are equivalent to one ?fth of tax revenues,
and they are sometimes signi?cantly larger than those from the in?ation tax. Second,
the results of a multivariate analysis point at an important connection between ?scal
variables and the presence of the liquidation e?ect. While this is consistent with the
argument by Montiel (2003) about ?nancial repression having a ?scal origin, further
work is required to understand whether the links between the variables are causal.
Most of the sample period coincides with the Bretton Woods era, which was
characterized by the presence of tightly regulated capital ?ows. This appears to
have facilitated domestic policies that kept interest rates on government debt arti-
?cially low. Further evidence of the presence of ?nancial repression and its e?ect
is provided by looking at the return of government bonds for the period 1900-2000.
The real returns on bonds were signi?cantly higher both in the period before and
after 1940-1980. The presence of frictions in ?nancial markets during this period
is also apparent when looking at the average real returns on Treasury Bills, which
were negative in all of the countries in the sample between 1940 and 1980.
The abnormally low real return for bonds during the period of ?nancial re-
pression is also re?ected in the equity premium. The equity premium, calculated
over 30-year rolling periods, was the highest during the decades after World War II
that are studied in this paper. In the United States, the average equity premium
from 1870 to the present is 4.4 percent while the average equity premium during
1945-1980 is almost twice as large at 8.3 percent. These ?ndings may be relevant
for understanding the equity premium puzzle, and suggest that studies trying to as-
sess prospective risk should take into account the signi?cant e?ect that government
117
intervention can have on the return of di?erent assets.
Real growth is unlikely to play a major role in reducing debt burdens at least in
the next few years. Moreover, conditional on being able to implement ?scal austerity
measures, the evidence on their success is limited to a few countries (Perotti, 2011).
The options left are explicit defaults and restructurings, or the mechanisms studied
in this paper. In the last three decades, the world has moved towards a more
?nancially liberalized environment, which means that the magnitudes found here
may not be representative of what could happen in the future. There have been
some recent regulatory changes, however, that suggest governments may still be able
to pay low real returns on their debt in di?cult times. That, together with the scale
and breadth of how in?ation was used to liquidate government debt in the period
under study, suggests we should pay more attention to the use and implications of
this debt-reduction mechanism.
118
Figure 3.1: Debt-to-GDP, Real Growth Rate and Distribution of Real Returns
Panel A: Debt-to-GDP Ratios
Panel B: Real Growth Rate (5-year moving average)
Panel C: Distribution of Real Interest Rates
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
50
100
150
200
250
1830 1842 1854 1866 1878 1890 1902 1914 1926 1938 1950 1962 1974 1986 1998
D
o
m
e
s
t
i
c
D
e
b
t
-
t
o
-
G
D
P
(
i
n
p
e
r
c
e
n
t
s
)
UK
US
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
-10
-5
0
5
10
15
1835 1847 1859 1871 1883 1895 1907 1919 1931 1943 1955 1967 1979 1991 2003
(
R
e
a
l
G
D
P
g
r
o
w
t
h
-
5
-
y
e
a
r
m
o
v
i
n
g
a
v
e
r
a
g
e
,
i
n
p
e
r
c
e
n
t
s
)
US
UK
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
5
10
15
20
25
30
35
-9% -6% -3% 0% 3%
P
r
o
b
a
b
i
l
i
t
y
D
e
n
s
i
t
y
F
u
n
c
t
i
o
n
(
i
n
p
e
r
c
e
n
t
)
US
UK
119
Figure 3.2: Frequency Distributions of Nominal and Ex Post Real Yield to Maturity
(YTM)
AUSTRALIA
INDIA
IRELAND
-2
8
18
28
38
48
58
68
-8% -6% -4% -2% 0% 2% 4% 6% 8%
P
r
o
b
a
b
i
l
i
t
y
D
e
n
s
i
t
y
F
u
n
c
t
i
o
n
(
i
n
%
)
Nominal YTM Ex Post Real YTM
-5
0
5
10
15
20
25
30
35
40
-4.0% -2.0% 0.0% 2.0% 4.0% 6.0% 8.0%
P
r
o
b
a
b
i
l
i
t
y
D
e
n
s
i
t
y
F
u
n
c
t
i
o
n
(
i
n
%
)
Nominal YTM Ex Post Real YTM
-5.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
-9% -8% -6% -5% -3% -2% 0% 2% 3% 5% 6% 8% 9% 11% 12%
P
r
o
b
a
b
i
l
i
t
y
D
e
n
s
i
t
y
F
u
n
c
t
i
o
n
(
i
n
%
)
Nominal YTM Ex Post Real YTM
120
Figure 3.3: Maturity Structure
INDIA
JAPAN
UNITED STATES
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1949 1953 1957 1961 1965 1969 1973 1977
Short Term (1 year)
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Short Term ( 1 Year)
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1944 1948 1952 1956 1960 1964 1968 1972 1976
Short Term ( 1 year)
121
Figure 3.4: Rolling 30-year Equity Premium
UNITED KINGDOM
UNITED STATES
0
0
0
0
0
1
1
1
1
1
1
-2.5
-0.5
1.5
3.5
5.5
7.5
9.5
1829 1849 1869 1889 1909 1929 1949 1969 1989 2009
(
P
e
r
c
e
n
t
a
g
e
P
o
i
n
t
s
)
Equity
Premium
Equity
Premium -
Real Growth
1900-2010* 2.1 0.3
1945-1980 6.7 4.7
* excluding 1945-1980
0
0
0
0
0
1
1
1
1
1
1
0
2
4
6
8
10
12
1899 1909 1919 1929 1939 1949 1959 1969 1979 1989 1999 2009
(
P
e
r
c
e
n
t
a
g
e
P
o
i
n
t
s
)
Equity
Premium
Equity
Premium -
Real Growth
1900-2010* 4.4 1.0
1945-1980 8.3 4.5
* excluding 1945-1980
122
Table 3.1: Countries in the Sample and Sample Periods
Country Period Change in debt-to-GDP in percentage points (p.p.)
Argentina 1942-1980 15.2 p.p. (from 42.2 to 27.0)
Australia 1945-1980 124.6 p.p (from 143.8 to 19.3)
Belgium
a
1945-1974 74.4 p.p. (from 112.8 to 38.4)
France
b
1945-1980* 97.9 p.p. (from 104.1 to 6.2)
India 1949-1980 (1.4) p.p. (from 27.5 to 28.9)
Ireland 1960-1990 1.8 p.p. (from 54.7 to 52.8)
Italy
c
1946-1980 11.6 p.p. (from 37.2 to 25.6)
Japan 1946-2008 (82.6) p.p. (from 82.6 to 164.2)
South Africa 1945-1980 40.6 p.p. (from 73.0 to 32.4)
Sweden 1945-1990 13.9 p.p. (from 52.0 to 38.1)
United Kingdom 1945-1980 169.1 p.p. (from 210.0 to 40.9)
United States 1945-1980 85.0 p.p. (from 118.4 to 33.4)
a
Missing data for 1964-1968
b
Missing data 1953-1958 and 1960-1963
c
The debt ratio in 1942 was 118 percent of GDP.
Table 3.2: Summary of In?ation and Real Interest Rate (in percent)
Country
In?ation CIR HPR
Average Median St. Dev. Average Median St. Dev. Average Median St. Dev.
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Argentina 53.4 21.7 81.0 -19.4 -14.1 18.3 -14.1 -11.0 15.9
Australia 6.4 4.3 5.4 -1.7 -0.3 4.5 n.a. n.a. n.a.
Belgium 3.9 3.2 4.0 -0.6 0.2 4.3 1.0 2.0 5.8
France 12.2 6.0 17.7 -7.3 -2.4 12.7 n.a. n.a. n.a.
India 5.2 4.4 7.5 -0.7 -0.4 6.8 -1.3 -0.7 6.9
Ireland 8.5 6.7 6.2 -1.3 -0.8 4.3 n.a. n.a. n.a.
Italy 8.3 4.6 10.1 -1.6 0.4 6.8 -0.6 0.7 5.7
Japan 27.2 5.6 73.7 -1.5 1.2 12.3 n.a. n.a. n.a.
South Africa 5.2 4.1 3.8 -0.5 0.2 2.7 n.a. n.a. n.a.
Sweden 5.9 5.7 3.9 0.2 0.5 3.4 -0.9 -0.2 4.7
UK 6.3 4.2 5.8 -1.4 -0.6 3.7 -1.2 -0.1 7.3
US 4.6 3.2 4.4 -0.8 0.3 4.1 -0.4 0.0 4.2
Average 12.3 6.1 18.6 -3.1 -1.3 7.0 -2.5 -1.3 7.2
Notes: See Table 3.1 for sample period.
a
Missing data for 1964-1968
b
Missing data 1953-1958 and 1960-1963
123
Table 3.3: Incidence of Liquidation Years
Country Period
Share of Liquidation Years for di?erent subperiod
Full Sample 1945-1956 1957-1968 1969-1980 1981-1993 1994-2008
(1) (2) (3) (4) (5) (6)
Argentina 1942-1980 92 80 100 100 - -
Australia 1945-1980 53 67 8 83 -
Belgium
a
1945-1974 48 42 14 100 - -
France
b
1945-1980 77 75 50 92 - -
India 1949-1980 53 25 67 58 - -
Ireland 1960-1990 68 - 78 92 30 -
Italy 1946-1980 49 36 25 83 - -
Japan 1946-2008 35
c
64 42 75 0 7
South Africa 1945-1980 47 58 8 75 - -
Sweden 1945-1990 48 42 50 83 10 -
United Kingdom 1945-1980 58 67 25 83 - -
United States 1945-1980 50 58 17 75 - -
Average 56 56 40 83 13 7
Notes: Share of liquidation years is calculated as the ratio between number of years in which the real return
was negative and the total number of years in the corresponding subperiod. The measure of real interest rate
is the Contractual Interest Rate (CIR).
- Subperiods not included in the sample of the country
a
Missing data for 1964-1968
b
Missing data 1953-1958 and 1960-1963
c
The share of liquidation years for the period 1946-1980 is 60 percent
Table 3.4: Liquidation Rate
Country
Average Liquidation E?ect Minimum
Full Sample 1945-1956 1957-1968 1969-1980 1981-1993 1994-2008 (Year)
(1) (2) (3) (4) (5) (6) (7)
Argentina 21.4 13.0 17.0 34.2 - - 72.3 (1976)
Australia 4.6 6.8 0.9 3.2 - - 15.1 (1952)
Belgium 4.2 6.0 1.0 3.1 - - 9.6 (1974)
France 9.8 26.8 1.4 2.9 - - 41.2 (1946)
India 5.4 6.0 4.8 5.8 - - 17.4 (1974)
Ireland 3.4 - 1.0 4.4 5.7 - 12.7 (1975)
Italy 6.0 13.3 2.1 4.2 - - 27.6 (1947)
Japan 13.2 35.0 2.2 3.7 * 0.1 78.5 (1946)
South Africa 3.0 2.7 0.5 3.3 - - 6.8 (1975)
Sweden 2.6 4.7 1.2 2.4 2.9 - 11.9 (1951)
United Kingdom 3.5 2.7 0.5 5.1 - - 10.9 (1975)
United States 3.5 4.1 0.1 3.7 - - 13.7 (1946)
Average 6.7 11.0 2.7 6.3 4.3 0.1
Notes: The liquidation rate is absolute value of the real interest rate during liquidation years.
See table 3.3 for sample period. The measure of real interest rate is the Contractual Interest Rate.
- Subperiods not included in the sample of the country
* Subperiods which are part of the sample but have no LE years
124
Table 3.5: Liquidation Revenues
Country
Average Liquidation E?ect Revenues as percentage of GDP
Full Sample 1945-1956 1957-1968 1969-1980 1981-1993 1994-2008
(1) (2) (3) (4) (5) (6)
Argentina 3.1 3.8 3.0 2.5 -
Australia 3.3 6.7 0.6 0.8 - -
Belgium 2.5 4.9 0.6 1.3 - -
France 3.8 12.4 0.2 0.2 - -
India 1.5 1.6 1.4 1.5 - -
Ireland 1.8 - 0.5 2.2 2.9 -
Italy 1.6 13.3 0.4 1.1 - -
Japan 5.9 17.8 0.1 0.6 * 0.1
South Africa 1.3 1.5 0.2 1.2 - -
Sweden 0.8 1.6 0.3 0.6 1.3 -
United Kingdom 3.0 4.4 0.6 2.6 - -
United States 2.3 4.3 0.0 1.2 - -
Notes:See Table 3.3 for sample period.The measure of real interest rate is the Contractual Interest Rate.
- Subperiods not included in the sample of the country
* Subperiods which are part of the sample but have no LE years
Table 3.6: Comparison Liquidation Revenues as percentage of:
Country
GDP Tax Revenues
CIR HPR CIR HPR
Average Median Average Median Average Median Average Median
Argentina 3.1 2.4 3.1 2.2 38.3 27.3 39.0 28.6
Australia 3.3 1.1 n.a. n.a. 12.9 4.4 n.a. n.a.
Belgium 2.5 1.2 3.5 3.5 18.6 10.3 23.9 18.9
France 3.8 0.2 35.3 1.1
India 1.5 1.6 1.5 1.8 27.2 27.6 27.2 28.4
Ireland 1.8 1.1 n.a. n.a. 7.9 6.1 n.a. n.a.
Italy 1.6 0.8 1.6 0.7 24.6 5.4 26.5 4.9
Japan 5.9 0.4 n.a. n.a. 37.9 3.0 n.a. n.a.
South Africa 1.3 1.4 n.a. n.a. 8.0 6.8 n.a. n.a.
Sweden 0.8 0.4 1.3 0.9 4.4 2.1 4.4 2.1
United Kingdom 3.0 1.9 3.1 1.8 18.8 11.0 19.6 10.0
United States 2.3 0.7 2.7 1.3 13.4 3.9 15.9 7.1
Notes:See Table 3.3 for sample period.
125
Table 3.7: Comparison of Median In?ation between LE Years and 1930-2010
Contractual Interest Rate
Liquidation Years 1930-2009
Argentina 21.1 15.3
Australia 9.0 3.8
Belgium 7.6 2.6
India 9.4 5.6
Ireland 10.0 3.2
Italy 11.3 4.5
Japan 8.3 3.0
South Africa 7.5 5.8
Sweden 7.0 4.0
United Kingdom 8.3 3.8
United States 6.0 3.0
Holding Period Return
Liquidation Years 1930-2009
Argentina 21.4 15.3
Belgium 8.6 2.6
India 9.3 5.6
Italy 12.2 4.5
Sweden 7.0 4.0
United Kingdom 11.9 3.8
United States 5.6 3.0
126
T
a
b
l
e
3
.
8
:
R
e
g
r
e
s
s
i
o
n
s
f
o
r
I
n
?
a
t
i
o
n
E
x
p
e
c
t
a
t
i
o
n
s
A
r
g
e
n
t
i
n
a
A
u
s
t
r
a
l
i
a
B
e
l
g
i
u
m
F
r
a
n
c
e
I
n
d
i
a
I
r
e
l
a
n
d
I
t
a
l
y
J
a
p
a
n
S
o
u
t
h
A
f
r
i
c
a
S
w
e
d
e
n
U
n
i
t
e
d
K
i
n
g
d
o
m
U
n
i
t
e
d
S
t
a
t
e
s
C
o
n
s
t
a
n
t
0
.
9
2
0
.
0
2
-
0
.
3
4
0
.
0
0
-
0
.
0
5
0
.
3
6
0
.
0
2
-
0
.
5
2
-
0
.
0
3
0
.
0
5
-
0
.
0
2
0
.
0
2
(
3
.
1
5
)
(
2
.
7
)
-
4
.
7
6
-
0
.
0
5
(
-
1
.
1
8
)
(
2
.
9
4
)
(
1
.
4
1
)
(
-
4
.
3
9
)
(
-
0
.
9
2
)
(
2
.
0
8
)
(
-
0
.
5
1
)
(
2
.
8
6
)
I
n
?
(
T
-
1
)
-
0
.
0
8
-
0
.
6
5
0
.
3
4
-
0
.
5
2
-
0
.
3
2
-
0
.
1
1
-
0
.
1
4
-
0
.
3
2
-
0
.
3
0
-
0
.
5
2
-
0
.
2
4
(
-
1
.
9
6
)
(
-
5
.
3
6
)
1
.
5
8
-
3
.
5
3
(
-
2
.
1
6
)
(
-
6
.
0
4
)
(
-
4
.
7
6
)
(
-
1
.
8
6
)
(
-
2
.
2
)
(
-
3
.
7
3
)
(
-
2
.
1
6
)
I
n
?
(
T
-
2
)
0
.
0
4
0
.
3
7
0
.
6
0
0
.
1
2
-
0
.
0
2
0
.
0
2
-
0
.
0
2
-
0
.
0
9
0
.
1
6
0
.
1
5
0
.
2
3
(
1
.
1
)
(
3
.
0
7
)
3
.
4
2
0
.
5
8
(
-
0
.
0
2
)
(
0
.
7
8
)
(
-
0
.
7
3
)
(
-
0
.
5
3
)
(
1
.
1
4
)
(
1
.
1
2
)
(
2
.
0
1
)
I
n
?
(
T
-
3
)
0
.
1
6
-
0
.
1
2
0
.
2
6
-
0
.
0
8
0
.
0
7
0
.
9
6
-
0
.
5
8
(
1
.
7
8
)
(
-
4
.
0
5
)
(
2
.
7
1
)
I
n
?
(
T
-
4
)
0
.
4
8
0
.
0
7
0
.
0
5
2
.
8
3
0
.
4
6
(
3
.
3
5
)
G
M
1
(
T
-
1
)
-
0
.
2
5
-
0
.
3
5
-
0
.
0
8
-
0
.
3
9
(
-
4
.
1
3
)
(
-
2
.
3
3
)
(
-
4
.
0
5
)
(
-
2
.
4
1
)
G
M
1
(
T
-
2
)
-
0
.
0
1
-
0
.
1
5
0
.
0
5
0
.
1
3
(
-
0
.
1
1
)
(
-
0
.
9
)
(
3
.
3
5
)
(
0
.
6
7
)
G
M
1
(
T
-
3
)
-
0
.
5
0
-
0
.
0
8
0
.
2
1
(
-
2
.
4
)
(
-
0
.
7
6
)
(
1
.
2
5
)
G
M
1
(
T
-
4
)
-
0
.
3
1
(
-
3
.
4
7
)
G
r
o
w
t
h
(
T
-
1
)
0
.
0
0
0
.
0
0
0
.
0
0
(
-
3
.
2
6
)
(
1
.
5
1
)
(
-
2
.
0
9
)
G
r
o
w
t
h
(
T
-
2
)
0
.
0
0
(
-
1
.
1
4
)
T
r
e
n
d
-
0
.
0
3
0
.
0
3
0
.
0
0
0
.
0
4
0
.
0
0
0
.
0
0
0
.
0
1
(
-
1
.
0
3
)
5
.
1
6
(
3
.
2
4
)
(
3
.
9
1
)
(
1
.
7
1
)
(
0
.
9
7
)
(
0
.
8
7
)
T
r
e
n
d
2
0
.
0
0
0
.
0
0
(
2
.
0
7
)
-
5
.
3
3
T
r
e
n
d
3
0
.
0
0
0
.
0
0
0
.
0
0
(
-
1
.
5
4
)
(
3
.
0
7
)
(
1
.
3
5
)
R
2
_
b
a
r
0
.
5
2
0
.
6
3
0
.
6
1
0
.
4
7
0
.
2
7
0
.
5
5
0
.
8
9
0
.
6
2
0
.
4
7
0
.
2
7
0
.
4
5
0
.
5
3
D
u
r
b
i
n
W
a
t
s
o
n
1
.
9
3
1
.
8
6
1
.
8
6
1
.
4
4
2
.
0
4
1
.
8
0
1
.
5
3
1
.
1
9
1
.
9
7
1
.
9
5
1
.
8
7
2
.
4
7
F
S
i
g
.
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
2
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
O
b
s
.
3
8
3
6
2
4
2
6
3
2
3
1
3
6
6
3
3
6
4
6
4
6
3
6
127
Table 3.9: In?ation Surprises and Liquidation Years
Country
Share of Overlap Liquidation Years
In?ation Suprises CIR HPR
Lower Upper Lower Upper Lower Upper
Argentina 42 17 43 17 32 4
Australia 25 8 42 16
Belgium 8 4 18 9 13 13
France 15 12 20 15
India 9 6 18 6 18 6
Ireland 6 3 10 5
Italy 22 14 39 28 39 28
Japan 14 8 39 22
South Africa 19 8 35 18
Sweden 9 4 18 9 18 9
UK 17 8 29 14 29 14
US 11 9 22 17 17 17
Average 17 8 28 15 24 13
Notes: In?ation suprises are de?ned as years where the realized
in?ation rate is two standard deviations above the estimated expected
in?ation rate. The actual standard errors cannot be obtained but
a lower and upper bound for them.
Table 3.10: Comparison between Ex Ante and Ex Post Yield to Maturity for secu-
rities issued within sample period (in percent)
Country
Nominal YTM at issuance Ex-post real YTM as of time of issuance
Number of bonds
Average Maximum Minimum Average Maximum Minimum
Australia
a
4.2 5.4 2.0 -0.6 4.7 -12.3 50
India
b
5.3 6.5 3.6 -2.3 4.6 -5.0 98
Ireland
c
8.4 14.6 5.4 -2.2 3.5 -10.2 26
a
Bonds issued between 1945-1968
b
Bonds issued between 1960-1978
c
Bonds issued between 1965-1975
128
Table 3.11: Comparison Liquidation E?ect Revenues and In?ation Tax (as %GDP)
Country Subperiod In?ation Tax LE Revenues CIR
G
r
o
u
p
1
Belgium 1945-1954 13.8 23.3
1955-1964 4.1 3
1965-1974 1.1 7.3
India 1949-1959 8.4 6.3
1960-1969 11.5 10.9
1970-1980 15.5 13.9
Ireland 1960-1969 13.8 7.6
1970-1979 20.4 25.5
1980-1990 10.6 11.2
Sweden 1945-1956 6.5 10.5
1957-1968 4.7 2.1
1969-1979 6.8 5.2
1980-1990 5.0 2.8
United Kingdom 1945-1956 6.4 35.7
1957-1968 3.8 1.7
1969-1980 9.4 28.8
G
r
o
u
p
2
Australia 1945-1956 43.6 65.3
1957-1968 7.8 0.5
1969-1980 22.6 14.8
France 1945-1956 74.1 97.5
1957-1968 16.1 0.8
1969-1980 23.7 2.1
Japan 1946-1956 15.3 73.6
1957-1968 28.0 1.1
1969-1980 37.8 6.3
1981-1993 9.1 0.0
1994-2008 1.9 0.1
United States 1945-1956 17.6 26.8
1957-1968 8.5 0.1
1969-1980 21.0 17.0
G
r
o
u
p
3
Argentina 1945-1956 61.4 53.5
1957-1968 42.4 37.8
1969-1980 50.3 40.2
Italy 1946-1956 29.7 19.2
1957-1968 15.8 1.8
1969-1980 117.1 23.4
South Africa 1945-1956 23 12
1957-1968 8 0
1969-1980 24 18
129
Table 3.12: Stock of Debt under di?erent paths for the in?ation rate
Country Period
Debt/GDP Debt/GDP (End)
Initial Actual Median
In?ation
In?=Nominal
Int Rate
In?=2%
(1) (2) (3) (4) (5)
Argentina 1942-1980 42.0 10.4 181.4 2914.2 8080.2
Australia 1945-1980 145.4 18.1 50.3 40.5 98.9
Belgium
a
1945-1974* 112.8 38.4 54.5 45.5 64.2
France
b
1945-1980* 0.0 0.0 0.0 0.0 0.0
India 1949-1980 26.2 25.4 24.3 39.0 71.1
Ireland 1960-1990 61.7 48.4 242.3 53.1 346.4
Italy 1946-1980 67.0 26.2 97.0 58.9 220.4
Japan 1947-1980 27.5 39.1 257.4 137.6 355.9
South Africa 1945-1980 56.6 32.1 28.5 43.3 109.9
Sweden 1945-1990 52.6 39.2 91.0 37.6 224.8
United Kingdom 1945-1980 236.2 44.4 111.6 82.7 211.7
United States 1945-1980 116.0 31.6 50.7 33.8 71.7
Notes: Initial D/GDP refers to the debt ratio in the ?rst year of the sample period for each country, and the
D/GDP (end) is the debt ratio in the last year of the sample.
a
Missing data for 1964-1968
b
Missing data 1953-1958 and 1960-1963
Table 3.13: Panel Estimation Results- Dependent Variable: CIR
(1) (2) (3) (4) (5)
intgdp -0.230*** -0.218*** -0.221*** -0.221*** -0.220***
(-16.07) (-18.81) (-12.84) (-18.70) (-18.74)
defgdp 0.681*** 0.676*** 0.670***
(8.67) (9.59) (9.53)
trgdp 0.0606
(0.62)
stlt 0.0722 0.0750
(1.57) (1.61)
cbind -0.0876
(-1.85)
?nrep
Constant -0.0286 0.0563 -0.0207 0.154*** 0.179***
(-0.72) (1.35) (-0.39) (3.40) (3.88)
Country FE Yes Yes Yes Yes Yes
Time FE Yes Yes Yes Yes Yes
N 446 445 443 434 432
Notes: t-statistics in parenthesis
* p
Financial repression is any of the measures that governments employ to channel funds to themselves, that, in a deregulated market, would go elsewhere. Financial repression can be particularly effective at liquidating debt.
ABSTRACT
Title of dissertation: A STUDY OF DEBT AND INFLATION
DURING A PERIOD OF FINANCIAL REPRESSION
María Belén Sbrancia, Doctor of Philosophy, 2012
Dissertation directed by: Carmen M. Reinhart
Peterson Institute for International Economics
Carlos A. Végh
Department of Economics
The large accumulation of public debt which took place as a consequence
of the recent ?nancial crisis, poses the question of how governments are going to
reduce their debts. In this dissertation I look at prior episodes where most advanced
economies were highly indebted to understand the possible courses of action that
may be available to governments in the future.
Chapter 1, written with Carmen M. Reinhart, documents the role played by ?-
nancial repression and in?ation in reducing government debt after the end of World
War II. This mechanism represents a more subtle and gradual form of debt re-
structuring. Evidence on the evolution and distribution of several measures of real
interest rates show that during the period 1945-1980 interest rates were markedly
more negative than what they were both in the period before World War II and
after 1980. When looking at the real interest rate on the portfolio of domestic debt,
real interest rates were negative half of the time for the 12 countries in the sample.
A negative real interest rate on the portfolio of domestic debt represents a saving
on interest payments for the government. In this case, savings on interest payments
averaged 2 to 3 percent of GDP. The chapter also documents the series of defaults,
restructurings and conversions that took place after the end World War I and the
Great Depression.
Chapter 2 builds on the results of the ?rst chapter. It provides a conceptual
framework to understand the di?erent ways through which in?ation can a?ect the
value of government debt, and also how to think of ?nancial repression as a restruc-
turing mechanism. Several econometric exercises are performed to disentangle the
relative contribution of ?nancial repression and unanticipated in?ation. The results
point to ?nancial repression combined with in?ation (i.e., a nonmarket interest rate
combined with in?ation) as a relatively more important factor explaining the debt
reduction in the post-World War II period. Finally, the question of how investors
were a?ected during this period is explored. It is shown that the equity premium
almost doubled during this period. This result suggests that the presence of ?nan-
cial repression, during the period 1945-1980, could help explain to some extent the
equity premium puzzle.
A STUDY OF DEBT AND INFLATION DURING A PERIOD OF
FINANCIAL REPRESSION
by
María Belén Sbrancia
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial ful?llment
of the requirements for the degree of
Doctor of Philosophy
2012
Advisory Committee:
Dr. Carmen M. Reinhart, Co-Chair
Professor Carlos Végh, Co-Chair
Professor Anton Korinek
Professor Pablo D’Erasmo
Professor Phillip L. Swagel
c Copyright by
María Belén Sbrancia
2012
Dedication
A mis abuelos Olga, Elidia, Atilio y Santín porque gracias a sus esfuerzos y
dedicación fue posible que hoy yo estuviera aquí. A mis papás, Gladys y Guido, por
todo el amor y apoyo que me han brindado en estos 27 años. A Tim, el amor de mi
vida, por ser mi compañero.
ii
Acknowledgments
I would like to thank my main advisors, Carmen Reinhart and Carlos Végh,
for their guidance, support and criticism along the way which have helped me com-
plete this dissertation. They are both outstanding academics and excellent people
who made a big di?erence, both professionally and personally, in my experience at
Maryland. I owe to Carmen my passion for empirical work, she has been a devoted
mentor who has inspired me in so many di?erent ways. She supported early on
when I started to work on my dissertation and encouraged me to pursue this topic.
I feel extremely lucky to have had the opportunity to learn so much from her. The
?rst chapter of this dissertation is written jointly with her. I don’t think there many
advisors like Carlos who is always available to talk, and who excels at the art of
being rigorous and supportive at the same time. He has taught me the importance
of keeping things simple. As he would often say "you don’t need an AK47 to kill a
?y."
I am also grateful to Anton Korinek for his comments, help and sometimes
hard questions which have helped me to judge my work from a di?erent angle and
become a better researcher. I would also like to thank all the other professors at
Maryland who have provided comments and suggestions along the way.
I owe immense gratitude to Adrian Guissarri who introduced me to the fasci-
nating ?eld of Economics, and encouraged me to pursue a doctoral degree.
I would like to extend my thanks to the administrative sta? at the Economics
Department, specially Vickie Fletcher, for their dedication, time and help during this
iii
time. I am also grateful to the sta? at the Library of the Congress who provided
me with invaluable help while I put together the database for this dissertation.
Thanks to my parents. I wouldn’t be here if it wasn’t for them. Thanks for all
your support, love, and for helping me pursue my dreams. I will always be grateful to
my dad for teaching me to persevere, work hard, and be an honest person. And very
special thanks to my mom, one of the strongest persons that I know, for listening to
me, for teaching me to be optimistic, and for always believing in me. To Nazareno
for constantly reminding me that "it’s not such a big deal" and to relax. To María
José for distracting me with fashion and music talks, and cheering me during these
years.
My deepest thanks go to my husband and best friend, Tim, for his uncon-
ditional love and support. This journey would have been a lot harder without his
jokes and company.
And last but not least thanks to all my good friends (Armando, Carla, Car-
olina, Cesar, Fernando, Florencia, Francisca, Giorgo, Jose?na, Laura, Lisa, Mariana,
Nieves, Pablo, Paula, Rong, Tiziana, Ximena) with whom I have shared so many
special and fun moments throughout the years.
iv
Table of Contents
List of Tables vii
List of Figures viii
1 Introduction 1
2 The Liquidation of Government Debt (with Carmen M. Reinhart) 5
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Default, Restructuring and Conversions: Highlights from 1920s-1950s 12
2.2.1 Global debt surges and their resolution . . . . . . . . . . . . . 13
2.2.2 Default, restructurings and forcible conversions in the 1930s . 14
2.3 Financial Repression: policies and evidence from real interest rates . . 19
2.3.1 Selected ?nancial regulation measures during the "era of ?-
nancial repression" . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.2 Real Interest Rates . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 The Liquidation of Government Debt: Conceptual and Data Issues . 23
2.4.1 Benchmark basic estimates of the "liquidation e?ect" . . . . . 25
2.4.2 An alternative measure of the liquidation e?ect based on total
returns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.4.3 The role of in?ation and currency depreciation . . . . . . . . 28
2.5 The Liquidation of Government Debt: Empirical Estimates . . . . . . 29
2.5.1 Incidence and magnitude of the "liquidation tax" . . . . . . . 30
2.5.2 Estimates of the Liquidation E?ect . . . . . . . . . . . . . . . 32
2.6 In?ation and Debt Reduction . . . . . . . . . . . . . . . . . . . . . . 34
2.7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3 Debt and In?ation during a Period of Financial Repression 61
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.2 Conceptual Framework . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.3 Empirical Measures and Data . . . . . . . . . . . . . . . . . . . . . . 77
3.3.1 Empirical Measures . . . . . . . . . . . . . . . . . . . . . . . . 77
3.3.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.4 The Liquidation E?ect from the Perspective of the Government . . . 81
3.4.1 In?ation and Real Interest Rates on Debt Portfolios . . . . . . 82
3.4.2 Incidence and Magnitude of the Liquidation E?ect . . . . . . . 84
3.4.3 Liquidation Revenues for the Government . . . . . . . . . . . 87
3.4.4 The Role of In?ation Expectations and Financial Repression . 89
3.4.5 Are the Results Biased by Bonds Issued Before 1945? . . . . . 95
3.4.6 Comparison to In?ation Tax . . . . . . . . . . . . . . . . . . . 98
3.4.7 E?ect on the Stock of Debt . . . . . . . . . . . . . . . . . . . 101
3.4.8 Understanding What A?ects the Probability of a Liquidation
Year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
3.5 How did Investors Fare During this Period? . . . . . . . . . . . . . . 108
v
3.5.1 Stocks, Bills and Bonds during the Sample Period . . . . . . . 109
3.5.2 A Longer Historical Perspective . . . . . . . . . . . . . . . . . 111
3.5.3 The Equity Premium Puzzle Revisited . . . . . . . . . . . . . 113
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
A Evidence of the Presence of Financial Repression 134
A.1 United States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
A.2 United Kingdom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
A.3 Japan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
A.4 France . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
A.5 Italy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
A.6 Australia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
A.7 India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
B Appendix Tables and Literature Review 157
C Data Appendix and Glossary 163
C.1 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
C.2 Variable De?nition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
C.3 Data Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Bibliography 168
vi
List of Tables
2.1 Episodes of Domestic Debt Conversions, Default or Restructuring,1920s-
1950s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Selected Measures Associated with Financial Repression . . . . . . . 45
2.3 Incidence and Magnitude of the Liquidation of Public Debt: Selected
Countries, 1945-1980 . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.4 Incidence of Liquidation Years for Di?erent Real Interest Rate Thresh-
olds: Selected Countries, 1945-1980 . . . . . . . . . . . . . . . . . . . 58
2.5 Government Revenues (interest cost savings) from the "Liquidation
E?ect:" per year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.6 Debt Liquidation through Financial Repression: Selected Countries,
1945-1955 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.7 In?ation Performance during Major Domestic Public Debt Reduction
Episodes: 28 Countries, 1790-2009 . . . . . . . . . . . . . . . . . . . 60
3.1 Countries in the Sample and Sample Periods . . . . . . . . . . . . . . 123
3.2 Summary of In?ation and Real Interest Rate (in percent) . . . . . . . 123
3.3 Incidence of Liquidation Years . . . . . . . . . . . . . . . . . . . . . . 124
3.4 Liquidation Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.5 Liquidation Revenues . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
3.6 Comparison Liquidation Revenues as percentage of: . . . . . . . . . . 125
3.7 Comparison of Median In?ation between LE Years and 1930-2010 . . 126
3.8 Regressions for In?ation Expectations . . . . . . . . . . . . . . . . . . 127
3.9 In?ation Surprises and Liquidation Years . . . . . . . . . . . . . . . . 128
3.10 Comparison between Ex Ante and Ex Post Yield to Maturity for
securities issued within sample period (in percent) . . . . . . . . . . . 128
3.11 Comparison Liquidation E?ect Revenues and In?ation Tax (as %GDP)129
3.12 Stock of Debt under di?erent paths for the in?ation rate . . . . . . . 130
3.13 Panel Estimation Results- Dependent Variable: CIR . . . . . . . . . . 130
3.14 Comparison of Real Returns Bills, Bonds and Equity 1945-1980 (in
percent) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
3.15 Percentage of the time stocks outperform bonds for di?erent holding
periods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.16 Real Returns on Bonds and Bills during 1900-2000 . . . . . . . . . . 132
3.17 Real Returns on Stocks during 1900-2000 . . . . . . . . . . . . . . . . 133
B.1 Real Interest Rates during Financial Repression and Post-Liberalization157
B.2 Measuring "Taxes" from Financial Repression: Selected Papers . . . . 160
C.1 Country Speci?c Data Sources . . . . . . . . . . . . . . . . . . . . . . 166
C.2 India: Composition of Domestic Debt (as percentage of total domestic debt)167
C.3 US: Composition of Domestic Debt (as percentage of total domestic debt) . 167
vii
List of Figures
2.1 Surges in Central Government Public Debts and their Resolution:
Advanced Economies and Emerging Markets, 1900-2011 . . . . . . . . 39
2.2 Average Ex-post Real Rate on Treasury Bills: Advanced Economies
and Emerging Markets, 1945-2009 (3-year moving averages, in percent) 40
2.3 Average Ex-post Real Discount Rate: Advanced Economies and Emerg-
ing Markets, 1945-2009 (3-year moving averages, in percent) . . . . . 41
2.4 Average Ex-post Real Rate on Deposits: Advanced Economies and
Emerging Markets, 1945-2009 (3-year moving averages, in percent) . . 42
2.5 Real Interest Rates Frequency Distributions: Advanced Economies,
1945-2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.6 Real Deposit Interest Rates Frequency Distributions: United King-
dom, 1880-2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.1 Debt-to-GDP, Real Growth Rate and Distribution of Real Returns . . 119
3.2 Frequency Distributions of Nominal and Ex Post Real Yield to Ma-
turity (YTM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
3.3 Maturity Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.4 Rolling 30-year Equity Premium . . . . . . . . . . . . . . . . . . . . . 122
viii
Chapter 1
Introduction
As a result of the recent ?nancial crisis, the public debt-to-GDP ratios of most
advanced economies have reached historically high levels. A similar phenomena was
observed after the end of World War I and the Great Depression, and again after
the end of World War II.
1
What is novel about this episode is that the debt ratios
of the private sector (both household and corporate) are also at historically high
levels. As of the end of 2009, the total debt-to-GDP ratio of Japan stood at 477
percent, 466 percent in the UK, and almost 300 percent in the US.
2
A high level of private indebtedness could worsen even further the public ?-
nances of several countries. In the past, private debts before a crisis have usually
become public debt afterwards. Reinhart (2012) points at the cases of Ireland were
public debt was around 25 percent of GDP in 2007 and has since then increased to
107 percent as of the end of 2011; and also to the case of the US where the federal
debt increased by 25 percentage points as a result of the transfer of the debts of
Fannie Mae and Freddie Mac. An added element of concern is that there is increas-
ing evidence that high debt ratios can negatively a?ect real growth once they reach
a certain threshold (See for instance, Reinhart and Rogo?, 2010; Checherita and
1
For instance, the debt ratio for advanced economies, calculated as a simple average of the
debt-to-GDP ratios of 19 advanced countries, reached 79 percent in 1922, 73 percent in 1932, 100
percent in 1946, and 70 percent in 2009.
2
Total debt includes: government, non?nancial businesses, households, and ?nancial
institutions.http://www.gfmag.com/tools/global-database/economic-data/10403-total-debt-to-
gdp.html#axzz1rMxbYikW
1
Rother, 2010; Kuman and Woo, 2010), making it less likely that countries will be
able to grow their way out of debt (at least in the short term).
Revisiting similar episodes in the past can be useful for understanding the
possible courses of action currently available to governments. This dissertation looks
at how governments have reduced high debt burdens in the past, with particular
emphasis to the post-World War II period. World War I and Great Depressions debts
were reduced primarily via default, restructurings and conversions. There were also
some cases of hyperin?ation such as Germany and Hungary. Following World War
II there were some defaults on external debt (i.e., Italy and Japan) but the majority
of the debts -which were denominated in local currency- were reduced through a
combination of ?nancial repression and in?ation. This mechanism represents a more
subtle and gradual form of debt restructuring or taxation.
The term "?nancial repression" is used to refer to a situation characterized by:
i) numerous policies and regulations which introduce frictions in ?nancial markets,
and ii) large participation of nonmarket players. The policies that will be partic-
ularly relevant are those which create captive investors for government debt, and
hence allow the government to issue debt at a rate below what the market would
charge absent any restrictions. Being able to issue debt at a below market interest
rate represents a saving in interest payments for the government. When combined
with an in?ation rate above the nominal interest rate, this leads to negative real
interest rates that e?ectively reduce government debt. An important observation is
that this mechanism does not require unanticipated in?ation in order to be e?ective.
The main database was constructed from primary sources, and replicates the
2
domestic debt portfolios of 12 countries. These portfolios include detailed infor-
mation on each security outstanding every year, and covers for most countries the
period 1945-1980. The countries in the sample are: Argentina, Australia, Belgium,
France, India, Ireland, Italy, Japan, South Africa, Sweden, the United Kingdom, and
the United States.These countries exhibit di?erent levels of economic development,
di?erent degrees of involvement in World War II, di?erent in?ationary and debt
histories. The diversity across countries is important to understand how general the
results are.
Real interest rates were signi?cantly lower during the ?nancial repression pe-
riod (1945-1980) than both in the periods before and after. This is true regardless
of the measure of interest rate that is used (whether central bank discount, treasury
bills, deposit, or lending rates and whether for advanced or emerging markets). For
the advanced economies, real ex-post interest rates were negative in about half of the
years of the ?nancial repression era compared with less than 15 percent of the time
since the early 1980s. When looking at the portfolio of domestic debt, real interest
rates were also negative half of the time between 1945-1980. This high incidence of
negative real interest rates implied large savings for the government, which averaged
between 2 to 3 percent of GDP per year.
The dissertation is structured in two chapters. Chapter 2 starts by document-
ing the large incidence of defaults following the World War I and Great Depression.
Evidence on the evolution and distribution of several di?erent measures of real in-
terest rates for advanced and emerging economies is also provided. The chapter also
documents the importance of ?nancial repression as a restructuring mechanism for
3
government debt after World War II. It contains ?rst pass estimates which show
that the e?ects are quantitatively important. Chapter 3 complements the results
of the previous chapter in three important ways. First, it presents a conceptual
framework which is useful to understand the di?erent channels through which in?a-
tion can reduce government debt, and also how to think of ?nancial repression as
a restructuring mechanism. Second, through di?erent empirical exercises, it shows
that ?nancial repression in combination with in?ation is the most important channel
through which debt was reduced. Finally, the paper looks at the key features of this
period of ?nancial repression and how the returns of other assets were a?ected. The
main ?nding of that section is that the abnormally low return on government bonds
is also re?ected in the equity premium, which almost doubled in the US during the
period 1945-1980.
Since the beginning of the 1980s the world has moved toward a more liber-
alized environment, which raises the question of how relevant the ?ndings of this
dissertation may be to explain the debt reduction that will need to take place in
the future. However, since the outbreak of the crisis there have been several regula-
tory changes which resemble those in place during the period 1945-1980 (Reinhart,
Kirkeegard, and Sbrancia; 2011). An element to keep in mind is that many of the
policies in place during the post-World War II period were not designed to reduce
government debt. On the contrary, they were designed to preserve and guarantee
the stability of the ?nancial system (i.e., ceilings on deposit rates). There were
however "unintended" consequences of these policies, which facilitated a relatively
quick debt reduction.
4
Chapter 2
The Liquidation of Government Debt (with Carmen M. Reinhart)
2.1 Introduction
Some people will think the 2
3
4
nonmarketable bond is a trick issue. We
want to meet that head on. It is. It is an attempt to lock up as much
as possible of these longer-term issues.
Assistant Secretary of the Treasury William McChesney Martin Jr.
1
The decade that preceded the outbreak of the subprime crisis in the summer
of 2007 produced a record surge in private debt in many advanced economies, in-
cluding the United States. The period prior to the 2001 burst of the "tech bubble"
was associated with a marked rise in the leverage of non?nancial corporate business;
in the years 2001-2007, debts of the ?nancial industry and households reached un-
precedented heights.
2
The decade following the crisis may yet mark a record surge
in public debt during peacetime, at least for the advanced economies. It is not
surprising that debt reduction, of one form or another, is a topic that is receiving
1
FOMC minutes, March 1â??2, 1951, remarks on the 1951 conversion of short-term marketable
US Treasury debts for 29-year nonmarketable bonds. Martin subsequently became chairman of
the Board of Governors, 1951â??70.
2
The surge in private debt is manifest in both the gross external debt ?gures of the private
sector (see Lane and Milesi-Ferretti, 2010, for careful and extensive historical documentation since
1970 and Reinharthttp://www.carmenreinhart.com/ for a splicing of their data with the latest
IMF/World Bank ?gures) and domestic bank credit (as documented in Reinhart, 2010). Relative
to GDP, these debt measures reached unprecented heights during 2007-2010 in many advanced
economies.
5
substantial attention in academic and policy circles alike.
3
Throughout history, debt/GDP ratios have been reduced by (i) economic
growth; (ii) substantive ?scal adjustment/austerity plans; (iii) explicit default or
restructuring of private and/or public debt; (iv) a sudden surprise burst in in?ation;
and (v) a steady dosage of ?nancial repression that is accompanied by an equally
steady dosage of in?ation. (Financial repression is de?ned in Box 1) It is critical
to clarify that options (iv) and (v) are viable only for domestic-currency debts.
Since these debt-reduction channels are not necessarily mutually exclusive, histori-
cal episodes of debt-reduction have owed to a combination of more than one of these
channels.
4
Hoping that substantial public and private debt overhangs are resolved by
growth may be uplifting, but it is not particularly practical from a policy stand-
point. The evidence, at any rate, is not particularly encouraging, as high levels
of public debt appear to be associated with lower growth.
5
The e?ectiveness of
?scal adjustment/austerity in reducing debt -and particularly, their growth conse-
quences (which are the subject of some considerable debate)- is beyond the scope
of this paper. Reinhart and Rogo? (2009 and 2011) analyze the incidence of ex-
plicit default or debt restructuring (or forcible debt conversions) among advanced
3
Among recent studies, see for example, Alesina and Ardagna (2009), IMF (2010), Lilico,
Holmes and Sameen (2009) on debt reduction via ?scal adjustment and Sturzenegger and Zettle-
meyer (2006), Reinhart and Rogo? (2009) and sources cited therein on debt reduction through
default and restructuring.
4
For instance, in analyzing external debt reduction episodes in emerging markets, Reinhart,
Rogo?, and Savastano (2003) suggest that default and debt/restructuring played a leading role
in most of the episodes they identify. However, in numerous cases the debt restructurings (often
under the umbrella of IMF programs) were accompanied by debt repayments associated with some
degree of ?scal adjustment.
5
See Checherita and Rother (2010), Kumar and Woo (2010), and Reinhart and Rogo? (2010).
6
economies (through and including World War II episodes) and emerging markets as
well as hyperin?ation as debt reduction mechanisms.
The aim of this paper is to document the more subtle and gradual form of debt
restructuring or "taxation" that has occurred via ?nancial repression (as de?ned in
Box 1). We show that such repression helped reduce lofty mountains of public debt
in many of the advanced economies in the decades following World War II and
subsequently in emerging markets, where ?nancial liberalization is of more recent
vintage.
6
We ?nd that ?nancial repression in combination with in?ation played an
important role in reducing debts. In?ation need not take market participants en-
tirely by surprise and, in e?ect, it need not be very high (by historic standards). In
e?ect, ?nancial repression via controlled interest rates, directed credit, and persis-
tent, positive in?ation rates is still an e?ective way of reducing domestic government
debts in the world’s second largest economy–China.
7
Prior to the 2007 crisis, it was deemed unlikely that advanced economies could
experience ?nancial meltdowns of a severity to match those of the pre-World War II
era; the prospect of a sovereign default in wealthy economies was similarly unthink-
able.
8
Repeating that pattern, the ongoing discussion of how public debts have
6
In a recent paper, Aizenman and Marion (2010) stress the important role played by in?ation
in reducing U.S. World War II debts and develop a framework to highlight how the government
may be tempted to follow that route in the near future. However, the critical role played by
?nancial repression (regulation) in keeping nominal interest rates low and producing negative real
interest rates was not part of their analysis.
7
Bai et. al. (1999), for example, present a framework that provides a general rationale for
?nancial repression as an implicit taxation of savings. They argue that when e?ective income-
tax rates are very uneven, as common in developing countries, raising some government revenue
through mild ?nancial repression can be more e?cient than collecting income tax only.
8
The literature and public discussion surrounding "the great moderation" attests to this benign
view of the state of the macroeconomy in the advanced economies. See, for example, McConnell
and Perez-Quiros (2000).
7
been reduced in the past has focused on the role played by ?scal adjustment. It
thus appears that it has also been collectively "forgotten" that the widespread sys-
tem of ?nancial repression that prevailed for several decades (1945-1980s) worldwide
played an instrumental role in reducing or "liquidating" the massive stocks of debt
accumulated during World War II in many of the advanced countries, United States
inclusive.
9
We document this phenomenon.
The next section discusses how previous "debt-overhang" episodes have been
resolved since 1900. There is a brief sketch of the numerous defaults, restructurings,
conversions (forcible and "voluntary") that dealt with the debts of World War I
and the Great Depression. This narrative, which follows Reinhart and Rogo? (2009
and 2011), primarily serves to highlight the substantially di?erent route taken after
World War II to deal with the legacy of high war debts. Section III provides a short
description of the types of ?nancial sector policies that facilitated the liquidation
of public debt. Hence, our analysis focuses importantly on regulations a?ecting
interest rates (with the explicit intent on keeping these low) and on policies creating
"captive" domestic audiences that would hold public debts (in part achieved through
capital controls, directed lending, and an enhanced role for nomarketable public
debts).
We also focus on the evolution of real interest rates during the era of ?nancial
repression (1945-1980s). We show that real interest rates were signi?cantly lower
during 1945-1980 than in the freer capital markets before World War II and after ?-
9
For the political economy of this point see the analysis presented in Alesina, Grilli, and
Milesi Ferretti (1993). They present a framework and stylized evidence to support that strong
governments coupled with weak central banks may impose capital controls so as to enable them to
raise more seigniorage and keep interest rates arti?cially low-facilitating domestic debt reduction.
8
nancial liberalization. This is the case irrespective of the interest rate used–whether
central bank discount, treasury bills, deposit, or lending rates and whether for ad-
vanced or emerging markets. For the advanced economies, real ex-post interest rates
were negative in about half of the years of the ?nancial repression era compared with
less than 15 percent of the time since the early 1980s.
In Section IV, we provide a basic conceptual framework for calculating the "?-
nancial repression tax," or more speci?cally, the annual "liquidation rate" of govern-
ment debt. Alternative measures are also discussed. These exercises use a detailed
data base on a country’s public debt pro?le (coupon rates, maturities, composition,
etc.) from 1945 to 1980 constructed by Sbrancia (2011). This "synthetic" public
debt portfolio re?ects the actual shares of debts across the di?erent spectra of ma-
turities as well as the shares of marketable versus nonmarketable debt (the latter
involving both securitized debt as well as direct bank loans).
Section V presents the central ?ndings of the paper, which are estimates of the
annual "liquidation tax" as well as the incidence of liquidation years for ten coun-
tries (Argentina, Australia, Belgium, India, Ireland, Italy, South Africa, Sweden,
the United Kingdom, and the United States). For the United States and the United
Kingdom, the annual liquidation of debt via negative real interest rates amounted
to 2 to 3 percent of GDP on average per year. Such annual de?cit reduction quickly
accumulates (even without any compounding) to a 20-30 percent of GDP debt re-
duction in the course of a decade. For other countries that, recorded higher in?ation
rates the liquidation e?ect was even larger. As to the incidence of liquidation years,
Argentina sets the record with negative real rates recorded in all years but one from
9
1945 to 1980.
Section VI examines the question of whether in?ation rates were systematically
higher during periods of debt reduction in the context of a broader 28-country sample
that spans both the heyday of ?nancial repression and the periods before and after.
We describe the algorithm used to identify the largest debt reduction episodes on
a country-by-country basis and, show that in 21 of the 28 countries in?ation was
higher during the larger debt reduction periods.
Finally, we discuss some of the implications of our analysis for the current debt
overhang and highlight areas for further research. Two appendices to this paper: (i)
compare our methodology to other approaches in the literature that have been used
to measure the extent of ?nancial repression or calculate the ?nancial repression
tax; (ii) provide country-speci?c details on the behavior of real interest rates across
regimes; and (iii) describe the coverage and extensive sources for the data compiled
for this study.
Box 1: Financial Repression De?ned
The pillars of "Financial repression"
The term ?nancial repression was introduced in the literature by the works
of Edward Shaw (1973) and Ronald McKinnon (1973). Subsequently, the term
became a way of describing emerging market ?nancial systems prior to the
widespread ?nancial liberalization that began in the 1980 (see Agenor and Mon-
tiel, 2008, for an excellent discussion of the role of in?ation and Giovannini
10
and de Melo, 1993 and Easterly, 1989 for country-speci?c estimates). However,
as we document in this paper, ?nancial repression was also the norm for ad-
vanced economies during the post-World War II period and in varying degrees
up through the 1980s. We describe here some of its main features.
(i) Explicit or indirect caps or ceilings on interest rates, particularly
(but not exclusively) those on government debts. These interest rate ceilings
could be e?ected through various means including: (a) explicit government reg-
ulation (for instance, Regulation Q in the United States prohibited banks from
paying interest on demand deposits and capped interest rates on saving deposits);
(b) ceilings on banks’ lending rates, which were a direct subsidy to the govern-
ment in cases where it borrowed directly from the banks (via loans rather than
securitized debt); and (c) interest rate cap in the context of ?xed coupon rate
nonmarketable debt or (d) maintained through central bank interest rate targets
(often at the directive of the Treasury or Ministry of Finance when central bank
independence was limited or nonexistent). Allan Meltzer’s (2003) monumental
history of the Federal Reserve (Volume I) documents the US experience in this
regard; Alex Cukierman’s (1992) classic on central bank independence provides
a broader international context.
(ii) Creation and maintenance of a captive domestic audience that
facilitated directed credit to the government. This was achieved through multi-
ple layers of regulations from very blunt to more subtle measures. (a) Capital
account restrictions and exchange controls orchestrated a "forced home bias" in
11
the portfolio of ?nancial institutions and individuals under the Bretton Woods
arrangements. (b) High reserve requirements (usually non-remunerated) as a
tax levy on banks (see Brock, 1989, for an insightful international comparison).
Among more subtle measures, (c) "prudential" regulatory measures requiring
that institutions (almost exclusively domestic ones) hold government debts in
their portfolios (pension funds have historically been a primary target). (d)
Transaction taxes on equities (see Campbell and Froot, 1994) also act to direct
investors toward government (and other) types of debt instruments. And (e)
prohibitions on gold transactions.
(iii) Other common measures associated with ?nancial repression aside
from the ones discussed above are, (a) direct ownership (e.g., in China or India)
of banks or extensive management of banks and other ?nancial institutions (e.g.,
in Japan) and (b) restricting entry into the ?nancial industry and directing credit
to certain industries (see Beim and Calomiris, 2000).
2.2 Default, Restructuring and Conversions: Highlights from 1920s-
1950s
Peaks and troughs in public debt/GDP are seldom synchronized across many
countries’ historical paths. There are, however, a few historical episodes where global
(or nearly global) developments, be it a war or a severe ?nancial and economic
crisis, produce a synchronized surge in public debt, such as the one recorded for
12
advanced economies since 2008. Using the Reinhart and Rogo? (2011) database for
70 countries, Figure 2.1 provides central government debt/GDP for the advanced
and emerging economies subgroups since 1900. It is a simple arithmetic average
that does not assign weight according to country size.
2.2.1 Global debt surges and their resolution
An examination of these two series identi?es a total of ?ve peaks in world
indebtedness. Three episodes (World War I, World War II, and the Second Great
Contraction, 2008-present) are almost exclusively advanced economy debt peaks;
one is unique to emerging markets (1980s debt crisis followed by the transition
economies’ collapses); and the Great Depression of the 1930s is common to both
groups. World War I and Depression debts were importantly resolved by widespread
default and explicit restructurings or predominantly forcible conversions of domestic
and external debts in both the now-advanced economies, and the emerging markets.
Notorious hyperin?ation in Germany, Hungary and other parts of Europe violently
liquidated domestic-currency debts. Table 2.1 and the associated discussion provide
a chronology of these debt resolution episodes. As Reinhart and Rogo? (2009 and
2011) document, debt reduction via default or restructuring has historically been
associated with substantial declines in output in the run-up to as well as during the
credit event and in its immediate aftermath.
The World War II debt overhang was importantly liquidated via the combina-
tion of ?nancial repression and in?ation, as we shall document. This was possible
13
because debts were predominantly domestic and denominated in domestic curren-
cies. The robust post-war growth also contributed importantly to debt reduction
in a way that was a marked contrast to the 1930s, when the combined e?ects of
de?ation and output collapses worked to worsen the debt/GDP balance in the way
stressed by Irving Fisher (1933).
The resolution of the emerging market debt crisis involved a combination of
default or restructuring of external debts, explicit default, or ?nancial repression
on domestic debt. In several episodes, notably in Latin America, hyperin?ation in
the mid-to-late 1980s and early 1990s completed the job of signi?cantly liquidating
(at least for a brief interlude) the remaining stock of domestic currency debt (even
when such debts were indexed, as was the case of Brazil).
10
2.2.2 Default, restructurings and forcible conversions in the 1930s
Table 2.1 lists the known "domestic credit events" of the Depression. Default
on or restructuring of external debt (see the extensive notes to the table) also of-
ten accompanied the restructuring or default of the domestic debt. All the Allied
governments, with the exception of Finland, defaulted on (and remained in default
through 1939 and never repaid) their World War I debts to the United States as
economic conditions deteriorated worldwide during the 1930s.
11
Thus, the high debts of World War I and the subsequent debts associated with
the Depression of the 1930s were resolved primarily through default and restructur-
10
Backward-looking indexation schemes are not particularly e?ective in hyperin?ationary con-
ditions.
11
Finland, being under threat of Soviet invasion at the time, maintained payments on its debts
to the United States so as to maintain the best possible relationship.
14
ing. Neither economic growth nor in?ation contributed much. In e?ect, for all 21
now-advanced economies, the median annual in?ation rate for 1930-1939 was barely
above zero (0.4 percent).
12
Real interest rates remained high through signi?cant
stretches of the decade.
It is important to stress that during the period after World War I the gold
standard was still in place in many countries, which meant that monetary policy was
subordinated to keep a given gold parity. In those cases, in?ation was not a policy
variable available to policymakers in the same way that it was after the adoption of
?at currencies.
12
See Reinhart and Reinhart (2010).
15
Table 2.1: Episodes of Domestic Debt Conversions, Default or Restructuring,1920s-
1950s
Country Dates Commentary
Australia 1931/1932 The Debt Conversion Agreement Act in 1931/32 which
appears to have done something similar to the later NZ
induced conversion. See New Zealand entry.
1
Bolivia 1927 Arrears of interest lasted until at least 1940.
Canada (Al-
berta)
April 1935 The only province to default - which lasted for about 10
years.
China 1932 First of several "consolidations", monthly cost of domes-
tic service was cut in half. Interest rates were reduced
to 6 percent (from over 9 percent)-amortization periods
were about doubled in length.
France 1932 Various redeemable bonds with coupons between 5 and 7
percent, converted into a 4.5 percent bond with maturity
in 75 years.
Greece 1932 Interest on domestic debt was reduced by 75 percent
since 1932; Domestic debt was about 1/4 of total public
debt.
Italy November 6th,
1926
Issuance of Littorio. There were 20.4 billion lire subject
to conversion, of which 15.2 billion were "Buoni Ordi-
nari" (short term securities).
Italy February 3rd,
1934
5 percent Littorio (see entry above) converted into 3.5
percent Redimibile
Mexico 1930s Service on external debt was suspended in 1928. Dur-
ing the 1930s, interest payments included "arrears of
expenditure and civil and military pensions."
16
Table 2.1 – Continued from previous page
Country Dates Commentary
New Zealand 1933 In March 1933 the New Zealand Debt Conversion Act
was passed providing for voluntary conversion of internal
debt amounting to 113 million pounds to a basis of 4 per
cent for ordinary debt and 3 per cent for tax-free debt.
Holders had the option of dissenting but interest in the
dissented portion was made subject to an interest tax of
33.3 per cent.
1
Peru 1931 After suspending service on external debt on May 29,
Peru made "partial interest payments" on domestic
debt.
Romania February 1933 Redemption of domestic and foreign debt is suspended
(except for three loans).
Spain October 1936-
April 1939
Interest payments on external debt were suspended, ar-
rears on domestic debt service.
United States 1933 Abrogation of the gold clause. In e?ect, the U.S. refused
to pay Panama the annuity in gold due to Panama ac-
cording to a 1903 treaty. The dispute was settled in 1936
when the US paid the agreed amount in gold balboas.
United King-
dom
1932 Most of the outstanding WWI debt was consolidated
into a 3.5 percent perpetual annuity. This domestic debt
conversion was apparently voluntary. However, some of
the WWI debts to the United States were issued under
domestic (UK) law (and therefore classi?ed as domestic
debt) and these were defaulted on following the end of
the Hoover 1931 moratorium.
17
Table 2.1 – Continued from previous page
Country Dates Commentary
Uruguay November 1, 1932
- February, 1937
After suspending redemption of external debt on Jan-
uary 20, redemptions on domestic debt were equally sus-
pended.
Austria December 1945 Restoration of schilling (150 limit per person). Remain-
der placed in blocked accounts. In December 1947, large
amounts of previously blocked schillings invalidated and
rendered worthless. Temporary blockage of 50 percent
of deposits.
Germany June 20, 1948 Monetary reform limiting 40 Deutschemark per person.
Partial cancellation and blocking of all accounts.
Japan March 2, 1946-
1952
After in?ation, exchange of all bank notes for new is-
sue (1 to 1) limited to 100 yen per person. Remaining
balances were deposited in blocked accounts.
Russia 1947 The monetary reform subjected privately held currency
to a 90 percent reduction.
April 10 1957 Repudiation of domestic debt (about 253 billion rubles
at the time).
Sources: Reinhart and Rogo? (2011) and the authors.
1
See Schedvin (1970) and Prichard (1970), for accounts of the Australian and New Zealand
conversions, respectively, during the Depression. Michael Reddell kindly alerted us to these
episodes and references. Alex Pollock pointed out the relevance of widespread restrictions
on gold holdings in the United States and elsewhere during the ?nancial repression era.
Notes: We have made signi?cant further progress in sorting out the defaults on World War
I debts to the United States, notably by European countries. In all cases these episodes are
classi?ed as a default on external debts. However, in some case â??such as the UK–some of
the WWI debts to the US were also issued under the domestic law and, as such, would also
qualify as a domestic default. The external defaults on June 15, 1934 included: Austria,
Belgium, Czechoslovakia, Estonia, France, Greece, Hungary, Italy, Latvia, Poland, United
Kingdom. Only Finland made payments. See New York Times, June 15, 1934.
18
2.3 Financial Repression: policies and evidence from real interest rates
2.3.1 Selected ?nancial regulation measures during the "era of ?nancial
repression"
One salient characteristic of ?nancial repression is its pervasive lack of trans-
parency. The reams of regulations applying to domestic and cross-border ?nancial
transactions and directives cannot be summarized by a brief description. Table 2.2
makes this clear by providing a broad sense of the kinds of regulations on inter-
est rates and cross-border and foreign exchange transactions and how long these
lasted since the end of World War II in 1945. A common element across countries
"?nancial architecture" not brought out in Table 2.2 is that domestic government
debt played a dominant role in domestic institutions’ asset holdings–notably that
of pension funds. High reserve requirements, relative to the current practice in ad-
vanced economies and many emerging markets, were also a common way of taxing
the banks not captured in our minimalist description. The interested reader is re-
ferred to Brock (1989) and Agenor and Montiel (2008), who focus on the role of
reserve requirements and their link to in?ation (see also Appendix Table B.2 and
accompanying discussion.)
19
2.3.2 Real Interest Rates
One of the main goals of ?nancial repression is to keep nominal interest rates
lower than would otherwise prevail. This e?ect, other things equal, reduces the
governments’ interest expenses for a given stock of debt and contributes to de?cit
reduction. However, when ?nancial repression produces negative real interest rates,
this also reduces or liquidates existing debts. It is a transfer from creditors (savers)
to borrowers (in the historical episode under study here–the government).
The ?nancial repression tax has some interesting political-economy properties.
Unlike income, consumption, or sales taxes, the "repression" tax rate (or rates) are
determined by ?nancial regulations and in?ation performance that are opaque to
the highly politicized realm of ?scal measures. Given that de?cit reduction usually
involves highly unpopular expenditure reductions and (or) tax increases of one form
or another, the relatively "stealthier" ?nancial repression tax may be a more politi-
cally palatable alternative to authorities faced with the need to reduce outstanding
debts. As discussed in Obstfeld and Taylor (2004) and others, liberal capital- market
regulations (the accompanying market-determined interest rates) and international
capital mobility reached their heyday prior to World War I under the umbrella of
the gold standard. World War I and the suspension of convertibility and interna-
tional gold shipments it brought, and, more generally, a variety of restrictions on
cross-border transactions were the ?rst blows to the globalization of capital. Global
capital markets recovered partially during the roaring twenties, but the Great De-
pression, followed by World War II, put the ?nal nails in the co?n of laissez faire
20
banking. It was in this environment that the Bretton Woods arrangement of ?xed
exchange rates and tightly controlled domestic and international capital markets
was conceived.
13
In that context, and taking into account the major economic dis-
locations, scarcities, etc. which prevailed at the closure of the second great war, we
witness a combination of very low nominal interest rates and in?ationary spurts of
varying degrees across the advanced economies. The obvious result were real inter-
est rates -whether on treasury bills (Figure 2.2), central bank discount rates (Figure
2.3), deposits (Figure 2.4), or loans (not shown)- that were markedly negative during
1945-1946.
For the next 35 years or so, real interest rates in both advanced and emerging
economies would remain consistently lower than the eras of freer capital mobility
before and after the ?nancial repression era. In e?ect, real interest rates (Figures
2.2-2.4) were on average negative.
14
Binding interest rate ceilings on deposits (which
kept real ex post deposit rates even more negative than real ex-post rates on treasury
bills, as shown in Figures 2.2 and 2.4) "induced" domestic savers to hold government
bonds. What delayed the emergence of leakages in the search for higher yields
(apart from prevailing capital controls) was that the incidence of negative returns
on government bonds and on deposits was (more or less) a universal phenomenon
at this time.
15
The frequency distributions of real rates for the period of ?nancial
13
In a framework where there are both tax collection costs and a large stock of domestic govern-
ment, Aizenman and Guidotti, (1994) show how a government can resort to capital controls (which
lower domestic interest rates relative to foreign interest rates) to reduce the costs of servicing the
domestic debt.
14
Note that real interest rates were lower in a high-economic-growth period of 1945 to 1980
than in the lower growth period 1981-2009; this is exactly the opposite of the prediction of a basic
growth model and therefore indicative of signi?cant impediments to ?nancial trade
15
A comparison of the return on government bonds to that of equity during this period and its
21
repression (1945-1980) and the years following ?nancial liberalization (roughly 1981-
2009 for the advanced economies) shown in the three panels of Figure 2.5, highlight
the universality of lower real interest rates prior to the 1980s and the high incidence
of negative real interest rates.
Such negative (or low) real interest rates were consistently and substantially
below the real rate of growth of GDP, this is consistent with the observation of
Elmendorf and Mankiw (1999) when they state "An important factor behind the
dramatic drop (in US public debt) between 1945 and 1975 is that the growth rate of
GNP exceeded the interest rate on government debt for most of that period." They
fail to explain why this con?guration should persist over three decades in so many
countries.
Real interest rates on deposits were negative in about 60 percent of the obser-
vations. In e?ect, real ex-post deposit rates were below one percent about 83 percent
of the time. Appendix Table B.1, which shows for each country average real interest
rates during the ?nancial repression period (the dates vary, as highlighted in Table
B.1, depending on when interest rates were liberalized) and thereafter, substantiates
our claims that low and negative real interest rates (by historical standards) were
the norm across countries with very di?erent levels of economic development.
The preceding analysis sets the general tone of what to expect, in terms of
real rates of return on a portfolio of government debt, during the era of ?nancial
repression. For the United States, for example, Homer and Sylla (1963) describe
connection to "the equity premium puzzle" can be found in Sbrancia (2011).
22
1946-1981 as the second (and longest) bear bond market in US history.
16
To reit-
erate the point that the low real interest rates of the ?nancial repression era were
exceptionally low in relation to not only the post-liberalization period but also the
more liberal ?nancial environment of pre-World War II, Figure 2.6 plots the fre-
quency distribution of real interest rates on deposits for the United Kingdom over
three subperiods, 1880-1939,
17
1945-1980, and 1981-2010.
The preceding analysis of real interest rates despite being qualitatively sugges-
tive falls short of providing estimates of the magnitude of the debt-servicing savings
and outright debt liquidation that accrued to governments during this extended pe-
riod. To ?ll in that gap the next section outlines the methodological approach we
follow to quantify the ?nancial repression tax, while Section 5 presents the main
results.
2.4 The Liquidation of Government Debt: Conceptual and Data Issues
This section discusses the data and methodology we develop to arrive at esti-
mates of how much debt was liquidated via a combination of low nominal interest
rates and higher in?ation rates, or what we term "the liquidation e?ect."
18
Data requirements. Reliable estimates of the liquidation e?ect require con-
siderable data, most of which are not readily available from even the most com-
prehensive electronic databases. Indeed, most of the data used in these exercises
16
They identify 1899-1920 as the ?rst US bear bond market.
17
Excluding the WWI period.
18
Table B.2 and its accompanying discussion also examines other approaches to quantifying the
?nancial repression tax.
23
come from a broad variety of historical government publications, many which are
quite obscure, as detailed in the Data Appendix. The calculation of the "liquidation
e?ect" is a clear illustration of a case where the devil lies in the details, as the struc-
ture of government debt varies enormously across countries and within countries
over time. Di?erences in coupon rates, maturity, distribution of marketable and
nonmarketable debt, and securitized debt versus loans from ?nancial institutions
importantly shape the overall cost of debt ?nancing for the government. There is
no "single" government interest rate (such as a 3-month T-bill or a 10-year bond)
that is appropriate to apply to a hybrid debt stock. The starting point to come up
with a measure that re?ects the true cost of debt ?nancing is a reconstruction of
the government’s debt pro?le over time.
Sample. We employ two samples in our empirical analysis. We use the
database from Sbrancia (2011) of the government’s debt pro?les for 10 countries
(Argentina, Australia, Belgium, India, Ireland, Italy, South Africa, Sweden, the
United Kingdom, and the United States). These were constructed from primary
sources over the period 1945-1990 where possible or over shorter intervals (deter-
mined by data availability) for a subset of the sample. For the benchmark or basic
calculations (described below), this involves data on a detailed composition of debt,
including maturity, coupon rate, and outstanding amounts by instrument. For a
more comprehensive measure, which takes into account capital gains or losses of
holding government debt, bond price data are also required. In all cases, we also
use o?cial estimates of consumer price in?ation, which at various points in history
24
may signi?cantly understate the true in?ation rates.
19
Data on Nominal GDP and
government tax revenues are used to express the estimates of the liquidation e?ect
as ratios that are comparable across time and countries.
For our broader analysis of the behavior of in?ation during major debt reduc-
tion episodes, which has far less demanding data requirements (domestic public debt
outstanding/GDP and in?ation rates) our sample broadens to 28 countries from all
regions for 1790-2010 (or subsamples therein). The countries and their respective
coverage are listed in Appendix C.
2.4.1 Benchmark basic estimates of the "liquidation e?ect"
The debt portfolio. We construct a "synthetic portfolio"
20
for the government’s
total debt stock at the beginning of the year (?scal or calendar, as noted). This
portfolio re?ects the actual shares of debts across the di?erent spectra of maturities
as well as the shares of marketable versus nonmarketable debt.
Interest rate on the portfolio. The "aggregate" nominal interest rate for a
particular year is the coupon rate on a particular type of debt instrument weighted
by that instrument’s share in the total stock of debt.
21
We then aggregate across
19
This is primarily due to the existence of price controls which were mainly imposed during
WWII and remained for several years after the end of the con?ict. See Friedman and Schwartz
(1982) for estimates of the actual price level in the US and UK, and Wiles (1952) for post-World
War II United Kingdom.
20
The term "synthetic" is used in the sense that a hypothetical investor holds the total portfolio
of government debt at the beginning of the period, which is de?ned as either the beginning of the
calendar year or the ?scal year, depending on how the debt data is reported by the particular
country. Country speci?cs are detailed in the data appendix. The weights in this hypothetical
portfolio are given by the actual shares of each component of debt in the total domestic debt of
the government.
21
Giovannini and de Melo (1993) state "the choice of a "representative" interest rate on domestic
liabilities an almost impossible task and because there are no reliable breakdowns of domestic and
foreign liabilities by type of loan and interest rate charged." This is precisely the almost impossible
25
all debt instruments. The real rate of interest,
r
t
=
i
t?1
??
t
1?
t
(2.1)
is calculated on an ex-post basis using CPI in?ation for the corresponding
one-year period. It is a before-tax real rate of return (excluding capital gains or
losses).
22
A de?nition of debt "liquidation years." Our benchmark calculations de?ne
a liquidation year, as one in which the real rate of interest (as de?ned above) is
negative (below zero). This is a conservative de?nition of liquidation year; a more
comprehensive de?nition would include periods where the real interest rate on gov-
ernment debt was below a "market" real rate.
23
It is possible that if the equilibrium
interest rate was negative, using this de?nition would be overestimating the actual
e?ect.
Savings to the government during liquidation years. This concept
captures the savings (in interest costs) to the government from having a negative real
interest rate on government debt. (As noted it is a lower bound on saving of interest
costs, if the benchmark used assumed, for example a positive real rate of, say, two or
three percent.) These savings can be thought of as having "a revenue-equivalent" for
the government, which like regular budgetary revenues can be expressed as a share
task we undertake here. Their alternative methodology is described in appendix Table B.2
22
Some of the observations on in?ation are su?ciently high to make the more familiar linear
version of the Fisher equation a poor approximation.
23
However, determining what such a market rate would be in periods of pervasive ?nancial
repression requires assumptions about whether real interest rates during that period would have
comparable to the real interest that prevailed in period when market were liberalized and prices
were market determined.
26
of GDP or as a share of recorded tax revenues to provide standard measures of the
"liquidation e?ect" across countries and over time. The saving (or "revenue") to
the government or the "liquidation e?ect" or the "?nancial repression tax" is
the real (negative) interest rate times the "tax base," which is the stock of domestic
government debt outstanding.
2.4.2 An alternative measure of the liquidation e?ect based on total
returns
Thus far, our measure of the liquidation a?ect has been con?ned to savings
to the government by way of annual interest costs. However, capital losses (if bond
prices fall) may also contribute importantly to the calculus of debt liquidation over
time. This is the case because the market value of the debt will actually be lower than
its face value. The market value of government debt obviously matters for investors’
wealth but also measures the true capitalized value of future coupon and interest
payments. Moreover, a government (or its central bank) buying back existing debt
could directly and immediately lower the par value of existing obligations. Once
we take into account potential price changes, the total nominal return or holding
period return (HPR) for each instrument is given by:
HPR
t
=
(P
t
?P
t?1
) + C
t
P
t?1
(2.2)
where and are the prices of the bond at time and respectively, and is the annual
interest payment (i.e., the nominal coupon rate).
27
We use this total return measure as a supplement rather than as our core or
benchmark "liquidation measure" (despite the fact that it incorporates more infor-
mation on the performance of the bond portfolio).
24
Bond price data are available
only for a subset of the securities that constitute the government portfolio and, more
generally, consistent time series price data are more di?cult to get for some of the
countries in our sample. It is also worth noting that while price movements for
di?erent bonds are generally in the same direction during a particular year, there
are signi?cant di?erences in the magnitudes of the price changes. This cross-bond
variation in price performance makes it di?cult to infer the price of nonmarketable
debt (for which there are no price data altogether), as well as marketable bonds
for which there is no price data. As before, we de?ne "liquidation years" as those
periods in which the real return of the portfolio is negative.
2.4.3 The role of in?ation and currency depreciation
The idea of governments using in?ation to liquidate debt is hardly a new one
since the widespread adoption of ?at currency, as discussed earlier. It is obvious that
for any given nominal interest rate a higher in?ation rate reduces the real interest
rate on the debt, thus increasing the odds that real interest rates become negative
and the year is classi?ed as a "liquidation year." Furthermore, it is also evident that
for any year that is classi?ed as a liquidation year the higher the in?ation rate (for
a given coupon rate) the higher the saving to the government.
24
See for example, Calvoâ??s (1989) framework which highlights the role of in?ation in debt
liquidation even in the presence of short-term debt.
28
2.5 The Liquidation of Government Debt: Empirical Estimates
This section presents estimates of the "liquidation e?ect" for ten advanced
and emerging economies for most of the post-World War II period. Our main interest
lies in the period prior to the process of ?nancial liberalization that took hold during
the 1980s-that is, the era of ?nancial repression. However, as noted, this three-plus
decade-long stretch is by no means uniform. The decade immediately following
World War II was characterized by a very high public debt overhang-legacy of the
war, a higher incidence of in?ation, and often multiple currency practices (with
huge black market exchange rate premiums) in many advanced economies.
25
The
next decade (1960s) was the heyday of the Bretton Woods system with heavily
regulated domestic and foreign exchange markets and more stable in?ation rates
in the advanced economies (as well as more moderate public debt levels). The
1970s was quite distinct from the prior decades, as leakages in ?nancial regulations
proliferated, the ?xed exchange rate arrangements under Bretton Woods among the
advanced economies broke down, and in?ation began to resurface in the wake of
the global oil shock and accommodative monetary policies in the United States and
elsewhere. To this end, we also provide estimates of the liquidation of government
debt for relevant subperiods.
25
See De Vries (1969), Horse?eld (1969), Reinhart and Rogo? (2002).
29
2.5.1 Incidence and magnitude of the "liquidation tax"
Table 2.3 provides information on a country-by-country basis for the period
under study; the incidence of debt liquidation years (as de?ned in the preceding
section); the listing of the liquidation years; the average (negative) real interest rate
during the liquidation years; and the minimum real interest rate recorded (and the
year in which that minimum was reached). Given its notorious high and chronic in-
?ation history coupled with heavy-handed domestic ?nancial regulation and capital
controls during 1944-1974, it is not surprising that Argentina tops the list. Almost
all the years (92 percent) were recorded as liquidation years, as the Argentine real
ex-post interest rates were negative in every single year during 1944-1980 except
for 1953 (a just de?ationary year). For India, that share was 53 percent (slightly
more than one half of the 1949-1980 observations recorded negative real interest
rates). Before reaching the conclusion that this debt liquidation through ?nancial
repression was predominantly an emerging market phenomenon, it is worth noting
that for the United Kingdom the share of liquidation years was about 58 percent
during 1945-1980. For the United States, the world’s ?nancial center, half of the
years during that same period Treasury debt had negative real interest rates.
As to the magnitudes of the ?nancial repression tax (Table 2.3), real inter-
est rates were most negative for Argentina (reaching a minimum of -72.3 percent in
1976). The share of domestic government debt in Argentina (and other Latin Amer-
ican countries) in total (domestic plus external) public debt was substantial during
1900-1950s; it is not surprising that in light of these real rates the domestic debt
30
market all but disappeared and capital ?ight marched upwards (capital controls
notwithstanding). By the late 1970s Argentina and many other chronic in?ation
countries were predominantly relying on external debt.
26
Italian real interest rates
right after World War II were as negative as 28 percent (in 1947). For the Unites
States real rates were on average -7 percent during 1945-1947 (on average the US
had -3.5 percent real rates during the liquidation years).
There are two distinct patterns in the ten-country sample evident from an
inspection of the timing of the incidence and magnitude of the negative real rates.
The ?rst of these is the cases where the negative real rates (?nancial repression
tax) were most pronounced in the years following World War II (as war debts were
importantly in?ated away). This pattern is most evident in Australia, the United
Kingdom and the United States, although negative real rates re-emerge following
the breakdown of Bretton Woods in 1974-1975. Then there are the cases where
there is a more persistent or chronic reliance on ?nancial repression throughout the
sample as a way of funding government de?cits and/or eroding existing government
debts. The cases of Argentina and India in the emerging markets and Belgium and
Sweden in the advanced economies stand out in this regard.
The preceding analysis, as noted, adopts a very narrow, conservative calcula-
tion of both the incidence of the "liquidation e?ect" or the ?nancial repression tax.
Much of the literature on growth, as well as standard calibration exercises involving
subjective rates of time preference assume benchmark real interest rates of three
percent per annum and even higher. Thus, a threshold that only examines periods
26
See Reinhart and Rogo? (2011)’s forgotten history of domestic debt.
31
where real interest rates were actually negative is bound to underestimate the inci-
dence of "abnormally low" real interest rates during the era of ?nancial repression
(approximately taken to be 1945-1980). To assess the incidence of more broadly
de?ned low real interest rates, Table 3.3 presents for the 10 core countries the share
of years where real returns on a portfolio of government debt (as de?ned earlier)
were below zero (as in Table 2.3), one, two, and three percent, respectively.
27
In the era of ?nancial repression that we examine here, real ex post interest
rates on government debt reached three percent in only two years in the United
States; in e?ect in nearly 60 percent of the years real interest rates were below one
percent. The incidence of "abnormally low" real interest rates is comparable for
the United Kingdom and Australia-both countries had sharp and relatively rapid
declines in public debt to GDP following World War II.
28
Even in countries with
substantial economic and ?nancial volatility during this period (including Ireland,
and Italy), real interest rates on government debt above three percent were relatively
rare (accounting for no more than 20 percent of the observations).
2.5.2 Estimates of the Liquidation E?ect
Having documented the high incidence of "liquidation years" (even by conser-
vative estimates), we now calculate the magnitude of the savings to the government
(?nancial repression tax or liquidation e?ect). These estimates take "the tax rate"
27
An alternative strategy would be to use a growth model to calibrate the relationship between
the real interest rate and output growth for the counterfactual of free markets. That, however,
would make the results model speci?c.
28
"Abnormally low" by the historical standards which include periods of liberalized ?nancial
markets before and after 1945-1980; see Homer and Sylla’s (2005) classic book for a comprehensive
and insightful history of interest rates.
32
(the negative real interest rate) and multiply it by the "tax base" or the stock of
debt. Table 2.5 reports these estimates for each country.
The magnitudes are in all cases non-trivial, irrespective of whether we use the
benchmark measure that is exclusively based on interest rate (coupon yields) or
the alternative measure that includes capital gains (or losses) for the cases where
the bond price data is available. For the United States and the United Kingdom
the annual liquidation of debt via negative real interest rates amounted on average
to 2 and 3 percent of GDP a year. Obviously, annual de?cit reduction of 2 to 3
percent of GDP quickly accumulates (even without any compounding) to a 20 to
30 percent of GDP debt reduction in the course of a decade. Interestingly (but
not entirely surprising), the average annual magnitude of the liquidation e?ect for
Argentina is about the same as that of the UK, despite the fact that the average real
interest rate averaged about -3.5 percent for the UK and -21 percent for Argentina
during liquidation years in the 1945-1980 repression era. Just as money holdings
secularly shrink during periods of high and chronic in?ation, so does the domestic
debt market.
29
Argentina’s "tax base" (domestic public debt) shrank steadily during
this period; at the end of World War II nearly all public debt was domestic and by
the early 1980s domestic debt accounted for less than
1
2
of total public debt. Without
the means to liquidate external debts, Argentina defaulted on its external obligations
in 1982.
Countries like Ireland, India, Sweden and South Africa that did not experience
a massive public debt build-up during World War II recorded more modest annual
29
These issues are examined in Reinhart and Rogo? (2011).
33
savings (but still substantive) during the heyday of ?nancial repression.
30
2.6 In?ation and Debt Reduction
We have argued that in?ation is most e?ective in liquidating government debts
(or debts in general), when interest rates are not able to respond to the rise in in?a-
tion and in in?ation expectations.
31
This disconnect between nominal interest rates
and in?ation can occur if: (i) the setting is one where interest rates are either admin-
istered or predetermined (via ?nancial repression, as described); (ii) all government
debts are ?xed-rate and long maturities and the government has no new ?nancing
needs (even if there is no ?nancial repression the long maturities avoid rising interest
costs that would otherwise prevail if short maturity debts needed to be rolled over);
and (iii) all (or nearly all) debt is liquidated in one "surprise" in?ation spike. Our
attention thus far has been con?ned to the ?rst on that list, the ?nancial repres-
sion environment. The second scenario, where governments only have long-term,
?xed-rate debt outstanding and have no new ?nancing needs (de?cits) remain to be
identi?ed (however, we have a sense such episodes are relatively rare). This leaves
the third case where debts are swiftly liquidated via an in?ation spike (or perhaps
more appropriately surge). To attempt to identify potential episodes of the latter,
we conduct a simple exercise.
32
30
It is important to note that while ?nancial repression wound down in most of the advanced
economies in the sample by the mid 1980s, it has persisted in varying degrees in India through the
present (with its system of state-owned banks and widespread capital controls) and in Argentina
(except for the years of the "Convertibility Plan," April 1991-December 2001).
31
That is, the coe?cient in the Fisher equation is less than one.
32
See Chapter 3, Section 4.4 for an exercise where in?ation expectations are estimated.
34
The exercise begins by identifying debt-reduction episodes and then focusing
on the largest of these. Any decline in debt/GDP over a three year window classi?es
as a debt-reduction episode. For this pool of debt-reduction episodes, we construct
their frequency distribution (for each country) and focus on the lower (ten percent)
tail of the distribution to identify the "largest" three-year debt reduction episodes.
This algorithm biases our selection of episodes toward the more sudden (or abrupt)
ones (even if these are later reversed) which might a priori be attributable to some
combination of a booming economy, a substantive ?scal austerity plan, or a burst
in in?ation/liquidation, or explicit default or restructuring. A milder but steady
debt reduction process that lasts over many years would be identi?ed as a series of
episodes-but if the decline in debt over any particular three-year window is modest
it may not be large enough to fall in the lower ten percent of all the observations.
This exercise helps ?ag episodes where in?ation is likely to have played a signif-
icant role in public debt reduction but does not provide estimates of how much debt
was liquidated (as in the preceding analysis). Because we only require information
on domestic public debt/GDP and in?ation, we expand our coverage to 28 countries
predominantly (but not exclusively) over 1900-2009. Thus, we are not exclusively
focusing on the period of ?nancial repression but examining more broadly the role
of in?ation and debt reduction in the countries’ histories. Table 8 lists the largest
debt reduction episodes by country, the last year of the 3-year episode is shown for
each country; the year that appears in italics represents the largest single-episode
of debt reduction. The next two columns of the table are devoted to the average
and median in?ation performance during the debt reduction episodes listed in the
35
second column in comparison to the in?ation performance (average and median) for
the full sample (the coverage, which varies by country, is shown in Table A.3). In 22
of 28 countries, in?ation is signi?cantly higher in the episodes of debt reduction than
for the full sample. In the extreme cases, it is the wholesale liquidation of domestic
debt, such as during the German hyperin?ation of the early 1920s and the long-
lasting Brazilian and Argentine hyperin?ations of the early 1990s. Even without
these extreme cases, the in?ation di?erentials between the debt reduction episodes
and the full sample are suggestive of the use of in?ation (intentionally or because it
became unmanageable) to reduce (or liquidate) government debts even in periods
outside the era of heavy ?nancial repressions. The evidence is only suggestive of
this interpretation, as no explicit causal pattern is tested.
2.7 Concluding Remarks
The substantial tax on ?nancial savings imposed by the ?nancial repression
that characterized 1945-1980 was a major factor explaining the relatively rapid re-
duction of public debt in a number of the advanced economies. This fact has been
largely overlooked in the literature and discussion on debt reduction. The UK’s his-
tory o?ers a pertinent illustration. Following the Napoleonic Wars, the UK’s public
debt was a staggering 260 percent of GDP; it took over 40 years to bring it down to
about 100 percent (a massive reduction in an era of price stability and high capital
mobility anchored by the gold standard). Following World War II, the UK’s public
36
debt ratio was reduced by a comparable amount in 20 years.
33
The ?nancial repression route taken at the creation of the Bretton Woods
system was facilitated by initial conditions after the war, which had left a legacy of
pervasive domestic and ?nancial restrictions. Indeed, even before the outbreak of
World War II, the pendulum had begun to swing away from laissez-faire ?nancial
markets toward heavier-handed regulation in response to the widespread ?nancial
crises of 1929-1931. But one cannot help thinking that part of the design principle of
the Bretton Woods system was to make it easier to work down massive debt burdens.
The legacy of ?nancial crisis made it easier to package those policies as prudential.
To deal with the current debt overhang, similar policies to those documented here
may re-emerge in the guise of prudential regulation rather than under the politically
incorrect label of ?nancial repression. Moreover, the process where debts are being
"placed" at below market interest rates in pension funds and other more captive
domestic ?nancial institutions is already under way in several countries in Europe.
There are many bankrupt (or nearly so) pension plans at the state level in the
United States that bear scrutiny (in addition to the substantive unfunded liabilities
at the federal level).
Markets for government bonds are increasingly populated by nonmarket play-
ers, notably central banks of the United States, Europe and many of the largest
emerging markets, calling into question what the information content of bond prices
33
Peak debt/GDP was 260.6 in 1819 and 237.9 percent in 1947. Real GDP growth was about
the same during the two debt reduction periods (1819-1859) and (1947-1967), averaging about 2.5
percent per annum (the comparison is not exact as continuous GDP data begins in 1830). As such,
higher growth cannot obviously account for the by far faster debt reduction following World War
II.
37
are relatively to their underlying risk pro?le. This decoupling between interest rates
and risk is a common feature of ?nancially repressed systems. With public and
private external debts at record highs, many advanced economies are increasingly
looking inward for public debt placements. While to state that initial conditions on
the extent of global integration are vastly di?erent at the outset of Bretton Woods
in 1946 and today is an understatement, the direction of regulatory changes have
many common features. The incentives to reduce the debt overhang are more com-
pelling today than about half a century ago. After World War II, the overhang
was limited to public debt (as the private sector had painfully deleveraged through
the 1930s and the war); at present, the debt overhang many advanced economies
face encompasses (in varying degrees) households, ?rms, ?nancial institutions and
governments.
38
Figure 2.1: Surges in Central Government Public Debts and their Resolution: Ad-
vanced Economies and Emerging Markets, 1900-2011
0
20
40
60
80
100
120
1901 1911 1921 1931 1941 1951 1961 1971 1981 1991 2001 2011
WWI and Depression debts
(advanced economies:
default, restructuring and
conversions--a few
hyperinflations)
Advanced
economies
Emerging
Markets
Great depression
debts
(emerging markets-default)
WWII debts:
(Axis countries: default and
financial repression/inflation
Allies: financial
repression/inflation)
1980s Debt Crisis
(emerging
markets:default,
restructuring, financial
repression/inflation and
several hyperinflations)
Second Great
Contraction
(advanced
economies)
Sources: Reinhart (2010), Reinhart and Rogo? (2009 and 2011), sources cited
therein and the authors.
Notes: Listed in parentheses below each debt-surge episode are the main mech-
anisms for debt resolution besides ?scal austerity programs which were not imple-
mented in any discernible synchronous pattern across countries in any given episode.
Speci?c default/restructuring years by country are provided in the Reinhart-Rogo?
database and a richer level of detail for 1920s-1950s (including various conversions
are listed in Table 1). The "typical" forms of ?nancial repression measures are
discussed in Box 1 and greater detail for the core countries are provided in Table 2.
39
Figure 2.2: Average Ex-post Real Rate on Treasury Bills: Advanced Economies and
Emerging Markets, 1945-2009 (3-year moving averages, in percent)
1945-1980 1981-2009
-1.6 2.8
-1.2 2.6
Average Real Treasury Bill Rate
Advanced economies
Emerging markets
-15.0
-10.0
-5.0
0.0
5.0
10.0
1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
`
Advanced economies
(3-year moving average)
Emerging Markets
(3-year moving average)
Financial Repression Era
Sources: International Financial Statistics, International Monetary Fund, various
sources listed in the Data Appendix, and authors’ calculations.
Notes:The advanced economy aggregate comprises: Australia, Belgium, Canada,
Finland, France, Germany, Greece, Ireland, Italy, Japan, New Zealand, Sweden,
the United States, and the United Kingdom. The emerging market group consists
of: Argentina, Brazil, Chile, Colombia, Egypt, India, Korea, Malaysia, Mexico,
Philippines, South Africa, Turkey and Venezuela. The average is unweighted and
the country coverage is somewhat spotty prior for emerging markets to 1960
40
Figure 2.3: Average Ex-post Real Discount Rate: Advanced Economies and Emerg-
ing Markets, 1945-2009 (3-year moving averages, in percent)
1945-1980 1981-2009
-1.1 2.7
-5.3 3.8
Average Real Discount Rate
Advanced economies
Emerging markets
-20.0
-15.0
-10.0
-5.0
0.0
5.0
10.0
15.0
20.0
1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
`
Advanced economies
(3-year moving average)
Emerging Markets
(3-year moving average)
Financial Repression Era
Sources: International Financial Statistics, International Monetary Fund, various
sources listed in the Data Appendix, and authors’ calculations.
Notes:The advanced economy aggregate comprises: Australia, Belgium, Canada,
Finland, France, Germany, Greece, Ireland, Italy, Japan, New Zealand, Sweden,
the United States, and the United Kingdom. The emerging market group consists
of: Argentina, Brazil, Chile, Colombia, Egypt, India, Korea, Malaysia, Mexico,
Philippines, South Africa, Turkey and Venezuela. The average is unweighted and
the country coverage is somewhat spotty prior for emerging markets to 1960
41
Figure 2.4: Average Ex-post Real Rate on Deposits: Advanced Economies and
Emerging Markets, 1945-2009 (3-year moving averages, in percent)
1945-1980 1981-2009
-1.94 1.35
-4.01 2.85
Average Real Interest Rate on Deposits
Advanced economies
Emerging markets
-15.0
-10.0
-5.0
0.0
5.0
10.0
1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
`
Advanced economies
(3-year moving average)
Emerging Markets
(3-year moving average)
Financial Repression Era
Sources: International Financial Statistics, International Monetary Fund, various
sources listed in the Data Appendix, and authors’ calculations.
Notes:The advanced economy aggregate comprises: Australia, Belgium, Canada,
Finland, France, Germany, Greece, Ireland, Italy, Japan, New Zealand, Sweden,
the United States, and the United Kingdom. The emerging market group consists
of: Argentina, Brazil, Chile, Colombia, Egypt, India, Korea, Malaysia, Mexico,
Philippines, South Africa, Turkey and Venezuela. The average is unweighted and
the country coverage is somewhat spotty prior for emerging markets to 1960
42
Figure 2.5: Real Interest Rates Frequency Distributions: Advanced Economies,
1945-2009
Treasury bill rate
Discount rate
Deposit rate
1945-1980 1981-2009
0 46.9 10.5
1 percent 61.6 25.2
2 percent 78.6 36.2
3 percent 88.6 55.0
Real Interest rate on T-bills
Share of obsevations at or below:
-0.2
1.8
3.8
5.8
7.8
9.8
11.8
13.8
15.8
17.8
19.8
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11
1945-1980 1981-2009
1945-1980 1981-2009
0 41.9 11.6
1 percent 54.6 23.5
2 percent 69.4 37.0
3 percent 82.1 54.9
Real discount rate
Share of obsevations at or below:
-0.2
1.8
3.8
5.8
7.8
9.8
11.8
13.8
15.8
17.8
19.8
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9
1945-1980
1981-2009
1945-1980 1981-2009
0 58.8 24.6
1 percent 82.7 58.0
2 percent 94.7 85.4
3 percent 98.4 96.6
Real Interest rate on deposits
Share of obsevations at or below:
-0.2
4.8
9.8
14.8
19.8
24.8
29.8
34.8
39.8
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
1945-1980
1981-2009
43
Figure 2.6: Real Deposit Interest Rates Frequency Distributions: United Kingdom,
1880-2010
0.0
0.1
0.2
0.3
0.4
0.5
0.6
-4 -2 0 2 4 6 8
1945-1980
1981-2010
1880-1939
Sources: International Financial Statistics, International Monetary Fund, various
sources listed in the Data Appendix, and authors’ calculations.
44
Table 2.2: Selected Measures Associated with Financial Repression
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Argentina 1977-82, 1987, and 1991-2001, Ini-
tial liberalization in 1977 was re-
versed in 1982. Alfonsin govern-
ment undertook steps to deregulate
the ?nancial sector in October 1987,
some interest rates being freed at
that time. The Convertibility Plan
-March 1991-2001, subsequently re-
versed.
1977-82 and 1991-2001. Between 1976
and 1978 multiple rate system was uni-
?ed, foreign loans were permitted at
market exchange rates, and all forex
transactions were permitted up to US$
20,000 by September 1978. Controls
on in?ows and out?ows loosened over
1977-82. Liberalization measures were
reversed in 1982. Capital and exchange
controls eliminated in 1991 and rein-
stated on December 2001.
Australia 1980, Deposit rate controls lifted in
1980. Most loan rate ceilings abol-
ished in 1985. A deposit subsidy
program for savings banks started in
1986 and ended in 1987.
1983, capital and exchange controls
tightened in the late 1970’s, after the
move to indirect monetary policy in-
creased capital in?ows. Capital account
liberalized in 1983.
45
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Brazil 1976-79 and 1989 onwards, interest
rate ceilings removed in 1976, but
reimposed in 1979. Deposit rates
fully liberalized in 1989. Some loan
rates freed in 1988. Priority sec-
tors continue to borrow at subsidized
rates. Separate regulation on inter-
est rate ceilings exists for the micro-
?nance sector
1984, System of comprehensive foreign
exchange controls abolished in 1984. In
the 1980’s most controls restricted out-
?ows. In the 1990’s controls on in?ows
were strengthened and those on out-
?ows loosened and (once again) in 2010.
Canada 1967, with the revision of the Bank
Act in 1967, interest rates ceilings
were abolished. Further liberalizing
measures were adopted in 1980 (al-
lowing foreign banks entry into the
Canadian market) and 1986.
1970, mostly liberal regime.
Chile 1974 but deepens after 1984, com-
mercial bank rates liberalized in
1974. Some controls reimposed in
1982. Deposit rates fully market de-
termined since 1985. Most loan rates
are market determined since 1984.
1979, capital controls gradually eased
since 1979. Foreign portfolio and di-
rect investment is subject to a one year
minimum holding period. During the
1990s, foreign borrowing is subject to a
30% reserve requirement.
46
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Colombia 1980, most deposit rates at com-
mercial banks are market determined
since 1980; all after 1990. Loan rates
at commercial banks are market de-
termined since the mid-70’s. Re-
maining controls lifted by 1994 in all
but a few sectors. Some usury ceil-
ings remain.
1991, capital transactions liberalized in
1991. Exchange controls were also re-
duced. Large capital in?ows in the
early 90’s led to the reimposition of re-
serve requirements on foreign loans in
1993.
Egypt 1991, interest rates liberalized.
Heavy "moral suasion" on banks
remains.
1991, Decontrol and uni?cation of the
foreign exchange system. Portfolio
and direct investment controls partially
lifted in the 90’s.
47
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Finland 1982, gradual liberalization 1982-91.
Average lending rate permitted to
?uctuate within limits around the
Bank of Finland base rate or the av-
erage deposit rate in 1986. Later in
the year regulations on lending rates
abolished. In 1987, credit guide-
lines discontinued, the Bank of Fin-
land began open market operations
in bank CD’s and HELIBOR market
rates were introduced. In 1988, ?oat-
ing rates allowed on all loans.
1982.Gradual liberalization 1982-91.
Foreign banks allowed to establish sub-
sidiaries in 1982. In 1984, domestic
banks allowed to lend abroad and in-
vest in foreign securities. In 1987, re-
strictions on long-term foreign borrow-
ing on corporations lifted. In 1989, re-
maining regulations on foreign currency
loans were abolished, except for house-
holds. Short-term capital movements
liberalized in 1991. In the same year,
households were allowed to raise foreign
currency denominated loans.
France 1984, interest rates (except on sub-
sidized loans) freed in 1984. Sub-
sidized loans now available to all
banks, are subject to uniform inter-
est ceiling.
1986, in the wake of the dollar crisis
controls on in/out?ows tightened. The
extensive control system established by
1974, remains in place to early 80’s.
Some restrictions lifted in 1983-85. In-
?ows were largely liberalized over 1986-
88. Liberalization completed in 1990.
48
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Germany 1980, interest rates freely market de-
termined from the 70’s to today. In
the year indicated, further liberaliza-
tions were undertaken.
1974. Mostly liberal regime in the late
60’s, Germany experiments with con-
trols between 1970-73. Starting 1974,
controls gradually lifted, and largely
eliminated by 1981.
India 1992. Complex system of regulated
interest rates simpli?ed in 1992. In-
terest rate controls on D’s and com-
mercial paper eliminated in 1993 and
the gold market is liberalized. The
minimum lending rate on credit over
200,000 Rs eliminated in 1994. In-
terest rates on term deposits of over
two years liberalized in 1995.
1991. Regulations on portfolio and di-
rect investment ?ows eased in 1991.
The exchange rate was uni?ed in
1993/94. Out?ows remained restricted,
and controls remained on private o?-
shore borrowing.
Italy 1983. Maximum rates on deposits
and minimum rates on loans set
by Italian Banker’s Association un-
til 1974. Floor prices on government
bonds eliminated in 1992.
1985. Continuous operation of ex-
change controls in the 70’s. Fragile BoP
delays opening in early 80’s. Start-
ing in 1985, restrictions are gradually
lifted. All remaining foreign exchange
and capital controls eliminated by May
1990.
49
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Japan 1979. Interest rate deregulation
started in 1979. Gradual decontrol
of rates as money markets grow and
deepen after 85. Interest rates on
most ?xed-term deposits eliminates
by 1993. Non time deposits rates
freed in 1994. Lending rates mar-
ket determined in the 90’s (though
they started in 1979, both exter-
nal and domestic liberalizations were
very gradual and cautious).
1979. Controls on in?ows eased after
1979. Controls on out?ows eased in
the mid-80s. Forex restrictions eased in
1980. Remaining restrictions on cross
border transactions removed in 1995.
Korea 1991. Liberalizing measures adopted
in the early 80’s aimed at privati-
zation and greater managerial lee-
way to commercial banks. Signif-
icant interest rate liberalization in
four phases. Signi?cant interest rate
liberalization in four phases in the
90’s: 1991, 1993-94 and 1997. Most
interest rate deregulated by 1995, ex-
cept demand deposits and govern-
ment supported lending.
1991. Current account gradually liber-
alized between 1985-87, and article VIII
accepted in 1988. Capital account grad-
ually liberalized, starting in 1991, usu-
ally following domestic liberalization.
Restrictions on FDI and portfolio in-
vestment loosened in the early 90’s. Be-
ginning with out?ows, in?ows to secu-
rity markets allowed cautiously only in
the mid 90’s. Complete liberalization
planned for 2000.
50
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Malaysia 1978-1985 and 1987 onwards. Ini-
tially liberalized in 1978. Controls
were reimposed in the mid-80’s (es-
pecially 1985-87) and abandoned in
1991.
1987. Measures for freer in/out?ows of
funds taken in 1973. Further ease of
controls in 1987. Some capital controls
reimposed in 1994. Liberalization of the
capital account was more modest, and
followed that of the current account.
Mexico 1977, deepens after 1988.Time de-
posits with ?exible interest rates
below a ceiling permitted in 1977.
Deposit rates liberalized in 1988-
89. Loan rates have been liberalized
since 1988-89 except at development
banks.
1985. Historically exchange regime
much less restrictive than trade regime.
Further gradual easing between mid-
1985 to 1991. 1972 Law gave gov-
ernment discretion over the sectors in
which foreign direct investment was
permitted. Ambiguous restrictions on
fdi rationalized in 1989. Portfolio ?ows
were further decontrolled in 1989.
New
Zealand
1984. Interest rate ceilings removed
in 1976 and reimposed in 1981. All
interest rate controls removed in the
summer of 1984.
1984. All controls on inward and out-
ward Forex transactions removed in
1984. Controls on outward investment
lifted in 1984. Restrictions on foreign
companies’ access to domestic ?nancial
markets removed in 1984.
51
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Philippines 1981. Interest rate controls mostly
phased out between 1981-85. Some
controls reintroduced during the ?-
nancial crisis of 1981-87. Cartel-like
interest rate ?xing remains preva-
lent.
1981. Foreign exchange and invest-
ment controlled by the government in
the 70’s. After the 1983 debt crisis the
peso was ?oated but with very limited
interbank forex trading. O?-?oor trad-
ing introduced in 1992. Between 1992-
95 restrictions on all current and most
capital account transactions were elim-
inated. Outward investment limited to
$6 mill/person/year
South
Africa
1980. Interest rate controls re-
moved in 1980. South Africa Reserve
Bank relies entirely on indirect in-
struments. Primary, Secondary and
Interbank markets active and highly
developed. Stock Exchange modern
with high volume of transactions.
1983. Partially liberalized regime. Ex-
change controls on non-residents abol-
ished in 1983. Limits still apply on pur-
chases of forex for capital and current
transactions by residents. Inward in-
vestment unrestricted, outward is sub-
ject to approval if outside Common
Monetary Area. Several types of ?nan-
cial transactions subject to approval for
monitoring and prudential purposes.
52
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Sweden 1980. Gradual liberalization in the
early 80’s. Ceilings on deposit rates
abolished in 1978. In 1980, controls
on lending rates for insurance compa-
nies were removed, as well as a tax on
bank issues of certi?cate of deposits.
Ceilings on bank loan rates were re-
moved in 1985.
1980. Gradual liberalization between
1980-90. Foreigners allowed to hold
Swedish shares in 1980. Forex controls
on stock transactions relaxed in 1986-
88, and residents allowed to buy for-
eign shares in 1988-89. In 1989 foreign-
ers were allowed to buy interest bear-
ing assets and remaining forex controls
were removed. Foreign banks were al-
lowed subsidiaries in 1986, and opera-
tion through branch o?ces in 1990.
53
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Thailand 1989. Removal of ceilings on in-
terest rates begins in 1989. Ceil-
ing on all time deposits abolished
by 1990. Ceilings on saving de-
posits rates lifted in 1992. Ceilings
on ?nance companies borrowing and
lending rates abolished in 1992.
1991. Liberalized capital movements
and exchange restrictions in successive
waves between 1982-92. Article VIII
accepted and current account liberal-
ization in 1990, capital account liber-
alization starting in 1991. Aggressive
policy to attract in?ows, but out?ows
freed more gradually. Restrictions on
export of capital remain. The reserve
requirement on short-term foreign bor-
rowing in 7%. Currency controls intro-
duced in May-June 1997. These con-
trols restricted foreign access to baht in
domestic markets and from the sale of
Thai equities. Thailand relaxed limits
on foreign ownership of domestic ?nan-
cial institutions in October of 1997.
54
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
Turkey 1980-82 and 1987 onwards. Liberal-
ization initiated in 1980 but reversed
by 1982. Interest rates partially
deregulated again in 1987, when
banks were allowed to ?x rates sub-
ject to ceilings determined by the
Central Bank. Ceilings were later re-
moved and deposit rates e?ectively
deregulated. Gold market liberalized
in 1993.
1989. Partial external liberalization in
the early 80’s, when restrictions on in-
?ows and out?ows are maintained ex-
cept for a limited set of agents whose
transactions are still subject to con-
trols. Restrictions on capital move-
ments ?nally lifted after August 1989.
United
Kingdom
1981. The gold market, closed in
early World War II, reopened only in
1954. The Bank of England stopped
publishing the Minimum Lending
Rate in 1981. In 1986, the govern-
ment withdrew its guidance on mort-
gage lending.
1979. July 79: all restrictions on
outward FDI abolished, and outward
portfolio investment liberalized. Oct
1979: Exchange Control Act of 1947
suspended, and all remaining barriers
to inward and outward ?ows of capital
removed.
55
Table 2.2 – Continued from previous page
Country Domestic Financial Regulation Cap. Account - Exchange Restrictions
United
States
1982. 1951-Treasury accord/debt
conversion swapped marketable
short term debt for nonmarketable
29-year bond. Regulation Q sus-
pended and S&Ls deregulated in
1982. In 1933, President Franklin
D. Roosevelt prohibits private hold-
ings of all gold coins, bullion, and
certi?cates. On December 31, 1974,
Americans are permitted to own
gold, other than just jewelry.
1974. In 1961 Americans are forbid-
den to own gold abroad as well as at
home. A broad array of controls were
abolished in 1974.
Venezuela 1991-94 and 1996 onwards. Interest
rate ceilings removed in 1991, reim-
posed in 1994, and removed again in
1996. Some interest rate ceilings ap-
ply only to institutions and individu-
als not regulated by banking author-
ities (including NGOs).
1989-94 and 1996 onwards. FDI regime
largely liberalized over 1989-90. Ex-
change controls on current and capi-
tal transactions imposed in 1994. The
system of comprehensive forex controls
was abandoned in April 1996. Controls
are reintroduced in 2003.
Sources: Reinhart and Reinhart (2011) and sources cited therein. See
also FOMC minutes, March 1-2, 1951 for US debt conversion particulars,http://www.micro?nancegateway.org/p/site/m/template.rc/1.26.9055/ on current ceilings
and related practices applied to micro?nance, and National Mining Association (2006) on
measures pertaining to gold.
Notes: Liberalization year(s) in italics.
56
Table 2.3: Incidence and Magnitude of the Liquidation of Public Debt: Selected
Countries, 1945-1980
Country Period Share of Liquidation Negative Real Interest Rate
Liquidation
Years
Years Liquidation Years
Average Min(Year)
(1) (2) (3) (4) (5) (6)
Argentina 1942-
1980
92.3 1944-1952,1954-
1980
21.4 72.3 (1976)
Australia 1945-
1980
52.8 1947-1953, 1956-
1957, 1971-1980
4.6 15.1 (1952)
Belgium
1
1945-
1974
48 1945-
1948,1951,1963,1969-
1974
4.2 9.6 (1974)
India 1949-
1980
53 1949,1951,1957,1959-
1960,1964-
1968,1970,1972-
1975,1977,1980
5.4 17.4 (1974)
Ireland 1960-
1990
65.4 1961-1965, 1967-
1977, 1979-1983
3.4 12.7 (1975)
Italy
2
1946-
1980
48.6 1946-1947,1950-
1951,1962-1964,
1970, 1972-1980
6 27.63
(1947)
3
South
Africa
1945-
1980
47.2 1947-1949, 1951-
1953, 1955,
1958, 1972-1980
3 6.8 (1975)
Sweden 1945-
1990
47.8 1947-1948,
1951-1952, 1956-
1958,1960,1962,1965-
1966,1970-
1978,1980-1981
2.6 11.9 (1951)
United
Kingdom
1945-
1980
58.3 1948-
1953,1955-1956,
1958,1962,1965,
1969, 1971-1977,
1979-1980
3.5 10.9 (1975)
United
States
1945-
1980
50 1945-1948,1950-
1951,1956-
1957,1968-
1970,1973-
1975,1977-1980
3.5 13.7 (1946)
Notes: Share of liquidation years is de?ned as the number of years during which the
real interest rate on the portfolio is negative divided by the total number of years
as noted in column (2). The real interest rate is calculated as de?ned in equation (1).
1
No data available for 1964-1968
2
The average and minimum real interest rate during liquidation years were calculated
over the period 1945-1970 to exclude war years
3
In 1944, the negative real return was 82.3 percent and in 1945 it was 46.6 percent.
57
Table 2.4: Incidence of Liquidation Years for Di?erent Real Interest Rate Thresh-
olds: Selected Countries, 1945-1980
Country Period
Share of Years with Real Interest Rate below:
0 percent 1 percent 2 percent 3 percent
Argentina 1942-1980 92.3 92.3 94.9 94.9
Australia 1945-1980 52.3 63.9 83.3 94.4
Belgium
1
1945-1974 48.0 65.4 72 80
India 1949-1980 53.0 62.5 71.9 78.1
Ireland 1960-1990 65.4 74.2 77.4 80.6
Italy
2
1946-1980 48.6 62.9 65.7 82.9
South Africa 1945-1980 47.2 61.1 77.8 97.2
Sweden 1945-1990 47.8 52.2 69.6 82.6
United Kingdom 1945-1980 58.3 72.2 86.1 97.2
United States 1945-1980 50.0 55.6 86.1 94.4
Notes: Share of liquidation years is de?ned as the number of years during which the
real interest rate on the portfolio is negative divided by the total number of years
as noted in column (2). The real interest rate is calculated as de?ned in equation (2.1).
1
No data available for 1964-1968
2
The average and minimum real interest rate during liquidation years were calculated
over the period 1945-1970 to exclude war years.
Table 2.5: Government Revenues (interest cost savings) from the "Liquidation Ef-
fect:" per year
Country Period
Benchmark Measure Alternative Measure
%
GDP
% Tax
Rev-
enues
%
GDP
% Tax
Rev-
enues
Argentina 1942-1980 3.1 38.3 3.1 39
Australia 1945-1980 3.3 12.9 n.a. n.a.
Belgium 1945-1974 2.5 18.6 3.5 23.9
India 1949-1980 1.5 27.2 1.5 27.2
Ireland 1960-1990 1.8 7.9 n.a. n.a.
Italy 1946-1980 1.6 24.5 1.6 26.5
South Africa 1945-1980 1.3 8 n.a. n.a.
Sweden 1945-1990 0.8 4.4 1.3 4.4
United Kingdom 1945-1980 3.0 18.8 3.1 19.6
United States 1945-1980 2.3 13.4 2.7 15.9
Sources: See data appendix and sources cited therein and authors’ calculations.
58
Table 2.6: Debt Liquidation through Financial Repression: Selected Countries,
1945-1955
Country
Public debt/GDP Annual average: 1946-1955
1945 1955 (actual) 1955 without "?nancial repression in?ation
repression revenue"/GDP
savings" (est.)
4
Australia 143.8 66.3 195.7 7.4 8.6
Belgium
1
112.6 63.3 130.1 5.7 3.4
Italy
2
66.9 38.1 120.2 13.3 9.6
Sweden 52 29.6 72.6 5.2 3.8
United Kingdom
3
215.6 138.2 233.8 2.6 3.9
United States 116 66.2 143.8 5.6 4.1
1
The debt-to-GDP ratio corresponds to 1946
2
Italy was in default on its external debt 1940-1946
3
The savings from ?nancial repression are a lower bound, as we use the "o?cial" consumer
price index for this period in the calculations and in?ation is estimated to have been substantially
higher than the o?cial ?gure (see for example Friedman and Schwartz, 1982).
4
The simple cumulative annual savings without compounding.
Notes: The peaks in debt/GDP were: Italy 129.0 in 1943; United Kingdom 237.9 in 1947;
United States 121.3 in 1946. An alternative interpretation of the ?nancial repression revenue
is simply as savings in interest service on the debt.
Sources: See data appendix B and sources cited therein and authors’ calculations; for
debt/GDP see Reinhart (2010) and Reinhart and Rogo? (2011b).
59
Table 2.7: In?ation Performance during Major Domestic Public Debt Reduction
Episodes: 28 Countries, 1790-2009
Country
Major Debt Reduction Episodes* Full Sample
Dates
In?ation In?ation
Average Median Average Median
Argentina 1900-
1902,1990,2006-
2007
479.8 8.2 82.5 8.6
Australia 1948,1949-1953 10.3 9.3 3.0 2.5
Belgium 1925-28, 1949 10.7 12.8 2.0 1.9
Brazil 1990-1992,1995-
1996
898.2 980.2 111.3 11.3
Canada 1948,1949-1952 7.3 5.3 3.2 2.5
Chile 1993-1997, 2004-
2007
7.7 6.1 17.7 5.5
Colombia 2008, 2009 8.5 6.3 12.6 10.8
Egypt 2008 12.0 8.6 11.7 9.9
Finland 1946-1949 34.5 24.9 10.4 3.9
France 1924, 1926-1927,
1938
11.1 12.6 6.4 2.7
Germany 1922, 1923 5555049529.6 1764.7 231460401.3 2.3
Greece 1925-1927 23.7 12.8 8.0 5.1
India 1958, 1996, 2006 7.1 6.2 6.6 6.2
Ireland 1972, 1982, 1998 9.8 8.6 5.9 3.7
Italy 1945, 1946-1948 106.7 44.3 10.6 2.6
Japan 1898, 1912-1913 7.6 6.7 3.6 2.6
Korea 1986 2.5 2.5 6.3 4.6
Malaysia 1995 8.4 8.8 6.9 5.4
Mexico 1991, 1992, 1993 18.9 20.0 13.3 5.6
New Zealand 1935-1937, 1950-
1952
4.9 5.3 4.2 2.8
Phillipines 1998, 2007-2008 7.2 7.7 7.7 6.2
South Africa 1935, 1952, 1981,
2001-2002
7.0 6.6 5.8 4.9
Sweden 1948, 1952, 1989,
2001-2003, 2009
4.7 3.2 4.4 3.2
Thailand 1989-1990 4.4 4.6 4.8 3.8
Turkey 1943, 2006-2008 23.2 9.2 25.3 9.7
UK 1836, 1846, 1854,
1936, 1940, 1948-
1950,1951-1954
4.7 3.7 2.7 1.8
US 1794-1796, 1881-
1882, 1948-1952,
1953, 1957, 1966
4.0 2.6 1.6 1.7
Venezuela 1989, 1997-1998,
2006-2007
41.6 29.5 11.4 5.8
* Consists of the 10% largest reductions for each country.
Notes: shown in italics the year of the major reduction
60
Chapter 3
Debt and In?ation during a Period of Financial Repression
3.1 Introduction
In many countries, the current ratio of government debt to GDP is at histor-
ically high levels. The 2008 ?nancial crisis has created a situation where several
countries are at risk of defaulting on their debt, and many more are struggling with
the economic and political changes needed to reduce their debt to more sustain-
able levels. Broadly speaking, the di?erent alternatives available to governments to
reduce their debt burdens are: (i) growth; (ii) ?scal adjustment (i.e., increases in
taxes and reductions in government spending); (iii) outright default or restructur-
ing; and (iv) in?ation (via in?ation surprises or a combination of ?nancial repression
and in?ation). While there is some empirical evidence on the consequences of de-
fault (Reinhart and Rogo?, 2009; Sturzenegger and Zettlelmeyer, 2006; Borensztein
and Panizza, 2006) and on the potential for ?scal austerity/restructuring (Alesina
and Ardagna, 2010; Perotti, 2011), there is limited empirical evidence on whether
in?ation can reduce the real value of debt.
Revisiting similar episodes in the past can be useful for understanding the
possible courses of action currently available to governments. In this paper, I explore
the relationship between in?ation and debt in the years after the Second World War
61
in 12 countries with very di?erent economic characteristics.
1
Many governments
had high debt levels at the end of the Second World War, which then declined over
the following decades. I construct a detailed database of government debt portfolios
spanning three or more decades for each country and analyze the role of in?ation
in reducing real debt, examine the circumstances under which this occurs, and the
implications of this for both the government and investors. The primary focus of
the paper is to understand whether this is an empirically important phenomenon,
rather than the desirability or optimality of using in?ation to erode the real value
of debt.
To understand the magnitude and nature of the e?ects, it is important to
identify the channels through which in?ation can have an e?ect on debt. The pre-
vious literature on how in?ation may reduce debt is largely theoretical, and focused
on the role of unanticipated in?ation in reducing government debt. Several papers
have looked at the time inconsistency problem present when debt is denominated in
nominal terms, where governments may be tempted to use unanticipated in?ation
to reduce the real value of their debt (Barro and Gordon, 1983; Grossman and Van
Huyck, 1984, 1985; Grossman, 1987, 1988). In this paper, I identify three channels
through which in?ation can have an e?ect on debt: unanticipated in?ation, in?a-
tion in combination with ?nancial repression, and via changes in the market value
of debt.
The term "?nancial repression" is used to refer to a situation characterized by:
1
The countries in the sample are: Argentina, Australia, Belgium, France, India, Ireland, Italy,
Japan, South Africa, Sweden, the United Kingdom, and the United States.
62
i) numerous policies and regulations which introduce frictions in ?nancial markets,
and ii) large participation of nonmarket players. The list of policies is large; some
examples of ?nancial repression are: ceilings on interest rates, directed lending, cap-
ital controls.
2
The policies that will be particularly relevant for this paper are those
which create captive investors for government debt, and hence allow the government
to issue debt at a rate below what the market would charge absent any restrictions.
Being able to issue debt at a below market interest rate represents a saving in in-
terest payments for the government. When combined with an in?ation rate above
the nominal interest rate, this leads to negative real interest rates that e?ectively
reduce government debt. This mechanism can be present even when in?ation is fully
anticipated.
There are several channels through which in?ation can a?ect debt: unantici-
pated in?ation, ?nancial repression combined with in?ation, and via changes in the
market value of debt. I develop a conceptual framework in order to understand how
those channels can be measured separately and together. While it is not possible
to directly observe in?ation expectations and quantify the e?ect of ?nancial repres-
sion on market interest rates, I show that the net e?ects of these channels must be
large whenever real interest rates are negative; that is, when real interest payments
are negative (which can be thought of as a revenue for the government). My pri-
mary empirical strategy is to focus on years where this occurs, which are labeled
liquidation years.
3
2
For a more detailed de?nition, see Appendix ?? and Reinhart and Sbrancia (2011).
3
By focusing on years with negative real interest rates, the estimates may be overestimating
the actual e?ect if for instance the "natural" interest rate in the economy was negative. On the
other hand, the estimates may be underestimating the actual e?ect whenever the "natural" interest
63
On average, real interest rates on the overall portfolios of domestic government
debt were negative in half of the years in the sample.
4
The predominant pattern that
emerges across countries is a high incidence of liquidation years in the period imme-
diately after the end of WWII, and again during the 1970s. In the United States,
50 percent of the years between 1945 and 1980 were liquidation years. Estimates
of the implicit revenues for the governments in the 12 countries, which are labeled
liquidation revenues, average between two and three percent of GDP in liquidation
years.
5
I show that the e?ects are similar whether I measure debt at face value or
at market value, which suggests that changes in the market value of debt are not
responsible for these large e?ects. I conduct several exercises to disentangle the rel-
ative contributions of the other two channels, unanticipated in?ation and ?nancial
repression. First, I estimate in?ation expectations to see whether in?ation surprises
can account for the liquidation years. Across the 12 countries, only 15 percent of
liquidation years are ones where there is an in?ation surprise. This suggests that,
for the period under consideration, ?nancial repression (combined with in?ation) is
important for explaining the high incidence of liquidation years. In a second ex-
ercise, I test whether the overall results are biased by bonds issued prior to 1945.
The thinking behind doing this is that, if the overall results are primarily due to
unanticipated in?ation, then newly- issued bonds should re?ect higher in?ation ex-
rate is positive.
4
Historically most of the domestic debt has been denominated in local currency. Notable
exceptions are Mexican Tesobonos in the 1980’s and Brazil dollar-denominated bonds a decade
after.
5
These revenues estimates do not includes the revenues from seigniorage. A comparison to
seigniorage revenues is performed in section 3.4.
64
pectations. For bonds issued after 1945, I ?nd negative real interest rates were as
common as in the full sample of bonds. This failure of markets to respond provides
further evidence that there was ?nancial repression during this period.
To put the magnitude of the liquidation revenues into perspective, I compare
them to in?ation tax revenues.
6
The liquidation e?ect revenues are consistently
larger than in?ation tax revenues at the beginning of the sample period, when debt
levels were high. In some countries, such as the United Kingdom, the liquidation
e?ect revenues are larger than in?ation tax revenues throughout the whole sample
period. This ?nding may help to explain why Reinhart and Rogo? (2009) ?nd that
the debt stock was signi?cantly larger than the money stock in many episodes of
high and hyperin?ation, as the gains from in?ating away debt may be larger than
in?ation tax revenues during these periods.
I also gain a greater understanding of the role of in?ation by examining the
cumulative e?ect of in?ation on the stock of debt under plausible alternative in?a-
tion paths. I ?nd that these cumulative e?ects are large. For example, if annual
in?ation rate had remained constant at two percent throughout the 1945-1980 pe-
riod, the debt-to-GDP ratio in 1980 would have been 40 percentage points higher
in the United States, 167 percentage points higher in the United Kingdom, and 81
percentage points higher in Australia.
An important ?nding of the paper is that in?ation need not be particularly
high in order to obtain a sizable reduction of the debt. Except in Argentina and Italy,
6
The term "in?ation tax” refers to the component of seigniorage that would be collected in
steady state.
65
median in?ation during liquidation years is below 10 percent. Average in?ation is
four percentage points higher than median in?ation rates over the 1930-2010 period.
A multivariate analysis is conducted to understand how country characteris-
tics and di?erent factors a?ect the incidence of the liquidation e?ect. I ?nd that
two variables strongly and positively correlated with liquidation years are interest
payments (as a proportion of GDP) and the size of the de?cit relative to GDP. The
signi?cance of these ?scal variables points at the presence of important links be-
tween ?scal and monetary policies. I ?nd that neither the share of short term debt
nor central bank independence have have strong relationships with the incidence of
liquidation years.
The last section of the paper focuses on the implication for investors. If Trea-
sury bonds are used as a benchmark in the valuation of other assets, then it is
important to understand how the presence of ?nancial repression a?ected ?nancial
markets more broadly.
7
I examine returns on bonds and other ?nancial assets for
more than 100 years, and ?nd that the returns on bonds were low during the 1945-
1980 period relative to other periods. Furthermore, for all of the countries in the
sample the return from investing in stocks during the 1945-1980 period was much
higher than the return from investing in government bonds. These di?erences do
not appear to be explained by stock return volatility or risk. The equity premium
is shown to be relatively high during this period. For instance, the average equity
premium - calculated as the excess return of stocks on T-Bills over rolling 30-year
7
For instance, Treasury bonds serve as a benchmark in the valuation of corporate bonds
(Crabbe and Fabbozi, 2002).
66
periods - in the US averaged 8.3 percent over 1945-1980, compared to an equity
premium of 4.4 percent in the other years between 1870 and 2010. These patterns
are consistent with the presence of ?nancial repression keeping returns on bonds
arti?cially low over the 1945-1980 period.
The paper contributes to several literatures. First, it extends the results of
Reinhart and Sbrancia (2011) who establish the importance of ?nancial repression
as a restructuring mechanism for government debt. In their paper, the authors
provide ?rst pass estimates which show that the e?ects are quantitatively important.
This paper complements the results of Reinhart and Sbrancia (2011) in three main
ways. First, I examine the di?erent channels through which in?ation can reduce
government debt and consider how to think of ?nancial repression as a restructuring
mechanism. Second, through di?erent empirical exercises, I show that ?nancial
repression in combination with in?ation is the most important channel through
which debt was reduced. Finally, the paper looks at the key features of this period
of ?nancial repression and how the returns of other assets were a?ected.
A related literature looks at the implications of in?ation on a government’s re-
ported ?scal position, and particularly how in?ation a?ects measures of ?scal de?cit
(Siegel,1979; Tanzi et al., 1987; Persson et al.,1996). The results in this paper
provide empirical support for the theoretical literature which has argued that gov-
ernment debt could not be reduced systematically by unanticipated in?ation (Calvo
and Guidotti, 1993). The results also point to ?nancial repression as an important
source of revenue, which could be important for rationalizing why countries seem
to set in?ation rates above what would be optimal if they were just maximizing
67
seigniorage revenues (Calvo and Leiderman, 1992).
The paper also adds to the literature on ?nancial repression, which has pri-
marily focused on the economic growth implications in emerging economies (Mc
Kinnon, 1973; Shaw, 1973; King and Levine, 1993; Fry, 1997). I show that ?nan-
cial repression can be important in advanced economies, and may generate revenues
for governments by reducing their debt burdens.
8
This complements work by Gio-
vannini and de Melo (1991), who measure the revenues from ?nancial repression
by comparing the interest paid by the government on its external and domestic
debt, and Aizenman and Guidotti’s (1990) theoretical work analyzing under what
conditions it may be optimal to impose capital controls.
Finally, the paper contributes to growing e?orts to use historical documents
and government reports to understand key issues in international ?nance. Long time
series on public debt were uncommon until the publication of Reinhart and Rogo?
(2008, 2009). I construct a database on the domestic debt portfolios of 12 countries
for three or more decades after the end of WWII. The database contains a detailed
description of the di?erent instruments that constitute the stock of debt in a given
year together with their coupon rates, maturity date, outstanding amount and in
some cases prices at di?erent points in time.
The next section presents the conceptual framework. Section 3.3 describes the
data and the empirical measures constructed. In Section 3.4, I present the results
for the liquidation e?ect from the perspective of the government. In Section 3.5,
8
For evidence on the presence of ?nancial repression in these countries, please see the Appendix
?? in this paper and Reinhart and Sbrancia (2011).
68
a broader analysis of ?nancial markets during the period of ?nancial repression is
provided, together with its implications for investors. Section 3.6 concludes.
3.2 Conceptual Framework
The ?rst step is to understand the channels through which in?ation can a?ect
the value of debt. When discussing how debts may be "in?ated away," researchers
usually think of a higher than expected in?ation rate eroding the real value of the
debt. However, ?nancial repression (by which governments issue debt at below mar-
ket interest rates) is an additional and important channel through which in?ation
can reduce government debt. The primary objective of this section is to separate, at
least conceptually, the contribution of unanticipated in?ation and ?nancial repres-
sion in the liquidation of government debt.
The consolidated budget constraint for the government is obtained by com-
bining the budget constraints of the ?scal and monetary authorities. This budget
constraint makes explicit the tight linkage that exists between monetary and ?scal
policy. In real terms the consolidated budget constraint is given by:
g
t
+
1 + i
t?1
1 + ?
t
b
t?1
= ?
t
+ b
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.1)
On the left side are the expenditures in a given year: government spending
(g
t
) and the real interest payments on the real stock of debt, which depends on
the nominal interest rate set in the previous period (i
t?1
), the in?ation rate in the
69
current period (?
t
), and the debt from the previous period (b
t?1
).
9
The real interest
rate paid on the stock of debt issued in the previous period is an ex post real interest
rate, since it is determined by the realized rate of in?ation. The right side contains
the sources of income: revenues (?
t
), newly issued real debt (b
t
), and the seigniorage
revenues from printing money, where h
t
is the real monetary base.
10
While in?ation
a?ects seigniorage revenues as well as other items of the budget constraint, I ignore
those e?ects as the focus on the paper is on sources of revenue which have a direct
impact on a government’s real debt payments.
11
These e?ects are compared to
seigniorage revenues in Section 3.4.
The budget constraint can be re-written in terms of the ex post real interest
rate (r
P
t
) as follows:
g
t
+ (1 + r
P
t
)b
t?1
= ?
t
+ b
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.2)
Two additional de?nitions of interest rates are required to capture the role of
unanticipated in?ation and ?nancial repression. The ?rst one is the ex ante real
interest rate. This is the interest rate that is expected to be earned in period t, as
9
Expressing the budget in terms of a one-period bond simpli?es the notation without changing
the implications that would be derived from explicitly considering a richer maturity structure.
10
Seigniorage is the change in the nominal monetary base relative to the previous period, and
divided by the current price level. It arises from two sources as shown below:
H
t
?H
t?1
P
t
= (h
t
?h
t?1
) +
?
t
1 +?
t
h
t
The ?rst component of seigniorage comes from changes in the real stock of monetary base. The
second comes from a depreciation in the outstanding stock of real balances, and is sometimes
referred to as in?ation tax. In steady state, only the second component will be positive.
11
See Persson, Persson and Svensson (1996) for a study on the overall ?scal gains from an
increase in the in?ation rate in Sweden.
70
of period t ?1. It is determined by the nominal interest rate i
t?1
and the expected
in?ation rate ?
e
t
.
The second interest rate de?nition identi?es the e?ect of ?nancial repression.
The free market interest rate (i
F
t?1
) is the interest rate that would be observed in
the absence of ?nancial frictions. If the government issues debt at a below-market
interest rate, then i
F
t?1
> i
t?1
.
The three relevant interest rates are:
1 + r
P
t
=
1 + i
t?1
1 + ?
t
Ex post real interest rate (3.3)
1 + r
A
t
=
1 + i
t?1
1 + ?
e
t
Ex ante real interest rate (3.4)
1 + r
F
t
=
1 + i
F
t?1
1 + ?
e
t
Ex ante free market real interest rate (3.5)
These terms can be incorporated into the government budget constraint. After
some algebraic manipulations, the following equation is obtained:
12
g
t
+(1+r
F
t
)b
t?1
? (1 + r
A
t
)
?
t
??
e
t
1 + ?
t
b
t?1
. ¸¸ .
Unanticipated
In?ation E?ect (A)
?
i
F
t?1
?i
t?1
1 + ?
e
t
b
t?1
. ¸¸ .
Financial
Repression E?ect (B)
= ?
t
+b
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.6)
The "unanticipated in?ation e?ect" is the di?erence between realized and ex-
pected in?ation multiplied by the real cost of previous period stock of debt, while
12
The term,
1+it?1+t?1
1+?
e
t
where
t?1
= i
F
t?1
?i
t?1
, is added and subtracted from the left-hand
side of equation (3.1).
71
the "?nancial repression e?ect" is the di?erence between the free market and actual
nominal interest rate multiplied by the real stock of debt from the previous period.
To better understand this equation, note that if there were no ?nancial frictions
that would cause i
t?1
to be di?erent from i
F
t?1
, and if actual in?ation was equal to
expected in?ation, then the last two terms on the left side would be equal to zero.
In this case, (1 + r
F
t
) would be both the ex ante and ex post real interest rate,
and there would be no savings in interest payments for the government from either
source.
Whenever the actual in?ation rate is above the expected in?ation rate, the
unanticipated in?ation e?ect will be positive and the government will save on interest
payments by the amount given by this term.
13
The opposite is true when expected
in?ation is higher than the actual in?ation rate. The ?nancial repression e?ect will
be positive and represent savings for the government when the nominal interest rate
does not re?ect the true cost of borrowing for the government, so that the actual
nominal interest rate is below the free market interest rate.
Both e?ects can be present at the same time. In this case, ?nancial repression
has an indirect e?ect on the size of the unanticipated in?ation e?ect. This indirect
e?ect comes from the fact that the ex ante real interest rate (r
A
t
) will be lower than
what it would be in the absence of frictions. In other words, for a given
?t??
e
t
1+?t
, the
savings from unanticipated in?ation will be lower in the presence of ?nancial repres-
sion. This interaction is potentially important when modeling in?ation expectations
13
When the only friction is the di?erence between actual and expected in?ation, it follows that
r
F
t
= r
A
t
72
and in?ation surprises, although it will not be separately estimated in this paper.
The Consolidated Budget Constraint at Market Value
Up to this point debt has been expressed at face value. Expressing the consol-
idated budget constraint with debt at market value allows the identi?cation of an
additional e?ect that comes from changes in the market value of debt. This will be
important for understanding returns for investors, and the response of the market in
the presence of the e?ects. In a well-functioning ?nancial system, changes in in?a-
tion expectations are going to a?ect expected returns and should be re?ected in the
price of government bonds. The (real) market value consolidated budget constraint
is given by:
g
t
+ i
t?1
B
t?1
P
t
+
P
B
t
B
t?1
P
t
= ?
t
+
P
B
t
B
t
P
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.7)
Where P
B
t
stands for the price of the debt at time t, and B
t
is the nominal
amount outstanding of debt at time t. The real market value of debt is de?ned as
ˆ
b
t
=
P
B
t
Bt
Pt
. Using these de?nitions, equation (3.7) can be re-written as:
g
t
+
i
t?1
1 + ?
t
ˆ
b
t?1
P
B
t?1
+
ˆ
b
t?1
1 + ?
t
+
P
B
t
?P
B
t?1
P
B
t?1
ˆ
b
t?1
1 + ?
t
= ?
t
+
ˆ
b
t
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.8)
Note that this equation is similar to equation (3.1), apart from the extra term
that captures changes in the market value of debt, where
P
B
t
?P
B
t?1
P
B
t?1
is the rate of
change in the market value of debt.
14
Following some algebraic manipulations, a
14
Changes in the market price of debt a?ect only the principal and not the interest payment
73
version of equation (3.6) is obtained:
g
t
+ i
F
t?1
b
t?1
? i
t?1
?
t
??
e
t
1 + ?
t
b
t?1
. ¸¸ .
Unanticipated
In?ation E?ect
?
i
F
t?1
?i
t?1
1 + ?
e
t
b
t?1
. ¸¸ .
Financial
Repression E?ect
?
P
B
t?1
?P
B
t
P
B
t?1
ˆ
b
t?1
1 + ?
t
. ¸¸ .
Valuation E?ect
=
?
t
+
_
ˆ
b
t
?
ˆ
b
t?1
1 + ?
t
_
+
_
h
t
?
h
t?1
1 + ?
t
_
(3.9)
When using debt at market values it is possible to distinguish between three
e?ects: a valuation e?ect, the unanticipated in?ation e?ect, and the ?nancial repres-
sion e?ect. The unanticipated in?ation and ?nancial repression e?ects are identical
to the corresponding terms in equation (3.6), and can be interpreted in the same
way. The valuation e?ect has an easy interpretation: when the prices of the govern-
ment bonds go down there is an implicit capital gain for the government due to a
lower value of its liabilities. On the other hand, when prices are increasing there is
an implicit capital loss due to an increase in the value of the government liabilities.
This does not represent a change in the cash payments the government makes, but
a change in the market value of its debt.
Measurement Issues
Equation (3.6) identi?es the di?erent elements required to estimate the sources
of interest payment savings for the government at face value. Similarly, equation
term
_
ˆ
bt?1
P
B
t?1
= b
t?1
_
.
74
(3.9) identi?es the elements required to estimate the sources of interest payment
savings for the government at market value. A central challenge is that, in both
cases, it is not possible to directly observe in?ation expectations and free market
interest rates.
The ?rst approach to dealing with this is to focus on instances when the
net e?ect of in?ation expectations and ?nancial repression is so large that one or
both must be present. When real interest payments are negative, they constitute
a revenue rather than an expenditure for the government. In equation (3.2), when
debt is at face value, this will be the case when r
P
t
< 0. In these years, the sum of
the unanticipated in?ation e?ect and the ?nancial repression e?ect is large enough
to outweigh the free market interest payments, which is given by the second term
on the left hand side of the equation. Given that government debt is liquidated in
any year where real interest payments are negative, those years will be de?ned as
liquidation years.
15
I will refer to the e?ect of in?ation on debt as the liquidation
e?ect and the revenues for the government from this source as liquidation revenues,
which will be calculated in any single year as:
Liquidation Revenues = Negative Real Interest Rate×
Outstanding Stock of Domestic Debt
15
Reinhart and Rogo? (2009) refer to the use of in?ation to erode the value of government
debts as default via in?ation. I use the term "liquidation e?ect” to allow for the possibility that
the in?ation was not caused by a deliberate government action.
75
This de?nition provides a lower bound for the e?ect of in?ation on debt, since
the combined e?ect of unanticipated in?ation and ?nancial repression needs to out-
weigh free market interest payments.
16
It is possible however that the estimates
may be overestimating the actual e?ect. This would be the case for instance if the
"equilibrium interest rate" was negative.
17
There is also a possible source of mea-
surement error arising from the fact that ?nancial repression may have had an e?ect
on the size of the debt stock (the tax base). The relevant question is what would
have been the size of the debt stock absent ?nancial repression.
The measure captures the net e?ect of in?ation expectations and ?nancial
repression on domestic debt, when these are su?ciently large to satisfy the de?nition
(assuming a positive natural interest rate). In order to gain some understanding of
the relative contribution of each, in?ation expectations are estimated. This is done
in Section 3.4 and provides a range for the contribution of in?ation expectations.
By considering the remainder of the total e?ect as due to ?nancial repression, this
approach also provides some understanding of the importance of ?nancial repression.
Similar approaches are used when it comes measuring debt at market value. As
shown in equation (3.8), real interest payments will be negative when
i
t?1
+(P
B
t
?P
B
t?1
)
P
B
t?1
(1+?t)
<
0. The only di?erence is the additional term that re?ects changes in the market value
of government debt. This component turns out to be small, so that the face and
market value measures produce similar results. To the extent that the term re?ects
changes in expectations for future in?ation, this term provides additional informa-
16
Additional assumptions are used in Section 3.4 to understand the likely frequency of instances
where these e?ects are positive but not large enough to satisfy the de?nition adopted here.
17
Available estimates for the US and other countries show a positive and above 1% equilibrium
interest rate from 1960 (Laubach and Williams, 2003; Benati and Vitale, 2007).
76
tion about the role of in?ation expectations for the countries and periods under
study.
18
3.3 Empirical Measures and Data
This section presents the empirical measures constructed to measure the e?ect
of in?ation on government debt, and the data used to calculate those measures. De-
tailed information on the overall portfolio is necessary to obtain accurate estimates
because there is no single interest rate that is going to re?ect the ?nancing cost of
the government.
3.3.1 Empirical Measures
Two measures of nominal interest rates are constructed, one corresponding to
when debt is expressed at face value and the other when debt is at market value. The
face value measure is the Contractual Interest Rate (CIR), which is the coupon rate
at which the bond was issued. From the perspective of the government it represents
the annual interest cost of each security. The CIR is consistent with the accounting
method used by the government.
The market value measure is the Holding Period Return (HPR). The HPR is
the nominal return for a security bought at the beginning of a year and sold at the
18
This term could also re?ect changes in the general economic environment. Siegel (1979)
decomposes the changes in the real market value of debt coming from changes in the general price
level and changes in the price of government bonds. He ?nds that bond price changes are more
volatile but account for a small fraction of the total changes in the real market value of debt. An
additional point made by Siegel is that changes in the general environment will a?ect both the
liabilities and assets side of a government’s balance sheet, which may leave the wealth position of
the government largely unchanged.
77
end, where a year corresponds to the ?scal year of each country.
19
The HPR re?ects
changes in the market interest rate and is the proper way to measure the (before
tax) return on government bonds for investors.
HPR
t
=
C
t
+
_
P
B
t
?P
B
t?1
_
P
B
t?1
(3.10)
The Holding Period Return comes from two sources: the annual interest pay-
ments and any capital gains (losses) coming from increases (decreases) in the price
of the bond. Apart from minor di?erences in notation, this expression is the same
than the one obtained in the conceptual framework.
Both the CPR and HPR are nominal measures. Their real counterparts are
obtained using in?ation data from each country’s Consumer Price Index and that
corresponds to the annual in?ation rate during the ?scal year. Since the rate of
in?ation is sometimes large, the following formula is used to obtain the real return
for security i:
r
i
t
=
x
i
t
??
t
1 + ?
t
(3.11)
Where x
i
t
is the nominal return (either CIR, HPR) for security i at time t.
The last step consists of calculating the real interest rate for the whole portfolio
of government securities. This is done by calculating the weighted average of the real
interest rates of each security, where the weights represent the amount outstanding
19
For the US data as of end of December is used, given that monthly data was available, due
to changes in the ?scal year during the sample period.
78
of that security relative to the total outstanding of all securities. In the case of HPR
the total amount outstanding corresponds to the sum of the amounts outstanding
of the securities for which the measure was calculated. This is done, in order to
have the weights adding up to 1.
Portfolio Real Return (R
t
) =
Nt
t=1
r
i
t
Outstanding Amount
i
Total Outstanding Amount
(3.12)
where N
t
equals the total amount of securities at each point in time.
A liquidation year takes place whenever the real interest rate on the overall
portfolio is negative. This de?nition is a lower bound for the actual e?ect in the
case of the CIR measure since only cases where the combined e?ect of unanticipated
in?ation and ?nancial repression is large enough to make the real interest rate neg-
ative are considered. In the case of the HPR measure there exists the theoretical
possibility that it may be capturing changes in the general economic environment
and not the e?ects studied in this paper. As discussed in the previous section, em-
pirical evidence suggests this should account at most for a small fraction of the total
cases.
Finally, by saving in interest payments there is an implicit revenue for the
government in years of liquidation e?ect which can be calculated as:
Liquidation Revenues = Negative Real Interest Rate×Outstanding Stock of Domestic Debt
79
3.3.2 Data
A database was constructed from primary sources and includes 12 advanced
and developing countries. The sources are usually publications by the ?scal authority
or the central bank of each country; they are listed in Appendix ?? . The stock of
domestic debt in each year consists of the full list of securities outstanding at the
end of that year. For each security, I collected data on the outstanding amount,
maturity date, and coupon rate. Additional information was collected, depending
on its availability, including the price at which the securities were issued, and the
share of marketable and non-marketable debt.
There are 12 countries in the sample: Argentina, Australia, Belgium, France,
India, Ireland, Italy, Japan, South Africa, Sweden, United Kingdom and the United
States. The data on securities prices required to calculate the Holding Period Return
were available for Argentina, Belgium, India, Italy, Sweden, United Kingdom and
the United States. The sample period in each country generally covers 1945 to 1980.
The exceptions are India, which only has data after its independence in 1949, and
Belgium, and France, where data is unavailable for 11, and 8 years respectively.
Data is available for Ireland, Japan and Sweden that extends beyond 1980.
20
Table
3.1 lists the sample periods covered for all 12 countries.
Nine of these countries are advanced economies; the three developing countries
are Argentina, India and South Africa. These countries had quite di?erent economic
outcomes during the sample period. They also di?er in the degree to which they
20
More recent data is also available for other countries in the sample, but has not yet been
compiled.
80
were involved in World War II, and the challenges they faced at the end of it. There
are also large di?erences in their debt over time. Table 3.1 also shows the change in
the debt-to-GDP ratio from 1945 to 1980 for each country. Many of these countries
experienced large reductions in their debt over this period. The UK has the largest
decline in the debt ratio, from 210 to 41 percentage points. In the United States,
the ratio decreased from 118 to 33 percentage points. While debt ratios generally
declined over time, there are several countries in which the debt ratio remained
fairly constant or even increased. The diversity of the types and experiences of
these countries will be important to understand how general the results are, as well
as, how they vary depending on the countries’ economic characteristics..
It is important that the database covers all of a country outstanding securities,
as the composition of debt varies over time. For example, in the United States,
Treasury Bills constituted 6.5 percent of the total domestic debt in 1946 and 25.1
percent of the total in 1976, while non-marketable securities accounted for 22.7
percent in 1946, 16.7 percent in 1966 and 35.4 percent in 1976. As another example,
the share of marketable rupee loans in India went from 59 percent in 1950 to 39
percent in 1970. The composition of the debt portfolio of India and the US at
di?erent dates is shown in Table C.3 of Appendix ??.
3.4 The Liquidation E?ect from the Perspective of the Government
In this section, I begin by describing the in?ation in the 12 countries and real
interest rates on their debt portfolios over the sample period. The incidence and
81
magnitude of the liquidation e?ect is then outlined, together with a description of
the associated revenues for the government.
Several additional exercises are presented, which are aimed at understanding
what mechanisms are driving the aggregate patterns. First, in?ation expectations
are estimated to separate the relative contribution coming from unanticipated in-
?ation and ?nancial repression. Next, I focus on bonds issued within the sample
period in order to understand whether returns on new bond issues responded to
low real returns. This furthers our understanding of the role of ?nancial repression
during this period, as poor returns on new bond issues suggests a captive audience
for those bonds. The magnitude of the revenues from the liquidation e?ect are then
put in perspective by comparing it to those from the in?ation tax, and by looking
at the cumulative e?ect it had on the stock of debt. Finally, a multivariate analysis
is used to identify which country characteristics are particularly relevant to explain
the incidence of the liquidation e?ect.
3.4.1 In?ation and Real Interest Rates on Debt Portfolios
Before getting into an examination of the liquidation e?ect, it is important to
understand the in?ation rates and the distribution of real interest rates during the
period under study. The ?rst three columns of Table 3.2 shows several statistics for
the in?ation rates in the di?erent countries. The average in?ation rate in nine of the
countries is in single digits during this period, while the median in?ation is in single
digits in all of the countries except Argentina. In the United States the average
in?ation rate was 4.6 percent and the median in?ation rate was 3.2 percent, while in
82
the United Kingdom the average in?ation was 6.3 percent and median in?ation 4.2
percent. High in?ation rates in France, Italy, and Japan in the years immediately
after WWII declined signi?cantly in the 1950s.
Table 3.2 also shows the arithmetic mean, the median and the standard de-
viation for the real interest rate on the debt of each country. Columns (4) to (6)
show the results when the Contractual Interest Rate is used, and columns (7) to (9)
show the results when the Holding Period Return is used to measure real returns.
When the CIR is used, the average real interest rate is negative in all countries
except Sweden. The real interest rate is negative in seven of the eight countries
where the HPR can be calculated, with Belgium the only exception. Median real
interest rates are also low: for example, using the CIR, six of the 12 countries have
a median real return that is negative and the median real return is never larger than
1.2 percent. Using the CIR, the median real interest rate was -0.6 percent in the
United Kingdom and 0.3 percent in the United States. In the whole sample, the
fraction of the observations where the real interest rate is above three percent is,
on average, 11 percent; it is 5.6 percent in the United States and 2.8 percent in the
United Kingdom.
Two countries have particularly poor real returns. Argentina has the largest
negative mean and median real interest rate: the median real interest was -14.1
according to CIR and -11.0 according to HPR. France has the next largest average
and median negative real interest rates, which may be explained by two main facts.
First, the missing years of 1953-1958 and 1960-1963 perhaps contributes to that
pattern. In most countries, the real interest rate is above the average and median
83
values in those years. Second, even if France did not lose WWII, it had been
occupied by German forces during the War, which left the country and the economy
in a delicate situation. The average in?ation rate was 40 percent between 1945 and
1950, while the (weighted) average interest rate on the debt was 2.7 percent during
the same period.
It is clear that the results that follow are not going to be driven by a small
number of years in each country. The incidence of negative real interest rates during
the period 1945-80 was consistently high across the countries, and the distributions
of real returns were skewed towards negative values.
21
3.4.2 Incidence and Magnitude of the Liquidation E?ect
Liquidation years are years where the real interest rate on the debt portfolio
is negative. Table 3.3 shows the incidence of liquidation years for each country
according to the CIR measure. Column (1) contains the share of liquidation years
for the full sample period of each country, and the following columns show the share
of liquidation years for di?erent subperiods. The average share of liquidation years
is 56 percent for the full sample period. Excluding Argentina, which is an outlier in
the sample, it is 53 percent. Liquidation years comprise 50 percent of the years in
the United States and 58 percent of the years in the United Kingdom.
When looking at changes in the share of liquidation years across subperiods,
two patterns can be identi?ed. The most common one is a high incidence of liquida-
21
Reinhart and Sbrancia (2011) show the distributions of real deposit rates, discount rates and
T-Bill rates in a larger groups of advanced and emerging economies. In all cases the distributions
for the period 1945-1980 are to the left from those before 1945 and after 1980.
84
tion years immediately after the end of WWII, a lower incidence between 1957-1968,
and a higher incidence again in the 1970s. This is the case for eight countries: Aus-
tralia, Belgium, France, Italy, Japan, South Africa, the UK and the US. In all of
these countries, there is a higher incidence during the period 1969-1980 than during
1945-1956. These are typically countries where the debt ratios were high at the end
of WWII. The low incidence period of 1957-1968 coincides with the golden era of
Bretton Woods, while the high incidence during the 1970s occurs at a time when a
surge in the price of commodities led to an increase in the in?ation rates of most
countries.
The second pattern occurs in the case of Argentina, India, Ireland and Sweden,
who exhibit a reasonably constant incidence of liquidation years across the subpe-
riods. Argentina is the most extreme case, with almost every single year satisfying
the de?nition of liquidation year. In India, there is a lower incidence in the ?rst
subperiod and the subsequent increase in the share of liquidation years in the other
two subperiods. While the debt ratio in India did not vary much during the period
under study, this is explained by the average in?ation rate during 1949-1956 (0.2
percent) being signi?cantly lower than in the other subperiods (6.6 percent during
1957-1968 and 7.0 percent during 1969-1980). In the case of Ireland, the lower share
of liquidation years during the 1980s can be attributed to a higher average nominal
interest rate and a lower in?ation rate. In Sweden, the average in?ation rate during
1969-1980 was 8.4 percent, twice as large the average in?ation for the two previous
subperiods. This, combined with the fact that the nominal interest rate did not
increase by the same proportion, explains the higher incidence of liquidation years
85
during 1969-1980 relative to 1945-1968.
An interesting observation is that the incidence of liquidation years diminishes
in the countries for which the sample period extends beyond 1980, when most of
the controls were lifted and the era ?nancial liberalization began. In Ireland, the
incidence for 1969-1980 was 92 percent, whereas for 1981-1990 it was 30 percent.
It went from 83 percent to 10 percent over the same subperiods in Sweden, while
there is only one liquidation year in Japan after 1980. The period after 1980 is a
period of greater ?nancial liberalization, and the reduced incidence of negative real
returns seems to be associated with higher nominal interest rates and lower in?ation
rates. To illustrate, in Sweden the average nominal interest rate was 6.2 percent and
average in?ation 8.4 percent during 1969-1980, whereas the average nominal interest
rate was 10.9 percent and average in?ation 7.6 percent during 1981-1990. Even if
in?ation did not go down by much on average the nominal interest rate increased
signi?cantly.
Table 3.4 shows the magnitude of the liquidation e?ect in these liquidation
years. For any given year, the liquidation e?ect corresponds to the absolute value of
a negative real interest rate. Column (1) shows the average liquidation e?ect for the
full sample period, while the next ?ve columns show the average liquidation e?ects
for the same subperiods as in Table 3. Excluding Argentina and Japan, which have
large average liquidation e?ects of 21.4 and 13.2 percent respectively, the average
liquidation rate was 4.6 percent. The subperiods results show a similar pattern
to that observed in the previous table, namely that liquidation rates were higher
both after WWII and during the 1970’s. Column (1) of the table shows the largest
86
negative real interest rate for each country, together with the year in which it took
place. For roughly half of the countries, the minimum real interest rate took place
in the years immediately after the end of WWII, and for the other half it occurred
in the early 1970s. The minimum real interest rate in the United States was -13.7
percent in 1946, while in the UK the minimum was -10.9 percent in 1975.
3.4.3 Liquidation Revenues for the Government
The implicit revenues for the government are calculated as a percentage of
GDP and presented in Table 3.5. The way to interpret these estimates is that, if
the average liquidation revenues were two percent, it means that the government’s
de?cit was, on average, two percentage points of GDP lower during liquidation years
and led to savings on interest payments equivalent to two percent of GDP. Revenues
are determined by the magnitude of the liquidation e?ect and the size of the stock
of debt.
The average liquidation e?ect revenues during the entire sample period gener-
ally lies between 1.5 and 3.8 percent of GDP, with Sweden (0.8 percent) and Japan
(5.9 percent) the only countries outside of this range. Table 5 also shows average
revenues within subperiods. Initially high debt-to-GDP ratios means revenues were
highest in the period immediately after WWII. In all countries, the average revenue
is higher for 1945-1956 than for the full sample period. In the United States, average
revenues relative to GDP were 4.3 percent during 1945-1956 and 2.3 percent for the
full period, with both the liquidation rate and the debt ratio highest in that ?rst
subperiod. In the case of Australia, revenues during 1945-1956 (6.7 percent) were
87
twice as high as the average for the full sample (3.3 percent).Debt-to-GDP ratios
were relatively low in India, Ireland, South Africa and Sweden, which is re?ected in
lower average revenues. In India, where both the debt-to-GDP ratio and the liqui-
dation e?ect rate were constant over the full sample, the revenues were reasonably
constant across subperiods as well. Italy also exhibits relatively low revenues, the
large reduction in the debt ratio was between 1942-1947 which explains the high
value in the ?rst subperiod.
It is helpful to also compare liquidation e?ect revenues to tax revenues in those
years. Table 3.6 presents a comparison of the liquidation revenues relative to both
GDP and tax revenues for the Holding Period Return measure as well as the CIR
measure. Tax revenues do not include those from the in?ation tax. Results for
the CIR and HPR measures are similar, which suggests that the valuation e?ect
is generally small. The revenues from the liquidation e?ect can be sizable when
expressed in terms of tax revenues, as they average 20 percent of tax revenues. A
comparison to the revenues from the in?ation tax is presented later in the section.
An important ?nding of this paper is that the in?ation rate does not need to be
very high. Table 3.7 compares the median in?ation rate -due to the presence of very
high in?ation in some of the countries- during liquidation years and during 1930-
2010. The reason to start in 1930 is to focus on the period after the gold standard
when there was a change in the way monetary policy was conducted. Two thirds
of the countries in the sample have median in?ation below 10 percent both during
liquidation years and during 1930-2010. The di?erence is as low as 1.8 percentage
points for South Africa. In the US, median in?ation during liquidation years was
88
6.0 percent, in contrast to 3.0 percent during 1930-2010. In the UK median in?ation
was 8.3 percent, 4.5 percentage points higher than the historical median.
Higher in?ation rates led in some cases to a faster liquidation of the debt,
whereas in other cases this did not happen. In Argentina, in?ation rates were high
during most of the period (53.4 percent on average) but no large reduction in the
debt ratio is observed, possibly because the government kept running de?cits which
forced it to keep borrowing. In contrast, Italy and Japan had very high in?ation
(around 500 percent) at the end of WWII. This spike in in?ation in Italy reduced
the debt ratio from 118 percent in 1942 to 21 percent in 1947.
22
In the Japanese
case there is no good data for GDP during the war but di?erent estimates put the
debt ratio over 110 percent, in 1951 the debt-to-GDP ratio had reached 10 percent.
In both cases the sample was started in 1946 after the end of WWII, in the Italian
case the spike in in?ation was prior to that which is re?ected in a lower incidence
and liquidation rate as well as revenues.
23
If one were to include 1942-1945 to the
Italian sample the revenues numbers for the ?rst subperiod would be very similar
to those of Japan.
3.4.4 The Role of In?ation Expectations and Financial Repression
The measures presented so far do not distinguish between the relative contri-
butions of in?ation surprises and ?nancial repression. The goal of this section is to
estimate in?ation expectations, in orde to identify the relative contribution of each
22
Average real growth was -4 percent between 1942 and 1947
23
The year 1945 was considered as the end of the war despite the fact that Italy surrender in
September of 1943.
89
factor.
24
The empirical strategy to estimate in?ation expectations follows Fama (1975)
and Mishkin (1981), who were interested in testing for market e?ciency. An advan-
tage of this method is that it allows standard errors to be obtained. The analysis
is centered on the Fisher equation, in which the nominal interest rate at time t is
equal to the real interest rate plus the rate of in?ation that is expected between t?1
and t:
i
t
= r
A
t
+ ?
e
t
(3.13)
Where i
t
represents the nominal interest rate between t ? 1 and t, r
A
t
is the
real interest rate expected to be earned between t ?1 and t, and ?
e
t
is the in?ation
rate expected by the market between t ?1 and t.
Contrary to the ex ante real interest rate, which is determined by the expected
rate of in?ation, the ex post real interest rate is determined by the actual in?ation
between t ?1 and t:
24
There are several ways in which in?ation expectations have been estimated. One is to estimate
the anticipated component of in?ation (or money supply) with an ARMA or ARIMA process and
take the residuals as the unanticipated component (Barro, 1978). Other techniques include using
a Kalman ?lter (Burmeister, Wall, and Hamilton, 1986) or indexed bond yields to recover the
in?ation expectations (Deacon and Derry, 1994). The latter one is not a possible alternative in
this case because indexed bonds were issued after the end of the sample period. For example,
they were ?rst introduced in the United Kingdom in 1981 and in the United States in 1997.
While there are surveys on in?ation expectations, such as the Livingston Survey and the Thomson
Reuters/University of Michigan Survey in the United States, such surveys are not widely available
during this period.
90
r
P
t
= i
t
??
t
(3.14)
= r
A
t
?(?
t
??
e
t
) (3.15)
Under the assumption of rational expectations in the bond market, the forecast
error of in?ation should be uncorrelated with information available at t ?1, which
implies that:
E (?
t
??
e
t
|?
t?1
) = 0 (3.16)
Where ?
t?1
denotes the information set at t ? 1. That is, given all of the
available information at t?1, on average the di?erence between actual and expected
in?ation is equal to zero.
Using variables X
t?1
that are part of the information set ?
t?1
, the following
equation can be written:
r
A
t
= X
t?1
? + u
t
(3.17)
The error term u
t
is also determined at t ?1 and is assumed to have a mean
of zero, constant variance, and to be serially uncorrelated. Equation (3.17) can be
substituted into (3.14) to obtain:
r
P
t
= X
t?1
? + u
t
?
t
(3.18)
91
where
t
= ?
e
t
??
t
.
Contrary to (3.17), equation (3.18) can be estimated. Mishkin shows that the
OLS estimate of ? from (3.17) and (3.18) are equal in expectation. The variance-
covariance matrix derived from equation (3.18) will be larger than the one resulting
from (3.17), and the estimates of ? from equation (3.18) will be less precise.
The variables included in X
t?1
are: in?ation rate, money growth rate, real
GDP growth, and trend variables. These variables are the variables usually used in
this literature because of their high correlation with the ex ante real interest rate.
The estimates are used to obtain the following series for the ex-ante real interest
rate:
ˆ
r
A
t
= X
t?1
ˆ
?
r
P (3.19)
Combining this with equation (3.13), estimates for in?ation expectations can
be obtained:
ˆ
?
e
t
= i
t
?
ˆ
r
A
t
= i
t
?X
t?1
ˆ
?
r
P (3.20)
The exact standard errors for the estimates cannot be obtained, as the variance
of the within-sample error depends on the relative size of the variance of u
t
and the
variance of
t
. However, lower and upper bounds for the errors can be obtained.
I estimated this equation as follows. The ?rst step was to run the regression
with one explanatory variable at a time, varying the number of lags of that variable
and selecting the lag structure with the highest adjusted R
2
. Of the regressions
92
with the di?erent explanatory variables, the regression with the highest adjusted
R
2
was chosen. In the second step, a second variable was added to this regression
and the process repeated to choose the variable and lag structure that results in the
highest adjusted R
2
. This is done for the subsequent variables until adjusted R
2
is maximized.
25
This iterative process determined the regression to be estimated,
which was tested for serial correlation in the errors and for heteroscedasticity.
Table 3.8 contains the results of the regressions for each country when CIR
measures are used.
26
The explanatory variables are denoted as follows: in?ation
rate (INFL), growth of money supply (GM1), real growth in GDP (GROWTH),
and a trend variable (TREND). The estimated equation varies country by country,
re?ecting di?erent sets of variables that provided the best ?t. All of the regres-
sions include lagged values of in?ation and/or money growth; the ?rst lag of both
measures is generally negative and statistically signi?cant. Lagged values of growth
are included in several countries, although the magnitude of the coe?cients is never
particularly large. These country-speci?c speci?cations generally explain around
half of the variation in in?ation.
These regressions provide a basis on which to bound in?ation expectations.
27
When people form their expectations about future in?ation, they are likely to have
a distribution for future in?ation rather than a point estimate. For this reason, an
in?ation surprise year is said to take place whenever the actual in?ation rate is two
25
Darrat (1985) applies a similar method to calculate in?ation expectations but with an autore-
gressive process.
26
Similar results are obtained using the HPR, when it is available.
27
As a robustness exercise, in?ation expectations were also estimated by ?tting an autoregressive
process. This generates similar estimates of in?ation expectations. Also, for the United States,
these estimates were found to be similar to the Livingston survey on in?ation expectations.
93
standard deviations above the estimated expected rate of in?ation.
Table 3.9 shows the share of in?ation surprises relative to the total number
of years in the sample for each country, as well as the overlap between in?ation
surprises years and liquidation years. The results are presented using both the
lower and upper bound estimates for the standard errors. The average share of
in?ation surprises each county has is 8 percent using upper bound standard errors
and 17 percent using the lower bound. The frequency of in?ation surprise years in
liquidation years is 15 to 28 percent, depending on which estimate for the standard
errors is used. In either case, they constitute a minority of cases, which suggests
that in?ation surprises are not the primary cause of liquidation years.
For most countries, in?ation surprises are concentrated immediately after the
end of World War II and during the 1970s. It is worth noting that, after the end of
World War I, most countries experienced low in?ation rates as they tried to return
to the gold standard. This led many people to expect low in?ation rates after
World War II, and many economists thought that the biggest challenge after the
war would be slow growth and high unemployment (Studenski and Krooss, 1963).
What actually happened is that average in?ation rate in the decade after WWII
was 7 percentage points higher than the average in?ation rate in the decade after
WWI. The other period with high incidence of in?ation surprise years, in the 1970s,
corresponds to a period of oil shocks and a surge in the price of commodities.
The main conclusion from this exercise is that ?nancial repression appears to
have been more important than in?ation surprises in reducing government debt.
94
3.4.5 Are the Results Biased by Bonds Issued Before 1945?
In?ation rates were lower before WWII which, absent any restrictions, should
be re?ected in lower nominal interest rates. If a large proportion of the bonds in a
country’s portfolio were issued prior to 1945, then the results may be biased by the
returns on these bonds.
The richness of the database allows me to look at the issuance patterns over
time, to get a sense of what fraction of debt is issued before 1945 and the importance
of this issue. In most countries, many new securities were issued every period and
securities issued after 1945 quickly account for most of the debt. The exception is
Japan, where there were not many instruments issued between the end of WWII
and 1965. The maturity structure also means that most of the debt matures within
the sample period; as perpetual securities, such as the consols in the UK, account
for small fractions of the overall portfolios.
It may still be the case that the liquidation e?ects are not common for debt
issued after 1945. To check this, I analyze the performance of bonds issued after
1945 in Australia, India and Ireland, which are countries for which the necessary
data were available. The Yield to Maturity (YTM) is used to study the performance
of the bonds. The YTM is a measure of the per-period return an investor expects
to receive by holding a security until its maturity date. It is sometimes referred to
as the internal rate of return. For a given bond, the YTM at time t is given by the
following equation:
95
P
t
=
T
t=1
C
t
(1 + r)
t
+
Principal
(1 + r)
T
(3.21)
Where C
t
is the coupon payment, P
t
is the price of the security at time t, and
r is the YTM. The YTM is the rate of discount at which the present value of the
promised future cash ?ows is equal to the price of the security.
As mentioned before, the YTM should re?ect the investor’s in?ation expecta-
tions at the time the bond was issued. Investors are likely to have in mind a possible
distribution for future in?ation. While this is not directly observable, two extreme
assumptions about the real returns and in?ation expected by investors can provide
lower and upper bounds on the nominal YTM. First, to calculate a lower bound,
suppose investors expected a zero real return and so expected an in?ation rate equal
to the YTM for each period. Alternatively, suppose investors expected a real return
equal to the nominal YTM, so that they expected an in?ation rate equal to zero.
This can be thought of an upper bound, and these extremes can be used to bound
the ex ante real YTM as of time of issuance.
The ?nal step in calculating the real return consists of expressing the cash ?ows
of each security for each year until maturity in real terms. The year of issuance for
each security is used to de?ate the cash?ows. Having obtained these real cash ?ows,
and using equation (3.21), I can calculate the YTM that would deliver that stream
of cash?ows given the price at which the security was issued. This is the ex post
real YTM as of time of issuance, and provides a measure of real returns within the
sample period based on expectations at time of issuance.
96
Australian data is available for the period 1945-1968, Indian data for 1960-
1978, and Irish data for 1965-1975. The data for each security issued within these
years consists of the price and date of issuance, maturity date, and coupon rate. All
of the securities mature before 2010.
Figure 3.2 shows, at the time securities were issued, the frequency distribution
for both the ex ante real YTM and ex post real YTM. When it is assumed that
the expected in?ation rate is equal to the YTM in each period, there is a mass
concentrated at zero which is shown by the vertical solid line in each panel. It
follows then that, even without knowing what in?ation expectations investors had,
the actual distribution for the ex ante real YTM should be somewhere between the
two solid black lines. As the ?gures for the three countries show, the distribution for
the ex post real return is always to the left of the ex ante, which suggests negative
real returns were common. In the case of India and Ireland, the overlap between the
two distributions is close to zero. The share of observations with negative ex post
real YTM is 54 percent in Australia, 85 percent in Ireland, and 94 percent in India.
Summary information on the nominal and ex post real YTM appears in Table
3.10. For each country, I show the average, maximum and minimum for the nominal
YTM at issuance and the ex post real YTM as of time of issuance. On average, the
ex post real YTM was negative in all countries.
The ?ndings of this exercise show that the results presented in the previous
subsection are not biased by securities that were issued before 1945, and that neg-
ative real returns were common for securities issued within the sample period. The
results provide further evidence that the presence of ?nancial repression is an im-
97
portant factor for explaining the incidence and magnitude of the liquidation e?ect.
3.4.6 Comparison to In?ation Tax
In?ation has usually been considered as a tax on real cash balances (Friedman,
1971). It is important to compare the revenues from the liquidation e?ect to those
from the in?ation tax, in order to understand the incentives of governments to
use in?ation as a broader source of revenue. Calvo and Leiderman (1992) showed
evidence that in some situations the observed in?ation rate is above the rate at
which seigniorage is maximized. The authors argued that countries were sometimes
setting in?ation rates in the ine?cient side of the La?er Curve. However, if debt
liquidation revenues are an additional source of revenue generated by the in?ation
rate, then it may not be that the observed in?ation rates were too high but that the
tax bases were larger than previously thought. Support for this possibility comes
from Reinhart and Rogo? (2008, 2009), who noted during several episodes of high
in?ation that the debt-to-GDP ratio was much higher than the ratio of the monetary
stock to GDP.
The in?ation tax can be collected every period, which is not generally the case
for the "liquidation e?ect" tax. For this reason, instead of comparing year-on-year
revenues from each source, a comparison is conducted over subperiods. The revenues
from the in?ation tax are calculated as:
In?ation Tax =
?
t
1 + ?
t
? Money Supply
98
Where ?
t
is the in?ation rate. This is the component of seigniorage that would
be collected in steady state (see footnote 10). The monetary aggregate used is M1,
which is the most liquid monetary aggregate. This is the measure used by Rodriguez
(1994) and Easterly et al. (1995) for instance. If the monetary base was used instead
of M1, then the estimates for the in?ation tax would be lower because the monetary
base is a fraction of M1.
After calculating the revenues from each source, the sample is split into subpe-
riods of at least 10 years so that each country (except Ireland) has three subperiods
that cover the 1945-1980 period. Using the o?cial CPI, the revenues for each year
are expressed in constant terms. The base year is the ?rst year in each subperiod.
To illustrate, total revenues for the subperiod 1945-1956 are expressed in 1945 dol-
lars. The revenues are then added up and expressed in terms of the GDP of the
base year.
The results of such comparison are displayed in Table 3.11, with countries
grouped according to the main patterns that can be observed. The ?rst group of
countries has liquidation revenues that are consistently higher or similar to those
from the in?ation tax across the subperiods. These countries are Belgium, India,
Ireland, Sweden and the United Kingdom. In Belgium and the UK, the liquidation
e?ect dominated the in?ation tax both in the decade after 1945 and in the 1970s.
For India, Ireland and Sweden, the revenues from each source are broadly similar in
all subperiods.
The second group of countries have liquidation revenues which are relatively
large in the ?rst subperiod (normally 1945-1956), then have relatively larger in?ation
99
tax revenues in later periods. Countries in this group are Australia, France, Japan,
and the United States. Their liquidation e?ect revenues are higher than those from
the in?ation tax in the ?rst subperiod, when debt was very large. In the subsequent
subperiods, when the debt stock had been reduced and the incidence and magnitude
of the liquidation e?ect declined, the in?ation tax revenues were relatively higher.
In Australia and the United States the revenues from the liquidation e?ect increase
again during 1969-1980 but do not surpass those from the in?ation tax. The case of
Japan
28
highlights the point that in some episodes of high in?ation the revenue from
liquidating debt were higher than those from the in?ation tax. For the period 1946-
1956 the revenues from liquidation e?ect were 73.6 percent whereas the revenues
from in?ation tax were 15.3 percent.
In the third group of countries, which are Argentina, Italy, and South Africa,
the revenues from the in?ation tax are higher than those from the liquidation e?ect
in all subperiods. Debt in Italy had been reduced before the start of the ?rst
subperiod, whereas the money supply remained higher for a longer period of time.
In South Africa, the debt ratio was low and in?ation averaged a relatively low 3.5
percent between 1945 and 1972.
Despite the fact that the liquidation e?ect may not be collected every year,
during periods of high incidence of the liquidation e?ect the revenues obtained from
this source are usually higher than the revenues from in?ation tax. Liquidation rev-
enues were relatively high when the debt stocks were large. Looking at the revenues
from both sources together, this suggests that the total tax revenues generated by
28
Similar results are found (though not reported) for Italy for the period 1942-1947.
100
in?ation can be signi?cantly higher than previously thought. The e?ective tax base,
at least during this period of high incidence of negative real interest rates, should
be thought of as the stock of domestic debt in addition to the money stock.
3.4.7 E?ect on the Stock of Debt
To this point, I have focused on the e?ect of in?ation on the government debt
on a year-on-year basis. There is also a cumulative e?ect on the stock of debt, as by
paying lower interest on its debt the government has a lower de?cit which a?ects its
needs for new debt issuance and future interest payments. One way to capture the
magnitude of this cumulative e?ect is by assessing what each country’s debt-to-GDP
ratio would have under plausible alternative in?ation paths. As it will be shown, a
small di?erence in the average in?ation rate can have large e?ects on the stock of
debt.
To capture the e?ect on the stock of debt, the following equation of motion
for the government debt is used:
B
t
P
t
rGDP
t
= (1 + i
t?1
??
t?1
?g
t?1
)
B
t?1
P
t?1
rGDP
t?1
+
def
t?1
P
t
rGDP
t
(3.22)
Where B
t
is the stock of domestic debt, P
t
is the implicit price level, rGDP
t
is the real GDP, i
t
is the nominal interest rate, ?
t
is the net in?ation rate, g
t
is the
net real growth rate and def
t
is the primary de?cit.
A series for the primary de?cit is generated using the estimated values for the
101
nominal interest rate, together with observed values for the real stock of debt, real
growth and in?ation. I assume that the primary de?cit remains unchanged under
the di?erent in?ation scenarios. Debt ratios are then obtained for di?erent in?ation
paths.
Three alternative paths for the in?ation rate are assumed: (i) the in?ation rate
is equal to the country’s median in?ation between 1930 and 2010, (ii) the in?ation
rate is equal to the (weighted) average nominal interest rate of the corresponding
year, and (iii) the in?ation rate is equal to 2 percent.
29
These di?erent alternatives
help to understand the e?ect that in?ation can have in shaping debt dynamics.
Under the ?rst scenario, I assume that the experience in each country during the
sample period is comparable to its in?ation rate during a longer period of time.
With the second scenario, I compare the actual debt dynamics to a situation where
real interest rates are zero in every period. By assuming an in?ation rate of 2
percent, which is a common in?ation target nowadays in many countries, in the
third scenario it is possible to explore how the debt ratios in these countries would
have evolved if that in?ation target had been in place.
Table 3.12 shows the results of the analysis. Column (1) shows the debt-to-
GDP ratios at the start of the sample period (normally 1945), while Column (2)
shows the debt-to- GDP ratios at the end of the sample period (normally 1980).
Columns (3) to (5) show what the debt-to-GDP ratio would have been at the end
of the sample period under the three alternative in?ation paths.
The exercise highlights the cumulative e?ect of in?ation on the dynamics of
29
Under (i) and (iii), the in?ation rate remains constant over time.
102
the debt- to-GDP ratio. Compounding means that relatively small di?erences in
the in?ation rate has large long-term e?ects on debt-to-GDP ratio. For example,
the United States had an actual debt-to-GDP ratio in 1980 of 32 percent. Under
the scenario where the in?ation rate is equal to the median in?ation rate over 1930-
2010, which is 3.0 percent, the ratio would have been 51 percent in 1980; if the
in?ation rate had been a constant 2 percent the ratio would have been 72 percent
in 1980. The cumulative liquidation e?ect implied a reduction of 40 percentage
points, relative to a scenario where the annual in?ation rate had been 2 percent.
For the same scenario of a constant 2 percent in?ation rate, the di?erence between
the observed and estimated debt-to-GDP ratios in the United Kingdom would have
been of 167 percentage points, leaving the debt ratio at 212 percent. In Argentina,
where actual in?ation was signi?cantly higher than the alternatives proposed, the
estimated debt-to-GDP ratios are implausible high. This is also the case for Italy
and Japan if the sample is started in the ?nal years of WWII.
In the case of Japan the estimated ratios are shown until 1980 to make the
results comparable to those of other countries. For the period 1947-1980 the esti-
mated ratios are signi?cantly higher than the actual debt ratio which should not be
surprising given that median in?ation was 6 percent during those years. The ap-
proach can be also used to show what the debt dynamics in Japan would have been
if in?ation had been higher during the period 1981-2008. The in?ation rates during
this period were markedly lower than the in?ation rates in the previous forty years.
Median in?ation between 1981-2008 was 0.7 percent. If in?ation had remained at 2
percent, conducting the same exercise as before for 1980-2008, the debt ratio would
103
have been 115 percent in 2008.
30
The actual debt-to-GDP ratio in 2008 was 167
percent.
3.4.8 Understanding What A?ects the Probability of a Liquidation Year
In order to understand under which circumstances negative real interest rates
are most likely to occur, it is important to understand how di?erent factors a?ect
the return on the portfolio of government debt. For instance, does the share of
short-term debt in the portfolio increase the probability of a liquidation year? I
run a panel-data model to understand such factors. The results should be taken as
identifying conditional correlations rather than causal relationships.
The estimated model is given by:
r
CIR
it
= ?
i
+ ?
t
+X’
it
? + u
it
i = 1, ..., N
u
it
= ?
i
u
i,t?1
+
it
t = 1, ..., T
Where i is the country identi?er and t the time period, N is the total number
of countries and T is the number of time periods. The dependent variable is the
ex post real interest rate on the portfolio given by the CIR measure. The speci?-
cation includes country ?xed e?ects (?
i
) to control for constant di?erences between
countries, and time ?xed e?ects (?
t
) to control for shocks common to all countries.
30
The estimated debt ratio is 87 percent under the scenario of in?ation equal to historical
median in?ation and 70 percent under the scenario of in?ation equal to average nominal interest
rate.
104
The matrix X includes a constant and a number of explanatory variables, which
are detailed in the next paragraph. The error structure allows for correlation across
countries and for autocorrelation over time with country speci?c autocorrelation
coe?cients (?
i
). The estimator is the pooled least-square estimator with an AR(1)
for the errors.
31
The explanatory variables included in the regression are: the ratio of interest
payments to GDP (intgdp); the ratio of de?cit to GDP (defgdp); the ratio of tax
revenues to GDP (trgdp); the share of short term debt (st); and an indicator for
central bank independence (cbind). Higher interest payments could increase a gov-
ernment’s reliance on other sources of ?nancing and make in?ation more attractive.
A higher de?cit and lower tax revenues could have a similar e?ect. One would expect
that a higher share of short-term debt would be associated with a lower probability
of a liquidation year, because the government will need to re?nance its debt sooner
in the market. The idea behind Central Bank independence is that an independent
Central Bank may be less willing to ?nance government de?cits with in?ation, which
could lead to a higher real interest rate. The sources for these variables are detailed
in the Data Appendix.
Table 3.13 shows the results from estimating this equation. The variable with
the greatest explanatory power is the interest payments-to-GDP (intgdp) variable,
which is shown in Column (1). When either de?cit-to-GDP or tax revenues-to-GDP
are added, the estimated coe?cients have the expected signs in each case, but tax
31
The results are robust to other estimators such as Feasible Generalized Least Squares, or
allowing for other types of autocorrelation, such as a common AR(1) parameter.
105
revenues are not statistically signi?cant. This is the case both under di?erent error
structures and di?erent estimations. The results for both regressions are shown in
Columns (2) and (3) of Table 3.13. The negative coe?cient on the interest payments
variable means that increases in the variable are associated with lower real interest
rates. A higher value of the de?cit variable, a less negative de?cit or higher surplus,
have a positive conditional correlation with real interest payments.
32
In all of the speci?cations, the share of short-term debt is not statistically
signi?cant at the ?ve percent level. This could appear at ?rst as a surprising re-
sult given the argument that, as the government attempts to in?ate away its debt,
investors would seek protection from in?ation through shorter maturities and index-
ation (Blanchard et al., 1985; Spaventa, 1986). The result provides further evidence
that there were restrictions on the degree to which ?nancial markets could respond.
Further evidence of this lack of market response is provided in Figure 3.3, which
shows the maturity structure for debt in India, Japan, and the United States. The
dark gray area is short term debt maturing in less than a year and the lighter gray
area is debt with a maturity period longer than one year.
33
In all three countries,
the share of short-term debt represents less than 50 percent of the debt throughout
the sample period. In the United States, the share of short-term debt was highest
during the 1950s and early 1960s, and then declines to around 20 percent of to-
tal debt after 1965. In Japan, the share of short-term debt initially increases and
peaks around 1960, and then steadily declines to around 10 percent of the total debt
32
The de?cit variable is de?ned as the di?erence between revenues and spending and expressed
as a share of GDP.
33
The shares of debt remain fairly constant even when medium term debt -debt maturing in 1
to 5 years- is included.
106
by 1980. While India exhibits the largest share of short-term debt throughout the
whole sample period, after an increase in the ?rst decade the fraction of short-term
debt is reasonably stable. The fact that despite the large incidence of negative real
interest rate, governments were able to issue debts at a fairly unchanged interest
rate and still at long maturities can be interpreted as further evidence of ?nancial
repression during the period 1945-1980.
The last variable added, Column (5), is Central Bank Independence (cbind)
which is not statistically signi?cant at the ?ve percent level. One possible explana-
tion is that there are not many observations after 1980 when central banks became
more independent. An alternative explanation could be related to Alesina and Sum-
mer’s (1993) ?nding that, although a higher degree of central bank independence
is associated to a lower in?ation rate, there is no correlation between central bank
independence and real interest rates levels or variability.
A government’s ?scal position is an important determinant of the likelihood of
a liquidation year. After controlling for several factors, including the share of short-
term debt, worsening government ?nances are associated with lower real interest
rates on government debt. Montiel (2003) argued that ?nancial repression has a
?scal origin, in the sense that the inability of a government to collect revenues from
traditional sources forces it to seek other sources of revenues. In the case of ?nancial
repression, this constitutes an implicit tax on the ?nancial sector. The results of this
exercise suggest that there is a high conditional correlation between ?scal variables
and negative real interest rates, but the exercise cannot determine whether these
are causal relationships.
107
3.5 How did Investors Fare During this Period?
In this section, I compare the relative performance of bills, bonds and stocks.
In order to understand investors’ options and whether ?nancial repression policies
were key to the low returns on government bonds over the 1945-1980 period, I
compare the returns on bonds to other ?nancial assets during this period as well
as in the decades before 1945 and after 1980. The returns on bonds during the
1945-1980 were poor relative to the return on stocks, but higher than that on other
assets during the same period.
This suggests governments were able to generate demand for their securities,
and provides further evidence that the presence of ?nancial repression during the
sample period a?ected the return on government bonds. During World War II,
countries imposed controls on the economy in order to obtain the physical and
monetary resources necessary to ?ght the war.
34
The objective of governments was
the same across countries: to direct funds towards government bonds while keeping
the interest rate on their securities at low levels. Although the War o?cially ended
in 1945, it took several decades to dismantle the restrictions that had been imposed
during the war. Direct evidence on policies implemented during the sample period
can be found in Appendix sec:historyFR.
34
Oosterlinck (2010) discusses the situation in the French stock market during the war; and also
refers to the situation in other countries such as Belgium, the UK and the US.
108
3.5.1 Stocks, Bills and Bonds during the Sample Period
Table 3.14 shows the real returns using the portfolio of government debt us-
ing the Contractual Interest Rate and Holding Period Return measures, together
with real returns on bills (short-term debt) and the stock market. For each of the
variables, I present the geometric return, the arithmetic return and the standard
deviation of the real returns. The geometric (or compound) return is the most
appropriate measure to compare investments over a longer period of time (Bodie,
Kane and Marcus, 2009), but I show the results are also similar when the arithmetic
return is used. This information is provided for every country except Argentina and
Ireland, as they had no stock market data that included dividends.
The most accurate measure on government bonds is the HPR, since it incor-
porates capital gains and losses. The CIR is also presented, however, because the
HPR is not available for all of the countries. The results are generally similar, al-
though the average real returns measured by CIR are one percentage point lower
than those measured by HPR. This suggests that capital gains partially compensated
investors for low interest payments. As a reminder, the measures for the return on
government bonds are for the portfolio of government securities, and hence contain
securities of various maturities. The return for short term government securities,
bills, is presented separately as well.
The stock market returns were calculated using the most comprehensive index
available for each country. In all cases, the indexes include both dividends and
price changes. Details on the indexes used and their sources are included in ??.
109
The results for Japan and Sweden correspond to the period 1945-1980, so that the
experience in these countries can be compared to that of the other countries in the
sample.
In all countries, the average real return from the stock market is positive
whereas the real return from government bonds is negative (except for Belgium
under the HPR measure). This is true for both the short term and overall portfolio
of government securities. The fact that the geometric mean is negative implies that
someone who invested at the beginning of the period recovered less than the initial
investment at the end of the sample period. For an investor who invested $1,000 in
US government bonds in 1945, he would have received $824 in 1980 (measured in
1945 dollars). An investor in the United Kingdom would have recovered 58 percent
of their initial investment during the same sample period.
The standard deviation for the portfolio of debt is lower than the standard
deviation for stocks, which implies that there is a trade-o? between risk and return..
A simple way to look at this trade-o? is to compute the percentage of the time that
stocks outperformed bonds for holding periods of di?erent lengths. In Table 3.15, I
present these percentages for holding periods of one, two, ?ve, ten, 20 and 30 years
for both the CIR and HPR measures. The way to read the table is as follows: in
Australia, stocks perform better than bonds in 71 percent of all two year holding
periods, 96 percent of 10 year holding periods and 100 percent of the 20 and 30 year
holding periods. In all of the countries and under both measures, stocks tend to
outperform bonds regardless of the holding period. In around 60 percent of the one
year holding periods, stocks perform better than bonds. This suggests that, even
110
over short investment horizons, stocks outperform bonds and make it unlikely that
risk could explain the di?erences in returns between bonds and stocks.
The calculations reported in this table are before-tax returns. In order to be
able to calculate after-tax returns, one would need historical information on the
tax codes of the di?erent countries. Siegel (2008) calculated historical after-tax
returns for the United States, and ?nds that stocks to have a tax advantage relative
to bonds. This occurs because relatively most of the return on stocks come from
capital gains rather than dividend payments, and capital gain taxes can be deferred
until the assets are sold so that the investment grows at the before-tax rate. The
di?culty of extending this to other countries is that, while dividend and income
taxes are usually higher than capital gain taxes, government securities often receive
special tax treatment. This special treatment for government bonds makes it hard
to know which group of assets had a tax advantage in other countries.
35
3.5.2 A Longer Historical Perspective
Dimson, Marsh, and Stauton (2002) and Siegel (2008) provide evidence that
investing in the stock market is the best alternative for investors with a long time
horizon, at least in advanced economies. They do this by comparing the returns on
equity, bonds and bills in samples that span more than 100 years.
The results presented by these authors are useful to put the returns during the
1945-1980 period into historical perspective. Siegel (2008), who focuses on the US
35
For instance, in the US municipal bonds are tax-exempt. In the UK private investors in Gilts
are not liable for capital taxes.
111
economy, reports that the annual real return on long-term government bonds was 4.8
percent over 1802-1870, 3.7 percent over 1871-1925, and 2.4 percent between 1926-
2006. If the last subperiod is further split into smaller subperiods, it becomes clear
that the low real returns are driven by the negative average real returns between
1945-1980. This suggests that the 1945-1980 period was distinctive in terms of the
poor real returns on US government bonds.
Dimson et al. (2002) reported real returns on bonds, bills and stocks for all of
the countries in my sample except Argentina and India. Tables 3.16 and 3.17 show
these returns over 1900 to 2000, as well as over 1900-1939, 1940-1979 and 1980-
2000. The returns on bonds for both 1900-1939 and 1980-2000 periods were higher
than those during the 1940-1979 period in all of the countries except Belgium. In
Australia, the annual real return on bonds from 1900-2000 was 1.1 percent while the
real return between 1940-1980 was -2.8 percent. In the United States, the equivalent
numbers are 1.6 percent and -1.8 percent respectively.
Bills are short-term securities, which should quickly re?ect changes in in?ation
expectations and market interest rates in a well-functioning market. The fact that
the average real return on bills was negative in all countries during 1940-1979 o?ers
supportive evidence of the presence of ?nancial repression. Average real returns on
bills are rarely negative in earlier and later periods.. Across the ten countries for
which this information is available, the average real return on bills was 1.0 percent
between 1900-1939, -3.6 percent between 1940-1979, and 3.5 percent between 1980-
2000.
In contrast to the performance of government bonds and bills, the real return in
112
the stock market was positive in all subperiods in all countries (with the exception of
Italy between 1940-1979). There is no common pattern for stock returns across the
1900-1939 and 1940-1979 subperiods, while the return for the 1980-2000 subperiod
was markedly higher. This last observation will be important next, when the equity
premium is re-examined.
3.5.3 The Equity Premium Puzzle Revisited
The equity premium is the excess return on equities over a risk-free asset, such
as US T-bills, and is a key variable in many asset pricing models. Mehra and Prescott
published a paper titled The Equity Premium: A Puzzle?" in 1985, where they found
that the historical equity premium was far higher than the premium that could
be rationalize by a standard neoclassical model. Speci?cally, the observed equity
premium over a 100 year period was more than 6 percent, whereas the premium
predicted by the model was 1.4 percent. An additional ?nding at odds with the
empirical evidence was that the risk free rate predicted by the model was 13.2
percent, whereas the average risk free rate observed in the data was less than 1
percent.
A large literature has attempted to explain this puzzle. Mehra (2006) summa-
rizes the di?erent explanations that have been provided, dividing the explanations
into those that are risk-based and those that are not. Under the ?rst category,
Mehra groups explanations that have been successful at obtaining a risk free rate
in line with the empirical evidence but have failed to explain the equity premium.
113
Examples of this are models which propose alternative preferences (e.g., habit for-
mation), di?erent probability distributions (e.g., adding a disaster state), and be-
havioral models where agents are not fully rational. Mehra concludes that: "The
di?culty that several model classes have collectively had in explaining the equity
premium as a compensation for bearing risk leads us to conclude that perhaps it is
not a "risk premium" but rather due to other factors." Papers that explore explana-
tions that are not risk- ased use models which take into account market frictions or
allow for incomplete markets, by adding characteristics like borrowing constraints,
transaction costs and taxes.
36
The particularly low bond returns during the 1945-1980 period may account
for some of the equity premium. To examine this, I calculate the equity premium
for rolling 30-year periods over time spans in the United States and the United
Kingdom.
37
These are plotted in Figure 3.4, with the US results shown in Panel
A and the UK results in Panel B. The shaded areas correspond to the period of
?nancial repression (1945-1980). The data for the UK is available for the 1800-2010
period and for the US it is available for the 1870-2010 period, so that the ?rst 30-year
period is 1829 and 1899 respectively.
The ?rst observation is that the equity premium has varied signi?cantly over
time, and it has even been negative at times in the United Kingdom. The second,
more important observation is that the peak in the equity premium in both countries
coincides with the period of ?nancial repression. Excluding 1945-1980, the average
36
In addition to Mehra (2006), Kocherlakota (1996), Cochrane (1997) and Campbell (1999,
2001) also o?er surveys of the literature on the equity premium.
37
Each observation is the geometric mean for the equity premium over the preceding 30 years
114
equity premium for the period 1900-2010 is 2.1 percent in the UK and 4.4 percent
in the US. This compares with the average equity premia for the 1945-1980 period
of 6.7 percent in the UK and 8.3 percent in the US.
McGrattan and Prescott (2003) argue that one should take into account the
role of taxes, diversi?cation costs and regulatory constraints as determinants of
the equity premium. Although some of their results are encouraging, further work
should be done to try to measure the e?ect of ?nancial regulations, including taxes
and other restrictions, in shaping the equity premium.
The observations of this section have potentially important implications for
understanding the equity premium and the equity premium puzzle. First, a success-
ful model of the equity premium should take into account the e?ect of regulations
and institutional background. Trying to explain the average equity premium with-
out acknowledging its large variations over time may not be particularly helpful to
model prospective risk. In addition, if one believes that to some extent it is possible
a return to a more tightly regulated ?nancial system, then one should expect to
observe a larger equity premium.
3.6 Conclusion
As a result of the recent ?nancial crisis, the public debt ratios of advanced
economies have increased to levels not seen since World War II. This raises questions
as to how governments will reduce their debt burden. Studying how countries coped
with similar situations in the past can shed light on the e?ectiveness and implications
115
of the di?erent alternatives. In this paper, I provide empirical evidence that in?ation
reduced post-World War II debts. I show the conditions under which in?ation was
e?ective and the implications for both governments and investors.
In combination with ?nancial repression, in?ation was an e?ective mechanism
to reduce large debt burdens. Negative real interest rates were common and large in
magnitude across the 12 countries under study. On average, the real interest rates
on the portfolio of domestic government securities were negative in half of the obser-
vations in my sample. Implicit government revenues averaged two to three percent
of GDP. When the sample is broken into smaller subperiods, it becomes apparent
that there was a high incidence of the e?ect both in the years after the end of World
War II and during the 1970s. In France, Italy and Japan, the countries which ex-
perienced the highest in?ation rates after the war, the revenues as a proportion of
GDP averaged 12 to 17 percent between 1945 and1956.
A conceptual framework was developed to identify the channels through which
in?ation can have an e?ect in reducing government debt, and to show how to think
of ?nancial repression as a restructuring mechanism. Several exercises were con-
ducted to separate their relative contribution. The results consistently point to the
importance of ?nancial repression, rather than unanticipated in?ation, in explain-
ing the high incidence of negative real interest rates. For instance, when in?ation
expectations are measured I ?nd that in?ation surprises occur in only 15 percent of
liquidation years.
Of the various exercises focused on governments, two are worth emphasizing.
First, when compared to other sources of government revenue, liquidation revenues
116
are large. On average, liquidation revenues are equivalent to one ?fth of tax revenues,
and they are sometimes signi?cantly larger than those from the in?ation tax. Second,
the results of a multivariate analysis point at an important connection between ?scal
variables and the presence of the liquidation e?ect. While this is consistent with the
argument by Montiel (2003) about ?nancial repression having a ?scal origin, further
work is required to understand whether the links between the variables are causal.
Most of the sample period coincides with the Bretton Woods era, which was
characterized by the presence of tightly regulated capital ?ows. This appears to
have facilitated domestic policies that kept interest rates on government debt arti-
?cially low. Further evidence of the presence of ?nancial repression and its e?ect
is provided by looking at the return of government bonds for the period 1900-2000.
The real returns on bonds were signi?cantly higher both in the period before and
after 1940-1980. The presence of frictions in ?nancial markets during this period
is also apparent when looking at the average real returns on Treasury Bills, which
were negative in all of the countries in the sample between 1940 and 1980.
The abnormally low real return for bonds during the period of ?nancial re-
pression is also re?ected in the equity premium. The equity premium, calculated
over 30-year rolling periods, was the highest during the decades after World War II
that are studied in this paper. In the United States, the average equity premium
from 1870 to the present is 4.4 percent while the average equity premium during
1945-1980 is almost twice as large at 8.3 percent. These ?ndings may be relevant
for understanding the equity premium puzzle, and suggest that studies trying to as-
sess prospective risk should take into account the signi?cant e?ect that government
117
intervention can have on the return of di?erent assets.
Real growth is unlikely to play a major role in reducing debt burdens at least in
the next few years. Moreover, conditional on being able to implement ?scal austerity
measures, the evidence on their success is limited to a few countries (Perotti, 2011).
The options left are explicit defaults and restructurings, or the mechanisms studied
in this paper. In the last three decades, the world has moved towards a more
?nancially liberalized environment, which means that the magnitudes found here
may not be representative of what could happen in the future. There have been
some recent regulatory changes, however, that suggest governments may still be able
to pay low real returns on their debt in di?cult times. That, together with the scale
and breadth of how in?ation was used to liquidate government debt in the period
under study, suggests we should pay more attention to the use and implications of
this debt-reduction mechanism.
118
Figure 3.1: Debt-to-GDP, Real Growth Rate and Distribution of Real Returns
Panel A: Debt-to-GDP Ratios
Panel B: Real Growth Rate (5-year moving average)
Panel C: Distribution of Real Interest Rates
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
50
100
150
200
250
1830 1842 1854 1866 1878 1890 1902 1914 1926 1938 1950 1962 1974 1986 1998
D
o
m
e
s
t
i
c
D
e
b
t
-
t
o
-
G
D
P
(
i
n
p
e
r
c
e
n
t
s
)
UK
US
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
-10
-5
0
5
10
15
1835 1847 1859 1871 1883 1895 1907 1919 1931 1943 1955 1967 1979 1991 2003
(
R
e
a
l
G
D
P
g
r
o
w
t
h
-
5
-
y
e
a
r
m
o
v
i
n
g
a
v
e
r
a
g
e
,
i
n
p
e
r
c
e
n
t
s
)
US
UK
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
5
10
15
20
25
30
35
-9% -6% -3% 0% 3%
P
r
o
b
a
b
i
l
i
t
y
D
e
n
s
i
t
y
F
u
n
c
t
i
o
n
(
i
n
p
e
r
c
e
n
t
)
US
UK
119
Figure 3.2: Frequency Distributions of Nominal and Ex Post Real Yield to Maturity
(YTM)
AUSTRALIA
INDIA
IRELAND
-2
8
18
28
38
48
58
68
-8% -6% -4% -2% 0% 2% 4% 6% 8%
P
r
o
b
a
b
i
l
i
t
y
D
e
n
s
i
t
y
F
u
n
c
t
i
o
n
(
i
n
%
)
Nominal YTM Ex Post Real YTM
-5
0
5
10
15
20
25
30
35
40
-4.0% -2.0% 0.0% 2.0% 4.0% 6.0% 8.0%
P
r
o
b
a
b
i
l
i
t
y
D
e
n
s
i
t
y
F
u
n
c
t
i
o
n
(
i
n
%
)
Nominal YTM Ex Post Real YTM
-5.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
-9% -8% -6% -5% -3% -2% 0% 2% 3% 5% 6% 8% 9% 11% 12%
P
r
o
b
a
b
i
l
i
t
y
D
e
n
s
i
t
y
F
u
n
c
t
i
o
n
(
i
n
%
)
Nominal YTM Ex Post Real YTM
120
Figure 3.3: Maturity Structure
INDIA
JAPAN
UNITED STATES
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1949 1953 1957 1961 1965 1969 1973 1977
Short Term (1 year)
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Short Term ( 1 Year)
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1944 1948 1952 1956 1960 1964 1968 1972 1976
Short Term ( 1 year)
121
Figure 3.4: Rolling 30-year Equity Premium
UNITED KINGDOM
UNITED STATES
0
0
0
0
0
1
1
1
1
1
1
-2.5
-0.5
1.5
3.5
5.5
7.5
9.5
1829 1849 1869 1889 1909 1929 1949 1969 1989 2009
(
P
e
r
c
e
n
t
a
g
e
P
o
i
n
t
s
)
Equity
Premium
Equity
Premium -
Real Growth
1900-2010* 2.1 0.3
1945-1980 6.7 4.7
* excluding 1945-1980
0
0
0
0
0
1
1
1
1
1
1
0
2
4
6
8
10
12
1899 1909 1919 1929 1939 1949 1959 1969 1979 1989 1999 2009
(
P
e
r
c
e
n
t
a
g
e
P
o
i
n
t
s
)
Equity
Premium
Equity
Premium -
Real Growth
1900-2010* 4.4 1.0
1945-1980 8.3 4.5
* excluding 1945-1980
122
Table 3.1: Countries in the Sample and Sample Periods
Country Period Change in debt-to-GDP in percentage points (p.p.)
Argentina 1942-1980 15.2 p.p. (from 42.2 to 27.0)
Australia 1945-1980 124.6 p.p (from 143.8 to 19.3)
Belgium
a
1945-1974 74.4 p.p. (from 112.8 to 38.4)
France
b
1945-1980* 97.9 p.p. (from 104.1 to 6.2)
India 1949-1980 (1.4) p.p. (from 27.5 to 28.9)
Ireland 1960-1990 1.8 p.p. (from 54.7 to 52.8)
Italy
c
1946-1980 11.6 p.p. (from 37.2 to 25.6)
Japan 1946-2008 (82.6) p.p. (from 82.6 to 164.2)
South Africa 1945-1980 40.6 p.p. (from 73.0 to 32.4)
Sweden 1945-1990 13.9 p.p. (from 52.0 to 38.1)
United Kingdom 1945-1980 169.1 p.p. (from 210.0 to 40.9)
United States 1945-1980 85.0 p.p. (from 118.4 to 33.4)
a
Missing data for 1964-1968
b
Missing data 1953-1958 and 1960-1963
c
The debt ratio in 1942 was 118 percent of GDP.
Table 3.2: Summary of In?ation and Real Interest Rate (in percent)
Country
In?ation CIR HPR
Average Median St. Dev. Average Median St. Dev. Average Median St. Dev.
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Argentina 53.4 21.7 81.0 -19.4 -14.1 18.3 -14.1 -11.0 15.9
Australia 6.4 4.3 5.4 -1.7 -0.3 4.5 n.a. n.a. n.a.
Belgium 3.9 3.2 4.0 -0.6 0.2 4.3 1.0 2.0 5.8
France 12.2 6.0 17.7 -7.3 -2.4 12.7 n.a. n.a. n.a.
India 5.2 4.4 7.5 -0.7 -0.4 6.8 -1.3 -0.7 6.9
Ireland 8.5 6.7 6.2 -1.3 -0.8 4.3 n.a. n.a. n.a.
Italy 8.3 4.6 10.1 -1.6 0.4 6.8 -0.6 0.7 5.7
Japan 27.2 5.6 73.7 -1.5 1.2 12.3 n.a. n.a. n.a.
South Africa 5.2 4.1 3.8 -0.5 0.2 2.7 n.a. n.a. n.a.
Sweden 5.9 5.7 3.9 0.2 0.5 3.4 -0.9 -0.2 4.7
UK 6.3 4.2 5.8 -1.4 -0.6 3.7 -1.2 -0.1 7.3
US 4.6 3.2 4.4 -0.8 0.3 4.1 -0.4 0.0 4.2
Average 12.3 6.1 18.6 -3.1 -1.3 7.0 -2.5 -1.3 7.2
Notes: See Table 3.1 for sample period.
a
Missing data for 1964-1968
b
Missing data 1953-1958 and 1960-1963
123
Table 3.3: Incidence of Liquidation Years
Country Period
Share of Liquidation Years for di?erent subperiod
Full Sample 1945-1956 1957-1968 1969-1980 1981-1993 1994-2008
(1) (2) (3) (4) (5) (6)
Argentina 1942-1980 92 80 100 100 - -
Australia 1945-1980 53 67 8 83 -
Belgium
a
1945-1974 48 42 14 100 - -
France
b
1945-1980 77 75 50 92 - -
India 1949-1980 53 25 67 58 - -
Ireland 1960-1990 68 - 78 92 30 -
Italy 1946-1980 49 36 25 83 - -
Japan 1946-2008 35
c
64 42 75 0 7
South Africa 1945-1980 47 58 8 75 - -
Sweden 1945-1990 48 42 50 83 10 -
United Kingdom 1945-1980 58 67 25 83 - -
United States 1945-1980 50 58 17 75 - -
Average 56 56 40 83 13 7
Notes: Share of liquidation years is calculated as the ratio between number of years in which the real return
was negative and the total number of years in the corresponding subperiod. The measure of real interest rate
is the Contractual Interest Rate (CIR).
- Subperiods not included in the sample of the country
a
Missing data for 1964-1968
b
Missing data 1953-1958 and 1960-1963
c
The share of liquidation years for the period 1946-1980 is 60 percent
Table 3.4: Liquidation Rate
Country
Average Liquidation E?ect Minimum
Full Sample 1945-1956 1957-1968 1969-1980 1981-1993 1994-2008 (Year)
(1) (2) (3) (4) (5) (6) (7)
Argentina 21.4 13.0 17.0 34.2 - - 72.3 (1976)
Australia 4.6 6.8 0.9 3.2 - - 15.1 (1952)
Belgium 4.2 6.0 1.0 3.1 - - 9.6 (1974)
France 9.8 26.8 1.4 2.9 - - 41.2 (1946)
India 5.4 6.0 4.8 5.8 - - 17.4 (1974)
Ireland 3.4 - 1.0 4.4 5.7 - 12.7 (1975)
Italy 6.0 13.3 2.1 4.2 - - 27.6 (1947)
Japan 13.2 35.0 2.2 3.7 * 0.1 78.5 (1946)
South Africa 3.0 2.7 0.5 3.3 - - 6.8 (1975)
Sweden 2.6 4.7 1.2 2.4 2.9 - 11.9 (1951)
United Kingdom 3.5 2.7 0.5 5.1 - - 10.9 (1975)
United States 3.5 4.1 0.1 3.7 - - 13.7 (1946)
Average 6.7 11.0 2.7 6.3 4.3 0.1
Notes: The liquidation rate is absolute value of the real interest rate during liquidation years.
See table 3.3 for sample period. The measure of real interest rate is the Contractual Interest Rate.
- Subperiods not included in the sample of the country
* Subperiods which are part of the sample but have no LE years
124
Table 3.5: Liquidation Revenues
Country
Average Liquidation E?ect Revenues as percentage of GDP
Full Sample 1945-1956 1957-1968 1969-1980 1981-1993 1994-2008
(1) (2) (3) (4) (5) (6)
Argentina 3.1 3.8 3.0 2.5 -
Australia 3.3 6.7 0.6 0.8 - -
Belgium 2.5 4.9 0.6 1.3 - -
France 3.8 12.4 0.2 0.2 - -
India 1.5 1.6 1.4 1.5 - -
Ireland 1.8 - 0.5 2.2 2.9 -
Italy 1.6 13.3 0.4 1.1 - -
Japan 5.9 17.8 0.1 0.6 * 0.1
South Africa 1.3 1.5 0.2 1.2 - -
Sweden 0.8 1.6 0.3 0.6 1.3 -
United Kingdom 3.0 4.4 0.6 2.6 - -
United States 2.3 4.3 0.0 1.2 - -
Notes:See Table 3.3 for sample period.The measure of real interest rate is the Contractual Interest Rate.
- Subperiods not included in the sample of the country
* Subperiods which are part of the sample but have no LE years
Table 3.6: Comparison Liquidation Revenues as percentage of:
Country
GDP Tax Revenues
CIR HPR CIR HPR
Average Median Average Median Average Median Average Median
Argentina 3.1 2.4 3.1 2.2 38.3 27.3 39.0 28.6
Australia 3.3 1.1 n.a. n.a. 12.9 4.4 n.a. n.a.
Belgium 2.5 1.2 3.5 3.5 18.6 10.3 23.9 18.9
France 3.8 0.2 35.3 1.1
India 1.5 1.6 1.5 1.8 27.2 27.6 27.2 28.4
Ireland 1.8 1.1 n.a. n.a. 7.9 6.1 n.a. n.a.
Italy 1.6 0.8 1.6 0.7 24.6 5.4 26.5 4.9
Japan 5.9 0.4 n.a. n.a. 37.9 3.0 n.a. n.a.
South Africa 1.3 1.4 n.a. n.a. 8.0 6.8 n.a. n.a.
Sweden 0.8 0.4 1.3 0.9 4.4 2.1 4.4 2.1
United Kingdom 3.0 1.9 3.1 1.8 18.8 11.0 19.6 10.0
United States 2.3 0.7 2.7 1.3 13.4 3.9 15.9 7.1
Notes:See Table 3.3 for sample period.
125
Table 3.7: Comparison of Median In?ation between LE Years and 1930-2010
Contractual Interest Rate
Liquidation Years 1930-2009
Argentina 21.1 15.3
Australia 9.0 3.8
Belgium 7.6 2.6
India 9.4 5.6
Ireland 10.0 3.2
Italy 11.3 4.5
Japan 8.3 3.0
South Africa 7.5 5.8
Sweden 7.0 4.0
United Kingdom 8.3 3.8
United States 6.0 3.0
Holding Period Return
Liquidation Years 1930-2009
Argentina 21.4 15.3
Belgium 8.6 2.6
India 9.3 5.6
Italy 12.2 4.5
Sweden 7.0 4.0
United Kingdom 11.9 3.8
United States 5.6 3.0
126
T
a
b
l
e
3
.
8
:
R
e
g
r
e
s
s
i
o
n
s
f
o
r
I
n
?
a
t
i
o
n
E
x
p
e
c
t
a
t
i
o
n
s
A
r
g
e
n
t
i
n
a
A
u
s
t
r
a
l
i
a
B
e
l
g
i
u
m
F
r
a
n
c
e
I
n
d
i
a
I
r
e
l
a
n
d
I
t
a
l
y
J
a
p
a
n
S
o
u
t
h
A
f
r
i
c
a
S
w
e
d
e
n
U
n
i
t
e
d
K
i
n
g
d
o
m
U
n
i
t
e
d
S
t
a
t
e
s
C
o
n
s
t
a
n
t
0
.
9
2
0
.
0
2
-
0
.
3
4
0
.
0
0
-
0
.
0
5
0
.
3
6
0
.
0
2
-
0
.
5
2
-
0
.
0
3
0
.
0
5
-
0
.
0
2
0
.
0
2
(
3
.
1
5
)
(
2
.
7
)
-
4
.
7
6
-
0
.
0
5
(
-
1
.
1
8
)
(
2
.
9
4
)
(
1
.
4
1
)
(
-
4
.
3
9
)
(
-
0
.
9
2
)
(
2
.
0
8
)
(
-
0
.
5
1
)
(
2
.
8
6
)
I
n
?
(
T
-
1
)
-
0
.
0
8
-
0
.
6
5
0
.
3
4
-
0
.
5
2
-
0
.
3
2
-
0
.
1
1
-
0
.
1
4
-
0
.
3
2
-
0
.
3
0
-
0
.
5
2
-
0
.
2
4
(
-
1
.
9
6
)
(
-
5
.
3
6
)
1
.
5
8
-
3
.
5
3
(
-
2
.
1
6
)
(
-
6
.
0
4
)
(
-
4
.
7
6
)
(
-
1
.
8
6
)
(
-
2
.
2
)
(
-
3
.
7
3
)
(
-
2
.
1
6
)
I
n
?
(
T
-
2
)
0
.
0
4
0
.
3
7
0
.
6
0
0
.
1
2
-
0
.
0
2
0
.
0
2
-
0
.
0
2
-
0
.
0
9
0
.
1
6
0
.
1
5
0
.
2
3
(
1
.
1
)
(
3
.
0
7
)
3
.
4
2
0
.
5
8
(
-
0
.
0
2
)
(
0
.
7
8
)
(
-
0
.
7
3
)
(
-
0
.
5
3
)
(
1
.
1
4
)
(
1
.
1
2
)
(
2
.
0
1
)
I
n
?
(
T
-
3
)
0
.
1
6
-
0
.
1
2
0
.
2
6
-
0
.
0
8
0
.
0
7
0
.
9
6
-
0
.
5
8
(
1
.
7
8
)
(
-
4
.
0
5
)
(
2
.
7
1
)
I
n
?
(
T
-
4
)
0
.
4
8
0
.
0
7
0
.
0
5
2
.
8
3
0
.
4
6
(
3
.
3
5
)
G
M
1
(
T
-
1
)
-
0
.
2
5
-
0
.
3
5
-
0
.
0
8
-
0
.
3
9
(
-
4
.
1
3
)
(
-
2
.
3
3
)
(
-
4
.
0
5
)
(
-
2
.
4
1
)
G
M
1
(
T
-
2
)
-
0
.
0
1
-
0
.
1
5
0
.
0
5
0
.
1
3
(
-
0
.
1
1
)
(
-
0
.
9
)
(
3
.
3
5
)
(
0
.
6
7
)
G
M
1
(
T
-
3
)
-
0
.
5
0
-
0
.
0
8
0
.
2
1
(
-
2
.
4
)
(
-
0
.
7
6
)
(
1
.
2
5
)
G
M
1
(
T
-
4
)
-
0
.
3
1
(
-
3
.
4
7
)
G
r
o
w
t
h
(
T
-
1
)
0
.
0
0
0
.
0
0
0
.
0
0
(
-
3
.
2
6
)
(
1
.
5
1
)
(
-
2
.
0
9
)
G
r
o
w
t
h
(
T
-
2
)
0
.
0
0
(
-
1
.
1
4
)
T
r
e
n
d
-
0
.
0
3
0
.
0
3
0
.
0
0
0
.
0
4
0
.
0
0
0
.
0
0
0
.
0
1
(
-
1
.
0
3
)
5
.
1
6
(
3
.
2
4
)
(
3
.
9
1
)
(
1
.
7
1
)
(
0
.
9
7
)
(
0
.
8
7
)
T
r
e
n
d
2
0
.
0
0
0
.
0
0
(
2
.
0
7
)
-
5
.
3
3
T
r
e
n
d
3
0
.
0
0
0
.
0
0
0
.
0
0
(
-
1
.
5
4
)
(
3
.
0
7
)
(
1
.
3
5
)
R
2
_
b
a
r
0
.
5
2
0
.
6
3
0
.
6
1
0
.
4
7
0
.
2
7
0
.
5
5
0
.
8
9
0
.
6
2
0
.
4
7
0
.
2
7
0
.
4
5
0
.
5
3
D
u
r
b
i
n
W
a
t
s
o
n
1
.
9
3
1
.
8
6
1
.
8
6
1
.
4
4
2
.
0
4
1
.
8
0
1
.
5
3
1
.
1
9
1
.
9
7
1
.
9
5
1
.
8
7
2
.
4
7
F
S
i
g
.
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
2
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
0
O
b
s
.
3
8
3
6
2
4
2
6
3
2
3
1
3
6
6
3
3
6
4
6
4
6
3
6
127
Table 3.9: In?ation Surprises and Liquidation Years
Country
Share of Overlap Liquidation Years
In?ation Suprises CIR HPR
Lower Upper Lower Upper Lower Upper
Argentina 42 17 43 17 32 4
Australia 25 8 42 16
Belgium 8 4 18 9 13 13
France 15 12 20 15
India 9 6 18 6 18 6
Ireland 6 3 10 5
Italy 22 14 39 28 39 28
Japan 14 8 39 22
South Africa 19 8 35 18
Sweden 9 4 18 9 18 9
UK 17 8 29 14 29 14
US 11 9 22 17 17 17
Average 17 8 28 15 24 13
Notes: In?ation suprises are de?ned as years where the realized
in?ation rate is two standard deviations above the estimated expected
in?ation rate. The actual standard errors cannot be obtained but
a lower and upper bound for them.
Table 3.10: Comparison between Ex Ante and Ex Post Yield to Maturity for secu-
rities issued within sample period (in percent)
Country
Nominal YTM at issuance Ex-post real YTM as of time of issuance
Number of bonds
Average Maximum Minimum Average Maximum Minimum
Australia
a
4.2 5.4 2.0 -0.6 4.7 -12.3 50
India
b
5.3 6.5 3.6 -2.3 4.6 -5.0 98
Ireland
c
8.4 14.6 5.4 -2.2 3.5 -10.2 26
a
Bonds issued between 1945-1968
b
Bonds issued between 1960-1978
c
Bonds issued between 1965-1975
128
Table 3.11: Comparison Liquidation E?ect Revenues and In?ation Tax (as %GDP)
Country Subperiod In?ation Tax LE Revenues CIR
G
r
o
u
p
1
Belgium 1945-1954 13.8 23.3
1955-1964 4.1 3
1965-1974 1.1 7.3
India 1949-1959 8.4 6.3
1960-1969 11.5 10.9
1970-1980 15.5 13.9
Ireland 1960-1969 13.8 7.6
1970-1979 20.4 25.5
1980-1990 10.6 11.2
Sweden 1945-1956 6.5 10.5
1957-1968 4.7 2.1
1969-1979 6.8 5.2
1980-1990 5.0 2.8
United Kingdom 1945-1956 6.4 35.7
1957-1968 3.8 1.7
1969-1980 9.4 28.8
G
r
o
u
p
2
Australia 1945-1956 43.6 65.3
1957-1968 7.8 0.5
1969-1980 22.6 14.8
France 1945-1956 74.1 97.5
1957-1968 16.1 0.8
1969-1980 23.7 2.1
Japan 1946-1956 15.3 73.6
1957-1968 28.0 1.1
1969-1980 37.8 6.3
1981-1993 9.1 0.0
1994-2008 1.9 0.1
United States 1945-1956 17.6 26.8
1957-1968 8.5 0.1
1969-1980 21.0 17.0
G
r
o
u
p
3
Argentina 1945-1956 61.4 53.5
1957-1968 42.4 37.8
1969-1980 50.3 40.2
Italy 1946-1956 29.7 19.2
1957-1968 15.8 1.8
1969-1980 117.1 23.4
South Africa 1945-1956 23 12
1957-1968 8 0
1969-1980 24 18
129
Table 3.12: Stock of Debt under di?erent paths for the in?ation rate
Country Period
Debt/GDP Debt/GDP (End)
Initial Actual Median
In?ation
In?=Nominal
Int Rate
In?=2%
(1) (2) (3) (4) (5)
Argentina 1942-1980 42.0 10.4 181.4 2914.2 8080.2
Australia 1945-1980 145.4 18.1 50.3 40.5 98.9
Belgium
a
1945-1974* 112.8 38.4 54.5 45.5 64.2
France
b
1945-1980* 0.0 0.0 0.0 0.0 0.0
India 1949-1980 26.2 25.4 24.3 39.0 71.1
Ireland 1960-1990 61.7 48.4 242.3 53.1 346.4
Italy 1946-1980 67.0 26.2 97.0 58.9 220.4
Japan 1947-1980 27.5 39.1 257.4 137.6 355.9
South Africa 1945-1980 56.6 32.1 28.5 43.3 109.9
Sweden 1945-1990 52.6 39.2 91.0 37.6 224.8
United Kingdom 1945-1980 236.2 44.4 111.6 82.7 211.7
United States 1945-1980 116.0 31.6 50.7 33.8 71.7
Notes: Initial D/GDP refers to the debt ratio in the ?rst year of the sample period for each country, and the
D/GDP (end) is the debt ratio in the last year of the sample.
a
Missing data for 1964-1968
b
Missing data 1953-1958 and 1960-1963
Table 3.13: Panel Estimation Results- Dependent Variable: CIR
(1) (2) (3) (4) (5)
intgdp -0.230*** -0.218*** -0.221*** -0.221*** -0.220***
(-16.07) (-18.81) (-12.84) (-18.70) (-18.74)
defgdp 0.681*** 0.676*** 0.670***
(8.67) (9.59) (9.53)
trgdp 0.0606
(0.62)
stlt 0.0722 0.0750
(1.57) (1.61)
cbind -0.0876
(-1.85)
?nrep
Constant -0.0286 0.0563 -0.0207 0.154*** 0.179***
(-0.72) (1.35) (-0.39) (3.40) (3.88)
Country FE Yes Yes Yes Yes Yes
Time FE Yes Yes Yes Yes Yes
N 446 445 443 434 432
Notes: t-statistics in parenthesis
* p