Description
The purpose of this paper is to show that state economic policies, in addition to state
economic performance, impact state bond ratings.
Journal of Financial Economic Policy
Economic freedom, the cost of public borrowing, and state bond ratings
Peter T. Calcagno J ustin D. Benefield
Article information:
To cite this document:
Peter T. Calcagno J ustin D. Benefield, (2013),"Economic freedom, the cost of public borrowing, and state
bond ratings", J ournal of Financial Economic Policy, Vol. 5 Iss 1 pp. 72 - 85
Permanent link to this document:
http://dx.doi.org/10.1108/17576381311317790
Downloaded on: 24 January 2016, At: 21:46 (PT)
References: this document contains references to 23 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 301 times since 2013*
Users who downloaded this article also downloaded:
Simplice A. Asongu, (2013),"Real and monetary policy convergence: EMU crisis to the CFA zone", J ournal
of Financial Economic Policy, Vol. 5 Iss 1 pp. 20-38 http://dx.doi.org/10.1108/17576381311317763
Wenling Lu, David A. Whidbee, (2013),"Bank structure and failure during the financial crisis", J ournal of
Financial Economic Policy, Vol. 5 Iss 3 pp. 281-299 http://dx.doi.org/10.1108/J FEP-02-2013-0006
Eric Osei-Assibey, Baimba Augustine Bockarie, (2013),"Bank risks, capital and loan supply: evidence from
Sierra Leone", J ournal of Financial Economic Policy, Vol. 5 Iss 3 pp. 256-271 http://dx.doi.org/10.1108/
J FEP-09-2012-0041
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Economic freedom, the cost
of public borrowing, and state
bond ratings
Peter T. Calcagno
Department of Economics and Finance, College of Charleston,
Charleston, South Carolina, USA, and
Justin D. Bene?eld
Department of Finance, Auburn University, Auburn, Alabama, USA
Abstract
Purpose – The purpose of this paper is to show that state economic policies, in addition to state
economic performance, impact state bond ratings.
Design/methodology/approach – Using a sample of 39 states over the period 1998-2008, regression
analysis is employed to determine whether various measures of economic freedom contribute to state
bond ratings.
Findings – After controlling for common factors such as state per-capita income, unemployment,
the ratio of tax revenue to income, state debt as a percentage of government revenue, and
public corruption, results suggest that greater economic freedom is associated with higher
bond ratings. For example, a one standard deviation increase in Area 2 of the Economic Freedom of
North America index (Takings and Taxation) would be associated with a 0.36 increase in Moody’s
bond rating for that state, which translates to approximately a $247 lower cost per million dollars
of debt.
Originality/value – This study contributes to the empirical state bond rating literature
by highlighting that states with greater economic freedom have higher bond ratings and, therefore,
pay lower borrowing costs than their counterparts with lower economic freedom index scores.
Keywords United States of America, Fiscal policy, Bonds, Public ?nance, Borrowing, Bond rating,
Economic freedom, State ?scal policy
Paper type Research paper
Introduction
In the aftermath of the most recent economic crisis, Moody’s Investors Services
(Moody’s) announced their decision to consider ?ve states for possible bond rating
downgrades: Maryland, New Mexico, South Carolina, Tennessee, and Virginia.
The primary point of concern cited by Moody’s was the reliance of these states on
intergovernmental transfers (Everett, 2012). As it happens, these downgrades did not
take place, although the ?ve states were placed in a negative outlook status. However,
in January 2012 Moody’s did downgrade two states: Connecticut and Washington.
As economic conditions for states change with regard to unfunded obligations,
outstanding debt, and revenue received from a federal government with its own bond
rating issues, Moody’s and other credit rating agencies have begun scrutinizing state
bond ratings more closely (Moody’s Investors Service, 2012a, b).
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JEL classi?cation – G10, H71, H74
Journal of Financial Economic Policy
Vol. 5 No. 1, 2013
pp. 72-85
qEmerald Group Publishing Limited
1757-6385
DOI 10.1108/17576381311317790
JFEP
5,1
72
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The issue of sovereign credit ratings on an international level has been explored by
numerous authors, including seminal work in the area by Cantor and Packer (1996).
More recently, Roychoudhury and Lawson (2010) and Qi et al. (2010) have incorporated
measures of economic freedom, or some very similar metric, into the analysis of
international sovereign credit ratings and international corporate bond ratings,
respectively. As noted by Roychoudhury and Lawson (2010), economic freedom has
been considered by relatively fewempirical studies in the ?nance literature. Depken and
Lafountain (2006) conduct an analysis similar to the aforementioned studies of
international credit ratings, but their analysis focuses on US state-level bond ratings and
the negative effects of public corruption on those ratings (and, thus, on the price of
borrowing for states). The current study also considers state bond ratings, but from
a slightly different perspective that incorporates the work done using measures of
economic freedom at the international level. Speci?cally, this study considers whether
changes at the state-level for a state’s economic freedomindexscore might affect a state’s
bond ratings. Holding all else equal, an improved state bond rating should translate into
lower borrowing costs for state governments.
Despite its reliance on the work of Qi et al. (2010) and Roychoudhury and Lawson
(2010), the current study does depart somewhat from the analyses undertaken in those
studies. For example, Qi et al. (2010) focus on the role of political and legal institutions in
determininginternational corporate bondratings. While political andlegal institutions are
certainly very important, this study instead focuses on the role of economic institutions.
Speci?cally, the question of interest is the degree to which measures of economic freedom
in?uence bond ratings. Roychoudhury and Lawson (2010) consider this same research
question, but they use Economic Freedomof the World Index values to analyze sovereign
debt ratings at the international level. Roychoudhury and Lawson (2010) do indeed ?nd
that a higher economic freedom index score lowers the cost of borrowing; unfortunately,
using international data on sovereign debt ratings is fraught with potential unobserved
variable issues. Using bondratingdata fromUS states ameliorates some of these potential
unobserved variable concerns, as the same basic legal and political structure is already in
place across these various sub-national governments. Thus, ?nding higher bond ratings
associatedwithincreasedeconomic freedomusingUSstate-level datawouldsubstantially
bolster the results obtained by Roychoudhury and Lawson (2010).
Despite the similarities in legal and political structure, there exist substantial
variations across states with regard to the type of ?scal policy pursued and the type of
institutional framework utilized. Therefore, the analysis conducted here focuses on
whether different levels of economic freedom in different states can explain observed
variation in US state bond ratings. The remainder of the paper is organized as follows:
the second section provides background from previous studies, the third section
describes the data and the methodology employed, the fourth section presents the
results obtained, and the ?fth section offers concluding remarks.
Literature review
The importance of credit ratings for both sovereign and corporate bonds in driving the
?owof international capital has been a topic of frequent inquiry. The topic has generated
enough interest, in fact, that a full treatment of all related works is beyond the scope of
the current study; however, Reinhart and Rogoff (2004) provide an excellent literature
review on the broad topic of credit ratings and ?ow of funds. Butler and Fauver (2006)
Economic
freedom
73
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and Qi et al. (2010) examine the role played by political and legal institutions in
determining sovereign credit ratings. Both studies ?nd that countries with more stable
legal and political institutions have higher sovereign credit ratings, ceteris paribus.
Without question, legal and political institutions are important in determining
governmental credit ratings; however, the importance of actual economic data from the
governmental unit being rated cannot be overstated, nor can it be overlooked. Cantor
and Packer (1996) present compelling empirical evidence indicating that a country’s
sovereign credit rating re?ects macro-economic conditions in that country. Their study
provides the basis for the macro-economic variables included in many subsequent
studies of governmental credit ratings, including the current study.
In addition, a number of studies, including Liu and Thakor (1984), Bayoumi et al.
(1995), Depken and Lafountain (2006), and Roychoudhury and Lawson (2010), provide
clear evidence of the link between sovereign credit ratings and borrowing costs. Thus,
inferences about borrowing costs can readily be made from results involving credit
ratings. However, of the studies mentioned, only Roychoudhury and Lawson (2010)
speci?cally examines the role of economic freedomin the determination of governmental
credit ratings. As previously mentioned, a focus on state bond ratings for US states
arguably provides a better laboratory to investigate the importance of economic freedom
in governmental bond ratings than does a focus on sovereign bond ratings at the
international level, such as is seen in Roychoudhury and Lawson (2010).
Turning to the literature dealing more speci?cally with state-level government debt,
the focus has primarilybeenonthe determination of bondyields andpremia. Poterba and
Rueben (1999) brie?y review the literature dealing with state bonds as they examine the
role of state ?scal institutions (e.g. budget rules such as limits on expenditure and debt) in
the determination of yields on US municipal bonds. Bayoumi et al. (1995) consider
whether credit markets exhibit a type of self-enforcing discipline mechanism that is
triggered when states increase borrowing levels. Many of the earlier studies analyzing
some aspect of state bond issues, including Goldstein and Woglom(1992), Bayoumi et al.
(1995) and Lowry (2001), use the Advisory Council on Intergovernmental Relations
(ACIR) index to evaluate differences in ?scal policies and budget rules across states.
These earlier studies consistently ?nd that states subject to anti-de?cit provisions tend to
have lower levels of future debt and, related to that lower debt level, reduced credit risk.
More recently, Wagner (2004) introduces an additional ?scal institution, budget
stabilization funds, into the analysis of state-level credit ratings, ?nding that strict
enforcement of budget stabilization funds lowers the cost of borrowing.
Several of the aforementioned studies, including the seminal 1984 study by Liu and
Thakor, make use of bond ratings from the major rating agencies, such as Moody’s or
Standard and Poor’s (S&P). However, the vast majority of studies that consider bond
ratings actually include them as an additional independent variable in explaining
either bond yields or premia over US Treasury debt. Despite the strong evidence cited
above in support of a direct link between state bond ratings and state bond yields, very
few studies have attempted to determine the factors that in?uence state bond ratings
by treating the bond rating as the dependent variable. In this respect, the current study
is most similar to Depken and Lafountain (2006) and Roychoudhury and Lawson
(2010), since both studies treat credit ratings as the dependent variable – the former in
its examination of the impact of state-level data, including public corruption, on state
bond ratings and the latter in its examination of sovereign debt ratings with a focus
JFEP
5,1
74
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
on economic freedom as a key explanatory variable. To the authors’ knowledge, the
only prior study to use economic freedom in examining ?nancial markets at the US
state level is Lawson and Roychoudhury (2008). However, in that study, economic
freedom is used to explain equity markets, rather than debt markets, in US states.
Thus, modeling state bond ratings as a function of state-level economic freedom is a
unique contribution to this strand of the literature.
Data and methodology
Table I provides a list of variable names and de?nitions. For reporting the results,
particularly in the tables, the variable names from Table I are often used. Most of the
necessary data are readily available from public sources. The US Census Bureau web
site provides the data on state tax burdens, state unemployment levels, state average
incomes, state debt levels, state revenues, and state populations. Likewise, state bond
Bond_Rating The appropriate index score combining the state’s S&P, Moody’s, and Fitch’s
bond ratings
Bond_Supp The appropriate index score combining the state’s S&P, Moody’s, and Fitch’s
bond ratings, with missing state bond ratings proxied using largest city bond
ratings when available
State_Tax State tax revenue as a percentage of personal income
Debt_Percent State debt level as a percentage of government revenue
Income_PC Real state per capita income, in thousands (2005 dollars)
Unemp Annual state unemployment
Corruption Federal public corruption convictions per 100,000 population
1998 One if year is 1998, zero otherwise
1999 One if year is 1999, zero otherwise
2000 One if year is 2000, zero otherwise
2001 One if year is 2001, zero otherwise
2002 One if year is 2002, zero otherwise
2003 One if year is 2003, zero otherwise
2004 One if year is 2004, zero otherwise
2005 One if year is 2005, zero otherwise
2006 One if year is 2006, zero otherwise
2007 One if year is 2007, zero otherwise
2008 One if year is 2008, zero otherwise
EFNA – All EFNA Overall Score for all levels of government
Area_1 – All EFNA Size of Government Score for all levels of government
Area_2 – All EFNA Takings and Discriminatory Taxation Score for all levels of government
Area_3 – All EFNA Labor Market Freedom Score for all levels of government
EFNA EFNA Overall Score for state and local government only
Area_1 EFNA Size of Government Score for state and local government only
Area_2 EFNA Takings and Discriminatory Taxation Score for state and local
government only
Area_3 EFNA Labor Market Freedom Score for state and local government only
PRI_Fiscal Paci?c Research Institute Fiscal Policy Score
PRI_Reg Paci?c Research Institute Regulatory Score
PRI_Jud Paci?c Research Institute Judicial Freedom Score
PRI_Gov Paci?c Research Institute Government Size Score
PRI_Wel Paci?c Research Institute Welfare Spending Score
Note: EFNA – Economic Freedom of North America
Table I.
Variable de?nitions
Economic
freedom
75
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ratings, and city bond ratings where applicable, are published in the Statistical
Abstract of the United States, which is also available on the Census Bureau site.
Table II provides summary statistics for these variables, as well as the economic
freedom indices described.
The only necessary data not easily obtained from a public source are the economic
freedom indices. The primary source for data on economic freedom is the
Economic Freedom Index of North America (EFNA) published by the Fraser Institute
(Ashby et al., 2010). EFNA is an index measured on a scale from zero to ten with
three sub-categories that combine to form the overall index. Area 1 measures Size
of Government, Area 2 measures Takings and Discriminatory Taxation, and Area
3 measures Labor Market Freedom. A higher overall index value indicates a greater
degree of economic freedom within the state. The overall index levels and the three area
scores are analyzed in separate models to determine what impact they may have on state
bond ratings. In addition to these comprehensive indices, the Fraser Institute also
publishes an EFNA index and the same three area indices using values solely from
Mean Median SD Min. Max.
Bond_Rating 0.920 0.909 0.055 0.634 1.000
Bond_Supp 0.921 0.909 0.055 0.634 1.000
State_Tax 0.067 0.065 0.016 0.029 0.284
Debt_Percent 0.402 0.353 0.296 0.013 1.681
Income_PC 3.302 3.227 0.519 2.284 5.258
Unemp 4.694 4.604 1.143 2.258 8.300
Corruption 0.342 0.267 0.317 0.000 2.552
1998 0.091 0.000 0.288 0.000 1.000
1999 0.091 0.000 0.288 0.000 1.000
2000 0.091 0.000 0.288 0.000 1.000
2001 0.091 0.000 0.288 0.000 1.000
2002 0.091 0.000 0.288 0.000 1.000
2003 0.091 0.000 0.288 0.000 1.000
2004 0.091 0.000 0.288 0.000 1.000
2005 0.091 0.000 0.288 0.000 1.000
2006 0.091 0.000 0.288 0.000 1.000
2007 0.091 0.000 0.288 0.000 1.000
2008 0.091 0.000 0.288 0.000 1.000
EFNA – All 6.896 6.894 0.548 5.206 8.478
Area_1 – All 7.186 7.246 0.940 2.494 9.097
Area_2 – All 6.266 6.263 0.632 3.540 8.463
Area_3 – All 7.237 7.264 0.555 5.659 8.416
EFNA 7.179 7.236 0.612 5.353 8.394
Area_1 7.201 7.329 0.942 3.744 8.894
Area_2 7.315 7.374 0.664 5.589 8.957
Area_3 7.022 6.906 0.728 5.702 8.828
PRI_Fiscal 23.920 23.715 3.686 16.570 31.460
PRI_Reg 13.480 13.070 2.300 9.956 19.090
PRI_Jud 13.792 13.765 3.734 7.667 23.330
PRI_Gov 25.446 25.217 6.228 10.710 37.714
PRI_Wel 25.280 25.840 7.080 9.220 39.220
Notes: EFNA – Economic Freedom of North America; PRI – Paci?c Research Institute
Table II.
Summary statistics
JFEP
5,1
76
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
each state government, i.e. net of federal government transfers to the state. The
comprehensive EFNA and area indices that include federal government transfers are
denoted by the addition of the label “ALL” (e.g. EFNA – ALL, AREA_1 – ALL, etc.),
while the EFNA and area indices that exclude federal government transfers are simply
labeled EFNA, AREA_1, etc. The economic freedom index values net of federal
government transfers are also analyzed in a second set of models to ascertain whether
the in?uence of federal government transfers on state economic freedom rankings is
important in determining state bond ratings. EFNA is highly correlated with economic
growth and well-being, which could also be measured using state income and
state unemployment. Since income and unemployment are two of the other independent
variables speci?ed in the models, only lagged values of the economic freedomindices are
included in an attempt to minimize potential multicollinearity and endogeneity issues.
As noted in the literature review, prior studies of state level bond markets have used
a variety of ?scal institutions to estimate bond yields. While most prior studies have
not included EFNA as an independent variable, a strong argument along two
dimensions can be made regarding its appropriateness in models of bond yields and
ratings. First, states that have greater economic freedomare more likely to have adopted
anti-de?cit ?scal institutions of the type often measured in prior literature by the ACIR
index[1]. Several of the ?scal institutions can have multiple interpretations based on the
rules associated with that particular institution in any given state. For instance, the
“balanced budget rule” cited in several prior studies is actually an amalgamation of
several rules of varying degrees of restrictiveness ranging from a simple requirement
that the governor propose a balanced budget to a strict prohibition that the state cannot
carry a de?cit. Thus, proper interpretation of discrete ?scal institutions variables is
dependent upon having all potential rule andimplementation variations accounted for in
the model[2]. This potential speci?cation issue is exactly what the ACIR index was
designed to address. Given that the EFNA index is also a continuous variable, like the
ACIR, it should also serve to limit potential speci?cation problems that might arise if
economic freedom were measured via discrete variables, as would be necessary in the
absence of the EFNA index.
Second, there are two different aspects of institutional quality through which
economic freedom can affect bond ratings. The more important of these two aspects is
that states with anti-de?cit ?scal institutions would be more likely to exhibit greater
economic freedom. The other facet of institutional quality through which economic
freedom might impact bond ratings is found in Hall and Sobel (2007), Campbell and
Rodgers (2007) and Campbell et al. (2007). Speci?cally, states that have greater
economic freedom (i.e. higher institutional quality) exhibit higher rates of economic
growth, personal income, and entrepreneurship as measured by ?rm births and deaths.
Granted, these ?scal institutions would have a more indirect effect on bond ratings, but
the effect would still be potentially relevant to the determination of state bond ratings.
For these two reasons, EFNA should be seen as a broader measure of institutional
quality that re?ects more than merely ?scal conditions and ?scal institutions. In this
way, the use of EFNA is an important contribution to the bond rating literature.
As an additional robustness check, the state level economic freedom index
published by the Paci?c Research Institute (PRI) is used as an alternative measure
of economic freedom. Since this index is not published on an annual basis, the
most recent PRI index scores from 2008 are used to estimate a cross-sectional model
Economic
freedom
77
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
for that year. The PRI index evaluates economic freedom broken down into ?ve
categories, or sectors: ?scal, regulatory, judicial, government size, and welfare spending.
The individual sector scores used to compute the overall PRI index have various ranges
depending in large part on how many indicators were used in computing that sector
score and how the indicators were ultimately weighted. It should be noted that the
PRI index does not assign scores in the same way as the EFNA, in that lower PRI
index values mean greater economic freedom. The interested reader is referred to
McQuillan et al. (2008) for additional information.
The statistical methodology employed is very straightforward. Using state-level
data for the years 1998-2008, ordinary least squares (OLS) regression is utilized in the
?rst set of estimations to analyze the relationship between the chosen independent
variables and state bond ratings. Since the index of bond ratings used is bounded from
above at 1.000, a censored regression, or tobit estimation, is utilized as a robustness
check on the results observed from the OLS estimation. Both statistical techniques are
very well-understood and exceedingly common in the literature. Therefore, little time is
devoted to an explanation of their use. The speci?c OLS models estimated are:
Bond_Rating
i;t
¼ b
0
þ b
1
X
i;t
þ b
2
Corruption
i;t
þ b
3
Year
t
þb
4
EFNA
i;t 21
þ 1
i;t
ð1Þ
Bond_Rating
i;t
¼ b
0
þ b
1
X
i;t
þ b
2
Corruption
i;t
þ b
3
Year
t
þb
4
EFNA_Area
k;i;t 21
þ 1
i;t
ð2Þ
where Bond_Rating
i,t
is the bond rating for state i at time t, X
i,t
is a vector of state
economic control variables as further de?ned in Table I, Corruption
i,t
is the number
of federal convictions of public of?cials, EFNA
i,t21
is the lagged value of the overall
economic freedom index and EFNA_Area
k,i,t21
is the lagged value of economic
freedom for area k (where k ¼ 1 2 3), Year
t
is an indicator variable for the year, and
1
i,t
is the normally distributed error term.
The methodology for computing the state bond rating index (Bond_Rating) is,
however, not a technique that can simply be mentioned and dismissed. To begin, the
individual state bond ratings are collected for each sample year (i.e. the rating at the
end of the last quarter of the year) for each state from the three major bond rating
providers: S&P, Moody’s, and Fitch Ratings (Fitch’s). S&P has a total of 22 rating
designations, Moody’s has 21, and Fitch’s has 20; the complete list of all possible
ratings for each agency is provided in the Appendix.
Usingthe same ratingscheme as DepkenandLafountain(2006), eachratingdesignation
is assigned a number value, with one denoting the lowest rating designation used by that
rating provider. For example, the lowest rating designation given by S&P (“D”) would be
assigned a value of one and the highest rating designation given by S&P (“AAA”) would
be assigned a value of 22, since S&Phas 22 total ratingdesignations. These cardinal scores
are then normalized by the number of possible rankings for each provider. The ?nal step is
to average the available normalized scores for each state. It shouldbe noted here that not all
states have debt issues that are ratedinall sample years (i.e. a state couldhave zero ratings).
Also, ratings may be available from one, two, or all three ratings providers.
In an effort to present as complete a picture as possible, and as a way of checking
the robustness of the primary results, bond ratings for the largest city in a state were
substituted if no state bond rating was available in a particular sample year[3].
JFEP
5,1
78
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Despite this attempt to include all states in the analysis, there are still some missing
observations. In some cases, the largest city may not have had debt outstanding
(or rated), while in other cases, the largest city may not have been large enough to be
reported in the Statistical Abstract of the United States (e.g. Idaho).
We estimate equation (3) using a tobit model:
Bond_Rating
*
i;t
¼ b
0
þ b
1
X
i;t
þ b
2
Corruption
i;t
þ b
3
Year
t
þ b
4
EFNA_Area
k;i;t 21
þ 1
i;t
ð3Þ
where:
Bond_Rating
*
i;t
¼
Bond_Rating
i;t
; if Bond_Rating
i;t
. 0
0; if Bond_Rating
i;t
# 0
8
<
:
It is assumed that 1 , N(0,s
2
I), and Bond_Rating
*
i,t
is the latent variable used in the
tobit estimation. The same set of independent variables is utilized as in equation
and (2), replacing the dependent variable Bond_Rating with Bond_Rating
*
i,t
.
Results
Table III reports OLS results for state bond ratings using the various economic freedom
indices described. Model 1 uses the comprehensive EFNA index, which includes federal
government spendingat the state level. Model 2 uses the three area scores underlying the
comprehensive EFNA index, Models 3 and 4 include the comprehensive and
area-speci?c EFNA index scores net of federal government transfers, and Model 5 is
the cross-sectional analysis utilizing the PRI sector scores for 2008. As can be seen in the
table, results on the control variables are very consistent across the models using the
various EFNA indices. A higher state debt level relative to state government revenue,
a higher unemployment rate, a higher level of public corruption, and a lower percent
of state tax revenue relative to personal income are all associated with lower state bond
ratings, with the only exception being state tax revenue as a percent of personal income
in Model 1, which is insigni?cant. Per capita income is not related to state bond ratings in
three of the models, but it is marginally signi?cantly positively related to bond ratings in
Model 2. These ?ve control variables do not register as signi?cant in Model 5, which is
not surprising. Recall that Model 5 uses the PRI index for 2008, which results ina sharply
reduced sample size.
Turning to the variables of interest, the economic freedom indices, Table III shows
that, regardless of the index chosen, there is at least some positive and statistically
signi?cant impact on state bond ratings. There is no difference between results using the
EFNA index whether federal government expenditures at the state level are considered
(Model 1) or not (Model 3). In both cases, a higher EFNA index value is highly
signi?cantly related to a higher state bondrating, which is as expected, given that higher
EFNA index values indicate greater economic freedom. Using the EFNA area-speci?c
scores, some difference is noted depending on whether federal government spending
at the state level is included (Model 2) or not (Model 4). Speci?cally, when federal
government spending at the state level is excluded (Model 4), Size of Government
(Area 1) moves from insigni?cant in Model 2 to marginally signi?cant and positive in
Model 4 and Labor Market Freedom (Area 3) changes from statistically signi?cantly
Economic
freedom
79
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
positive in Model 2 to insigni?cant in Model 4. Area 2, which measures Takings and
Discriminatory Taxation, is highly statistically signi?cant in both models. Changes in
signi?cance for these variables based solely on whether federal government spending at
the state level is included when the area scores are computed indicates that federal
government spending can substantially in?uence economic freedom at the individual
state level. In Model 5, the only two sectors that enter the model as signi?cant predictors
of state bond ratings are the Regulation and Welfare Spending sectors. It is worthy of
note that both of thementer with the expected negative signs (recall that in the PRI index
sector scores, a lower value indicates greater economic freedom).
Table IV presents an alternative approach to estimating the relationship between
the independent variables and state bond ratings. The bond rating classi?cation scheme
described in the data and methodology section produces an index of bond ratings that is
bounded from above at 1.000. Since OLS is known to produce inconsistent parameter
Variable Model 1 Model 2 Model 3 Model 4 Model 5
Constant 0.798
* * *
0.624
* * *
0.721
* * *
0.716
* * *
1.046
* * *
State_Tax 0.161 0.371
* *
0.413
* * *
0.475
* * *
0.052
Debt_Percent 20.039
* * *
20.041
* * *
20.037
* * *
20.041
* * *
20.165
Income_PC 20.008 0.011
*
0.004 0.001 20.009
Unemp 20.013
* * *
20.009
* * *
20.009
* * *
20.012
* * *
20.001
Corruption 20.030
* * *
20.039
* * *
20.039
* * *
20.043
* * *
20.011
1999 0.002 0.002 0.001 0.001
2000 0.006 0.003 0.005 0.004
2001 0.010 0.002 0.005 0.006
2002 0.016 20.003 0.013 0.015
2003 0.010 20.015 0.009 0.010
2004 0.002 20.026
* *
0.004 0.004
2005 20.001 20.034
* * *
0.002 0.004
2006 20.020
*
20.055
* * *
20.020
*
20.019
*
2007 20.014 20.051
* * *
20.017 20.015
2008 0.005 20.035
* *
20.002 0.002
EFNA – All 0.032
* * *
Area_1 – All 20.006
Area_2 – All 0.018
* * *
Area_3 – All 0.036
* * *
EFNA 0.032
* * *
Area_1 0.006
*
Area_2 0.024
* * *
Area_3 0.005
PRI_Fiscal 0.001
PRI_Reg 20.007
* *
PRI_Jud 0.003
PRI_Gov 0.001
PRI_Wel 20.002
*
R
2
(%) 24.2 27.5 25.2 27.0 28.9
Adjusted R
2
(%) 21.7 24.9 22.8 24.4 10.7
F-statistic 9.99
* * *
10.55
* * *
10.56
* * *
10.28
* * *
1.59
n 519 519 519 519 50
Note: Signi?cant at:
*
10,
* *
5 and
* * *
1 percent levels
Table III.
OLS regression results
for state bond ratings
using various economic
freedom indices
JFEP
5,1
80
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
estimates when dealing with a bounded dependent variable, censored regression, or tobit
estimations, are utilized as an alternative. Given space limitations, tobit results are
reported for onlytwo models: the EFNAindex usingarea-speci?c scores withand without
federal government spending at the state level, or Models 2 and 4, respectively. As can be
seen in the table, the tobit results for the control variables are entirely consistent with the
OLS results from Table III. Area 1 (Size of Government) is marginally signi?cantly
negative in Model 2 and marginally signi?cantly positive in Model 4. These results
suggest that federal government transfers are particularly important when considering
size of state government. Speci?cally, when including federal government transfers, it
seems that greater Area 2 economic freedom index scores still suffer because of the
perception that federal monies are, at least to some extent, subsidizing the size of state
government. When state spending is considered in the absence of federal
government transfers, smaller government seems to be more closely associated with
Gain/loss in bond rating from 1 SD change
S&P Moody’s Fitch’s
Variable
Model 2 marginal
impact
Model 4 marginal
impact
Model
2
Model
4
Model
2
Model
4
Model
2
Model
4
State_Tax 0.410
* * *
0.514
* * *
0.15 0.18 0.14 0.17 0.13 0.17
Debt_Percent 20.043
* * *
20.042
* * *
20.28 20.27 20.27 20.26 20.26 20.25
Income_PC 0.013
*
0.001 0.15 0.02 0.14 0.01 0.13 0.01
Unemp 20.010
* * *
20.013
* * *
20.26 20.32 20.25 20.30 20.24 20.29
Corruption 20.040
* * *
20.043
* * *
20.28 20.30 20.27 20.28 20.25 20.27
1999 0.002 0.000 0.01 0.00 0.01 0.00 0.01 0.00
2000 0.002 0.003 0.01 0.02 0.01 0.02 0.01 0.02
2001 0.001 0.004 0.00 0.03 0.00 0.03 0.00 0.03
2002 20.005 0.014 20.00 0.09 20.00 0.09 20.00 0.08
2003 20.020 0.010 20.13 0.06 20.12 0.06 20.11 0.06
2004 20.033
* *
0.003 20.21 0.02 20.20 0.02 20.19 0.02
2005 20.043
* * *
0.001 20.27 0.01 20.26 0.01 20.25 0.01
2006 20.067
* * *
20.024
*
20.42 20.15 20.40 20.14 20.38 20.14
2007 20.062
* * *
20.019 20.39 20.12 20.37 20.12 20.36 20.11
2008 20.044
* * *
20.002 20.28 20.01 20.27 20.01 20.25 20.01
Area_1 – All 20.007
*
20.15 20.15 20.14
Area_2 – All 0.022
* * *
0.31 0.29 0.28
Area_3 – All 0.040
* * *
0.49 0.47 0.45
Area_1 0.006
*
0.13 0.13 0.12
Area_2 0.026
* * *
0.38 0.36 0.34
Area_3 0.007
*
0.11 0.10 0.10
Log-
likelihood 567.99 562.59
Pseudo R
2
(%) 18.16 17.03
Test for zero
slopes 174.57
* * *
163.77
* * *
n 519 519
Notes: Signi?cant at:
*
10,
* *
5 and
* * *
1 percent levels; the gain or loss in bond rating is computed as
the marginal impact times the standard deviation times the number of ratings for a rating agency; for
example, the loss in S&P bond rating from a one standard deviation increase in unemployment equals
20.010 £ 1.143 £ 22 ¼ 0.26 for Model 2
Table IV.
Tobit estimation results
for state bond ratings
using state-level
Economic Freedom of
North America indices
Economic
freedom
81
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
greater ?scal responsibility and higher bond ratings. Area 3 (Labor Market Freedom) is
marginally signi?cantly positive in Model 4. In all three of these marginally signi?cant
cases, the marginal impact of a one standard deviation change in that area index score is
economically very small. In the other three cases (Area 2 in both models and Area 3 in
Model 2), the economic impact varies fromroughlyone-third of a bondratingto one-half of
a bond rating for a one standard deviation change in the area index score.
To illustrate, Table IV includes estimates from the Depken and Lafountain (2006)
method of examining the economic impact of increasing the EFNAindex by one standard
deviation above the mean. Rubinfeld (1973) demonstrated that a rating change from
Moody’s AAA to Moody’s AA status (at the time of his writing, Moody’s did not
differentiate within investment grades) for a municipal bond issuer correlated with an
increase in interest rates of approximately 20.6 basis points. Using the current
Moody’s rating system, this would indicate an approximate 6.87 basis point increase for a
Moody’s rating downgrade from Aaa to Aa1. For comparison purposes regarding the
scale of ratings changes associated with economic freedom, Table IV shows a one
standard deviation increase in public corruption above the national mean would result in
a 20.27 change in the Moody’s state bond rating, which would increase the cost of debt
service by approximately $185 per year per million dollars of debt (assuming an original
interest rate of 5 percent). Similarly, a one standard deviation increase in Area 2 (Takings
and Taxation) would result in a 0.36 increase in Moody’s state rating. Thus, greater
economic freedom related to takings and taxation would lower the cost of debt by
approximately $247 per year per million dollars of debt.
Table V, reports results fromthe same EFNAmodels (Models 1-4) speci?edinTable III.
However, the bond rating index is augmented via the substitution of the bond rating for
the largest city in a state if the state has no bond rating available. This is admittedly an
imperfect solution, but the exclusion of several states from the estimation is, arguably, a
more troubling issue. Unfortunately, even allowing this substitution does not enable the
inclusion of all states in the estimation. For several states that have no bond rating, there
are either no cities with bond ratings or no cities large enough to have their bond ratings
reported in any easily accessible manner. Nonetheless, the sample was increased by a
relative few observations, and the results can be seen in Table V. As shown there, the
EFNA index and various area-speci?c scores perform exactly the same as in Table III.
Conclusions
This study estimates the impact of various indices of economic freedom on US
state-level bond ratings. Using OLS and censored regression techniques, the Economic
Freedom of North America index, its three area components, and the ?ve sector scores
underlying the PRI economic freedom index are found to signi?cantly impact state
bond ratings. The results provide substantial support for the notion that states
exhibiting greater levels of economic freedom also enjoy higher bond ratings, with the
associated lower borrowing costs that such ratings entail. Comparison of the results
using area-speci?c EFNA scores shows that federal government expenditures at the
state level can in?uence the manner in which the three areas of economic freedom
impact state bond ratings. Policy implications are quite clear. First, states that pursue
policies aimed at achieving greater economic freedom will be favorably positioned
to receive higher bond ratings than they would in the absence of such policies.
Second, state politicians should be cognizant of the impact of their decisions about
JFEP
5,1
82
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
which federal monies to take, and the manner in which those funds are taken, on their
state’s economic freedom, and thus, their state’s bond ratings.
Notes
1. The Advisory Commission on Intergovernmental Relations (ACIR), which produced the ACIR
indexonanti-de?cit ?scal institutions, nolonger exists. It is nowknownas the AmericanCouncil
onIntergovernmental Relations. Inaddition, the Signi?cant Features of Fiscal Federalismreport,
which provided much of the data in the ACIR index, has not been published since 1995.
2. Wagner and Sobel (2006) ?nd that states with already-existing tax and expenditure limitations
were more likelyto adopt statutorybudget stabilizationfunds but were less likelyto adopt such
funds if they also included stringent deposit and withdrawal rules. This suggests that at least
some budget stabilization funds were adopted to circumvent existing ?scal constraints.
3. It should be noted that 11 states do not issue general obligation bonds. These states
include Arizona, Arkansas, Colorado, Idaho, Indiana, Iowa, Kansas, Kentucky, Nebraska,
South Dakota, and Wyoming. These 11 states are the ones for which an attempt is made
to proxy the state’s bond rating using the bond rating for the state’s largest city.
Variable Model 1 Model 2 Model 3 Model 4
Constant 0.803
* * *
0.640
* * *
0.731
* * *
0.730
* * *
State_Tax 0.173 0.372
* *
0.414 0.467
* * *
Debt_Percent 20.039
* * *
20.041
* * *
20.038
* * *
20.042
* * *
Income_PC 20.008 0.009 0.004 20.000
Unemp 20.013
* * *
20.010
* * *
20.010
* * *
20.012
* * *
Corruption 20.030
* * *
20.040
* * *
20.039
* * *
20.043
* * *
1999 0.002 0.002 0.001 0.001
2000 0.006 0.003 0.005 0.004
2001 0.011 0.003 0.006 0.007
2002 0.018 0.001 0.015 0.017
*
2003 0.012 20.012 0.012 0.013
2004 0.004 20.022
*
0.006 0.007
2005 0.001 20.030
* *
0.005 0.007
2006 20.019
*
20.052
* * *
20.019
*
20.018
2007 20.013 20.048
* * *
20.016 20.013
2008 0.006 20.031
* *
20.000 0.003
EFNA – All 0.032
* * *
Area_1 – All 20.005
Area_2 – All 0.017
* * *
Area_3 – All 0.035
* * *
EFNA 0.032
* * *
Area_1 0.007
* *
Area_2 0.023
* * *
Area_3 0.004
R
2
(%) 25.6 28.6 26.5 28.2
Adjusted R
2
(%) 23.3 26.1 24.2 25.7
F-statistic 11.16
* * *
11.47
* * *
11.67
* * *
11.26
* * *
n 535 535 535 535
Notes: Signi?cant at:
*
10,
* *
5 and
* * *
1 percent levels; results using the Paci?c Research Institute
measures of economic freedom are not included since no state bond rating proxies were used in the
?nal sample year
Table V.
OLS regression results
for supplemented state
bond ratings using
various economic
freedom indices
Economic
freedom
83
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
References
Ashby, N.J., Karabegovic´, A., McMahon, F. and Bueno, A. (2010), Economic Freedom of North
America 2010, Fraser Institute, Vancouver.
Bayoumi, T., Goldstein, M. and Woglom, G. (1995), “Do credit markets discipline sovereign
borrowers? Evidence from US States”, Journal of Money, Credit, and Banking, Vol. 27,
pp. 1046-59.
Butler, A.W. and Fauver, L. (2006), “Institutional environment and sovereign credit ratings”,
Financial Management, Vol. 35, pp. 53-79.
Campbell, N.D. and Rodgers, T.M. (2007), “Economic freedom and net business formation”, Cato
Journal, Vol. 27 No. 1, pp. 23-36.
Campbell, N.D., Rogers, T.M. andHeriot, K.C. (2007), “The economic freedomindexas adeterminant
of ?rm births and ?rm deaths”, Southwest Business and Economics Journal, pp. 37-50.
Cantor, R. and Packer, F. (1996), “Determinants and impact of sovereign credit ratings”, Federal
Reserve Bank of New York Economic Polciy Review, pp. 37-53.
Depken, C.A. and Lafountain, C.L. (2006), “Fiscal consequences of public corruption: empirical
evidence from state bond ratings”, Public Choice, Vol. 126, pp. 75-85.
Everett, B. (2012), Politico, March 8, available at: Politico.com:dyn.politico.com
Goldstein, M. and Woglom, G. (1992), “Market-based ?scal discipline in monetary unions:
evidence from the US municipal bond market”, in Canzoneri, M., Grilli, V. and
Masson, P. (Eds), Establishing a Central Bank: Issues in Europe and Lessons from the
United States, Cambridge University, Cambridge.
Hall, J.C. and Sobel, R.S. (2007), “Institutions, entrepreneurship, and regional differences in
economic growth”, Southern Journal of Entrepreneurship, Vol. 1, pp. 70-96.
Lawson, R.A. and Roychoudhury, S. (2008), “Economic freedom and equity prices among US
States”, The Credit and Financial Management Review, Vol. 14, pp. 25-35.
Liu, P. and Thakor, A.V. (1984), “Interest yields, credit ratings, andeconomic characteristics of state
bonds: an empirical analysis”, Journal of Money, Credit and Banking, Vol. 16, pp. 344-51.
Lowry, R.C. (2001), “A visible hand? Bond markets, political parties, balanced budget laws, and
state government debt”, Economics and Politics, Vol. 13, pp. 49-72.
McQuillan, L.J., Maloney, M.T., Daniels, E. and Eastwood, B.M. (2008), US Economic Freedom
Index: 2008 Report, Paci?c Research Institute, San Francisco, CA.
Moody’s Investors Service (2012a), Moody’s Downgrades State of Connecticut General Obligation
Bonds To Aa3 From Aa2 Rating and Assigns Stable Outlook, January 20, available at:
www.moodys.com
Moody’s Investors Service (2012b), Moody’s Revises State of Washington Rating Outlook to
Negative from Stable and Af?rms Aa1G.O, January 30, available at: www.moodys.com
Poterba, J.M. and Rueben, K. (1999), “State ?scal institutions and the US municipal bond market”,
in Poterba, J.M. (Ed.), Fiscal Institutions and Fiscal Performance, University of Chicago,
Chicago, IL.
Qi, Y., Roth, L. and Wald, J.K. (2010), “Political rights and the cost of debt”, Journal of Financial
Economics, Vol. 95, pp. 202-26.
Reinhart, C.M. and Rogoff, K.S. (2004), “Serial default and the “Paradox” of rich-to poor capital
?ows”, American Economic Review, Vol. 94, pp. 53-9.
Roychoudhury, S. and Lawson, R.A. (2010), “Economic freedom and sovereign credit ratings and
default risk”, Journal of Financial Economic Policy, Vol. 2, pp. 149-62.
JFEP
5,1
84
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Rubinfeld, D.L. (1973), “Credit ratings and the market for general obligation municipal bonds”,
National Tax Journal, Vol. 26 No. 1, pp. 17-27.
Wagner, G.A. (2004), “The bond market and ?scal institutions: have budget stabilization funds
reduced state borrowing costs?”, National Tax Journal, Vol. 57 No. 4, pp. 785-804.
Wagner, G.A. and Sobel, R.S. (2006), “State budget stabilization fund adoption: preparing for the
next recession or circumventing ?scal constraints?”, Public Choice, Vol. 126, pp. 177-99.
Appendix
Corresponding author
Justin D. Bene?eld can be contacted at: [email protected]
S&P Moody’s Fitch’s
AAA Aaa AAA
AA þ Aa1 AA þ
AA Aa2 AA
AA 2 Aa3 AA 2
A þ A1 A þ
A A2 A
A 2 A3 A 2
BBB þ Baa1 BBB þ
BBB Baa2 BBB
BBB 2 Baa3 BBB 2
BB þ Ba1 BB þ
BB Ba2 BB
BB 2 Ba3 BB 2
B þ B1 B þ
B B2 B
B 2 B3 B 2
CCC þ Caa1 CCC
CCC Caa2 CC
CCC 2 Caa3 C
CC Ca D
C C
D
Table AI.
Bond ratings used
by the three major
rating agencies
Economic
freedom
85
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_705910591.pdf
The purpose of this paper is to show that state economic policies, in addition to state
economic performance, impact state bond ratings.
Journal of Financial Economic Policy
Economic freedom, the cost of public borrowing, and state bond ratings
Peter T. Calcagno J ustin D. Benefield
Article information:
To cite this document:
Peter T. Calcagno J ustin D. Benefield, (2013),"Economic freedom, the cost of public borrowing, and state
bond ratings", J ournal of Financial Economic Policy, Vol. 5 Iss 1 pp. 72 - 85
Permanent link to this document:
http://dx.doi.org/10.1108/17576381311317790
Downloaded on: 24 January 2016, At: 21:46 (PT)
References: this document contains references to 23 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 301 times since 2013*
Users who downloaded this article also downloaded:
Simplice A. Asongu, (2013),"Real and monetary policy convergence: EMU crisis to the CFA zone", J ournal
of Financial Economic Policy, Vol. 5 Iss 1 pp. 20-38 http://dx.doi.org/10.1108/17576381311317763
Wenling Lu, David A. Whidbee, (2013),"Bank structure and failure during the financial crisis", J ournal of
Financial Economic Policy, Vol. 5 Iss 3 pp. 281-299 http://dx.doi.org/10.1108/J FEP-02-2013-0006
Eric Osei-Assibey, Baimba Augustine Bockarie, (2013),"Bank risks, capital and loan supply: evidence from
Sierra Leone", J ournal of Financial Economic Policy, Vol. 5 Iss 3 pp. 256-271 http://dx.doi.org/10.1108/
J FEP-09-2012-0041
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Economic freedom, the cost
of public borrowing, and state
bond ratings
Peter T. Calcagno
Department of Economics and Finance, College of Charleston,
Charleston, South Carolina, USA, and
Justin D. Bene?eld
Department of Finance, Auburn University, Auburn, Alabama, USA
Abstract
Purpose – The purpose of this paper is to show that state economic policies, in addition to state
economic performance, impact state bond ratings.
Design/methodology/approach – Using a sample of 39 states over the period 1998-2008, regression
analysis is employed to determine whether various measures of economic freedom contribute to state
bond ratings.
Findings – After controlling for common factors such as state per-capita income, unemployment,
the ratio of tax revenue to income, state debt as a percentage of government revenue, and
public corruption, results suggest that greater economic freedom is associated with higher
bond ratings. For example, a one standard deviation increase in Area 2 of the Economic Freedom of
North America index (Takings and Taxation) would be associated with a 0.36 increase in Moody’s
bond rating for that state, which translates to approximately a $247 lower cost per million dollars
of debt.
Originality/value – This study contributes to the empirical state bond rating literature
by highlighting that states with greater economic freedom have higher bond ratings and, therefore,
pay lower borrowing costs than their counterparts with lower economic freedom index scores.
Keywords United States of America, Fiscal policy, Bonds, Public ?nance, Borrowing, Bond rating,
Economic freedom, State ?scal policy
Paper type Research paper
Introduction
In the aftermath of the most recent economic crisis, Moody’s Investors Services
(Moody’s) announced their decision to consider ?ve states for possible bond rating
downgrades: Maryland, New Mexico, South Carolina, Tennessee, and Virginia.
The primary point of concern cited by Moody’s was the reliance of these states on
intergovernmental transfers (Everett, 2012). As it happens, these downgrades did not
take place, although the ?ve states were placed in a negative outlook status. However,
in January 2012 Moody’s did downgrade two states: Connecticut and Washington.
As economic conditions for states change with regard to unfunded obligations,
outstanding debt, and revenue received from a federal government with its own bond
rating issues, Moody’s and other credit rating agencies have begun scrutinizing state
bond ratings more closely (Moody’s Investors Service, 2012a, b).
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JEL classi?cation – G10, H71, H74
Journal of Financial Economic Policy
Vol. 5 No. 1, 2013
pp. 72-85
qEmerald Group Publishing Limited
1757-6385
DOI 10.1108/17576381311317790
JFEP
5,1
72
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The issue of sovereign credit ratings on an international level has been explored by
numerous authors, including seminal work in the area by Cantor and Packer (1996).
More recently, Roychoudhury and Lawson (2010) and Qi et al. (2010) have incorporated
measures of economic freedom, or some very similar metric, into the analysis of
international sovereign credit ratings and international corporate bond ratings,
respectively. As noted by Roychoudhury and Lawson (2010), economic freedom has
been considered by relatively fewempirical studies in the ?nance literature. Depken and
Lafountain (2006) conduct an analysis similar to the aforementioned studies of
international credit ratings, but their analysis focuses on US state-level bond ratings and
the negative effects of public corruption on those ratings (and, thus, on the price of
borrowing for states). The current study also considers state bond ratings, but from
a slightly different perspective that incorporates the work done using measures of
economic freedom at the international level. Speci?cally, this study considers whether
changes at the state-level for a state’s economic freedomindexscore might affect a state’s
bond ratings. Holding all else equal, an improved state bond rating should translate into
lower borrowing costs for state governments.
Despite its reliance on the work of Qi et al. (2010) and Roychoudhury and Lawson
(2010), the current study does depart somewhat from the analyses undertaken in those
studies. For example, Qi et al. (2010) focus on the role of political and legal institutions in
determininginternational corporate bondratings. While political andlegal institutions are
certainly very important, this study instead focuses on the role of economic institutions.
Speci?cally, the question of interest is the degree to which measures of economic freedom
in?uence bond ratings. Roychoudhury and Lawson (2010) consider this same research
question, but they use Economic Freedomof the World Index values to analyze sovereign
debt ratings at the international level. Roychoudhury and Lawson (2010) do indeed ?nd
that a higher economic freedom index score lowers the cost of borrowing; unfortunately,
using international data on sovereign debt ratings is fraught with potential unobserved
variable issues. Using bondratingdata fromUS states ameliorates some of these potential
unobserved variable concerns, as the same basic legal and political structure is already in
place across these various sub-national governments. Thus, ?nding higher bond ratings
associatedwithincreasedeconomic freedomusingUSstate-level datawouldsubstantially
bolster the results obtained by Roychoudhury and Lawson (2010).
Despite the similarities in legal and political structure, there exist substantial
variations across states with regard to the type of ?scal policy pursued and the type of
institutional framework utilized. Therefore, the analysis conducted here focuses on
whether different levels of economic freedom in different states can explain observed
variation in US state bond ratings. The remainder of the paper is organized as follows:
the second section provides background from previous studies, the third section
describes the data and the methodology employed, the fourth section presents the
results obtained, and the ?fth section offers concluding remarks.
Literature review
The importance of credit ratings for both sovereign and corporate bonds in driving the
?owof international capital has been a topic of frequent inquiry. The topic has generated
enough interest, in fact, that a full treatment of all related works is beyond the scope of
the current study; however, Reinhart and Rogoff (2004) provide an excellent literature
review on the broad topic of credit ratings and ?ow of funds. Butler and Fauver (2006)
Economic
freedom
73
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and Qi et al. (2010) examine the role played by political and legal institutions in
determining sovereign credit ratings. Both studies ?nd that countries with more stable
legal and political institutions have higher sovereign credit ratings, ceteris paribus.
Without question, legal and political institutions are important in determining
governmental credit ratings; however, the importance of actual economic data from the
governmental unit being rated cannot be overstated, nor can it be overlooked. Cantor
and Packer (1996) present compelling empirical evidence indicating that a country’s
sovereign credit rating re?ects macro-economic conditions in that country. Their study
provides the basis for the macro-economic variables included in many subsequent
studies of governmental credit ratings, including the current study.
In addition, a number of studies, including Liu and Thakor (1984), Bayoumi et al.
(1995), Depken and Lafountain (2006), and Roychoudhury and Lawson (2010), provide
clear evidence of the link between sovereign credit ratings and borrowing costs. Thus,
inferences about borrowing costs can readily be made from results involving credit
ratings. However, of the studies mentioned, only Roychoudhury and Lawson (2010)
speci?cally examines the role of economic freedomin the determination of governmental
credit ratings. As previously mentioned, a focus on state bond ratings for US states
arguably provides a better laboratory to investigate the importance of economic freedom
in governmental bond ratings than does a focus on sovereign bond ratings at the
international level, such as is seen in Roychoudhury and Lawson (2010).
Turning to the literature dealing more speci?cally with state-level government debt,
the focus has primarilybeenonthe determination of bondyields andpremia. Poterba and
Rueben (1999) brie?y review the literature dealing with state bonds as they examine the
role of state ?scal institutions (e.g. budget rules such as limits on expenditure and debt) in
the determination of yields on US municipal bonds. Bayoumi et al. (1995) consider
whether credit markets exhibit a type of self-enforcing discipline mechanism that is
triggered when states increase borrowing levels. Many of the earlier studies analyzing
some aspect of state bond issues, including Goldstein and Woglom(1992), Bayoumi et al.
(1995) and Lowry (2001), use the Advisory Council on Intergovernmental Relations
(ACIR) index to evaluate differences in ?scal policies and budget rules across states.
These earlier studies consistently ?nd that states subject to anti-de?cit provisions tend to
have lower levels of future debt and, related to that lower debt level, reduced credit risk.
More recently, Wagner (2004) introduces an additional ?scal institution, budget
stabilization funds, into the analysis of state-level credit ratings, ?nding that strict
enforcement of budget stabilization funds lowers the cost of borrowing.
Several of the aforementioned studies, including the seminal 1984 study by Liu and
Thakor, make use of bond ratings from the major rating agencies, such as Moody’s or
Standard and Poor’s (S&P). However, the vast majority of studies that consider bond
ratings actually include them as an additional independent variable in explaining
either bond yields or premia over US Treasury debt. Despite the strong evidence cited
above in support of a direct link between state bond ratings and state bond yields, very
few studies have attempted to determine the factors that in?uence state bond ratings
by treating the bond rating as the dependent variable. In this respect, the current study
is most similar to Depken and Lafountain (2006) and Roychoudhury and Lawson
(2010), since both studies treat credit ratings as the dependent variable – the former in
its examination of the impact of state-level data, including public corruption, on state
bond ratings and the latter in its examination of sovereign debt ratings with a focus
JFEP
5,1
74
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
on economic freedom as a key explanatory variable. To the authors’ knowledge, the
only prior study to use economic freedom in examining ?nancial markets at the US
state level is Lawson and Roychoudhury (2008). However, in that study, economic
freedom is used to explain equity markets, rather than debt markets, in US states.
Thus, modeling state bond ratings as a function of state-level economic freedom is a
unique contribution to this strand of the literature.
Data and methodology
Table I provides a list of variable names and de?nitions. For reporting the results,
particularly in the tables, the variable names from Table I are often used. Most of the
necessary data are readily available from public sources. The US Census Bureau web
site provides the data on state tax burdens, state unemployment levels, state average
incomes, state debt levels, state revenues, and state populations. Likewise, state bond
Bond_Rating The appropriate index score combining the state’s S&P, Moody’s, and Fitch’s
bond ratings
Bond_Supp The appropriate index score combining the state’s S&P, Moody’s, and Fitch’s
bond ratings, with missing state bond ratings proxied using largest city bond
ratings when available
State_Tax State tax revenue as a percentage of personal income
Debt_Percent State debt level as a percentage of government revenue
Income_PC Real state per capita income, in thousands (2005 dollars)
Unemp Annual state unemployment
Corruption Federal public corruption convictions per 100,000 population
1998 One if year is 1998, zero otherwise
1999 One if year is 1999, zero otherwise
2000 One if year is 2000, zero otherwise
2001 One if year is 2001, zero otherwise
2002 One if year is 2002, zero otherwise
2003 One if year is 2003, zero otherwise
2004 One if year is 2004, zero otherwise
2005 One if year is 2005, zero otherwise
2006 One if year is 2006, zero otherwise
2007 One if year is 2007, zero otherwise
2008 One if year is 2008, zero otherwise
EFNA – All EFNA Overall Score for all levels of government
Area_1 – All EFNA Size of Government Score for all levels of government
Area_2 – All EFNA Takings and Discriminatory Taxation Score for all levels of government
Area_3 – All EFNA Labor Market Freedom Score for all levels of government
EFNA EFNA Overall Score for state and local government only
Area_1 EFNA Size of Government Score for state and local government only
Area_2 EFNA Takings and Discriminatory Taxation Score for state and local
government only
Area_3 EFNA Labor Market Freedom Score for state and local government only
PRI_Fiscal Paci?c Research Institute Fiscal Policy Score
PRI_Reg Paci?c Research Institute Regulatory Score
PRI_Jud Paci?c Research Institute Judicial Freedom Score
PRI_Gov Paci?c Research Institute Government Size Score
PRI_Wel Paci?c Research Institute Welfare Spending Score
Note: EFNA – Economic Freedom of North America
Table I.
Variable de?nitions
Economic
freedom
75
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ratings, and city bond ratings where applicable, are published in the Statistical
Abstract of the United States, which is also available on the Census Bureau site.
Table II provides summary statistics for these variables, as well as the economic
freedom indices described.
The only necessary data not easily obtained from a public source are the economic
freedom indices. The primary source for data on economic freedom is the
Economic Freedom Index of North America (EFNA) published by the Fraser Institute
(Ashby et al., 2010). EFNA is an index measured on a scale from zero to ten with
three sub-categories that combine to form the overall index. Area 1 measures Size
of Government, Area 2 measures Takings and Discriminatory Taxation, and Area
3 measures Labor Market Freedom. A higher overall index value indicates a greater
degree of economic freedom within the state. The overall index levels and the three area
scores are analyzed in separate models to determine what impact they may have on state
bond ratings. In addition to these comprehensive indices, the Fraser Institute also
publishes an EFNA index and the same three area indices using values solely from
Mean Median SD Min. Max.
Bond_Rating 0.920 0.909 0.055 0.634 1.000
Bond_Supp 0.921 0.909 0.055 0.634 1.000
State_Tax 0.067 0.065 0.016 0.029 0.284
Debt_Percent 0.402 0.353 0.296 0.013 1.681
Income_PC 3.302 3.227 0.519 2.284 5.258
Unemp 4.694 4.604 1.143 2.258 8.300
Corruption 0.342 0.267 0.317 0.000 2.552
1998 0.091 0.000 0.288 0.000 1.000
1999 0.091 0.000 0.288 0.000 1.000
2000 0.091 0.000 0.288 0.000 1.000
2001 0.091 0.000 0.288 0.000 1.000
2002 0.091 0.000 0.288 0.000 1.000
2003 0.091 0.000 0.288 0.000 1.000
2004 0.091 0.000 0.288 0.000 1.000
2005 0.091 0.000 0.288 0.000 1.000
2006 0.091 0.000 0.288 0.000 1.000
2007 0.091 0.000 0.288 0.000 1.000
2008 0.091 0.000 0.288 0.000 1.000
EFNA – All 6.896 6.894 0.548 5.206 8.478
Area_1 – All 7.186 7.246 0.940 2.494 9.097
Area_2 – All 6.266 6.263 0.632 3.540 8.463
Area_3 – All 7.237 7.264 0.555 5.659 8.416
EFNA 7.179 7.236 0.612 5.353 8.394
Area_1 7.201 7.329 0.942 3.744 8.894
Area_2 7.315 7.374 0.664 5.589 8.957
Area_3 7.022 6.906 0.728 5.702 8.828
PRI_Fiscal 23.920 23.715 3.686 16.570 31.460
PRI_Reg 13.480 13.070 2.300 9.956 19.090
PRI_Jud 13.792 13.765 3.734 7.667 23.330
PRI_Gov 25.446 25.217 6.228 10.710 37.714
PRI_Wel 25.280 25.840 7.080 9.220 39.220
Notes: EFNA – Economic Freedom of North America; PRI – Paci?c Research Institute
Table II.
Summary statistics
JFEP
5,1
76
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
each state government, i.e. net of federal government transfers to the state. The
comprehensive EFNA and area indices that include federal government transfers are
denoted by the addition of the label “ALL” (e.g. EFNA – ALL, AREA_1 – ALL, etc.),
while the EFNA and area indices that exclude federal government transfers are simply
labeled EFNA, AREA_1, etc. The economic freedom index values net of federal
government transfers are also analyzed in a second set of models to ascertain whether
the in?uence of federal government transfers on state economic freedom rankings is
important in determining state bond ratings. EFNA is highly correlated with economic
growth and well-being, which could also be measured using state income and
state unemployment. Since income and unemployment are two of the other independent
variables speci?ed in the models, only lagged values of the economic freedomindices are
included in an attempt to minimize potential multicollinearity and endogeneity issues.
As noted in the literature review, prior studies of state level bond markets have used
a variety of ?scal institutions to estimate bond yields. While most prior studies have
not included EFNA as an independent variable, a strong argument along two
dimensions can be made regarding its appropriateness in models of bond yields and
ratings. First, states that have greater economic freedomare more likely to have adopted
anti-de?cit ?scal institutions of the type often measured in prior literature by the ACIR
index[1]. Several of the ?scal institutions can have multiple interpretations based on the
rules associated with that particular institution in any given state. For instance, the
“balanced budget rule” cited in several prior studies is actually an amalgamation of
several rules of varying degrees of restrictiveness ranging from a simple requirement
that the governor propose a balanced budget to a strict prohibition that the state cannot
carry a de?cit. Thus, proper interpretation of discrete ?scal institutions variables is
dependent upon having all potential rule andimplementation variations accounted for in
the model[2]. This potential speci?cation issue is exactly what the ACIR index was
designed to address. Given that the EFNA index is also a continuous variable, like the
ACIR, it should also serve to limit potential speci?cation problems that might arise if
economic freedom were measured via discrete variables, as would be necessary in the
absence of the EFNA index.
Second, there are two different aspects of institutional quality through which
economic freedom can affect bond ratings. The more important of these two aspects is
that states with anti-de?cit ?scal institutions would be more likely to exhibit greater
economic freedom. The other facet of institutional quality through which economic
freedom might impact bond ratings is found in Hall and Sobel (2007), Campbell and
Rodgers (2007) and Campbell et al. (2007). Speci?cally, states that have greater
economic freedom (i.e. higher institutional quality) exhibit higher rates of economic
growth, personal income, and entrepreneurship as measured by ?rm births and deaths.
Granted, these ?scal institutions would have a more indirect effect on bond ratings, but
the effect would still be potentially relevant to the determination of state bond ratings.
For these two reasons, EFNA should be seen as a broader measure of institutional
quality that re?ects more than merely ?scal conditions and ?scal institutions. In this
way, the use of EFNA is an important contribution to the bond rating literature.
As an additional robustness check, the state level economic freedom index
published by the Paci?c Research Institute (PRI) is used as an alternative measure
of economic freedom. Since this index is not published on an annual basis, the
most recent PRI index scores from 2008 are used to estimate a cross-sectional model
Economic
freedom
77
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
for that year. The PRI index evaluates economic freedom broken down into ?ve
categories, or sectors: ?scal, regulatory, judicial, government size, and welfare spending.
The individual sector scores used to compute the overall PRI index have various ranges
depending in large part on how many indicators were used in computing that sector
score and how the indicators were ultimately weighted. It should be noted that the
PRI index does not assign scores in the same way as the EFNA, in that lower PRI
index values mean greater economic freedom. The interested reader is referred to
McQuillan et al. (2008) for additional information.
The statistical methodology employed is very straightforward. Using state-level
data for the years 1998-2008, ordinary least squares (OLS) regression is utilized in the
?rst set of estimations to analyze the relationship between the chosen independent
variables and state bond ratings. Since the index of bond ratings used is bounded from
above at 1.000, a censored regression, or tobit estimation, is utilized as a robustness
check on the results observed from the OLS estimation. Both statistical techniques are
very well-understood and exceedingly common in the literature. Therefore, little time is
devoted to an explanation of their use. The speci?c OLS models estimated are:
Bond_Rating
i;t
¼ b
0
þ b
1
X
i;t
þ b
2
Corruption
i;t
þ b
3
Year
t
þb
4
EFNA
i;t 21
þ 1
i;t
ð1Þ
Bond_Rating
i;t
¼ b
0
þ b
1
X
i;t
þ b
2
Corruption
i;t
þ b
3
Year
t
þb
4
EFNA_Area
k;i;t 21
þ 1
i;t
ð2Þ
where Bond_Rating
i,t
is the bond rating for state i at time t, X
i,t
is a vector of state
economic control variables as further de?ned in Table I, Corruption
i,t
is the number
of federal convictions of public of?cials, EFNA
i,t21
is the lagged value of the overall
economic freedom index and EFNA_Area
k,i,t21
is the lagged value of economic
freedom for area k (where k ¼ 1 2 3), Year
t
is an indicator variable for the year, and
1
i,t
is the normally distributed error term.
The methodology for computing the state bond rating index (Bond_Rating) is,
however, not a technique that can simply be mentioned and dismissed. To begin, the
individual state bond ratings are collected for each sample year (i.e. the rating at the
end of the last quarter of the year) for each state from the three major bond rating
providers: S&P, Moody’s, and Fitch Ratings (Fitch’s). S&P has a total of 22 rating
designations, Moody’s has 21, and Fitch’s has 20; the complete list of all possible
ratings for each agency is provided in the Appendix.
Usingthe same ratingscheme as DepkenandLafountain(2006), eachratingdesignation
is assigned a number value, with one denoting the lowest rating designation used by that
rating provider. For example, the lowest rating designation given by S&P (“D”) would be
assigned a value of one and the highest rating designation given by S&P (“AAA”) would
be assigned a value of 22, since S&Phas 22 total ratingdesignations. These cardinal scores
are then normalized by the number of possible rankings for each provider. The ?nal step is
to average the available normalized scores for each state. It shouldbe noted here that not all
states have debt issues that are ratedinall sample years (i.e. a state couldhave zero ratings).
Also, ratings may be available from one, two, or all three ratings providers.
In an effort to present as complete a picture as possible, and as a way of checking
the robustness of the primary results, bond ratings for the largest city in a state were
substituted if no state bond rating was available in a particular sample year[3].
JFEP
5,1
78
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Despite this attempt to include all states in the analysis, there are still some missing
observations. In some cases, the largest city may not have had debt outstanding
(or rated), while in other cases, the largest city may not have been large enough to be
reported in the Statistical Abstract of the United States (e.g. Idaho).
We estimate equation (3) using a tobit model:
Bond_Rating
*
i;t
¼ b
0
þ b
1
X
i;t
þ b
2
Corruption
i;t
þ b
3
Year
t
þ b
4
EFNA_Area
k;i;t 21
þ 1
i;t
ð3Þ
where:
Bond_Rating
*
i;t
¼
Bond_Rating
i;t
; if Bond_Rating
i;t
. 0
0; if Bond_Rating
i;t
# 0
8
<
:
It is assumed that 1 , N(0,s
2
I), and Bond_Rating
*
i,t
is the latent variable used in the
tobit estimation. The same set of independent variables is utilized as in equation
and (2), replacing the dependent variable Bond_Rating with Bond_Rating
*
i,t
.
Results
Table III reports OLS results for state bond ratings using the various economic freedom
indices described. Model 1 uses the comprehensive EFNA index, which includes federal
government spendingat the state level. Model 2 uses the three area scores underlying the
comprehensive EFNA index, Models 3 and 4 include the comprehensive and
area-speci?c EFNA index scores net of federal government transfers, and Model 5 is
the cross-sectional analysis utilizing the PRI sector scores for 2008. As can be seen in the
table, results on the control variables are very consistent across the models using the
various EFNA indices. A higher state debt level relative to state government revenue,
a higher unemployment rate, a higher level of public corruption, and a lower percent
of state tax revenue relative to personal income are all associated with lower state bond
ratings, with the only exception being state tax revenue as a percent of personal income
in Model 1, which is insigni?cant. Per capita income is not related to state bond ratings in
three of the models, but it is marginally signi?cantly positively related to bond ratings in
Model 2. These ?ve control variables do not register as signi?cant in Model 5, which is
not surprising. Recall that Model 5 uses the PRI index for 2008, which results ina sharply
reduced sample size.
Turning to the variables of interest, the economic freedom indices, Table III shows
that, regardless of the index chosen, there is at least some positive and statistically
signi?cant impact on state bond ratings. There is no difference between results using the
EFNA index whether federal government expenditures at the state level are considered
(Model 1) or not (Model 3). In both cases, a higher EFNA index value is highly
signi?cantly related to a higher state bondrating, which is as expected, given that higher
EFNA index values indicate greater economic freedom. Using the EFNA area-speci?c
scores, some difference is noted depending on whether federal government spending
at the state level is included (Model 2) or not (Model 4). Speci?cally, when federal
government spending at the state level is excluded (Model 4), Size of Government
(Area 1) moves from insigni?cant in Model 2 to marginally signi?cant and positive in
Model 4 and Labor Market Freedom (Area 3) changes from statistically signi?cantly
Economic
freedom
79
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
positive in Model 2 to insigni?cant in Model 4. Area 2, which measures Takings and
Discriminatory Taxation, is highly statistically signi?cant in both models. Changes in
signi?cance for these variables based solely on whether federal government spending at
the state level is included when the area scores are computed indicates that federal
government spending can substantially in?uence economic freedom at the individual
state level. In Model 5, the only two sectors that enter the model as signi?cant predictors
of state bond ratings are the Regulation and Welfare Spending sectors. It is worthy of
note that both of thementer with the expected negative signs (recall that in the PRI index
sector scores, a lower value indicates greater economic freedom).
Table IV presents an alternative approach to estimating the relationship between
the independent variables and state bond ratings. The bond rating classi?cation scheme
described in the data and methodology section produces an index of bond ratings that is
bounded from above at 1.000. Since OLS is known to produce inconsistent parameter
Variable Model 1 Model 2 Model 3 Model 4 Model 5
Constant 0.798
* * *
0.624
* * *
0.721
* * *
0.716
* * *
1.046
* * *
State_Tax 0.161 0.371
* *
0.413
* * *
0.475
* * *
0.052
Debt_Percent 20.039
* * *
20.041
* * *
20.037
* * *
20.041
* * *
20.165
Income_PC 20.008 0.011
*
0.004 0.001 20.009
Unemp 20.013
* * *
20.009
* * *
20.009
* * *
20.012
* * *
20.001
Corruption 20.030
* * *
20.039
* * *
20.039
* * *
20.043
* * *
20.011
1999 0.002 0.002 0.001 0.001
2000 0.006 0.003 0.005 0.004
2001 0.010 0.002 0.005 0.006
2002 0.016 20.003 0.013 0.015
2003 0.010 20.015 0.009 0.010
2004 0.002 20.026
* *
0.004 0.004
2005 20.001 20.034
* * *
0.002 0.004
2006 20.020
*
20.055
* * *
20.020
*
20.019
*
2007 20.014 20.051
* * *
20.017 20.015
2008 0.005 20.035
* *
20.002 0.002
EFNA – All 0.032
* * *
Area_1 – All 20.006
Area_2 – All 0.018
* * *
Area_3 – All 0.036
* * *
EFNA 0.032
* * *
Area_1 0.006
*
Area_2 0.024
* * *
Area_3 0.005
PRI_Fiscal 0.001
PRI_Reg 20.007
* *
PRI_Jud 0.003
PRI_Gov 0.001
PRI_Wel 20.002
*
R
2
(%) 24.2 27.5 25.2 27.0 28.9
Adjusted R
2
(%) 21.7 24.9 22.8 24.4 10.7
F-statistic 9.99
* * *
10.55
* * *
10.56
* * *
10.28
* * *
1.59
n 519 519 519 519 50
Note: Signi?cant at:
*
10,
* *
5 and
* * *
1 percent levels
Table III.
OLS regression results
for state bond ratings
using various economic
freedom indices
JFEP
5,1
80
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
estimates when dealing with a bounded dependent variable, censored regression, or tobit
estimations, are utilized as an alternative. Given space limitations, tobit results are
reported for onlytwo models: the EFNAindex usingarea-speci?c scores withand without
federal government spending at the state level, or Models 2 and 4, respectively. As can be
seen in the table, the tobit results for the control variables are entirely consistent with the
OLS results from Table III. Area 1 (Size of Government) is marginally signi?cantly
negative in Model 2 and marginally signi?cantly positive in Model 4. These results
suggest that federal government transfers are particularly important when considering
size of state government. Speci?cally, when including federal government transfers, it
seems that greater Area 2 economic freedom index scores still suffer because of the
perception that federal monies are, at least to some extent, subsidizing the size of state
government. When state spending is considered in the absence of federal
government transfers, smaller government seems to be more closely associated with
Gain/loss in bond rating from 1 SD change
S&P Moody’s Fitch’s
Variable
Model 2 marginal
impact
Model 4 marginal
impact
Model
2
Model
4
Model
2
Model
4
Model
2
Model
4
State_Tax 0.410
* * *
0.514
* * *
0.15 0.18 0.14 0.17 0.13 0.17
Debt_Percent 20.043
* * *
20.042
* * *
20.28 20.27 20.27 20.26 20.26 20.25
Income_PC 0.013
*
0.001 0.15 0.02 0.14 0.01 0.13 0.01
Unemp 20.010
* * *
20.013
* * *
20.26 20.32 20.25 20.30 20.24 20.29
Corruption 20.040
* * *
20.043
* * *
20.28 20.30 20.27 20.28 20.25 20.27
1999 0.002 0.000 0.01 0.00 0.01 0.00 0.01 0.00
2000 0.002 0.003 0.01 0.02 0.01 0.02 0.01 0.02
2001 0.001 0.004 0.00 0.03 0.00 0.03 0.00 0.03
2002 20.005 0.014 20.00 0.09 20.00 0.09 20.00 0.08
2003 20.020 0.010 20.13 0.06 20.12 0.06 20.11 0.06
2004 20.033
* *
0.003 20.21 0.02 20.20 0.02 20.19 0.02
2005 20.043
* * *
0.001 20.27 0.01 20.26 0.01 20.25 0.01
2006 20.067
* * *
20.024
*
20.42 20.15 20.40 20.14 20.38 20.14
2007 20.062
* * *
20.019 20.39 20.12 20.37 20.12 20.36 20.11
2008 20.044
* * *
20.002 20.28 20.01 20.27 20.01 20.25 20.01
Area_1 – All 20.007
*
20.15 20.15 20.14
Area_2 – All 0.022
* * *
0.31 0.29 0.28
Area_3 – All 0.040
* * *
0.49 0.47 0.45
Area_1 0.006
*
0.13 0.13 0.12
Area_2 0.026
* * *
0.38 0.36 0.34
Area_3 0.007
*
0.11 0.10 0.10
Log-
likelihood 567.99 562.59
Pseudo R
2
(%) 18.16 17.03
Test for zero
slopes 174.57
* * *
163.77
* * *
n 519 519
Notes: Signi?cant at:
*
10,
* *
5 and
* * *
1 percent levels; the gain or loss in bond rating is computed as
the marginal impact times the standard deviation times the number of ratings for a rating agency; for
example, the loss in S&P bond rating from a one standard deviation increase in unemployment equals
20.010 £ 1.143 £ 22 ¼ 0.26 for Model 2
Table IV.
Tobit estimation results
for state bond ratings
using state-level
Economic Freedom of
North America indices
Economic
freedom
81
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
greater ?scal responsibility and higher bond ratings. Area 3 (Labor Market Freedom) is
marginally signi?cantly positive in Model 4. In all three of these marginally signi?cant
cases, the marginal impact of a one standard deviation change in that area index score is
economically very small. In the other three cases (Area 2 in both models and Area 3 in
Model 2), the economic impact varies fromroughlyone-third of a bondratingto one-half of
a bond rating for a one standard deviation change in the area index score.
To illustrate, Table IV includes estimates from the Depken and Lafountain (2006)
method of examining the economic impact of increasing the EFNAindex by one standard
deviation above the mean. Rubinfeld (1973) demonstrated that a rating change from
Moody’s AAA to Moody’s AA status (at the time of his writing, Moody’s did not
differentiate within investment grades) for a municipal bond issuer correlated with an
increase in interest rates of approximately 20.6 basis points. Using the current
Moody’s rating system, this would indicate an approximate 6.87 basis point increase for a
Moody’s rating downgrade from Aaa to Aa1. For comparison purposes regarding the
scale of ratings changes associated with economic freedom, Table IV shows a one
standard deviation increase in public corruption above the national mean would result in
a 20.27 change in the Moody’s state bond rating, which would increase the cost of debt
service by approximately $185 per year per million dollars of debt (assuming an original
interest rate of 5 percent). Similarly, a one standard deviation increase in Area 2 (Takings
and Taxation) would result in a 0.36 increase in Moody’s state rating. Thus, greater
economic freedom related to takings and taxation would lower the cost of debt by
approximately $247 per year per million dollars of debt.
Table V, reports results fromthe same EFNAmodels (Models 1-4) speci?edinTable III.
However, the bond rating index is augmented via the substitution of the bond rating for
the largest city in a state if the state has no bond rating available. This is admittedly an
imperfect solution, but the exclusion of several states from the estimation is, arguably, a
more troubling issue. Unfortunately, even allowing this substitution does not enable the
inclusion of all states in the estimation. For several states that have no bond rating, there
are either no cities with bond ratings or no cities large enough to have their bond ratings
reported in any easily accessible manner. Nonetheless, the sample was increased by a
relative few observations, and the results can be seen in Table V. As shown there, the
EFNA index and various area-speci?c scores perform exactly the same as in Table III.
Conclusions
This study estimates the impact of various indices of economic freedom on US
state-level bond ratings. Using OLS and censored regression techniques, the Economic
Freedom of North America index, its three area components, and the ?ve sector scores
underlying the PRI economic freedom index are found to signi?cantly impact state
bond ratings. The results provide substantial support for the notion that states
exhibiting greater levels of economic freedom also enjoy higher bond ratings, with the
associated lower borrowing costs that such ratings entail. Comparison of the results
using area-speci?c EFNA scores shows that federal government expenditures at the
state level can in?uence the manner in which the three areas of economic freedom
impact state bond ratings. Policy implications are quite clear. First, states that pursue
policies aimed at achieving greater economic freedom will be favorably positioned
to receive higher bond ratings than they would in the absence of such policies.
Second, state politicians should be cognizant of the impact of their decisions about
JFEP
5,1
82
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
which federal monies to take, and the manner in which those funds are taken, on their
state’s economic freedom, and thus, their state’s bond ratings.
Notes
1. The Advisory Commission on Intergovernmental Relations (ACIR), which produced the ACIR
indexonanti-de?cit ?scal institutions, nolonger exists. It is nowknownas the AmericanCouncil
onIntergovernmental Relations. Inaddition, the Signi?cant Features of Fiscal Federalismreport,
which provided much of the data in the ACIR index, has not been published since 1995.
2. Wagner and Sobel (2006) ?nd that states with already-existing tax and expenditure limitations
were more likelyto adopt statutorybudget stabilizationfunds but were less likelyto adopt such
funds if they also included stringent deposit and withdrawal rules. This suggests that at least
some budget stabilization funds were adopted to circumvent existing ?scal constraints.
3. It should be noted that 11 states do not issue general obligation bonds. These states
include Arizona, Arkansas, Colorado, Idaho, Indiana, Iowa, Kansas, Kentucky, Nebraska,
South Dakota, and Wyoming. These 11 states are the ones for which an attempt is made
to proxy the state’s bond rating using the bond rating for the state’s largest city.
Variable Model 1 Model 2 Model 3 Model 4
Constant 0.803
* * *
0.640
* * *
0.731
* * *
0.730
* * *
State_Tax 0.173 0.372
* *
0.414 0.467
* * *
Debt_Percent 20.039
* * *
20.041
* * *
20.038
* * *
20.042
* * *
Income_PC 20.008 0.009 0.004 20.000
Unemp 20.013
* * *
20.010
* * *
20.010
* * *
20.012
* * *
Corruption 20.030
* * *
20.040
* * *
20.039
* * *
20.043
* * *
1999 0.002 0.002 0.001 0.001
2000 0.006 0.003 0.005 0.004
2001 0.011 0.003 0.006 0.007
2002 0.018 0.001 0.015 0.017
*
2003 0.012 20.012 0.012 0.013
2004 0.004 20.022
*
0.006 0.007
2005 0.001 20.030
* *
0.005 0.007
2006 20.019
*
20.052
* * *
20.019
*
20.018
2007 20.013 20.048
* * *
20.016 20.013
2008 0.006 20.031
* *
20.000 0.003
EFNA – All 0.032
* * *
Area_1 – All 20.005
Area_2 – All 0.017
* * *
Area_3 – All 0.035
* * *
EFNA 0.032
* * *
Area_1 0.007
* *
Area_2 0.023
* * *
Area_3 0.004
R
2
(%) 25.6 28.6 26.5 28.2
Adjusted R
2
(%) 23.3 26.1 24.2 25.7
F-statistic 11.16
* * *
11.47
* * *
11.67
* * *
11.26
* * *
n 535 535 535 535
Notes: Signi?cant at:
*
10,
* *
5 and
* * *
1 percent levels; results using the Paci?c Research Institute
measures of economic freedom are not included since no state bond rating proxies were used in the
?nal sample year
Table V.
OLS regression results
for supplemented state
bond ratings using
various economic
freedom indices
Economic
freedom
83
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
References
Ashby, N.J., Karabegovic´, A., McMahon, F. and Bueno, A. (2010), Economic Freedom of North
America 2010, Fraser Institute, Vancouver.
Bayoumi, T., Goldstein, M. and Woglom, G. (1995), “Do credit markets discipline sovereign
borrowers? Evidence from US States”, Journal of Money, Credit, and Banking, Vol. 27,
pp. 1046-59.
Butler, A.W. and Fauver, L. (2006), “Institutional environment and sovereign credit ratings”,
Financial Management, Vol. 35, pp. 53-79.
Campbell, N.D. and Rodgers, T.M. (2007), “Economic freedom and net business formation”, Cato
Journal, Vol. 27 No. 1, pp. 23-36.
Campbell, N.D., Rogers, T.M. andHeriot, K.C. (2007), “The economic freedomindexas adeterminant
of ?rm births and ?rm deaths”, Southwest Business and Economics Journal, pp. 37-50.
Cantor, R. and Packer, F. (1996), “Determinants and impact of sovereign credit ratings”, Federal
Reserve Bank of New York Economic Polciy Review, pp. 37-53.
Depken, C.A. and Lafountain, C.L. (2006), “Fiscal consequences of public corruption: empirical
evidence from state bond ratings”, Public Choice, Vol. 126, pp. 75-85.
Everett, B. (2012), Politico, March 8, available at: Politico.com:dyn.politico.com
Goldstein, M. and Woglom, G. (1992), “Market-based ?scal discipline in monetary unions:
evidence from the US municipal bond market”, in Canzoneri, M., Grilli, V. and
Masson, P. (Eds), Establishing a Central Bank: Issues in Europe and Lessons from the
United States, Cambridge University, Cambridge.
Hall, J.C. and Sobel, R.S. (2007), “Institutions, entrepreneurship, and regional differences in
economic growth”, Southern Journal of Entrepreneurship, Vol. 1, pp. 70-96.
Lawson, R.A. and Roychoudhury, S. (2008), “Economic freedom and equity prices among US
States”, The Credit and Financial Management Review, Vol. 14, pp. 25-35.
Liu, P. and Thakor, A.V. (1984), “Interest yields, credit ratings, andeconomic characteristics of state
bonds: an empirical analysis”, Journal of Money, Credit and Banking, Vol. 16, pp. 344-51.
Lowry, R.C. (2001), “A visible hand? Bond markets, political parties, balanced budget laws, and
state government debt”, Economics and Politics, Vol. 13, pp. 49-72.
McQuillan, L.J., Maloney, M.T., Daniels, E. and Eastwood, B.M. (2008), US Economic Freedom
Index: 2008 Report, Paci?c Research Institute, San Francisco, CA.
Moody’s Investors Service (2012a), Moody’s Downgrades State of Connecticut General Obligation
Bonds To Aa3 From Aa2 Rating and Assigns Stable Outlook, January 20, available at:
www.moodys.com
Moody’s Investors Service (2012b), Moody’s Revises State of Washington Rating Outlook to
Negative from Stable and Af?rms Aa1G.O, January 30, available at: www.moodys.com
Poterba, J.M. and Rueben, K. (1999), “State ?scal institutions and the US municipal bond market”,
in Poterba, J.M. (Ed.), Fiscal Institutions and Fiscal Performance, University of Chicago,
Chicago, IL.
Qi, Y., Roth, L. and Wald, J.K. (2010), “Political rights and the cost of debt”, Journal of Financial
Economics, Vol. 95, pp. 202-26.
Reinhart, C.M. and Rogoff, K.S. (2004), “Serial default and the “Paradox” of rich-to poor capital
?ows”, American Economic Review, Vol. 94, pp. 53-9.
Roychoudhury, S. and Lawson, R.A. (2010), “Economic freedom and sovereign credit ratings and
default risk”, Journal of Financial Economic Policy, Vol. 2, pp. 149-62.
JFEP
5,1
84
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Rubinfeld, D.L. (1973), “Credit ratings and the market for general obligation municipal bonds”,
National Tax Journal, Vol. 26 No. 1, pp. 17-27.
Wagner, G.A. (2004), “The bond market and ?scal institutions: have budget stabilization funds
reduced state borrowing costs?”, National Tax Journal, Vol. 57 No. 4, pp. 785-804.
Wagner, G.A. and Sobel, R.S. (2006), “State budget stabilization fund adoption: preparing for the
next recession or circumventing ?scal constraints?”, Public Choice, Vol. 126, pp. 177-99.
Appendix
Corresponding author
Justin D. Bene?eld can be contacted at: [email protected]
S&P Moody’s Fitch’s
AAA Aaa AAA
AA þ Aa1 AA þ
AA Aa2 AA
AA 2 Aa3 AA 2
A þ A1 A þ
A A2 A
A 2 A3 A 2
BBB þ Baa1 BBB þ
BBB Baa2 BBB
BBB 2 Baa3 BBB 2
BB þ Ba1 BB þ
BB Ba2 BB
BB 2 Ba3 BB 2
B þ B1 B þ
B B2 B
B 2 B3 B 2
CCC þ Caa1 CCC
CCC Caa2 CC
CCC 2 Caa3 C
CC Ca D
C C
D
Table AI.
Bond ratings used
by the three major
rating agencies
Economic
freedom
85
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_705910591.pdf