Description
This paper aims to study the interplay of fiscal policy and asset prices in a time-varying
fashion.
Journal of Financial Economic Policy
A time-varying approach to analysing fiscal policy and asset prices in South Africa
Rangan Gupta Charl J ooste Kanyane Matlou
Article information:
To cite this document:
Rangan Gupta Charl J ooste Kanyane Matlou , (2014),"A time-varying approach to analysing fiscal policy
and asset prices in South Africa", J ournal of Financial Economic Policy, Vol. 6 Iss 1 pp. 46 - 63
Permanent link to this document:http://dx.doi.org/10.1108/J FEP-01-2013-0003
Downloaded on: 24 January 2016, At: 21:48 (PT)
References: this document contains references to 68 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 209 times since 2014*
Users who downloaded this article also downloaded:
Matthew Kofi Ocran, (2011),"Fiscal policy and economic growth in South Africa", J ournal of Economic
Studies, Vol. 38 Iss 5 pp. 604-618http://dx.doi.org/10.1108/01443581111161841
Saumya Ranjan Dash, J itendra Mahakud, (2015),"Market anomalies, asset pricing models, and stock
returns: evidence from the Indian stock market", J ournal of Asia Business Studies, Vol. 9 Iss 3 pp. 306-328http://dx.doi.org/10.1108/J ABS-06-2014-0040
Pablo A. Caron, Marcela A. Cruchaga, Axel E. Larreteguy, (2015),"Sensitivity analysis of finite volume
simulations of a breaking dam problem", International J ournal of Numerical Methods for Heat & Fluid
Flow, Vol. 25 Iss 7 pp. 1718-1745http://dx.doi.org/10.1108/HFF-10-2014-0308
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
A time-varying approach
to analysing ?scal policy and
asset prices in South Africa
Rangan Gupta, Charl Jooste and Kanyane Matlou
Department of Economics, University of Pretoria, Pretoria, South Africa
Abstract
Purpose – This paper aims to study the interplay of ?scal policy and asset prices in a time-varying
fashion.
Design/methodology/approach – Using South African data since 1966, the authors are able to
study the dynamic shocks of both ?scal policy and asset prices on asset prices and ?scal policy based
on a time-varying parameter vector autoregressive (TVP-VAR) model. This enables the authors to
isolate speci?c periods in time to understand the size and sign of the shocks.
Findings – The results seem to suggest that at least two regimes exist in which expansionary ?scal
policy affected asset prices. From the 1970s until 1990, ?scal expansions were associated with
declining house and slightly increased stock prices. The majority of the ?rst decade of 2000 had asset
prices increasing when ?scal policy expanded. On the other hand, increasing asset prices reduced
de?cits for the majority of the sample period, while the recent ?nancial crises had a marked change on
the way asset prices affect ?scal policy.
Originality/value – This is the ?rst attempt in the literature of ?scal policy and asset prices to use a
TVP-VAR model to not only analyse the impact of ?scal policy on asset prices, but also the feedback
from asset prices to ?scal policy over time.
Keywords House prices, TVP-VAR, Stock prices, Countercyclical ?scal policy
Paper type Research paper
1. Introduction
The recent global ?nancial crisis demonstrates that boom/bust cycles in asset prices
can dramatically affect macroeconomic stability, especially output and price stability.
Hence, the importance of monetary and ?scal policy in sustaining economic growth
during and after the ?nancial crisis has become a dominant area of study. Analysts
typically focus on monetary policy to consider the linkages between economic policy
and asset markets[1]. Whilst monetary policy dominated the ?eld of academic and
policy discussions on controlling elements of the business cycle, ?scal policy has
become key after monetary policy reached the zero interest rate lower bound and
became ineffective in stimulating demand during the recent recession (Feldstein, 2009).
Large and persistent ?scal stimulus, however, can lead to long-term unsustainability of
sovereign ?nances as seen when analysing current government bond markets
(Schuknecht et al., 2009). Researchers need to disentangle this effect, however, from the
mess left by ?nancial institutions in Europe and the USA. Furthermore, this may lead
to business cycle de-synchronization (Ra?q and Mallick, 2008; Mallick and Mohsin,
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JEL classi?cation – C11, C15, C32, H30, H61
The authors would like to thank two anonymous referees and Jouchi Nakajima for many
helpful comments. The usual disclaimer applies.
Journal of Financial Economic Policy
Vol. 6 No. 1, 2014
pp. 46-63
qEmerald Group Publishing Limited
1757-6385
DOI 10.1108/JFEP-01-2013-0003
JFEP
6,1
46
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
2007, 2010) or negatively affect the nexus between monetary and ?nancial stability
(Castro, 2011; Granville and Mallick, 2009; Sousa, 2010a).
Despite the large number of studies analysing the macroeconomic effects of ?scal
policy (see Mountford and Uhlig, 2009; Afonso and Sousa, 2012 for detailed reviews), the
importance of asset markets over the business cycle (Afonso and Sousa, 2011; Agnello and
Sousa, 2014; Iacoviello, 2010, 2011), andthe feedback of asset prices to ?scal policy(refer to
Agnello et al., 2012a for a detailed review), an important gap in the literature exists
regarding the empirical relationship between ?scal policy actions and developments in
asset prices and in turn, the possible feedback from asset prices to ?scal policy stance,
especially in emerging market economies. This study concentrates on South Africa, given
our familiarity with the economic structure of the economy. In South Africa, non-housing
wealth (housing wealth) equals 49.95 per cent (31.13 per cent) of household’s total assets
and 61.59 per cent (38.41 per cent) of household’s net worth in 2011 (Aye et al., 2013a).
Hence, it is not surprising that recent evidence show that Aron and Muellbauer (2013),
Das et al. (2011), Ncube and Ndou (2011), Apergis et al. (2013), Simo-Kengne et al. (2012,
2013a), Peretti et al. (2012) and Aye et al. (2013b) there are signi?cant spill-overs onto
consumption and output from not only the stock market, but also the housing market.
Also, as highlighted by the time-varying approaches of Peretti et al. (2012) and Aye et al.
(2014a), the South African economy began slowing by the end of 2007, as the stock and
housing markets entered deep bear markets (Venter, 2011; Simo-Kengne et al., 2013b).
This paper attempts to contribute to the existing literature, and hence our main
contribution, by focussing on the consequences of ?scal policy/asset price shocks on
asset prices/?scal policy in speci?c periods and over different regimes. This study
focuses on the interplay of South African asset prices and ?scal policy. Time varying
parameters in a model which links these variables in a simultaneous setup enables a
bird’s eye view of certain events and periods such as the recent ?nancial crisis.
In particular, we analyse not only the effects of ?scal policy shocks, but also look at
asset price shocks to understand its impact on ?scal variables, and to the extent that we
?nd a link between them, we look at the magnitude and the persistence of the effects.
2. Literature review
The behaviour of asset markets and their prices emerges as an important factor for
the decision making of ?nancial institutions, homeowners and consumers, businesses,
and policy makers. The linkages between the ?nancial market and the banking
system, the housing sector, and the credit market produced strong and powerful effects
in the course of the ?nancial turmoil (Afonso and Sousa, 2011). According to the
European Central Bank (2010), a variety of mechanisms exist through which asset
prices can affect consumption spending. For example, a wealth effect working through
consumers and a “q-effect”[2] working through businesses can affect asset prices.
House price bubbles, which arose in most developed and emerging-market countries
prior to the ?nancial crisis, led to unsustainable borrowing by homeowners to ?nance
consumption against “seemingly” permanent increases in their equity holdings.
If q increases as a result of an increase in equity prices, the ?rm can raise more capital
by issuing new equity. This makes it more attractive for ?rms to raise new capital,
thus increasing investment demand, which may, in turn, lead to higher prices for goods
and services. Additional effects can stem from residential property prices, which, via
higher wage demands by workers, may lead to increases in both the prices of goods
Time-varying
approach
47
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and services and, therefore, consumer prices. Finally, movements in asset prices can
signi?cantly affect business and consumer con?dence. Hence, researchers now focus
their attention on the relationship between macroeconomic variables, wealth, and asset
returns (see Sousa, 2010b, c; Afonso and Sousa, 2011, 2012; Agnello and Sousa, 2014;
Peretti et al., 2012; Simo-Kengne et al., 2013a for detailed literature reviews).
Our understanding of the transmission of ?scal policy innovations to asset markets
is limited, however, exists because of the few studies concentrating on US and
industrialized European markets (e.g. Afonso and Sousa, 2011; Agnello and Sousa,
2014, and references cited therein). Various channels exist whereby ?scal policy can
affect stock and housing markets (Afonso and Sousa, 2011; Agnello and Sousa, 2014).
For instance, ?scal policy can in?uence stock markets via its effect on sovereign risk
spreads. These spreads, in turn, re?ect the ?nancing capacity of government as well as
investor expectations. When the markets deem that ?scal policy is stable, then an
in?ow of capital causes the exchange rate to appreciate and subsequently to reduce
pressures on central bank authorities to raise interest rates. Since demand for
government bonds strengthen, the overall bond yield curve falls, which affects the
stock market. Increasing public de?cits through the government’s wage bill, however,
can lead to a deteriorating lending environment, as this could lead to an increase in the
demand for credit that pushes interest rates higher. Consequently, the present
discounted value of the cash-?ows generated by stocks falls, the markets require a
higher risk premium, and stock prices shrink. Finally, unsound ?scal policies can
prompt a loss in the con?dence of home-currency assets and generate a rebalancing of
asset portfolio composition away from domestic assets toward foreign assets.
Fiscal policy can also affect housing markets. For example, taxes on housing capital
gains and the imputed rental housing value, ?scal subsidies and value added taxes
(VAT) on purchases of new homes, and the tax deductibility of mortgage payments and
housing rents can affect house prices via their effects on households’ disposable income
and the demand of housing. An indirect effect of ?scal spending through the wage bill
and government infrastructure spending can lead to both increases and decreases in the
demand for housing. More broadly, the deterioration of the ?scal stance and uncertainty
about the long-run sustainability of public ?nances can affect long-term interest rates
and negatively impinge on the ?nancing conditions for mortgages, pushing house prices
downwards. Hence, we should not neglect the role of ?scal policy in explaining both
housing market developments and stock market dynamics.
As discussed above, changes in stock or house prices can in?uence consumption.
However, it is the variation in the ?nancial and housing wealth that can produce
substantial variation in personal savings. In a Keynesian setup, when the corporate
sector does not compensate the change in household savings, it is then left for
the government to allow for a variation in its own savings and, thereby, to smooth
the ?uctuations in national saving that originates from movements in asset prices.
There is, however, the other (more neoclassical) line of thinking which recommends
very little government intervention in the case of declining asset prices, as they argue
that it is intervention and regulation that caused cyclical ?uctuations in the ?rst place
(Andre et al., 2012; Aye et al., 2014b). Also, Blake et al. (1988) and Lossani and Tirelli
(1994) suggest that ?scal policy rules can be designed to steer national wealth to its
target value point to accommodate wealth expansions when the wealth level is below a
certain target value. Moreover, a tax increase may reduce the incentive to accumulate
JFEP
6,1
48
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
wealth since it reduces the incentive to earn income and increases the incentive to
consume, which will have negative repercussions for the whole economy. Under such
assumptions, time-varying models that account for the “state” of the economy and the
“state” of asset values are useful to disentangle the relationship between ?scal policy
and wealth dynamics.
In terms of the effects of asset price shocks on ?scal policy, Schuknecht and
Eschenbach (2002) conduct an empirical study of changes in real estate and asset price
on the ?scal balance across 17 OECD countries (OECD, 2012). The paper ?nds that asset
prices affect ?scal balances through the revenue channel; capital gains, turnover related
taxes as well as wealth effects and their impact on consumption are found to have an
impact on the ?scal balance. The study ?nds that, on average, a 10 per cent change in real
estate and stock prices have a similar effect on the ?scal balance as a 1 per cent change
in output. Tagkalakis (2011) augments a ?scal policy reaction function with ?nancial
variables for OECD countries. Looking at the impact on the ?scal balance, current
expenditure and current revenue, the author ?nds that an increase in asset prices has
a positive impact on the ?scal balance. Furthermore, the paper ?nds that residential price
changes play a bigger role in their effect on the budget balance, compared to commercial
property price and equity price changes. Agnello et al. (2012a, b) look at the impact of
asset market developments on ?scal policy. Employing both linear and non-linear
speci?cations, the study estimates a ?scal policy rule that includes ?nancial, as well as
housing wealth. The authors ?nd that in the linear speci?cation, spending does not react
to asset prices, but taxes and the primary surplus fall in reaction to a rise in stock prices,
and rise when house prices increase. Declining asset prices are also associated with
declining revenue collections, especially where capital and dividend taxes apply. Any
ramp up in listed company pro?ts will result in a higher dividends tax while increasing
house prices will increase revenue collected from capital gains.
A few studies (Du Plessis et al., 2007, 2008; Jooste et al., 2012) employ structural
VARs and vector error-correction (VEC) models, time-varying VARs, and dynamic
stochastic general equilibrium (DSGE) models to analyse simultaneously the effects of
business cycle, monetary policy, and ?scal policy shocks on output, consumption,
in?ation, and interest rates in South Africa. To the best of our knowledge, Aye et al.
(2014b) is the ?rst study to analyse simultaneously the effects of these shocks on South
African asset prices. That said, the literature on the effect of monetary policy on asset
prices in South Africa includes numerous studies. A number of those studies examine
the effects of monetary policy on equity prices (returns) in South Africa (Smal and de
Jager, 2001; Coetzee, 2002; Prinsloo, 2002; Durham, 2003; Hewson and Bonga-Bonga,
2005; Alam and Uddin, 2009; Chinzara, 2010; Mallick and Sousa, 2011; Mangani, 2011;
Muroyiwa, 2011), mainly based on (structural) VAR models and, at times, panel
data approaches that include South Africa. On the other hand, we know of only four
studies – Kasai and Gupta (2010), Gupta et al. (2010), Ncube and Ndou (2011) and
Simo-Kengne et al. (2013b) – that analyse the role played by the housing market in the
monetary policy transmission mechanism, using the effect of monetary policy shocks
on house prices in structural, factor-augmented, and Markov-switching VAR models.
These studies generally show that contractionary monetary policy leads to lower stock
and house prices.
This paper builds on the work of Aye et al. (2014b) that uses a sign-restriction
approach to capture the effects of ?scal policy shocks on asset prices. The theory-based
Time-varying
approach
49
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sign-restriction method allows this paper to identify shocks, such as tax
announcements and anticipation effects, on the macro economy. Aye et al. (2014b)
separate their results into expected and unexpected ?scal policy changes which
overcome problems of correctly identifying the shocks. The study shows that an
unanticipated and anticipated government revenue shock leads both house prices and
stock prices to decline. Anticipated government revenue shocks impact stock prices
more negatively and has a more persistent impact. The impact of an unanticipated
government spending shock has hardly any effect on house prices. Stock prices
respond positively. However, anticipated government spending shocks increase house
prices, but reduce equity prices. Our paper will benchmark the results against Aye et al.
(2014b) for South Africa. Barring the recent related paper by Agnello et al. (2012b),
which uses time varying transition probabilities in a two-state Markov-switching
framework to analyse the response of ?scal variables to asset prices for the US
economy[3], our paper is the ?rst in the literature of ?scal policy and asset prices to use
a time-varying VAR model to analyse not only the above relationship, but also the
effect of ?scal policy on asset prices. Note that, unlike the above approach of
Agnello et al. (2012b), the time-varying VAR approach allows us to treat each point in
time as a speci?c regime (rather than just assuming speci?c number of regimes) with
smooth transition across regimes, and also allow for stochastic volatility, ignoring
which, leads to biased estimates (see Section 3 for further details).
The rest of the paper unfolds as follows: Section 3 describes the empirical
methodology while Section 4 describes the data transformations and empirical results.
Finally, Section 5 concludes.
3. Empirical methodology
A vector autoregression (VAR), proposed by Sims (1980), has become a popular
technique used in econometric analysis and is adaptable to a vast array of economic
settings (Baltagi, 2011). In this study, a TVP-VAR model with stochastic volatility is
used. The TVP-VAR is common in the analysis of macroeconomic issues and allows us
to capture the time-varying nature of the underlying structure in the economy in
a ?exible and robust manner (Nakajima, 2011). The parameters in the VARspeci?cation
are assumed to follow a ?rst order random walk process, thereby incorporating both
temporary and permanent changes to the parameters. The inclusion of stochastic
volatility is an important aspect in this TVP-VAR model. In many situations,
a data-generating process of economic variables seems to have drifting coef?cients and
shocks of stochastic volatility. In that case, the application of a time-varying parameter
model but with constant volatility may result in biased estimations of the time-varying
coef?cients, since a possible variation of the volatility in disturbances is ignored. The
TVP-VAR model with stochastic volatility avoids this misspeci?cation and re?ects
simultaneous relations among variables of the model and heteroscedasticity of the
innovations (Primiceri, 2005)[4]. Although stochastic volatility makes the estimation
dif?cult due to the intractability of the likelihood function, the model can be estimated
using Markov Chain Monte Carlo (MCMC) methods in the context of a Bayesian
inference. Measuring the responses over time lends insight into the timing aspect of
government shocks on asset prices and analyses periods in which shocks were most
signi?cant. This variation over time can for example help to explain how ?scal shocks
relate to assets during booms and busts, and more recently, the ?nancial crisis.
JFEP
6,1
50
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Following Nakajima (2011), this paper estimates a time-varying parameter VAR
model with stochastic volatility of the form:
y
t
¼ c
t
þ B
1t
y
t21
þ · · · þ B
st
y
t2s
þ e
t
; e
t
, Nð0; V
t
Þ; ð1Þ
for t ¼ s þ 1, . . . , n, where y
t
is a (k £ 1) vector of observed variables, B
1t
, . . . , B
st
are
(k £ k) matrices of time-varying coef?cients, and V
t
is a (k £ k) time-varying
covariance matrix. A recursive identi?cation scheme is assumed by the decomposition
of V
t
¼ A
21
t
S
t
S
t
A
0
21
t
, where A
t
is a lower-triangle matrix with diagonal elements
equal to one, and S
t
¼ diagðs
1t
; . . . ; s
kt
Þ. Let us de?ne b
t
as the stacked row vector of
B
1t
, . . . , B
st
; a
t
is the stacked row vector of the free lower-triangular elements of A
t
; and
h
t
¼ ðh
1t
; . . . ; h
kt
Þ where h
jt
¼ log s
2
jt
. The time-varying parameters are assumed to
follow a random walk process:
b
tþ1
¼ b
t
þy
bt
;
a
tþ1
¼ a
t
þy
at
;
h
tþ1
¼ h
t
þy
ht
;
1
t
y
bt
y
at
y
ht
_
_
_
_
_
_
_
_
_
_
_
_
_
_
, N 0;
I 0 0 0
0 S
b
0 0
0 0 S
a
0
0 0 0 S
h
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
;
for t ¼ s þ 1, . . . , n, with e
t
¼ A
21
t
S
t
1
t
where S
a
and S
h
are diagonal, b
sþ1
,
Nðm
bo
; S
bo
Þ; a
sþ1
, Nðm
ao
; S
ao
Þ; and h
sþ1
, Nðm
ho
; S
ho
Þ[5]. A Bayesian inference is
used to estimate the TVP-VAR models via MCMC methods. The goal of MCMC
methods is to assess the joint posterior distributions of the parameters of interest
under certain prior probability densities that are set in advance. We assume the
following priors, as in Nakajima (2011): S
b
, IWð25; 0:01I Þ, ðS
a
Þ
22
i
, Gð4; 0:02Þ;
ðS
h
Þ
22
i
, Gð4; 0:02Þ; where ðS
a
Þ
22
i
and ðS
h
Þ
22
i
are the ith diagonal elements in S
a
and
S
h
, respectively. IW and G denotes the inverse Wishart and the gamma distributions,
respectively. For the initial set of the time-varying parameter, ?at priors are set such
that: m
bo
¼ m
ao
¼ m
ho
¼ 0 and S
bo
¼ S
ao
¼ S
ho
¼ 10 £ I :
3.1 Data description
Three variables are used in the analysis, with the sample period covering
1966:Q1-2012:Q2. We source the government’s budget balance data from the
South African Reserve Bank (where government revenue is subtracted from
government expenditure and expressed as a percent of GDP), Bloomberg for the All
Share Index and Amalgamated Bank of South Africa for the house price index. Both the
asset prices are expressed in real terms by de?ating the respective nominal series by the
CPI index. Note that all the variables were obtained in their seasonally adjusted form.
Log values of real house and stock prices are differenced to induce stationarity, and are
also standardised so that we can compare the magnitude of effect of ?scal policy across
the two asset prices, and also, the differences in the size of the feedback of the two asset
prices on ?scal policy behaviour. We use house, BB and JSE as short hand for the house
price index, the budget balance and the Johannesburg Stock Exchange All Share Index.
The variables pass the usual unit root tests namely, an augmented Dickey and Fuller
(ADF) (1981), Phillips and Perron (1988), Dickey-Fuller test with generalized least
Time-varying
approach
51
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
squares detrending (DF-GLS), the Kwiatkowski et al. (1992) test; the Elliot, Rothenberg,
and Stock (ERS) (1996) point optimal test, the Ng and Perron (2001) modi?ed versions of
the PP (NP-MZt) test and the ERS point optimal (NP-MPT) test. The stable[6] TVP-VAR
is estimated based on two lags, as was unanimously suggested by all the popular
lag-length tests, namely, the sequential modi?ed LR test statistic, the Akaike
information criterion, the Schwarz information criterion, applied to a constant parameter
VAR[7]. Accounting for stationarity and lags, our effective sample period start from
1966:4.
4. Empirical results
Anecdotal evidence shows that prior to the ?nancial crisis of 2008/2009, house prices
on average increased by 19 percent (2000-2007). The government recorded budget
surpluses in 2006/2007 and 2007/2008 and JSE All Share Index was growing at high
rates. The ?nancial crisis had a signi?cant impact on house prices and was
accompanied with decline in house prices by 0.2 per cent in 2009. The government, in
response to declining aggregate demand, increased government expenditure (together
with automatic effects of tax revenue decline) and as a consequence widened the
budget de?cit by 2.1 percent of GDP in 2008/2009 and 5.6 percent of GDP in 2009/2010.
While it is dif?cult to infer a direct relationship between government expenditure and
house prices, a simple multiplier (calculated as the ratio of the percentage change in
house prices to the percentage change in government expenditure) show that the
pre-crisis multiplier was stable at around 1.8 while during the crisis the multiplier
declined to 0.2 providing the ?rst indication of a nonlinear relationship.
The posterior estimates fromthe TVP-VARwere obtained after 10,000 samples were
drawn, with the ?rst 1,000 draws discarded. These posterior estimates for the means,
along with those for the standard deviations, the 95 per cent credibility intervals[8], the
convergence diagnostic (CD) due to Geweke (1992) and the inef?ciency factors
are presented in Table I[9]. The 95 per cent credibility intervals include the estimates for
the posterior means, and the CD statistics do not allow us to reject a null hypothesis of
convergence to the posterior distribution at a signi?cance level of 5 per cent. In general,
the inef?ciency factors are relatively low. We can thus conclude that the MCMC
algorithm is an ef?cient method of producing the posterior draws. Figure 1 presents the
estimation results of the TVP-VAR model with stochastic volatility.
Figure 2 plots the posterior estimates of stochastic volatility for each of the
variables used in the TVP-VAR. The estimates for the stochastic volatility shock hint
to at least two regimes; one characterised by a pre-1990 era and one of a post-1990 era.
Parameter Mean SD 95% intervals CD Inef.
(S
b
)
1
0.0374 0.0094 (0.0240,0.0596) 0.134 48.31
(S
b
)
2
0.0271 0.0049 (0.0192,0.0383) 0.237 30.12
(S
a
)
1
0.0538 0.0142 (0.0336,0.0911) 0.099 52.26
(S
a
)
2
0.0499 0.0114 (0.0329,0.0779) 0.938 26.80
(S
h
)
1
0.0826 0.0389 (0.0431,0.1886) 0.142 183.30
(S
h
)
2
0.0800 0.0342 (0.0398,0.1725) 0.862 156.84
Note: The estimates of S
b
and S
a
are multiplied by 100
Table I.
Selected estimation
results
JFEP
6,1
52
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 1.
Moments and posterior
distributions
N
o
t
e
s
:
S
a
m
p
l
e
a
u
t
o
c
o
r
r
e
l
a
t
i
o
n
s
(
t
o
p
p
a
n
e
l
)
,
s
a
m
p
l
e
p
a
t
h
s
(
m
i
d
d
l
e
p
a
n
e
l
)
,
a
n
d
p
o
s
t
e
r
i
o
r
d
e
n
s
i
t
i
e
s
(
b
o
t
t
o
m
p
a
n
e
l
)
;
t
h
e
e
s
t
i
m
a
t
e
s
o
f
?
b
a
n
d
?
a
a
r
e
m
u
l
t
i
p
l
i
e
d
b
y
1
0
0
Time-varying
approach
53
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Coincidentally this shift overlaps with South Africa’s democratic transition with the
release of Nelson Mandela. The non-steady stochastic volatility justi?es the use of a
TVP-VAR model to capture possible regime and period speci?c changes.
Impulse responses are used as a tool to capture the macroeconomic dynamics in the
estimated VAR system. For a standard constant parameter VAR model, the impulse
responses are drawn for each set of two variables, whereas for a TVP-VAR model,
the impulse responses can be drawn in an additional dimension, as the responses are
computed at all points in time using the time-varying parameters. There are several
ways to simulate the impulse responses based on the parameter estimates of the
TVP-VAR model. Following Nakajima (2011), we compute the impulse responses by
?xing an initial shock size equal to the time-series average of stochastic volatility over
the sample period, and using the simultaneous relations at each point in time, for
considering the comparability over time. We identify the three structural shocks (house
demand, ?scal policy and stock demand (portfolio)) using a recursive or Cholesky
identi?cation scheme, as obtained based on the lower-triangular matrix A
t
. We order
the variables as follows: house, BB and JSE following Agnello and Sousa (2014). The
ordering of the two asset prices relative to the ?scal policy instrument is quite intuitive:
the stock price is ordered last as it refers to assets that are traded in markets where
auctions take place instantaneously. While, the house price was ordered ?rst in the
Figure 2.
Posterior estimates
for the stochastic
volatility of the
structural shock
Notes: Top panel presents the data values; bottom panel depicts the posterior mean estimates
(solid line) and 95 percent credible intervals (dotted lines) for stochastic volatility of a
structural shock
JFEP
6,1
54
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
system to account for the fact that housing markets are inherently sticky and so house
prices do not immediately reach the equilibrium after a shock. Then there is a
“time-to-build” argument suggesting that it takes time for developers to bring new
houses to the market or to work out of inventories when demand increases. Further,
the matching between buyers and sellers requires time, and ?nally, one needs to also
account for important transaction costs inherent to trading housing up or down.
To compute the recursive innovation of the variable, the estimated time-varying
coef?cients are used from the current date to future periods. Around the end of the
sample period, the coef?cients are set constant in future periods for convenience.
Although a time series of impulse responses for selected horizons or impulse responses
for selected periods are often exhibited in the literature, one could draw a
three-dimensional plot for the time-varying impulse responses.
Figure 3 plots the mean impulse response function of asset prices in reaction to a
shock in ?scal policy. Contemporaneous ?scal policy shocks are analysed over
different horizons, over time and in terms of magnitude. House prices respond with a
lag due to the VAR ordering. This response is mainly positive following a ?scal policy
shock. As discussed earlier, negative tax shocks have wealth effects that could lead to
a higher demand of assets. This can also occur with a rise in government spending,
especially when spending is concentrated around wage increases. Its amplitude varies
over time with the most signi?cant impact being during the ?nancial crisis in 2010
(with a multiplier of 0.4). The impulse responses of house prices peak four quarters
ahead before dissipating. Apart from the late 1980s and the shock in 2009/2010,
Figure 3.
Impulse response function
of ?scal policy following a
shock to real house prices
0.4
0.2
0
–0.2
2010
2000
1990
1980
1970
2
4
6
8
10
12
1.5
1
0.5
0
–0.5
2010
2000
1990
1980
1970
2
4
6
8
10
12
M
a
g
n
i
t
u
d
e
M
a
g
n
i
t
u
d
e
Y
e
a
r
I
R
F
h
o
r
i
z
o
n
Y
e
a
r
I
R
F
h
o
r
i
z
o
n
House response to BB JSE response to BB
Time-varying
approach
55
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
house price responses were relatively short lived (six quarters). In comparison, equity
price responses are rather mute following a budget shock. In terms of size, the initial
impact is largest (the multiplier is not signi?cantly bigger than 0.5 over towards the
end of the sample period). Increases are quickly met with a decline in equity prices
which would suggest that markets are quick to adjust to any ?scal news.
Accumulating over the impulse horizon reveals that ?scal shocks have a negligible
impact on stock prices. It is also interesting to note that the contemporaneous impact of
?scal shocks on equity prices tapered down since the 1970s with an initial impact of
close to 1. From Figure 3 it becomes clear that house prices respond more starkly
compared to equity prices following a ?scal shock. The shock also lasts longer on
house prices. Quite surprising is that house prices responded more than equity prices
during the 2008/2009 ?nancial crisis. One hypothesis could be that the ?nancial crisis
represented a liquidity and solvency problem more relevant for asset classes such as
houses than investments into the equity market.
This study also looks at the impact of shocks in stock and house prices on the
government’s budget balance. A priori, one would expect that an increase in asset prices
will lead to higher tax revenues through property and capital gains taxes which should
lead to a contracting budget de?cit. The channels through which asset price shocks
effect the budget balance is rather complex and requires a detailed decomposition of
expectations, output and interest rates. For one, equity price increases could exert
upward pressure on government bond interest rates if there is a substitution away from
bonds. This in turn will lead to a rising de?cit. After a period government would collect
revenue from dividend pay outs which should reduce the de?cit. In essence the shock of
asset prices to the balance relies on the net effect revenue gains minus the effect of
a potential increase in debt service costs. Figure 4 show that higher house prices had
a negative and signi?cant effect on the budget de?cit (de?cit reducing) throughout our
sample and at various horizons. Although the contemporaneous impact seems constant,
the impact was largest during the ?nancial crisis. House price shocks also lasts for
almost two years (in 2008-2012) compared to a relatively short-lived outcome in the 1970s
and 1980s. One of the reasons why house price shocks had a larger impact on the balance
is due to modernisation processes of municipalities and tax collecting authorities which
made collecting taxes more ef?cient. This part of our sample was also coincidentally
associated with South Africa’s, and for that matter the majority of the Western world,
housing boom period. House price shocks during this period are slightly more persistent
and have a bigger impact on the budget de?cit. However, during the ?nancial crisis
house price shocks had a smaller impact of the de?cit.
Finally, the right hand side of Figure 4 shows the propagation of stock price shocks
to the budget balance. The impact of stock price shocks on the budget de?cit varies
between positive and negative during different horizons. The period during the
?nancial crisis highlights that a rise in equity prices increased the de?cit. As shown by
Aye et al. (2013b) asset price shocks lead to an increase in interest rates. The interest
rate channel could be used to motivate how equity price shocks could lead to an
increase in the budget de?cit, especially if it causes a substitution away from bonds to
equities. That being said, stock price shocks have only a transitory impact on the
budget as the shocks dissipate already after six to eight quarters[10].
These results are in line with Aye et al. (2014b): expansionary ?scal shocks lead to
an increase in asset prices, but are only transitory. As mentioned earlier, it could be
JFEP
6,1
56
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that markets are quick to adjust when hit by shocks. Usually expansionary ?scal
policy ?nds its way in the public discourse which would allow markets to anticipate
?scal shocks more reliably. This would only strengthen an argument for the short lived
response of ?scal shocks. The results show that on average, an expansionary ?scal
policy shock has a small impact on house prices (as was the case in Aye et al. (2014b)
given a spending shock). The impact, however, became more pronounced at the onset
of the ?nancial crisis which would suggest that effects are ampli?ed under distressed
economic conditions. Furthermore, looking at the impact of ?scal shocks and asset
prices could provide a channel through which an explanation can be given for private
investment being crowded out when ?scal policy expands. However, this requires a
more detailed analysis that is beyond the scope of this study.
On the other hand, asset price shocks represent a possible increase in revenue
collection which should effectively reduce budget de?cits. The results in this paper
con?rm that asset price shocks reduced the de?cits. With new modernisation
programmes from the revenue collecting authorities and a bigger emphasis and tax
broadening, asset price shocks have had a larger impact on the budget post-2000.
5. Conclusion
This paper uses a three variable (stock prices, house prices and government’s budget
balance) TVP-VAR with stochastic volatility to study the simultaneous impact of ?scal
shocks on asset prices, and asset price shocks on ?scal policy. We ?nd that ?scal
shocks had a small impact on asset prices which is in line with Aye et al. (2014b).
Figure 4.
Impulse response function
of ?scal policy following a
shock to real stock prices
BB response to House BB response to JSE
0
–0.1
–0.2
–0.3
–0.4
–0.5
–0.6
2010
2000
1990
1980
1970
2
4
6
8
10
12
0.5
–0.5
–1
0
–1.5
2010
2000
1990
1980
1970
2
4
6
8
10
12
M
a
g
n
i
t
u
d
e
M
a
g
n
i
t
u
d
e
Y
e
a
r
I
R
F
h
o
r
i
z
o
n
Y
e
a
r
I
R
F
h
o
r
i
z
o
n
Time-varying
approach
57
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The results show that ?scal policy and asset price shocks have varying impacts over
time. Furthermore, the results show that extreme economic events such as the recent
?nancial crisis change the impact of these shocks. The recent ?nancial crisis shows
that ?scal shocks amplify the effects on house prices while the effects on equity prices
become more subdued. Information ?ows are a lot more transparent and ef?cient in
equity markets while house price stickiness could explain the different responses
between the asset classes. At times when the budget balance seemed unsustainable,
a consolidating budget balances seem to have a positive impact on asset prices while
increasing asset prices reduces de?cits as tax collections improve. However, as shown
in the study of Aye et al. (2014b) increasing taxes will limit the amount of revenue
collected as it reduces real asset prices. This suggests that consolidation effects, when
considering the impact on asset prices, should happen through spending channels. The
policy implication is that expansionary government decisions have clear wealth
effects, but this depends on speed at which information can be absorbed. In the case of
equity markets, it would seem that information is relatively quickly absorbed and thus
the impact on equity markets would depend on their views on the sustainability of
?scal balances and the impact of spending and tax decisions.
Notes
1. For detailed international literature reviews on studies involving monetary policy and asset
prices, see Bjørnland and Leitemo (2009), Iglesias and Haughton (2011), Gupta et al. (2012a, b)
and Bjørnland and Jacobsen (2014).
2. Tobin’s q equals the ratio of the stock market value of a ?rm to the replacement cost of its
capital.
3. This paper tested for nonlinear effects of asset prices on the US ?scal policy. By modeling
government spending and taxes as time-varying transition probability Markovian
processes, the authors found that taxes signi?cantly adjust in a nonlinear fashion to asset
prices. In particular, taxes respond to housing and (to a smaller extent) to stock prices
changes during normal times. However, at periods characterized by high ?nancial volatility,
government taxation only counteracts stock market developments (and not the dynamics of
the housing sector). On the other hand government spending, is neutral vis-a` -vis the asset
market cycles.
4. We also estimated a TVP-VAR model without stochastic volatility. Though, the qualitative
behaviour of the impulse responses, in general, is very similar, the impulse responses are
very volatile. Also, quantitatively, the effects are larger as well. This is not surprising, since
we do not allow for heteroscedastic disturbances, the parameters are not only estimated with
less precision, but also tends to be biased upwards to in?ate the multipliers (Nakajima, 2011).
The details of these results are available upon request from the authors.
5. For a comprehensive analysis of the TVP-VAR methodology and the estimation algorithm
(Nakajima, 2011).
6. The constant parameter VAR is found to be stable as all roots were found to lie within the
unit circle.
7. Complete details of the unit root, stability and lag length tests are available from the authors
upon request.
8. Credibility intervals are used in the Bayesian paradigm as opposed to “con?dence” intervals
which belong in the frequentist realm.
JFEP
6,1
58
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
9. Geweke (1992) suggests the comparison between the ?rst n
0
draws and the last n
1
draws,
dropping out the middle draws, to check for convergence in the Markov chain. The CD
statistics are computed as follows:
CD ¼ ð x
0
2 x
1
Þ=
?????????????????
^ s
2
0
n
0
þ
^ s
2
1
n
1
¸
where:
x
j
¼
1
n
j
_ _
mjþnj21
i¼mj
x
ði Þ
where x
j
being the ith draw and ^ s
2
j
=n
j
_ _
is the standard error of x
j
, respectively, for j ¼ 0, 1.
If the sequence of the MCMC sampling is stationary, it converges to a standard normal
distribution. We set m
0
¼ 1, n
0
¼ 1,000, n
1
¼ 5,001 and n
2
¼ 5,000. ^ s
2
j
is computed using a
Parzen window with bandwidth (B
m
) ¼ 500. The inef?ciency parameter is de?ned as:
1 þ 2
B
m
i¼1
r
i
;
where r
i
is the sample autocorrelation at lag s, which is computed to measure how well the
MCMC chain mixes.
10. We also estimated a constant parameter VAR and analysed the effect of a ?scal shock on
house prices and stock prices, as well as, the response of ?scal policy to shocks on the asset
prices. Realizing that the shape of the impulse response in the constant VAR model is
associated with the average level of the response in the TVP-VAR model to some extent, we
basically obtain the general conclusions of the TVP-VAR model. The only exception being
the negative, but insigni?cant, impact of the budget balance on house prices for the ?rst two
quarters. Again, the possible biasedness in the parameter estimates for not allowing
time-variation, as well as, stochastic volatility in the error structure could be a reason behind
such contradictory impulse behaviour (Nakajima, 2011). The details of these results are
available upon request from the authors.
References
Afonso, A. and Sousa, R.M. (2011), “What are the effects of ?scal policy on asset markets?”,
Economic Modelling, Vol. 28, pp. 1871-1890.
Afonso, A. and Sousa, R.M. (2012), “The macroeconomic effects of ?scal policy”, Applied
Economics, Vol. 44 No. 34, pp. 4439-4454.
Agnello, L. and Sousa, R.M. (2014), “Fiscal policy and asset prices”, Bulletin of Economic
Research (in press).
Agnello, L., Castro, V. and Sousa, R.M. (2012a), “How does ?scal policy react to wealth
composition and asset prices?”, Journal of Macroeconomics, Vol. 34 No. 3, pp. 874-890.
Agnello, L., Dufrenot, G. and Sousa, R.M. (2012b), “Adjusting the US ?scal policy for asset prices:
evidence from a TVP-MS framework”, NIPE Working Papers, NIPE, Universidade do
Minho, Braga.
Alam, M.M. and Uddin, M.G. (2009), “Relationship between interest rate and stock price:
empirical evidence from developed and developing countries”, International Journal of
Business and Management, Vol. 4 No. 3, pp. 43-51.
Time-varying
approach
59
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Andre, C., Gupta, R. and Kanda, P.T. (2012), “Do house prices impact consumption and interest
rate? Evidence from OECD countries using an agnostic identi?cation procedure”, Applied
Economics Quaterly, Vol. 58 No. 1, pp. 19-70.
Apergis, N., Simo-Kengne, B.D. and Gupta, R. (2013), “The long-run relationship between
consumption, house prices and stock prices in South Africa: evidence from provincial-level
data”, Department of Economic Working Paper No. 2013/26, University of Pretoria,
Pretoria.
Aron, J. and Muellbauer, J. (2013), “Wealth, credit conditions and consumption: evidence from
South Africa”, Review of Income and Wealth, Vol. 59, Supplement S1, pp. S161-S196.
Aye, G.C., Balcilar, M. and Gupta, R. (2013a), “Long- and short-run relationships between house
and stock prices in South Africa: a nonparametric approach”, Journal of Housing Research,
Vol. 22 No. 2, pp. 203-219.
Aye, G.C., Gupta, R. and Modise, P. (2014a), “Do stock prices impact consumption and interest
rates in South Africa? Evidence from a time varying vector autoregressive model”, Journal
of Emerging Market Finance (in press).
Aye, G.C., Gupta, R., Kaninda, A., Nyakabawo, W. and Razak, A. (2013b), “House price, stock
price and consumption in South Africa? A structural VAR approach”, Corporate
Ownership and Control, Vol. 10 No. 2, pp. 585-589.
Aye, G.C., Balcilar, M., Gupta, R., Jooste, C., Miller, S. and Ozdemir, Z. (2014b), “Fiscal policy
shocks and the dynamic asset prices: the South African experience”, Public Finance Review
(in press).
Baltagi, B.H. (2011), Econometrics, Springer, Berlin.
Bjørnland, H.C. and Jacobsen, D.H. (2014), “House prices and stock prices: different roles in the
US monetary transmission mechanism”, Scandinavian Journal of Economics (in press).
Bjørnland, H.C. and Leitemo, K. (2009), “Identifying the interdependence between US monetary
policy and the stock market”, Journal of Monetary Economics, Vol. 56 No. 2, pp. 275-282.
Blake, A., Vines, D. and Weale, M. (1988), “Wealth targets, exchange rate targets and
macroeconomic policy”, Discussion Paper No. 246.
Castro, V. (2011), “Can central banks’ monetary policy be described by a linear (augmented)
Taylor rule or by a nonlinear rule?”, Journal of Financial Stability, Vol. 7 No. 4, pp. 228-246.
Chinzara, Z. (2010), “Macroeconomic uncertainty and emerging stock market volatility: the case
for South Africa”, Economic Research Southern Africa Working Paper No. 187.
Coetzee, C.E. (2002), “Monetary conditions and stock returns: a South African case study”,
EconWPA, 0205002.
Das, S., Gupta, R. and Kanda, P.T. (2011), “Bubbles in South African house prices and their
impact on consumption”, Journal of Real Estate Literature, Vol. 19 No. 1, pp. 71-92.
Dickey, D. and Fuller, W. (1981), “Likelihood ratio statistics for autoregressive time series with
a unit root”, Econometrica, Vol. 49, pp. 1057-1072.
Du Plessis, S., Smit, B. and Sturzenegger, F. (2007), “The cyclicality of monetary and ?scal policy
in South Africa since 1994”, South African Journal of Economics, Vol. 75 No. 3, pp. 391-411.
Du Plessis, S., Smit, B. and Sturzenegger, F. (2008), “Identifying aggregate supply and demand
shocks in South Africa”, Journal of African Economies, Vol. 17 No. 5, pp. 765-793.
Durham, J.B. (2003), “Monetary policy and stock prices returns”, Financial Analysts Journal,
Vol. 59 No. 4, pp. 26-35.
Elliot, G., Rothenberg, T.J. and Stock, J.H. (1996), “Ef?cient tests for an autoregressive unit root”,
Econometrica, Vol. 64, pp. 813-836.
JFEP
6,1
60
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
European Central Bank (2010), “Asset price bubbles and monetary policy revisited”, ECB
Monthly Bulletin, November, pp. 71-83.
Feldstein, M. (2009), “Rethinking the role of ?scal policy”, Contributed paper at the January 2009
Meeting of the American Economic Association.
Geweke, J. (1992), “Evaluating the accuracy of sampling-based approaches to the calculation of
posterior moments”, in Bernado, J.M., Berger, J.O., Dawid, A.P. and Smith, A.F.M. (Eds),
Bayesian Statistics 4, Oxford University Press, New York, NY, pp. 169-188.
Granville, B. and Mallick, S.K. (2009), “Monetary and ?nancial stability in the Euro area:
procyclicality versus trade-off”, Journal of International Financial Markets, Institutions
and Money, Vol. 19, pp. 662-674.
Gupta, R., Jurgilas, M. and Kabundi, A. (2010), “The effect of monetary policy on real house price
growth in South Africa: a factor-augmented vector autoregression (FAVAR) approach”,
Economic Modelling, Vol. 27 No. 1, pp. 315-323.
Gupta, R., Jurgilas, M., Miller, S.M. and van Wyk, D. (2012a), “Financial market liberalization,
monetary policy, and housing sector dynamics”, International Business and Economics
Research Journal, Vol. 11 No. 1, pp. 69-82.
Gupta, R., Kabundi, A., Jurgilas, M. and Miller, S.M. (2012b), “Monetary policy and housing
sector dynamics in a large-scale Bayesian vector autoregressive model”, International
Journal of Strategic Property Management, Vol. 16 No. 1, pp. 1-20.
Hewson, M. and Bonga-Bonga, L. (2005), “The effects of monetary policy shocks on stock returns
in South Africa: a structural vector error correction model”, paper presented at the
Economic Society of South Africa Conference, Durban, South Africa.
Iacoviello, M. (2010), “Housing in DSGE models: ?ndings and new directions”, in de Bandt, O.,
Knetsch, T., Pen˜ alosa, J. and Zollino, F. (Eds), Housing Markets in Europe:
A Macroeconomic Perspective, Springer, Berlin, pp. 3-16.
Iacoviello, M. (2011), “Housing wealth and consumption”, International Encyclopedia of Housing
and Home, Elsevier, Amsterdam.
Iglesias, E.M. and Haughton, A.Y. (2011), “Interaction between monetary policy and stock prices:
a comparison between the Caribbean and the US”, Contributed paper at the Conference on
the Economy (COTE) 2011, The University of the West Indies, St Augustine, Trinidad and
Tobago, October.
Jooste, C., Liu, G. and Naraidoo, R. (2012), “Analysing the effects of ?scal policy shocks in the
South African economy”, Working Paper Series No. WP 2012-06, University of Pretoria,
Pretoria.
Kasai, N. and Gupta, R. (2010), “Financial liberalisation and the effectiveness of monetary policy
on house prices in South Africa”, ICFAI Journal of Monetary Economics, Vol. 8 No. 4,
pp. 59-74.
Kwiatkowski, D., Phillips, P., Schmidt, P. and Shin, J. (1992), “Testing the null hypothesis
of stationarity against the alternative of a unit root”, Journal of Econometrics, Vol. 54,
pp. 159-178.
Lossani, M. and Tirelli, P. (1994), “Simple rules, ?scal policy and wealth targets in a monetary
union: an alternative to the Maastricht guidelines”, Scottish Journal of Political Economy,
Vol. 41 No. 4, pp. 372-393.
Mallick, S.K. and Mohsin, M. (2007), “Monetary policy in high in?ation open economies: evidence
fromIsrael andTurkey”, International Journal of Finance and Economics, Vol. 12, pp. 405-415.
Mallick, S.K. and Mohsin, M. (2010), “On the real effects of in?ation in open economies: theory
and empirics”, Empirical Economics, Vol. 39, pp. 643-673.
Time-varying
approach
61
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Mallick, S.K. and Sousa, M.R. (2011), “In?ationary pressures and monetary policy: evidence
from BRICS economies”, paper presented at the Quantitative and Qualitative Analysis in
Social Sciences Conference, available at: www.qass.org.uk/2011-May_Brunel-conference/
Mallick.pdf
Mangani, R. (2011), “Monetary policy, structural breaks and JSE returns”, Investment Analysts
Journal, Vol. 73, pp. 27-35.
Mountford, A. and Uhlig, H. (2009), “What are the effects of ?scal policy shocks?”, Journal of
Applied Econometrics, Vol. 24, pp. 960-992.
Muroyiwa, B. (2011), “Identifying the interdependence between South Africa’s monetary policy
and the stock market”, unpublished MSc thesis, Department of Economics and Economic
History, Rhodes University, Grahamstown.
Nakajima, J. (2011), “Time-varying parameter VAR model with stochastic volatility: an overview
of methodology and empirical applications”, Monetary and Economic Studies, Vol. 29,
pp. 107-142.
Ncube, M. and Ndou, E. (2011), “Monetary policy transmission, house prices and consumer
spending in South Africa: an SVAR approach”, Working Paper No. 133, African
Development Bank Group, Abidjan.
Ng, S. and Perron, P. (2001), “Lag length selection and the construction of unit root tests with
good size and power”, Econometrica, Vol. 69, pp. 1519-1554.
OECD (2012), “Debt and macroeconomic stability”, WP(2012)22, OECD, Paris.
Peretti, V., Gupta, R. and Inglesi-Lotz, R. (2012), “Do house prices impact consumption and
interest rate in South Africa? Evidence from a time varying vector autoregressive model”,
Economics, Financial Markets and Management, Vol. 7 No. 4, pp. 101-120.
Phillips, P. and Perron, P. (1988), “Testing for a unit root in time series regression”, Biometrika,
Vol. 75, pp. 335-346.
Primiceri, G. (2005), “Time varying structural vector autoregressions and monetary policy”,
Review of Economic Studies, Vol. 72, pp. 821-852.
Prinsloo, J.W. (2002), “Household debt, wealth and savings”, Quarterly Bulletin, December,
pp. 63-78.
Ra?q, M.S. and Mallick, S.K. (2008), “The effect of monetary policy on output in EMU3: a sign
restriction approach”, Journal of Macroeconomics, Vol. 30, pp. 1756-1791.
Schuknecht, L. and Eschenbach, F. (2002), “Asset prices and ?scal balances”, ECB Working
Paper No. 141, European Central Bank, Frankfurt.
Schuknecht, L., von Hagen, J. and Wolswijk, G. (2009), “Government risk premiums in the bond
market: EMU and Canada”, European Journal of Political Economy, Vol. 25, pp. 371-384.
Simo-Kengne, B.D., Bittencourt, M. and Gupta, R. (2012), “House prices and economic growth in
South Africa: evidence from provincial-level data”, Journal of Real Estate Literature,
Vol. 20, pp. 97-117.
Simo-Kengne, B.D., Bittencourt, M. and Gupta, R. (2013a), “The impact of house prices on
consumption in South Africa: evidence from provincial-level panel VARs”, Housing
Studies, Vol. 28 No. 8, pp. 1133-1154.
Simo-Kengne, B.D., Balcilar, M., Gupta, R., Aye, G.C. and Reid, M. (2013b), “Is the relationship
between monetary policy and house prices asymmetric in South Africa? Evidence from
a Markov-switching vector autoregressive model”, Economic Modelling, Vol. 32 No. 1,
pp. 161-171.
Sims, C.A. (1980), “Macroeconomics and reality”, Econometrica, Vol. 48, pp. 1-48.
JFEP
6,1
62
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Smal, M.M. and de Jager, S. (2001), “The monetary policy transmission mechanism in
South Africa”, Occasional Paper No. 16, South Africa Reserve Bank, Pretoria.
Sousa, R.M. (2010a), “Consumption (dis)aggregate wealth, and asset returns”, Journal of
Empirical Finance, Vol. 17, pp. 606-622.
Sousa, R.M. (2010b), “Housing wealth, ?nancial wealth, money demand and policy rule: evidence
from the Euro area”, The North American Journal of Economics and Finance, Vol. 21,
pp. 88-105.
Sousa, R.M. (2010c), “The consumption-wealth ratio and asset returns: the Euro area, the UK and
the US”, Journal of Money, Investment and Banking, Vol. 16, pp. 12-29.
Tagkalakis, A. (2011), “Fiscal policy and ?nancial market movements”, Journal of Banking and
Finance, Vol. 35, pp. 231-251.
Venter, J.C. (2011), “Business cycles in South Africa during the period 2007-2009”, Quarterly
Bulletin, June, pp. 61-66.
Corresponding author
Rangan Gupta can be contacted at: [email protected]
Time-varying
approach
63
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_278999260.pdf
This paper aims to study the interplay of fiscal policy and asset prices in a time-varying
fashion.
Journal of Financial Economic Policy
A time-varying approach to analysing fiscal policy and asset prices in South Africa
Rangan Gupta Charl J ooste Kanyane Matlou
Article information:
To cite this document:
Rangan Gupta Charl J ooste Kanyane Matlou , (2014),"A time-varying approach to analysing fiscal policy
and asset prices in South Africa", J ournal of Financial Economic Policy, Vol. 6 Iss 1 pp. 46 - 63
Permanent link to this document:http://dx.doi.org/10.1108/J FEP-01-2013-0003
Downloaded on: 24 January 2016, At: 21:48 (PT)
References: this document contains references to 68 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 209 times since 2014*
Users who downloaded this article also downloaded:
Matthew Kofi Ocran, (2011),"Fiscal policy and economic growth in South Africa", J ournal of Economic
Studies, Vol. 38 Iss 5 pp. 604-618http://dx.doi.org/10.1108/01443581111161841
Saumya Ranjan Dash, J itendra Mahakud, (2015),"Market anomalies, asset pricing models, and stock
returns: evidence from the Indian stock market", J ournal of Asia Business Studies, Vol. 9 Iss 3 pp. 306-328http://dx.doi.org/10.1108/J ABS-06-2014-0040
Pablo A. Caron, Marcela A. Cruchaga, Axel E. Larreteguy, (2015),"Sensitivity analysis of finite volume
simulations of a breaking dam problem", International J ournal of Numerical Methods for Heat & Fluid
Flow, Vol. 25 Iss 7 pp. 1718-1745http://dx.doi.org/10.1108/HFF-10-2014-0308
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
A time-varying approach
to analysing ?scal policy and
asset prices in South Africa
Rangan Gupta, Charl Jooste and Kanyane Matlou
Department of Economics, University of Pretoria, Pretoria, South Africa
Abstract
Purpose – This paper aims to study the interplay of ?scal policy and asset prices in a time-varying
fashion.
Design/methodology/approach – Using South African data since 1966, the authors are able to
study the dynamic shocks of both ?scal policy and asset prices on asset prices and ?scal policy based
on a time-varying parameter vector autoregressive (TVP-VAR) model. This enables the authors to
isolate speci?c periods in time to understand the size and sign of the shocks.
Findings – The results seem to suggest that at least two regimes exist in which expansionary ?scal
policy affected asset prices. From the 1970s until 1990, ?scal expansions were associated with
declining house and slightly increased stock prices. The majority of the ?rst decade of 2000 had asset
prices increasing when ?scal policy expanded. On the other hand, increasing asset prices reduced
de?cits for the majority of the sample period, while the recent ?nancial crises had a marked change on
the way asset prices affect ?scal policy.
Originality/value – This is the ?rst attempt in the literature of ?scal policy and asset prices to use a
TVP-VAR model to not only analyse the impact of ?scal policy on asset prices, but also the feedback
from asset prices to ?scal policy over time.
Keywords House prices, TVP-VAR, Stock prices, Countercyclical ?scal policy
Paper type Research paper
1. Introduction
The recent global ?nancial crisis demonstrates that boom/bust cycles in asset prices
can dramatically affect macroeconomic stability, especially output and price stability.
Hence, the importance of monetary and ?scal policy in sustaining economic growth
during and after the ?nancial crisis has become a dominant area of study. Analysts
typically focus on monetary policy to consider the linkages between economic policy
and asset markets[1]. Whilst monetary policy dominated the ?eld of academic and
policy discussions on controlling elements of the business cycle, ?scal policy has
become key after monetary policy reached the zero interest rate lower bound and
became ineffective in stimulating demand during the recent recession (Feldstein, 2009).
Large and persistent ?scal stimulus, however, can lead to long-term unsustainability of
sovereign ?nances as seen when analysing current government bond markets
(Schuknecht et al., 2009). Researchers need to disentangle this effect, however, from the
mess left by ?nancial institutions in Europe and the USA. Furthermore, this may lead
to business cycle de-synchronization (Ra?q and Mallick, 2008; Mallick and Mohsin,
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JEL classi?cation – C11, C15, C32, H30, H61
The authors would like to thank two anonymous referees and Jouchi Nakajima for many
helpful comments. The usual disclaimer applies.
Journal of Financial Economic Policy
Vol. 6 No. 1, 2014
pp. 46-63
qEmerald Group Publishing Limited
1757-6385
DOI 10.1108/JFEP-01-2013-0003
JFEP
6,1
46
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
2007, 2010) or negatively affect the nexus between monetary and ?nancial stability
(Castro, 2011; Granville and Mallick, 2009; Sousa, 2010a).
Despite the large number of studies analysing the macroeconomic effects of ?scal
policy (see Mountford and Uhlig, 2009; Afonso and Sousa, 2012 for detailed reviews), the
importance of asset markets over the business cycle (Afonso and Sousa, 2011; Agnello and
Sousa, 2014; Iacoviello, 2010, 2011), andthe feedback of asset prices to ?scal policy(refer to
Agnello et al., 2012a for a detailed review), an important gap in the literature exists
regarding the empirical relationship between ?scal policy actions and developments in
asset prices and in turn, the possible feedback from asset prices to ?scal policy stance,
especially in emerging market economies. This study concentrates on South Africa, given
our familiarity with the economic structure of the economy. In South Africa, non-housing
wealth (housing wealth) equals 49.95 per cent (31.13 per cent) of household’s total assets
and 61.59 per cent (38.41 per cent) of household’s net worth in 2011 (Aye et al., 2013a).
Hence, it is not surprising that recent evidence show that Aron and Muellbauer (2013),
Das et al. (2011), Ncube and Ndou (2011), Apergis et al. (2013), Simo-Kengne et al. (2012,
2013a), Peretti et al. (2012) and Aye et al. (2013b) there are signi?cant spill-overs onto
consumption and output from not only the stock market, but also the housing market.
Also, as highlighted by the time-varying approaches of Peretti et al. (2012) and Aye et al.
(2014a), the South African economy began slowing by the end of 2007, as the stock and
housing markets entered deep bear markets (Venter, 2011; Simo-Kengne et al., 2013b).
This paper attempts to contribute to the existing literature, and hence our main
contribution, by focussing on the consequences of ?scal policy/asset price shocks on
asset prices/?scal policy in speci?c periods and over different regimes. This study
focuses on the interplay of South African asset prices and ?scal policy. Time varying
parameters in a model which links these variables in a simultaneous setup enables a
bird’s eye view of certain events and periods such as the recent ?nancial crisis.
In particular, we analyse not only the effects of ?scal policy shocks, but also look at
asset price shocks to understand its impact on ?scal variables, and to the extent that we
?nd a link between them, we look at the magnitude and the persistence of the effects.
2. Literature review
The behaviour of asset markets and their prices emerges as an important factor for
the decision making of ?nancial institutions, homeowners and consumers, businesses,
and policy makers. The linkages between the ?nancial market and the banking
system, the housing sector, and the credit market produced strong and powerful effects
in the course of the ?nancial turmoil (Afonso and Sousa, 2011). According to the
European Central Bank (2010), a variety of mechanisms exist through which asset
prices can affect consumption spending. For example, a wealth effect working through
consumers and a “q-effect”[2] working through businesses can affect asset prices.
House price bubbles, which arose in most developed and emerging-market countries
prior to the ?nancial crisis, led to unsustainable borrowing by homeowners to ?nance
consumption against “seemingly” permanent increases in their equity holdings.
If q increases as a result of an increase in equity prices, the ?rm can raise more capital
by issuing new equity. This makes it more attractive for ?rms to raise new capital,
thus increasing investment demand, which may, in turn, lead to higher prices for goods
and services. Additional effects can stem from residential property prices, which, via
higher wage demands by workers, may lead to increases in both the prices of goods
Time-varying
approach
47
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and services and, therefore, consumer prices. Finally, movements in asset prices can
signi?cantly affect business and consumer con?dence. Hence, researchers now focus
their attention on the relationship between macroeconomic variables, wealth, and asset
returns (see Sousa, 2010b, c; Afonso and Sousa, 2011, 2012; Agnello and Sousa, 2014;
Peretti et al., 2012; Simo-Kengne et al., 2013a for detailed literature reviews).
Our understanding of the transmission of ?scal policy innovations to asset markets
is limited, however, exists because of the few studies concentrating on US and
industrialized European markets (e.g. Afonso and Sousa, 2011; Agnello and Sousa,
2014, and references cited therein). Various channels exist whereby ?scal policy can
affect stock and housing markets (Afonso and Sousa, 2011; Agnello and Sousa, 2014).
For instance, ?scal policy can in?uence stock markets via its effect on sovereign risk
spreads. These spreads, in turn, re?ect the ?nancing capacity of government as well as
investor expectations. When the markets deem that ?scal policy is stable, then an
in?ow of capital causes the exchange rate to appreciate and subsequently to reduce
pressures on central bank authorities to raise interest rates. Since demand for
government bonds strengthen, the overall bond yield curve falls, which affects the
stock market. Increasing public de?cits through the government’s wage bill, however,
can lead to a deteriorating lending environment, as this could lead to an increase in the
demand for credit that pushes interest rates higher. Consequently, the present
discounted value of the cash-?ows generated by stocks falls, the markets require a
higher risk premium, and stock prices shrink. Finally, unsound ?scal policies can
prompt a loss in the con?dence of home-currency assets and generate a rebalancing of
asset portfolio composition away from domestic assets toward foreign assets.
Fiscal policy can also affect housing markets. For example, taxes on housing capital
gains and the imputed rental housing value, ?scal subsidies and value added taxes
(VAT) on purchases of new homes, and the tax deductibility of mortgage payments and
housing rents can affect house prices via their effects on households’ disposable income
and the demand of housing. An indirect effect of ?scal spending through the wage bill
and government infrastructure spending can lead to both increases and decreases in the
demand for housing. More broadly, the deterioration of the ?scal stance and uncertainty
about the long-run sustainability of public ?nances can affect long-term interest rates
and negatively impinge on the ?nancing conditions for mortgages, pushing house prices
downwards. Hence, we should not neglect the role of ?scal policy in explaining both
housing market developments and stock market dynamics.
As discussed above, changes in stock or house prices can in?uence consumption.
However, it is the variation in the ?nancial and housing wealth that can produce
substantial variation in personal savings. In a Keynesian setup, when the corporate
sector does not compensate the change in household savings, it is then left for
the government to allow for a variation in its own savings and, thereby, to smooth
the ?uctuations in national saving that originates from movements in asset prices.
There is, however, the other (more neoclassical) line of thinking which recommends
very little government intervention in the case of declining asset prices, as they argue
that it is intervention and regulation that caused cyclical ?uctuations in the ?rst place
(Andre et al., 2012; Aye et al., 2014b). Also, Blake et al. (1988) and Lossani and Tirelli
(1994) suggest that ?scal policy rules can be designed to steer national wealth to its
target value point to accommodate wealth expansions when the wealth level is below a
certain target value. Moreover, a tax increase may reduce the incentive to accumulate
JFEP
6,1
48
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
wealth since it reduces the incentive to earn income and increases the incentive to
consume, which will have negative repercussions for the whole economy. Under such
assumptions, time-varying models that account for the “state” of the economy and the
“state” of asset values are useful to disentangle the relationship between ?scal policy
and wealth dynamics.
In terms of the effects of asset price shocks on ?scal policy, Schuknecht and
Eschenbach (2002) conduct an empirical study of changes in real estate and asset price
on the ?scal balance across 17 OECD countries (OECD, 2012). The paper ?nds that asset
prices affect ?scal balances through the revenue channel; capital gains, turnover related
taxes as well as wealth effects and their impact on consumption are found to have an
impact on the ?scal balance. The study ?nds that, on average, a 10 per cent change in real
estate and stock prices have a similar effect on the ?scal balance as a 1 per cent change
in output. Tagkalakis (2011) augments a ?scal policy reaction function with ?nancial
variables for OECD countries. Looking at the impact on the ?scal balance, current
expenditure and current revenue, the author ?nds that an increase in asset prices has
a positive impact on the ?scal balance. Furthermore, the paper ?nds that residential price
changes play a bigger role in their effect on the budget balance, compared to commercial
property price and equity price changes. Agnello et al. (2012a, b) look at the impact of
asset market developments on ?scal policy. Employing both linear and non-linear
speci?cations, the study estimates a ?scal policy rule that includes ?nancial, as well as
housing wealth. The authors ?nd that in the linear speci?cation, spending does not react
to asset prices, but taxes and the primary surplus fall in reaction to a rise in stock prices,
and rise when house prices increase. Declining asset prices are also associated with
declining revenue collections, especially where capital and dividend taxes apply. Any
ramp up in listed company pro?ts will result in a higher dividends tax while increasing
house prices will increase revenue collected from capital gains.
A few studies (Du Plessis et al., 2007, 2008; Jooste et al., 2012) employ structural
VARs and vector error-correction (VEC) models, time-varying VARs, and dynamic
stochastic general equilibrium (DSGE) models to analyse simultaneously the effects of
business cycle, monetary policy, and ?scal policy shocks on output, consumption,
in?ation, and interest rates in South Africa. To the best of our knowledge, Aye et al.
(2014b) is the ?rst study to analyse simultaneously the effects of these shocks on South
African asset prices. That said, the literature on the effect of monetary policy on asset
prices in South Africa includes numerous studies. A number of those studies examine
the effects of monetary policy on equity prices (returns) in South Africa (Smal and de
Jager, 2001; Coetzee, 2002; Prinsloo, 2002; Durham, 2003; Hewson and Bonga-Bonga,
2005; Alam and Uddin, 2009; Chinzara, 2010; Mallick and Sousa, 2011; Mangani, 2011;
Muroyiwa, 2011), mainly based on (structural) VAR models and, at times, panel
data approaches that include South Africa. On the other hand, we know of only four
studies – Kasai and Gupta (2010), Gupta et al. (2010), Ncube and Ndou (2011) and
Simo-Kengne et al. (2013b) – that analyse the role played by the housing market in the
monetary policy transmission mechanism, using the effect of monetary policy shocks
on house prices in structural, factor-augmented, and Markov-switching VAR models.
These studies generally show that contractionary monetary policy leads to lower stock
and house prices.
This paper builds on the work of Aye et al. (2014b) that uses a sign-restriction
approach to capture the effects of ?scal policy shocks on asset prices. The theory-based
Time-varying
approach
49
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sign-restriction method allows this paper to identify shocks, such as tax
announcements and anticipation effects, on the macro economy. Aye et al. (2014b)
separate their results into expected and unexpected ?scal policy changes which
overcome problems of correctly identifying the shocks. The study shows that an
unanticipated and anticipated government revenue shock leads both house prices and
stock prices to decline. Anticipated government revenue shocks impact stock prices
more negatively and has a more persistent impact. The impact of an unanticipated
government spending shock has hardly any effect on house prices. Stock prices
respond positively. However, anticipated government spending shocks increase house
prices, but reduce equity prices. Our paper will benchmark the results against Aye et al.
(2014b) for South Africa. Barring the recent related paper by Agnello et al. (2012b),
which uses time varying transition probabilities in a two-state Markov-switching
framework to analyse the response of ?scal variables to asset prices for the US
economy[3], our paper is the ?rst in the literature of ?scal policy and asset prices to use
a time-varying VAR model to analyse not only the above relationship, but also the
effect of ?scal policy on asset prices. Note that, unlike the above approach of
Agnello et al. (2012b), the time-varying VAR approach allows us to treat each point in
time as a speci?c regime (rather than just assuming speci?c number of regimes) with
smooth transition across regimes, and also allow for stochastic volatility, ignoring
which, leads to biased estimates (see Section 3 for further details).
The rest of the paper unfolds as follows: Section 3 describes the empirical
methodology while Section 4 describes the data transformations and empirical results.
Finally, Section 5 concludes.
3. Empirical methodology
A vector autoregression (VAR), proposed by Sims (1980), has become a popular
technique used in econometric analysis and is adaptable to a vast array of economic
settings (Baltagi, 2011). In this study, a TVP-VAR model with stochastic volatility is
used. The TVP-VAR is common in the analysis of macroeconomic issues and allows us
to capture the time-varying nature of the underlying structure in the economy in
a ?exible and robust manner (Nakajima, 2011). The parameters in the VARspeci?cation
are assumed to follow a ?rst order random walk process, thereby incorporating both
temporary and permanent changes to the parameters. The inclusion of stochastic
volatility is an important aspect in this TVP-VAR model. In many situations,
a data-generating process of economic variables seems to have drifting coef?cients and
shocks of stochastic volatility. In that case, the application of a time-varying parameter
model but with constant volatility may result in biased estimations of the time-varying
coef?cients, since a possible variation of the volatility in disturbances is ignored. The
TVP-VAR model with stochastic volatility avoids this misspeci?cation and re?ects
simultaneous relations among variables of the model and heteroscedasticity of the
innovations (Primiceri, 2005)[4]. Although stochastic volatility makes the estimation
dif?cult due to the intractability of the likelihood function, the model can be estimated
using Markov Chain Monte Carlo (MCMC) methods in the context of a Bayesian
inference. Measuring the responses over time lends insight into the timing aspect of
government shocks on asset prices and analyses periods in which shocks were most
signi?cant. This variation over time can for example help to explain how ?scal shocks
relate to assets during booms and busts, and more recently, the ?nancial crisis.
JFEP
6,1
50
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Following Nakajima (2011), this paper estimates a time-varying parameter VAR
model with stochastic volatility of the form:
y
t
¼ c
t
þ B
1t
y
t21
þ · · · þ B
st
y
t2s
þ e
t
; e
t
, Nð0; V
t
Þ; ð1Þ
for t ¼ s þ 1, . . . , n, where y
t
is a (k £ 1) vector of observed variables, B
1t
, . . . , B
st
are
(k £ k) matrices of time-varying coef?cients, and V
t
is a (k £ k) time-varying
covariance matrix. A recursive identi?cation scheme is assumed by the decomposition
of V
t
¼ A
21
t
S
t
S
t
A
0
21
t
, where A
t
is a lower-triangle matrix with diagonal elements
equal to one, and S
t
¼ diagðs
1t
; . . . ; s
kt
Þ. Let us de?ne b
t
as the stacked row vector of
B
1t
, . . . , B
st
; a
t
is the stacked row vector of the free lower-triangular elements of A
t
; and
h
t
¼ ðh
1t
; . . . ; h
kt
Þ where h
jt
¼ log s
2
jt
. The time-varying parameters are assumed to
follow a random walk process:
b
tþ1
¼ b
t
þy
bt
;
a
tþ1
¼ a
t
þy
at
;
h
tþ1
¼ h
t
þy
ht
;
1
t
y
bt
y
at
y
ht
_
_
_
_
_
_
_
_
_
_
_
_
_
_
, N 0;
I 0 0 0
0 S
b
0 0
0 0 S
a
0
0 0 0 S
h
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
;
for t ¼ s þ 1, . . . , n, with e
t
¼ A
21
t
S
t
1
t
where S
a
and S
h
are diagonal, b
sþ1
,
Nðm
bo
; S
bo
Þ; a
sþ1
, Nðm
ao
; S
ao
Þ; and h
sþ1
, Nðm
ho
; S
ho
Þ[5]. A Bayesian inference is
used to estimate the TVP-VAR models via MCMC methods. The goal of MCMC
methods is to assess the joint posterior distributions of the parameters of interest
under certain prior probability densities that are set in advance. We assume the
following priors, as in Nakajima (2011): S
b
, IWð25; 0:01I Þ, ðS
a
Þ
22
i
, Gð4; 0:02Þ;
ðS
h
Þ
22
i
, Gð4; 0:02Þ; where ðS
a
Þ
22
i
and ðS
h
Þ
22
i
are the ith diagonal elements in S
a
and
S
h
, respectively. IW and G denotes the inverse Wishart and the gamma distributions,
respectively. For the initial set of the time-varying parameter, ?at priors are set such
that: m
bo
¼ m
ao
¼ m
ho
¼ 0 and S
bo
¼ S
ao
¼ S
ho
¼ 10 £ I :
3.1 Data description
Three variables are used in the analysis, with the sample period covering
1966:Q1-2012:Q2. We source the government’s budget balance data from the
South African Reserve Bank (where government revenue is subtracted from
government expenditure and expressed as a percent of GDP), Bloomberg for the All
Share Index and Amalgamated Bank of South Africa for the house price index. Both the
asset prices are expressed in real terms by de?ating the respective nominal series by the
CPI index. Note that all the variables were obtained in their seasonally adjusted form.
Log values of real house and stock prices are differenced to induce stationarity, and are
also standardised so that we can compare the magnitude of effect of ?scal policy across
the two asset prices, and also, the differences in the size of the feedback of the two asset
prices on ?scal policy behaviour. We use house, BB and JSE as short hand for the house
price index, the budget balance and the Johannesburg Stock Exchange All Share Index.
The variables pass the usual unit root tests namely, an augmented Dickey and Fuller
(ADF) (1981), Phillips and Perron (1988), Dickey-Fuller test with generalized least
Time-varying
approach
51
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
squares detrending (DF-GLS), the Kwiatkowski et al. (1992) test; the Elliot, Rothenberg,
and Stock (ERS) (1996) point optimal test, the Ng and Perron (2001) modi?ed versions of
the PP (NP-MZt) test and the ERS point optimal (NP-MPT) test. The stable[6] TVP-VAR
is estimated based on two lags, as was unanimously suggested by all the popular
lag-length tests, namely, the sequential modi?ed LR test statistic, the Akaike
information criterion, the Schwarz information criterion, applied to a constant parameter
VAR[7]. Accounting for stationarity and lags, our effective sample period start from
1966:4.
4. Empirical results
Anecdotal evidence shows that prior to the ?nancial crisis of 2008/2009, house prices
on average increased by 19 percent (2000-2007). The government recorded budget
surpluses in 2006/2007 and 2007/2008 and JSE All Share Index was growing at high
rates. The ?nancial crisis had a signi?cant impact on house prices and was
accompanied with decline in house prices by 0.2 per cent in 2009. The government, in
response to declining aggregate demand, increased government expenditure (together
with automatic effects of tax revenue decline) and as a consequence widened the
budget de?cit by 2.1 percent of GDP in 2008/2009 and 5.6 percent of GDP in 2009/2010.
While it is dif?cult to infer a direct relationship between government expenditure and
house prices, a simple multiplier (calculated as the ratio of the percentage change in
house prices to the percentage change in government expenditure) show that the
pre-crisis multiplier was stable at around 1.8 while during the crisis the multiplier
declined to 0.2 providing the ?rst indication of a nonlinear relationship.
The posterior estimates fromthe TVP-VARwere obtained after 10,000 samples were
drawn, with the ?rst 1,000 draws discarded. These posterior estimates for the means,
along with those for the standard deviations, the 95 per cent credibility intervals[8], the
convergence diagnostic (CD) due to Geweke (1992) and the inef?ciency factors
are presented in Table I[9]. The 95 per cent credibility intervals include the estimates for
the posterior means, and the CD statistics do not allow us to reject a null hypothesis of
convergence to the posterior distribution at a signi?cance level of 5 per cent. In general,
the inef?ciency factors are relatively low. We can thus conclude that the MCMC
algorithm is an ef?cient method of producing the posterior draws. Figure 1 presents the
estimation results of the TVP-VAR model with stochastic volatility.
Figure 2 plots the posterior estimates of stochastic volatility for each of the
variables used in the TVP-VAR. The estimates for the stochastic volatility shock hint
to at least two regimes; one characterised by a pre-1990 era and one of a post-1990 era.
Parameter Mean SD 95% intervals CD Inef.
(S
b
)
1
0.0374 0.0094 (0.0240,0.0596) 0.134 48.31
(S
b
)
2
0.0271 0.0049 (0.0192,0.0383) 0.237 30.12
(S
a
)
1
0.0538 0.0142 (0.0336,0.0911) 0.099 52.26
(S
a
)
2
0.0499 0.0114 (0.0329,0.0779) 0.938 26.80
(S
h
)
1
0.0826 0.0389 (0.0431,0.1886) 0.142 183.30
(S
h
)
2
0.0800 0.0342 (0.0398,0.1725) 0.862 156.84
Note: The estimates of S
b
and S
a
are multiplied by 100
Table I.
Selected estimation
results
JFEP
6,1
52
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 1.
Moments and posterior
distributions
N
o
t
e
s
:
S
a
m
p
l
e
a
u
t
o
c
o
r
r
e
l
a
t
i
o
n
s
(
t
o
p
p
a
n
e
l
)
,
s
a
m
p
l
e
p
a
t
h
s
(
m
i
d
d
l
e
p
a
n
e
l
)
,
a
n
d
p
o
s
t
e
r
i
o
r
d
e
n
s
i
t
i
e
s
(
b
o
t
t
o
m
p
a
n
e
l
)
;
t
h
e
e
s
t
i
m
a
t
e
s
o
f
?
b
a
n
d
?
a
a
r
e
m
u
l
t
i
p
l
i
e
d
b
y
1
0
0
Time-varying
approach
53
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Coincidentally this shift overlaps with South Africa’s democratic transition with the
release of Nelson Mandela. The non-steady stochastic volatility justi?es the use of a
TVP-VAR model to capture possible regime and period speci?c changes.
Impulse responses are used as a tool to capture the macroeconomic dynamics in the
estimated VAR system. For a standard constant parameter VAR model, the impulse
responses are drawn for each set of two variables, whereas for a TVP-VAR model,
the impulse responses can be drawn in an additional dimension, as the responses are
computed at all points in time using the time-varying parameters. There are several
ways to simulate the impulse responses based on the parameter estimates of the
TVP-VAR model. Following Nakajima (2011), we compute the impulse responses by
?xing an initial shock size equal to the time-series average of stochastic volatility over
the sample period, and using the simultaneous relations at each point in time, for
considering the comparability over time. We identify the three structural shocks (house
demand, ?scal policy and stock demand (portfolio)) using a recursive or Cholesky
identi?cation scheme, as obtained based on the lower-triangular matrix A
t
. We order
the variables as follows: house, BB and JSE following Agnello and Sousa (2014). The
ordering of the two asset prices relative to the ?scal policy instrument is quite intuitive:
the stock price is ordered last as it refers to assets that are traded in markets where
auctions take place instantaneously. While, the house price was ordered ?rst in the
Figure 2.
Posterior estimates
for the stochastic
volatility of the
structural shock
Notes: Top panel presents the data values; bottom panel depicts the posterior mean estimates
(solid line) and 95 percent credible intervals (dotted lines) for stochastic volatility of a
structural shock
JFEP
6,1
54
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
system to account for the fact that housing markets are inherently sticky and so house
prices do not immediately reach the equilibrium after a shock. Then there is a
“time-to-build” argument suggesting that it takes time for developers to bring new
houses to the market or to work out of inventories when demand increases. Further,
the matching between buyers and sellers requires time, and ?nally, one needs to also
account for important transaction costs inherent to trading housing up or down.
To compute the recursive innovation of the variable, the estimated time-varying
coef?cients are used from the current date to future periods. Around the end of the
sample period, the coef?cients are set constant in future periods for convenience.
Although a time series of impulse responses for selected horizons or impulse responses
for selected periods are often exhibited in the literature, one could draw a
three-dimensional plot for the time-varying impulse responses.
Figure 3 plots the mean impulse response function of asset prices in reaction to a
shock in ?scal policy. Contemporaneous ?scal policy shocks are analysed over
different horizons, over time and in terms of magnitude. House prices respond with a
lag due to the VAR ordering. This response is mainly positive following a ?scal policy
shock. As discussed earlier, negative tax shocks have wealth effects that could lead to
a higher demand of assets. This can also occur with a rise in government spending,
especially when spending is concentrated around wage increases. Its amplitude varies
over time with the most signi?cant impact being during the ?nancial crisis in 2010
(with a multiplier of 0.4). The impulse responses of house prices peak four quarters
ahead before dissipating. Apart from the late 1980s and the shock in 2009/2010,
Figure 3.
Impulse response function
of ?scal policy following a
shock to real house prices
0.4
0.2
0
–0.2
2010
2000
1990
1980
1970
2
4
6
8
10
12
1.5
1
0.5
0
–0.5
2010
2000
1990
1980
1970
2
4
6
8
10
12
M
a
g
n
i
t
u
d
e
M
a
g
n
i
t
u
d
e
Y
e
a
r
I
R
F
h
o
r
i
z
o
n
Y
e
a
r
I
R
F
h
o
r
i
z
o
n
House response to BB JSE response to BB
Time-varying
approach
55
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
house price responses were relatively short lived (six quarters). In comparison, equity
price responses are rather mute following a budget shock. In terms of size, the initial
impact is largest (the multiplier is not signi?cantly bigger than 0.5 over towards the
end of the sample period). Increases are quickly met with a decline in equity prices
which would suggest that markets are quick to adjust to any ?scal news.
Accumulating over the impulse horizon reveals that ?scal shocks have a negligible
impact on stock prices. It is also interesting to note that the contemporaneous impact of
?scal shocks on equity prices tapered down since the 1970s with an initial impact of
close to 1. From Figure 3 it becomes clear that house prices respond more starkly
compared to equity prices following a ?scal shock. The shock also lasts longer on
house prices. Quite surprising is that house prices responded more than equity prices
during the 2008/2009 ?nancial crisis. One hypothesis could be that the ?nancial crisis
represented a liquidity and solvency problem more relevant for asset classes such as
houses than investments into the equity market.
This study also looks at the impact of shocks in stock and house prices on the
government’s budget balance. A priori, one would expect that an increase in asset prices
will lead to higher tax revenues through property and capital gains taxes which should
lead to a contracting budget de?cit. The channels through which asset price shocks
effect the budget balance is rather complex and requires a detailed decomposition of
expectations, output and interest rates. For one, equity price increases could exert
upward pressure on government bond interest rates if there is a substitution away from
bonds. This in turn will lead to a rising de?cit. After a period government would collect
revenue from dividend pay outs which should reduce the de?cit. In essence the shock of
asset prices to the balance relies on the net effect revenue gains minus the effect of
a potential increase in debt service costs. Figure 4 show that higher house prices had
a negative and signi?cant effect on the budget de?cit (de?cit reducing) throughout our
sample and at various horizons. Although the contemporaneous impact seems constant,
the impact was largest during the ?nancial crisis. House price shocks also lasts for
almost two years (in 2008-2012) compared to a relatively short-lived outcome in the 1970s
and 1980s. One of the reasons why house price shocks had a larger impact on the balance
is due to modernisation processes of municipalities and tax collecting authorities which
made collecting taxes more ef?cient. This part of our sample was also coincidentally
associated with South Africa’s, and for that matter the majority of the Western world,
housing boom period. House price shocks during this period are slightly more persistent
and have a bigger impact on the budget de?cit. However, during the ?nancial crisis
house price shocks had a smaller impact of the de?cit.
Finally, the right hand side of Figure 4 shows the propagation of stock price shocks
to the budget balance. The impact of stock price shocks on the budget de?cit varies
between positive and negative during different horizons. The period during the
?nancial crisis highlights that a rise in equity prices increased the de?cit. As shown by
Aye et al. (2013b) asset price shocks lead to an increase in interest rates. The interest
rate channel could be used to motivate how equity price shocks could lead to an
increase in the budget de?cit, especially if it causes a substitution away from bonds to
equities. That being said, stock price shocks have only a transitory impact on the
budget as the shocks dissipate already after six to eight quarters[10].
These results are in line with Aye et al. (2014b): expansionary ?scal shocks lead to
an increase in asset prices, but are only transitory. As mentioned earlier, it could be
JFEP
6,1
56
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that markets are quick to adjust when hit by shocks. Usually expansionary ?scal
policy ?nds its way in the public discourse which would allow markets to anticipate
?scal shocks more reliably. This would only strengthen an argument for the short lived
response of ?scal shocks. The results show that on average, an expansionary ?scal
policy shock has a small impact on house prices (as was the case in Aye et al. (2014b)
given a spending shock). The impact, however, became more pronounced at the onset
of the ?nancial crisis which would suggest that effects are ampli?ed under distressed
economic conditions. Furthermore, looking at the impact of ?scal shocks and asset
prices could provide a channel through which an explanation can be given for private
investment being crowded out when ?scal policy expands. However, this requires a
more detailed analysis that is beyond the scope of this study.
On the other hand, asset price shocks represent a possible increase in revenue
collection which should effectively reduce budget de?cits. The results in this paper
con?rm that asset price shocks reduced the de?cits. With new modernisation
programmes from the revenue collecting authorities and a bigger emphasis and tax
broadening, asset price shocks have had a larger impact on the budget post-2000.
5. Conclusion
This paper uses a three variable (stock prices, house prices and government’s budget
balance) TVP-VAR with stochastic volatility to study the simultaneous impact of ?scal
shocks on asset prices, and asset price shocks on ?scal policy. We ?nd that ?scal
shocks had a small impact on asset prices which is in line with Aye et al. (2014b).
Figure 4.
Impulse response function
of ?scal policy following a
shock to real stock prices
BB response to House BB response to JSE
0
–0.1
–0.2
–0.3
–0.4
–0.5
–0.6
2010
2000
1990
1980
1970
2
4
6
8
10
12
0.5
–0.5
–1
0
–1.5
2010
2000
1990
1980
1970
2
4
6
8
10
12
M
a
g
n
i
t
u
d
e
M
a
g
n
i
t
u
d
e
Y
e
a
r
I
R
F
h
o
r
i
z
o
n
Y
e
a
r
I
R
F
h
o
r
i
z
o
n
Time-varying
approach
57
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The results show that ?scal policy and asset price shocks have varying impacts over
time. Furthermore, the results show that extreme economic events such as the recent
?nancial crisis change the impact of these shocks. The recent ?nancial crisis shows
that ?scal shocks amplify the effects on house prices while the effects on equity prices
become more subdued. Information ?ows are a lot more transparent and ef?cient in
equity markets while house price stickiness could explain the different responses
between the asset classes. At times when the budget balance seemed unsustainable,
a consolidating budget balances seem to have a positive impact on asset prices while
increasing asset prices reduces de?cits as tax collections improve. However, as shown
in the study of Aye et al. (2014b) increasing taxes will limit the amount of revenue
collected as it reduces real asset prices. This suggests that consolidation effects, when
considering the impact on asset prices, should happen through spending channels. The
policy implication is that expansionary government decisions have clear wealth
effects, but this depends on speed at which information can be absorbed. In the case of
equity markets, it would seem that information is relatively quickly absorbed and thus
the impact on equity markets would depend on their views on the sustainability of
?scal balances and the impact of spending and tax decisions.
Notes
1. For detailed international literature reviews on studies involving monetary policy and asset
prices, see Bjørnland and Leitemo (2009), Iglesias and Haughton (2011), Gupta et al. (2012a, b)
and Bjørnland and Jacobsen (2014).
2. Tobin’s q equals the ratio of the stock market value of a ?rm to the replacement cost of its
capital.
3. This paper tested for nonlinear effects of asset prices on the US ?scal policy. By modeling
government spending and taxes as time-varying transition probability Markovian
processes, the authors found that taxes signi?cantly adjust in a nonlinear fashion to asset
prices. In particular, taxes respond to housing and (to a smaller extent) to stock prices
changes during normal times. However, at periods characterized by high ?nancial volatility,
government taxation only counteracts stock market developments (and not the dynamics of
the housing sector). On the other hand government spending, is neutral vis-a` -vis the asset
market cycles.
4. We also estimated a TVP-VAR model without stochastic volatility. Though, the qualitative
behaviour of the impulse responses, in general, is very similar, the impulse responses are
very volatile. Also, quantitatively, the effects are larger as well. This is not surprising, since
we do not allow for heteroscedastic disturbances, the parameters are not only estimated with
less precision, but also tends to be biased upwards to in?ate the multipliers (Nakajima, 2011).
The details of these results are available upon request from the authors.
5. For a comprehensive analysis of the TVP-VAR methodology and the estimation algorithm
(Nakajima, 2011).
6. The constant parameter VAR is found to be stable as all roots were found to lie within the
unit circle.
7. Complete details of the unit root, stability and lag length tests are available from the authors
upon request.
8. Credibility intervals are used in the Bayesian paradigm as opposed to “con?dence” intervals
which belong in the frequentist realm.
JFEP
6,1
58
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
9. Geweke (1992) suggests the comparison between the ?rst n
0
draws and the last n
1
draws,
dropping out the middle draws, to check for convergence in the Markov chain. The CD
statistics are computed as follows:
CD ¼ ð x
0
2 x
1
Þ=
?????????????????
^ s
2
0
n
0
þ
^ s
2
1
n
1
¸
where:
x
j
¼
1
n
j
_ _
mjþnj21
i¼mj
x
ði Þ
where x
j
being the ith draw and ^ s
2
j
=n
j
_ _
is the standard error of x
j
, respectively, for j ¼ 0, 1.
If the sequence of the MCMC sampling is stationary, it converges to a standard normal
distribution. We set m
0
¼ 1, n
0
¼ 1,000, n
1
¼ 5,001 and n
2
¼ 5,000. ^ s
2
j
is computed using a
Parzen window with bandwidth (B
m
) ¼ 500. The inef?ciency parameter is de?ned as:
1 þ 2
B
m
i¼1
r
i
;
where r
i
is the sample autocorrelation at lag s, which is computed to measure how well the
MCMC chain mixes.
10. We also estimated a constant parameter VAR and analysed the effect of a ?scal shock on
house prices and stock prices, as well as, the response of ?scal policy to shocks on the asset
prices. Realizing that the shape of the impulse response in the constant VAR model is
associated with the average level of the response in the TVP-VAR model to some extent, we
basically obtain the general conclusions of the TVP-VAR model. The only exception being
the negative, but insigni?cant, impact of the budget balance on house prices for the ?rst two
quarters. Again, the possible biasedness in the parameter estimates for not allowing
time-variation, as well as, stochastic volatility in the error structure could be a reason behind
such contradictory impulse behaviour (Nakajima, 2011). The details of these results are
available upon request from the authors.
References
Afonso, A. and Sousa, R.M. (2011), “What are the effects of ?scal policy on asset markets?”,
Economic Modelling, Vol. 28, pp. 1871-1890.
Afonso, A. and Sousa, R.M. (2012), “The macroeconomic effects of ?scal policy”, Applied
Economics, Vol. 44 No. 34, pp. 4439-4454.
Agnello, L. and Sousa, R.M. (2014), “Fiscal policy and asset prices”, Bulletin of Economic
Research (in press).
Agnello, L., Castro, V. and Sousa, R.M. (2012a), “How does ?scal policy react to wealth
composition and asset prices?”, Journal of Macroeconomics, Vol. 34 No. 3, pp. 874-890.
Agnello, L., Dufrenot, G. and Sousa, R.M. (2012b), “Adjusting the US ?scal policy for asset prices:
evidence from a TVP-MS framework”, NIPE Working Papers, NIPE, Universidade do
Minho, Braga.
Alam, M.M. and Uddin, M.G. (2009), “Relationship between interest rate and stock price:
empirical evidence from developed and developing countries”, International Journal of
Business and Management, Vol. 4 No. 3, pp. 43-51.
Time-varying
approach
59
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Andre, C., Gupta, R. and Kanda, P.T. (2012), “Do house prices impact consumption and interest
rate? Evidence from OECD countries using an agnostic identi?cation procedure”, Applied
Economics Quaterly, Vol. 58 No. 1, pp. 19-70.
Apergis, N., Simo-Kengne, B.D. and Gupta, R. (2013), “The long-run relationship between
consumption, house prices and stock prices in South Africa: evidence from provincial-level
data”, Department of Economic Working Paper No. 2013/26, University of Pretoria,
Pretoria.
Aron, J. and Muellbauer, J. (2013), “Wealth, credit conditions and consumption: evidence from
South Africa”, Review of Income and Wealth, Vol. 59, Supplement S1, pp. S161-S196.
Aye, G.C., Balcilar, M. and Gupta, R. (2013a), “Long- and short-run relationships between house
and stock prices in South Africa: a nonparametric approach”, Journal of Housing Research,
Vol. 22 No. 2, pp. 203-219.
Aye, G.C., Gupta, R. and Modise, P. (2014a), “Do stock prices impact consumption and interest
rates in South Africa? Evidence from a time varying vector autoregressive model”, Journal
of Emerging Market Finance (in press).
Aye, G.C., Gupta, R., Kaninda, A., Nyakabawo, W. and Razak, A. (2013b), “House price, stock
price and consumption in South Africa? A structural VAR approach”, Corporate
Ownership and Control, Vol. 10 No. 2, pp. 585-589.
Aye, G.C., Balcilar, M., Gupta, R., Jooste, C., Miller, S. and Ozdemir, Z. (2014b), “Fiscal policy
shocks and the dynamic asset prices: the South African experience”, Public Finance Review
(in press).
Baltagi, B.H. (2011), Econometrics, Springer, Berlin.
Bjørnland, H.C. and Jacobsen, D.H. (2014), “House prices and stock prices: different roles in the
US monetary transmission mechanism”, Scandinavian Journal of Economics (in press).
Bjørnland, H.C. and Leitemo, K. (2009), “Identifying the interdependence between US monetary
policy and the stock market”, Journal of Monetary Economics, Vol. 56 No. 2, pp. 275-282.
Blake, A., Vines, D. and Weale, M. (1988), “Wealth targets, exchange rate targets and
macroeconomic policy”, Discussion Paper No. 246.
Castro, V. (2011), “Can central banks’ monetary policy be described by a linear (augmented)
Taylor rule or by a nonlinear rule?”, Journal of Financial Stability, Vol. 7 No. 4, pp. 228-246.
Chinzara, Z. (2010), “Macroeconomic uncertainty and emerging stock market volatility: the case
for South Africa”, Economic Research Southern Africa Working Paper No. 187.
Coetzee, C.E. (2002), “Monetary conditions and stock returns: a South African case study”,
EconWPA, 0205002.
Das, S., Gupta, R. and Kanda, P.T. (2011), “Bubbles in South African house prices and their
impact on consumption”, Journal of Real Estate Literature, Vol. 19 No. 1, pp. 71-92.
Dickey, D. and Fuller, W. (1981), “Likelihood ratio statistics for autoregressive time series with
a unit root”, Econometrica, Vol. 49, pp. 1057-1072.
Du Plessis, S., Smit, B. and Sturzenegger, F. (2007), “The cyclicality of monetary and ?scal policy
in South Africa since 1994”, South African Journal of Economics, Vol. 75 No. 3, pp. 391-411.
Du Plessis, S., Smit, B. and Sturzenegger, F. (2008), “Identifying aggregate supply and demand
shocks in South Africa”, Journal of African Economies, Vol. 17 No. 5, pp. 765-793.
Durham, J.B. (2003), “Monetary policy and stock prices returns”, Financial Analysts Journal,
Vol. 59 No. 4, pp. 26-35.
Elliot, G., Rothenberg, T.J. and Stock, J.H. (1996), “Ef?cient tests for an autoregressive unit root”,
Econometrica, Vol. 64, pp. 813-836.
JFEP
6,1
60
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
European Central Bank (2010), “Asset price bubbles and monetary policy revisited”, ECB
Monthly Bulletin, November, pp. 71-83.
Feldstein, M. (2009), “Rethinking the role of ?scal policy”, Contributed paper at the January 2009
Meeting of the American Economic Association.
Geweke, J. (1992), “Evaluating the accuracy of sampling-based approaches to the calculation of
posterior moments”, in Bernado, J.M., Berger, J.O., Dawid, A.P. and Smith, A.F.M. (Eds),
Bayesian Statistics 4, Oxford University Press, New York, NY, pp. 169-188.
Granville, B. and Mallick, S.K. (2009), “Monetary and ?nancial stability in the Euro area:
procyclicality versus trade-off”, Journal of International Financial Markets, Institutions
and Money, Vol. 19, pp. 662-674.
Gupta, R., Jurgilas, M. and Kabundi, A. (2010), “The effect of monetary policy on real house price
growth in South Africa: a factor-augmented vector autoregression (FAVAR) approach”,
Economic Modelling, Vol. 27 No. 1, pp. 315-323.
Gupta, R., Jurgilas, M., Miller, S.M. and van Wyk, D. (2012a), “Financial market liberalization,
monetary policy, and housing sector dynamics”, International Business and Economics
Research Journal, Vol. 11 No. 1, pp. 69-82.
Gupta, R., Kabundi, A., Jurgilas, M. and Miller, S.M. (2012b), “Monetary policy and housing
sector dynamics in a large-scale Bayesian vector autoregressive model”, International
Journal of Strategic Property Management, Vol. 16 No. 1, pp. 1-20.
Hewson, M. and Bonga-Bonga, L. (2005), “The effects of monetary policy shocks on stock returns
in South Africa: a structural vector error correction model”, paper presented at the
Economic Society of South Africa Conference, Durban, South Africa.
Iacoviello, M. (2010), “Housing in DSGE models: ?ndings and new directions”, in de Bandt, O.,
Knetsch, T., Pen˜ alosa, J. and Zollino, F. (Eds), Housing Markets in Europe:
A Macroeconomic Perspective, Springer, Berlin, pp. 3-16.
Iacoviello, M. (2011), “Housing wealth and consumption”, International Encyclopedia of Housing
and Home, Elsevier, Amsterdam.
Iglesias, E.M. and Haughton, A.Y. (2011), “Interaction between monetary policy and stock prices:
a comparison between the Caribbean and the US”, Contributed paper at the Conference on
the Economy (COTE) 2011, The University of the West Indies, St Augustine, Trinidad and
Tobago, October.
Jooste, C., Liu, G. and Naraidoo, R. (2012), “Analysing the effects of ?scal policy shocks in the
South African economy”, Working Paper Series No. WP 2012-06, University of Pretoria,
Pretoria.
Kasai, N. and Gupta, R. (2010), “Financial liberalisation and the effectiveness of monetary policy
on house prices in South Africa”, ICFAI Journal of Monetary Economics, Vol. 8 No. 4,
pp. 59-74.
Kwiatkowski, D., Phillips, P., Schmidt, P. and Shin, J. (1992), “Testing the null hypothesis
of stationarity against the alternative of a unit root”, Journal of Econometrics, Vol. 54,
pp. 159-178.
Lossani, M. and Tirelli, P. (1994), “Simple rules, ?scal policy and wealth targets in a monetary
union: an alternative to the Maastricht guidelines”, Scottish Journal of Political Economy,
Vol. 41 No. 4, pp. 372-393.
Mallick, S.K. and Mohsin, M. (2007), “Monetary policy in high in?ation open economies: evidence
fromIsrael andTurkey”, International Journal of Finance and Economics, Vol. 12, pp. 405-415.
Mallick, S.K. and Mohsin, M. (2010), “On the real effects of in?ation in open economies: theory
and empirics”, Empirical Economics, Vol. 39, pp. 643-673.
Time-varying
approach
61
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Mallick, S.K. and Sousa, M.R. (2011), “In?ationary pressures and monetary policy: evidence
from BRICS economies”, paper presented at the Quantitative and Qualitative Analysis in
Social Sciences Conference, available at: www.qass.org.uk/2011-May_Brunel-conference/
Mallick.pdf
Mangani, R. (2011), “Monetary policy, structural breaks and JSE returns”, Investment Analysts
Journal, Vol. 73, pp. 27-35.
Mountford, A. and Uhlig, H. (2009), “What are the effects of ?scal policy shocks?”, Journal of
Applied Econometrics, Vol. 24, pp. 960-992.
Muroyiwa, B. (2011), “Identifying the interdependence between South Africa’s monetary policy
and the stock market”, unpublished MSc thesis, Department of Economics and Economic
History, Rhodes University, Grahamstown.
Nakajima, J. (2011), “Time-varying parameter VAR model with stochastic volatility: an overview
of methodology and empirical applications”, Monetary and Economic Studies, Vol. 29,
pp. 107-142.
Ncube, M. and Ndou, E. (2011), “Monetary policy transmission, house prices and consumer
spending in South Africa: an SVAR approach”, Working Paper No. 133, African
Development Bank Group, Abidjan.
Ng, S. and Perron, P. (2001), “Lag length selection and the construction of unit root tests with
good size and power”, Econometrica, Vol. 69, pp. 1519-1554.
OECD (2012), “Debt and macroeconomic stability”, WP(2012)22, OECD, Paris.
Peretti, V., Gupta, R. and Inglesi-Lotz, R. (2012), “Do house prices impact consumption and
interest rate in South Africa? Evidence from a time varying vector autoregressive model”,
Economics, Financial Markets and Management, Vol. 7 No. 4, pp. 101-120.
Phillips, P. and Perron, P. (1988), “Testing for a unit root in time series regression”, Biometrika,
Vol. 75, pp. 335-346.
Primiceri, G. (2005), “Time varying structural vector autoregressions and monetary policy”,
Review of Economic Studies, Vol. 72, pp. 821-852.
Prinsloo, J.W. (2002), “Household debt, wealth and savings”, Quarterly Bulletin, December,
pp. 63-78.
Ra?q, M.S. and Mallick, S.K. (2008), “The effect of monetary policy on output in EMU3: a sign
restriction approach”, Journal of Macroeconomics, Vol. 30, pp. 1756-1791.
Schuknecht, L. and Eschenbach, F. (2002), “Asset prices and ?scal balances”, ECB Working
Paper No. 141, European Central Bank, Frankfurt.
Schuknecht, L., von Hagen, J. and Wolswijk, G. (2009), “Government risk premiums in the bond
market: EMU and Canada”, European Journal of Political Economy, Vol. 25, pp. 371-384.
Simo-Kengne, B.D., Bittencourt, M. and Gupta, R. (2012), “House prices and economic growth in
South Africa: evidence from provincial-level data”, Journal of Real Estate Literature,
Vol. 20, pp. 97-117.
Simo-Kengne, B.D., Bittencourt, M. and Gupta, R. (2013a), “The impact of house prices on
consumption in South Africa: evidence from provincial-level panel VARs”, Housing
Studies, Vol. 28 No. 8, pp. 1133-1154.
Simo-Kengne, B.D., Balcilar, M., Gupta, R., Aye, G.C. and Reid, M. (2013b), “Is the relationship
between monetary policy and house prices asymmetric in South Africa? Evidence from
a Markov-switching vector autoregressive model”, Economic Modelling, Vol. 32 No. 1,
pp. 161-171.
Sims, C.A. (1980), “Macroeconomics and reality”, Econometrica, Vol. 48, pp. 1-48.
JFEP
6,1
62
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Smal, M.M. and de Jager, S. (2001), “The monetary policy transmission mechanism in
South Africa”, Occasional Paper No. 16, South Africa Reserve Bank, Pretoria.
Sousa, R.M. (2010a), “Consumption (dis)aggregate wealth, and asset returns”, Journal of
Empirical Finance, Vol. 17, pp. 606-622.
Sousa, R.M. (2010b), “Housing wealth, ?nancial wealth, money demand and policy rule: evidence
from the Euro area”, The North American Journal of Economics and Finance, Vol. 21,
pp. 88-105.
Sousa, R.M. (2010c), “The consumption-wealth ratio and asset returns: the Euro area, the UK and
the US”, Journal of Money, Investment and Banking, Vol. 16, pp. 12-29.
Tagkalakis, A. (2011), “Fiscal policy and ?nancial market movements”, Journal of Banking and
Finance, Vol. 35, pp. 231-251.
Venter, J.C. (2011), “Business cycles in South Africa during the period 2007-2009”, Quarterly
Bulletin, June, pp. 61-66.
Corresponding author
Rangan Gupta can be contacted at: [email protected]
Time-varying
approach
63
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
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_278999260.pdf