Transcript Academy of Economic Studies Doctoral School of Finance and

Academy of Economic Studies
Doctoral School of Finance and Banking
A COMPARATIVE PERSPECTIVE ON THE
SUSTAINABILITY OF FISCAL POLICY IN HUNGARY,
POLAND AND ROMANIA
MSc Student: Dorin Mantescu
Supervisor: Professor Moisă Altăr
Academy of Economic Studies
Doctoral School of Finance and Banking
• The recent preoccupation and reform of the Stability and
Growth Pact reiterated the need to measure in a
corresponding way the sustainability of the fiscal and
budget policies.
• Determining the extent to which fiscal policy is
sustainable proved highly contentious, controversial and
difficult.
• It is generally agreed that a fiscal policy stance is
sustainable if it satisfies the government’s intertemporal
budget constraint. In practice, this does not solve the
problem due to the fact that the government may
announce its intention and plan for offsetting current
deficits and debts by generating future surpluses.
Academy of Economic Studies
Doctoral School of Finance and Banking
• There are overall two main streams in the economic
literature regarding the assessment of the fiscal policy
sustainability, the prospective and retrospective
approaches.
• The main pillars for the prospective approach are
represented by
- fiscal gap indicator (proposed by Blanchard and others1990),
- stress testing by targeting primary surpluses typically
aiming at the reduction of the public debt and alternative
scenarios with two standard deviations shocks to interest
rates and growth
- value at risk which forecasts the deficit and the public
debt (Androgue, applied to Central American Countries).
- stochastic simulation models of debt accumulation.
Variance of debt increases with the forecast horizon,
giving rise to a “fan chart”.
Academy of Economic Studies
Doctoral School of Finance and Banking
• The retrospective approach is trying to answer about the
implications on the fiscal policy sustainability of the
maintenance of the historical policies into the future.
• Several well-known approaches.
- calculation of the fiscal gap proposed by Blanchard and
others (1990) which evaluates the historical primary
surplus against debt stabilizing benchmark.
- the second one is represented by the analysis of the
stationarity of the budget deficit proposed by Hamilton
and Flavin (1986), Wilcox (1989), Trehan and Walsh
(1990, 1991) and Hakkio and Rush (1991). Solvency is
guaranteed by the stationarity of the real deficit, but debt
may still rise.
- cointegration of the revenues and expenditures is
proposed by Bohn (1991), Hakkio and Rush (1991),
Ahmed and Rogers (1995) where the solvency is
guaranteed by the cointegration of primary expenditures,
revenues and debt.
Academy of Economic Studies
Doctoral School of Finance and Banking
- The last one is represented by the link between the
primary surplus and debt, where the solvency is
guaranteed by the positive relationship between primary
surplus and debt.
- I focus on the retrospective methodologies for assessing
the sustainability of the fiscal policy in order to avoid
additional uncertainty regarding the hypothesis regarding
governments behavior and the possible courses of policy
decision making.
Academy of Economic Studies
Doctoral School of Finance and Banking
• Fiscal sustainability analysis is based on the
government budget constraint.
• (1) Pt*Gt+(1+Rt)Bt-1=Bt+ΔMt+Pt*Tt
• Where Pt represents the general level of prices,
gt is the real government expenditure level,
including the transfers toward population, Rt is
the average interest rate paid to the debt
contracted at the end of t-1, Bt represents the
nominal value of the liabilities issued at the end
of the t period, Mt is the money supply provided
by the central bank at the beginning of the
period t, while Tt represents the real taxes level.
Academy of Economic Studies
Doctoral School of Finance and Banking
• The real budgetary constraint may be written in real
terms, as shares in GDP. Dividing the equation 1 by the
general level of prices (Pt) and the real national income
(yt) we obtain:
• (2) bt/yt=(1+ρt)*bt-1/yt-1+dt/yt
• where bt represents the real stock of public debt,
ρt=(1+Rt)/(1+Лt)(1+ψt) is the real interest rate adjusted
with the real GDP growth (ψt) while dt is the real primary
deficit, calculated as the total deficit minus the interest
payments associated to the public debt. The dynamic
equation of the public debt as percent of GDP is an
equation in discreet time, where is supposed that the
public debt has a one year maturity.
Academy of Economic Studies
Doctoral School of Finance and Banking
A)
•
•
•
•
•
•
•
•
ρ constant
If ρ>0 , the equation (2) can be resolved forward giving
the following result:
(3) Bt/yt=1/(1+ρ)*Et(bt+1/yt+1-dt+1/yt+1)=
(1+ρ)-n*et(bt+n/yt+n)-∑(1+ρ)-s*dt+s/yt+s
When n→∞, there is obtained the following
transversality condition
(4) lim (1+ ρ)-n*Et(bt+n/yt+n)=0
According to Trehan and Walsh (1991), the budget
process is called sustainable if the present value of the
future stock of public debt converge toward zero. This
condition excludes a Ponzi type scheme.
Academy of Economic Studies
Doctoral School of Finance and Banking
• The intertemporal budgetary constraint obtained on the
basis of the last two equations represents the departure
point of view for the majority of the empirical estimations
of the fiscal sustainability,
• (5) bt/yt= ∑(1+ρ)-s*-dt+s/yt+s.
• The public sector is solvent if the present value of
the future primary surpluses is equal to the current
value of the public debt, implying a zero value for the
long term public debt.
• The first test regarding the fiscal sustainability was
proposed by Hamilton and Flavin in an article published
in American Economic Review. The solution proposed by
Hamilton and Flavin (1986) is that the dynamic equation
of the public debt as a share in GDP to be written in the
following form:
• (8) bt/yt=(1+ ρt)*bt-1/yt-1+dt/yt+υt
Academy of Economic Studies
Doctoral School of Finance and Banking
• Iterating forward, the above equation implies the
same transversality condition, the difference
being related to the inclusion of an error term.
Hamilton and Flavin propose a test with the null
hypothesis represented by the following
relationship:
• (9) bt/yt=lim(1+ ρ)-s*(-dt+s/yt+s+υt+s).
• The test research if there is rejected or not the
null hypothesis against the alternative one:
lim(1+ ρ)-n*Et(bt+n/yt+n)=a*(1+r)t
• If a=0 then it is a necessary and sufficient
condition for the process described by the share
of the public debt in GDP to be stationary.
Academy of Economic Studies
Doctoral School of Finance and Banking
•
•
•
•
•
•
•
ρ is variable
If ρ is variable, the budgetary constraint can be solved
forward, obtaining;
(10) bt/yt=Et(П 1/(1+ ρt+s)bt+n/yt+n)-Et(∑ 1/(1+
ρt+s)*dt+s/yt+s)
Where δt,s= П 1/(1+ ρt+s)<=1, for s>=1.
The intertemporal budgetary constraint has the same
interpretation supposing that the present value of the
future surpluses is sufficient to cover the current debt.
For the analysis of the fiscal sustainability is necessary
to define the following variables:
xt= δt,n*bt/yt; zt= δt,n*dt/yt, where Δxt=zt.
The transversality condition becomes: lim Et(xt+n)=0
and the intertemporal budget constraint becomes: xt=lim Et(∑ zt+s).
Academy of Economic Studies
Doctoral School of Finance and Banking
• Wilcox shows in 1989 that the fiscal sustainability is
attained when xt is a stationary process with a zero
average. However, Uctum and Wickens (2000)
demonstrates a more general result, showing that this is
a condition sufficient but not necessary, the fiscal
sustainability being observed also when xt is a process
integrated of first order and zt is a stationary process
with a zero average. On the other hand, while the real
interest rate is adjusted with the economic growth,
Trehan and Walsh (1991) are reaching the conclusion
that the fiscal policy is sustainable if the total deficit
is stationary. From an econometric point of view this
condition is similar with the recognition of a cointegration
relationship between the primary surplus and public debt
where the cointegration vector is equal to (1,-r) or the
existence of cointegration relationship between total
revenues, total expenditures excluding the interest rate
ones and the public debt where the cointegratio vector is
equal with (1,-1,-r).
Academy of Economic Studies
Doctoral School of Finance and Banking
• The empirical testing of the intertemporal budgetary
constraint and the transversality condition by the
cointegration method is presented by Trehan and Walsh
(1988 and 1991) and Hakkio and Rush (1991). In order
for the public debt to converge toward zero, the primary
deficit in the average has to be zero. If the total
governmental expenditures and the total governmental
revenues are integrated of first order, then the sufficient
condition for the convergence of the budget deficit
toward zero is that the respective two variables are
cointegrated with a cointegration vector of (-1,1).
Academy of Economic Studies
Doctoral School of Finance and Banking
• While the intertemporal budget constraint is generally written as
equation (3), there is a need for an alternative equation to derive a
more pragmatic approach for testing.
(11) Rt = a + b GGt+ t
• The null hypothesis is b=1 and t is stationary. This condition
guarantees the existence of a cointegration relationship between the
total real consolidated revenues and total real consolidated
expenditures, implicitly guaranteeing that the government is
respecting its present value budget constraint.
• Cointegration of total government expenditures GG and R is
consistent with McCallum’s (1984) discussion of the government’s
intertemporal budget constraint. McCallum argued that a constant,
positive primary deficit cannot be financed entirely by bond sales
while a constant positive total deficit can be sustained from the
financing point of view.
Academy of Economic Studies
Doctoral School of Finance and Banking
• All of the tests advocated by Engle and Granger involve estimating
the so-called equilibrium or cointegration regression
• (15) Rt = a + b GGt+ t.
• The first three tests focus on the stationarity of the residuals while
the rest are examining whether the total consolidated revenues and
expenditures are obeying an error correction process.
• The first test involves the Durbin Watson statistic from the
equilibrium regression. If the Durbin Watson test is large, the two
series are cointegrated. The second method involves the testing of
the existence of an unit root of the time series of the residuals from
the equilibrium relationship. In this context if there is a unit root, then
the total consolidated real expenditures and total consolidated real
revenues are not cointegrated. Thus there is considered the
following regression and there is studied if the estimated coefficient,
, equals zero.
Academy of Economic Studies
Doctoral School of Finance and Banking
• If the coefficient is significantly different from zero, u is
stationary and the hypothesis of cointegration is
accepted. The second test is represented by an
augmented Dickey-Fuller test (ADF), including additional
lags u in the regression equation.
• (16)  u t = -u t-1 +e t
• The next pair of tests focus on the question if the
estimated coefficients 1 and 2 are jointly significant,
then the total consolidated expenditures and total
consolidated revenues have an error correction form and
are cointegrated.
•  Rt = c1 +11 Rt-1 +12GGt-1+t
• GGt= c2 +21 Rt-1 +22GGt-1+ Rt +t.
Academy of Economic Studies
Doctoral School of Finance and Banking
• Bohn (1998) sustains that some of the sustainability tests based on
the cointegration may produce misleading results. As an alternative
test, the author proposes to analyze the relationship between
primary deficit and public debt. If the primary deficit responds in a
positive way to the developments in the public debt than the public
debt is sustainable. The relationship proposed to be tested is of the
following form:
• (18) st=ρt*dt+μt
• For every t, ρ>0 a constant and μt a variable which includes other
factors. If μt is stationary then the fiscal policy is sustainable and the
government observes the intertemporal budget constraint.
• The existence of this reaction function ensures that the increase of
the public debt as a share of GDP decreases in every period by (1ρ) compared to a Ponzi scheme. In this way, the public debt ahead
with n period is reduced by a factor (1- ρ)n. For every positive ρ this
implies that the following relationship is respected:
• Et(ut,n*dt+n)=(1- ρ)n*dt→0.
Academy of Economic Studies
Doctoral School of Finance and Banking
• Results
Hungarian fiscal policy
• Regarding the data, the unit root tests are showing that the real
consolidated revenues series are stationary in level, logarithm and
per GDP forms while the real consolidated expenditures are nonstationary in level, logarithms and per GDP forms. In this context,
there is no long term relationship between the real total consolidated
government expenditures and real total consolidated government
revenues. Intuitively, this implies that the government is violating the
long term budget constraint because the total real government
consolidated expenditures have a tendency to grow while the real
total consolidated government revenues are stationary. In this
context, the limit term derived in equation 4 does not converge to
zero and implicitly the fiscal and budgetary policies promoted by the
Hungarian government along the time interval studied are not
sustainable.
Academy of Economic Studies
Doctoral School of Finance and Banking
• The regression of the real expenditures as a dependent
variable against the real revenues as an independent
variable, shows in the both cases (in levels and
logarithm) that the real expenditures are increasing in
the average quicker than the real revenues. Thus, a one
percent increase in the real revenues leads to an
average increase of 1.14 percent of the real
expenditures. Moreover, an one billion forints increase of
the real revenues leads in the average to an increase of
1.26 billion forints of the real expenditures. The
coefficient attached to the independent variable has a
high level of significance in the both cases, levels and
logarithm
Academy of Economic Studies
Doctoral School of Finance and Banking
• The regression of the real primary deficit as a share in
GDP as an independent variable against the real public
debt as a share in GDP shows that the primary surplus is
not responsive in the average to the developments in the
real public debt. A one percent of GDP increase in the
public debt level leads in the average to an increase of
0.06 percent of GDP of the average primary deficit. The
coefficient attached to the independent variable is highly
significant and its sign is negative. Moreover, the test of
the stationarity of the residuals series rejects the null
hypothesis
Academy of Economic Studies
Doctoral School of Finance and Banking
• Polish fiscal policy
• The total consolidated real revenues and expenditures are not
stationary being integrated of first order. The budget deficit (in levels
and per GDP), public debt (levels, per GDP and in logarithm) are not
stationary and are integrated of first order.
• The regression of the real expenditures as a dependent variable
against the real revenues as an independent variable, shows in the
both cases (in levels and logarithm) that the real expenditures are
increasing in the average quicker than the real revenues. Thus, a
one percent increase in the real revenues leads to an average
increase of 1,0009 percent of the real expenditures. Moreover, an
one million euro increase of the real revenues leads in the average
to an increase of 1,085 million euro of the real expenditures. The
coefficient attached to the independent variable has a high level of
significance in the both cases, levels and logarithm.
Academy of Economic Studies
Doctoral School of Finance and Banking
• The regression of the real primary deficit as an
independent variable against the real public debt shows
that the primary surplus as percent of GDP is not
responsive in the average to the developments in the
real public debt. A one percent of GDP increase in the
public debt leads in the average to an increase by 0.025
percent of GDP in the real primary deficit. The coefficient
attached to the independent variable is highly significant
and its sign is negative. The residuals generated by this
equation are stationary.
Academy of Economic Studies
Doctoral School of Finance and Banking
Response to Cholesky One S.D. Innovations ± 2 S.E.
Response of REALEXP_SA to REALEXP_SA
Response of REALEXP_SA to REALREV_SA
700
700
600
600
500
500
400
400
300
300
200
200
100
100
0
0
-100
-100
-200
-200
1
2
3
4
5
6
7
8
9
10
1
Response of REALREV_SA to REALEXP_SA
3
4
5
6
7
8
9
10
Response of REALREV_SA to REALREV_SA
600
600
500
500
400
400
300
300
200
200
100
100
0
0
-100
-100
-200
2
-200
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Academy of Economic Studies
Doctoral School of Finance and Banking
• As can be observed from the above graphs, the
response of the real expenditures to the innovations of
the real revenues, is important. The real expenditures
have a rapid increasing tendency on the short term and
are still increasing on the long term although at a more
slower pace. Moreover, the response of the real
revenues to the innovations of the real expenditures is
significant on the short term. However, the response is
losing steam on the medium and long term.
• The Johansen cointegration test does not indicate the
existence of a long term cointegration relationship
between the total real consolidated revenues and the
total real consolidated expenditures
Academy of Economic Studies
Doctoral School of Finance and Banking
• Romanian fiscal policy
• The stationarity tests, represented by the
Augmented Dickey Fuller Test and by the
Phillips Peron tests are showing that the total
consolidated real revenues and expenditures
are not stationary being integrated of first order.
The primary budget deficit (per GDP), public
debt (levels, per GDP and in logarithm) are
stationary.
Academy of Economic Studies
Doctoral School of Finance and Banking
• If revenues and expenditures are non-stationary, the
existence of a cointegration relationship between them is
key for the government to obey its long term budget
constraint. The results of the tests reports that the
residuals obtained from the regression of the total
government expenditures as an dependent variable
against the total government revenues as an
independent variable are stationary. There is a whole
number of tests which are pointing to this direction,
including the regression of the first difference of the
residuals against its first order lags and the realization of
the Dickey-Fuller, Augmented Dickey-Fuller and PhillipsPerron tests. As observed from the below table, the tstatistics and the associated probabilities are
demonstrating in both levels and logarithms that the
residuals are stationary and implicitly the government is
obeying its long term budget constraint.
Academy of Economic Studies
Doctoral School of Finance and Banking
• The regression of the real expenditures as a dependent
variable against the real revenues as an independent
variable, shows in the both cases (in levels and
logarithm) that the real expenditures are increasing in
the average slower than the real revenues. Thus, a one
percent increase in the real revenues leads to an
average increase of 0.86 percent of the real
expenditures. Moreover, an one billion RON increase of
the real revenues leads in the average to an increase of
0.94 billion RON of the real expenditures. The
coefficient attached to the independent variable has a
high level of significance in the both cases, levels and
logarithm.
Academy of Economic Studies
Doctoral School of Finance and Banking
• The regression of the real primary deficit as an
independent variable against the real public debt
shows that the primary surplus is responsive in
the average to the developments in the real
public debt. A one percent of GDP increase in
the real public debt leads in the average to a
0.41 percent of GDP (Annex, Table 10) diminish
in the real primary deficit. The coefficient
attached to the independent variable is highly
significant and its sign is positive. Moreover, the
residuals generated by this equation are
stationary.
Academy of Economic Studies
Doctoral School of Finance and Banking
• The next step in our approach is to study the relationship
between the public revenues and expenditures and the
sustainability of the budget deficits between the period
2000-2006. The exclusion of the crisis 1999 year
(Romania was about to collapse from the financial point
of view due to the weak international context and the
macroeconomic vulnerabilities associated with the high
level of short term debt contracted in the previous years)
from the dataset may lead to a better understanding on
the more recent developments in the Romanian public
finance area. This later period overlaps to a large extent
with the finalization from the implementation point of
view of the first IMF agreement after successive
previous failures in the post 1989 era and with the later
EU accession drive.
• We develop a vector autoregressive analysis of
Academy of Economic Studies
Doctoral School of Finance and Banking
Response to Cholesky One S.D. Innovations ± 2 S.E.
Response of CHELT_SA to CHELT_SA
Response of CHELT_SA to VEN_SA
16000
16000
12000
12000
8000
8000
4000
4000
0
0
-4000
-4000
1
2
3
4
5
6
7
8
9
10
1
Response of VEN_SA to CHELT_SA
12000
8000
8000
4000
4000
0
0
2
3
4
5
6
7
8
9
3
4
5
6
7
8
9
10
Response of VEN_SA to VEN_SA
12000
1
2
10
1
2
3
4
5
6
7
8
9
10
Academy of Economic Studies
Doctoral School of Finance and Banking
• As can be observed from the above graph, the response of the real
expenditures to the innovations of the real revenues, is very small
(actually the real expenditures are not responding to innovations in
the total real consolidated revenues). This finding is in line with the
actual policy choice promoted in this period in the public finance
area which had a clear orientation in reducing the excess aggregate
demand by the diminish of the budget deficit. The beginning of the
year revenue forecasts were usually based on very cautious
hypothesis regarding the economic growth, inflation and exchange
rate assumptions. The additional revenues observed along the year
due to higher than initially projected nominal GDP growth, improved
revenue collections and more favorable exchange rate
developments were used in the average to consolidate additionally
the budget deficit trough a late autumn budget rectification.
Academy of Economic Studies
Doctoral School of Finance and Banking
• Moreover, there is a clear evidence of a long term
response in the real revenues as a result of the
innovations incurred in the real expenditures. This could
be an additional outcome supported by the evidence in
the budgetary policymaking of the last several years
when the additional real expenditures were
accommodated trough the consolidation of the revenues
by promoting a coherent combination of policies aiming
at diminishing the direct tax rates and implementing
coherent measures to enlarge the tax base and improve
the tax administration and collection. However, more
probably, this may point out also to a “liquidity illusion”
created by the strong economic growth process and the
associated performance of the public revenues, leading
to the expectation of future gains in revenues and
neglecting their sensitivity to the economic cycle and
possible sustainability problems.
Academy of Economic Studies
Doctoral School of Finance and Banking
• The Johansen cointegration test indicates the existence
of a long term cointegration relationship between the
total real consolidated revenues and the total real
consolidated expenditures.
• The long term cointegration relationship between real
expenditures and real revenues in levels is of the
following form: Chelt_sa=39265.43+0.42*Ven_sa. The
relationship is significant from the T-statistics test result
for the coefficients and the coefficient atatched to the
real revenues level is smaller than unity.
Academy of Economic Studies
Doctoral School of Finance and Banking
• Conclusions
• The fiscal policies promoted by the Polish and Hungarian
governments in the last several years are not obeying
the intertemporal budget constraint and implicitly are
lacking sustainability. This outcome is pointed out by a
large set of indicators, including the analysis of the
stationarity of the public debt and deficits, the regression
of the primary surplus against the debt and the analysis
of the existence of a long term cointegration relationship
between the total government revenues and total
government expenditures.
• The fiscal policy promoted by the Romanian government was
sustainable. However, there is evidence of a “real revenue illusion”
due most probably to the change of vision in the last two years,
when the government promoted a more prociclical policy
characterized by increased budget deficits and by the full use of the
collected revenues in order to finance additional expenditures
(rather than to consolidate the budget deficit).
Academy of Economic Studies
Doctoral School of Finance and Banking
• Although, the public debt is very low and such policy would not
create sustainability problems in the short and medium term, it
should be seen with caution. The increased current account deficit
may pose sustainability problems as a result to the expansionary
aggregate demand and the worsening of the saving-investment
balance in the both public and private sectors. Moreover, this policy
exposes the public budget to adverse macroeconomic shocks such
as a possible slowing down of the economic growth and the
associated negative impact on both public revenues and
expenditures (implicitly a higher budget deficit). Such a policy
outcome is undesirable given the close to 3 percent of GDP actual
deficit targets for 2007 and 2008. The paper should be developed
additionally in order to study in a correlated perspective the
sustainability of both public and private sector developments with a
peculiar focus on the relationship between the sustainability of the
current account deficits and the sustainability of the budget deficits.
Academy of Economic Studies
Doctoral School of Finance and Banking
Academy of Economic Studies
Doctoral School of Finance and Banking