Download ppt presentation

Twin Deficits in Greece
Theodore Panagiotidis
Hellenic Observatory and University of Macedonia
1
Introduction
•
•
•
•
Twin Deficits Hypothesis: fiscal shocks that cause deterioration of the
government’s budget also worsen a country’s current account balance.
Current Account Deficit = Investment - Private Saving + Budget deficit
Trade Deficit: when consumption exceeds production
Ricardian equivalence: consumers internalize the government’s budget
constraint and any tax change does not affect their spending. It does not matter
whether the government finances its spending through debt or tax increase.
Total level of demand remains unchanged.
2
Background
•
•
•
•
Two distinct ways of thinking about the link between fiscal policy and the current
account.
Mundell-Fleming model: with flexible exchange rates, fiscal deficits appreciate the
currency: a higher relative price of domestic goods crowds out net export. If fiscal
deficits also raise the interest rate, the resulting external imbalance may be migitated
because of a simultaneous fall in domestic investment. This model stresses changes in
terms of trade and interest rates.
Intertemporal approach to the current account emphasize consumption smoothing
and optimal intertemporal investment decisions.
Refined Twin Deficit Hypothesis: twin deficits are more likely if (1) the economy is
relative open i.e. highly integrated into world markets and (2) fiscal expansions are
persistent (Corsetti and Muller 2006)
3
Literature I
•
•
•
•
•
•
Early studies apply OLS regressions to cross country data. Positive and significant
relationship between current account deficits and budget deficits is found (Milne 1977
and Bernheim 1987).
Supportive evidence of the twin deficit hypothesis also provided by simulation studies
(Bryat et al 1988 and Zietz and Pemberton 1990).
VAR models:
Abell (1990) employed a sever variable VAR with first differenced data for the US
from 1979-85. Granger causality and IRF suggest that the two deficits are twins and
that budget deficits affect current account deficits through the transmission
mechanisms of interest rates and exchange rates.
Bachman (1992) and Rosensweig & Tallman (1993) through unrestricted VAR in
differences: government budget deficits have a sizeable effect on trade deficits.
Enders & Lee (1990) employ quarterly data 1947-1987 and a six variable structural
VAR with differenced data. Supports the Ricardian equivalence: no evidence that BD
raises CA.
4
Literature II
•
•
•
•
•
•
Cointegration studies (require both CA and BD to be nonstationary, problems?)
Bachman 1992 fails to find any evidence of cointegration (LR relationship).
Dibooglou (1997) does find evidence of cointegration and argues that budget deficits
and increases in real interest rates are associated with current account deficits.
Holmes (2010) employ a nonparametric test for nonlinear co-trending and found that
deficits are stationary around a nonlinear deterministic trend and are co-trended
insofar as they share a common nonlinear deterministic trend.
VAR in levels
Kim & Roubini (2008) employ a VAR in levels for the post-Bretton-Woods periods
and find that increases in the US government budget deficits actually improve the US
current account balance (result opposite to the standard theoretical prediction). Data
are quarterly from 1973-2004:1. The empirical results (IRF) suggest that government
deficit shocks improve the current account. The current account improvement resulted
from a partial Ricardian behavior of private saving (increase) and a fall in investment (a
crowding out effect): twin divergence rather than twin deficit. CA counter cyclical:
increase in output (productivity shock) increases the demand for foreign goods and
worsens the CA. Private savings increase to compensate for the gov saving decrease
(Ricardian equivalence) (partial increase).
5
Literature III
•
•
•
•
•
•
Corsetti & Muller (2006): In the US and Australia, relative less open and government
spending shocks less persistent, ca impact on fiscal policy is rather limited
Grier & Ye (2009): Quarterly observations from 1948Q1 to 2005Q1. There is no longrun twin relationship between CA and BD in the US. In the short-run there is a
significant positive and persistent short-run effect of budget shocks on the CA (SR:
VAR-GARCH, demeaned data, GIRF and VD).
Different samples, different variables, different econometric methodology…different
results.
Chinn (2005): US world’s largest debtor nation. Suggest three policies to reduce the
deficit substantially so they will not be no further increases in its indebtedness to the
rest of the world: i) A concerted effort to reduce the federal budget deficit, ii)
Reducing the quantity of oil imports, iii) Coordinating a revaluation of East Asian
currencies.
Evidence for Greece
Vamvoukas (1999) employed data from 1948 to 1994 for output, prices, budget deficit
and trade deficit. Methodology includes unit root tests and cointegration. Results from
the VECM suggest a predominantly unidirectional causality from budget deficit to
trade deficit in the LR and in the SR.
6
• The Greek Economy in graphs…
7
0
-2
-4
-6
-8
-10
-12
-14
-16
50
55
60
65
70
75
80
Budget Deficit
85
90
95
00
05
10
Current Account
8
Debt % of GDP
160
140
120
100
80
60
40
20
0
50
55
60
65
70
75
80
85
90
95
00
05
10
9
GDP Growth
15.0
12.5
10.0
7.5
5.0
2.5
0.0
-2.5
-5.0
50
55
60
65
70
75
80
85
90
95
00
05
10
10
Government Spending % GDP
40
36
32
28
24
20
16
50
55
60
65
70
75
80
85
90
95
00
05
10
11
Exports/Imports
55
50
45
40
35
30
25
50
55
60
65
70
75
80
85
90
95
00
05
10
12
• Correlation between BD and CA: 0.10
• Correlation between BD and CA once the
cyclical component is removed (HP filter): 0.17
• In the US the correlation is -0.16 (Kim &
Rounibi 2008)
• Corestti and Muller: -0.24 US, -0.26 UK, -0.16
Canada and -0.37 in Australia using only the
cyclical components.
• Greece: -0.00159 (p-value 0.91) only the cyclical
components
13
-2
-4
-6
-8
-10
-12
50
55
60
65
70
75
80
CA HP
85
90
95
00
05
10
BD HP
14
VAR
• Annual Observations from 1960 till 2010.
• Unit Root Tests: ADF, KPSS, nonlinear unit root tests KSS and Unit Root
tests with endogenous breaks.
• Evidence for I(0) mainly from the nonlinear unit root tests.
• Vector Autoregressive Model (VAR): RGDP , BD, CA and Long Term
Interest rates
• Lag order AIC (2 lags)
• SR: positive but insignificant the lags of BD for the CA equation.
• Generalised Impulse response Functions: conceptual experiment, ordering of
the variables.
• Variance Decompositions
• Response Standard Errors from MC 1000 repetitions
• Variations of the VAR do not alter the results
15
Unit Root & Stationarity tests
Current Account
Budget Deficit
Debt
Exports/Imports
GDP Growth
Government Spending
Long Term Interest rates
cv 10%
Levels
-2.848394
-2.175386
2.050193
-1.701652
-1.475524
-1.526953
-1.60038
p -value
0.0577
0.2175
0.9999
0.4231
0.5389
0.5121
0.4748
KPSS
0.382527
0.361141
0.983643
0.230266
0.758385
0.633804
0.214096
0.347
KSS 1
-1.524872
-2.058281
0.818821
-0.651826
-2.510606
0.094645
-1.253082
KSS 2
-1.732349
-2.643842
0.113563
-3.101221
-1.891222
-1.840952
-1.741872
-1.92
-2.66
KSSUR
3 with Structural Break
Break Date
-4.30529
-1.7779
1985
-2.374034
-2.8222
1981
-2.858409
-2.8605
1990
trend
-3.089579
-2.3215
1979
-2.582601
-4.5023
1974
-2.320748
-1.2235
1988
-1.874075
-1.618
1980
-3.13
-2.58
16
Response to Generalized One S.D . Innov ations ± 2 S.E.
Response of RGDP to RGDP
Response of RGDP to BD
Response of RGDP to CA
Response of RGDP to IR_LONG_T ERM
60,000
60,000
60,000
60,000
40,000
40,000
40,000
40,000
20,000
20,000
20,000
20,000
0
0
0
0
-20,000
-20,000
-20,000
-20,000
-40,000
-40,000
1
2
3
4
5
6
7
8
9
10
-40,000
1
2
Response of BD to RGDP
3
4
5
6
7
8
9
10
-40,000
1
2
Response of BD to BD
3
4
5
6
7
8
9
10
1
Response of BD to CA
3
3
3
2
2
2
2
1
1
1
1
0
0
0
0
-1
-1
-1
-1
-2
1
2
3
4
5
6
7
8
9
10
-2
1
2
Response of CA to RGDP
3
4
5
6
7
8
9
10
2
Response of CA to BD
3
4
5
6
7
8
9
10
1
Response of CA to CA
2
2
1
1
1
1
0
0
0
0
-1
-1
-1
-1
-2
2
3
4
5
6
7
8
9
10
-2
1
Response of IR_LONG_TERM to RGDP
2
3
4
5
6
7
8
9
10
Response of IR_LONG_TERM to BD
2
3
4
5
6
7
8
9
10
Response of IR_LONG_T ERMto CA
1
4
4
2
2
2
2
0
0
0
0
-2
-2
-2
-2
-4
-4
-4
-4
-6
2
3
4
5
6
7
8
9
10
-6
1
2
3
4
5
6
7
8
9
10
7
8
9
10
2
3
4
5
6
7
8
9
10
2
3
4
5
6
7
8
9
10
Response of IR_LONG_T ERM to IR_LONG_T ERM
4
1
6
-2
1
4
-6
5
Response of CA to IR_LONG_T ERM
2
1
4
-2
1
2
-2
3
Response of BD to IR_LONG_T ERM
3
-2
2
-6
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
17
Va riance Decompositio n ± 2 S.E.
Percent RGDP variance due to RGDP
Percent RGDP variance due to BD
Percent RGDP variance due to CA
Percent RGDP variance due to IR_LONG_T ERM
120
120
120
120
80
80
80
80
40
40
40
40
0
0
0
0
-4 0
-40
1
2
3
4
5
6
7
8
9
10
-40
1
2
Percent BD variance due to RGDP
3
4
5
6
7
8
9
10
-40
1
Percent BD variance due to BD
2
3
4
5
6
7
8
9
10
1
Percent BD variance due to CA
120
120
120
80
80
80
80
40
40
40
40
0
0
0
0
-40
1
2
3
4
5
6
7
8
9
10
-40
1
2
Percent CA variance due to RGDP
3
4
5
6
7
8
9
10
Percent CA variance due to BD
2
3
4
5
6
7
8
9
10
1
Percent CA variance due to CA
120
120
80
80
80
80
40
40
40
40
0
0
0
0
-40
2
3
4
5
6
7
8
9
10
-40
1
Percent IR_LONG_TERMvariance due to RGDP
2
3
4
5
6
7
8
9
10
2
3
4
5
6
7
8
9
10
Percent IR_LONG_T ERMvariance due to CA
1
120
120
120
80
80
80
80
40
40
40
40
0
0
0
0
-40
1
2
3
4
5
6
7
8
9
10
-40
1
2
3
4
5
6
7
8
9
10
7
8
9
10
2
3
4
5
6
7
8
9
10
2
3
4
5
6
7
8
9
10
Percent IR_LONG_TERMvariance due to IR_LONG_T ERM
120
-4 0
6
-40
1
Percent IR_LONG_T ERMvariance due to BD
5
Percent CA variance dueto IR_LONG_T ERM
120
1
4
-40
1
120
-4 0
3
Percent BD variance dueto IR_LONG_T ERM
120
-4 0
2
-40
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
18
Replace Real GDP with GDP Growth
R e s ponse to Generalized O ne S.D . In nov ations ± 2 S.E.
Respons e of GDP _GROW T H to GDP _GROWT H
Res ponse of GDP _GROWT H to B D
Res pons e of GDP _GROW T H to CA
Res pons e of GDP _GROW T H to IR_LONG_T E RM
4
4
4
4
3
3
3
3
2
2
2
2
1
1
1
1
0
0
0
0
-1
-1
-1
-1
-2
-2
1
2
3
4
5
6
7
8
9
10
-2
1
2
Res pons e of B D to GDP _GROW T H
3
4
5
6
7
8
9
10
-2
1
2
Res pons e of B D to B D
3
4
5
6
7
8
9
10
1
Res pons e of B D to CA
3
3
3
2
2
2
2
1
1
1
1
0
0
0
0
-1
-1
-1
-1
-2
-2
-2
2
3
4
5
6
7
8
9
10
1
2
Res pons e of CA to GDP _GROW T H
3
4
5
6
7
8
9
10
2
Res pons e of CA to B D
3
4
5
6
7
8
9
10
1
Res pons e of CA to CA
3
3
2
2
2
2
1
1
1
1
0
0
0
0
-1
-1
-1
-1
-2
-2
-2
3
4
5
6
7
8
9
10
1
Res pons e of IR_LONG_TE RM to GDP _GROW TH
2
3
4
5
6
7
8
9
10
Res pons e of IR_LONG_T E RM to B D
2
3
4
5
6
7
8
9
10
Res ponse of IR_LONG_T E RM to CA
1
4
4
2
2
2
2
0
0
0
0
-2
-2
-2
-2
-4
-4
-4
-4
-6
-6
-6
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
7
8
9
10
2
3
4
5
6
7
8
9
10
2
3
4
5
6
7
8
9
10
Res pons e of IR_LONG_T E RM to IR_LONG_T E RM
4
2
6
-2
1
4
1
5
Res pons e of CA to IR_LONG_T E RM
3
2
4
-2
1
3
1
3
Res pons e of B D to IR_LONG_T E RM
3
1
2
-6
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
19
Ala Grier & Ye (2009)
R e s pons e to Generalized O ne S.D . In nov ations ± 2 S.E.
Respons e of GDP _GROW T H to GDP _GROWT H
Res ponse of GDP _GROWT H to B D
Res pons e of GDP _GROW T H to CA
Res pons e of GDP _GROW T H to IR_LONG_T E RM
4
4
4
4
3
3
3
3
2
2
2
2
1
1
1
1
0
0
0
0
-1
-1
-1
-1
-2
-2
1
2
3
4
5
6
7
8
9
10
-2
1
2
Res pons e of B D to GDP _GROW T H
3
4
5
6
7
8
9
10
-2
1
2
Res pons e of B D to B D
3
4
5
6
7
8
9
10
1
Res pons e of B D to CA
3
3
3
2
2
2
2
1
1
1
1
0
0
0
0
-1
-1
-1
-1
-2
-2
-2
2
3
4
5
6
7
8
9
10
1
2
Res pons e of CA to GDP _GROW T H
3
4
5
6
7
8
9
10
2
Res pons e of CA to B D
3
4
5
6
7
8
9
10
1
Res pons e of CA to CA
3
3
2
2
2
2
1
1
1
1
0
0
0
0
-1
-1
-1
-1
-2
2
3
4
5
6
7
8
9
10
-2
1
Res pons e of IR_LONG_TE RM to GDP _GROW TH
2
3
4
5
6
7
8
9
10
Res pons e of IR_LONG_T E RM to B D
2
3
4
5
6
7
8
9
10
Res ponse of IR_LONG_T E RM to CA
1
4
4
2
2
2
2
0
0
0
0
-2
-2
-2
-2
-4
2
3
4
5
6
7
8
9
10
-4
1
2
3
4
5
6
7
8
9
10
7
8
9
10
2
3
4
5
6
7
8
9
10
2
3
4
5
6
7
8
9
10
Res pons e of IR_LONG_T E RM to IR_LONG_T E RM
4
1
6
-2
1
4
-4
5
Res pons e of CA to IR_LONG_T E RM
3
1
4
-2
1
3
-2
3
Res pons e of B D to IR_LONG_T E RM
3
1
2
-4
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
20
Granger causality
•
•
TEST FOR GRANGER-CAUSALITY:
H0: "BD" do not Granger-cause "GDP_GROWTH, CA, IR_LONG_TERM"
•
•
Test statistic l = 2.6202
pval-F( l; 6, 160) = 0.0189
•
•
TEST FOR INSTANTANEOUS CAUSALITY:
H0: No instantaneous causality between "BD" and "GDP_GROWTH, CA, IR_LONG_TERM"
•
•
Test statistic: c = 3.4413
pval-Chi( c; 3) = 0.3285
21
22
23
24
Growth and Debt
• Debt: accumulated BD.
• Is growth possible in time of debt?
• Reinhart and Rogoff (2010): threshold effect
above 90%.
• Criticism Irons and Bivens (2010):correlation
diff to causality, lower growth may cause debt to
increase, static and not dynamics, statistical
significance etc.
25
Greece
15
10
5
0
160
-5
120
80
40
0
50
55
60
65
70
75
DEBT
80
85
90
95
00
05
10
GDP_GROWTH
26
US
.20
.15
GROWTH
.10
.05
.00
-.05
-.10
-.15
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
DEBT
27
15.0
12.5
GDP_GROWTH
10.0
7.5
5.0
2.5
0.0
-2 . 5
-5 . 0
0
20
40
60
80
100
120
140
160
DEBT
28
Greece
US
Quantile Process Es timates (95% CI)
Quantile Process Estimates (95% CI)
C
C
.16
16
.12
12
.08
8
.04
4
.00
0
-.04
-4
0.0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.1
0.2
0.3
0.4
1.0
0.5
0.6
0.7
0.8
0.9
1.0
0.6
0.7
0.8
0.9
1.0
Quantile
Quantile
DEBT
DEBT
.050
.00
.025
-.02
.000
-.04
-.025
-.050
-.06
-.075
-.08
-.100
0.0
-.10
0.0
0.1
0.2
0.3
0.4
0.5
Quantile
0.6
0.7
0.8
0.9
1.0
0.1
0.2
0.3
0.4
0.5
Quantile
29
Conclusions
• Positive (low) correlation between the two
• No long-run relationship between BD and CA.
• Positive but insignificant response of fiscal policy to
current account. Increased importance as horizon
expands.
• Twin divergence hypothesis is rejected, weak support
for the twin deficits.
• High GDP Growth even when debt is around 100% in
Greece
30