4470
Views
1829
Downloads |
Fiscal convergence and sustainability in the European Union
Vladimir Arčabić* 
Vladimir Arčabić
Affiliation: University of Zagreb, Faculty of Economics and Business, Department of Macroeconomics and Economic Development, Zagreb, Croatia
 0000-0003-4173-8637
Correspondence
varcabic@efzg.hr
Article | Year: 2018 | Pages: 353 - 380 | Volume: 42 | Issue: 4 Received: June 1, 2018 | Accepted: October 10, 2018 | Published online: December 14, 2018
|
FULL ARTICLE
FIGURES & DATA
REFERENCES
CROSSMARK POLICY
METRICS
LICENCING
PDF
|
(a)
Government Debt convergence results
|
|
log(t)
|
All
Countries
|
|
|
γ
|
-0.253*
|
|
|
|
|
|
t-stat
|
-22.13
|
|
|
|
|
|
Club classification
|
|
log(t)
|
Club1
[19]
|
Club2
[9]
|
|
|
|
|
γ
|
-0.00900
|
0.560
|
|
|
|
|
t-stat
|
-0.686
|
6.100
|
|
|
|
|
(b)
Government Revenues convergence results
|
|
|
log(t)
|
All
Countries
|
|
|
|
|
γ
|
-0.729*
|
|
|
|
|
t-stat
|
-33.34
|
|
|
|
|
Club classification
|
|
|
log(t)
|
Club1
[19]
|
Club2
[5]
|
Club3
[2]
|
Group4
[2]
|
|
|
γ
|
0.00700
|
0.792
|
0.114
|
-3.378*
|
|
|
t-stat
|
0.527
|
22.75
|
0.395
|
-2.779
|
|
|
(c)
Government Expenditures convergence results
|
|
log(t)
|
All
Countries
|
|
|
|
|
|
γ
|
-1.075*
|
|
|
|
|
|
t-stat
|
-10.68
|
|
|
|
|
|
Club classification
|
|
log(t)
|
Club1
[5]
|
Club2
[11]
|
Club3
[6]
|
Club4
[3]
|
Club5
[2]
|
|
γ
|
0.284
|
0.264
|
0.113
|
0.851
|
-0.125
|
|
t-stat
|
1.016
|
16.05
|
8.963
|
9.936
|
-0.154
|
|
Final classification
|
|
|
log(t)
|
Club1
[5]
|
Club2
[11]
|
Club3
[9]
|
Club4
[2]
|
|
|
γ
|
0.284
|
0.264
|
0.169
|
-0.125
|
|
|
t-stat
|
1.016
|
16.05
|
14.93
|
-0.154
|
|
Note: The table presents γ coefficient from log t regression together with t-statistics. * Marks a rejection of convergence at 5% level. Numbers in brackets are number of countries in the club. Club classification is a result of the initial clustering algorithm. Final classification is a result after club merging. Final classification is presented only when club merging is significant. Countries that form different clubs are presented in figure 3.
Note: Convergence clubs are in squares, non-convergent groups are in circles.
2a: Percent of countries
converging to the average gov. debt (%)
|
|
ADF
|
Lee & Strazicich
|
Enders & Lee
|
|
EU
[28]
|
3.57
|
3.57
|
7.14
|
|
Club
1 [19]
|
0.00
|
5.26
|
5.26
|
|
Club
2 [9]
|
22.22
|
0.00
|
0.00
|
2b: Percent of
countries converging to the average gov. revenues (%)
|
|
ADF
|
Lee & Strazicich
|
Enders & Lee
|
|
EU
[28]
|
35.71
|
85.71
|
46.43
|
|
Club
1 [19]
|
42.11
|
94.74
|
57.89
|
|
Club
2 [5]
|
40.00
|
100.00
|
40.00
|
|
Club
3 [2]
|
0.00
|
100.00
|
100.00
|
2c: Percent of
countries converging to the average gov. expenditures (%)
|
|
ADF
|
Lee & Strazicich
|
Enders & Lee
|
|
EU
[28]
|
39.29
|
78.57
|
46.43
|
|
Club
1 [5]
|
40.00
|
100.00
|
40.00
|
|
Club
2 [11]
|
54.55
|
90.91
|
81.81
|
|
Club
3 [9]
|
33.33
|
77.78
|
55.56
|
|
Club
4 [2]
|
100.00
|
100.00
|
100.00
|
|
Models
|
Model 1
Benchmark
|
Model 2
Club 1
|
Model 3
Club 2
|
Model 4
EU-15
|
Model 5
EU-13
|
Model 6
debt ≥ 90%
|
Model 7
debt < 90%
|
|
Variables
|
Surplus
|
Surplus
|
Surplus
|
Surplus
|
Surplus
|
Surplus
|
Surplus
|
|
Surplus (lagged) (
)
|
0.502***
|
0.415***
|
0.390***
|
0.547***
|
0.308**
|
0.573**
|
0.526***
|
|
|
(0.08)
|
(0.09)
|
(0.10)
|
(0.06)
|
(0.11)
|
(0.21)
|
(0.10)
|
|
Debt (
)
|
-0.018
|
-0.009
|
-0.043
|
-0.029
|
0.040
|
0.036
|
-0.057*
|
|
|
(0.01)
|
(0.01)
|
(0.12)
|
(0.02)
|
(0.03)
|
(0.08)
|
(0.03)
|
|
Output Gap (
)
|
0.286***
|
0.413***
|
0.435***
|
0.394***
|
0.388**
|
0.297**
|
0.305***
|
|
|
(0.05)
|
(0.07)
|
(0.13)
|
(0.04)
|
(0.15)
|
(0.09)
|
(0.07)
|
|
Constant
|
0.776
|
0.122
|
2.030
|
1.879
|
-2.351*
|
-4.298
|
2.459*
|
|
|
(0.76)
|
(1.06)
|
(5.63)
|
(1.23)
|
(1.15)
|
(8.69)
|
(1.44)
|
|
Observations
|
1,708
|
1,159
|
549
|
915
|
793
|
281
|
1,427
|
|
Number of countries
|
28
|
19
|
9
|
15
|
13
|
9
|
26
|
|
F-test
|
0
|
3.40e-06
|
0.00139
|
0
|
2.43e-05
|
0.00352
|
0
|
|
# of instruments
|
23
|
14
|
11
|
14
|
11
|
11
|
20
|
|
Hansen test
|
0.222
|
0.100
|
0.371
|
0.175
|
0.102
|
0.506
|
0.166
|
|
Models
|
Model 1
Benchmark
|
Model 2
Club 1
|
Model 3
Club 2
|
Model 4
EU-15
|
Model 5
EU-13
|
Model 6
debt ≥ 90%
|
Model 7
debt < 90%
|
|
Variables
|
Surplus
|
Surplus
|
Surplus
|
Surplus
|
Surplus
|
Surplus
|
Surplus
|
|
Surplus (lagged) (
)
|
0.512***
|
0.515***
|
0.491***
|
0.637***
|
0.336***
|
0.248**
|
0.611***
|
|
|
(0.06)
|
(0.07)
|
(0.12)
|
(0.04)
|
(0.07)
|
(0.10)
|
(0.09)
|
|
Debt (
)
|
0.006
|
0.004
|
0.025
|
0.003
|
0.029**
|
0.031**
|
-0.009
|
|
|
(0.01)
|
(0.01)
|
(0.02)
|
(0.00)
|
(0.01)
|
(0.01)
|
(0.01)
|
|
Output Gap (
)
|
0.176***
|
0.172***
|
0.209**
|
0.226***
|
0.197***
|
0.213
|
0.133***
|
|
|
(0.04)
|
(0.04)
|
(0.09)
|
(0.03)
|
(0.05)
|
(0.12)
|
(0.04)
|
|
Constant
|
-0.614*
|
-0.713
|
-0.998
|
-0.327
|
-1.831***
|
-3.951**
|
0.168
|
|
|
(0.36)
|
(0.42)
|
(0.71)
|
(0.35)
|
(0.40)
|
(1.48)
|
(0.63)
|
|
Observations
|
1,708
|
1,159
|
549
|
915
|
793
|
281
|
1,427
|
|
Number of countries
|
0.304
|
0.298
|
0.335
|
0.457
|
0.173
|
0.123
|
0.384
|
|
F-test
|
28
|
19
|
9
|
15
|
13
|
9
|
26
|
|
# of instruments
|
0.512***
|
0.515***
|
0.491***
|
0.637***
|
0.336***
|
0.248**
|
0.611***
|
|
Hansen test
|
(0.06)
|
(0.07)
|
(0.12)
|
(0.04)
|
(0.07)
|
(0.10)
|
(0.09)
|
Note: Standard errors in parentheses, ***, **, and * mark statistical significance at 1%, 5%, and 10% level. Model 1 is the benchmark model. Models 2 and 3 include countries from endogenous debt convergence clubs 1 and 2, respectively. Models 4 and 5 include EU-15 and EU-13 countries, and models 6 and 7 include subsamples with government debt ≥ 90% and debt < 90% of GDP, respectively.
|
Country
|
Revenues
|
Expenditures
|
Debt
|
Surplus
|
|
Mean
|
St. dev.
|
Mean
|
St. dev.
|
Mean
|
St. dev.
|
Mean
|
St. dev.
|
|
Belgium
|
0.498
|
0.014
|
0.519
|
0.029
|
102.371
|
6.191
|
1.489
|
2.662
|
|
Bulgaria
|
0.375
|
0.032
|
0.378
|
0.035
|
25.430
|
12.762
|
0.809
|
3.995
|
|
Czech R.
|
0.397
|
0.021
|
0.427
|
0.030
|
34.493
|
6.377
|
-1.570
|
2.538
|
|
Denmark
|
0.542
|
0.012
|
0.537
|
0.025
|
41.284
|
6.727
|
2.350
|
2.937
|
|
Germany
|
0.439
|
0.011
|
0.453
|
0.017
|
69.791
|
6.443
|
0.853
|
1.749
|
|
Estonia
|
0.380
|
0.027
|
0.376
|
0.035
|
6.770
|
2.386
|
0.605
|
2.605
|
|
Ireland
|
0.331
|
0.031
|
0.374
|
0.095
|
63.992
|
35.900
|
-3.019
|
8.925
|
|
Greece
|
0.427
|
0.042
|
0.501
|
0.052
|
135.084
|
31.814
|
-2.996
|
4.643
|
|
Spain
|
0.380
|
0.016
|
0.417
|
0.035
|
64.440
|
24.433
|
-1.729
|
4.398
|
|
France
|
0.506
|
0.016
|
0.544
|
0.023
|
78.400
|
13.788
|
-1.601
|
1.526
|
|
Croatia
|
0.433
|
0.030
|
0.478
|
0.030
|
56.491
|
19.248
|
-1.880
|
2.281
|
|
Italy
|
0.453
|
0.019
|
0.485
|
0.019
|
115.085
|
12.457
|
1.345
|
1.277
|
|
Cyprus
|
0.365
|
0.030
|
0.395
|
0.055
|
73.230
|
21.307
|
-0.192
|
5.510
|
|
Latvia
|
0.352
|
0.022
|
0.374
|
0.036
|
26.770
|
14.433
|
-1.088
|
3.003
|
|
Lithuania
|
0.341
|
0.013
|
0.367
|
0.039
|
28.502
|
10.293
|
-1.275
|
3.675
|
|
Luxembourg
|
0.434
|
0.013
|
0.418
|
0.026
|
14.941
|
7.215
|
1.360
|
1.821
|
|
Hungary
|
0.444
|
0.021
|
0.490
|
0.023
|
69.585
|
8.796
|
-0.745
|
3.255
|
|
Malta
|
0.384
|
0.020
|
0.416
|
0.023
|
66.048
|
4.001
|
0.266
|
2.870
|
|
Netherlands
|
0.429
|
0.009
|
0.446
|
0.021
|
56.298
|
8.245
|
-0.165
|
2.086
|
|
Austria
|
0.489
|
0.012
|
0.513
|
0.019
|
76.310
|
6.738
|
0.363
|
1.796
|
|
Poland
|
0.395
|
0.013
|
0.435
|
0.018
|
48.860
|
5.134
|
-1.724
|
1.608
|
|
Portugal
|
0.415
|
0.021
|
0.469
|
0.035
|
92.343
|
30.325
|
-1.906
|
3.499
|
|
Romania
|
0.335
|
0.016
|
0.367
|
0.033
|
25.756
|
10.096
|
-1.550
|
3.267
|
|
Slovenia
|
0.434
|
0.010
|
0.472
|
0.048
|
44.276
|
22.349
|
-1.759
|
4.888
|
|
Slovakia
|
0.373
|
0.024
|
0.417
|
0.039
|
43.263
|
8.989
|
-1.969
|
2.157
|
|
Finland
|
0.531
|
0.013
|
0.520
|
0.040
|
46.337
|
10.241
|
1.891
|
3.288
|
|
Sweden
|
0.520
|
0.019
|
0.515
|
0.016
|
42.539
|
4.814
|
1.568
|
1.730
|
|
UK
|
0.378
|
0.012
|
0.420
|
0.036
|
60.967
|
22.826
|
-2.558
|
2.478
|
Figure 1Relative transition paths DISPLAY Figure
Table 1log t convergence test results and convergence clubs classification DISPLAY Table
Figure 2Rolling window estimation of log t regression DISPLAY Figure
Figure 3Convergence clubs DISPLAY Figure
Figure 4β-convergence in clubs DISPLAY Figure
Table 2Club convergence using unit root tests DISPLAY Table
Table 3Results of fiscal sustainability analysis using system GMM-CCE model DISPLAY Table
Table 4Results of fiscal sustainability analysis using FE model DISPLAY Table
Figure A1Government revenues and expenditures as a percent of GDP DISPLAY Figure
Figure A2Primary surplus and government debt as a percent of GDP DISPLAY Figure
Table A1Descriptive statistics DISPLAY Table
* The author would like to thank Junsoo Lee, Robert J. Sonora, Hrvoje Šimović, Josip Tica, Tomislav Globan, Kyoung Tae Kim, Ozana Nadoveza Jelić, Irena Raguž Krištić, Yu You, and two anonymous referees for helpful comments on the paper.
1 1997-2003 period is considered.
2 For each variable we run a separate test.
3 Phillips and Sul (  2007; 2009) provide more technical details of the test. For empirical analysis we use a set of procedures described in Du ( 2017).
4 To form a group, the t-statistic for parameter γ from log t regression must be t > -1.65.
5 Bohn ( 2005; 2007) criticize fiscal sustainability analysis based on unit root and cointegration techniques popularized by Trehan and Walsh ( 1988) and Hamilton and Flavin ( 1986). He argues that such techniques are not capable of rejecting sustainability hypothesis because the relevant debt variables are necessary stationary after a finite number of differencing and thus in compliance with the intertemporal budget constraint (IBC).
6 We use first differencing instead of forward orthogonal deviaton (FOD) because our data set is a balanced panel. Refer to Arellano and Bover ( 1995) and Blundell and Bond ( 1998) for complete technical details.
7 Indeed, when we apply Pesaran ( 2015) test for weak cross-sectional dependence to the model, the null hypothesis of cross-sectional independence can be easily rejected.
8 EU core countries are Austria, Belgium, Denmark, Finland, France, Germany, Luxembourg, Netherlands, Sweden, and UK. Other 18 countries form EU periphery.
9 Analysis of factors and criteria to which countries converge is beyond the scope of this paper.
10 We consider government debt convergence clubs only, but clubs are fairly homogeneous between fiscal variables, as discussed. In addition, some government revenues and expenditures convergence clubs include only a few countries, which is impractical for panel data analysis.
11 Only Model 6 in table 4 has a positive, but insignificant output gap.
|
|
December, 2018
IV/2018 |
