Table 5
Size of government in panel
GMM and FE estimates
|
(1) |
(2) |
(3) |
(4) |
(5) |
(6) | |
|
Dep var |
last_cgexp |
cgexp |
last_cgexp |
last_cgexp |
cgexp |
cgexp |
|
maj |
1.08 |
-0.86 | ||||
|
(1.67) |
(0.60) | |||||
|
district |
0.25 |
0.42 | ||||
|
(0.08)*** |
(0.15)*** | |||||
|
threshold |
-0.12 |
-0.18 | ||||
|
(0.17) |
(0.19) | |||||
|
coalition |
6.51 |
0.94 | ||||
|
(3.12)** |
(0.38)** | |||||
|
single |
-1.79 |
-0.52 | ||||
|
(4.31) |
(0.33) | |||||
|
L(1) dep var |
0.53 |
0.93 |
0.46 |
0.66 |
0.98 |
0.98 |
|
(0.21)** |
(0.03)*** |
(0.18)** |
(0.17)*** |
(0.04)*** |
(0.04)*** | |
|
L(2) dep var |
-0.08 |
-0.14 |
-0.13 | |||
|
(0.03)** |
(0.04)*** |
(0.04)*** | ||||
|
Periods |
legislatures |
years |
legislatures |
legislatures |
years |
years |
|
Over-id |
5.35(3) |
10.24(9) |
7.11(9) |
8.66(3)** |
9.49(3)** | |
|
AR(2) |
0.76 |
0.52 |
0.96 | |||
|
Estimation |
GMM |
FE |
GMM |
GMM |
FEIV |
FEIV |
|
Obs. |
111 |
889 |
105 |
105 |
822 |
822 |
|
N countries |
34 |
36 |
32 |
32 |
34 |
34 |
Standard errors in parentheses, robust in cols (1),(3)--(4) * significant at 10%; ** significant at 5%; *** significant at 1%.
Estimation methods: FE: with country fixed effects. FEIV: fixed-effects instrumental-variable estimates.
GMM Arellano-Bond estimates (all variables in first differences).
Specification : always included: constant, lpop, prop65, ygap; in cols (1), (3)-(4) leg_length.
In col (1), instruments for lagged last_cgexp is one further lag of this variable.
In cols (3)-(4), type of government (coalition or single) is treated as endogenous with one lag of this variable plus maj, district, and threshold as instruments.
In cols (5)-(6), instruments for type of government is lagged type of government, plus maj, district, and threshold.
AR(2) Arellano-Bond test for absence of second order serial correlation.
Over-id: Test of over-identifying restrictions, distributed like Chi2 with degrees of freedoms in parenthesis;
critical values at 5% confidence: 7.81 cols (1), (5)-(6); 16.92 cols (3)-(4).