|
(1)
|
(2)
|
(3)
|
(4)
|
(5)
|
(6)
|
(7)
|
(8)
|
---|
Constant
|
0.49 (23.16)
|
− 4.08 (22.68)
|
34.4 (30.08)
|
20.49 (27.04)
|
9.51 (21.39)
|
21.26 (33.18)
|
31.69 (24.38)
|
14.7 (20.47)
|
Δ(TD-MD)/GDP
|
14.52 (21.88)
| |
− 0.47 (24.09)
| |
17.37 (21.26)
| |
− 9.48 (23.62)
| |
ΔMD/GDP
|
99.99*** (33.12)
|
103.1*** (20.47)
| | |
38.9* (23.01)
|
64.81** (25.29)
| | |
ΔMout,b(t-1)
|
0.03 (0.29)
| | |
0.46** (0.2)
|
0.46** (0.15)
| | |
0.52*** (0.14)
|
Adjusted R2
|
0.43
|
0.47
|
− 0.04
|
0.18
|
0.28
|
0.13
|
− 0.02
|
0.25
|
Box-Ljung (3)
|
0.33
|
0.31
|
0.04
|
0.16
|
0.76
|
0
|
0
|
0.4
|
Observations
|
27
|
28
|
28
|
27
|
45
|
46
|
46
|
45
|
- This table reports regressions with ΔMout,b as dependent variable. The dataset contains yearly observations. Column 1 displays the baseline regression, whereas columns 2 to 8 display specific robustness checks. Column 1–4 is based on yearly observations from 1980 to 2008 and column 5 to 8 from 1970 to 2016. ***, **, and * denote statistical significance (two-tailed) at the 1%, 5%, and 10% significance level respectively. Standard errors are reported in parentheses. For columns 1, 4, and 8, heteroscedasticity-consistent (Huber-White) standard errors are used. Column 7 displays Newey-West standard errors due to serial correlation. Due to the small number of observations, for columns 2, 3, 5, and 6 we use normal standard errors that are larger than the Huber-White or Newey-West standard errors. To test for serial correlation of the error term, we performed the Box-Ljung test with a lag of 3 (considering the small sample size)