Amparo Castelló-Climent, Universidad Carlos III de Madrid ... - Ivie

Table 3- Random and Fixed effects
Random effects
Fixed effects
(1) (2) (3) (4) (5) (6)
constant 0.069 ∗∗∗ 0.112 ∗∗∗ 0.118 ∗∗∗ 0.371 ∗∗∗ 0.460 ∗∗∗ 0.433 ∗∗∗
(0.025) (0.028) (0.030) (0.064) (0.069) (0.068)
lny -0.005 -0.009 ∗∗∗ -0.009 ∗∗∗ -0.043 ∗∗∗ -0.046 ∗∗∗ -0.048 ∗∗∗
(0.003) (0.003) (0.003) (0.007) (0.007) (0.007)
Gini y -0.021 -0.009 0.102 ∗∗ 0.102 ∗∗∗
(0.021) (0.021) (0.039) (0.038)
Gini h -0.043 ∗∗∗ -0.042 ∗∗∗ -0.062 ∗∗ -0.062 ∗∗
(0.013) (0.014) (0.027) (0.026)
Educf -0.006 -0.011 ∗∗ -0.010 ∗∗ 0.002 0.001 0.001
(0.004) (0.004) (0.005) (0.007) (0.007) (0.006)
Educm 0.010 ∗∗ 0.014 ∗∗∗ 0.013 ∗∗∗ 0.004 0.003 0.004
(0.004) (0.004) (0.004) (0.007) (0.007) (0.007)
pi -0.017 ∗∗∗ -0.017 ∗∗∗ -0.017 ∗∗∗ -0.017 ∗∗ -0.015 ∗∗ -0.015 ∗∗
(0.006) (0.006) (0.006) (0.007) (0.007) (0.007)
Time
dummies yes yes yes yes yes yes
R 2 0.146 0.180 0.181 0.257 0.251 0.279
Countries 55 55 55 55 55 55
Obs 250 250 250 250 250 250
Period 1965-2000 1965-2000 1965-2000 1965-2000 1965-2000 1965-2000
Note- Robust standard errors in parenthesis. *** 1 per cent significance level,
** 5 per cent significance level, * 10 per cent significance level. See Table 2
for definition of variables. R 2 is overall for random effectsandwithinforfixed
effects. Time dummies for the years 1965, 1970, 1975, 1980, 1985 and 1990
are included in columns (1)-(6).
Table 4 reports the results of the GMM estimators treating Gini y , Gini h ,
Educf and Educm as endogenous. 11 Columns (1)-(3) display the results of the
first difference GMM estimator also used by Forbes (2000). The results show a
negative and statistically significant coefficients of the initial per capita income
and the price of investment. With regard to the inequality variables, on the
one hand, as it was the case in the fixed effects model, the income inequality
coefficient is positive and statistically significant in all cases, corroborating
Forbes’ (2000) results. On the other hand, the coefficient of education inequality
remains always negative and it is also statistically significant. Therefore, the results
point out that the negative effect on growth from human capital inequality,
found in Castello and Domenech (2002), is not due to omitted variables bias.
Nevertheless, Bond et al. (2001) suggest using system GMM in the estimation
of growth equations since first differences GMM estimators may be biased
11 In order to compare the estimated coefficients of equation (1) and (3) the results obtained
using the GMM estimator are shown according to the transformations displayed after equation
(3).
17