GMM estimates imply that TFP is negative at about 0.4 percent whereas GMM estimate
implies a positive rate of growth of 0.16 percent.
|
Table 1 Estimates of Production Function | |||
|
OLS |
GLS |
GMM* | |
|
A |
-0.773 (0.00) |
-1.611 (0.00) |
-0.011 (0.07) |
|
g |
-0.714E-02 (0.00) |
-0.427E-02 (0.00) |
0.163E-02 |
|
α |
0.858 |
0.197 (0.00) |
0.204 |
|
2 R |
0.877 |
0.872 |
0.002 |
|
Notes: p-values in the parenthesis below the coefficients. | |||
Since these estimates have some limitations we shall present now the SGMM
estimates in Table 2. In the first column of Table 2 estimates of the modified production
function where TFP depends only on trend and GLO are presented. All the coefficients are
significant and the estimate of profit share at 0.37 is close to its stylized value of one third in
the growth accounting exercises. The coefficient of trend at -0.026 implies that the overall
TFP is negative at -2.6 percent per year. However, GLO has a small but significant growth
___2
effect on TFP. The R of the levels equation is high and that of the first differences low but
an improvement over the GMM estimate in Table 1. The average value of GLO ranged from
a high of 40 for Nigeria to a low of 16.5 for Burundi. To offset the negative value of TFP it
is necessary for GLO to increase by another 8 points even for Nigeria.
We have added other control variables viz., GRAT, INSTI and DLP to the estimate of
the equation with GLO alone, but none of the coefficients of these control variables are
significant. These estimates are not reported to conserve space. Therefore, we have