Mapping growth into economic development: has elite political instability mattered in Sub-Saharan Africa?

American Journal of Economics and Sociology, The,  Nov, 2004  by Augustin Kwasi Fosu

• E-mail
• Print
• Link

Table 1
Regression Results: Elite Political Instability and Economic
Development in SSA
(absolute values of the t ratio in parentheses)

Eqn.    Const.      Y         P1        p2         y*p1

A. Ordinary Least Squares (OLS)
(A1)    0.062   0.012 (a)     --       --          --
               (5.61)
(A2.1)  0.079   0.011 (a)   -0.005     --          --
               (5.06)       (1.55)
(A2.2)  0.083   0.010 (a)     --    -0.011 (b)     --
               (5.14)               (2.39)
(A3.1)  0.073   0.013 (a)   -0.001     --       -0.0016
               (4.98)       (0.33)              (1.33)
(A3.2)  0.075   0.012 (a)     --    -0.003         --
               (5.33)               (0.45)
(A4.1)  0.070   0.013 (a)     --       --       -0.0019 (b)
               (6.29)                           (2.07)
(A4.2)  0.072   0.013 (a)     --       --          --
               (6.83)

B. Instrumental Variable Estimation (IVE)
(B1)    0.063   0.013 (a)     --       --          --
                (4.67)
(B2.1)  0.077   0.011 (a)   -0.005     --          --
                (4.00)      -1.33
(B2.2)  0.082   0.011 (a)     --     0.011 (b)     --
                (4.07)              (2.21)
(B3.1)  0.064   0.016 (a)    0.000     --       -0.0026 (c)
                (4.49)      (0.10)              (1.98)
(B3.2)  0.061   0.016 (a)     --     0.002         --
                (4.93)              (0.32)
(B4.1)  0.065   0.016 (a)     --       --       -0.0025 (b)
                (5.65)                          (2.47)
(B4.2)  0.065   0.016 (a)     --       --          --
                (6.43)

Eqn.       y*p2      Adj.[R.sup.2]   SEE     B

A. Ordinary Least Squares (OLS)
(A1)       --            0.521      0.032  0.821

(A2.1)     --            0.544      0.031  0.765

(A2.2)     --            0.593      0.030  1.24

(A3.1)     --            0.558      0.031  1.88

(A3.2)  -0.0028          0.615      0.029  2.05
        (1.59)
(A4.1)     --            0.572      0.030  2.06

(A4.2)  -0.0034 (a)      0.627      0.028  2.13
        (2.94)

B. Instrumental Variable Estimation (IVE)
(B1)       --            0.425      0.035  0.450

(B2.1)     --            0.441      0.035  0.678

(B2.2)     --            0.498      0.033  0.932

(B3.1)     --            0.497      0.033  1.55

(B3.2)  -0.0044 (b)      0.580      0.030  2.02
        (2.46)
(B4.1)     --            0.516      0.032  1.54

(B4.2)  -0.0040 (a)      0.594      0.029  1.96
        (3.49)

(a) Statistically significant at the 0.01 level (two-tailed).

(b) Statistically significant at the 0.05 level (two-tailed).

(c) Statistically significant at the 0.10 level (two-tailed).

The dependent variable is HD185-HD170, where HD185 and HD170 are the
values of the United Nations Human Development Index for 1985 and 1970,
respectively, both from united Nations, Human Development Report, 1991.
The independent variable y is the mean annual growth rate of GDP over
the 1967-1985 period, derived from World Bank, World Tables, 1988-89.
The variables p1 and p2 denote elite political instability; p1 is the
principal component of the incidents of successful coups d'etat (SC),
attempted but unsuccessful coups (AC), and coup plots (CP) over the
1967-1985 period (see text for details), using data from Patrick
McGowan, Intervention Event File 1986; p2 is SC only. SEE and
Adj.[R.sup.2] are the standard error of estimate and the adjusted
coefficient of determination, respectively. B is the Bartlett
likelihood ratio test statistic to test for heteroscedasticity; it is
distributed as chi-square with 2 degrees of freedom in the present
sample. The sample consists of 29 SSA countries. IVE results are
obtained by using predicted values of y, where labor force l,
investment k, exports x, pl, and a constant serve as the instrumental
variables (see text for details). Note that SEE and Adj.[R.sup.2] are
not comparable across OLS and IVE, since SEE and [R.sup.2] tend to be
overestimated and underestimated, respectively, under IVE. Hence, the t
ratios for the IVE are actually slightly more significant than are
indicated herein.

Table 2
Summary Statistics of Regression Variables and Partial Effects
and Elasticities: Economic Development Versus GDP Growth
and Elite Political Instability

    Mean   Std.Dev.  Effect *  Elasticity **

b   0.107   0.046       --          --
y   3.52    2.86      0.011        0.362
p1  1.88    1.74     -0.0088       0.155
p2  1.17    1.28     -0.0141       0.154

* Computed at the means of the variables, using Equations B4.1 and B4.2
of Table 1, respectively. The effect of y is calculated as 0.016
-0.0040(1.17), p1 as -0.0025(3.52), and p2 as -0.0040(3.52).

** Calculated as the absolute value of the respective estimated effect
times the ratio of the mean of the independent variable to the mean
of b.

Note: See Table 1 for the definitions and data sources of the variables.