Using a method similar to that used by Dickens et al (1999), this section examines the
effect o f minimum wage on wages throughout the distribution. Specifically, the
average monthly wage as dependent variable is divided into each percentile (decile) in
the wage distribution, while the independent variables follow the previous section
estimates. In other words, the effect of minimum wage on wages is estimated at each
decile in the wage distribution. Besides showing the full effect o f minimum wage
throughout the wage distribution, this investigation is also useful to check the
effectiveness o f different minimum wage measures relating to the wage effects. This
section continues to compare the minimum wage measures using the dynamic
Arellano-Bond method, given the biased estimate o f the simple fixed effect and the
PCSE methods findings.
Table 5.10 Effect on Wages Throughout the Distribution using Dynamic Arellano-Bond Estimates Log o f Real M in im u m W a g e (1) D ynam ic A -B Fraction A ffected (2) D y n a m ic A -B Fraction A t (3) D ynam ic A-B Fraction B e lo w (4) D ynam ic A-B
Deciles C o ef P value C o ef P value C o ef P value C o ef P value
1st 0.014 0.863 0.205 0.038 -0.282 0.690 -0.947 0.000 2^d 0.063 0.291 0.260 0.001 -0.933 0.023 -0.767 0.000 3rd 0.051 0.349 0.372 0.000 -1.453 0.000 -0.693 0.000 4th 0.066 0.172 0.246 0.000 -1.438 0.000 -0.578 0.000 5th 0.115 0.008 0.194 0.000 -1.126 0.000 -0.395 0.000 6th 0.101 0.009 0.123 0.011 -1.226 0.000 -0.288 0.000 7th 0.079 0.013 0.087 0.029 -1.100 0.000 -0.156 0.020 8th 0.098 0.001 0.138 0.000 -0.864 0.007 -0.084 0.187 9th 0.047 0.191 0.116 0.017 -0.758 0.048 -0.081 0.283 Observation Number of Areas 338 26 338 26 338 26 338 26
Note: D ependent variables are measured as the natural logs o f the real monthly average wage at each decile point o f distribution. Dynam ic Arellano-Bond estim ate results include year dum m ies and are estimated using Arellano-B ond one-step GM M estimator.
In the first column o f table 5.10, the effect of minimum wage on wage throughout the
distribution is estimated using the log o f real minimum wage as the minimum wage
measure. Similar to the previous section, the wage equation is instrumented using the
one year lagged value o f the log o f real minimum wage in order to control for the
endogeneity problem. The result shows that there is no significant effect o f minimum
wage at the lowest point o f wage distribution. The main reason is the fact that there is
a non-compliance issue with the minimum wage policy in Indonesia, which is likely to
be found at the lowest part o f the wage distribution. Specifically, the log o f real
minimum wage shows positive and significant effects only from the 5th to the 8th
deciles o f the distribution, while at the bottom and upper levels o f the distribution the
effects o f minimum wage are not significantly different from zero. The result
potentially suggests that the maximum impact is on the part o f the distribution
occupied by the minimum wage itself. In addition, the result also suggests the
presence o f spillover effect, where the minimum wage effect is strong up to the 8th
decile. As pointed out by Card and Krueger (1995), the spillover effect is defined as
an increase in the wages of workers who are already paid above the minimum wage
level before the minimum wage increases. This effect is present due to the fact that the
wage o f workers who are already paid above the minimum wage level should not be
the same level or lower than the minimum wage workers’ wages. Compared to the
other developing countries, Gindling and Terrell (2007) found significant effects
throughout the distribution using the log o f real minimum wage measure, also
suggesting the presence o f spillover effect, but the largest effect was in the 3rd decile
Moreover, my result shows that the fraction affected has significant and positive
effects on wages throughout the distribution, representing an extensive spillover
effect. Consistent with the log o f real minimum wage estimate, the effect o f minimum
wage is not strongest at the bottom o f the distribution, suggesting that minimum wage
is not effective in helping the lowest part o f the wage distribution. Specifically, this
result indicates an inverse U-shaped relationship o f distribution where the 3rd decile is
the highest point of the wage effects, confirming the part o f the distribution mostly
affected by the minimum wage policy. In comparison, the effect o f minimum wage on
average wage at the 30th percentile (0.372) is about three times bigger than at the 9th
decile (0.116) These results are consistent with the view that the main goal o f the
Indonesian minimum wage seems to be an instrument for raising the standard of living
o f workers, suggesting an extensive “spillover effect” o f the minimum wage
(Manning, 2003a). Comparing this result to the similar fraction affected measure in
the Colombian case, Arango and Pachon (2004) found a positive effect on the family
income above the 25th percentile o f family earnings distribution and between the 45th
and 60th percentiles o f the individual earnings distribution.
In the next two columns, the effects o f minimum wage on wages throughout the
distribution are estimated using the fraction at and the fraction below measures.
Consistent with the previous section estimates, the fraction at and the fraction below
shows irrelevant negative effects o f minimum wage on wage throughout the
distribution. Once again, these results confirm that the fraction at and the fraction
below are poor measures o f minimum wage effects on wages in the case o f Indonesia.
affected measures are generally more effective in estimating the effects o f minimum
wage on wage throughout the distribution.