Log of Real Minimum Wage
(1)
Fraction Affected (2)
Coef P value Coef P value
Ln Real W ages -1.0954 0.000 -1.3028 0.000 Urban -0.6177 0.023 -1.3121 0.000 Youth 0.1079 0.890 -0.2380 0.745 Women -0.5762 0.392 -0.5945 0.444 Industry -0.1779 0.807 -0.3949 0.581 Trade -1.0036 0.140 -0.8105 0.254 Services 2.8741 0.000 1.5240 0.030 Construction 4.3107 0.004 2.0987 0.153 High school 0.1123 0.816 1.1585 0.014 University 3.3257 0.027 9.5215 0.000 Year 1991 0.0776 0.111 -0.4211 0.000 Year 1992 0.1676 0.003 0.1167 0.053 Year 1993 0.3346 0.000 0.1149 0.059 Year 1994 0.3607 0.001 0.2001 0.003 Year 1995 0.4621 0.000 0.0568 0.364 Year 1996 0.6174 0.000 0.2061 0.007 Year 1997 0.7530 0.000 0.0379 0.621 Year 1998 0.5110 0.000 0.1621 0.009 Year 1999 0.5866 0.000 -0.3131 0.001 Year 2000 0.7972 0.000 0.0877 0.142 Year 2001 0.8765 0.000 0.2224 0.001 Year 2002 0.8322 0.000 0.1392 0.067 Year 2003 0.8170 0.000 -0.0293 0.613 Constant 25.1213 0.000 -0.0168 0.699 Observations 364 338 Number of Areas 26 26 Within group R-sq 0.489 0.336
Note: All regressions include province dummies.
Com pared to the m ethods in the previous section, the structural em ploym ent demand
equation provides evidence that the m inim um wage affects em ploym ent indirectly
through the w age rate, w hich is consistent w ith the standard theory. As m entioned
above, standard theory suggests that m inim um wage affects em ploym ent if there is a
significant effect on the wage paid to the worker. Therefore, we argue that this
specification is m ore com plete and relevant than the m odel in the previous section
using separate reduced form equations in explaining the effect o f m inim um wage on
em ploym ent. In addition, this m ethod does not suffer from the endogeneity problem
and contem poraneous correlation across equations in separate reduced form equation.
A s a result, we argue that the structural em ploym ent dem and equation is the m ore
preferred specification, showing a sim ple interaction betw een w age and em ploym ent
equations.
5.6. Conclusions
This chapter analyses the effects o f changes in the m inim um wage on wages and
em ploym ent in Indonesia using a provincial panel data set from 1989 to 2003 across 26
provinces. Com pared to the other developing country studies, the m inim um wage
policy in Indonesia provides a unique characteristic as the m inim um wage is set
differently across regions (provinces). Follow ing Lemos (2004d and 2005), this study
has applied five different m easures o f m inim um wage using four standard panel data
estim ates, including the sim ple fixed effect, Panel Corrected Standard Error (PCSE),
instrum ental variable and Dynam ic Arellano Bond estimates.
A lthough the findings vary across different m easures and m ethods, generally they
support the standard com petitive m odel, suggesting that an increase in the m inim um
wage increases average wages and causes a reduction in covered sector em ployment.
U nlike previous studies, the result shows that there is no significant impact o f the
m inim um wage on total paid em ploym ent because o f the non-com pliance problem in
Indonesia. This result implies that total paid em ploym ent is not a valid m easure for the
em ploym ent effect o f m inim um wage in Indonesia. Standard com petitive m arket model
alw ays assum ed that the m inim um wage com pliance is high and this is not the case
w ith Indonesia w here there is a high proportion o f uncovered sector em ployment. This
em ploym ent effect o f m inim um wage in Indonesia. In practice, the coefficients change
dram atically when paid em ploym ent in the covered sector is used as the dependent
variable.
Table 5.21 the Effects of Minimum Wage across Measures and Estimates
Log of Real Minimum Wage (1) T ou gh n ess (2) Fraction Affected (3) Fraction Below (4) Fraction At (5) P an el A: W age Equation
Sim ple Fixed 0.1656 - 0.2935 -0.1599 -0.9075
Effects (0.00) (0.021) (0.087) (.019) PCSE 0.1669 - 0.2881 -0.1766 -0.8040 (0.00) (0.009) (0.098) (0.042) Instrumental 0.2613 - - 0.1648 -4.839 Variable (0.002) (0.522) (0.209) Dynamic A-B 0.1221 - 0.099 - - (0.008) (0.071)
P an el B: E m p loym ent Equation (C overed Sector)
Simple Fixed -0.4597 -0.8349 -1.0463 -1.3057 -1.0175 Effects (0.000) (0.000) (0.000) (0.000) (0.081) PCSE -0.4175 -0.6851 -0.9590 -1.2063 - (0.000) (0.000) (0.000) (0.000) Instrumental - -1.5660 -6.2963 -1.7307 -11.8607 Variable (0.000) (0.251) (0.000) (0.091) Dynamic A-B -0.3010 -0.6052 - -1.0445 - (0.016) (0.000) (0.000)
Table 5.21 presents a com parison table sum m arizing the basic results o f wage and
em ploym ent equations across m easures and estim ates which were analyzed above. In
practice, the reduced form s o f wage and em ploym ent are estim ated separately
(independently) in explaining the effects o f m inim um wage on wage and em ployment.
Com paring m inim um wage m easures, the findings suggest that the log o f real
m inim um wage is superior to the other m easures. The log o f real m inim um wage
consistently shows a significant impact across different estim ates, providing well-
suited m easurem ent particularly w hen the m inim um wage varies across regions.
A lthough the log o f real m inim um wage is endogenous in the wage equation, it is
exogenously determ ined with em ploym ent, providing a “cleaner” m easure o f the
m inim um wage effect on em ployment. Toughness suffers from the endogeneity bias
m easures are endogenously determ ined w ith em ploym ent because they m easure in a
sim ilar w ay with em ploym ent in the left-hand-side equation, providing a strongly
sim ultaneous correlation with em ployment. U sing the log o f real m inim um wage, it is
suggested that an increase o f the m inim um wage by 10% increases the average wage o f
paid em ploym ent in the covered sector by 1.22% -2.61% and decreases covered sector
em ploym ent by 3.01% -4.17% , depending on the m ethod used.
In com parison across the basic m odel, the D ynam ic A rellano-B ond m ethod tends to
provide a m ore efficient result using wage and em ploym ent equation as it m easures the
adjustm ent o f em ploym ent to changes in labour dem and and supply. In contrast, the
PC SE and sim ple fixed effect estimates tend to suffer from the endogeneity problem as
they assum e exogeneity across variables.
In order to check the robustness o f the result obtained by changing specification, the
effect o f the m inim um wage on wages and em ploym ent is also estim ated using
different specifications, including the Seem ingly U nrelated Regression (SUR) and
structural em ploym ent dem and equation. SUR is estim ated due to the fact that the error
term s across w age and em ploym ent equations are likely to be correlated. However, the
SU R estim ate does not control for the endogeneity problem in the m inim um wage
m easures, suggesting that the results are potentially m eaningless.
On the other hand, the structural em ploym ent dem and equation estimate shows in a
different way that the m inim um wage affects em ploym ent indirectly through the wage
rate, show ing a sim ple correlation betw een wage and em ploym ent equations as
endogeneity problem and contem poraneous correlation across equations, providing a
m ore efficient estim ate than separate reduced form equations. As a result, we conclude
that the structural em ploym ent dem and equation is the m ore preferred specification
com pared to the previous estim ates using separate reduced form wage and em ploym ent
equations. In addition, although the specification differs from the previous estimate,
the finding supports the previous evidence using separate reduced form equations that
the m inim um wage is positively associated w ith average wages and negatively
associated w ith em ploym ent. The coefficient o f m inim um wage is also similar,
suggesting that the findings are strong across different specifications.
Thus far, this chapter showed that an increase in m inim um wage reduces paid
em ploym ent in the covered sector (as part o f the em ploym ent in the formal sector).
This chapter therefore provided a basic analysis o f the im pact o f m inim um wage on
w ages and em ploym ent in Indonesia. A further area o f research revealed by these
findings is how the m inim um wage policy affects em ploym ent in the uncovered sector.
The effect o f m inim um wage on em ploym ent in the uncovered sector shows a
displacem ent effect that m ight exist as the result o f decreasing paid em ploym ent (the
covered sector). The full effect o f m inim um wage on em ploym ent in the covered and
uncovered sectors w ill be discussed in the next chapter.