William Davidson Institute Working Paper 479
The measures differ in their sensitivity to income variations at different levels of the
distribution. For equi-distant transfers, the Gini index is considered to be more sensitive to
changes around the mode, while the coefficient of variation is more sensitive to transfers at the
top of the distribution. The mean logarithmic deviation is relatively more responsive to changes
at the lower end of the distribution.
To test whether the level of the minimum wage and changes in the minimum wage affect
wage dispersion in the covered and the uncovered sectors, we estimate the following equation
separately for the covered and uncovered sectors:
6 12
InDit = αο + αι (MWi/W) +∑βiIi + ∑γt Tt + μ (4)
i=1 t=1
where the subscripts i = 1...7 are for each industry,5 and t =1...13 for years (1980-1992) and:
• D= CV,G,T is a measure of hourly wage dispersion within an industry and sector in each of
the 13 years.
• ΜW∕W is the lowest average hourly minimum wage set by the government for each industry
divided by the average hourly wage in each industry. This variable captures the “toughness”
of the minimum wage.
• I are industry specific dummies, with domestics as the base.
• T are annual dummies for 1980-1992.
Industry and time dummies are added to control for industry and time fixed effects, such
as changes in aggregate output and shocks over time. We also test whether the impact of the
minimum wage is non-linear (with a quadratic specification) and whether the minimum wage has
a lagged effect, using a one-year lag.