where i I = G T represents the group of individual who are actually affected by the minimium
wage increase. Assume Yi satisfies
where b represents the time effect that is invariant of group variable Gi
, and g is the group effect which is the invariant of the time variable Ti
. The DID estimator can be written as:
Equation (3) indicates that without a minimum wage increase, changes in employment
and working hours between the control and the experiment group should be the same.
However, a minimum wage increase induces the experiment group to deviate from this
common trend. Therefore, to estimate the treatment effect, we need to subtract the population
average differences of the control group from those of the experiment group.
Assume that a minimum wage increase has the same effect on every observation (i.e.
1 0 Y Y i i - =t ), and also consider other factors that may affect the outcome variable. We can
obtain:
Yi Ti Gi i i i =a + b +g +t I X + + d e . (4)
Equation (4) is a regression-adjusted DID model (Angrist and Pischke, 2009).