minimum wage can only affect the experiment group, and the control group is used to
capture changes in employment and working hours without aminimum wage. The experiment
group includes provincesthat had at least one minimum wage increase during the specified
period, while the control group includes provinces that did not increase their minimum
wage during the sample period. Each observation i has a group dummy {0,1} Gi Î (1 for
the experiment group and 0 for the control group) and a time dummy {0,1} i
T Î (1 for after
the experiment period and 0 for before the experiment period). For a random sample from the
population, Gi
and Ti
can be considered random variables. Let Yi
denote the outcome
variable, which represents employment status and working hoursin thisstudy. The observed
data for i is( , , ) Yi G Ti i . Using the potential outcome framework proposed by Rubin (1974),
we let 0
Yi
denote the outcome of observation i without treatment (i.e. with no minimum
wage increase), and let 1
Yi
denote the outcome under treatment. The observed outcome Yi
can be obtained as follows: