Difference-in-Differences (DiD) model will be applied to the Lithuania‘s case in order to evaluate the effect of policy change of the increasing minimum wage on employment. The treatment group in the model will be formed of workers earning less than the newly increased minimum wage at the time of introducing a higher minimum wage and, following this, the control group will consist of workers earning more.
I will use individual-level data from the Lithuanian Labour Force Survey. I plan to analyse the period 2013-2018.
The quarter before the minimum wage increase will be my treatment period and then control period will be three quarters after. For example, if the increase happens in January 2013, the 4th quarter of 2012 will be my treatment period and the 3rd quarter of 2013 will be the control period. However, minimum wage increases in Lithuania were not regular, so I face the issue of combining the dataset (see the dates below). In many other studies of specific countries, the increases are yearly and regular so dataset is smooth. In Lithuania's case the increases are not smooth, we have years without increases (2017), we have years with two increases (2016) and we have a long period between raises (i.e. January 2013 and then September 2014).
Q: Is it appropriate to combine the data for all years (since 2013) using the logic that for each increase I take one quarter before its increase (treatment period) and then my control period will be after 3 quarters of certain increase? Is the method correct?
Dates of increases: 2018 January; 2017 - no increase; 2016 January & July (increase was twice in 2016); 2015 July; 2014 September; 2013 January; 2012 August; 2011 - no increase.
