Fig. 1.
The different DiD designs.
Note: The figure depicts the different transition and control groups and their respective numbers of observations. The sizes of control groups I.2 and II.2 (people who remain subsidized employed) slightly vary since the different sets of control variables cause missing values to a different extent. For our consistency checks on direct transitions between unemployment and regular employment, we rely on 589 transitions from regular employment to unemployment (control group of 22,696 regularly employed people) and on 1,032 observations of unemployed people who become regularly employed (control group of 12,024 unemployed people).