“…The task was to evaluate empirically whether the JA provisions have causally affected firms' employment behavior. According to the approaches and notation of Blundell and Costas (2002), Bandik andKarpaty (2011), andZeli (2018), we adopted a two-stage strategy: we constructed a sample of matched beneficiary and non-beneficiary firms, then we estimated the DiD coefficient for this matched sample. Let TC β{0,1} be an indicator of whether firm i is treated (i.e., the firm is exploiting JA social security relief or the possibility of easily dismissing newly hired workers) in time period t, and let π¦ π,π‘+π 1 be the employment at time t+s (s>0 after the first deduction year).…”