Do payroll tax subsidies reduce undeclared work? Evidence from Korea

Using administrative panel data sets on over 300,000 …rms in Korea, we examine the e¤ects of subsidizing social security contributions on the number of workers covered by the social security. Exploiting variations in location and size of …rms which a¤ected eligibility for the program, we implement a di¤erence-in-di¤erences (DD) and a di¤erence-in-di¤erence-in-di¤erences (DDD) estimation in order to identify the e¤ect of subsidies. Our DD estimates show that the subsidy program increased the number of covered workers by 1.34 percent while the DDD estimates indicate an increase in the number of covered workers by 0.71 percent. These estimates are smaller in size than those reported in previous studies implying massive …scal drains from the subsidy program. The minimal e¤ects and large deadweight losses of the subsidy scheme may be explained by the design features of the subsidy program in Korea.