Book Description

The impact of the tax and benefit system on work incentives is a salient issue in labor and public economics. There is, however, relatively little analysis of the joint work and tax evasion incentive effects introduced by social insurance programs. This paper evaluates the behavioral responses of workers in these dimensions. Using a quasi-experimental approach, it studies a large scale expansion of an employment based benefit in the social insurance system of a middle income country, Uruguay. The policy change extended the coverage of an in-work and payroll tax financed health insurance program to the dependent children of private sector salaried workers. The extension only applied to full-time registered employees - those complying with payroll tax and social insurance contribution requirements. The results indicate that those who benefited from the reform responded to these financial incentives as predicted by economic theory, significantly increasing their labor force participation and hours of work, with most of this increase in registered (or “on”-the-books) employment. The reform only required one registered employee within the household to warrant the extended coverage, and the results point to a disincentive effect for secondary workers, with reduced labor supply in the form of registered employment. Besides the reduction in off-the-books employment (a fall in outright tax evasion), the analysis uncovers an additional adjustment along a further dimension of evasion. Exploiting an original feature of the data, the results indicate that the reform induced an increase in underreporting of salaried earnings for registered employees. These results, driven mainly by workers in small firms, suggest some degree of collusion between employers and employees to deceive the tax authority, with workers receiving the additional benefit introduced by the reform without incurring the full cost of the higher tax liability. The conclusion illustrates how these additional margins of adjustment to tax and benefits, not contemplated by the canonical model, can have relevant consequences for policy and for the design of social insurance programs.