Principal investigator: Bruno Benevit
Original title: The Incidence of Payroll Taxation: Evidence from Chile
Author Jonathan Gruber
Location of the Intervention: Chile
Sample Size: 71.000 Companies
Sector: Job market
Primary Variable of Interest: Salaries and Jobs
Type of Intervention: Tax reduction
Methodology: OLS, Triple Difference and IV
Summary
Payroll taxation is a large and growing source of public funding in the United States and the rest of the world. However, this type of revenue is linked to firms' payroll costs and may lead to a reduction in employment levels. The objective of this article was to evaluate how the reduction in payroll taxes resulting from the privatization of the Chilean social security system in the 1980s impacted wages and employment. Using OLS, Triple Difference, and instrumental variables methodologies, the author identifies that the incidence of payroll taxation fell entirely on wages. No effects on employment levels were identified. Several robustness strategies corroborate these findings.
- Policy Problem
The increasing tax burden around the world throughout the last century has been widely criticized for several reasons. Critics argue that increases in payroll taxes lead to higher labor costs, decreasing the competitiveness of a country's producers and potentially causing underemployment. This explanation was corroborated by the persistently high level of unemployment observed in Europe after the continuous increase in payroll tax rates since 1960. However, the costs of such tax increases will only affect employment if it is not possible to pass them on through wage reductions. In other words, if there is freedom to transfer tax costs to wages, there will be no resulting unemployment. In this sense, measuring the incidence of payroll taxes is essential to assess the impacts of this form of tax collection.
- Implementation and Evaluation Context
Payroll taxes accounted for 12,4% of US federal revenues in 1960, reaching a rate of 38% in 1993 (Gruber, 1997). The same pattern was observed in other nations of the Organisation for Economic Co-operation and Development (OECD) during the period 1965-88, rising from 19% to 25% of national tax revenues.
However, there was a drastic change in payroll taxation in Chile with the privatization of its Social Security and Disability Insurance programs in May 1981, which until then had been financed by a substantial tax on workers' payrolls. In 1980, the average payroll tax rate for manufacturing companies was 30%, while the average rate for workers was 12%. There was a maximum level of taxable income. In 1980, this maximum was approximately 528.000 pesos annually, corresponding to more than 5 times the average earnings of blue-collar workers in my study and more than twice the average earnings of white-collar workers observed in the sample of this study.
The Chilean Social Security system began in 1924, evolving in the 1970s into a standard pay-as-you-go defined benefit plan. Social Security “contributions” were paid to “Social Security Institutions” (IPSs) in Chile. The main IPSs were intended for white-collar, blue-collar, and public sector worker groups. Although other IPSs existed for other worker groups, only 5% of private sector workers were covered by them in 1980. The contributions collected by the IPSs funded old-age pensions. In addition, contributions also funded disability and maternity benefits, family allowances (annual benefits paid for each child), unemployment insurance, and workers' compensation. These benefits were funded by payroll taxes levied on companies and workers.
- Policy/Program Details
With the privatization of Social Security, individuals became responsible for their own retirement, instead of it being used to finance the consumption of current retirees. Individual retirement savings were invested in one of several competing private pension funds, and the savings could be converted into a lump sum payment or an annuity after retirement. Privatization also brought about major changes in the financing of social security. Privatized pensions and disability insurance were financed by a mandatory 13% contribution from the earnings of both blue-collar and white-collar workers, with no contribution from employers. The system maintained a minimum benefit, financed by general revenues.
Participation in this new system was optional for existing workers and mandatory for new entrants to the workforce. Among existing workers, 85% to 90% joined the new system. After privatization, there was a drop in the average payroll tax rate for manufacturing companies to 8,5% in 1982.
- Method
To assess the effects of the privatization of Chile's Social Security programs on wages and employment, this study uses data from the Chilean business census for the period 1979 to 1986. All companies with more than 10 employees were considered. The data includes information on employment, wages, and payroll taxes paid by the firm for both white-collar and blue-collar categories. All values were deflated using sector-specific price indices.
This study analyzes the effect of taxation on a given company's salaries before and after a policy change, based on a change in its average tax rate. The outcome variables analyzed relate to salaries and employment of worker groups. The explanatory variable related to the payroll tax rate was constructed using total tax payments and total salaries per worker category, considering each firm and year.
Due to data limitations, two empirical strategies were employed to circumvent the problem of spurious variation in the tax rate. The first considers variations in relation to the firm, the worker group, the year, and the interactions between these three variables. The second strategy involves the use of instrumental variables (IV) correlated with the actual tax rate and independent of the error terms of the outcome variable. The objective of this IV is to control for the possible bias caused by heterogeneity in the reduction of rates among firms and groups.
This article presents an analysis of temporal trends in contribution rates and wages from 1979 to 1986. To control for the effects of the recession experienced by the country in the early 80s, the analysis considers the periods 1979-80 and 1984-85. The analysis of the impacts on wages and employment used four models considering 3.305 firms per year. The first model estimated, using the OLS (One-Size Logical Scale), the first average difference between wages, employment, and rates. The second employed the triple difference, where the change in wages is identified by the relative change in wages within a factory for white-collar and blue-collar workers as a function of the relative changes in their contribution rates. The third model used a contribution rate innovation (VRI) for each group of workers (white-collar or blue-collar) based on the rate of the other group. Finally, the fourth estimated model used a contribution rate innovation created from a set of 13 dummies of regional areas. With the exception of the triple difference model, all models present disaggregated estimates with respect to worker groups.
- Main results
The results regarding temporal trends indicate that average real wages per worker increased by 27% for blue-collar workers and 29% for white-collar workers. This amount is considerably higher than the nominal wage increase of 18% (9% real) established for the year. From 1982 onwards, a wage reduction was observed. However, the author highlights that these results may be associated with the economic recession that caused a 14,1% drop in Chilean GDP. Regarding contributions, constant declines were observed until 1985.
The results regarding the effect of the reduction in the contribution rate on workers' wages demonstrate strong evidence of the transmission of the tax amount to workers' wages and no decrease in employment. In the OLS and triple difference models, the results suggest a full transfer of taxes to wages. Observing the disaggregated analyses of the first model, it was identified that the transfer mechanism was 50% higher for white-collar workers, indicating that white-collar workers may perceive stronger links between tax payments and social benefits.
The results of the models that employed IVs presented coefficients distinct from the first analysis. The results of the IV model that simulated contribution rates through group rates maintain the same inference identified in the previous results, although with larger magnitudes for white-collar workers, reinforcing the interpretation of "overvaluation" of social benefits by this group. The results of the second IV model, however, indicate the complete transfer of contribution rates only to the wages of blue-collar workers.
- Lessons in Public Policy
In this article, the authors analyzed the impacts of the privatization of Social Security and Disability Insurance programs in Chile in the early 1980s. Through various methodologies, the authors sought to mitigate the problems associated with spurious variations in error terms. The evidence found indicates that the reduction in company costs resulting from the reduction in payroll taxation was fully passed on to workers through wage increases. Furthermore, no significant effects on employment levels were identified. Analyses based on inverse variables suggest that possible measurement problems at the firm level were not relevant to the estimates found.
The authors, however, highlight the limitations of such results and their broader interpretations, emphasizing the need for greater attention to the structural sources of wage and employment changes in countries. The perception of social benefits, the elasticity of labor supply and demand, and the level of inflation in the economy are fundamental aspects for evaluating the impact of this type of policy.
References
Gruber, J. (1997), “The Incidence of Payroll Taxation: Evidence from Chile”, Journal of Labor Economics, Vol. 15 No. S3, pp. S72–S101.
