Principal investigator: Viviane Pires Ribeiro
Article title: INFORMAL LABOR AND THE EFFICIENCY COST OF SOCIAL PROGRAMS: EVIDENCE FROM THE BRAZILIAN UNEMPLOYMENT INSURANCE PROGRAM
Article authors: François Gerard and Gustavo Gonzaga
Location of the intervention: Brazil
Sample size: 27 Brazilian states
Sector: Labor market
Type of InterventionCost-efficiency analysis of unemployment insurance.
Primary variable of interest: Unemployment insurance
Evaluation method: Others Canonical model
Evaluation Context
The informal sector, the part of an economy that escapes government monitoring, represents a larger share of employment in developing countries. In a context of high informality, conventional wisdom holds that social programs – transfer programs and social insurance programs – impose high efficiency costs, especially when they require beneficiaries to be informally unemployed. In this scenario, the concern is that these social programs and the availability of informal work opportunities will diminish incentives to work in the formal sector.
Despite this widespread view, the evidence behind it remains limited. Firstly, partly due to data constraints, few studies estimate the impact of social programs on employment choices in developing countries. Secondly, studies that conclude that social programs induce some beneficiaries not to work in the formal sector typically lack a theoretical framework to interpret this evidence in relevant efficiency-equity terms or tradeoff"between efficiency and safety."
Intervention Details
Gerard and Gonzaga (2018) study the efficiency cost of increases in unemployment insurance benefits in a context of high informality, combining an ideal unemployment insurance framework and empirical evidence for Brazilian unemployment insurance beneficiaries. The analysis is primarily based on two sets of administrative data. RAIS is a legally mandated employer-employee dataset of the universe of formal employees, including public employees. RAIS contains information on job security, age, gender, education, sector, size and location, reason for dismissal, and, since 2002, hiring and termination dates for each one-year work period. The main analysis uses data from 2005 to 2010.
Microdata from two surveys conducted by the Brazilian Institute of Geography and Statistics (IBGE) were used. Both surveys inquire about the labor market status of each household member over ten years of age, including information on salary, job ownership, and formal employment contracts. State-level representative household surveys (PNAD) were used annually to measure informality rates in the 27 Brazilian states. Monthly labor force surveys (PME) were also used to assess the importance of informal work opportunities for laid-off formal employees and to explore their need for unemployment insurance.
Brazil is an interesting empirical setting for several reasons. First, unemployment insurance has existed in Brazil for many years, so workers are aware of its associated incentives. Second, the longest potential duration of unemployment insurance in the country was five months. Third, Brazil is not an outlier in terms of informality: its average informality rate is close to the average in Latin American countries. Fourth, there is considerable heterogeneity in informality rates across Brazilian labor markets. Lastly, the authors had access to comprehensive administrative data, which, combined with quasi-experimental variation in unemployment insurance benefits, allowed them to estimate all the statistics that go into the efficiency cost measure.
Methodology Details
The methodology used by Gerard and Gonzaga (2018) was the canonical model of optimal unemployment insurance, which specifies the tradeoff between workers' needs for insurance and the efficiency cost of distorting their incentives to return to formal employment. The model was combined with evidence drawn from comprehensive administrative data to quantify the efficiency cost of increases in the potential duration of Brazilian unemployment insurance.
Using administrative data, the authors documented the patterns of insurance benefit collection and formal re-employment, which differ from those observed in developed countries. Subsequently, they sought to document how these patterns compare across Brazilian labor markets with different rates of informality. This allowed them to provide a cost-efficiency estimate that can be compared with estimates from countries with low informality.
Results
The results show that the average duration of unemployment insurance paid is high compared to the potential duration of the insurance, as most insurance claimants exhaust their benefits; for example, more than 80% of those eligible for five months of unemployment insurance do so. This is only about 35% in the US, where unemployment insurance is typically eligible for 24 weeks. This difference arises from the fact that the proportion of workers who find new formal employment is lower when they are eligible for unemployment insurance in Brazil. The proportion who find new formal employment increases soon after the insurance is exhausted, suggesting clear behavioral responses to unemployment insurance incentives. Despite the existence of behavioral responses, the efficiency cost of increasing unemployment insurance benefits may not be relatively high. Furthermore, laid-off workers may choose to work informally for reasons unrelated to the incentives of this type of insurance.
Estimates indicate that a one-month increase in the potential duration of unemployment insurance leads to a large increase in the average duration of insurance paid (0,86 months). Behavioral responses account for 14,6% of the increase in insurance duration. Thus, workers postpone formal re-employment by 0,39 months and reduce, on average, the time spent formally employed by 0,24 months.
Analysis of the 27 Brazilian states shows that the cost of efficiency is lower in states with higher levels of informality. Furthermore, the impacts of behavioral responses on insurance duration, time spent outside of formal employment, and time spent formally employed are all low (in absolute terms).
Finally, the authors resorted to the marginal value of insurance, given that the results suggest a shift in the political debate towards the actual need for insurance by workers. In this sense, an efficiency cost of $0,2 per $1 implies that the welfare effect of an increase in the duration of unemployment insurance would be positive if the average marginal utility of $1 were at least 20% higher for mechanical beneficiaries than for formal jobs.
Lessons in Public Policy
Gerard and Gonzaga (2018) study the efficiency cost of increasing unemployment insurance benefits in a context of high informality, combining an ideal unemployment insurance framework and empirical evidence for Brazilian unemployment insurance beneficiaries. Their findings contradict widespread claims in political circles that raise concerns about the problem of habitual moral hazard – that unemployment insurance distorts incentives to return to formal work – thus preventing the existence or expansion of unemployment insurance in this context. Therefore, the authors argue that the associated efficiency cost is not necessarily high. Indeed, the efficiency cost is low in Brazil when compared to countries with low informality.
Furthermore, the results have implications for other policies aimed at helping formal workers in developing countries. Unemployment insurance savings accounts are sometimes presented as an alternative to unemployment insurance in these countries. In this sense, there is also evidence that Brazilian workers are willing to negotiate low formal wages for mandatory benefits, including benefits related to job loss.
References
Gerard, F., & Gonzaga, G. (2016). Informal labor and the efficiency cost of social programs: evidence from the Brazilian unemployment insurance program (No. w22608). National Bureau of Economic Research.
