Principal investigator: Eduarda Miller de Figueiredo
Authors: François Gerard and Joana Naritomi
Location of the Intervention: São Paulo
Sample Size: 77.892 dismissal events
Sector: Labor market
Primary Variable of Interest: Consumer spending
Type of Intervention: Resignation
Methodology: Differences-in-Differences
Summary
After being laid off, workers are impacted by a negative and permanent shock to their income. However, at the same time, there is a positive transitory income shock with severance pay or unemployment insurance. This article aimed to observe a context in which the incentives to smooth consumption are particularly strong. Using data for São Paulo and applying the Difference-in-Differences model, changes in consumer spending for displaced workers are estimated for the 12 months before and after layoff. The estimations suggest that all groups of workers experienced a sharp and significant increase in expenses after dismissal.
- Policy Problem
Literature documents a high sensitivity of consumption to positive transitory income shocks (Shapiro, 2005; Olafsson and Pagel, 2018). Severance pay levels lead to higher consumption levels in the first months after dismissal. That is, displaced workers tend to have financial resource limitations; however, displaced workers can smooth consumption far beyond unemployment insurance if they save severance pay more slowly (Chetty, 2008; Ganong and Noel, 2019).
In this article, the authors observe a context in which the incentives to smooth consumption are particularly strong. They examine the spending behavior of individuals who receive a lump-sum severance payment, i.e., a transitory income shock, at the same time as they are laid off and therefore experience a permanent negative income shock. Standard economic models with forward-looking agents typically predict that a large portion of this fixed amount should be saved to smooth out the consumption losses resulting from the permanent negative shock.
- Implementation and Evaluation Context
The empirical analysis was conducted in the state of São Paulo, the largest state in Brazil, which has a population of 42 million and accounts for 34% of the country's GDP. Informal workers in the state of São Paulo represent approximately 35% of employment in the private sector and are therefore not covered by commuting insurance programs.
In Brazil, there are three sources of job displacement insurance benefits for workers who are involuntarily dismissed from formal jobs in the private sector.
- Unemployment Insurance (UI): Workers who are laid off and have at least six months of employment are eligible for unemployment insurance benefits after a 30-day waiting period. The benefit is paid monthly, and its duration depends on the worker's total length of employment in all formal jobs during the 36 months prior to dismissal: 3, 4, or 5 payments if they have more than 6, 12, or 24 months of employment, respectively. The benefit amount depends on the average salary in the three months prior to dismissal.
- The Guarantee Fund for Length of Service (FGTS): Employers must deposit 8% of their workers' monthly salary into an account in each worker's name at a national bank. This account can be accessed if the worker is involuntarily dismissed, and because the interest rate is low and the account is illiquid, workers have strong incentives to withdraw the total balance when dismissed.
- Severance pay: Employers must pay a monthly salary to relocated workers as "notice period" for dismissal. They must also pay a "penalty" to relocated workers of 40% of the amount deposited in the FGTS (Brazilian severance fund) during the employment period.
- Policy/Program Details
For the analysis, a combination of administrative datasets was used. Data identifying formal workers, as well as information on age, race, sex, education, sector, hours, and other factors, were collected from RAIS (Annual Social Information Report). Data including worker ID, date, and benefit amount for all payments were captured from SD (Social Security System) records. Using the Nota Fiscal Paulista (NFP) program, the authors were able to combine formal employment and SD data with administrative expense data.
Therefore, all formal, full-time private sector employees in São Paulo who were dismissed between 2011 and 2013 were selected. Expense and employment data were obtained for at least one year before and one year after the dismissal for all workers with expense data. A sample of 77.892 dismissal events was obtained, and a reference sample of 156.11 dismissal events was randomly selected from 5% of the overall RAIS sample.
- Method
A Difference-in-Differences model was used to estimate changes in consumer spending for displaced workers in the 12 months before and after dismissal, compared to a control group of employed workers over the entire 25-month period. The control group was constructed by creating worker-month pairs for “placebo” displacement events in the months between 2011 and 2013. The control group contains 220.160 displacement events.
The dependent variable is consumer spending for displaced workers in the months surrounding their dismissal compared to a reference month. Additionally, fixed effects for each worker event (dismissal) were added, and standard errors were grouped by worker.
- Main results
The results show that total expenses increase by 31,4% and 37,7% in the month of dismissal and the following month, respectively. Estimates decrease rapidly, but remain positive in months 2 and 3 after dismissal, as shown in Panel A of Figure 1.
Figure 1: Profile of expenses related to travel.
In months 4 and 5, the estimates are negative and more stable, and in months 6 and 8 there is again a faster decrease, which is exactly in the months when workers exhaust their unemployment insurance benefits.
Panel B in Figure 1 compares the Difference-in-Differences results for laid-off workers. versus Workers who were dismissed for just cause and therefore ineligible for unemployment benefits. Estimates after the shift (period 0) show that workers dismissed for just cause experience a drop in consumer spending, reaching 26% in month 1. The large difference in spending profiles indicates that unemployment benefits in job displacement manage to offset spending patterns in dismissal without just cause. Estimates for dismissed workers remain low in subsequent months and below estimates for workers dismissed without just cause, although the proportion of re-employed workers is higher among those dismissed for just cause than those dismissed without just cause, which is consistent with unemployment benefits disincentives to job seeking.
Furthermore, estimates show that all groups of workers experienced a sharp and significant increase in expenses after dismissal. The authors even found the same increase in spending for workers who were immediately reemployed, supporting the argument that increased spending upon dismissal is unlikely due to a complementarity between leisure and expenses. It was also observed that workers who were immediately reemployed are the only group that did not experience a long-term loss, and that spending levels increase in the month following re-employment. Thus, there was an increase in spending of approximately 5% among reemployed workers in months 0 to 10 and approximately 10% among those reemployed after the end of unemployment benefits.
However, the authors point out that unemployment insurance benefits require workers to remain unemployed, which would lead to less distortion in job search efforts, as also discussed in Feldstein and Altman (2007).
- Lessons in Public Policy
The findings of the article demonstrate that installment payments of unemployment insurance can be more important, compared to lump-sum severance payments, in smoothing out post-layoff consumption. Given that the goal of a program is to provide security to displaced workers, the results imply that a lump-sum payment (severance pay) could undermine that goal.
References
GERARD, François; NARITOMI, Joana. Job displacement insurance and (the lack of) consumption-smoothing. American Economic Review, v. 111, no. 3, p. 899-942, 2021.
