Fabiano Dal-Ri is a postdoc at the Dyson School of Applied Economics and Management at Cornell University. He completed his Ph.D. in Business Economics at Insper in 2024.
Why Job Loss and Self-Employment Are Relevant Topics
Job loss can have a very negative impact on workers’ career trajectories. It is well-established, for example, that dismissed workers often face persistently lower wages when they find a new job – and that’s if they find a new job. After being fired, workers are more likely to leave the workforce altogether.
Given such a scenario, it is not surprising that job loss could lead workers to explore alternative career paths. In my paper, I focus on a particular trajectory that has become more relevant in recent years in both emerging and rich countries: self-employment.
Brazil is an exciting case to study self-employment because of a firm registration reform in 2009 that greatly reduced costs associated with opening and operating a self-employment business. The firms created under this new format are known as MEI firms (MEI stands for “microentrepreneur individuals” in Portuguese). The most direct effect was a massive increase in firm registration in Brazil, jumping from around 500,000 to more than 2 million new firms each year. While there is some evidence that this might reflect a formalization of businesses that would have been informal had the reform not taken place, we nevertheless observe many more formal firms, and, importantly for my work, we can now follow these self-employed workers in the data.
The Consequences of Job Loss
In my paper, I first aim to estimate the causal impact of job loss on self-employment decisions. Combining administrative data on the universe of formal job contracts in Brazil with identified data on business ownership, I construct a panel that allows me to track individuals through transitions to different employment statuses. To isolate the effect of job loss, I focus on workers fired during a mass displacement event, which I define as taking place when establishments lay off more than 30 percent of their workforce in a given year. To ensure an appropriate comparison group to these mass-displaced workers, I match them with similar non-displaced workers.
Using an event study design, I report that, for workers displaced in 2007 and 2008 (before the 2009 policy change), there is a 0.31 percentage-point increase in the probability that they transition to formal self-employment in the year of job separation. This represents a three-fold increase when compared to the pre-displacement probability. However, a striking difference emerges from comparing high- and low-income workers. While the displacement effect is larger for high-income workers (with a 0.79 percentage-point increase in self-employment probability, also a three-fold jump), I do not find any effect for low-income workers. The formal self-employment path was closed for them.
Jump to 2012 and 2013, however, and the scenario is drastically different. The displacement effect is now twice as large (an increase of 0.67 percentage points in the year after the displacement). While high-income workers continue to be more affected (1.54 percentage points), low-income workers arguably benefit the most. Due to the 2009 reform, formal self-employment is now a viable path for them, and I report a 0.57 percentage-point displacement effect for these workers.
Motivated by the finding that the 2009 policy change had heterogeneous impacts across the wage distribution, I proceed to understand the differential consequences of these formal self-employment spells. Now focusing on displaced workers only, I first show that, before and after 2009, workers who transition to formal self-employment following a job loss are less likely to return to wage employment. If they do, this will take longer to happen. A similar pattern is observed for both low- and high-income workers.
However, the conditions under which they return to wage employment are very different. I reveal that high-income workers who take the self-employment path face wages 12 percent smaller. In contrast, there is no effect on reemployment wages for low-income workers. Interestingly, the 2009 reform does not appear to alter these reemployment patterns.
In the paper, I also show three additional results. I reveal that minimum wage regulations place an upper bound on the extent to which the wages of low-income workers can be reduced following a job loss. I also provide suggestive evidence that the negative effect of MEI ownership for high-income workers appears to be associated with individuals who, based on observed covariates, we would typically expect to be successful self-employed workers – that is, those with higher education, managerial experience, and access to a substantial lump-sum severance pay. Lastly, my analysis reveals that, on average, high-income workers who open a MEI business reentry wage employment in low-skill occupations.
These results are consistent with three possible scenarios:
- Spending time in self-employment endeavors (particularly if unsuccessful and not aligned with previous experience) could result in losses in human capital.
- Short-lived self-employment spells might be interpreted as a negative signal for workers returning to wage employment.
- High-income MEI owners are potentially negatively selected on unobservable characteristics such as ability. If that’s the case, when they reentry wage employment they will earn less (since we expect wages to be correlated with ability).
So far, my paper underscores how a policy change reducing firm registration costs impacts individuals’ labor market decisions following a job loss. It reveals that, following the reduction in business registration costs, displaced Brazillian workers responded by transitioning to self-employment more often. In this dimension, there appear to be few negative effects for low-income workers, the primary demographic targeted by the reform. However, to fully understand the implications of this reform, it is crucial to disentangle the potential explanations behind the observed negative effects of self-employment spells on high-income workers. This will be the next phase of my research.
Interview with the author
Q: How did you come up with the idea for your paper?
I’ve always been interested in understanding why individuals decide to become self-employed workers or entrepreneurs, as opening a business is a big career change for many of them. In addition, in Brazil, we have a lot of small businesses closing within a short period after starting to operate. So it is arguably a risky choice for many individuals, and yet many of them pursue this path. There was also a firm registration reform in Brazil in 2009 (which I explore in the paper), and seeing many people decide to open new businesses after that provoked my interest. I must also mention that I’ve been working on another project aiming to understand the consequences of job loss in Brazil, and exploring self-employment decisions in this context was a natural extension, given my interest in the topic.
Q: What was the greatest challenge you faced when working on this paper?
The implications of job loss and self-employment decisions are numerous, making it difficult to pinpoint exactly which research question I wanted to explore first. It was only after conducting a more extensive literature review and some very helpful discussions with colleagues and professors that I settled on the idea of trying to understand displacement-driven self-employment decisions. Presenting the paper to different audiences also helped me think more about this project and how it could contribute to the existing literature.
Q: Do you have any advice for students who want to use a rich dataset like the ones you used for this paper?
I think it’s crucial to spend some time exploring these datasets, “playing” with the data. For example, tabulating each variable, checking raw correlations between some of them, and verifying whether there are missing observations. Sometimes, we lay our hands on these exciting datasets and jump directly into trying to answer our questions. However, it’s also important to understand in-depth the structure and the shortcomings of these datasets. It will help avoid mistakes and perhaps bring new ideas also worth researching.
Feature photo by Jonathan Weiss on Unsplash.
