The optimal design of redistributive and social insurance policies crucially depends on the nature of income inequality and risk across individuals in the economy. Therefore, a rapidly growing literature has been concerned with the measurement of income inequality and dynamics in the US and other high-income countries (Shorrocks, 1978, Katz and Autor, 1999, Kopczuk et al., 2010, Guvenen et al., 2014, 2019, Nakajima and Smirnyagin, 2019, Busch et al., 2020). However, less is known about these topics in a developing country context. This is despite the fact that redistributive and social insurance policies might be especially valuable in developing countries, where the average income level is lower and a larger share of the population works in the informal economy at near-subsistence levels of income (McCraig and Pavcnik, 2015, Dix-Carneiro et al., 2021).
To address this gap, in a new paper (Engbom et al. 2021) we combine rich administrative and household survey data in order to document a series of new facts on earnings inequality and dynamics in a developing country with a large informal sector, namely, Brazil. Among Brazilian metropolitan regions in 2004, 42% of all jobs were informal (i.e. without a formal work permit). At the same time, earnings inequality and informality rates significantly declined between the early 2000s and the late 2010s. This makes Brazil a particularly interesting setting to study for the purpose of understanding the nature of earnings inequality and dynamics in a developing country.
To make this study possible, we analyse high-quality administrative records from the Relação Anual de Informações Sociais (RAIS), which covers formal-sector employment in Brazil every year from 1985 to 2017, and detailed longitudinal household survey data from the Pesquisa Mensal de Emprego (PME), which covers individuals within households in Brazil’s six largest metropolitan regions at a monthly rate from 2002 to 2015. Altogether, these data cover up to 42 million formal-sector job spells per year in the RAIS data, and up to 34,000 households comprising 95,000 individuals in each month in the PME data.
Our main contribution is to dissect earnings inequality and dynamics in Brazil’s formal and informal sectors over this period using a combination of detailed administrative and household survey data. Our analysis complements previous work by Gomes et al. (2020), who also study earnings dynamics in Brazil using data from a different household survey for a shorter period. By combining administrative and household survey data, and by jointly studying earnings inequality and dynamics over a longer period, we provide a more complete picture of the Brazilian labour market than has been previously possible.
Over this period, there were several important changes in institutions and the macroeconomy in Brazil. Between 1985 and 2017, the Brazilian economy was characterised by rapid growth spurts interlaced with severe economic recessions. From the mid-1990s onwards, earnings inequality declined rapidly (Alvarez et al., 2018, Firpo and Portella, 2019), partly due to the concurrent rise of the real minimum wage (Engbom and Moser, 2021). Finally, while the informal employment rate remains substantial in Brazil, as in many other developing countries, there has been a pronounced trend of formalisation over a decade starting in 2004.
We document a significant decline in earnings inequality since the mid-1990s within the formal sector in Brazil. This is illustrated in the left panel of Figure 1, which shows various percentiles of the log earnings distribution between 1985 and 2017, all normalised to zero in 1995. Although real earnings grew throughout the earnings distribution over this period, real earnings growth was particularly pronounced at the bottom of the distribution. Consequently, earnings inequality as measured by the standard deviation of log earnings or the P90-P10 log earnings percentile ratio sharply decreased starting around 1995, as shown in the right panel of Figure 1.
Figure 1 Evolution of percentiles (left) and dispersion (right) of the log earnings distribution from 1985-2017
Source: RAIS, 1985-2017.
We also highlight that the large decline in inequality over this period is evident throughout careers, with inequality being lower at all points over the life cycle today relative to 20 years ago. Nevertheless, it is important to note that inequality actually increased at the very top of the earnings distribution, consistent with patterns observed in many high-income countries over this period.
In contrast to the large secular decline in inequality, the volatility of earnings has trended down more modestly in Brazil over this period. Figure 2 shows that between 1987 and 2014 there has been a pronounced decrease in the dispersion of five-year residual log earnings growth dispersion among men (left panel) and, though to a lesser extent, among women (right panel). At the same time, the tails of the distribution of earnings changes have become fatter over time. That is, it is more likely today relative to 20 years ago that a worker experiences a very large change in earnings, in both a positive and negative direction. There are also important differences in earnings dynamics over the life cycle and as a function of a worker’s income. Specifically, earnings are more volatile among young workers and poorer workers.
Figure 2 Evolution of five-year residual log earnings growth dispersion among men (left) and women (right) from 1987-2014
Source: RAIS, 1985-2017.
A rapidly expanding literature has successfully leveraged large scale administrative data on workers and firms to shed new light on labour market dynamics. An important concern with these studies is the extent to which such administrative data match up with what individuals self-report in surveys. We provide new evidence for this important debate by combining several survey data sources. The findings highlight similar trends in inequality and earnings dynamics in Brazil’s formal sector over this period, but important differences in levels between administrative and survey data. In particular, both the level and volatility of earnings is smaller in survey data relative to administrative data.
Over this period, the Brazilian labour market has become increasingly formalised. Figure 3 shows the share of informal employment between 2002 and 2015. This process of labour market formalisation is indeed evident throughout the earnings distribution. Moreover, in a purely accounting sense, it is primarily accounted for by a fall in the rate at which formal sector employees leave to the informal sector, although a higher transition rate from the informal to the formal sector also plays a role. In a similar spirit, the higher informality rate at the bottom of the earnings distribution is primarily accounted for by a higher transition rate from the formal to the informal sector, as opposed to a higher transition rate from the informal to the formal sector. Although these are pure accounting exercises, they shed important light on the potential drivers of labour market formalisation.
Figure 3 Evolution of share of informal employment from 2002 to 2015
Source: PME, 2002-2015.
We next turn to a comparison of earnings inequality and dynamics in the formal versus informal sector. In levels, inequality is higher in the informal sector. Moreover, the decline in inequality over time in Brazil is particularly pronounced in the formal sector. Since the minimum wage only applies in the formal sector, this observation is consistent with the argument that a rapid rise in the real minimum wage over this period has been an important driver of the fall in inequality in Brazil. That being said, the informal sector also experienced falling inequality.
Earnings are substantially more volatile in the informal sector relative to the formal sector, and especially volatile among those workers who switch sector, as shown in Figure 4. Consequently, ceteris paribus, one would expect the process of labour market formalisation over this period to have contributed to a fall in overall earnings volatility. Quantitatively, however, we find that the most important factor behind the overall decline in earnings volatility is a fall in volatility within sectors, in particular within the informal sector.
Figure 4 Densities of one-year residual earnings changes by origin and destination sector
Source: PME, 2002-2015
Our research efforts are part of a larger undertaking to understand the nature of income inequality and dynamics based on administrative data records under the umbrella of the Global Income Dynamics Project. As part of the project, a group of 13 (and counting) country teams compute harmonised statistics on earnings inequality and dynamics that will be disseminated for academic researchers and policymakers around the world via an online database that will be made available on a public website. The goal of the project is to provide harmonised statistics that can be used to calibrate income processes in micro- and macroeconomic models of the labour market, to guide the design of optimal redistributive and social insurance policies, and to inspire future research on earnings inequality and dynamics for a broad set of countries across the development spectrum.
Alvarez, J, F Benguria, N Engbom and C Moser (2018), “Firms and the Decline of Earnings Inequality in Brazil,” American Economic Journal: Macroeconomics 10(1): 149-189.
Busch, C, D Domeij, F Guvenen and R Madera (2020), “Skewed Idiosyncratic Income Risk over the Business Cycle: Sources and Insurance,” Mimeo.
Dix-Carneiro, R, P Goldberg, C Meghir and G Ulyssea (2015), “Informality and the effects of trade in developing countries”, VoxEU.org, 5 March.
Engbom, N, G Gonzaga, C Moser and R Olivieri (2021), “Earnings Inequality and Dynamics in the Presence of Informality: The Case of Brazil,” Mimeo.
Engbom, N and C Moser (2021), “Earnings Inequality and the Minimum Wage: Evidence from Brazil,” Mimeo.
Firpo, S and A Portella (2019), “Decline in Wage Inequality in Brazil: A Survey,” Policy Research Working Paper 9096, The World Bank.
Guvenen, F, F Karahan, S Ozkan and J Song (2019), “What Do Data on Millions of U.S. Workers Reveal about Life-Cycle Earnings Dynamics?” FRB of New York Staff Report, 710.
Guvenen, F, S Ozkan and J Song (2014), “The Nature of Countercyclical Income Risk,” Journal of Political Economy 122(3): 621-660.
Hoffmann, E B and D Malacrino (2019), “Employment Time and the Cyclicality of Earnings Growth,” Journal of Public Economics 169: 160-171.
Gomes, D B P, F S Iachan and C Santos (2020) “Labor Earnings Dynamics in a Developing Economy with a Large Informal Sector,” Journal of Economic Dynamics and Control 113.
Katz, L F and D H Autor (1999), “Changes in the Wage Structure and Earnings Inequality,” in O Ashenfelter and D Card (eds), Handbook of Labor Economics.
Kopczuk, W, E Saez and J Song (2010), "Earnings Inequality and Mobility in the United States: Evidence from Social Security Data since 1937," Quarterly Journal of Economics 125(1): 91-128.
McCaig, Brian and Nina Pavcnik (2015), “Creating Formal Sector Jobs in Low-Income Countries,” VoxEU.org, 4 August.
Nakajima, M and V Smirnyagin (2019), “Cyclical Labor Income Risk,” Mimeo.
Shorrocks, A (1978), “Income Inequality and Income Mobility,” Journal of Economic Theory 19(2): 376–393.