Labor market turnover and inequality in Latin America

Abstract

This paper describes the patterns of worker turnover in selected Latin American countries and their implications for wage inequality. We find that labor flows are as frequent as in many rich economies and that transitions involving informality are somewhat more likely, but even the formal market does not seem especially rigid. Job-to-job transitions are mostly wage-increasing. Job-to-job moves are especially positive for the worker when involving a transition from informality to formality, within formal jobs and from small to higher size firms. Wage gains are higher among younger workers who also face a more fluid labor market, with more frequent moves across jobs, and into nonemployment. We document a higher positive annual wage growth rate for job-to-job changers compared to stayers, consistent with turnover capturing the immediate gains from search behavior in the short run. Next, the paper analyzes wage growth by percentiles for all workers and job-to-job movers for each country over a more extended period. We find that human capital effects dominate the search effects in the long run, as human capital accumulates over time. Thus, long-run wage growth is lower for job changers than for stayers, so that, while in the short run the search effects tend to dominate those of human capital, in the long run the opposite occurs. As unskilled workers change jobs more frequently, this suggests that job changes are inequality-increasing in the long run. A potential explanation for limited wage growth in Latin American economies may include high informality rates. Policies to reduce wage inequality should focus on improving the conditions for positive turnover towards better investment and, thus, higher-quality jobs.

Introduction

Most economists agree that labor turnover in Latin America is very high, despite its strict labor regulations. Around 24–44% of the labor force separates from their jobs every year, with 50–70% of these exits going to another job.¹ This empirical phenomenon is primarily linked to the existence of an unregulated informal sector that accounts for a large share of the labor market.² Informal sector jobs last around three times less than formal sector jobs (e.g. Bosch and Maloney, 2010; Meghir et al., 2015; Narita, 2020), consistent with the lack of regulation and benefits in this sector.

High labor turnover may negatively affect the economy as it lowers on-the-job capital accumulation, leading to lower wage growth over the life cycle, as predicted by the human capital theory (Mincer, 1958, 1974; Ben-Porath, 1967 and Becker, 1975). However, not all transitions are bad, as gains from reallocating workers from low to high-productivity jobs can be substantial. Search models study job mobility as an outcome of the arrival of a job that can be rejected or accepted and explain why mobility alone can be a source of wage growth (e.g. Burdett, 1978). Evidence suggests that job search is important at the early stages of the worker’s career as it improves the matching with a suitable job in which workers can stay and get promoted over time (Mincer and Jovanovic, 1981; Topel and Ward, 1992).

More importantly, these different explanations have implications for wage inequality as workers are heterogeneous with respect to the level of investment in human capital, search intensity, job matching, and acceptance. For example, in many economies, less educated and young workers have the shortest job tenure, and thus face lower training rates and poor pay progression. Moreover, most of them face frequent transitions to the informal sector or out of the labor force. It is, therefore, likely that a high turnover in Latin American (LAC) countries will exacerbate existing labor market inequalities in the long run, but whether this happens or not is an empirical question and depend on the labor market institution of each country.

This paper uses household data to produce original results about job turnover and associated wage changes for workers with varying human capital levels in five different LAC countries. We start with a background discussion that will help interpret evidence on labor turnover and connections with wage inequality. Then, after briefly reviewing the literature on labor turnover in developed countries, this paper presents evidence on the patterns of labor mobility in Latin America. We investigate whether turnover is genuinely high in Latin America compared to rich countries and, if so, how this relates to the degree of stringency of labor market regulations. Then we provide detailed measures of job-to-job transitions, including switching of occupation, industry, firm size and formality status using longitudinal labor force surveys of Argentina (2003–2019), Brazil (2012–2019), Mexico (2005–2019), Ecuador (2008–2019) and Chile (2010–2019). We assess the potential wage gains or losses from turnover by looking at the fractions of job-to-job changers and job stayers with a wage increase or decrease, and the corresponding average real wage variation.

We find that job-to-job changes involve greater annual wage growth than remaining in jobs. This remains true when analyzing the formal sector separately, even though we would expect higher investment in human capital and, therefore, higher returns for staying in the same job in this sector. We also find that younger workers (aged 18–24) benefit relatively more from the positive effects of job-to-job changes, as expected. Next, we show that transitions involving informality are somewhat more likely in most countries, but even the formal market does not seem especially rigid. Most of these findings align with the fact that workers with low wage levels obtain higher annual gains, or lower losses, than other workers in a job change. We argue that a higher positive annual wage growth we find for job-to-job changers compared to stayers is due to short-run effects of turnover capturing the immediate gains from search behavior.

Next, we provide evidence of the relationship between labor turnover and wage inequality in the long run by looking at the distributions of wages over a longer period of time. By comparing observed and counterfactual wage growth by percentiles for job-to-job movers in each country, we present a decomposition exercise to analyze the contribution of job-to-job transitions to changes in wage inequality. Unlike in the individual-level analysis, job separations seem to be inequality-reducing, as they harm individuals at the lowest percentiles relatively less. In contrast, we find that human capital effects dominate the search effects in the long run, as human capital accumulates over time. Thus, long-run wage growth is lower for job changers than for stayers, so that, while in the short run the search effects tend to dominate those of human capital, in the long run the opposite occurs. As unskilled workers change jobs more frequently, this suggests that job changes are inequality increasing in the long run.

Finally, going beyond wages as a measure of job quality, this paper also contributes to the literature by studying the role of nonwage compensation for job turnover and wages. We find causal evidence for Brazil that the formal sector is less likely to hire and more likely to fire workers after the introduction of private health insurance (PHI) due to increased labor costs. To the extent that workers of different ages, education and gender are similarly affected, we argue that the effect of such policy on wage inequality is ambiguous. As for publicly provided nonwage benefits to informal workers and non-employed, we find lower inflows to the formal sector from nonemployment for women and from the informal sector for men in Mexico. In addition, as less educated and female workers are more affected by the introduction of non-contributory insurance, this may have contributed to delaying the decline of wage inequality in Mexico.

Conceptual issues: labor turnover and wage inequality

Labor turnover may have effects on wage inequality due to workers’ ex-ante heterogeneity implying that some workers benefit from mobility, while others do not. It can also affect inequality through workers’ ex-post heterogeneity due to e.g. occupational, or sectoral mobility leading to a different distribution of tenure and, thus, of human capital and wages. Finally, it may also reflect an environment where job-to-job transitions alone generate wage dispersion, as discussed below.

Negative turnover characterized by dismissals or instant transitions to worse jobs have negative implications for future wages. Following a dismissal, individuals face the risk of falling human capital accumulation and may be less likely to find a job paying the same or a higher wage. In this sense, negative turnover may increase wage inequality if the less paid or low educated and younger individuals are the ones who exit more often to unemployment or to worse jobs.

However, there are positive job-to-job separations, those involving transitions to better paid jobs, for example. In the search framework proposed by Burdett and Mortensen (1998), workers only change jobs if the pay exceeds their current wage. By paying higher wages, firms are thus able to attract workers from other firms offering lower wages. In equilibrium, since firms are homogeneous and thus have equal profits, firms that pay less employ fewer workers. As a result, even when workers are homogenous, the existence of search frictions and individuals searching on the job generates wage dispersion in equilibrium. All other factors being equal, this means that in a setting where positive job-to-job transitions take place more often wage inequality is higher, entirely motivated by a market structure argument rather than individual or job heterogeneity.

In an extension of this model, Bontemps et al. (2000) allow for heterogeneous job productivity showing that not only this implies a better fit of the wage data, but that productivity dispersion is an important determinant of the amount of wage inequality in the population. That said, job heterogeneity and selection of individuals across jobs imply that some individuals benefit more and some benefit less from job-to-job separations. Job changes in this case may increase wage inequality if e.g. low educated and younger individuals gain relatively less by moving jobs.

On the other hand, job search theories with learning suggest the importance of search especially for job-to-job moves at early stages of the worker’s career and for workers with low education. Limited information about the match quality, which is only revealed after production takes place, emphasizes the role of experimentation especially for such workers (Pries and Rogerson, 2005) This mechanism implies that hazard of job separation rises then declines over the life cycle and with education (Jovanovic, 1979; Rubinstein and Weiss, 2006).

Job-to-job turnover may also involve mobility across occupations, industries, firm size, and formal and informal sectors. Structural evidence for the U.S. suggests that occupational mobility and wage inequality are highly associated (Kambourov and Manovskii, 2009). An increase in occupational mobility accounts for over 90% of the increase in wage inequality between the 1970’s and the 1990’s. Kambourov and Manovskii (2008) document that there was a considerable increase in the fraction of workers switching occupations over the same period, and find substantial returns to tenure in an occupation, consistent with the occupation specificity of human capital.³ Occupational transitions influence wage inequality because mobility at this level impacts the distribution of occupational tenure and, thus, of human capital. According to such evidence, the extent of occupation mobility and returns to occupation tenure are important determinants of wage inequality.

Existing empirical evidence shows that developing countries have steeper wage-tenure profiles due to higher information frictions and thus lower initial wages (Donovan et al., 2022) Evidence of this has also been provided by Marinescu and Triyana (2016) who show substantially higher returns to tenure in the formal sector in developing countries than in the U.S. Regarding the intensity of mobility across occupations, sectors and firm size evidence for developing or LAC countries is scarce. The next sections of this paper will provide additional evidence on the types of job-to-job mobility for five LAC countries.

In the developing country context, with dual labor markets characterized by a formal and an informal sector, one could argue that job-to-job transitions favor a selected group of individuals that are able to move to the formal sector. In this sense, job-to-job transitions across formal and informal sectors tend to increase wage inequality if e.g. the low educated and younger individuals tend to move more often into informal jobs, i.e. those that are less productive and pay less on average. A large literature has documented that the transitions from unemployment and formal jobs into informal employment are higher for the young, women, and low-skill workers in Latin America (see Bosch and Maloney, 2010; Ulyssea, 2020). For a large set of developing and developed countries, Donovan et al. (2022) find that job-to-job switching rates are five times higher in poorer countries. These higher flows in poorer countries are driven by more frequent transitions within the informal sector rather from informal to formal sector.

As for job turnover induced by labor reforms in LAC, a reduction in job security is associated with an increase in employment exit rates in Colombia, Peru, and Argentina. Likewise, an increase in job security is linked to a decline in exit rates in Brazil. The evidence for Colombia shows that the rise in exit turnover is larger for middle-aged and older men employed in large firms, those who are more protected by security provisions (Kugler, 2000). Women and the young benefit more from higher entry rates from unemployment into the formal sector, however around 2/3 of formal hiring can be attributed to the use of temporary contracts. For Peru, blue-collar workers and, for Argentina, workers with college and those employed in large firms have a lower risk of job termination (Saavedra and Torero, 2004; Hopenhayn, 2004). In Brazil, an increase in job security, in particular the penalty paid by firms in case of unjustified dismissals, reduced fake layoffs (when the worker and the firm reach an agreement for a layoff so that the worker is entitled to collect the benefits of UI and severance savings however has to reimburse the firm for the firing penalty) with a greater decline for workers with more education (Gonzaga, 2003; de Barros and Corseuil, 2004) In terms of UI reforms, the estimated negative effect of a stricter UI eligibility on formal sector layoffs is greater for workers in small firms, with low education and in their first job, groups that have a higher UI replacement rate (Carvalho et al., 2018) Evidence for LAC countries are broadly consistent with that in OECD countries, reinforcing that a decrease in job security reduces income security of formerly protected workers but increase the job finding rate in the formal sector. It also suggests that job (as well as unemployment income) security provisions are inefficient and increase inequality (Heckman and Pages, 2000).

Regarding job turnover induced by trade reforms in LAC countries, early evidence using industry-level data provides little support for the view that trade openness would reduce inequality by enabling better labor reallocation of workers across sectors in developing countries (Goldberg and Pavcnik, 2007) In contrast, microdata evidence shows that the increased reallocation of workers across occupations and industries due to trade liberalization played a major role in the reduction of wage inequality in Brazil during 1988–1995 (Ferreira et al., 2010). However, other works suggest that labor market responses following trade liberalization may take several years (e.g. Dix-Carneiro, 2014; Dix-Carneiro and Kovak, 2017) In particular, Dix-Carneiro (2014) shows that trade reforms over the period 1995–2005 had important distributional effects with women, less educated, and older workers facing higher costs of switching sectors in Brazil. Such reforms also increased job displacement in Brazil (Menezes-Filho and Muendler, 2011) and Colombia (Coşar et al., 2016). Finally, Dix-Carneiro and Kovak (2019) find that labor reallocation to informality acted as a buffer against nonemployment in regions facing larger tariff cuts.⁴

Labor market dynamics: main facts

Turnover patterns in developed countries

Cross-country and cross-industry data show significant variation in job and worker flows (OECD, 2010)⁵ Using individual level data, Jolivet et al. (2006) draw on a panel of 10 European countries and the U.S. that allows following both employed and non-employed individuals yearly for up to 3 years or until their first change of status in the labor market which can correspond to a job-to-job, a job-to-nonemployment or a nonemployment-to-employment transition.⁶ Four main facts emerge from Table 1. First, across countries, job-to-job transitions are as important as transitions from job to nonemployment. Second, job-to-job turnover varies widely across countries, with high turnover in Denmark and the U.K., low turnover in Belgium, France, Italy, Portugal, and Spain, and the remaining countries, including the US, falling between these two. Third, there is no strong correlation between job loss rates and the job-to-job transitions. Most job loss rates range from 9 to 17%, excluding France (4%) and Spain (23%), and including countries known to have less stringent labor regulations such as the U.S. Fourth, most job-to-job transitions are associated with a wage increase, even though part of job-to-job transitions is constrained, or the wage is not the sole reason why workers move jobs, an issue we attempt to address in a later section.

By taking advantage of large sources of U.S. data collected over several decades,⁷  Rubinstein & Weiss (2006) conduct a separate analysis of mobility across employers, occupations and sectors. For sectoral and occupational mobility, transitions decline quickly with potential experience while the proportion of workers moving across employers initially increases in their first 2–3 years of experience and then declines, remaining at a relatively higher rate of ~15% per year at the end of worker’s career. This suggests the importance of job search earlier in the career and of sector and occupation capital rather than firm-specific capital (see also, Kambourov and Manovskii, 2009).

Turnover in Latin America

As with developed countries data, a major challenge with analyzing transition patterns in Latin America is to have harmonized data. A recent contribution is Beccaria and Maurizio (2020) who use household surveys from six LAC countries: Argentina, Brazil, Ecuador, Mexico, Paraguay and Peru.⁸ By looking at yearly job exit rates measured in two ways, including or excluding job-to-job changes, they find that the total exit rates for these countries range between 24 and 44%, with job exit rates to nonemployment being very similar between 11 and 13% for most countries, and slightly greater for Mexico and Peru. Comparing Table 1 and Table 2’s rates, we observe that the job separations to nonemployment in Latin America are close to those obtained for Germany, U.S., Denmark, Italy, and Portugal. Unlike many developed countries, 50 and 70% of job exits were for another job, making these the most frequent types of employment transitions in these LAC countries.⁹ These observed differences in the job-to-job turnover rates reveal an important feature in LAC countries’ data given the pronounced low levels of social protection, skills, and other issues that we will examine in more detail in the next section. Compared to the job-to-job rates in Table 1, the average yearly job-to-job rates across most of these LAC countries is similar to that observed for the U.S. and Ireland.

The role of informality

One important feature of labor markets in developing countries is informality, which corresponds to more than 40% of jobs in LAC countries.¹⁰ Unlike the informal sector, the formal sector is subject to labor market regulations such as minimum wages and employment protection, which can induce labor force adjustments including transitions across sectors. An analysis of sectoral transitions and duration in Argentina, Brazil, and Mexico shows that durations differ significantly across sectors but are very similar across these countries. On average, formal employment lasts 4.5 years, followed by informal self-employment which last for 2 years, and informal salaried jobs having the shortest duration of 1 year. Argentina shows relatively higher unemployment duration than Brazil and Mexico, consistent with stricter labor market rigidity (Bosch and Maloney, 2010). Across different measures, the most predominant fact is that flows between informal and formal salaried jobs are highly asymmetric with informal to formal flows being several times higher than the reverse flow. On the other hand, there is a high degree of symmetry in formal salaried-self-employment flows against the view of comparative advantage in formality, and supporting the idea that workers are taking advantage of profitable opportunities as they arrive (Bosch and Maloney, 2006, 2010). Also, mobility within the informal sector is significantly higher, around 3 times greater for informal than for formal salaried workers in Brazil (Meghir et al., 2015; Narita, 2020), again consistent with the lack of regulation and benefits in the informal sector.

Surprisingly, even when considering registered (formal sector) workers only, the total exit rates are not dissimilar across Argentina, Brazil, and the U.S., with monthly exit rates around 4–4.5% for Argentina, 2% for Brazil and 3% for the U.S., respectively (Beccaria and Maurizio, 2020  Narita, 2020).¹¹ However one could expect much lower exit rates for Argentina and Brazil, given stricter labor regulations in LAC countries. Using the OECD employment protection index, job security of permanent workers against individual dismissals is very low in the U.S. (0.5) compared to Brazil (1.84), which is a little below the average of OECD countries (2.05). Regulation of temporary employment is stricter in Argentina (3.0) and Brazil (4.1) than in OECD countries (2.1), and one of the least strict in the U.S. (0.3).¹²

In fact, as highlighted above, yearly job exit rates to nonemployment across these six LAC countries are very similar, even when compared to European countries known to have stricter labor market regulations such as Portugal, Italy and Germany. But, if (average) worker turnover is not a distinguishing feature of LAC countries, does it imply that labor turnover has a limited role to explain the high levels of inequality in the region? In the next sections, we provide further evidence on detailed measures of turnover in LAC that will help rationalizing the main insights in Section 3 and allow a discussion of the connections between workers’ mobility and wage inequality.

Further evidence for LAC

In this section we add to the existing evidence by providing detailed measures of job-to-job transitions including switching of occupation, industry, firm size and formality status. To assess potential gains or losses from turnover, we report the fraction of for job-to-job transitions and job stayers with a wage increase, decrease or same wage, as well as the corresponding yearly average real wage variation.

Data sources

We study a subset of five LAC countries, Argentina (2003–2019), Brazil (2012–2019), Mexico (2005–2019), Ecuador (2008–2019) and Chile (2010–2019). We use national household surveys containing a panel of individuals with which we can identify job-to-job transitions as they all have information on the job tenure.¹³ We track individuals’ employment status from their first interview until a year later and use the information on job tenure to verify whether the individual is in the same job or moved across different jobs.

We focus our main sample on males, unemployed or salaried workers working full-time (above 35 h/week) and aged 18–65 years.¹⁴ For each country, we construct detailed measures of annual flows out of employment as well as the wage variation for job changers and stayers. We do it for all workers and by formality status.¹⁵ We also analyze the main patterns of wage growth for workers among job movers across occupation, industry and firm sizes.

Main results

In this section, we present and discuss the main messages we obtain from descriptive analysis of annual flows. Our main results indicate that labor flows are as frequent as in many rich economies; transitions involving informality (at one or both ends) are somewhat more likely, but even the formal market does not seem specially rigid; job-to-job transitions are mostly wage-increasing, although about a third of them imply a wage cut; job-to-job moves are specially positive for the worker when involving a transition from informality to formality, but even transitions within the formal sector seem profitable; younger workers face a more fluid labor market, with more frequent moves across jobs, into nonemployment and also to formal states, which is consistent with their wages rising relatively more; workers in small firms obtain wage increases more often when moving to higher size firms than in the opposite direction. Finally, most of these findings align with the fact that workers with low wage levels obtain higher gains (or lower losses) than other workers in a job change.

Annual job flows are as frequent as in many rich economies

Figure 1a shows that, at one end of the spectrum, Ecuador and Argentina show relatively low percentages of job-to-job transitions (6.5% and 6.9% of employment in the first year, respectively), while at the other end, Brazil, Mexico and Chile show higher job-to-job transitions (respectively, 11.3%, 16.6% and 17.8%). Brazil, Mexico and Argentina have the highest job exit rates to nonemployment (from 15% to 17%) while in Ecuador and Chile such rates are ~10%. Such results are not so dissimilar compared to the range we observe in Table 1.

The informal sector is more fluid

The fraction of job-to-job transitions among informal sector workers is higher than for formal workers across all countries, as Figs. 1b and c show. While most formal job-to-job transitions occur within the formal sector, a large fraction of informal workers move towards the formal sector. However, the share of informal employees leaving to non-employment is well above that observed for formal employees.

More than half of job changers are gainers

When comparing wages before and after a job change (Figs. 2–6, panel a), we observe a large share of job changes associated with a wage cut in Ecuador, Mexico and Chile, 32%, 35% and 38%, respectively.¹⁶ This is above the average fraction of wage cut observed for many European countries and the U.S. (28%) as we observe in Table 1, which is similar to the fraction in Brazil and above that for Argentina, 17%. Panel (c) in Figs. 2 to 6 show that job stayers may also experience wage gains as well as losses, with a smaller fraction of gainers for most countries and a lower wage increase compared to that for job-to-job movers.¹⁷ This remains true even when analyzing separately the formal sector, in which we expect a larger investment in human capital and thus higher returns for job stayers.

Gains of job changing are larger for informal workers, especially when they move to the formal sector

We find higher gains for informal employees who have an opportunity to be employed in the formal sector but less so in the opposite direction (Panel (b) in Figs. 2–6). Nonetheless, Panel (d) in Figs. 2–6 show that the gains from job changing are larger compared to staying in same jobs for both formal and informal sector workers, even though we would expect stayers in the formal sector to have larger wage gains due to better quality and thus more human capital accumulation in such jobs.

Job-to-job transitions are higher among young and less educated individuals

A much higher fraction of young workers (aged 18–24) moves jobs in a year, 3 to 5 times higher than that for older workers (Fig. 7), consistent with lower search gains over the life cycle. In most countries, the majority of job-to-job transitions occur towards a formal or another formal job. Consistently, the wages of younger workers also rise more (Fig. 8). While the differences across education levels are small, job-to-job changes fall with education levels in most countries. In Mexico, around two thirds of job movers end up in a job or another job in the informal job among low education individuals (incomplete secondary). In contrast, in Brazil and Chile, the ratio is one third and one fifth, respectively. This explains relatively higher gains of job moving for low education workers in Brazil and Chile.

Transitions involving occupation or industry changes are frequent but there are more gainers among stayers

Around 36% to 54% of job-to-job transitions involve occupation changes (Fig. 9). In most countries, however, the fraction of gainers among workers who change occupation is lower than the fraction of gainers among those who stay in the same occupation. As expected, the latter is greater for non-clerical workers consistent with more human capital accumulation in non-clerical jobs. Similarly, Fig. 10 shows that industry changes also correspond to high fractions, 43% to 66%, of all job-to-job transitions and the fraction of wage gainers is higher for workers who do not switch industries, except for Chile and Ecuador. In most countries, leaving the tertiary sector involves relatively more wage gains, consistent with lower human capital accumulation in trade and services.

Workers in small firms, that move to higher size firms, obtain wage increases more often than in the opposite direction

Figure 11 displays transitions according to firm size defined in three categories,¹⁸ where we observe that 44% to 54% of all job-to-job transitions involve changing firm size. Wage increases are more frequent when moving from small to larger firms in most countries. Mexico is an exception, where transitions involving firm size change and wage gains are similar for all initial firm sizes. Such evidence is consistent with large firms being associated with more investment in training that possibly leads to higher opportunities for pay progression. However, to the extent that large firms are subject to heavier costs of regulation, this may prevent investment reducing workers’ compensation in large firms, such that transitioning towards large firms does not necessarily increase wages.

How wages change depending on the worker’s initial wage?

All facts above stated do not suggest a clear relationship between annual job transitions and wage inequality. To provide a more direct assessment of the relationship with inequality, we analyze the wage growth among job-to-job movers by initial wage quartile. We should expect that workers with low wage levels obtain higher gains/lower losses than other workers in a job change, if wage inequality is decreasing. In fact, this is the case in the countries analyzed as we see in Fig. 12. Workers in the first wage quartile obtain higher wage gains than those in higher quartiles in most countries. The except is Brazil, where gains are more evenly distributed across initial wages. However, in Brazil there are significant lower wage losses for workers in the first quartile, consistent with binding minimum wages. All such evidence aligns with our previous findings of higher wage gains for low education workers and those working in small and informal firms moving in the direction of larger and formal firms.

Growth incidence curves

In this section we provide a second look at the data, by providing more direct evidence of the relationship between wage inequality and labor turnover, while stressing previous findings and the discussion above. We present a decomposition exercise that attempts to quantify the contribution of yearly job-to-job transitions to the change in wage inequality. Specifically, we compare observed and counterfactual wage growth incidence curves for job-to-job movers.¹⁹ We focus this analysis on the two largest LAC countries in their respective sample period: Brazil (2012–2019) and Mexico (2005–2019).

In order to address the endogenous selection of workers into the sample of job-to-job changers, we follow the literature and estimate a discrete choice equation for the binary decision between changing and staying in the same job.²⁰ We allow a rich set of individual and job characteristics (age, squared age, cubic age, age and education interactions, industry, occupation, formal sector, household size and its square, head of household, and a year dummy), analogous to Ferreira and Barros (2005). The two panels of Fig. 13 plot the observed wage growth incidence curve between the first year (T0) and last year (T1) of each country data series, $g(p)=(w^{T1}(p)-w^{T0}(p))/w^{T0}(p)$⁠, and the counterfactual growth incidence curve: $g^s(p)=(w^s(p)-w^{T0}(p))/w^{T0}(p)$⁠, where $w^s$ correspond to the wage in T1 of the simulated job-to-job movers only.²¹

The dark line in Fig. 13 shows that wage inequality reduces in Mexico over the 2005–2019 period across all parts of the wage distribution. In Brazil, wages grow faster at lower percentiles and less so at the upper end, showing a reduction in the 90/10 differential in our sample for the 2012–2019 period. Importantly, the simulated growth incidence curves (gray lines) show that job-to-job turnover is inequality-reducing in both countries. In Mexico, job-to-job changers at the 10th percentile experience a wage growth of around 9% that is above the actual wage growth whereas those at the 90th percentile have lost more, almost 40%. For Brazil, the picture is similar with those at the 10^th percentile obtaining a 10% increase in real wages whereas those at the 90^th percentile having a wages loss of around 40%. However, the figure also shows a clear tradeoff. Mean wage growth over the considered time periods is lower among all job-to-job changers in Brazil and for the majority of those in Mexico, except the bottom 20%.

This apparent inconsistency with the analysis of turnover and wage growth at the individual level can be reconciled with the dynamic effects of turnover. The analysis of turnover and wage growth at the individual level can only be conducted within 1 year due to short panels, therefore it does not allow capturing long-run effects. In contrast, we can look at long-run effects by examining the incidence of growth at different points in the wage distribution over many years. Such differences we find in the relationship between wage growth and turnover in the short and long run can be rationalized by both the search theory and the theory of human capital. Search behavior suggest that the decisions to accept or turn down new offers (transitions) are triggered by shocks that can happen at any point in time. Particularly, positive shocks involve wage gains that are higher for the lowest paid. However, such gains tend to exhaust decreasing wage growth in subsequent periods, because high wage individuals are less likely to obtain greater offers. On the other hand, the human capital model would suggest that wage growth increases with experience (time) as individuals invest in training early in the career and collect gains later. These two complementary explanations imply that, in the short run, search effects tend to dominate those of human capital and, in the long run, the opposite. That search gains exist in the short run suggest that annual wage growth constructed from short panels would be greater for job-to-job changers than for stayers, whereas the wage growth by percentile of the wage distributions over a longer period would be lower for job-to-job changers than for stayers. As younger and less educated workers change jobs more frequently, job changes are inequality-increasing in the long run.

Finally, we complement our analysis by analyzing workers with high and low job tenure, where high tenure means the sample median. Both in Mexico and in Brazil, growth is higher among stayers for all workers as Fig. 14 illustrates. Unlike before, the lowest paid workers gain from staying longer in jobs in Mexico as expected since negative turnover is more frequent among workers with low education, those at the bottom of the wage distribution (as we can see in Fig. 7c). In this country, wage inequality falls within both groups, with low and high tenure, but more so among those workers with low tenure, whereas in Brazil the fall in wage inequality is driven by workers with low tenure. However, mean wage growth is lower for low tenure workers and, as less skilled workers tend to select into this group, it suggests that job separation increases inequality in the long run.

The role of nonwage compensation

Due to lack of data availability, it is hard to know how workers’ transitions are related to broader measures of job quality going beyond wages. Evidence shows that workers value job amenities such as job security and distance to work in several European countries, working times in France (Bonhomme and Jolivet, 2009), work conditions (Gronberg and Reed, 1994) and health insurance in the U.S. (Dey and Flinn, 2008). In this sense, there is scope for nonwage attributes to improve job quality but the impact on welfare inequality is unclear as it depends on how wages and employment adjust. In theory, nonwage job characteristics can drive both labor turnover and wage dispersion due to compensating differentials (Rosen, 1986). However empirical evidence on the correlation between wages and job amenities is quite mixed.

Search models provide useful insights to explain such conflicting results. If mobility is imperfect, no, or weak correlations may occur even if workers value job amenities. Particularly, when workers are constrained to move to better jobs or subject to reallocation shocks, firms have no incentive to compete against offers from other firms (Hwang et al., 1998; Bonhomme and Jolivet, 2009).²² This is consistent with weak or lack of evidence of compensating wage differentials. Overall, these results suggest that low opportunities for job upgrades and involuntary job changes may lead to higher inequality in job quality.

In this paper, we provide new evidence by exploiting within country variation in nonwage compensation such as the introduction of employment provided health insurance across firm in Brazil and the non-contributory health insurance across cities in Mexico to analyze the role of nonwage margins in job turnover and wages.

Employer provided health insurance

To examine whether and the extent to which PHI have effects on labor turnover, we use data on transitions between, into and exiting firms of the Brazilian formal sector over a one-year period considering only surviving firms, 2004–2018. We then link this dataset to administrative information on PHI contracts at firm-level in the same period. By exploiting the variation in the timing of implementation of the private insurance benefit across firms, Fig. 15a shows that formal firms hire 0.01p.p. (4.5%) less in Brazil since PHI increases labor costs.²³ This is driven firms hiring less from outside the formal sector than from other formal firms, as well as by workers with less than secondary education, across all ages and both genders (Figs. A.1–A.3) As for the exit turnover, we find that the exit rate from the formal sector increases by 4% while to other firms reduces by 6.6% (Figs. 15b and15c). Most exits from the formal sector are driven by both genders, workers aged 25–44, and those with high school education, likely due to greater costs with PHI provision for them (Figs. A.4–A.6). In addition, the negative effects on job-to-job turnover are consistent with the introduction of amenity in the current work. This seems driven by workers with high school education and aged 24 to 44, for which we find some (weak) evidence that they move to jobs that pay higher wages likely due to an increase in their reservation wages (Figs. A.7–A.11). In sum, firms tend to hire less workers and to fire more due to the introduction of PHI. Workers also tend to become more reluctant to move across jobs. These results yield unclear predictions regarding wage inequality, to the extent that more and less-disadvantaged workers are affected.

Non-contributory health insurance

Mexico introduced a large non-contributory health insurance scheme in 2004, Seguro Popular, extending health coverage to the uninsured: non-employed and informal sector workers. Using quarterly labor force panel data from Mexico 2000–2012, we follow Conti et al. (2023) and exploit the variation in the timing of implementation of the policy across municipalities.

Table 3 presents the effects of the introduction of Seguro Popular on several measures of turnover including exits to nonemployment as well as transitions between formal and informal jobs. For both men and women, we find that introducing health amenity outside the formal sector reduces entry into this sector. For men, transitions from informal to formal jobs decline, driven by those with very low education and older (Tables A.1–A.2) Women however stay more in nonemployment as transitions from nonemployment to the formal sector reduces among those with low education (Tables A.3–A.4).

As the tables show, when we condition the job-to-job transitions on changing wages, less educated males in the informal sector are less likely to move to jobs in the formal sector involving a pay cut, consistent with the health policy in the informal sector having a relatively high value for these workers such that it increases their reservation wages.

In sum, the above results show that nonwage benefits provided by formal sector firms lower formal demand reducing hiring in and increasing firing from the formal sector. As for nonwage benefits that are publicly provided to informal workers and the non-employed, we expect supply responses to be of first order and we do find lower inflows to the formal sector from nonemployment for women and from the informal sector for men. To the extent that the formal sector is associated with more productive jobs on average, these results suggest that nonwage aspects of jobs such as health insurance reduce productive transitions in Brazil and Mexico. In addition, as less educated and female workers are more affected, the introduction of non-contributory health policies in Mexico may be linked to higher wage inequality.

Conclusion

The purpose of the paper is to provide an analysis of the alternative explanations for the relationship between turnover, wage growth and inequality. First, it adds to the existing literature by providing detailed measures of job-to-job transitions, including switching of occupation, industry, firm size, and formality status. In addition, the paper reports the fraction of job-to-job transitions and job stayers with a wage increase, decrease, or same wage, as well as the corresponding yearly average real wage variation to assess potential gains or losses from turnover.

Using panels from five LAC countries and following individuals over 1 year, we find that labor flows are as frequent as in many rich economies and that transitions involving informality are somewhat more likely, but even the formal market does not seem especially rigid. Job-to-job transitions are mostly wage-increasing, although about a third of them imply a wage cut. Job-to-job moves are especially positive for the worker when involving a transition from informality to formality, within formal jobs, and from small to higher size firms. Wage gains are higher among younger workers who also face a more fluid labor market, with more frequent moves across jobs, into nonemployment and also to formal states, but their wages rise relatively more. Finally, most of these findings align with the fact that workers with low wage levels obtain higher gains, or lower losses, than other workers in a job change. We argue that a higher positive annual wage growth we find for job-to-job changers compared to stayers is due to short-run effects of turnover capturing the immediate gains from search behavior.

Second, the paper provides more direct evidence of the relationship between wage inequality and labor turnover. We present a decomposition exercise to analyze the contribution of workers' transitions to the change in wage inequality. We do so by comparing observed and counterfactual wage growth by percentiles for job-to-job movers for each country over a more extended period. We find that job-to-job changes are inequality-reducing, consistent with search gains associated with turnover exhausting more rapidly for the high-paid workers. These results remain valid even when looking at total job separations, including both job exits to nonemployment and job-to-job transitions.

However, these results also show a clear tradeoff. Mean wage growth over time is lower among all job exiters in Brazil and the majority in Mexico. This apparent inconsistency with the analysis of turnover and individual wage growth can be reconciled by the dynamic effects of turnover. Unlike the individual wage growth analysis using short panels, by following wage distributions over a more extended period, we are thus able to capture the long-run effects of turnover. As search gains tend to exhaust with experience or time, and because human capital accumulates over time, we expect that human capital effects dominate those of search in the long run. Thus, wage growth is lower for job changers than for stayers, so that, as unskilled workers change jobs more frequently, job-to-job transitions can be inequality-increasing in the long run.

A potential explanation for limited wage growth in LAC economies includes the high levels of informality and the drivers behind it. Barriers to formalization impeding access to more productive jobs limit human capital gains from turnover. In this context, policies aimed at reducing wage inequality should focus on improving the conditions for positive turnover towards better investment and, thus, higher-quality jobs.

STUDY FUNDING AND APC FUNDING

We thank the financial support from (i) the London School of Economics and the Inter-American Development Bank (IDB) for the execution of the chapter ‘Labor market turnover and inequality in Latin America’ for the Latin American and Caribbean Inequality Review (LACIR), and (ii) the IDB for covering travel expenses related to a presentation of this research at the WIDER Development Conference, Reducing inequality – the great challenge of our time, held on 5–7 October 2022 in Bogotá, Colombia.

CONFLICT OF INTEREST

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

AUTHORS' CONTRIBUTIONS

Naercio Menezes-Filho: Investigation, Validation, Visualization, Writing—original draft, Writing—review & editing. Renata Narita: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing.

DATA AVAILABILITY

All data are from secondary source.

• Encuesta Nacional de Empleo, Desempleo y Subempleo (ENEMDU) from Ecuador, accessed in 2022. Available at https://www.ecuadorencifras.gob.ec/enemdu-trimestral/#

• Encuesta Permanente de Hogares (EPH), from Argentina, accessed in 2022. Available at https://www.indec.gob.ar/indec/web/Institucional-Indec-BasesDeDatos

• Encuesta Nacional de Ocupación y Empleo (ENOE), from Mexico, accessed in 2022. Available at https://www.inegi.org.mx/programas/enoe/15ymas/#Microdatos

• Encuesta Suplementaria de Ingresos, from Chile, accessed in 2023. Available at: https://www.ine.gob.cl/estadisticas/sociales/ingresos-y-gastos/encuesta-suplementaria-de-ingresos

• Pesquisa Nacional por Amostra de Domicílios Contínua (PNADC), from Brazil, accessed in 2022. Available at https://www.ibge.gov.br/estatisticas/sociais/trabalho/9171-pesquisa-nacional-por-amostra-de-domicilios-continua-mensal.html?=&t=microdados

• Private health insurance, from Brazil, obtained in 2018 from Agência Nacional de Saúde Suplementar (ANS). Only Brazilians citizens (or institutions) can request information about health insurance using the access to information law (LAI). Details at: https://www.gov.br/acessoainformacao/pt-br/assuntos/pedidos. The availability of data is subject to approval by the responsible department, must be stored in a security office and used only for research purposes.

• Relação Anual de Informações Sociais (RAIS), from Brazil, obtained in 2018 from Ministério do Trabalho e Emprego. Only Brazilians citizens (or institutions) can request this matched employer-employee database. Details at: https://www.gov.br/pt-br/servicos/solicitar-acesso-aos-dados-identificados-rais-e-caged. The availability of data is subject to approval by the responsible department, must be stored in a security office and used only for research purposes.

• Encuesta Nacional de Empleo (ENE), from Mexico, accessed in 2022. Available at https://www.inegi.org.mx/programas/ene/2004/

• Padrón—Beneficiarios de Protección Social en Salud de Seguro Popular, from Mexico, accessed in 2022. Available at https://datos.gob.mx/busca/dataset/beneficiarios-de-proteccion-social-en-salud-de-seguro-popular. This data set can be accessed by requesting it to the Mexican authority which maintains the data.

Footnotes

REFERENCES

Appendix A

Appendix B: Data Details

We study a subset of five LAC countries, Argentina, Brazil, Mexico, Ecuador and Chile. We use national household surveys containing a panel of individuals with which we can identify job-to-job transitions as they all have information on the job tenure. We track individuals' employment status from their first interview until a year later and use the information on job tenure to verify whether the individual is in the same job or moved across different jobs. For each country, we construct detailed measures of job-to-job transitions by wage variation, including switching of occupation, industry, firm size, and formality status.

We focus our main sample on males, unemployed or salary workers working full-time (above 35 h/week) and aged 18–65 years. We exclude individuals with any missing wage, those with nominal wage below the 2nd percentile or above the 98th percentile of the wage distribution in each year, as well as those with nominal wage variation below the 2nd percentile or above the 98th percentile of the nominal wage variation distribution in each year. The variation in wages is calculated in nominal terms, although the classification between wage decrease, increase or same wage is made in real terms, after temporal correction.

The occupations and industries were grouped into 2 and 3 categories based on the identification of 1-dig: Clerical and Non-clerical, and Primary, Secondary and Tertiary. Clerical are Directors and managers, Scientific and intellectual professionals, Mid-level technicians and professionals, and administrative support staff. Non-Clerical are Service workers and shop and market vendors, Farmers and skilled agricultural, Officials, operators and craftsmen, Plant and machine operators and assemblers, Elementary and Military occupations. Primary is Agriculture, Secondary encompass Manufacturing and Construction, and Tertiary contemplates Trade, Transportation, Financial Services, Personal Services, Public Sector and Domestic services.

Ecuador

Data for Ecuador were taken from the Encuesta Nacional de Empleo, Desempleo y Subempleo (ENEMDU) for the period from 2008-Q1 to 2019-Q2, except 2017. The pairing of individuals in t and t + 12 by the identifier code constructed with the variables: household, gender and age. Transitions are identified if the individual says at t + 12 they have worked for less than 1 year in the current occupation/firm, if employed in both observations. Formal worker is those who work in firms that have a Single Taxpayer Registry (Cadastro Único de Contribuinte—RUC), described by the variable ‘secemp’. For time correction of wages, the index provided by INEC (Instituto Nacional de Estadística y Censos) corresponding to the Consumer Price Index (IPC) per quarter was used, based on prices in March/2022 (1st quarter/2022).

Argentina

For Argentina, we used the Permanent Survey of Homes (EPH) for the period from 2003-Q3 to 2019-Q4, except in quarters in which the survey did not occur and in which it is not possible to find individuals in two sequential years. The pairing of individuals in t and t + 12 by the identifier code constructed with the variables: families (‘codusu’), household (‘nro-hogar’), number of the person in the family (‘component’) and gender. Definition of the transition by working time, considering as a change if the worker has worked for less than 12 months in the occupation/firm in t + 12 (‘pp07a’), if employed in both observations. A worker who reports that the firm deducts an amount relating to social insurance (‘pp07h’) is considered formal. The size of firms is given in three categories, not allowing them to be standardized with other countries, being: up to five employees, from six to 40 employees and over 40 employees. Furthermore, 13% of workers do not report the size of their firm, so analyzes involving this information include a smaller number of observations. For wages’ time correction, three indices were used: (i) from 2003 to 2013 the Indice de Precios al Consumidor Grande Buenos Aires (IPC GBA), (ii) from 2014 to 2015 the Indice de Precios al Consumidor Nacional Urbano (IPC NU) and (iii) from 2016 onwards, the Consumer Price Index (IPC), based on prices in March/2022 (1st quarter/2022).

Mexico

Data for Mexico comes from the Encuesta Nacional de Ocupación y Empleo (ENOE) for the period from 2005-Q1 to 2019-Q4. The Ocupación y Empleo 1 (COE1T) database was used in the quarters in which the expanded questionnaire was applied (year 2005, first and second quarter of 2006, second quarter of 2007 and 2008 and the first semester of each year between 2009 and 2019). The pairing of individuals in t and t + 12 by the unique identifier code constructed from the combination of identifying variables. Transitions are identified if the individual claims to have been employed for less than 1 year in the current occupation/firm, if employed in both observations. The formality of employment is determined by the variable ‘mh_col’ which records cases in which the worker is in firms that do not have conventional accounting practices, where there is no written employment contract or there is no access to health institutions through work. Approximately 9% of workers do not report the size of their firm, so analyzes involving this information include a smaller number of observations. For the time correction of wages, the index provided by INEGI (Instituto Nacional de Estadística y Geografía) corresponding to the INPC (Indice Nacional de Precios al Consumidor) per quarter was used, based on prices in March/2022 (1st quarter/2022).

Chile

Data for Chile are taken from the Encuesta Suplementaria de Ingresos for the period from 2010 to 2019. This is an annual basis collected between October and December of each year. The pairing of individuals at t and t + 12 by the identification code available through the research. Definition of transition by working time at t + 12 as less than or equal 12 months in current occupation/firm, if employed in both observations. Workers who report that the employer collects amounts related to social security are considered formal. The firm size is given in three categories, not allowing them to be standardized with other countries, being: up to four employees, from five to 49 employees and over 49 employees. Furthermore, more than half of the workers (52%) do not report the size of the firm, so analyzes involving this information include a smaller number of observations. An even smaller group of observations (17%) report the firm's industry, so analyzes involving this information include a smaller number of observations. For wages’ time correction, the index provided by INE corresponding to the annual IPC was used, based on prices in March/2022 (1st quarter/2022).

Brazil

For Brazil, we drew on microdata from the Continuous National Household Sample Survey (PNADC) for the period from 2012 to 2019. The pairing of individuals in t and t + 12 by the unique identifier code constructed from the combination of identifying variables. Definition of transition by working time at t + 12 as less than or equal 12 months in current occupation/firm, if employed in both observations. The identification of formality considers whether workers have a signed work card (CTPS). Between 2015 and 2016, no information was collected on the size of the firm, therefore, the analyzes involving this information include a smaller number of observations and coverage period. For time correction of wages, we used the index provided by IBGE corresponding to the IPCA by State and quarter, based on prices in March/2022 (1st quarter/2022).

Author notes