Policy Research Working Paper 11116 Labor Market Scarring in a Developing Economy Stigma versus Lost Human Capital from Plant Closings in Mexico Francisco J. Arias Daniel Lederman Prosperity Vertical Vice Presidency May 2025 Policy Research Working Paper 11116 Abstract This paper estimates the magnitude of labor market scar- unobserved, time-invariant individual characteristics, the ring in a developing economy, a setting that has been impact of a plant closing declines from 11.9 to 8.2 percent. understudied by the labor scarring literature dominated These results imply that stigma in the labor market due to by advanced economies. The paper assesses the contribu- imperfect information about workers (captured by unob- tions of “stigma” versus “lost human capital,” which cause servable worker characteristics) accounts for 30.8 percent of earnings losses among displaced workers relative to non-dis- the average earnings losses, whereas lost employer-specific placed workers. The findings indicate that job separations human capital explains the remaining 69.2 percent. The caused by plant closings result in sizable and long-lasting paper explores the effects of job separations due to plant reductions in earnings, with an average decline of 7.5 per- closings on other labor market outcomes, including hours cent in hourly wages over a nine-year period. The estimate worked and informality, and provides estimates across gen- for one year after a plant closing is larger, at a decline of ders and levels of education. 10.8 percent. In a common sample, after controlling for This paper is a product of the Prosperity Vertical Vice Presidency. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at fariasvazquez@worldbank.org and dlederman@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Labor Market Scarring in a Developing Economy: Stigma versus Lost Human Capital from Plant Closings in Mexico Francisco J. Arias and Daniel Lederman1 Keywords: Labor market, job displacement, wages, education, difference in difference. JEL codes: I26, J01, J30, J60 1 Arias is a Senior Economist with the World Bank’s Macroeconomics, Investment and Trade Global Practice, and Lederman is Deputy Chief Economist for the Prosperity Vice-Presidency of the World Bank. We thank Diego Alvarez and Michelle Botello for their valuable research assistance. We gratefully acknowledge comments received from Joana Silva and Aart Kray on earlier versions of this paper. For information on the data or codes used for this study, please contact us at fariasvazquez@worldbank.org. All errors are the responsibility of the authors. The opinions expressed in this paper do not necessarily reflect the views of the World Bank, its Board of Directors, or the countries that they represent. 1. Introduction Labor reallocations and the flows of displaced workers reentering the workforce often accompany business cycles and other economic shocks. Yet to the extent that labor market distortions, including market failures, prevail in the labor market, displaced workers can experience declines in wages and other labor market outcomes such as informality well after the shock dissipates. Once formerly displaced workers reenter the workforce, wages and other job characteristics might decline in relation to what they would have received but for the job separation, a phenomenon commonly known as “labor market scarring.” Although there is a substantial literature on labor market scarring in advanced economies, there are few empirical estimates of this phenomenon in developing countries. There are a priori reasons why labor market scarring might not occur in developing countries, where a significant share of the labor force is informally employed. A key reason might be that the drivers of scarring, namely stigmatization of displaced workers and lost employer-specific human capital, might not be relevant in settings where informal employment is commonplace. Nonetheless, this paper aims to close this gap in the literature. Although earnings losses are the most common variable studied in the literature, the probability of working in the formal sector and hours of work are outcomes that deserve attention, especially in developing countries. This paper looks at how the Mexican labor market reabsorbs different types of displaced workers and provides empirical evidence of the magnitude of workers´ losses. We also explore the existence of heterogeneous effects by gender and education. Since stigma is associated with information asymmetries in the labor market, the rotative panel data from Mexico used in this paper enables an assessment of the quantitative contribution of stigma versus human capital losses driving labor market scarring in a developing country. That is, by comparing estimates of the wage losses of displaced workers across estimates with and without worker fixed 2 effects, we can infer the share of the wage losses that is likely due to labor market stigma. The rest of the paper is organized as follows. Section 2 discusses existing literature, highlighting the dearth of evidence from developing economies. Section 3 presents the data from the ENOE survey. Section 4 describes the econometric strategy, followed by the presentation of the results in Section 5. Section 6 provides a summary of the key findings and addresses potential areas for future research. 2. Related Literature In a recent literature review on unemployment scarring effects, Filomena (2021) reported 63 papers, of which only two were on developing economies. Cruces et al. (2012) look at the impact of youth unemployment and informality on adult labor market outcomes in Argentina and Brazil, finding that wage losses vanish over time. Petreski et al. (2017) found that short spells of unemployment in Macedonia reduce the probability of employment by 28%, and longer spells reduce the probability by 50%. Additionally, Eslava, Haltiwanger, Kugler and Kugler (2010) study the Colombian labor market and look at how trade liberalization affected workers displaced due to plant closings. The authors document the job reallocation that followed trade liberalization and analyze workers´ outcomes after they reentered the workforce. They find that displaced workers experience a reduction in wages and tenure (time in a job) and a lower probability of working in the formal sector within the first two years after their displacement. Yet they find that Colombian workers’ wages fully recover three years after a plant closing. For the case of Mexico, Kaplan, Martínez, and Robertson (2005) look at changes in formal wages following job displacement. They use administrative data from the Mexican Social Security Institute (IMSS in Spanish) to study the effect of job displacement on wages under different economic conditions. They find large wage losses when displacement occurs in periods of high 3 unemployment and in regions with lower economic activity. Their study is subject to two limitations, however. First, the Social Security records only collect information on formal employment and do not differentiate between informal employment, unemployment, and individuals who have left the labor force. Second, the Social Security records do not identify the type of job separation, and as the authors acknowledge, “the cause of separation is important because workers who leave voluntarily are more likely to have more positive economic prospects after they leave their current firm. Including voluntary separation would, therefore, bias the estimated effects of displaced workers upwards” (Kaplan et al., 2005, pp. 210). As mentioned, most of the literature looking at the effects of job displacement on workers’ labor outcomes has focused on the United States and developed economies. For instance, in one of the earliest studies on this topic, Ruhm (1991) finds long-lasting wage reductions for U.S. displaced workers of around 14 percent compared to wages of their non-displaced counterparts. Stevens (1997), using a longitudinal dataset, finds a drop in earnings of 30 percent at the time of displacement, which narrows to 10 percent after six years. In a seminal paper, using administrative data for workers in the state of Pennsylvania, Jacobson, LaLonde, and Sullivan (1993) look at the impact of large-scale layoffs on workers. They report earnings losses of more than 40 percent in the year of displacement, and six years after displacement earnings are still 25 percent below their pre-displacement level. In a related paper on the state of Connecticut, Couch and Placzek (2010) find that earnings losses for workers displaced through mass layoffs equal 32 percent at the time of job loss and six years later the reductions narrow to 14 percent. Davis and von Watcher (2011) calibrate a model of job search and unemployment fluctuations and find that earnings losses rise steeply with the unemployment rate at the time of displacement, concluding that tight labor market conditions at the time of displacement improve the medium- and long-term future earnings trajectory of displaced workers. In advanced economy settings, Bonikowska and Morissette (2012) look at the Canadian labor 4 market and find that better labor market conditions mitigate earnings losses for displaced workers. In periods of low unemployment, earnings losses are 10 percent, while in periods of high unemployment, they reach 21 percent. For Sweden, Eliason and Sotrrie (2006) use linked employer‐employee data and follow displaced workers over a pre-displacement period of 4 years and a post-displacement period of 11 years. They conclude that displaced workers experience both earnings losses and worsened labor market position not only during a transitory period of adjustment, but also in the long run. For Germany, Toulemon and Weber-Baghdiguian, (2016) find earnings losses of 10 percent but the effect vanishes after four years. They also find that other indicators of job quality are negatively affected. For the United Kingdom, Hijzen, Upward and Wright (2010) track workers for up to nine years after displacement and implement difference-in- differences estimators using propensity score matching methods. They report earnings losses during the first five years after displacement that range from 18 to 35 percent for workers whose firm closes, and from 14 to 25 percent for workers who exit a firm that suffers a mass layoff. 3. Data In this paper, we examine the impact of job displacement on labor market outcomes for Mexican workers, placing particular emphasis on the differential effects across gender and educational groups. The ENOE (Encuesta Nacional de Ocupación y Empleo) is a household survey that collects comprehensive data on labor force status, wages, employment, and demographic characteristics of the workforce. Conducted quarterly since 2005, it is representative at both national and state levels, encompassing approximately 100,000 households each quarter. The survey is designed as a rotating panel, where each household undergoes interviews for five consecutive quarters, with one-fifth of the sample being replaced quarterly. For the purposes of this analysis, we focus on individuals aged 15 to 65 years and utilize data from surveys with the extended questionnaire spanning from 2005 to 2019. 5 The survey employs an extended questionnaire to gather detailed information regarding individuals' employment histories, which allows us to explore the duration of workers’ losses due to job separations.2 Respondents are asked whether they have ever experienced a job separation that resulted in a temporary period of unemployment, and they must specify the reasons for their job separation. We categorize job separations into five mutually exclusive types: 1) plant closings, 2) quitting, 3) layoffs, 4) closing one's own business, and 5) other reasons including retirement. In our analysis, the benchmark category comprises workers who did not experience any job separation. This information shows that the effects of plant closings differ from other job separations, suggesting that plant closings are likely unrelated to worker characteristics. The selected sample consists of 1,673,293 employed workers. Some estimates use smaller samples depending on the specifications of the econometric models discussed below—specifically, the smallest sample has 34,043 high-educated workers. According to Table 1, 53.3% of workers did not experience a job separation; 8.4% reported experiencing a plant closing; 5.5% were laid off; 25.2% voluntarily resigned; 2.7% closed their own business; and 2.6% cited other reasons. Table 2 shows disparities in labor force status—employed, unemployed, or out of the labor force— between individuals who experienced a plant closing and those who faced other job separations. If plant closures are not related to worker characteristics, displaced workers should fare better than those terminated for cause. In fact, workers affected by plant closings are more likely to be employed, with 72.0% working at the time of the survey, compared to 61.0% of fired individuals. Additionally, 9.6% of workers from plant closings are unemployed, whereas this figure rises to 14.6% among laid-off workers. Table 2 shows that laid-off individuals are also more likely to exit 2 The extended questionnaire has a yearly frequency every second quarter. 6 the labor force, with 24.5% currently out of work compared to 18.5% of those affected by a plant closing. Notably, individuals who voluntarily quit their jobs —whether by quitting or closing their business— are less likely to be employed, with 51.6% of those who quit and 49.8% of those who closed their business reporting that they are out of the labor force. Table 3 presents a transition matrix linking types of displacement to occupational categories. It shows that workers from plant closings and layoffs have similar chances of returning to salaried jobs, with about 80% securing such positions and 15% moving into self-employment. In contrast, among those who voluntarily quit, 73.9% are now in salaried roles, and 18.3% have transitioned to self-employment. The descriptive evidence indicates that plant closures are less likely to be associated with worker characteristics compared to cases of firing or voluntary departures. 4. Econometric Approach The paper studies wages, hours of work, and formality status in the aftermath of job separations. After estimating the average impact of job displacement on these labor outcomes, we assess whether the impacts are transitory or permanent, and present results by educational groups and gender. Indeed, if stigmatization is the cause of labor market scarring, one would expect that the effects are transitory as new employers become familiar with a worker’s unobserved characteristics. If scarring results from the loss of employer-specific human capital that can be acquired through job experience, the effects are likely temporary. Thus, the duration of the effects of plant closings on labor market outcomes is uninformative about whether stigma or lost human capital explains the losses. The analyses in this paper focus on workers displaced due to plant closings. We should be cautious, as the relevant coefficients might be subject to endogeneity or selection bias to the extent that displaced workers are more likely to have unobservable characteristics that make them less 7 productive than the benchmark category of non-displaced workers. Nonetheless, the comparison of the estimates without and with controls for individual worker unobserved characteristics allows us to assess the contributions of lost human capital versus stigma to the average scarring effect. We include informal workers in our analysis, which is crucial, as they account for more than half of Mexico’s total employment in our data. To assess the impact of job displacement on current labor market outcomes, we analyze different forms of job displacement in comparison to the baseline category of non-displaced workers. The first model specification captures the average effects of distinct types of job displacement. The second specification examines the lagged effects of job displacement on current labor market outcomes, allowing us to distinguish between temporary and permanent effects. As some effects decrease over time, workers may see their labor outcomes become similar to those in the non- displacement group. The final specification leverages the rotating panel structure of the ENOE, which tracks individuals over a one-year period to estimate a difference-in-differences (DID) specification. This approach controls for a time-invariant individual heterogeneity term associated with stigmatization when worker characteristics remain unobserved. More formally, we first estimate the following specification: , + ∑ , + , , = + ′ (1) where Y represents the labor market outcome (log of real wages,3 hours of work, or formality status). We include a time-fixed effect, , which controls for economywide macroeconomic conditions. The matrix X includes a set of covariates, including gender, marital status, education, age, age squared, as well as state and industry fixed effects. The variable of interest is TD, which 3 We use the consumer price index to normalize nominal wages to 2005 prices. 8 reflects the type of displacement, including plant closings, lay off, quitting, and closed own business. The omitted category is non-displacement. We explore the existence of temporary and permanent effects of job displacement on current labor market outcomes. We modify the above specification by adding lagged indicators to the type of displacement. ,=9 ′ , + ∑ , ,,− + , , = + (2) ,=0 In equation 2, for example, for workers who experience a plant closing event three years prior to the survey interview, the coefficient α reflects the difference in the current labor outcome, Yi,t, in comparison to workers in the non-displacement category. Difference in Differences (DID) Equations 1 and 2 ignore unobservable worker characteristics that might be correlated with the probability of experiencing a plant closing and subsequent labor market outcomes. Since stigma due to imperfect information in the labor market is one of the mechanisms that produce scarring effects, the inclusion of individual worker fixed effects generates estimates of scarring without the stigma effect. To control for the effect of stigma, we estimate a difference-in-differences (DID) specification using the ENOE’s rotating panel structure that follows approximately 90,000 workers for one year. The difference between the DID and the previous OLS model can be interpreted as the effect of stigma, as opposed to the pure effect of lost human capital linked to the previous employer. The DID specification can be written as follows: , + , (3) , = + + ∑ + ∑ + ′ 9 We observe workers over two time periods, during which some experience job displacement and others do not. In the survey’s first interview (labeled t=1) and a follow-up one year later (labeled t=2). In equation 3, we have the time dummy, Dt which equals one if t=2 and zero if t=1; the indicator variable displacement with the omitted category of non-displacement (TDd). Here the coefficient of interest is which is the impact of plant closing – relative to non- displacement – on the labor market outcomes, Yi,t. After taking the first differences, we obtain equation 4, which removes the unobserved time invariant individual term (γi) as well as other time invariant variables. Hence, we estimate the following specification: ∆,2 = 1 + ∑ + ∆′,2 + ∆,2 (4) Unfortunately, the DID approach might still not be free of omitted variable bias, because the survey data do not provide information about employer-worker pairs. This might be problematic if the quality of the match (as opposed to the unobserved quality of the individual worker) is systematically correlated with both the probability of a plant closing and subsequent labor market outcomes of the worker. We return to this issue in the concluding section of the paper. 5. Results The findings in Table 4 indicate that workers displaced due to plant closings experience sizable and long- lasting wage reductions, with an average decline of -7.5 percent compared to non-displaced workers. Additionally, these displaced workers have an average increase of 1.4 hours worked per week and 2.6 percentage points in the probability of employment in the formal sector compared to their non-displaced counterparts. As shown in Table 4, plant closing have the largest average drop in wages, followed by laid- off workers with a decline of 6.9 percent, closed-own-business with 5.3 percent, and quit with 2.3 percent. In Figures 1 to 9, we decompose the average impacts of Table 1 over a nine-year period. 10 Figures 1 to 9 display lagged coefficients with confidence intervals, showing the impact of different types of job displacement on real wages, weekly working hours, and employment status in the formal sector. The results indicate that a plant closing event has a lasting negative impact on real wages, with a decline of 10.8 percent observed one-year after displacement, decreasing to 5.4 percent after four years, and further narrowing to 2 percent after nine years (Figure 1). Additionally, the increase in working hours among displaced workers is also persistent; these individuals work an average of 2.1 extra hours per week one year after displacement, which remains at an additional 0.7 hours after nine years compared to non-displaced workers (Figure 2). Plant closings have a short-term negative impact on the probability of working in the formal sector. Within the year of displacement, plant closing workers have a 2.1 percentage points reduction in the probability of being formal, but a year after the probability increases to 2.7 percentage points— above the probability of non-displaced workers— to reach 5.6 pp. after nine years (Figure 3). This finding is interesting for the discussion about remaining potential omitted variable bias, because if the quality of the match between employers and workers were an important determinant of worker performance after a separation due to a plant closing, then why should it not show up as a lasting negative impact on the probability of being formally employed? In the case of workers who quit, we find ambiguous results, as they experience, on average, a slight decline in wages. Still, once we disaggregate this impact over time, we observe that in the ninth year since quitting, they have higher wages of up to 10 percent relative to non-displaced workers (Figure 4). In terms of work hours, the impact disappears gradually for quitting workers; however, during the first three years after quitting, they work between 1.4 and 2.5 additional hours than non- displaced workers (figure 5). Finally, the estimates show that Q-workers have a permanent increase (range between 7 and 11 percentage points) in the probability of working in the formal sector compared to the non-displacement group (figure 6). Regarding workers who close their businesses, 11 they have a transitory wage decline that vanishes entirely in the fourth year (Figure 7), and the coefficients on weekly hours of work and probability of working in the formal sector vanish within the nine years (Figures 8 and 9). Results by Gender and Educational Categories During the sample period, unemployment rates in Mexico varied systematically across levels of education. Workers with more educational attainment had higher unemployment rates than workers with low education. Similarly, before the financial crisis of 2009 and after 2015, women consistently had higher unemployment rates than men (Figures 10 and 11). We find no evidence of gender-differentiated effects of plant closings. The results by gender indicate that the Mexican labor market reabsorbs displaced men and women similarly; however, point estimates suggest a slightly different effect of plant closing on wages, with a larger impact for males (a reduction of 8.0 percent) than for females (a reduction of 6.7 percent).4 We further explore this difference by estimating lagged coefficients of plant closing on real wages. The results in Figure 15 confirm the existence of short-lived gender differences, which operate only in the first two years after the plant closing event has occurred, notwithstanding, after the second year the recovering of wages follow a similar path for men and women with no statistically significant differences in the magnitude of the coefficients. In terms of hours of work, there is no gender-differentiated effect, as both men and women experience similar increases, which might be a coping mechanism to compensate for reduced wages. The results concerning the effect of plant closings on labor outcomes are also different across educational categories. Table 5 shows the findings. Highly educated workers have an average wage decline of 12 percent. As Figure 12 shows, the wages of this group initially dropped by 16 percent 4 In the Annex, tables A.1 and A.2 present average impacts by gender. 12 relative to their non-displaced counterparts, narrowing to 11.6 percent in the third year and reaching 6 percent in the ninth year. In contrast, for workers with low education, defined as having less than a high school degree, the average decline is 5.3 percent. For this group, the lagged decomposition shows that wages after one year drop by 9.0 percent relative to the control group with the same level of education. The gap narrows to 5 percent in the second year and vanishes after the sixth year. This finding confirms the intuition that scarring among low-skilled and informal workers is less punishing than for formal and more educated workers (Figure 12). Displaced workers due to plant closings have a permanent increase in work hours, and the impact on low-educated workers is stronger than for highly educated workers. For the low-education group, there is an impact of 2.8 additional hours per week, which is long-lasting over the nine-year horizon of the analysis. For the high-education group, the estimates point to an initial impact of 1.7 additional hours of work, with a gradual reduction over time that reaches 0.7 additional hour at the end of the nine-year horizon (Figure 13). This result may reflect that working more hours is a coping mechanism for low-educated individuals against the drop in wages. Finally, there is a notable difference in the impact on the probability of working in the formal sector (Figure 14). For the high-education group, there is a transitory negative impact, with an initial drop of -11.0 pp., which narrows to -3.0 pp. in the third year and vanishes after the sixth year. The opposite occurs for the low-education group, with a long-lasting increase in the probability of being formal, ranging between 7 and 10 pp. over the nine-year period. This differentiated impact across education categories is partially explained because low-education workers who experienced a plant closing were more likely to be formal than otherwise similar low-education workers (see Table A3). 13 Stigma versus Employer-Specific Human Capital Losses The difference-in-differences estimation adds to the empirical specification an unobserved time- invariant individual term that captures the stigma component of the scarring impact on wages. By comparing the results in Tables 4 and 5 with those in Table 6, we can obtain the relative magnitudes of both stigma and human capital losses that contribute to wage declines following a plant closing event. The DID results presented in Table 6 indicate that workers displaced by a plant closure in the previous year experience an average wage reduction of 8.2 percent, along with a moderate increase of 0.33 hours worked per week, and reduction of 9.0 percent in the probability of employment in the formal sector. Consequently, these results imply that stigma in the labor market due to imperfect information about workers (captured by unobservable worker characteristics) accounts for 23.6 percent (of the 10.8 percent estimated without individual worker fixed effects) of the average earnings losses whereas lost human capital accounts for the remaining 76.4 percent.5 The DID estimation reveals that the effect of plant closing on skilled and unskilled workers is heterogenous. Highly educated workers experience a wage reduction of 14.2 percent, which is larger than the 6.7 percent decline observed for low-educated workers. These findings also suggest that stigma plays a larger role in explaining the reduction of wages for low-educated workers than for their highly educated counterparts. In the estimations that do not account for individual worker fixed effects, low-educated workers saw a wage decline of 8.7 percent one year after the plant closure. In contrast, highly educated workers experienced a decrease of 16.0 percent. Consequently, the results from Table 6 and figure 12 indicate that for unskilled workers, stigma accounts for 22.4. percent of the wage decline. In contrast, stigma explains 11.1 percent of the skilled workers wage reduction. 5 Human capital loss is given by the ratio of the DID over the OLS coefficient (8.22 / 10.77), that is, 76.4 percent of the wage decline of 10.8% estimated without individual worker fixed effects. The stigma component is the remaining 23.6 percent. 14 The DID estimates lower the average increase in weekly work hours from 1.4 to 0.33 and increase the probability of working in the formal sector from 2.6 percent to 9.0 percent. There are differences across educational categories. The DID estimates a negligible impact on the weekly hours of work of highly educated workers and a slight decrease for low-educated workers (0.58 hours). In addition, the estimate indicates that there is a marked difference of plant closing on the probability of working in the formal sector, with the high education group exhibiting a reduction of -14.5 percentage points relative to non-displaced workers in the same educational group. In contrast, the low-education group has decrease of -8.0 percentage point relative to non-displaced workers in the same educational category. The DID estimation rejects the existence of gender-differentiated effects on wages, as both men and women have similar wage reductions of 8.4 and 8.1 percent, respectively. We observe no impact in weekly work hours for women and men. To further study potential biases in the DID estimates, figure 16 presents the estimates of the industry fixed effects on the probability of the occurrence of a plant closing event. Manufacturing is associated with a 14 percent higher probability of experiencing a plant closing, while agriculture is 36 percent less likely to experience this type of job separation. The results from figure 10 indicate the importance of controlling for workers’ industry of employment at the time of job separation. We re-estimate specification 2 with a subsample of workers for whom we can control for their industry of employment at the time of displacement and confirm that the results are robust to the inclusion of the industry of employment at the time of job displacement.7 Table A4 indicates that workers who experienced a plant closing event were more likely to be formally employed at that time than those who did not face displacement. To the extent that higher quality workers and higher quality employer-employee matches are in the manufacturing sector where large enterprises operate rather than in agriculture, combined with the results regarding formality status of workers, these results lend further credence to the speculation that the DID estimates in particular are unlikely to be 15 severely biased due to the omission of employer-employee fixed effects. Nonetheless, this remains a fertile area for future research. Comparing Estimates across Common Samples Since the estimation samples used in the DID are different from those used in the estimation of worker trajectories, it is worth asking whether the differences in the samples might explain the differences in the point estimates. The relevant results for the impact of plant closing on hourly wages appear in Table 7. Column 1 shows the results for the DID estimates for the various groups of workers, and Column 2 shows the results for the specification that excludes worker fixed effects. Column 3 shows the number of observations in the common samples. For the sake of completeness, Column 4 reports the estimates without worker fixed effects, followed by its number of observations in Column 5. The estimated effects of plant closings on hourly wages without fixed effects are similar across the two samples. The point estimates are 11.9 and 10.8 percent, the latter being the estimate for the 7 The ENOE collects information on the industry of employment at the time of job displacement for only a subsample of workers—those who experienced a job termination within the last two years of the survey interview. 16 large sample. Nevertheless, comparing the 11.9 percent decline in wages after one year since the plant closing to the 8.2 percent decline estimated with individual fixed effects in the DID framework implies that 30.8 percent of the decline is due to stigma, with the remaining portion due to lost employer-specific human capital. Moving down Table 7, the estimated contributions of stigma to labor-market scarring in terms of reductions in hourly wages are virtually identical across the two samples for the samples of all workers and male workers. The differences of the estimates across samples are notably higher for the samples of female workers and for the education groups. Importantly, the constant sample estimate for low-education workers with the common sample is slightly larger in magnitude than that with the largest plausible sample, as it increases from 8.7 to 9.2 percent. This finding is consistent with the view that scarring is lower among low-education workers than for high- education workers. Furthermore, the share of the scarring effect that is due to stigma is 27.0 percent in the small sample (of 96,732 observations) whereas if is 22.4 percent in the large sample (of 932,395 observations). This finding suggests that labor scarring among low-education workers, who are more likely to be informally employed, is not only much lower than for the other groups, but the share that is due to stigma might also be much lower. 6. Summary of Findings and Future Research The contribution of this paper is twofold. First, it expands the advanced economy predominant literature of labor market scarring to a developing country with a large prevalence of informal employment. Second, it explores the relative magnitude of the two components of scarring, namely the stigmatization of displaced workers and lost employer-specific human capital. The dataset allowed us to use plant closings as a proxy for job separations that are in principle unrelated to workers’ characteristics. 17 The results suggest that displaced workers from plant closings experience sizable and long-lasting wage reductions, with an average decline of -7.5 percent over a nine-year period. Disaggregating over a nine-year period, our robust estimates in Table 7 found that after the first year of plant closing, wages drop by 11.9 percent with respect to non-displaced workers, implying that stigma in the labor market due to imperfect information about workers, captured by unobservable worker characteristics, accounts for 30.8 percent of the average earnings losses, whereas lost human capital accounts for the remaining 69.2 percent. The results confirm the heterogeneous effects of plant closing on skilled and unskilled workers. Highly educated workers have a reduction in wages of 14.2 percent, two times the decline for low- educated workers of 6.7 percent. Our results indicate that scarring is less punishing among low- skilled and possibly informal workers, and that the role of stigma for these workers plays a more significant role. Stigma explains 27.0 percent of the wage losses for low-educated workers (and human capital the remaining 73.0 percent). In contrast, for the highly educated, stigma explains 33.6 percent of the wage losses (and human capital the remaining 66.4 percent). Finally, it is worth acknowledging that the DID estimates do not control for another potential source of omitted-variable bias: the quality of the match between employers and workers. That is, it is possible that the probability of a plant closing is systematically correlated with the quality of the match between the employer and its workers. Since the survey data, which nonetheless covers both informal and formal workers, does not offer information on employer-worker relationships, we are unable to control for this potential source of omitted variable bias. Future research should attempt to re-estimate the DID while controlling for employer-employee fixed effects, which requires administrative data, which would unfortunately leave out the informal labor market. 18 References Bonikowska, A. and Morissette, R. 2012. “Earnings Losses of Displaced Workers with Stable Labour Market Attachment: Recent Evidence from Canada”, Statistics Canada, Analytical Studies Branch. URL: https://www150.statcan.gc.ca/n1/en/catalogue/11F0019M2012346 Couch, K. A. and Placzek, D. W. 2010. “Earnings Losses of Displaced Workers Revisited”, American Economic Review, 100 (1), pp. 572-589. DOI: 10.1257/aer.100.1.572 Cruces, G; Ham, A. and Viollas, M. 2012. “Scarring effects of youth unemployment and informality: Evidence from Argentina and Brazil”, Center for Distributive, Labour and Social Studies. https://conference.iza.org/conference_files/YULMI2012/viollaz_m8017.pdf. Davis, S. and Von Wachter, T.M. 2011. “Recessions and the Costs of Job Loss”, NBER Working Paper 17638, https://www.nber.org/papers/w17638 Eliason, M. and Storrie, D. 2006. "Lasting or Latent Scars? Swedish Evidence on the Long-term Effects of Job Displacement", Journal of Labor Economics, 24(4), pp. 831–856. DOI: https://doi.org/10.1086/506487 Eslava, M., Haltiwanger, J., Kugler, A. and Kugler, M. 2010. “Trade liberalization and worker displacement: Evidence from trade reforms in Colombia”, Mimeo. URL: https://conference.iza.org/conference_files/worldb2011/kugler_a157.pdf Filomena, M. 2021. "Unemployment Scarring Effects: A Symposium on Empirical Literature," Working Papers 453, Universita' Politecnica delle Marche (I), Dipartimento di Scienze Economiche e Sociali, Italy. Hijzen, A., Upward, R. and Wright, P. 2010. "The Income Losses of Displaced Workers", The Journal of Human Resources, 45(1), pp. 243–269. DOI: 10.3368/jhr.45.1.243 INEGI .2007. “Cómo se hace la ENOE. Métodos y Procedimientos.” Available at: https://en.www.inegi.org.mx/contenidos/productos/prod_serv/contenidos/espanol/bvinegi/produc tos/nueva_estruc/702825190613.pdf Jacobson, R., LaLonde, R. and Sullivan, D. 1993. “Earnings Losses of Displaced Workers”, American Economic Review, 83, pp. 685-710. URL: https://www.jstor.org/stable/2117574 Kaplan, D., Martinez, G. and Robertson, R. 2005. “What Happens to Wages After Displacement?”, Journal of the Latin American and Caribbean Economic Association, 5(2), pp. 197-234. URL: https://mpra.ub.uni-muenchen.de/3079/ Messina J. and Silva, J. 2018. Wage Inequality in Latin America. Understanding the Past to Prepare for the Future. Washington, DC: The World Bank Group. Petreski, M; Mojsoska-Blazevski, N. and Bergolo, M. 2016. "Labor-Market Scars When Youth Unemployment Is Extremely High: Evidence from Macedonia," IZA Discussion Papers 10342, Institute of Labor Economics (IZA). Ruhm, C. 1991. “Are Workers Permanently Scarred by Job Displacement?”, American Economic Review, 81, pp. 319-324. URL: https://www.jstor.org/stable/2006805. 19 Stevens, A. H., 1997. “Persistent Effects of Job Displacement: The Importance of Multiple Job Losses”, Journal of Labor Economics, 13, pp. 165-188. DOI: https://doi.org/10.1086/209851. Toulemon, L. and Weber-Baghdiguian, L. 2016. “Long-term Impact of Job Displacement on Job Quality and Satisfaction: Evidence from Germany”, Paris School of Economics. Working Paper No. 2016-32. URL: https://halshs.archives-ouvertes.fr/halshs-01418183/. 20 Tables and Figures Table 1: Summary Statistics Observations (total) 2,707,562 Observations (employed workers) 1,673,293 Share of men 47.8 Average age 36.0 Type of displacement Share of married individuals 46.7 Plant Closing 8.4 Average years of education 9.6 Lay off 5.5 Size of locality (share of individuals) Resign 25.2 More than 100,000 60.4 Closed Own Business 2.7 Between 15,000-99,999 11.7 Retired 2.6 Between 2,500-14,999 12.0 Other 2.2 Less than 2,500 15.9 Non displaced workers 53.3 Weekly hours 43.1 Length of unemployment (months) Hourly wage 32.5 Plant Closing 3.2 Labor force composition Lay off 2.8 Employed 61.8 Resign 2.8 Unemployed 2.7 Closed Own Business 3.2 Out of labor force 35.5 Retired 4.7 Employed status Other 3.0 Employee 69.8 Measures of informality status Employer 5.0 Formal Employment 47.5 Self-employed 20.0 With access to health 41.5 Unpaid worker 5.1 With a secondary job 4.2 Source: ENOE 2005-2019. Descriptive statistics of labor market variables were calculated from a sample of 1,673,293 employed workers. Table 2. Types of job displacement and current labor force status Current labor force status Employed Unemployed Out of labor force Plant 72.0 9.6 18.5 Closing Type of job Lay off 60.9 14.6 24.5 displacement Resign 44.4 4.0 51.6 Closed own 44.8 5.4 49.8 business Source: ENOE 2005-2019. Note: “Current” means at the time of the survey. Number of observations: 2,189,112. 21 Table 3. Type of job displacement and current employment status Current employment status Unpaid Employee Employer Self-employed worker Plant 79.9 3.0 15.6 1.5 Closing Type of job Lay off 80.7 2.6 14.5 2.3 displacement Resign 73.9 5.3 18.3 2.6 Closed own 58.5 8.5 29.9 3.2 business Source: ENOE 2005-2019. Note: “Current” means at the time of the survey. Number of observations: 1,673.293. Table 4. Estimates of job displacements on labor market outcomes (1) (2) (3) Log of hourly wage Hours of work Formality Plant Closing -0.0749*** 1.4492*** 0.0262*** (0.006) (0.159) (0.004) Lay off -0.0687*** -0.1206 -0.0529*** (0.009) (0.174) (0.005) Quit -0.0229*** 1.7405*** 0.0401*** (0.005) (0.149) (0.004) Closed Own Business -0.0532*** 0.3016 -0.0714*** (0.005) (0.197) (0.007) Observations 1,229,280 1,579,584 1,671,031 R-squared 0.3239 0.1302 0.2644 Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.10. All regressions control for year of education, gender, marital status, age, age square, survey period, state fixed effects, and industry fixed effects. Standard errors are clustered at the state level. The omitted category is non-displaced workers. 22 Figure 1. Plant closing coefficients on log hourly wages 0.00 -0.020 -0.05 -0.054 -0.10 -0.108 -0.15 0 1 2 3 4 5 6 7 8 9 Years since plant closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2). The number of observations is 1,228,705. Figure 2. Plant closing coefficients on weekly hours of work 3.0 2.5 2.0 1.5 2.1 1.0 0.5 0.0 0 1 2 3 4 5 6 7 8 9 Years since plant closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2). The number of observations is 1,578,798. Figure 3. Plant closing coefficients on formality status 0.10 0.05 0.027 0.056 0.00 -0.021 -0.05 0 1 2 3 4 5 6 7 8 9 Years since plant closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2). The number of observations is 1,578,798. 23 Figure 4. Quitting coefficients on log hourly wages 0.15 0.10 0.05 0.02 0.00 -0.01 -0.05 0 1 2 3 4 5 6 7 8 9 Years since quitting Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2). The number of observations is 1,226,493. Figure 5. Quitting coefficients on weekly hours of work 4 3 2.5 2 1.4 1 0 -1 0 1 2 3 4 5 6 7 8 9 Years since quitting Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2). The number of observations is 1,575,408. Figure 6. Quitting coefficients on formality status 0.15 0.1 0.11 0.07 0.05 0 0 1 2 3 4 5 6 7 8 9 Years since quitting Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2). The number of observations is 1,575,408. 24 Figure 7. Closed own business coefficients on log hourly wages 0.10 0.05 0.00 -0.05 -0.05 -0.10 -0.12 -0.15 -0.20 0 1 2 3 4 5 6 7 8 9 Years since closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2). The number of observations is 1,229,139. Figure 8. Closed own business coefficients on weekly hours of work 2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 0 1 2 3 4 5 6 7 8 9 Years since closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2). The number of observations is 1,579,389. Figure 9. Closed own business coefficients on formality status 0.05 0.00 -0.05 -0.08 -0.10 -0.11 -0.15 0 1 2 3 4 5 6 7 8 9 Years since closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2). The number of observations is 1,579,389. 25 Figure 10. Unemployment Rates by Educational Attainment (% of the labor force) Less than a Primary School Diploma Primary Shool Diploma 8 Middle School Diploma More than a high school diploma 7.17 7 Total 6.90 6.15 6 5.75 5 3.90 4 3.39 3.26 3 3.12 2.27 2 1.36 1 0 Note: Data from the Mexican Institute of Statistics (INEGI, https://en.www.inegi.org.mx/temas/empleo/). Unemployed individuals are those without a job and actively looking for work the week before the survey interview. Figure 11. Unemployment Rates by Gender (% of the labor force of each group) Total Men Women 7 6.1 5.2 4.3 3.4 2.5 Note: Data from the Mexican Institute of Statistics (INEGI, https://en.www.inegi.org.mx/temas/empleo/). Unemployed individuals are those without a job and actively looking for work the week before the survey interview. 26 Table 5. Estimates of Job Displacement on Labor Market Outcomes by Educational Categories Log of hourly wage Hours of work Formality Primary and Primary and Primary and Tertiary Tertiary Tertiary Secondary Secondary Secondary Plant Closing -0.0536*** -0.1218*** 1.4798*** 0.8475*** 0.0461*** -0.0480*** (0.006) (0.006) (0.177) (0.125) (0.004) (0.005) Lay off -0.0624*** -0.1229*** -0.1674 -0.0864 -0.0377*** -0.1099*** (0.010) (0.007) (0.203) (0.137) (0.005) (0.007) Quit -0.0120** -0.0412*** 1.9146*** 0.9811*** 0.0545*** -0.0081 (0.006) (0.006) (0.159) (0.127) (0.004) (0.005) Closed Own Business -0.0285*** -0.1266*** 0.0998 0.6714*** -0.0561*** -0.1265*** (0.005) (0.012) (0.220) (0.238) (0.007) (0.009) Observations 932,874 296,406 1,157,290 422,294 1,221,797 449,234 R-squared 0.2072 0.2782 0.1300 0.1470 0.2277 0.1218 Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.10. All regressions control for year of education, gender, marital status, age, age square, survey period, state fixed effects, and industry fixed effects. Standard errors are clustered at the state level. The omitted category is non-displaced workers. Figure 12. Plant closing coefficients on log hourly wages: high and low education 0.05 0.00 -0.05 -0.09 -0.10 -0.15 -0.20 -0.16 Tertiary education -0.25 Primary and Secondary Education 0 1 2 3 4 5 6 7 8 9 Years since plant closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2) separated by educational category. The number of observations is 932,395 for the low education sample and 296,310 for the high education sample. 27 Figure 13. Plant closing coefficients on weekly hours of work: high and low education 5.0 Tertiary education Primary & Sec education 4.0 3.0 2.8 2.7 2.0 1.7 1.0 0.7 0.0 0 1 2 3 4 5 6 7 8 9 Years since plant closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2) separated by educational category. The number of observations is 932,395 for the low education sample and 296,310 for the high education sample. Figure 14. Plant closing coefficients on formality status: high and low education 0.15 0.07 0.10 0.10 0.05 0.00 -0.05 -0.10 -0.06 -0.03 -0.11 Tertiary education -0.15 Primary & Sec. education 0 1 2 3 4 5 6 7 8 9 Years since plant closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2) separated by gender. The number of observations is 753,126 for the male sample and 296,310 for the female sample. 28 Figure 15. Plant closing coefficients on log hourly wages: gender estimates 0.05 Male Female 0.00 -0.05 -0.10 -0.15 0 1 2 3 4 5 6 7 8 9 Years since plant closing Source: Authors’ calculations based on ENOE survey data from 2005 to 2019. Notes: This figure presents the estimates of coefficients from the model described by equation (2) separated by gender. The number of observations is 753,126 for the male sample and 296,310 for the female sample. Table 6. Difference-in-Differences Estimates (1) (2) (3) Log of hourly Obs Hours of work Obs Formality Obs wage All sample -0.0822*** 128,629 0.3371 196,438 -0.0902*** 213,624 (0.012) (0.237) (0.009) Male -0.0847*** 78,216 0.2668 119,698 -0.0995*** 129,907 (0.011) (0.297) (0.008) Female -0.0811*** 50,413 0.0008 76,740 -0.0703*** 83,717 (0.025) (0.528) (0.020) High Education -0.1422*** 31,897 -0.8955 55,153 -0.1453*** 60,740 (0.030) (0.535) (0.019) Low Education -0.0676*** 96,732 0.5815** 141,285 -0.0809*** 152,884 (0.011) (0.270) (0.008) Note: Robust standard errors in parentheses. ***p<0.01, **p<0.05, *p<0.10. All regressions control for marital status, survey period, state fixed effects, and industry fixed effects. Standard errors are clustered at the state level. The omitted category is non-displaced workers. 29 Figure 16. Industry fixed effects on probability of plant closing 0.50 0.35 0.20 0.05 -0.10 -0.25 -0.40 -0.55 Note: omitted category is other services. The number of observations is 273,736. Table 7. Plant closings and log hourly wages: Comparison of DID Results with Estimates without Worker Fixed Effects across Samples (1) (2) (3) (4) (5) (6) (7) Estimates without Fixed Effects: Estimates without Fixed Effects: Full DID Common Sample Sample After one- After one- Full Sample year Share Due Observation year Share Due Observation displacemen to Stigma s displacemen to Stigma s t t All sample -0.0822*** -0.1188*** 30.8% 128,629 -0.1077*** 23.7% 1,228,705 (0.012) (0.013) (0.009) Male -0.0847*** -0.1234*** 31.4% 78,216 -0.1156*** 26.7% 753,126 (0.011) (0.013) (0.008) Female -0.0811*** -0.1215*** 33.3% 50,413 -0.0936*** 13.4% 475,579 (0.025) (0.024) (0.012) High -0.1422*** Education -0.2143*** 33.6% 31,897 -0.1599*** 11.1% 296,310 (0.030) (0.030) (0.009) Low -0.0676*** Education -0.0926*** 27.0% 96,732 -0.0871*** 22.4% 932,395 (0.011) (0.013) (0.010) Notes: DID = Differences in Differences. Robust standard errors in parentheses. ***p<0.01, **p<0.05, *p<0.10. All regressions control for marital status, survey period, state fixed effects, and industry fixed effects. Standard errors are clustered at the state level. The omitted category is workers who were never displaced. 30 Annex Table A1. Male sample: estimates of job displacement on labor market outcomes (1) (2) (3) Log of hourly wage Hours of work Formality Plant Closing -0.0802*** 0.7610*** 0.0101*** (0.005) (0.140) (0.004) Lay off -0.0724*** -0.6139*** -0.0611*** (0.010) (0.182) (0.005) Quit -0.0209*** 1.7296*** 0.0421*** (0.006) (0.109) (0.005) Closed Own Busines -0.0527*** 0.3841** -0.0681*** (0.006) (0.176) (0.007) Retired -0.1364*** -3.6497*** -0.2221*** (0.013) (0.346) (0.012) Other -0.0889*** -0.9931*** -0.0920*** (0.006) (0.181) (0.005) Observations 753,619 966,028 1,018,629 R-squared 0.3197 0.1179 0.2818 Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.10. All regressions control for year of education, gender, marital status, age, age square, survey period, state fixed effects, and industry fixed effects. Standard errors are clustered at the state level. The omitted category is non-displaced workers. 31 Table A2. Female sample: estimates of job displacement on labor market outcomes (1) (2) (3) Log of hourly wage Hours of work Formality Plant Closing -0.0667*** 1.7968*** 0.0308*** (0.007) (0.215) (0.006) Fired -0.0701*** -0.0953 -0.0551*** (0.007) (0.205) (0.005) Quit -0.0291*** 1.0244*** 0.0167*** (0.006) (0.185) (0.004) Closed Own Business -0.0552*** -0.8703*** -0.1025*** (0.009) (0.280) (0.007) Retired -0.0830*** -4.2678*** -0.2740*** (0.023) (0.491) (0.012) Other -0.0855*** -1.0473*** -0.0793*** (0.013) (0.310) (0.007) Observations 475,661 613,556 652,402 R-squared 0.3383 0.0666 0.2813 Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.10. All regressions control for year of education, gender, marital status, age, age square, survey period, state fixed effects, and industry fixed effects. Standard errors are clustered at the state level. The omitted category is non- displaced workers. Table A3. Formality rate / Log of hourly wage (at t=0) Plant closing Non-displaced Low-educated 51.1 / 7.379 42.7 / 7.372 High-educated 71.8 / 7.896 78.9 / 8.109 Source: Authors calculation. Note: the sample is restricted to common sample of Table 7 (N=139,534). Formality and log hourly wage in the first interview of the rotative panel (t=0). 32 Table A4. Observable characteristics: plant closings versus non-displacement groups % Years Hours of % Female Age Formal Log wage Education work (at t=0) Non- displaced 42 37.2 10.00 0.51 7.52 42.13 Plant closing 36 38.3 9.97 0.54 7.48 46.65 Source: Authors calculations using ENOE 2005-2019. Note: Formality rate in the survey’s first interview. Formal status t=0 refers to the time of displacement. The number of observations is 170,779. Table A5. Distribution of employment across industries: Plant closings versus non-displacement groups. Non- Plant displaced closing Agriculture 10.2 4.9 Mining and electricity 1.0 1.2 Manufacturing 15.1 21.5 Construction 7.7 9.4 Wholesale 19.2 19.7 Hotels and restaurants 7.1 7.0 Transport and 4.4 7.8 communications Professional and financial 6.1 8.7 services Social services 12.5 5.8 Other services 16.6 14.1 Source: Authors’ calculations using ENOE 2005-2019. The number of observations is 170,779. 33