Minimum wage violations in a dual labor market with multiple floors

Transcription

1 Minimum wage violations in a dual labor market with multiple floors Andrés Ham Department of Agricultural and Consumer Economics University of Illinois at Urbana-Champaign This version: January 2015 Abstract Minimum wage compliance is studied by counting how many workers are paid below this floor. While useful, it is unable to identify the degree of underpayment. Recently, a family of violation indices has overcome this issue. I estimate these measures for Honduras, a dual labor market with multiple wage floors. First, I quantify industry-level violations by location and coverage status. Then, I estimate whether higher minimum wages affect non-compliance. The results reveal substantial variation in violations and that raising minimum wages leads to more underpaid workers but does not affect the degree by which they are underpaid. Keywords: minimum wages, compliance, incidence and depth of violations, dual labor markets, Honduras JEL Classification: J38, J42, O54 This paper was the fruit of attending the Reforming Minimum Wage and Labor Regulation Policies in Developing Economies Conference, held at Beijing Normal University in China from October 18-19, I would like to thank Ravi Kanbur, Haroon Bhorat, Li Shi, Tim Gindling, Lucas Ronconi, and Leonardo Bonilla for engaging discussions that contributed to this research; as well as conference participants for valuable feedback to a related study on minimum wages in Honduras. All remaining errors and omissions are entirely my sole responsibility and do not necessarily reflect the views of the University of Illinois. Contact: hamgonz2@illinois.edu. Mailing address: Department of Agricultural and Consumer Economics, 405 Mumford Hall, 1301 W. Gregory, Urbana, IL MC-710.

2 1 Introduction Most developing countries protect their workers by passing labor regulations, which generally include a minimum wage. However, the existence of legislation does not guarantee compliance (Basu et al., 2010). Therefore, the fundamental problem of labor regulation and minimum wages in particular, is enforcement (Ronconi, 2010). Previous estimates of non-compliance or minimum wage violation in developing countries estimate that 16-50% of workers are underpaid. Moreover, compliance rates are usually heterogeneous between and within countries. The level of violation ultimately depends on labor market attributes, including institutional factors, location, worker characteristics, and coverage status or informality. However, the majority of evidence on non-compliance is usually obtained by counting how many workers earn less than the minimum wage, also referred to as the incidence of violation. While this calculation remains useful, it tells one part of a larger story. This measure places equal weights on workers earning just below the minimum and those paid much less. For example, a construction worker earning one cent below the minimum is counted the same as another that makes only half the wage floor. Both employees are underpaid but find themselves in two clearly different situations, which the standard measure of non-compliance is unable to identify. Recently, Bhorat et al. (2013) proposed a family of violation indices which, similar to the Foster et al. (1984) poverty measures, can provide evidence on the incidence, depth, and severity of non-compliance with minimum wages. Their use facilitates distinguishing between compliance at the extensive and intensive margin. Extensive compliance counts whether workers effectively receive minimum wages, for which the traditional incidence measure is ideal. Intensive compliance captures the degree of violation by estimating the depth and severity of non-compliance. Therefore, these indices are able to count how many workers are underpaid but also how far they are from the minimum. 2

3 This paper employs these measures on household survey data from Honduras, a dual labor market with multiple floors. I pose two research questions. First, I ask whether industry-level compliance is heterogeneous, but take an additional step by highlighting urban-rural differences and duality in coverage. Second, since previous literature is mainly concerned with how enforcement affects compliance, I focus instead on if and how higher minimum wages affect the incidence and depth of minimum wage violations. I expect the results to contribute to the debate on non-compliance with labor regulations by providing new empirical evidence on minimum wage violation. For instance, while some studies do analyze industry-level heterogeneity in violations, fewer have focused on disparities among location and coverage status or informality. Additionally, at the time of writing, there remains no evidence on if and how higher minimum wages affect the incidence and depth of violations. Addressing these gaps would help researchers better understand the complexity of non-compliance in developing countries, and hopefully guide enforcement policies to improve the less than ideal level of minimum wage compliance. The main findings highlight the heterogeneity of non-compliance across industries, and reveal differences between covered and uncovered wage earners. On the one hand, covered workers are employed in industries with high non-compliance at the extensive margin, but are better off at the intensive margin since they are underpaid less. On the other hand, uncovered workers fare doubly worse because they ply their trades in industries with high non-compliance at both the extensive and intensive margin. I also conclude that higher minimum wages affect violation at the extensive but not the intensive margin. Therefore, after an increase, industries will pay a larger fraction of workers less than the minimum wage but not change the depth by which they underpay. The remainder of this paper is organized as follows. Section 2 provides background on minimum wage compliance in developing countries and then describes Honduras s labor market and minimum wage policy. Section 3 presents the data and empirical strategy 3

4 from which the results will be drawn. Section 4 analyzes industry-level heterogeneity and Section 5 estimates the effect of higher minimum wages on the incidence and depth of violations. Finally, Section 6 concludes and discusses some directions for future research. 2 Background 2.1 Minimum Wage Compliance in Developing Countries An implicit assumption when setting a minimum wage is that there will be full compliance. Empirically, this is not always true. For instance, Ashenfelter and Smith (1979) debunked this myth in the US, finding that over a third of workers earned below the minimum in Not surprisingly, in developing countries where enforcement is lower, the outlook for compliance is less encouraging (Grimshaw and Miozzo, 2003). Non-compliance or minimum wage violation in developing countries is widespread, as Maloney and Mendez (2004), Rani et al. (2013), and Del Carpio and Pabon (2014) find in their cross-country studies. In general, full compliance is not supported by the data, since between 16-50% of wage earners are paid below the minimum. Additionally, their results reveal that compliance rates are heterogeneous between and within countries. The level of violation ultimately depends on labor market attributes, including institutional factors, location, worker characteristics, and coverage status or informality. However, the majority of evidence on non-compliance is usually obtained by counting how many workers earn less than the minimum wage, also referred to as the incidence of violation. While this calculation remains useful, it tells one part of a larger story. This measure places equal weights on workers earning just below the minimum and those paid much less. For example, a construction worker earning one cent below the minimum is counted the same as another that makes only half the wage floor. Both employees are underpaid but find themselves in two clearly different situations, which the standard measure of non-compliance is unable to identify. 4

5 To single out these differences, Bhorat et al. (2013) propose a family of violation indices which, similar to the Foster et al. (1984) poverty measures, quantify the incidence, depth, and severity of non-compliance. Their use facilitates the distinction between compliance at the extensive and intensive margin. Extensive compliance counts whether workers effectively receive minimum wages, for which the traditional incidence measure is ideal. Intensive compliance captures the degree of violation by estimating the depth and severity of non-compliance. Therefore, these indices are able to count how many workers are underpaid but also how far they are from the minimum. Since their introduction, several papers have quantified minimum wage violation measures. Bhorat et al. (2012b) study the case of South Africa, where they find that on average, 44 percent of workers earn below the wage floor and that violated workers have an average shortfall that is 35% of the minimum wage. Rani et al. (2013) estimate that underpaid workers in their sample of eleven countries earn at most of the minimum wage, which often amounts to less than required living costs. Kanbur et al. (2013) evaluate violations in Chile, finding that 18.2% of workers are paid below the minimum, with an average shortfall of 24.5%. Additionally, most of these studies find heterogeneity in violations across industries and other employee characteristics. But not all studies are solely concerned with estimating the indices. For instance, Bhorat et al. (2012a) evaluate the determinants of the incidence and depth of noncompliance. In general, they infer that the depth of violation is driven by different underlying factors than its incidence. In particular, a variety of factors impact on the probability and depth of violation, including individual, industry, firm-level/contractual, and spatial characteristics. They find that some of these attributes were significant markers for the incidence of violation, but not its depth. Overall, these applications have provided two main findings. First, we should be concerned not only with the incidence of violation, but also by its depth since they are determined by different underlying factors. Second, there are important hetero- 5

6 geneities in the incidence and depth of violations across industries, and perhaps other labor market attributes. However, some topics remain unaddressed. On the one hand, while existing studies analyze heterogeneity in violations across industries, fewer have focused on disparities across location and coverage status or informality. On the other hand, at the time of writing, there remains no evidence on if and how higher minimum wages affect the incidence and depth of violations. Addressing these gaps would help researchers better understand the complexity of non-compliance in developing countries, and hopefully guide enforcement policies to improve the less than ideal level of minimum wage compliance. To provide evidence on these matters, I estimate the family of violation measures on household survey data from Honduras, a dual labor market with multiple floors. However, before proceeding to the estimates, I briefly describe the Honduran labor market and minimum wage policy. 2.2 Honduras s Labor Market and Minimum Wage Policy Honduras is the third poorest country in Latin America, behind Haiti and Nicaragua (Cordero, 2009). As recently as 2012, 61% of the adult population (15-65) was economically active. Nationwide employment and unemployment rates were 58.5% and 4%, respectively. Most employed individuals were either wage earners or self-employed workers, with the former usually located in cities and the latter in rural areas. The labor market has a large informal sector, estimated by SEDLAC (2013) at 63% under a productive definition and 94% using the social security criterion as defined in Gasparini and Tornarolli (2009). The minimum wage was first implemented in 1971 and has since become a standard component of labor regulation under the responsibility of the General Directorate of Wages (DGS, in Spanish). 1 Minimum wage policy has several distinctive characteristics 1 The minimum wage and the DGS were created by Decree 103, published on April 30th in La Gaceta, 6

7 in Honduras. First, they are set monthly, not hourly. Second, there have always been multiple minimum wages. As many as 36 different wage floors have co-existed, varying by industry, location, and firm size. Finally, while the minimum applies legally to covered or formal wage earners in the private sector, a number of other pay schemes use it as a benchmark (e.g. doctors, teachers, and most public employees). Figure 1 plots the evolution of nominal minimum wages, cost of living, and the poverty line, beginning in 1990 for which data on all three concepts are available. Over the past twenty years, the minimum wage in Honduras has been set above the poverty line, but below the estimated living cost. Upon closer inspection, I identify three broad periods in Honduran minimum wage policy. First, from , the minimum wage was low. Second, from , minimum wages increased somewhat, stabilizing between the poverty line and living costs. Finally, in 2009, minimum wages increased significantly due to an unexpected policy change. In 2009, the president unilaterally increased the minimum wage by 60 percent. To put this change into perspective, the highest increase during was 17%. Additionally, the policy also modified the payment structure of the minimum wage, simplifying the system from multiple values by industry and firm size towards location-based floors (urban and rural). The policy s implementation was immediate; the official announcement was made on December 28th, 2008, and the new pay scheme became effective on the first day of 2009 and was never reneged. Previous studies in Honduras have provided evidence on how minimum wages affect earnings and employment and their effectiveness as a poverty reduction tool. Nevertheless, while all these papers describe compliance in some manner, it is not their chief concern. For instance, Gindling and Terrell (2009) estimates the incidence of non-compliance during to provide context before presenting their estimated elasticities. Similarly, Gindling and Terrell (2010) describe the number of workers earning below the wage the official newspaper of the Honduran Senate. 7

8 floor before analyzing the relationship between minimum wages and poverty. Furthermore, Ham (2015) estimates the effect of the 2009 policy on earnings and employment, but mentions compliance in passing. Therefore, while there is knowledge about minimum wages and their effects in Honduras, the evidence on non-compliance is somewhat incomplete, something this study aims to remedy. Thus, the Honduran labor market poses an interesting case study for compliance. On the one hand, the characteristics of minimum wage setting and the large informal sector will allow analyzing heterogeneity in compliance across location and worker coverage. On the other hand, the 2009 policy provides large variation to determine if and how the incidence and depth of minimum wage violations change when minimum wages increase. These attributes will help address the identified research gaps in the minimum wage violation literature, and provide new empirical evidence to the labor regulation debate. 3 Data, Sample, and Empirical Strategy 3.1 Data To perform the analysis, I employ microdata from Honduras s Encuesta Permanente de Hogares de Propósitos Múltiples (EPHPM). The EPHPM is a nationally representative household survey usually conducted twice a year -in May and September- by the National Statistics Institute (INE). It contains detailed individual-level information on topics such as demographics, education, employment, and income for the Honduran population. 2 Furthermore, this data is augmented with decreed minimum wage values for each year published by the Ministry of Labor and Social Security (STSS), which are then imputed to workers using self-reported industry codes and firm size. 3 Since Gindling and Terrell (2009) have provided evidence on the incidence of min- 2 For a brief history of household surveys in Honduras, see Ham (2011). 3 The decrees and minimum wage tables are publicly available at the STSS website: 8

9 imum wage compliance in Honduras from , I focus on the subsequent period ( ). Surveys for each year are pooled into a single data set encompassing these eight years. I select the May rounds, except in 2005 and 2008 where I use the September survey. For the first case, this is because some of the required variables are not available in May. For the second case, the decision is due to data quality. The May 2008 survey overestimates incomes and differs from previous and following surveys in many important characteristics. 4 All variables are identically defined in each survey to ensure maximum comparability. 3.2 Sample Since the minimum wage in Honduras applies legally to wage earners in the private sector, I choose this as my population of interest. In 2012, these workers represented approximately one third of the country s total workforce. I further divide this group by location and coverage status to enrich the analysis. Worker location allows quantifying differences in minimum wage violations between urban and rural areas. As found by Gindling and Terrell (2009), resources devoted to guarantee minimum wage compliance in Honduras are limited, especially away from large cities. Coverage status helps determine how compliance varies for workers that (presumably) benefit from labor market institutions and those which do not. To classify workers by coverage status, I use a measure generally associated with labor informality that is based on the social security definition of labor protection. This criterion defines covered workers as those who are guaranteed certain benefits by their employers Gasparini and Tornarolli (2009). In the data, I consider wage earners to be covered if they receive at least one benefit from the following list: pensions, severance, paid vacations, overtime pay, accident insurance, christmas bonus, and mid-year bonus. If employees receive none of these benefits, they are classified as uncovered workers. 4 A comparative analysis of the household surveys is available from the author upon request. 9

10 The pooled data provides information on N = 66, 186 private sector wage earners during Approximately 54% of the sample lives in urban areas, with the remaining workers residing in the rural sector. About 45.7% receive employer benefits, which means that most workers in the sample are uncovered. The four main worker types I consider are distributed as follows: urban covered workers (39.8%), urban uncovered workers (20.3%), rural covered workers (9.9%), and rural uncovered workers (30%). These workers are employed in one of ten industries: agriculture, mining, manufacturing, utilities, construction, commerce or retail, transport, banking, services, and the export-driven maquila sector. Table 1 presents industry employment shares for each worker type. Urban covered employees in Honduras mostly work in commerce or retail, maquilas, and the manufacturing industry; which together account for 65% of their total employment share. Rural covered workers largely ply their trade in construction, followed closely by commerce or retail. Urban uncovered workers are typically employed in agriculture, the maquila sector, and manufacturing. Finally, rural uncovered wage earners are predominantly working in the agricultural sector. 3.3 Empirical Strategy To analyze the incidence and depth of minimum wage violations in Honduras, I employ the index proposed by Bhorat et al. (2013). This measure, V α, takes individual violations of the minimum wage and aggregates them in a manner similar to the Foster et al. (1984) indices of poverty: [{ (w V α = E [v(w m m } w) α ], w)] = E Here, w m is the monthly minimum wage and w is the actual monthly wage for workers in the sample. The value of α captures differing degrees of violation aversion. Therefore, α = 0 counts the fraction of workers that earn less than the minimum, α = 1 w m 10

11 quantifies the gap between actual pay and legislated minimum wages expressed as a fraction of the minimum, and α = 2 squares this gap to give more weight to large deviations. In what follows, I refer to these measures as the incidence (V 0 ), depth (V 1 ), and severity (V 2 ) of non-compliance, respectively. Since V 1 and V 2 are not directly interpretable, I also provide estimates of an alternative indicator used by Bhorat et al. (2013) to quantify the depth of minimum wage violations: V 1 /V 0. This ratio captures the percentage shortfall in the average pay of violated workers from the minimum wage. In other words, workers earning less than the mandated wage floor are paid on average V 1 /V 0 below the minimum. I present the estimates in two parts. In the first part, cross-sectional indices are calculated for the entire period ( ) to explore industry-level heterogeneity in minimum wage violations across location and coverage status. The findings will characterize current minimum wage compliance in the Honduran labor market, highlighting urban-rural differences and duality. In the second part, I split the time frame into two sub-periods, before and after the 2009 policy, to study the effect of higher minimum wages on violations. These results will identify which industries complied with the policy and which ones did not by measuring changes in the incidence of violations ( V 0 ) and their shortfall ( V 1 /V 0 ). Furthermore, since the aim is to compare minimum wage violations across industries and time, I bootstrap these indices to determine whether the estimated differences are statistically significant. In particular, point estimates of the main violation measures will be accompanied by 95% confidence intervals obtained from 500 replications. 5 5 The confidence intervals will be presented in graphical form, but the tables with these calculations are available from the author upon request. 11

12 4 The Incidence and Depth of Minimum Wage Violations in Honduras Before proceeding to the estimates, Table 2 shows average monthly minimum wages (w m ) and actual wages (w) by industry and worker type to provide some context. All values are expressed in local currency (Lempiras) deflated at December 1999 prices, so that these means represent real minimum wages and earnings. 6 On average, urban workers in Honduras are paid above the minimum wage regardless of their coverage status. The ratios between average wages and the minimum for covered and uncovered urban wage earners are 1.91 and 1.35, respectively. In contrast, rural wage earners are usually paid below legislated floors, with their ratios estimated at 0.95 for covered employees and 0.55 for uncovered laborers. Table 2 also highlights wage heterogeneity across industries, since some pay below the minimum and others above it. For instance, rural covered workers employed in transport, banking, services, and maquilas earn above the minimum but below this value in all remaining industries. These differences foreshadow that industry-level compliance will most likely vary. Therefore, I now turn to estimates of V α to explore heterogeneity in non-compliance in Honduras, focusing mainly on the incidence and depth of violations (the latter measured by V 1 /V 0 ). Tables 3 and 4 present estimates of V 0, V 1, V 2, and V 1 /V 0 for urban and rural workers, respectively. The values indicate a generalized level of incomplete compliance. In ascending order, 24.2% of covered urban workers, 31.9% of rural covered employees, 62.4% of uncovered urban wage earners, and 83% of uncovered rural workers earn below the minimum wage. The depth of these violations follow a similar pattern. I find that in urban areas, covered workers earn on average 21.4% below the minimum and the uncovered 22.6%. Shortfalls from the minimum are significantly worse in rural areas, 6 During the selected time frame, the average exchange rate was L per 1 US dollar (Source: Honduran Central Bank official exchange rates). 12

13 where covered and uncovered workers are underpaid by 45.2% and 53.8%. Point estimates for the main violation indicators by industry, V 0 and V 1 /V 0, with their respective confidence intervals, are shown in Figures 2 and 3. The horizontal line plots the means described in the preceding paragraph as a reference point to compare heterogeneity in industry-level violations. First, I observe differences in the incidence of minimum wage violations (V 0 ). Even though urban workers mostly earn above the minimum, there is some non-compliance. Moreover, the amount of violations fluctuates across industries, revealing statistically significant differences. For instance, covered workers are mostly employed in manufacturing, commerce or retail, and the maquila sector; which have above average incidence of violation. In turn, employers in construction, transport, and banking are below mean non-compliance. Uncovered wage earners in cities show similar patterns at the industrylevel, but the incidence of violation is much higher. Violation is especially high in sectors with many uncovered workers such as agriculture, manufacturing, and maquilas. However, the correlation coefficients between the share of covered and uncovered workers in each industry and V 0 are 0.18 and 0.50, although the latter is only marginally insignificant at the 0.10 level. Rural wage earners also experience heterogeneous compliance across industries. Covered workers are mostly employed in construction, commerce or retail, and agriculture, where non-compliance rates are 18%, 36.5%, and 37.4%, respectively. Besides construction, the two main employers of covered wage earners in rural areas seem to commit an above average number of violations. Uncovered workers in rural areas fare worse, since most of them are employed in agriculture, where 91.4% are paid less than the minimum wage. Correlation coefficients between the share of workers in each industry and V 0 are for covered wage earners and for uncovered employees, but are once again insignificantly different from zero, although just shy of 0.10 for uncovered workers. What about the depth of minimum wage violations? In urban areas, the estimates 13

14 of V 1 /V 0 reveal large differences across industries. On the low end, maquilas underpay covered workers by about 16.5% and uncovered workers by 38%. On the high end, the mining sector pays covered and uncovered wages that fall 40.6% and 50.4% short of the minimum. The remaining industries lie somewhere in between, but the general picture reveals that urban workers earn at most of the minimum wage if they are covered, and of what they should legally earn if they are uncovered. How much less do industries pay violated workers in urban areas? Covered workers are underpaid by 24.5% in manufacturing, 21.2% in commerce or retail, and by 16.5% in maquilas. The correlation between industry employment shares and V 1 /V 0 for these workers is and marginally insignificant at the.10 level, suggesting that more populated industries do not significantly underpay their employees. In contrast, uncovered laborers mostly work in agriculture, manufacturing, and maquilas, who pay 53%, 46.5%, and 38.3% below the minimum wage. The correlation between employment and violation depth is 0.19 and is not significant at any conventional level. In rural areas, the depth of minimum wage violations is higher for all industries. At most, rural wage earners make of the minimum wage if covered and when uncovered. Better paid industries in the rural sector include maquilas and transport, while underpaid industries encompass agriculture and mining. Covered wage earners outside cities tend to work in construction and commerce or retail, who underpay them by 21.6% and 26.5%. Uncovered employees in agricultural occupations earn less than half of the wage they are entitled to. The relationships between employment shares and violation depth are insignificant and estimated at and for covered and uncovered wage earners, respectively. Do V 0 and V 1 rank industries the same or do they capture different underlying factors? In general, the rank correlations between both measures are high, but not perfect. The coefficient is 0.709, 0.782, 0.794, and for urban covered, urban uncovered, rural covered, and rural uncovered workers. This implies than when extending compliance 14

15 analysis to include the depth of violations, some industries stand out more than others. Therefore, a high incidence of violations in one industry does not necessarily imply a stronger depth of the violation. In general, the previous findings highlight the heterogeneity of non-compliance across industries in Honduras, and reveal differences between covered and uncovered wage earners. On the one hand, covered workers are employed in industries with high noncompliance at the extensive margin, but are better off at the intensive margin since they are underpaid less. On the other hand, uncovered workers fare doubly worse because they ply their trades in industries with high non-compliance at both the extensive and intensive margin. In addition, rank correlations support that the depth of violation may be driven by different underlying factors than the incidence of violations, suggesting that standard measures of compliance may omit other relevant dimensions of non-compliance in developing country labor markets. 5 The Effect of Higher Minimum Wages on the Incidence and Depth of Violations While the above results provide evidence on heterogeneity in minimum wage violations for dual labor markets, they make one important assumption: that had no abrupt changes in minimum wage policy. As described previously, this was not the case in Honduras, since a large increase in the minimum wage became effective in I now examine how minimum wage violations changed between the pre-policy period ( ) and post-policy years ( ) to determine how compliance responds to higher wage floors. As mentioned in Section 2, nominal minimum wages rose by approximately 60% in Additionally, the policy also changed the payment structure of wage floors, moving from industry-level minimums to location-based floors (urban and rural). Therefore, 15

16 each industry was meant to experience a different increase, plotted in Figure 4. On average, urban workers were intended to receive higher increases than their rural counterparts. Some industries did not expect large gains, such as utilities and maquilas. However, other sectors anticipated to nearly double the wages they paid, including: mining, manufacturing, commerce or retail, and services. Tables 5 and 6 show that these large expected changes were not translated into effective compliance. Indeed, most industries did not comply with the 2009 minimum wage policy at the extensive margin. Before the increase, 16.9% of urban covered workers were paid below the minimum wage and the average depth of these violations was 19.1%. A higher fraction of urban uncovered workers earned less than the wage floor (48.4%), and were underpaid by almost 39%. Pre-policy violations were worse in rural areas, where 23.2% of covered wage earners suffered from non-compliance, with an estimated depth of 20%. Over 77% of uncovered rural employees did not receive the minimum wage and earned at most half of what they were entitled to. The tables also confirm industry-level heterogeneity in compliance before and after the policy, consistent with earlier findings. How did the incidence of violations respond to higher minimum wages? Figure 5 summarizes industry-level changes in V 0 before and after the policy. In urban areas, the average number of covered and uncovered workers paid below the minimum wage rose significantly by 15.1 and 25.5 percentage points, respectively. This increase was not uniform across industries. Incidence for covered workers rose substantially in agriculture, mining, and manufacturing, but did not change as much in services and the maquila sector. Uncovered workers earnings fell markedly below the minimum in commerce or retail, mining, manufacturing, and transport, but were relatively stable for services and maquilas. As before, I calculate the correlation coefficient between industry employment shares and V 0. For covered workers the estimated relationship is and is not significant. Similarly, for uncovered wage earners, the correlation is also low, , 16

17 and also insignificant. Hence, there seems to be no correlation between changes in the incidence of violations and the amount of workers in each industry. In rural areas, the number of violations significantly increased by 18 and 10.5 percentage points for covered and uncovered workers, respectively. Although somewhat lower than for the urban labor market, these results respond to the higher initial incidence of workers earning below the minimum outside cities. For covered workers, V 0 rose substantially in mining and transport but not as much in maquilas. For uncovered workers, mining and commerce were the highest non-compliers with the 2009 policy. Again, the correlation between employment shares and V 0 is not significant for either type of worker, for covered and for the uncovered. Figure 6 plots the changes in violation depth due to the 2009 policy. In ascending order, V 1 /V 0 changed by 3.6, 10.3, 2.9, and 7.9 percentage points for urban covered, urban uncovered, rural covered, and rural uncovered wage earners. While there are some industry-level differences, the depth of violations did not change significantly since all confidence intervals include zero. This implies that the minimum wage policy did not affect how much workers were underpaid, both on average and across industries. These findings suggest that increases in the minimum wage affect compliance at the extensive but not the intensive margin. Overall, the incidence of violation increased significantly for all industries, with substantial variation among them. However, while industries tend to pay workers below the minimum wage after an increase, the depth by which they underpay does not change significantly. 6 Conclusion This paper studies minimum wage compliance in Honduras, a dual labor market with multiple minimum wages. However, instead of only counting violations, I also focus on their depth by estimating the indices proposed by Bhorat et al. (2013). Using these 17

18 methods, I first analyze cross-sectional heterogeneity in compliance across industries, location, and coverage status or informality. Then, I estimate the effect of higher minimum wages on the incidence and depth of violation. The main findings highlight the heterogeneity of non-compliance across industries, and reveal important differences between covered and uncovered wage earners. On the one hand, covered workers are employed in industries with high non-compliance at the extensive margin, but are better off at the intensive margin since they are underpaid less. On the other hand, uncovered workers fare doubly worse because they ply their trades in industries with high non-compliance at both the extensive and intensive margin. I also conclude that higher minimum wages affect violation at the extensive but not the intensive margin. Therefore, after an increase, industries will pay a larger fraction of workers less than the minimum wage but not change the depth by which they underpay. Broadly, these results indicate that at least for wage earners, violation incidence and depth not only vary across industries, but also by location and coverage status or informality. Taking these factors into account when designing enforcement policies may enrich interventions aimed at increasing compliance, following the example of Gindling et al. (2014). However, while new findings are brought to the discussion, there remain other directions to explore. To name a few, future studies may estimate violation indices for workers other than wage earners, evaluate the impact of enforcement policies on both the incidence and depth of non-compliance, and attempt to characterize the role of minimum wages in segmented markets by implementing more accurate measures of the informal sector. Hopefully, more evidence will ensure labor regulations fulfill their ultimate objective, protecting the most vulnerable workers. 18

19 References Ashenfelter, O. and Smith, R. S. (1979). Compliance with the Minimum Wage Law. Journal of Political Economy, 87(2): Basu, A., Chau, N., and Kanbur, R. (2010). Turning a Blind Eye: Costly Enforcement, Credible Commitment and Minimum Wage Laws. Economic Journal, 120(543): Bhorat, H., Kanbur, R., and Mayet, N. (2012a). Estimating the Causal Effect of Enforcement on Minimum Wage Compliance: The Case of South Africa. Review of Development Economics, 16(4): Bhorat, H., Kanbur, R., and Mayet, N. (2012b). Minimum wage violation in South Africa. International Labour Review, 151(3): Bhorat, H., Kanbur, R., and Mayet, N. (2013). A Note on Measuring the Depth of Minimum Wage Violation. LABOUR, 27(2): Cordero, J. A. (2009). Honduras: Recent economic performance. CEPR Reports and Issue Briefs , Center for Economic and Policy Research (CEPR). Del Carpio, X. and Pabon, L. (2014). Minimum wage policy: Lessons with a focus on the ASEAN region. Working paper, The World Bank. Foster, J., Greer, J., and Thorbecke, E. (1984). Measures. Econometrica, 52(3): A Class of Decomposable Poverty Gasparini, L. and Tornarolli, L. (2009). Labor informality in Latin America and the Caribbean: Patterns and trends from household survey microdata. Revista Desarrollo y Sociedad, 63(1): Gindling, T. and Terrell, K. (2009). Minimum wages, wages and employment in various sectors in Honduras. Labour Economics, 16(3): Gindling, T. and Terrell, K. (2010). Minimum wages, globalization, and poverty in Honduras. World Development, 38(6): Gindling, T. H., Mossaad, N., and Trejos, J. D. (2014). The Consequences of Increased Enforcement of Legal Minimum Wages in a Developing Country: An Evaluation of the Impact of the Campaña Nacional de Salarios Mínimos in Costa Rica. IZA Discussion Papers 8253, Institute for the Study of Labor (IZA). Grimshaw, D. and Miozzo, M. (2003). Minimum wages and pay equity in Latin America: Identifying the employment and pay equity effects. ILO Working Papers , International Labour Organization. 19

20 Ham, A. (2011). La distribución del ingreso en Honduras: Un análisis de las principales tendencias desde los 1990s. Revista Centroamericana de Ciencias Sociales, VIII(2): Ham, A. (2015). The Effect of Minimum Wages on Earnings and Employment in Dual Labor Markets: Evidence from Honduras. Unpublished manuscript. Kanbur, R., Ronconi, L., and Wedenoja, L. (2013). International Labour Review, 152(3-4): Labour law violations in Chile. Maloney, W. and Mendez, J. (2004). Measuring the Impact of Minimum Wages. Evidence from Latin America. In Law and Employment: Lessons from Latin America and the Caribbean, pages National Bureau of Economic Research, Inc. Rani, U., Belser, P., Oelz, M., and Ranjbar, S. (2013). Minimum wage coverage and compliance in developing countries. International Labour Review, 152(3-4): Ronconi, L. (2010). Enforcement and Compliance with Labor Regulations. Industrial and Labor Relations Review, 63(4): SEDLAC (2013). Socio-Economic Database for Latin America and the Caribbean (CED- LAS and The World Bank). 20

21 Table 1 Share of wage earners by industry, average for Covered Uncovered Industry Urban Rural Urban Rural Agriculture, Fishing, and Hunting Mining Manufacturing Utilities Construction , Hotels, and Restaurants Transport, Storage, and Communications Banking, Financial, and Real Estate Services Communal, Social, and Personal Services Export sector (Maquila) Total share Source: Own calculations from pooled EPHPM surveys for Notes: All statistics are weighted using survey provided expansion factors. 21

22 Table 2 Real minimum wages and earnings by industry, average for Urban Industry Minimum wage Covered Uncovered Covered Uncovered Agriculture, Fishing, and Hunting 1, , , , Mining 1, , , , , Manufacturing 1, , , , , Utilities 2, , , , , Construction 1, , , , ,520.11, Hotels, and Restaurants 2, , , , , Transport, Storage, and Communications 2, , , , , Banking, Financial, and Real Estate Services 2, , , , , Communal, Social, and Personal Services 1, , , , , Export sector (Maquila) 1, , , , , Average 1, , , , , Source: Own calculations from pooled EPHPM surveys for Notes: All statistics are weighted using survey provided expansion factors. Values are deflated using the Honduran Central Bank s Consumer Price Index (Dec. 1999=100). During the time frame, average exchange rates were L per 1 US$. Rural

23 Table 3 Estimates of violation indices for urban wage earners, average for Covered Uncovered Industry V 0 V 1 V 2 V 1 /V 0 V 0 V 1 V 2 V 1 /V 0 23 Agriculture, Fishing, and Hunting Mining Manufacturing Utilities Construction , Hotels, and Restaurants Transport, Storage, and Communications Banking, Financial, and Real Estate Services Communal, Social, and Personal Services Export sector (Maquila) Average Source: Own calculations from pooled EPHPM surveys for Notes: All statistics are weighted using survey provided expansion factors.

24 Table 4 Estimates of violation indices for rural wage earners, average for Covered Uncovered Industry V 0 V 1 V 2 V 1 /V 0 V 0 V 1 V 2 V 1 /V 0 24 Agriculture, Fishing, and Hunting Mining Manufacturing Utilities Construction , Hotels, and Restaurants Transport, Storage, and Communications Banking, Financial, and Real Estate Services Communal, Social, and Personal Services Export sector (Maquila) Average Source: Own calculations from pooled EPHPM surveys for Notes: All statistics are weighted using survey provided expansion factors.

25 Table 5 Changes in violation indicators for urban wage earners, pre and post policy 25 V 0 V 1 /V 0 Industry Pre Post Difference Pre Post Difference Covered Agriculture, Fishing, and Hunting Mining Manufacturing Utilities Construction , Hotels, and Restaurants Transport, Storage, and Communications Banking, Financial, and Real Estate Services Communal, Social, and Personal Services Export sector (Maquila) Average Uncovered Agriculture, Fishing, and Hunting Mining Manufacturing Utilities Construction , Hotels, and Restaurants Transport, Storage, and Communications Banking, Financial, and Real Estate Services Communal, Social, and Personal Services Export sector (Maquila) Average Source: Own calculations from pooled EPHPM surveys for Notes: All statistics are weighted using survey provided expansion factors. Pre refers to and Post to

26 Table 6 Changes in violation indicators for rural wage earners, pre and post policy 26 V 0 V 1 /V 0 Industry Pre Post Difference Pre Post Difference Covered Agriculture, Fishing, and Hunting Mining Manufacturing Utilities Construction , Hotels, and Restaurants Transport, Storage, and Communications Banking, Financial, and Real Estate Services Communal, Social, and Personal Services Export sector (Maquila) Average Uncovered Agriculture, Fishing, and Hunting Mining Manufacturing Utilities Construction , Hotels, and Restaurants Transport, Storage, and Communications Banking, Financial, and Real Estate Services Communal, Social, and Personal Services Export sector (Maquila) Average Source: Own calculations from pooled EPHPM surveys for Notes: All statistics are weighted using survey provided expansion factors. Pre refers to and Post to

27 Figure 1 Trends in minimum wages, cost of living, and the poverty line Lempiras (Nominal) Nominal MW Cost of living Poverty line Source: Ministry of Labor and Social Security, 27

28 Figure 2 Incidence of minimum wage violations (V 0 ), average for Number of MW violations (V 0 ) Urban covered Number of MW violations (V 0 ) Rural covered Agriculture Manufacturing Mining Utilities Construction Transport Banking Services Export Agriculture Manufacturing Mining Utilities Construction Transport Banking Services Export Number of MW violations (V 0 ) Urban uncovered Number of MW violations (V 0 ) Rural uncovered Agriculture Manufacturing Mining Utilities Construction Transport Banking Services Export Agriculture Manufacturing Mining Utilities Construction Transport Banking Services Export Source: Own calculations from pooled EPHPM surveys for Notes: 95% confidence intervals calculated by 500 bootstrap replications accompany each point estimate. The horizontal line presents the average value of V 0 for each worker type and is taken from the last row in Tables 3 and 4. 28

29 Figure 3 Shortfall of wages from the minimum wage (V 1 /V 0 ), average for Shortfall of wages from the MW (V 1 /V 0 ) Urban covered Agriculture Manufacturing Mining Utilities Construction Transport Banking Services Export Shortfall of wages from the MW (V 1 /V 0 ) Rural covered Agriculture Manufacturing Mining Utilities Construction Transport Banking Services Export Shortfall of wages from the MW (V 1 /V 0 ) Urban uncovered Agriculture Manufacturing Mining Utilities Construction Transport Banking Services Export Shortfall of wages from the MW (V 1 /V 0 ) Rural uncovered Agriculture Manufacturing Mining Utilities Construction Transport Banking Services Export Source: Own calculations from pooled EPHPM surveys for Notes: 95% confidence intervals calculated by 500 bootstrap replications accompany each point estimate. The horizontal line presents the average value of V 1 /V 0 for each worker type and is taken from the last row in Tables 3 and 4. 29

30 Figure 4 Expected Industry-level changes in legislated minimum wages Urban Rural 30 % Change in Nominal MW % Change in Nominal MW Agriculture Mining Manufacturing Utilities Construction Transport Banking Services Export Agriculture Mining Manufacturing Utilities Construction Transport Banking Services Export Source: Own calculations from pooled EPHPM surveys for