World Bank 47071 Employment Policy Primer December 2008 No. 10 MiNiMuM WagEs iN DEvEloPiNg CouNtriEs: HElPiNg or HurtiNg WorkErs?* A large ongoing debate centers on the costs and ben- ers covered by minimum wage legislation and that the efits of increasing the minimum wage in an economy. effects tend to be stronger among low-wage workers. The Supporters of minimum wage legislation highlight that findings are quite limited and fairly inconclusive on the it can shift the earnings distribution in favor of low-paid indirect effects of increases in minimum wages on work- workers by shrinking the bottom tail of the distribution. ers in the uncovered sectors, where the legislation either Opponents highlight its disemployment effect and the does not apply or is not complied with. This is in part fact that it might reduce the share of earnings going to because of the diversity of this sector and in part because low-paid workers. of a paucity of appropriate data to test for these effects. In sum, although there is no robust empirical evi- "The goal of the minimum wage dence that increases in the minimum wage cause large disemployment effects, it is not clear that the result- is not, of course, to reduce ing wage gain among those who retain their jobs is employment, but to redistribute large enough to increase the share of earnings going to low-wage workers in the covered sector. The pay of the earnings to low-paid workers." displaced workers moving to the uncovered sector may R. Freeman, "The Minimum Wage as a Redistributive Tool," decrease, although the data supporting this is scant. Hence, the evidence is rather inconclusive on whether The Economic Journal minimum wages leave low-paid workers better off. This policy note reviews the literature on the effects theoretical Considerations of minimum wages on labor markets in developing coun- tries. We begin by elucidating the challenges to ascertain- From a theoretical perspective, it is very difficult to ing these effects, especially in developing economies predict the employment and wage effects of the mini- where a large segment of the workforce is not covered mum wage on workers in the covered and uncovered sec- by minimum wage legislation (uncovered sector). After tors. (See Box 1 below for a brief summary.) The overall summarizing the theoretical models and their predic- effect of the minimum wage on the economy depends tions, we review the empirical evidence of the impact of on multiple factors, including the degree of competi- minimum wage legislation on wages, employment, and tion in the labor market, relative level of the minimum unemployment in the covered and uncovered sectors of wage to the market clearing wage, structure (number) the labor market. of minimum wages, share of the workforce covered by The evidence strongly suggests that an increase in minimum wage legislation, degree of enforcement of the the minimum wage tends to have a positive wage effect legislation, and elasticities of demand in the covered and and a small negative employment effect among work- uncovered sectors. * Note prepared by Katherine Terrell (Ross School of Business and Ford School of Public Policy, University of Michigan) and Rita K. Almeida (Human Development Network Social Protection, World Bank). We thank the comments provided by Jean Fares (HDNSP, World Bank) and Milan Vodopivec (HDNSP, World Bank). Minimum Wages in Developing Countr ies: Helping or Hur ting Workers? Box 1: Implications of Theoretical Models wages. If the covered sector labor market responds for Wages and Employment in Covered and with an increase in employment, a backward shift Uncovered Sectors in labor supply to the uncovered sector could occur. This would lead to a reduction of employment and Competitive Imperfect Competition to an increase in wages. However, if the new hires in Model Model the covered sector are drawn from unemployment or Covered Sector If the MW is set above market Setting the MW at WW), W will ginal product of labor) => increase and employment in the uncovered sector. The theo- increase while E will decrease. in W and E. Setting MW>MPL will increase W but will decrease E. retical prediction from the monopsonistic model of Uncovered If a large number of displaced Ambiguous effect on W and E. setting the W>MPL is a reduction in employment, as Sector workers seek jobs in this sector, in the competitive labor market. Hence, one needs to it could lead to an increase in E take into account how high the minimum wage is set and decrease in W. relative to the marginal product of labor and whether markets are competitive or not. In a model of perfectly competitive labor market, uncovered sector and Non-Compliance an increase in the minimum wage (MW) set above the market clearing wage (W) will result in a reduction of Since the uncovered sector can be as large as 30­70 employment (E). The magnitude of the effect depends percent of the labor market in developing countries and on the size of the minimum wage increase and on the competitive two-sector model of minimum wages the elasticity of the labor demand. If the elasticity of predicts negative wage effects, it is especially important demand exceeds one in the covered sector, an increase to look at the "indirect" effects in this sector. Defining in the minimum wage will reduce rather than increase the uncovered sector, however, is a non-trivial task with the share of earnings going to low-wage workers. The potentially different implications for the effects of the impact on wages and employment in the uncovered minimum wage. The uncovered can be comprised of a sector is ambiguous. If the displaced workers seek jobs high-wage public sector (where the legislation does not in this sector, the shift in labor supply could lead to apply) plus a low-wage informal sector, such as work- an increase in E and a decrease in W. In this case, the ers in small firms (where the legislation is not typically extent to which E rises and W falls is a function of the enforced) and the self-employed (where the legislation size of the shift, the elasticity of demand in that market, might not apply). In some countries, such as Brazil and and the extent to which unemployment (U) increases Mexico, it also includes higher-wage workers in large as a result of the MW increases. If displaced workers firms without a formal registration card. flow into unemployment instead of to the uncovered Non-complianceinacoveredsectoressentiallymakes sector, there would be no impact on employment and it behave as an uncovered sector, and it should be treated wages. as such when thinking about the effects of an increase in Taking into account long-run general equilibrium the minimum wages. In most countries, including the effects, such as a diversion of capital flows from the U.S., relatively few resources are devoted to monitoring covered to the uncovered sector (in response to higher compliance in the covered sector. In developing coun- wages in the covered sector), can lead to an outward shift tries, there are even fewer resources. For example, in 2005 in labor demand in the uncovered sector, hence raising in Honduras the Ministry of Labor had only four inspec- both employment and wages in this sector. tors to enforce compliance with the labor law (including Departures from the perfectly competitive model the minimum wage) in the entire country. can lead to dramatic changes in the predicted effects Economists check for compliance with the mini- of the minimum wages. The model of a monopsonis- mum wage in several ways. First, one can look for spikes tic labor market predicts that a moderate increase in at the minimum wage (or at multiples of it) in the wage the minimum wage that is greater than the market distribution of the formal and informal sectors. While clearing wage but less than the marginal product of the evidence for the U.S. has generally found such spikes, labor (MPL) results in an increase in employment and the evidence for developing countries has been more 2 Minimum Wages in Developing Countr ies: Helping or Hur ting Workers? Figure 1.Wage Distribution in Brazil and Ecuador Brazil, 1999, Monthly Ecuador, 1998, Monthly 1.0 1.0 0.5 0.5 0.0 0.0 0 5 10 0 12 14 16 ln(monthly wage) ln(monthly wage) Formal sector Informal sector Formal sector Informal sector Source: Cunningham (2007). mixed.1 For example, Figure 1 shows a clear "spike" in universal or limited to private sector employees (or to the wage distribution in the covered sector in Brazil some economic activities); and, (iv) the level(s) is (are) though not in Ecuador. Second, one can calculate the set at the low end of the wage distribution (as in the average share of workers earning less than the mini- U.S.) or throughout the wage distribution (as in Costa mum wage. Eyraud and Saget (2005) find that this share Rica). As seen above, the effects, to the covered and ranges from 0.8 percent in the U.S. in 1973 to 15 percent uncovered sectors, also should depend on the degree of in Morocco in 1995­99 and 25 percent in Honduras in compliance with minimum wage legislation. 1999. Applying this measure to all workers, and not just covered workers, has the disadvantage of capturing other Covered Sector effects apart from compliance in the covered sector. One Early time series studies for the U.S. (using data from way to overcome this problem is to test whether the share the 1960s and 70s) showed that a 1 percent increase in the of workers earning the minimum wage is substantially minimum wage reduced teenage employment by 0.1­0.3 higher among uncovered than covered employees. Third, percent. However, more recent studies exploring micro one can test whether increases in minimum wages cause data in the 1980s and 90s tend to find smaller, and at increases in the average wage prevalent in the covered times insignificant, employment effects.3 Several expla- sector. An increase in the average wage can be interpreted nations have been offered for the insignificant employ- as evidence supporting the enforcement of the mini- ment effects in more recent years. They range from mon- mum. Using this approach, Gindling and Terrell (2007a, opsonistic labor markets and employers reducing other 2007b) find evidence that minimum wages are complied labor costs to compensate for a higher minimum wage with in both large and small firms in Costa Rica, but only to a lower level of the real minimum wage in the more in large firms in Honduras. recent period. The overall elasticities for developing countries Empirical Findings range from being quite large (­0.9 percent) in Puerto Rico after 1974 to no effects in Mexico in the 1980s. A growing literature is testing the effects of mini- 1 mum wages on employment and wages using data for Neumark et al. (2004) presents evidence for the U.S. Maloney and Núñez (2003) and Cunningham (2007) present evidence for several both developed and developing countries, especially Latin American Counties, including Argentina, Brazil, Chile, Colom- Latin America.2 This literature documents heterogeneity bia, Mexico, and Uruguay. Lemos (2006) also presents evidence for in the structure and coverage of minimum wages. Spe- Brazil, and Gindling and Terrell (2007a) for Costa Rica and (2007b) cifically, countries differ on whether: (i) one or multiple for Honduras. 2For recent and comprehensive reviews of the literature, see Cun- minimum wages are in effect; (ii) the minima are set nigham (2007) and Neumark and Wascher (2007). by occupation, industry, and/or region; (iii) coverage is 3See for example Brown (1999) and Card and Krueger (1995). 3 Minimum Wages in Developing Countr ies: Helping or Hur ting Workers? (See Box 2 below.) The difference in these elasticities the entire wage distribution (from domestic servants to can be traced to the level and structure of minimum university graduates), Gindling and Terrell (2007a) find wages. Where they are set at fairly high levels relative that only the wages and employment of individuals in to the market wage and enforced, minimum wages can the lower half of the wage distribution are impacted. generate larger disemployment effects. One explanation may be that the minimum wages are In fact, some evidence indicates that increases in the set at a level above the market clearing wage among low- minimum wage are not helping and may even be hurt- wage workers but not among high-wage workers. ing low-wage workers, both in the U.S. and developing In sum, we can conclude that in countries where the countries. For example, Neumark et al. (2004) conclude minimum wage is set at relatively low levels in the wage that low-wage workers in the U.S. are more adverse- distribution, increases in the minimum will improve ly affected by minimum wage increases than higher- the wages of the lower-wage workers and produce small wage workers. Although the wages of low-wage workers disemployment effects in the labor market. increase, their hours and employment decline such that a decline in total income earned by this group occurs. Uncovered Sector On the other hand, high-wage workers experience an The evidence for the effects of the minimum wage increase in earned income due to an increase in their on the uncovered sector is limited and inconclusive. This hours but no change in their wage. In Colombia, where is because the effects are more difficult to quantify, in the minimum is set relatively high, Arango and Pachón part because changes in employment and wages in the (2003) have found that the minimum wage improves the covered sector indirectly affect the uncovered sector (as earnings of families in the middle and upper part of the shown in Box 1) and in part because the uncovered sec- income distribution with net losses for those in the bot- tor is comprised of such heterogeneous labor markets. tom (due to significant negative employment effects). Moreover, until recently, the correct data (on individual In Brazil, where only one universal minimum has panels) have not been available to measure these effects. been set at relatively low levels for many years, evidence The extant evidence regarding the effects on wages attests that increases in the minimum wage raise the in this sector ranges from negative to positive. Most wages of low-wage workers earning below and around papers have found that wages in the uncovered sector the minimum wage more than the wages of higher-wage are not affected on average when this sector is defined as workers (Fajnzylber 2001).Nevertheless,it seems that the the self-employed, although some claim to have found low minimum wage in Brazil does not affect the earnings a positive effects as well.4 Alternatively, some authors and employment of those at the very bottom of the wage 4 distribution--those with no schooling (Cunningham For no significant effect see for example, Gindling and Terrell (2005, 2007b) for Costa Rica and Honduras, respectively, and Maloney and and Siga 2006). At the other extreme, in Costa Rica Núñez (2003) for Colombia. For a positive effect see for example, where minimum wages are set at different levels across Fajnzylber (2001). Box 2: Effects of the Minimum Wage on Employment in Selected Developing Countries Country MW Variation Elasticities of Employment Criticism Colombia Two federal MW increases. Small effects (­0.15). Selected group of workers. (Maloney and Núñez, 2003) Colombia and Mexico Regional variation in Mexico; transition from No effects in Mexico (­0.18 but insignificant); between -- (Bell, 1997) regional to national MW in Colombia. ­0.15 and ­0.33 for the unskilled in Colombia. Honduras More than 22 MWs set by industry, firm size, and Large effect in large firm covered sector -- (Gindling and Terrell, 2007b) region; changed annually. (­0.46); stronger for the lowest two education groups. Indonesia Cross-province variation in MW changes over 1990s. Small effects (­0.04). Unclear strength of identifying informa- (Rama, 2001) tion because low enforcement. Puerto Rico U.S. federal MW and cross-industry variation. Large effect (­0.91 after 1974). Krueger (1995) shows that results are (Castillo-Freeman and Freeman, 1992) fragile. 4 Minimum Wages in Developing Countr ies: Helping or Hur ting Workers? talk about a "lighthouse" effect of the minimum wage, countries is a challenging endeavor. The evidence to date where changes in its level serve as a reference for wages shows that increases in the minimum wage do raise the in the uncovered sector, defined as workers without a wages of some workers in the covered sector and cause signed registration card in some countries and workers small negative employment effects among others. The in micro firms in others.5 They conclude that in several total earnings of low-wage workers in the covered sector Latin American countries the minimum wages affect the do not seem to rise and, have been shown to fall, in at wage distribution of the informal sector positively both least one country (Honduras). There is little evidence as to at the minimum wage and at multiples of the minimum. whether the disemployed workers end up unemployed for There is also evidence that the public sector emulates a long time or find another job rapidly in the uncovered/ minimum wage increases in its wage structure when it is informal sector. There is even less of an understanding not formally covered (Gindling and Terrell 2007a). as to the indirect effects of minimum wages on the earn- Similarly,the impact on employment in the uncovered ings of these workers in the uncovered sector. Thus, it is sector is unclear. For example, in Brazil, where uncovered difficult to inform policymakers on the overall effect of was defined as workers without a "carteira de trabalho", minimum wages on all low-wage workers in an economy. some economists have found a positive employment effect Nevertheless, a number of issues are worth high- and others have found a negative one.6Similarly, when the lighting for policymakers. First, most of the evidence uncovered sector is defined as self-employed workers, for developing countries relies on the Latin American there are also mixed findings: for example, a small positive experience. Although this is an important and heteroge- effect in Costa Rica but no significant effect in Honduras.7 neous set of countries, evidence from other geographic The disemployment effect in the public sector appears to locals is lacking. For example, in Latin America the be insignificant or small where it has been measured.8 minimum wage tends to be set at relatively high levels, Because many of these findings are drawn from which might explain the disemployment effects among studies that do not use individual panel data, it is very the low-skilled, low-wage workers. In the former com- difficult to conclude what the indirect impact of mini- munist countries, where minimum wages tend to be set mum wage legislation is on the uncovered sector. at relatively low levels, the evidence is virtually null.9 Second, when considering changes in the mini- Unemployment mum, policymakers should also consider interactions There is considerable evidence that increases in the with other labor market institutions (e.g., unemploy- minimum wage also increases unemployment among ment benefits, hiring and firing legislation, and unions). covered sector workers but it is not clear whether low- Unfortunately most of the evidence to date tends to paid workers are affected more than higher-paid work- ignore these interactions, with other labor market insti- ers. For example, Maloney and Núñez (2003) find that tutions. However, they are likely to amplify or minimize increases in the minimum wage in Colombia increase the estimated effects of the minimum wages. Consider the probability that either a salaried worker or a self- for example, the impact of an increase of the minimum employed worker will become unemployed, and the wage on unemployment. Countries with strong social effect is larger for those earning up to 70 percent of the safety nets, like the former socialist countries in south minimum wage. Montenegro (2003) finds that increases and central Eastern Europe, are likely to be impacted sig- in the minimum wage are positively associated with nificantly while those with governments that provide few the unemployment rate in Chile. Gindling and Terrell if any unemployment benefits, like the Central American (2007b) find that increases in the minimum wage lead countries, are likely to be less impacted. Furthermore, the to an increased level of unemployment in Honduras; this may be concentrated only among workers with second- 5See for example, Cunningham (2007). 6 ary education rather than those with primary education. For positive effects see Carneiro and Corseuil (2001) and Lemos (2006), and for a negative effect see Fajnzylber (2001). 7Gindling and Terrell (2007a, 2007b). implementation Challenges 8See Lemos (2006) for Brazil and Gindling and Terrell (2007a) for Costa Rica. 9Since labor codes have recently become more flexible in these coun- Accessing the effects of minimum wages on labor tries, they would provide interesting comparative case studies to the market outcomes among low-wage workers in developing Latin American economies. 5 Minimum Wages in Developing Countr ies: Helping or Hur ting Workers? adjustment to a higher minimum wage is also affected Castillo-Freeman, A. and R. Freeman. "When the Minimum by two additional factors: (i) the degree of stringency of Wage Really Bites: The Effect of the U.S.-Led Minimum on Puerto Rico." In Immigration and the Work Force, firing costs in the country, which is likely to slow down edited by G. Borjas and R. Freeman. Chicago: University adjustment, and (ii) the other monetary or in-kind of Chicago Press, 1992. compensation (e.g., training, vacation days). It should Cunningham, W. 2007. "Minimum Wages and Social Policy: be noted that very little evidence exists on whether mini- Lessons from Developing Countries." Washington, D.C.: World Bank. mum wage jobs are "worse" jobs with respect to these Cunningham, W. and L. Siga. 2006. "Wage and Employment other non-cash dimensions. Finally, the extent to which Effects of Minimum Wages on Vulnerable Groups in the labor unions rely on the minimum wage when setting Labor Market: Brazil and Mexico." Washington, D.C.: wages also might affect how widely the minimum wage World Bank. Eyraud, F. and C. Saget. 2005. The Fundamentals of Minimum is implemented. For example, in countries with power- Wage Fixing. Geneva: International Labour Office. ful unions, like Bosnia or Brazil, one might expect larger Fajnzylber, Pablo. 2001."Minimum Wage Effects Throughout effects of the minimum wage. Therefore, the interaction the Wage Distribution: Evidence from Brazil's Formal and Informal Sectors." Unpublished paper, Department of the minima with these three known variables--unem- of Economics and CEDEPLAR, Universidade Federal do ployment benefits, hiring and firing legislation, and Belo Horizonte. unions--should also be considered when setting policy. Fajnzylber, P. and W. Maloney. 2001. "Comparing Labor Finally, economists still need to improve the knowl- Demand Elasticities across Countries: Dynamic Panel Estimates for Colombia, Chile and Mexico."Washington, edge base of the indirect effects of minimum wages on D.C.: World Bank. Mimeo. the uncovered sectors. This can be achieved by: (i) sepa- Freeman, Richard. 1996."The Minimum Wage as a Redistrib- rating the effects in each of the labor markets within this utive Tool," The Economic Journal, 106(436): 639­649. sector, such as the public sector, large versus small firms Gindling, T.H. and K. Terrell. 2007a. "The Effects of Multiple Minimum Wages throughout the Labor Market: The in the private sector, and the self-employed, and (ii) Case of Costa Rica," Labour Economics 14: 485­511. making more use of panel data on individuals (coupled ------. 2007b. Minimum Wages and the Welfare of Workers with a solid research design) to observe the transition of in Honduras. IZA Discussion Paper 2892. Bonn: Institute for the Study of Labor (IZA). workers across sectors as well as their labor market out- ------. 2005. "The Effect of Minimum Wages on Actual comes. Only then, will we better understand these pro- Wages in the Formal and Informal Sectors in Costa Rica," cesses and give a complete answer to our initial question, World Development 33(11): 1905­1921. "Do minimum wages help or hinder low-paid workers?" Lemos, S. 2006. "Minimum Wage Effects in a Developing Country." London: University College London. Unpub- references lished. Maloney, W. and J. Núñez. "Measuring the Impact of Mini- mum Wages Evidence from Latin America." In Law and Arango, C. and A. Pachón. 2003."Distributive Effects of Mini- Employment: Lessons from Latin America and the Carib- mum Wages on Household Incomes: Colombia 1997­ bean, edited by J. Heckman and C. Pagés. Chicago: Uni- 2002." Washington D.C.: Banco de la Republica/World versity of Chicago Press, 2003. Bank,mimeo. Montenegro, C. 2003. "Unemployment, Job Security and Bell, L. 1997. "The Impact of Minimum Wages in Mexico and Minimum Wages in Chile: 1960­2001." Washington, Colombia," Journal of Labor Economics 15(3): S102­S134. D.C.: World Bank. Brown, C. "Minimum Wages, Employment, and the Distribu- Neumark, D. and W. Wascher. 2007. Minimum Wages and tion of Income." In Handbook of Labor Economics, edited Employment. IZA Discussion Paper 2570. Bonn: Institute by O. Ashenfelter and D. Card. Amsterdam: North Hol- for the Study of Labor (IZA). land, 1999. Neumark, D., M. Schweitzer, and W. Wascher. 2004. "Mini- Card, D. and A.B. Krueger. Myth and Measurement: The New mum Wage Effects throughout the Wage Distribution," Economics of the Minimum Wage. Princeton: Princeton Journal of Human Resources 39(2): 425­450. University Press, 1995. Rama, M. 2001. "The Consequences of Doubling the Mini- Carneiro and Conseuil. 2001. "Os Impactos Do Salario Min- mum Wage: The Case of Indonesia," Industrial and Labor imo Sobre Emprego E Salarios No Brasil: Evidencias a Relations Review 54(4): 864­881. Partir De Dados Longitudinais E Series Temporais" IPEA Working Paper, 849. The findings, interpretations, and conclusions expressed herein are those of the author(s), and do not necessarily reflect the views of the International Bank for Reconstruction and Development / The World Bank and its affiliated organizations, or those of the Executive Directors of The World Bank or the governments they represent. 6