Minimum Wages in Sub-Saharan Africa: A Primer Haroon Bhorat, Ravi Kanbur, and Benjamin Stanwix The fraction of workers currently covered by minimum wages in Sub-Saharan Africa (SSA) is small, but as formality and urbanization increase, wage regulation will become increasingly relevant. In this analysis, we find that higher minimum wage values are as- sociated with higher levels of GDP per capita, in both SSA and non-SSA countries. Using two measures to assess the level at which minimum wages are set, we find that minimum wages in SSA countries are on average lower—relative to average wages—than most other comparable regions of the world. Thus, SSA as a whole reflects no particular bias toward a comparatively more pro–minimum wage policy. Within SSA, however, we ob- serve that low-income countries set relatively higher minimum wages than middle- or upper-income countries. We find significant variation in the detail of minimum wage re- gimes and schedules in the region, as well as large variations in compliance. Notably, sev- eral countries in SSA have relatively complex minimum wage schedules, and on average we find high levels of noncompliance among covered workers. We also summarize the limited research on the employment effects of minimum wages in SSA, which are consis- tent with global results. By and large, introducing and raising the minimum wage ap- pears to have small negative employment impacts or no statistically significant negative impacts. There are country studies, however, where substantial negative effects on em- ployment are reported—often for specific cohorts. The release of country-level earnings and employment data at regular intervals lies at the heart of a more substantive, country-focused minimum wage research agenda for Africa. Legislated minimum wages apply to many millions of workers around the world. Over the latter half of the 20th century, almost all countries in Sub-Saharan Africa (SSA) introduced some form of minimum wage legislation. In many cases, The World Bank Research Observer C The Author 2017. Published by Oxford University Press on behalf of the International Bank for Reconstruction and V Development / THE WORLD BANK. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com doi:10.1093/wbro/lkw007 Advance Access publication January 3, 2017 32:21–74 these laws apply to workers in specific industries or occupations, and in instances where wage levels are not set by the state, collectively bargained agreements may fix wages for specific sectors or occupations. Broadly, however, the introduction of national laws governing wages is part of an observed regulatory revival in low- and middle-income countries, where a range of labor regulations aimed at protect- ing low-paid workers have gradually been introduced (Piore and Schrank 2008). As will be documented in the paper, there has been widespread adoption of mini- mum wage legislation as a policy tool in SSA. And yet there has been little work on the nature, scope, and impact of such laws in the region. A key distinction must be made between the usually small, formal, wage- earning sector and usually large, informal, non-wage-earning sector in most African countries.1 This has implications for minimum wage policy and research. It is well known that in the overwhelming majority of SSA economies’ subsistence agriculture and, more recently, urban informal employment dominates the labor market. In many such countries, wage-earning employees only make up a small portion of the labor force. As figure 1, below, indicates for a selection of SSA coun- tries, it is only South Africa where formal salaried employees constitute the Figure 1. Wage-Earning Employees as a Percentage of Total Employment Percentage of wage-earning employees 80% 60% 40% 20% 0 Uganda Tanzania Group Average Namibia Mali Zambia Kenya SSA Average South Africa Sources: South Africa: Labour Market Dynamics Study (2013); Kenya: Kenya Integrated Household Budget Survey (2005–6); Uganda: Uganda National Panel Survey (2012); Mali: Rani et al. (2013); Zambia: Living Conditions Monitoring Survey (2010); Tanzania: Integrated Labour Force Survey (2005–6); Namibia: Labour Force Survey (2012); Bhorat, Naidoo, and Pillay (2015). 22 The World Bank Research Observer, vol. 32, no. 1 (February 2017) majority of the labor force. In the six other countries represented, salaried em- ployees account for less than half of the labor force, while the sample average is close to 20 percent. Minimum wages only apply to wage-earning employees and, in some cases, only to formal sector wage earners. Wage regulations thus cover only a minority of the total workforce in most SSA economies. This may go some way to explaining the dearth of existing research. However, (a) as the urban and formal sector grows, the experience of wage regulation in the currently covered areas will become sig- nificant, (b) there can be spill over or “lighthouse” effects on uncovered sectors,2 and (c) as we shall see, minimum wage policy is particularly progressive in the covered sectors in Sub-Saharan Africa. For these reasons, we believe that an em- pirical overview of minimum wages in the region is important for the current pol- icy and analytical discourse. It is a stylized fact that a large gap exists between de jure and de facto labor reg- ulation in most low- and middle-income (LMI) countries globally. Simply put, the levels of noncompliance with minimum wage laws are high. This has been shown in several studies (Bhorat et al. 2012; Rani et al. 2013; and Almeida and Ronconi 2012) and is supported by the evidence presented in this paper. While it could be argued that high levels of noncompliance make wage regulation irrelevant, the reasons for high levels of noncompliance are not yet adequately understood. A study of the enforcement of minimum wages is of interest in its own right but can also contribute to a broader research agenda on questions related to the rule of law. There is a paucity of available household survey data for many SSA countries, and particularly an absence of reliable information on earnings. This makes rigor- ous study on the impact of minimum wages difficult, particularly because any at- tempts to measure minimum wage impacts require pre- and postintervention data, and this is almost always absent for countries in SSA. The problem is brought into sharp relief when analyzing one of the most extensive and centralized bodies of in- formation on global minimum wages—the International Labour Organisation’s (ILO) TRAVAIL database and the ILO global wage database. While these databases contain detailed information on minimum wage frameworks, the method of setting wages in each of the ILO’s member states, and data on enforcement practices for almost every country, there is very limited information on wages, coverage, or lev- els of compliance for countries in Africa. This is because for many countries in the region reliable data on wages either do not exist or are not accessible. It is thus the purpose of this piece to provide a basic overview of minimum wage regimes in SSA, where data are available, as well as some more detailed work for several countries using household survey data. We begin with a brief literature re- view of more recent minimum wage work and build on Neumark and Wascher’s (2007) meta-review to provide an updated overview of the empirical wage- employment trade-off estimates, focused on LMI countries. We then document the Bhorat et al. 23 level of minimum wages in the region, present simple Kaitz ratios (the Kaitz ratio is the ratio of the minimum wage to the average wage) and use this to make compari- sons with wage levels and ratios elsewhere in the world. We proceed to examine com- pliance levels in seven SSA countries using household survey data and again compare the outcomes to estimates for a selection of non-SSA LMI countries. We end by reflecting on possible early lessons for wage-setting regimes in the region and on a research agenda to inform policy makers on key trade-offs in minimum wage setting. The Developing Country Literature: A Brief Overview In most SSA countries, minimum wage laws are used as a policy tool to achieve a number of objectives, and while there are large cross-country differences, wage legislation reveals considerable overlap. The stated objectives usually focus on pro- tecting vulnerable workers from extreme levels of low pay, addressing poverty by redistributing income from employers to low-wage employees, and encouraging la- bor productivity. It is well known that there are also risks associated with institut- ing wage floors, and in most cases country-level legislation recognizes the possible trade-offs. The costs can include increased unemployment in certain settings, ad- justed hours of work that disadvantage workers, and the movement of workers from formal to informal employment—where livelihoods are often more precari- ous. Rising wages as a result of minimum wage policy, while having positive ef- fects through driving demand, may also have some impact on the cost of living in the medium term if firms’ cross-price elasticity response to higher labor costs is to raise output prices.3 Beyond the various costs and benefits, which are the focus of a well-established body of literature for developed countries, the attendant issues of enforcement and compliance are central to any discussion of minimum wages in SSA. Indeed, it is arguable that most of the impact of a minimum wage policy is contingent on enforcement and compliance in LMI country settings. We proceed to briefly analyze the minimum wage literature here. The empirical work on minimum wages constitutes a large field that is now well established, with several recent books and papers dedicated to reviewing the main findings of this literature.4 Studies in the United States were the first to suggest that the textbook wage-employment trade-offs do not always hold in practice— that an increase in the minimum wage will not always perfectly predict a decrease in employment (the seminal paper in the new minimum wage literature being Card and Krueger 1994). In addition to the potential employment effects, mini- mum wage laws have been shown to induce adjustments in hours of work and nonwage benefits, as well as having possible “nonstandard effects” such as influenc- ing educational decisions and reservation wages.5 The limited but fast-growing body of work focused on LMI countries has emphasized that the impact of introducing, or 24 The World Bank Research Observer, vol. 32, no. 1 (February 2017) increasing, a minimum wage can have mixed impacts that are often crucially con- tingent on a variety of factors, including the level at which the minimum wage is set, broader economic conditions, the nature of the minimum wage intervention, po- litical economy factors, the enforcement regime, and so on. Existing studies provide evidence of negative employment effects in some cases but also evidence of no em- ployment declines in others, with a range of adjustments observed for hours of work, wages, and nonwage benefits. In Neumark and Wascher’s (2007) meta-review, they include 15 studies focused on eight developing countries over the period 1992 to 2006.6 The authors caution that studying minimum wage effects in developing countries is complicated, partly due to issues we have already mentioned above that relate to data availability and quality, but also because the results are often not easily generalizable across countries or sectors. The majority of findings reviewed by the authors reveal either no effects or small negative employment effects of minimum wages in LMI country settings. In Brazil, for example, disemployment effects and reduced work hours are seen to be mi- nor or nonexistent overall, but more pronounced for individuals with low skills and lower wages (Fajnzylber 2001; Lemos 2004, 2006, 2007; and Neumark et al. 2006). In Chile, increases in the minimum wage had negative employment effects for youth and unskilled workers but led to an increase in the employment of women (Montenegro and Page ´ s 2004). In Colombia, research suggests that disemployment effects were present, and these were higher for low-skilled workers (Bell 1997; and Maloney and Nun ~ ez Mendez 2004). In Costa Rica, increases in minimum wage also decreased employment and reduced hours worked by employees in covered sectors, especially those in the lower half of the skill distribution (Gindling and Terrell 2005, 2007). Similar results are found in Indonesia and Trinidad and Tobago (Alatas and Cameron 2003; Rama 2001; Suryahadi et al. 2003; Hyslop and Stillman 2004; Comola and De Mello 2011; and Strobl and Walsh 2003). In the years following the publication of Neumark and Wascher’s (2007) review paper, a growing body of research has focused on minimum wage effects in LMI countries (see table A4 in the Appendix for a detailed literature summary). In East Asia, a number of papers provide new evidence on minimum wage impacts: In Thailand, Del Carpio, Messina, and Sanz-de-Galdeano (2014) found small disem- ployment effects on female, elderly, and less-educated workers and large positive effects on the wages of prime-age male workers; in Vietnam, Del Carpio and Liang (2013) and Nguyen (2010) estimates showed that low-wage formal sector employ- ment is negatively affected. In the Phillipines, Lanzona (2012) and Del Carpio, Margolis, and Okamura (2013) observed disemployment effects only in sectors with high levels of compliance; and in Indonesia, two papers (Comola and De Mello 2011; and Harrison and Scorse 2010) found disemployment effects. In China, Wang and Gunderson (2011) and Fang and Lin (2013) provided some of the first estimates of minimum wage effects—which are generally negative. In Bhorat et al. 25 Latin America, several new papers built on existing work to provide new sector- specific results on employment, but also a nuanced focus on informality, social se- curity, and poverty (Ham 2013; Gindling 2014; and Kharmis 2013). The literature on SSA, however, remains rather limited, with published work ex- isting only for four countries, namely Ghana (Jones 1997), Kenya (Andalo n and Page ´ s 2008), Malawi (Livingstone 1995), and South Africa (Hertz 2005; Dinkelman and Ranchhod 2013; Bhorat et al. 2013, 2014a; Bhorat et al. 2014b; and Garbers 2015). The most comprehensive literature in the SSA region exists for South Africa and shows that while the introduction of minimum wages had a neg- ative impact on employment in agriculture, in all other covered sectors no employ- ment decreases were evident (Dinkelman and Ranchhod 2013; Bhorat et al. 2013; and Nattrass and Seekings 2014). In figure 2(a) and (b) below, we construct a graph of employment elasticities es- timated in the minimum wage literature described above. This includes the 98 Figure 2(a). Minimum Wage-Employment Elasticities for Low- and Middle-Income Countries 1 .5 Elasticity -.5 -1 -1.5 0 Czech Republic Colombia Honduras Honduras Indonesia Indonesia Indonesia China Indonesia Indonesia Thailand Indonesia Indonesia Indonesia Indonesia Indonesia Indonesia Ghana Ghana Brazil Puerto Rico Brazil Brazil Brazil Brazil Brazil Costa Rica Puerto Rico Puerto Rico Mexico South Africa South Africa Note: The upper dotted line is the median elasticity (-0.08); the lower dotted line is the mean elasticity (- 0.11). Source: Neumark and Wascher (2007) and authors’ calculations. 26 The World Bank Research Observer, vol. 32, no. 1 (February 2017) Figure 2(b). Minimum Wage-Employment Elasticities for High-Income Countries 2 0 Elasticity -2 -4 Canada Sweden Canada Canada Canada Canada Canada Canada Canada Canada Canada Canada Canada Portugal United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States United States France Greece Australia France Australia Australia France Australia France France Australia France Unites States Note: The upper dotted line is the median elasticity (-0.23); the lower dotted line is the mean elasticity (- 0.37). Source: Neumark and Wascher (2007) and authors’ calculations. papers reviewed in Neumark and Wascher’s (2007) work and 17 more recent studies focused on LMI countries not included in Neumark and Wascher (2007). The results include aggregate impacts for all workers but also the employment im- pacts for specific demographic groups, geographic locations, and sectors. Put differ- ently, where a study produced elasticity estimates for more than one cohort of workers, we include each estimate separately. Unlike in the Neumark and Wascher (2007) review, we have only included those estimates that were statisti- cally significant, and it is worth noting that 55 percent of reported elasticity esti- mates reviewed were not statistically significant. A detailed description of each of these studies is provided in the Appendix. In figure 2(a), the median elasticity is -0.08, represented by the upper dotted line on the figure, while the mean is -0.11, shown as the lower dotted line. In fig- ure 2(b), the averages are slightly higher and the upper dotted line represents a Bhorat et al. 27 median elasticity of -0.23, while the mean for upper-income countries is -0.37, shown as the lower dotted line. Overall, estimated employment elasticities range from 2.17 (Katz and Kreuger 1992) to -4.6 (Abowd, Kramarz, and Margolis 1999) in both samples. The mean and median elasticities suggest that on average the im- pacts of minimum wage hikes in the countries under review have been marginally negative. From the sample of 59 developed and 32 developing country estimates, 81 percent of the elasticities were negative while 19 percent were positive. It is im- portant to note that the absolute value of these coefficients on average is small. This would suggest that in general the minimum wage has either benign or slightly negative employment effects. However, the question of impact requires careful focused empirical investigation on the basis of each minimum wage inter- vention. In addition, it is crucial to note that in the developing country context dis- employment effects may be further mitigated due to signficant levels of noncompliance. We discuss this latter issue in greater detail below. There are of course a number of studies in both developed and developing coun- tries that do find substantial negative employment effects relative to the average elasticities presented above. These outlier results should not be dismissed. As Neumark and Wascher (2006) note, a study on Canadian teenagers by Baker et al. (1999) reports an average employment elasticity of -0.25, while work by Abowd et al. (1999) on France finds that the minimum wage had high negative impacts on minimum wage earning men aged 25–29 years (who are not protected by employment promotion contracts), where the employment elasticity was -4.6 relative to a comparable group of men who earn just above the minimum wage. In Indonesia, Suryahadi et al. (2003) find an overall negative employment elasticity of -.06, and this is more pronounced for women, where the elasticity is estimated at -.16. An interesting paper by Feliciano (1998) shows that a decline in real mini- mum wages in Mexico led to an increase in employment, particularly among women, where the elasticity ranges from -0.58 to -1.25. In Honduras, Gindling and Terrell (2007) find a wage employment elasticity of -.46 in the private sector, while in South Africa Bhorat et al. (2014) find a decrease in agricultural employ- ment of over 5 percent. In addition, the figures above underscore the fact that there is a range of poten- tial impacts of minimum wages on employment. The heterogeneity of outcomes, in LMI countries in particular, suggests that a variety of context-specific factors inter- act with the minimum wage. These could include: the level of the minimum wage relative to average wages, the size of the minimum wage increase, the sector under consideration (for example, whether it is a tradeable or nontradeable sector), the timing of wage changes, the change postlaw in the level of worker productivity, the enforcement regime, and the extent of compliance. While past increases in minimum wages have generally not had a large negative affect on employment, it is not the case that such positive outcomes will persist regardless of the level to 28 The World Bank Research Observer, vol. 32, no. 1 (February 2017) which a minimum wage is raised. There is a level beyond which a minimum wage will begin to negatively affect employment, and this level may differ across geo- graphic regions, sectors, and firms. Figure 3, below, visually represents this idea in a basic theoretical construct by presenting two different impact scenarios. Each of the two lines can be thought of as labor demand functions representing the employment response to increases in the minimum wage. The level of employment is on the y-axis, and the level of a minimum wage increase on the x-axis. It is clear that in the case of the upper line the minimum wage can be raised to WH (the high wage threshold) without any impact on employment. Beyond WH , however, any increase in the hypothetical minimum wage will begin to result in job losses. In the second case, the lower line, the broad relationship between the minimum wage and employment stays the same, but, crucially, employment is more sensitive to changes in the minimum wage where wages can only be raised to WL (the low wage threshold) before they begin to impact employment negatively. We could interpret the upper line, for example, as representing a nontradeable sector that can absorb wage increases more robustly, while the lower line would represent a sector more sensitive to changes in input costs and perhaps with more capital/labor replacement possibilities. These elements, together with the list of abovementioned factors that influence wage-employment trade-offs, all play a role in determining where the wage threshold lies for a given set of firms, workers, or sectors. Figure 3. The Relationship between Minimum Wage Adjustments and Employment Bhorat et al. 29 One distinct possibility is that minimum wages have larger negative effects when they are higher relative to average wages. In order to briefly examine this re- lationship further, we plot the statistically significant point estimates of employ- ment effects (shown in figures 2[a] and [b]) against the Kaitz ratio for each country in the study. The figure is shown in the Appendix (figure A1). While this is a fairly crude approach, the relationship in this case is not statistically signifi- cant. This could be for several data-specific reasons, such as, for example, that the employment effects are often calculated for specific subgroups or sectors within a country for which it is difficult to calculate Kaitz ratios. However, more impor- tantly, what this statistically insignificant relationship suggests is that the relative level at which the minimum wage is set is not necessarily the key, or indeed the only, factor that may impact the employment effects emanating from the mini- mum wage—a point we emphasize in figure 3 above. The functions above also suggest a key insight into estimated wage-employment elasticities: namely that they can be nonlinear. Put differently, if a 50 percent in- crease in the minimum wage results in a 10 percent drop in employment (where the wage-employment elasticity is thus -0.2), it is not the case that a 100 percent increase will result in a 20 percent drop in employment—the elasticity will not re- main at -0.2. The literature suggests that where increases in a minimum wage are large and immediate this can result in employment losses, especially for unskilled workers. More modest increases usually have very few observably adverse effects and may have large positive impacts on wages and a range of other labor market outcomes. The positive wage impacts in a legislated increase appear to hold even in situations of weak enforcement (Dinkelman and Ranchhod 2013; Bhorat et al. 2015). Minimum Wages in Sub-Saharan Africa Minimum wage systems in SSA, as elsewhere in the world, can be helpfully classi- fied into three broad categories: national, sectoral (occupational), and some combi- nation of the two (hybrid). Needless to say, a national minimum wage system is a single wage rate that applies to all workers, a sectoral system is one in which there are separated determinations for workers in particular sectors and/or occupations, and a combination or hybrid system can exist when the body that decides on mini- mum wage levels also decides to whom the wage applies and when this should change. Table A1 (see the Appendix) groups selected African countries into these three categories. The table serves to show the variety of systems that exist in the region. In a report by the ILO (2013), the minimum wage frameworks on the con- tinent are categorized in a similar manner, revealing that a majority of countries in Africa have adopted some form of sectoral or occupational minimum wage 30 The World Bank Research Observer, vol. 32, no. 1 (February 2017) structure with multiple wage rates rather than a single national minimum wage— approximately 30 percent versus 61 percent, respectively. Indeed, the percentage of African countries with multiple minimum wage systems is higher than any of the other world regions.7 A comprehensive account of wage-setting systems around the world can be found in Eyraud and Saget (2005). Beyond the diverse regulatory frameworks, the level at which the wage is set also varies substantially by the income group that a country falls into. We show this by grouping SSA countries into low-income (LI), lower-middle-income (LMI), and upper-middle-income (UMI) categories. We compare 37 SSA countries in these three groups to a comparable set of 67 non-SSA countries similarly grouped. Figure 4, below, presents average minimum wage levels by income group for SSA and non-SSA countries. The figure reveals substantial differences both across the three country income groups within SSA and compared to non-SSA countries. As expected, minimum wage levels are positively correlated with GNI per capita levels. In particular, for lower-income countries (LICs) in Africa with a GNI per capita of US$1,045 and be- low, mean minimum wages stood at an average of US$119 (PPP 2013). This in- creases by 19 percent to $142 for LMI economies and then further by 218 percent Figure 4. Monthly Minimum Wage Levels by Country Income Group, SSA and Non-SSA Countries (US$ PPP) 500 400 Minimum wage level (US$ PPP) 100 200 0 300 LI SSA LI non-SSA LMI SSA LMI non-SSA UMI SSA UMI non-SSA Notes: LI stands for low income (in black), LMI for lower-middle income (in dark gray), and UMI for upper- middle income (in light gray). Sources: ILO global wage database, World Bank WDI. Bhorat et al. 31 to $366 for UMI African economies. This relationship is explored in further detail below. In addition, our results suggest that minimum wage levels in SSA are on average lower than elsewhere in the world for countries in the same income group (it must be noted that in the LI group our sample of non-SSA countries is limited to four economies, while in the UMI group our sample of SSA countries is limited to five countries). In the LMI category, the difference in minimum wage levels is large and significant at the 5 percent level. In order to advance this notion of the relationship between country income lev- els and minimum wage levels, we produce two more figures (5[a] and [b], below). The figures plot levels of GDP per capita against the level of minimum wages for the same group of SSA and non-SSA countries used above. We find a relationship between GDP and the level of the minimum wage that is not dissimilar to what has been found for the relation between the poverty line and levels of consumption by Ravallion et al. (2009). The latter noted that based on cross-country evidence, the value of the national poverty line rises as average consumption levels rise Figure 5(a). Monthly Minimum Wages and GDP Per Capita (US$ PPP), Africa South Africa Gabon 400 Minimum wage (US$ PPP) Kenya 300 Equatorial Guinea Mauritania Chad Lesotho Nigeria Mauritius 200 Mozambique Côte d'Ivoire Tanzania, United Republic of Sudan Senegal Botswana Benin Congo Burkina Faso Angola Comoros Niger Madagascar Togo Gambia Mali Cameroon Ghana 100 Zambia Liberia Ethiopia Guinea-Bissau Democratic Republic of the Congo Malawi Burundi Uganda 0 5 6 7 8 9 10 Log of GDP per capita (US$ PPP) Fitted values Minimum wage (US$ PPP) Note: Sample based on 37 African economies, where the latest available data for each country was utilized. Sources: ILO global wage database, World Bank WDI. 32 The World Bank Research Observer, vol. 32, no. 1 (February 2017) Figure 5(b). Minimum Wages and GDP Per Capita (US$ PPP), Non-SSA Developing Countries 1000 Iran, Islamic Rep. 800 Minimum wage (US$ PPP) Paraguay Turkey Ecuador Romania Libya 600 Thailand Belarus Jordan Hungary Belize Costa Rica Brazil Colombia Malaysia Guatemala Peru Bulgaria Panama 400 Bolivia Montenegro Serbia Morocco Ukraine Albania Pakistan Honduras Bosnia and Herzegovina Jamaica China Macedonia, FYR Dominica Azerbaijan Nepal Afghanistan Mongolia Tunisia Philippines Nicaragua Guyana Armenia 200 Kazakhstan Papua New Guinea Bhutan Indonesia Turkmenistan India Fiji Samoa Lebanon El Salvador Mexico Lao PDR Moldova Sri Lanka Dominican Republic Tajikistan Vietnam Haiti Georgia Uzbekistan Solomon Islands Venezuela, RB Bangladesh Kyrgyzstan Cuba 0 6 7 8 9 10 Log of GDP per capita (US$ PPP) Fitted values Minimum wage (US$ PPP) Note: Sample based on 67 non-SSA developing economies, where the latest available data for each country was utilized. Sources: ILO global wage database, World Bank WDI. across economies. Levels of economic development are thus positively related to the country-based poverty line. This relationship appears to hold for minimum wage levels. As we illustrate then, for a sample of African economies and non- African developing countries, minimum wage levels are adjusted upwards along with the increases in GDP per capita. Specifically, in figure 5(a), the coefficient for the underlying relationship be- tween the log of GDP per capita and the minimum wage level in SSA is 59.42 (where the level of the minimum wage is on the y-axis). This relationship is similar but larger for the 67 non-SSA countries presented in figure 5b, where the coeffi- cient is 125.14. This suggests that across countries minimum wage levels in non- SSA countries are not only higher relative to levels of GDP compared to minimum wage levels in SSA countries, but also more responsive to increases in GDP relative to SSA countries. Bhorat et al. 33 In attempting to disaggregate minimum wage trends within SSA in greater de- tail, table 1 below presents country-level data for 21 countries. We focus on the average level of the minimum wage, the mean wage, and the Kaitz ratio (mini- mum-to-mean wage). Our estimates of country-based minimum wages represent the average of all schedules for those economies with more than one minimum. The Kaitz ratio, in turn, provides an indication of how high the minimum wage is set relative to average wages. Ideally we would present median wages and use the median wage to calculate the Kaitz ratio, but, unfortunately, data on median wages for countries in SSA is rare and would limit our sample even further. Table 1 reports the most recently available minimum wage rates, grouped ac- cording to country incomes. We convert all the data into current US$ (PPP) for the sake of comparison. The first column of minimum wage rates again makes it clear that despite substantial variation across individual economies, there is a clear trend showing that minimum wage levels covary positively with country income group. The group mean and median for UMI countries is thus more than double those of LI countries. For example, then, UMI economies such as Algeria, Gabon, and South Africa have legislated average minimum wage levels in excess of $400 per month. LI countries, on the other hand, have promulgated minimum wage lev- els as low as $26 a month (Burundi).8 The country GDP and minimum wage cor- relation, though, must be emphasized as an average effect, and from the table it is clear that there are outliers in each income group. The mean wage figures in column 2 reveal important cross-country differences, even within the country groups. Hence, as expected we observe a variance in the mean wages by country—which is broadly consistent with the GNI per capita of the country under scrutiny. There remains within this data, however, an interest- ing approach to measure the tendency of an economy, in a comparative sense, to- ward setting a higher minimum wage relative to other economies. One can think of this notion in the following manner: if the ratio of the minimum wage in coun- try i to the highest minimum wage country, country max, is higher than the ratio of the mean wage in country i to the highest mean wage country, country max— then this would reflect a relatively pro–minimum wage policy environment rela- tive to other countries. Simply put, we are measuring: 2 3 Wim m Wmax Wp ¼ 4 l5 Wi l Wmax where if Wp > 1, it reflects a relatively pro–minimum wage policy environment compared to other economies, whereas if Wp < 1, we suggest a minimum wage policy environment which is relatively benign. In the figure below, we provide estimates of Wp for a sample of 66 developing countries. An example from the 34 The World Bank Research Observer, vol. 32, no. 1 (February 2017) Figure 6. Degree of Relative Minimum Wage Policy Bias by Country 2.5 2 mean of Wp 1 1.5 .5 0 Bangladesh Bosnia and Herzegovina Congo, Rep. Senegal Uganda Gabon Montenegro Tanzania Panama Mongolia Ghana Philippines Albania Romania Thailand Guatemala Congo, Dem. Rep SSA Mean EAP Mean LAC Mean ECA Mean Swaziland Venezuela, RB Burundi El Salvador Ecuador Madagascar Honduras Cuba Botswana SA Mean Georgia Azerbaijan Moldova China Serbia Kenya Armenia Zambia Nicaragua Bulgaria Indonesia Ethiopia Algeria Lesotho Chad Burkina Faso Afghanistan Argentina Vietnam Malawi Belarus Uzbekistan South Africa Kazakhstan Jamaica Malaysia India Costa Rica Ukraine Tajikistan Bolivia Fiji Brazil Mexico Mauritius Kyrgyzstan Dominican Republic Macedonia, FYR Peru Pakistan Notes: See Mean on far right for legend: South Asia (SA), Sub-Saharan Africa (SSA), East Asia and Pacific (EAP), Latin America and the Caribbean (LAC), Europe and Central Asia (ECA). Sample based on 66 countries, where the latest available data for each country was utilized. Sources: ILO Global Wage database, World Bank WDI. data: Chad’s minimum wage mean is 22 percent of the region’s maximum mini- mum wage. Yet, its mean wage nationally is only 15 percent of the region’s maxi- mum mean wage. This suggests a relatively pro–minimum wage policy in Chad when compared with its mean wage differential relative to other developing coun- tries. Put in numerical terms using the formula presented above, Chad’s Wp value is 1.45, where a value over 1 suggests a pro–minimum wage policy. The data suggest that the majority of countries in our sample of 66 have a Wp ratio below 1, and countries in SSA do not appear to be large outliers in this re- gard. Indeed, the average figures for each region (presented on the right) suggest that the EAP, LAC, and ECA regions all have higher Wp ratios than SSA. The over- all sample average is 0.80. Figure 7, below, focuses only on the SSA region and presents the mean and me- dian Wp figures by country income group. There are two notable features here. Bhorat et al. 35 Figure 7. Average Wp Values for SSA by Country Income Group 1.5 1 Wp .5 0 UMI median UMI mean LMI median LMI mean LIC median LIC mean Sources: ILO Global Wage database, World Bank WDI. Firstly, the figure shows that the ratio of minimum wages in Africa to the maxi- mum minimum wage on the continent is greater than the similar ratio in relation to mean wages. Secondly, the LIC mean and median values are higher than both the LMI and UMI sample of countries. This would suggest that despite absolute minimum wage levels increasing with GNI per capita, low-income African econo- mies are relatively more pro–minimum wage in their policy setting when com- pared with the mean relative wages on the continent. Reverting back to table 1, we provide in column three, the common in-country measure of the extent to which the minimum wage has “bite”—namely the Kaitz ratio. The ratio is simply that of the minimum wage to the mean wage. The Kaitz ratio values range from 0.03 in Burundi to a value of 1.27 in the Democratic Republic of the Congo. The overall SSA mean is instructive and of use in and of itself: for the region, minimum wages are 37 percent of mean wages and 28 percent of the median wage. This ratio varies by income classification; so while the Kaitz is 0.28 at the mean for UMI African economies, it is 0.31 for LMI economies and 0.46 for LICs. We observe substantially higher ratios for LI countries, where the group median is 0.16 points above the median for LMI countries in the region and 0.26 points above the UMI median. Simply put, low-income African countries are setting higher minimum wages relative to their domestic mean wages, compared with both UMI and LMI economies in the region. When compared with other 36 The World Bank Research Observer, vol. 32, no. 1 (February 2017) Table 1. Monthly Average Minimum Wage Estimates: Sub-Saharan Africa Country Minimum wage (US$ PPP) Mean wage (US$ PPP) Kaitz ratio Low-income economies Burkina Faso 138 210 0.66 Burundi 26 256 0.10 Chad 239 371 0.64 Congo, Dem. Rep 68 53 1.27 Ethiopia 77 175 0.44 Madagascar 128 183 0.7 Malawi 49 368 0.13 Tanzania 149 624 0.24 Uganda 65 464 0.10 Group mean 104 300 0.46 Group median 77 256 0.44 Lower-middle-income economies Congo, Rep. 145 526 0.28 Ghana 128 469 0.27 Kenya 331 979 0.34 Lesotho 242 377 0.64 Senegal 148 983 0.15 Swaziland 94 815 0.12 Zambia 98 252 0.39 Group mean 169 629 0.31 Group median 145 526 0.28 Upper-middle-income economies Algeria 531 1,003 0.53 Botswana 148 1,287 0.12 Gabon 418 2,356 0.18 Mauritius 218 1,424 0.15 South Africa 517 1,251 0.41 Group mean 366 1,464 0.28 Group median 418 1,287 0.18 Total SSA mean 188 687 0.37 Total SSA median 145 469 0.28 Other regional averages LAC mean 369 937 0.46 (1.24)* LAC median 289 859 0.37 (1.32)* EAP mean 317 884 0.38 (1.03)* EAP median 284 739 0.32 (1.14)* SA mean 233 386 0.63 (1.70)* continued Bhorat et al. 37 Table 1. Continued Country Minimum wage (US$ PPP) Mean wage (US$ PPP) Kaitz ratio SA median 255 368 0.59 (2.11)* ECA mean 325 1,136 0.3 (0.81)* ECA median 344 1,183 0.28 (1.00)* Notes: SSA aggregate estimates based on 21 countries, EAP sample based on eight countries, LAC sample based on 16 countries, SA sample based on four countries, ECA sample based on 17 countries. The latest available data for each country was utilized. Wages are monthly, and where there is more than one minimum wage schedule, we have estimated the average minimum wage across the within-country schedules. All estimates are in current US$ PPP. Asterisk (*) indicates ratio of measure to SSA mean or median. Sources: ILO Global Wage database, World Bank WDI. regions, however, the data show that mean and median Kaitz ratios are lower in SSA. Indeed, the mean and median Kaitz ratios of other regions relative to that of SSA show that it is only the ECA mean that has a lower Kaitz than that of the SSA region. This provides some support for the evidence presented in figure 6. Ultimately, then, the cross-country evidence on minimum wages in Sub-Saharan Africa indicates firstly that if we simply group countries according to broad income levels, minimum wages in Sub-Saharan Africa appear in general to be set lower than those in the rest of the developing world. Secondly, the evidence illustrates a positive and linear relationship between GDP per capita and the level at which the minimum wage is set. Consistent with the positive country poverty line—GDP rela- tionship, and in keeping with other developing countries—African economies reveal an upward adjustment in the value of the minimum wage as the economies in the region grow and develop. Thirdly, when assessing the extent to which African gov- ernments may be more pro–minimum wage, or not, in setting wage levels compared to other regions, our cross-country evidence indicates that there is no tendency to- wards progressive minimum wage policy in our sample. Indeed, the regional Kaitz ratios presented in table 1 are lower for SSA than any other region except ECA. Finally, it is clear again that when examining the Kaitz index, while SSA economies overall are setting minimum wages at just over a third of the mean wage, substan- tially higher ratios are observed for low-income countries in the region. Variations in African Minimum Wage Schedules: A Country-Level Overview For the purposes of a cross-country comparison, the above estimates have aggregated across the different minimum wage schedules that exist within many economies. 38 The World Bank Research Observer, vol. 32, no. 1 (February 2017) There is, however, much granularity and nuance that is overlooked when presenting average minimum wage estimates by country in this fashion. We return here, then, to a more detailed consideration of minimum wage schedules, focusing on seven SSA economies for which we have the appropriate microdata. We have alluded to fact that many minimum wage systems in the region in- clude a range of sectoral and occupational schedules, and it is these complex schedules that make a single figure appear blunt as it masks substantial heteroge- neity at a country level. As an example, the case of Kenya’s minimum wage regime is instructive in terms of the complexity of minimum wage systems: Kenya has had an active minimum wage policy since it achieved independence in 1964. Minimum wages are set by Ministerial order following recommendations by a tri- partite council and public consultation (ILO 2014). The wage schedule that is pro- duced is intricate. Wages are set at different rates for agricultural and nonagricultural occupations, and within these categories there are geographical and occupational distinctions, each with a unique set of wages. Table A2 presents daily minimum wage rates for the agricultural sector, which is broken down into 10 categories according to the specific type of employee.9 Within the agricultural sector, daily wages can range from 203 Kenyan Shillings for a general worker to 370 Kenyan Shillings for a lorry or car driver. Table A3 details monthly minimum wages for the nonagricultural sector, which is firstly disaggregated by geographic region into three categories: cities, municipalities and town councils, and other areas. Within each of these three geographical areas, wages are further delineated across 15 different employee types. Monthly minimum wages for nonagricultural workers range from 5,217 to 22,070 Kenyan shillings. In total, Kenya has 55 sep- arate minimum wage rates. It is clear how the complexity of Kenya’s wage determination system makes ob- taining accurate aggregate estimates difficult. But it must be emphasized that Kenya is not an exception in this regard. South Africa’s legislated minimum wage schedule, for example, is even more complex for certain sectors, and in total the country currently has 124 different minimum wage rates. This excludes minimum wages agreed upon within Bargaining Councils, which would push the number of specific wage rates in the thousands. In table 2, below, we provide a brief overview of the number of minimum wage schedules for 12 countries in SSA. The data show that seven of the 12 countries in the sample have 10 or more wage schedules, with South Africa being an outlier in this sample. A differentiated, or complex, wage schedule can be useful in that it takes account of variations across worker skill levels, geographic regions, and sec- tors. Yet increasing levels of complexity also makes wage setting, as well as en- forcement and compliance, more difficult. The ILO (2014) suggests that in general the complexity of a wage schedule should be commensurate with the county’s re- source availability, where simpler wage schedules are more suitable if the Bhorat et al. 39 Table 2. Number of Minimum Wage Schedules by African Country Country Number of wage schedules Uganda 1 Mali 1 Ghana 1 Malawi 1 Nigeria 2 Botswana 10 Zambia 10 Tanzania 29 Namibia 32 Kenya 55 Ethiopia (public sector) 57 South Africa 124 Average 27 Source: ILO TRAVAIL Database. resources dedicated to minimum wage systems are few. A complex schedule such as Kenya’s certainly creates a more challenging set of rates to establish each year and enforce. Labor market regulations in several SSA countries would probably benefit from a re-assessment of current minimum wage systems in this regard, with a view toward greater simplification. Given the complexity of schedules, it is useful to look beyond cross-country ag- gregates, which, while useful, cannot provide a detailed picture or give any indica- tion toward levels of compliance with minimum wage laws. In order to do this, we make use of household survey data for seven SSA countries (South Africa, Uganda, Kenya, Zambia, Tanzania, Mali, and Namibia). As tables 2, A2, and A3 indicate, however, there remain significant complexities within several of these country minimum wage systems—many of which go be- yond the detail provided by a basic labor force survey. To deal with this, for the countries with more than one minimum wage we select and focus on two key sec- tors. The first, which we call the “lower floor” is a low-paid, unskilled sector cov- ered by a general minimum wage (usually agriculture or “general workers”), while the second sector we select is higher-paid and medium-skilled (such as retail trade or working as a clerk), and we call this the “upper floor.”10 This allows us to present the granularity of minimum wage schedules while at the same time enabling us to estimate more accurate Kaitz ratios (for specific sectors), as well as exploring levels of compliance (see the section Minimum Wage Compliance in Sub- Saharan Africa). To begin, we present kernel density estimates that provide a basic picture of where specific minimum wages are set relative to the wage distributions 40 The World Bank Research Observer, vol. 32, no. 1 (February 2017) Figure 8. (a) Distribution of Wages, Zambia (2010); (b) Distribution of Wages, Tanzania (2011); (c) Distribution of Wages, South Africa (2013); (d) Distribution of Wages, Namibia (2012); (e) Distribution of Wages, Uganda (2012); (f) Distribution of Wages, Kenya (2005/6). (a) (b) (c) (d) (e) (f) Source: Zambia: Living Conditions Monitoring Survey (2010); Tanzania: Integrated Labour Force Survey (2005/06); South Africa: Labour Market Dynamics Study (2013); Namibia: Labour Force Survey (2012); Uganda: Uganda National Panel Survey (2012); Kenya: Kenya Integrated Household Budget Survey (2005- 06). Bhorat et al. 41 in each country.11 We include all employed wage earners in the sample, regardless of occupation, sector, or whether they are covered by a minimum wage or not. In all cases, the distributional graphs indicate the lack of a traditional “spike” at either of minimum wage schedules. This is interesting in two respects: Firstly, un- like most of the developed country literature, this six-country sample of African economies shows no evidence of a spike in the wage distribution relevant to the level of the minimum. Secondly, though, this lack of a spike is also consistent with fairly fat tails to the left of both minimum wage schedules. Put differently, there is a significant share of earners not adhering to the minimum wage. This incidence and deficit of noncompliance is taken up in the next section. Ultimately, the notion of a subset of African economies displaying relatively high minimum wage levels that in turn coexist with a fair degree of independence from the country’s self- same wage distribution is a key result. Table 3, below, provides more detailed information on minimum wages for the six SSA countries in the figures above, as well as for Mali. In two of the seven countries in the table, there are single national minimum wages (Mali and Uganda), while in the other five there are detailed sectoral and occupational wage schedules. As noted above, for each of these five countries we select a lower and upper wage floor and identify the employees covered by these sectoral wage levels. For the two countries with a single national minimum wage, we simply include all workers classified as wage-earning employees in our analysis. The table provides an overview of wage levels, minimum wage levels, coverage, and we also draw at- tention to the difference between the level of the lower and upper floors, where ap- plicable. What emerges firstly and most obviously is the within-country variation in minimum wage levels. Among the seven countries profiled in table 3 and across the lower and upper floors in the same country, large variations are apparent both in the level of legis- lated minimum wages and in mean and median earnings. Column five shows that minimum wage levels differ substantially across the lower and upper floor categories within the same country, with wages for workers in lower floor, or unskilled catego- ries, set at 54 percent of wage levels for upper floor categories on average. While the heterogeneity in minimum wage levels makes for a complex wage schedule, it also appears to be reflective of the substantial intersectoral wage inequality, which is evi- dent in the differences in average wages for the same country. It is also worth noting the large difference between the mean and median earnings across any one category of workers, who are assigned here to a lower or upper floor. These estimates are sug- gestive of high levels of intrasector wage inequality—where mean wages are signifi- cantly larger than median wages. On average across the group of countries, mean wages (US$457) are 55 percent higher than median wages (US$251). Importantly, though, the data reveals an important characteristic of minimum wage setting in Africa—namely that in general these sectoral wages, given their 42 The World Bank Research Observer, vol. 32, no. 1 (February 2017) Table 3. Monthly Minimum Wages (US$ PPP) In Seven African Economies: Upper and Lower Floors Countries Sector % of total Minimum wage Lower/upper Mean Median employees (US$ PPP) floor wage wage South Africa (2013) Lower floor 4 441 0.65 558 329 Upper floor 13 680 1,475 526 Kenya (2005) Lower floor 7 116 0.44 146 96 Upper floor 2 264 470 353 Zambia (2010) Lower floor – 69 0.38 124 78 Upper floor 7 185 417 284 Tanzania (2007) Lower floor 61 162 0.59 100 52 Upper floor 2 274 190 87 Namibia (2012) Lower floor – 388 0.63 257 174 Upper floor 4 615 1,057 677 Uganda (2012) National 100 65 N/A 227 107 Mali (2013)*** National 100 132 N/A – – Mean 0.3 283 0.54 457 251 Notes: Figures only include those classified in household surveys as being wage-earning employees. The Ugandan minimum wage has not been updated since 1984, thus we take the 1984 rate and adjust for inflation to obtain a comparable 2012 figure. The data from Mali is taken from Rani et al. (2013), and we are unable to calcu- late mean or median wages. Wages are in current monthly US$ PPP. Sources: South Africa: Labour Market Dynamics Study (2013); Kenya: Kenya Integrated Household Budget Survey (2005–6); Uganda: Uganda National Panel Survey (2012); Mali: Rani et al. (2013); Zambia: Living Conditions Monitoring Survey (2010); Tanzania: Integrated Labour Force Survey (2005–6); Namibia: Labour Force Survey (2012). specific targeting—will not cover a large share of workers. Apart from Uganda and Mali, where a national minimum wage exists, coverage of most sectoral minimum wages is fairly low. In the pursuit of higher wages for vulnerable workers, it is pos- sible that many African economies do not end up covering a large share of the wage employed. To go beyond focusing on the variation in wages and wage levels, we examine the Kaitz ratios using both mean and median wages for workers in the lower and upper floor categories. These ratios give an indication of how high minimum wages are set relative to average wages within in each category and also provide suggestive evidence regarding compliance, which we take up in more detail shortly. Figure 9, below, presents the Kaitz ratios for mean wages across categories and for each country. The ratios vary widely across countries, as well as across the lower and upper floor categories. In Namibia, for example, for workers in the lower floor category (“general workers”), the minimum wage is set at 1.5 times the mean wage, while in Zambia this figure is 0.5. Overall, however, the average ratio is 0.93 among lower floor workers and 0.63 among upper floor workers. As a point of comparison, the average level of Bhorat et al. 43 Figure 9. Ratio of Minimum to Mean Wages, Seven African Economies 1.5 1 .5 0 Uganda Zambia Mali South Africa Kenya Namibia Tanzania Average Lower floor Upper floor Source: Authors’ calculations. minimum-to-mean wages for nine LMI countries outside of SSA12, presented in Rani et al. (2013), was 54 percent. Figure 10 presents the Kaitz ratio using median instead of mean wages, and the different results are stark, as table 3 suggested. The ratios are significantly higher, Figure 10. Ratio of Minimum to Median Wages, Seven African Economies 3 2 1 0 Uganda Mali Zambia Kenya South Africa Namibia Tanzania Average Lower floor Upper floor Source: Authors’ calculations. 44 The World Bank Research Observer, vol. 32, no. 1 (February 2017) with the average figures above 1, and the black dotted line makes this clear. Again, comparing these estimates against the sample of non-SSA countries from Rani et al. (2013), where the average ratio is 0.76, shows that indeed the figures are high. Overall, the Kaitz estimates for the SSA region highlight two main aspects of minimum wages: firstly, the country-specific nature of minimum wage frame- works, the level at which wages are set and the relation of this level to average wages for that covered group; and secondly, given that we are only focusing on covered workers in the two figures above, the data show that minimum wages are set high relative to average wages and are suggestive of significant noncompliance. This leads us to explore levels of noncompliance in more detail. Minimum Wage Compliance in Sub-Saharan Africa While the Kaitz ratios presented above are indicative of widespread noncompliance with minimum wage laws, a more robust method is required to investigate further. We apply an Index of Violation13 from Bhorat, Kanbur, and Mayet (2012) to cal- culate the level and depth of noncompliance, which provides a more Figure 11. Average Compliance Rates (V0), African and Developing Country Comparison .8 .6 .4 .2 0 Philippines Uganda Namibia Tanzania Indonesia Non-SSA Average Mali Mexico Zambia Brazil Kenya Costa Rica SSA Average India Viet Nam South Africa Peru Turkey Source: Authors’ calculations and Rani et al. (2013). Bhorat et al. 45 comprehensive picture. To do this, we combine the lower and upper floor catego- ries presented above where applicable, and we compare the resulting estimates with those of Rani et al. (2013). The Index of Violation allows us to calculate the level of noncompliance, or V0, which is simply the percentage of workers who earn below the minimum wage that applies to them. It also allows us to go beyond this to calculate the depth of this noncompliance, or V1, which measures how far below the minimum wage these workers earn on average. The data in figure 11 clearly show that for most countries in the SSA region, for the sectors we include, noncompliance is widespread. On average, 58 percent of workers earn below the minimum wage legislated for them. This is compared to an average of 30 percent for the non-SSA countries in the figure. This relationship holds for the mean and median, despite outlier economies such as Tanzania. The figure also highlights the cross-country differences in levels of noncompliance. In Zambia, approximately 36 percent of workers earn subminimum wages, while this rises to 80 percent in Tanzania. These numbers reinforce the picture provided by the Kaitz ratios above where minimum wage rates were shown to be set high rela- tive to average wages. Figure 12 moves beyond the level of compliance to explore how far below the minimum wage workers are on average. This gives a more nuanced picture of Figure 12. Average Depth of Noncompliance (V1), African and Developing Country Comparison .6 .4 V1 .2 0 Philippines Uganda Tanzania Namibia Indonesia Non-SSA Average Mali Brazil Zambia Kenya SSA Average India Costa Rica Mexico Viet Nam South Africa Peru Turkey Source: Authors’ calculations and Rani et al. (2013). 46 The World Bank Research Observer, vol. 32, no. 1 (February 2017) noncompliance and how severe noncompliance levels are when examining the av- erage distance below the minimum wage that workers are earning. The results suggest an interesting switch: while noncompliance levels were higher in the African sample of economies, relative levels of noncompliance are in fact higher in the non-African sample. Hence, our estimates show that the SSA re- gional average V1 estimate stands at 0.30 while the corresponding figure for the non-SSA countries is 0.35. This suggests, at least on the basis of this subsample of developing countries, that while Africa yields to higher levels of noncompliance than other developing countries, those who are below the minimum wage face a greater disadvantage in non-African economies. Put simply, absolute levels of non- compliance are higher in Africa, while relative levels of noncompliance are higher in non-African developing countries. Conclusion Most countries in Sub-Saharan Africa (SSA) have adopted minimum wage regula- tion. Although the sectors and fraction of workers covered are small given the low rates of formality and urbanization in SSA, as the number of covered workers grows, wage regulation will become increasingly significant for the economy as a whole. In addition, there can be spillover effects in uncovered sectors, which can be exacerbated, as we show in the paper, when wage regulation is particularly pro- gressive. Current experience with minimum wages in SSA is thus relevant for ana- lysts and policy makers as economies may consider redesigning their minimum wage architecture in particular, and active labor market policy in general. In examining the variety of minimum wage frameworks in the region, it is clear that the typology of minimum wage schedules, as well as the levels at which these minima are set, varies considerably across countries in Africa. Our evidence shows that higher minimum wage values are associated with higher GDP per capita. In addition, utilizing two different pieces of evidence, we illustrate that SSA does not seem to reflect a particular bias towards a more pro–minimum wage policy relative to other regions of the developing world. Importantly, however, we find that mini- mum wages in low-income countries in SSA are set at values relative to the coun- try’s mean wage that are higher than those in lower- and upper-middle income countries in Africa. There is limited research on the employment effect of minimum wages in SSA, but the findings for the four countries (Ghana, Kenya, Malawi, and South Africa) are consistent with the broad summary of global research. By and large, introducing and raising the minimum wage has a small negative impact or no measurable nega- tive impact. However, there is significant variation around this average finding, and some country studies do present evidence of employment losses in excess of 10 Bhorat et al. 47 percent in certain cases. In short, employment elasticities are neither constant nor linear. Where increases in a minimum wage are large and immediate, this can result in employment losses, but more modest increases usually have few observably ad- verse effects on employment and may have positive impacts on wages. The great variability in findings on employment can be explained partly by the great variation in the detail of the minimum wage regimes and schedules country by country, but also by the variations in compliance. Our data on minimum wage compliance in Africa illustrates that the continent has higher levels of absolute noncompliance when compared with other developing countries, but lower levels of relative noncompliance. However, there is considerable variation across coun- tries. We find that higher Kaitz indices are associated with higher levels of non- compliance, but more detailed explanation of noncompliance is an important item on the research agenda. This paper has provided an empirical overview of minimum wages in Sub- Saharan Africa. It is evident that research in this area is still at a very early stage, with this paper in essence attempting to craft the broad contours of the nature and extent of minimum wage setting on the continent. While work on minimum wages is fairly mature in many OECD countries, our understanding of minimum wage policy and its impact in SSA is not. This is in large part due to a severe lack of nec- essary data. The release of country-level earnings and employment data at regular intervals lies at the heart of a future country-focused minimum wage research agenda for Africa. Notes Haroon Bhorat is the Director of the Development Policy Research Unit, School of Economics, University of Cape Town, South Africa; Ravi Kanbur is the T. H. Lee Professor of World Affairs, International Professor of Applied Economics and Management, and Professor of Economics, Cornell University; and Benjamin Stanwix is a Senior Researcher in the Development Policy Research Unit, University of Cape Town, South Africa. The authors of this paper acknowledge Benjamin Jourdan for providing a number of graphs and tables used for the study. 1. In Sub-Saharan Africa, data indicates that approximately 19 percent of the labor force is in wage employment, while 74 percent is in agricultural or nonfarm self-employment (Bhorat, Naidoo, and Pillay 2015). In West Africa, for example, the informal sector accounts for approximately 50 per- cent of national output, over 80 percent of employment and 90 percent of new jobs (Benjamin, Golub, and Mbaye 2015). 2. While there have been several studies analyzing the lighthouse effect of minimum wage laws in many countries around the world, we know of no such work done for African countries. Related research on the impact of minimum wage laws in sectors with weak enforcement has been done in South Africa (see Dinkelman and Ranchhod 2013 and Bhorat, Kanbur, and Stanwix 2015). 3. Indeed, many countries have wage-setting legislation that takes cost of living measures into account when updating minimum wages. 4. Belman and Wolfson (2014); Neumark, Salas, and Wascher (2014); Schmitt (2013); Neumark and Wascher (2007, 2008). 48 The World Bank Research Observer, vol. 32, no. 1 (February 2017) 5. See, for example, Agell and Lommerud (1997) or for a discussion of the possible impacts of minimum wages from a behavioral perspective, see Falk, Fehr, and Zehnder (2006). 6. The country studies focus on Brazil, Chile, Colombia, Costa Rica, Indonesia, Mexico, Puerto Rico, and Trinidad and Tobago. Neumark and Wascher (2007) use the term “developing country” to describe what we have referred to as low- and middle-income countries. 7. The table below shows the percentage split across world regions between setting a national minimum wage versus some form or regional, sectoral, or occupational minimum wage. The percent- ages do not add up to 100 percent given that not all countries in a region have minimum wages. Source: ILO (2013). 8. It should be noted that in the case of Burundi the minimum wage has not been updated since 1988; the figure presented here is adjusted for inflation to 2012, priced and converted in US$ PPP. 9. Unskilled employee; stockman, herdsman, and watchman; skilled and semiskilled employees; house servant or cook; farm foreman; farm clerk; section foreman; farm artisan; tractor driver; com- bined harvester-driver; lorry driver or car driver. 10. The specific sectors for each country are as follows: South Africa Lower floor – Agriculture Upper floor – Wholesale and retail Kenya Lower floor – Agriculture Upper floor – Employee type [G], see table A2 Zambia Lower floor – General worker Upper floor – Clerks Tanzania Lower floor – Agriculture Upper floor – Hospitality Namibia (wages set by collective bargaining) Lower floor – General workers Upper floor – Clerks 11. We only do this for six countries as the data for Mali are taken from Rani et al. (2013). 12. Brazil, Costa Rica, India, Indonesia, Mexico, Philippines, Peru, Turkey, Viet Nam. 13. If we consider a distribution of wages F(w), where the density function is f(w) and the mini- mum wage is W m ; then an index of violation can be calculated as follows: ðm  W a W m À Wi V ðaÞ ¼ f ðwÞ; Wm 0 where Wi are individual wages and a is the “violation aversion” parameter such that when: a ¼ 0, V(a) measures the percentage of workers below the minimum; when a ¼ 1, V(a) measures the aver- age gap between W m ; and when a ¼ 2, V(a) is the squared violation gap. This index allows us to cal- culate a family of measures that capture both absolute (V0) and relative (V1, V2) levels of noncompliance. Bhorat et al. 49 References Abowd, J., F. Kramarz, and D. Margolis. 1999. “Minimum Wages and Employment in France and the United States.” Working Paper 6996. National Bureau of Economic Research, Cambridge, MA. Agell, J., and K. E. Lommerud. 1997. “Minimum Wages and the Incentives for Skill Formation.” Journal of Public Economics 64 (1) : 25–40. Alatas, V., and L. Cameron. 2003. “The Impact of Minimum Wages on Employment in a Low Income Country: An Evaluation Using the Difference-in-Differences Approach.” Working Paper 2985. World Bank, Policy Research Department, Washington, DC. Almeida, R., and L. Ronconi. 2012. “The Enforcement of Labor Law in the Developing World: Some Stylized Facts from Labor Inspections.” Paper Presented at the 7th Institute for the Study of Labor/ World Bank Conference on Employment and Developing, 2012–11, New Delhi, India. n, M., and C. Page Andalo ´ s. 2008. “Minimum Wages in Kenya.” Discussion Paper 3390. Institute for the Study of Labor, Bonn, Germany. Bell, L. A. 1997. “The Impact of Minimum Wages in Mexico and Colombia.” Journal of Labor Economics 15 (S3) : S102–35. Belman, D., and P. J. Wolfson. 2014. “What Does the Minimum Wage Do?” WE Upjohn Institute, Kalamazoo, MI. Benjamin, N., S. Golub, and A. A. Mbaye. 2015. “The Informal Sector, Employment, and Economic Growth: Recommendations for Effective Policies.” In E. Zedillo, O. Cattaneo, and H. Wheeler eds., AFRICA at a Fork in the Road. 163–77. New Haven, CT: Yale Center for the Study of Globalization. Bhorat, H., R. Kanbur, and N. Mayet. 2012. “Minimum Wage Violation in South Africa.” International Labour Review 151 (3) : 277–87. ———. 2013. “The Impact of Sectoral Minimum Wage Laws on Employment, Wages, and Hours of Work in South Africa.” IZA Journal of Labor and Development 2 (1) : 1–27. Bhorat, H., R. Kanbur, and B. Stanwix. 2014. “Estimating the Impact of Minimum Wages on Employment, Wages, and Non-Wage Benefits: The Case of Agriculture in South Africa.” American Journal of Agricultural Economics 96 (5) : 1,402–19. ———. 2015. “Minimum Wages in Sub-Saharan Africa: A Primer.” Discussion Paper 9204. Institute for the Study of Labor, Bonn, Germany. Bhorat, H., K. Naidoo, and K. Pillay. 2015. “Growth, Poverty and Inequality Interactions in Africa: An Overview of Key Issues.” Unpublished mimeograph. University of Cape Town, Development Policy Research Unit, Cape Town, South Africa. Card, D., and A. Krueger. 1994. “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania.” The American Economic Review 772–93. Comola, M., and L. De Mello. 2011. “How Does Decentralized Minimum Wage Setting Affect Employment and Informality? The Case of Indonesia.” Review of Income and Wealth 57 (S1) : S79–99. Del Carpio, X. N., and C. W. Liang. 2013. “The Impacts of Minimum Wages on Employment, Wages and Welfare: The Case of Vietnam.” Working Paper. World Bank, East Asia and Pacific Social Protection and Labor Unit, Washington, DC. Del Capio, X., D. Margolis, and Y. Okamura. 2013. “Effects of Minimum Wages on Labor Markets and Welfare in the Philippines: A Micro-Simulation-Based Estimation.” Working Paper. World Bank, East Asia and Pacific Social Protection and Labor Unit, Washington, DC. 50 The World Bank Research Observer, vol. 32, no. 1 (February 2017) Del Carpio, X., J. Messina, and A. Sanz-de-Galdeano. 2014. “Minimum Wage: Does It Improve Welfare in Thailand?” Discussion Paper 7911. Institute for the Study of Labor, Bonn, Germany. ———. 2014. “Minimum Wage: Does It Improve Welfare in Thailand?” Discussion Paper 7911. Institute for the Study of Labor, Bonn, Germany. Dinkelman, T., and V. Ranchhod. 2012. “Evidence on the Impact of Minimum Wage Laws in an Informal Sector: Domestic Workers in South Africa. Journal of Development Economics 99 (1) : 27–45. Dykstra, S., B. Dykstra, and J. Sandefur. 2014. “We Just Ran Twenty-Three Million Queries of the World Bank’s Website.” Working Paper 362. Center for Global Development, Washington, DC. Eyraud, F., and C. Saget. 2005. The Fundamentals of Minimum Wage Fixing. Geneva, Switzerland: International Labour Organization. Fajnzylber, P. 2001. “Minimum Wage Effects throughout the Wage Distribution: Evidence from Brazil’s Formal and Informal Sectors.” Unpublished manuscript. Universidade Federal Do Belo Horizonte, Department of Economics and CEDEPLAR, Minas Gerais, Brazil. Falk, A., E. Fehr, and C. Zehnder. 2006. “Fairness Perceptions and Reservation Wages: The Behavioral Effects of Minimum Wage Laws.” The Quarterly Journal of Economics 121 (4) : 1,347–81. Fang, T., and C. Lin. 2013. “Minimum Wages and Employment in China.” Discussion Paper 7813. Institute for the Study of Labor, Bonn, Germany. Garbers, C. 2015. “The Impact of the Agricultural Minimum Wage on Farmworker Employment in South Africa: A Fixed Effects Approach.” Unpublished masters thesis. Stellenbosch University, Cape Town, South Africa. Gindling, T. H. 2014. “Does Increasing the Minimum Wage Reduce Poverty in Developing Countries?” Working Paper. World of Labor, Institute for the Study of Labor, Bonn, Germany. Gindling, T. and K. Terrell. 2005. “The Effect of Minimum Wages on Actual Wages in Formal and Informal Sectors in Costa Rica.” World Development 33 (11) : 1,905–21. ———. 2007. “The Effects of Multiple Minimum Wages throughout the Labor Market: The Case of Costa Rica.” Labour Economics 485–511. Ham, A. 2013. “Revisiting the Effects of Minimum Wages in Developing Countries: Evidence from a Particular Policy Change in Honduras.” University of Illinois at Urbana–Champaign. Hyslop, D., and S. Stillman. 2004. “Youth Minimum Wage Reform and the Labour Market. Discussion Paper 1091. Institute for the Study of Labor, Bonn, Germany. Harrison, A., and J. Scorse. 2010. “Multinationals and Anti-Sweatshop Activism.” American Economic Review 100 (1) : 247–73. Hertz, T. 2005. “The Effect of Minimum Wages on the Employment and Earning of South Africa’s Domestic Service Workers.” Working Paper 05-120. Upjohn Institute, Kalamazoo, MI. ILO. 2013. Global Employment Trends for Youth, 2013: A Generation at Risk. Geneva, Switzerland: International Labour Organisation. Jones, P. 1997. “The Impact of Minimum Wage Legislation in Developing Countries Where Coverage Is Incomplete.” Working Paper WPS/98-2. University of Oxford, Center for the Study of African Economies, Oxford, UK. Katz, L., and A. Krueger. 1992. “The Effects of the Minimum Wage on the Fast Food Industry.” Industrial and Labor Relations Review 6–21. Kharmis, M. 2013. “Does the Minimum Wage Have a Higher Impact on the Informal than on the Formal Labour Market? Evidence from Quasi-Experiments.” Applied Economics 45 (4) : 477–95. Bhorat et al. 51 Lanzona, L. 2012. “The Impact of Minimum Wages on Output and Employment in Labor-Intensive Manufacturing Industries in the Philippines.” Background Paper for the Philippine Development Report, 2013. World Bank, Manila, Philippines. Lemos, S. 2004. “Minimum Wage Policy and Employment Effects: Evidence from Brazil.” Economia 219–66. ——— 2006. “The Minimum Wage Effects in a Developing Country.” Discussion Paper 01/06. University of Leicester, Leicester, UK. ——— 2007. “Minimum Wage Effects across the Private and Public Sectors in Brazil.” Journal of Development Studies 700–20. ——— 2009. “Minimum Wage Effects in a Developing Country.” Labour Economics 16 (2) : 224–37. Livingstone, I. 1995. “Minimum Wage Policy, Rural Development and Poverty Alleviation in Malawi.” Development and Change 26 (2) : 729–52. Maloney, W., and J. Nun~ ez Mendez. 2004. Measuring the Impact of Minimum Wages: Evidence from Latin America. Chicago: Chicago University Press. ´ s. 2004. “Who Benefits from Labor Market Regulations? Chile 1960– Montenegro, C., and C. Page 1998.” In J. Heckman, and C. Page´ s, eds., Law and Employment: Lessons from Latin America and the Caribbean, 401–34. Chicago: University of Chicago Press. Nattrass, N., and J. Seekings. 2014. “Job Destruction in Newcastle: Minimum Wage-Setting and Low- Wage Employment in the South African Clothing Industry.” Transformation: Critical Perspectives on Southern Africa 84 (1) : 1–30. Neumark, D., W. Cunningham, and L. Siga. 2006. “The Effects of the Minimum Wage in Brazil on the Distribution of Family Incomes: 1996–2001.” Journal of Development Economics 80 (1) : 136–59. Neumark, D., and W. Wascher. 2006. “Minimum Wages and Employment: A Review of Evidence from the New Minimum Wage Research” Working Paper W12663. National Bureau of Economic Research, Cambridge, MA. ———. 2007. “Minimum Wages and Employment.” Discussion Paper 2570. Institute for the Study of Labor, Bonn, Germany. ———. 2008. Minimum Wages. Cambridge, MA: MIT Press. Neumark, D., J. I. Salas, and W. Wascher. 2014. “Revisiting the Minimum Wage—Employment Debate: Throwing out the Baby with the Bathwater?” Industrial and Labor Relations Review 67 (3 Suppl) : 608–48. Nguyen, C. V. 2010. “The Impact of a Minimum Wage Increase on Employment, Wages and Expenditures of Low-Wage Workers in Vietnam.” Archive Paper 36751. Munich Personal RePEc, Munich, Germany. Piore, M. J., and A. Schrank. 2008. “Toward Managed Flexibility: The Revival of Labour Inspection in the Latin World.” International Labour Review 147 (1) : 1–23. Rama, M. 2001. “The Consequences of Doubling the Minimum Wage: the Case of Indonesia.” Industrial and Labor Relations Review 54 (4) : 864–81. Rani, U., P. Belser, M. Oelz, and S. Ranjbar. 2013. “Minimum Wage Coverage and Compliance in Developing Countries.” International Labour Review 152 (3–4) : 381–410. Ravallion, M., S. Chen, and P. Sangraula. 2009. “Dollar a Day Revisited.” World Bank Economic Review 23 (2) : 163–84. Schmitt, J. 2013. “Why Does the Minimum Wage Have No Discernible Effect on Employment?” Center for Economic and Policy Research 22 : 1–28. 52 The World Bank Research Observer, vol. 32, no. 1 (February 2017) Strobl, E., and F. Walsh. 2003. “Minimum Wages and Compliance: The Case of Trinidad and Tobago.” Economic Development and Cultural Change 51 (1) : 427–50. Suryahadi, A., W. Widyanti, D. Perwira, and S. Sumarto. 2003. “Minimum Wage Policy and Its Impact on Employment in the Urban Formal Sector.” Bulletin of Indonesian Economic Studies 39 (1) : 29–50. Wang, J., and M. Gunderson. 2011. “Minimum Wage Impacts in China: Estimates from a Prescribed Research Design, 2000–2007.” Contemporary Economic Policy 29 (3) : 392–406. Appendix Figure A1. Wage-Employment Elasticities and Kaitz Ratios 2 0 Elasticity -2 -4 .2 .3 .4 .5 .6 .7 Kaitz Ratio Elasticity Fitted values Source: Neumark and Wascher (2007) and authors’ calculations. Table A1. Minimum Wage System Classification for Selected African Countries National Sectoral/occupational Hybrid Algeria Botswana Burkina Faso Cameroon Kenya Angola Cote d’Ivoire Malawi Egypt Mozambique Ghana Rwanda Mali South Africa Uganda Tanzania Source: ILO TRAVAIL database. Bhorat et al. 53 Table A2. Agricultural Minimum Wages, Kenya, 2013 Type of employee Monthly wage (shillings) Unskilled employee 4,854.35 Stockman, herdsman, and watchman 5,606.05 House servant or cook 5,541.55 Farm foreman 8,757.20 Farm clerk 8,757.20 Section foreman 5,669.20 Farm artisan 5,802.05 Tractor driver 6,152.70 Combined harvester-driver 6,778.10 Lorry driver or car driver 7,113.25 Source: Ministry of Labour Social, Security and Services (2013). Table A3. Nonagricultural Monthly Minimum Wages, Kenyan Shillings, Kenya, 2013 Type of Nonagricultural industry in cities Nonagricultural industry at Nonagricultural industry employee (Nairobi, Mombasa, and Kisumu) all municipalities and town councils at all other areas a 9,780.95 9,024.10 5,217.95 b 10,563.60 9,372.20 6,028.95 c 10,911.70 10,116.15 6,223.65 d 11,085.70 10,316.00 8,361.35 e 12,654.90 11,838.70 9,679.05 f 13,201.55 12,184.25 10,071.05 g 15,064.65 13,772.75 11,743.30 h 16,602.85 15,259.35 13,606.40 i 18,329.25 17,101.80 15,434.70 j 20,283.90 18,940.40 17,644.60 k 22,070.95 20,769.95 19,474.20 l 13,201.55 12,184.25 10,071.05 m 16,602.85 15,259.35 13,580.60 n 17,932.15 17,101.55 15,434.70 o 22,070.95 20,769.95 19,474.20 Notes: a General labor, including cleaner, sweeper, gardener, children’s ayah, house servant, day watchman, messenger. b Miner, stone cutter, turn boy, waiter, cook, logger line cutter. c Night watchman. d Machine attendant, sawmill sawyer, machine assistant, mass production machinist, shoe cutter, bakery worker, bakery assis- tant, tailor’s assistant. e Machinist (made-to-measure), shoe upper preparer, chaplis maker, vehicle service worker (petrol and service stations), bakery plant hand, laundry operator, junior clerk, wheeled tractor driver (light). f Printing machine operator, bakery machine operator, plywood machine operator, sawmill dresser, shop assistant, machine tool operator, dough maker, table hand baker or confectioner, copy-typist, driver (cars and light vans). g Pattern designer (draughts-man), garment and dress cutter, single hand oven man, charge-hand baker, general clerk, tele- phone operator, receptionist, storekeeper. h Tailor, driver (medium-sized vehicle). i Dyer, crawler tractor driver, salesman. j Saw doctor, caretaker (buildings). k Cashier, driver (heavy commercial vehicle), salesman-driver. l Ungraded artisan. m Artisan grade III. n Artisan grade II. o Artisan grade I. Source: Ministry of Labour Social, Security and Services. 54 The World Bank Research Observer, vol. 32, no. 1 (February 2017) Table A4. International LMI Country Studies on Minimum Wages Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Bhorat et al. Kharmis (2013) Argentina Date: 1992–2005 Informal workers, workers without Data: Permanent Household social security contribution, Survey (EPH) for the year 1993 experienced significant wage increases and the Continuous when the minimum wage was raised, Permanent Household survey while formal workers did not. (EPH-C) for the year 2004 Fajnzylber Brazil Date: 1982–1997 Estimates suggest that Significant minimum wage effects (2001) Data: Brazil’s Monthly employment elasticities across the whole wage distribution Employment Survey, are negative for most and both in the formal and the individuals aged 15 to 65 years low-wage workers, being informal sectors. We also find that the lower in absolute value total impact of minimum wages on for formal salaried workers’ earnings (derived from workers (around -0.1 at current and lagged effects) is positive the bottom of the wage but smaller than the contemporaneous distribution) than for one. Other results include higher low-wage informal earnings elasticities for men, adults, salaried and self- and heads of households than for employed (between -0.25 women, teenagers, and nonheads, and -0.35). respectively. Foguel, Ramos, Brazil Date: 1982–1999 Increases in the value of A 10 percent increase in and Carneiro Data: Monthly Employment the official minimum unemployment causes a 1.2 percent (2001) Survey (Pesquisa Mensal de wage tend to decrease drop in the earnings of informal Emprego/IBGE), official formal employment (- workers, as opposed to a fall of only minimum wage rates from the 0.001 to -0.024) and 0.9 percent in the wages of formal Ministry of Labor, Brazilian increase informal workers. Institute of Geography and employment (0.0004 to Statistics 0.003). continued 55 Table A4. Continued 56 Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Lemos (2004) Brazil Date: 1982–2000 Small negative effects on An increase in the minimum wage Data: Brazil’s Monthly employment (-0.05), and strongly compresses the wage Employment Survey and it is dominated by the distribution. Brazilian Labour Ministry data hours effect, or reduction in hours worked. Lemos (2006) Brazil Date: 1982–2000 Minimum wage has no In the formal sector, a 1 percent Data: Brazil’s Monthly adverse effect on increase in the minimum wage Employment Survey and employment in Brazil increases the wages of those in the Brazilian Labour Ministry data between 1982 and 2000, 25th percentile by 0.33 percent and of despite the sizeable wage those in the 50th percentile by 0.10 effects found in both the percent (evaluated at the average formal and informal “fraction at” 11.6 percent). In the sectors. informal sector, it increases the wages of those in the 25th percentile by 0.31 percent and of those in the 50th percentile by 0.48 percent. Lemos (2007) Brazil Date: 1982–2000 No evidence of adverse In the private sector, a 10 percent Data: Brazil’s Monthly employment effects in increase in the real minimum wage Employment Survey and either the public or increases the wages of those in the Brazilian Labour Ministry data private sectors at the 20th percentile by 3.41 percent and of aggregate level or for those in the 30th percentile by 2.55 vulnerable groups such percent. In the public sector, it as teenagers, women, increases the wages of those in the and the low educated. 20th percentile by 1.79 percent and of Minimum wage policies those in the 30th percentile by 1.14 in Brazil appear to be a percent. potentially viable antipoverty instrument. The World Bank Research Observer, vol. 32, no. 1 (February 2017) continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Bhorat et al. Lemos (2009) Brazil Date: 1982–2004 No effect found in either Minimum wage compresses the wage Data: Brazil’s Monthly the formal or informal distribution of both the formal and Employment Survey and sector (0.080 to 0.358). informal sectors in Brazil in May Brazilian Labour Ministry data 1995. McIntyre Brazil Date: 1981–1999 (except 1991 Estimates reveal that the Mandated nonwage Lower minimum wages and laxer (2006) and 1994) minimum wage in Brazil benefits and the enforcement of the law both increase Data: Pesquisa Nacional de does not increase minimum wage law wages among the low-skilled and Amostra de Domicilios (PNAD) unemployment; rather, it have no effect on decrease wage inequality. raises formality. employment but do encourage informality and lower total compensation. Lower minimum wages encourage workers to formalize their benefits: a 10 percent decrease in the minimum wage increases by 1.9 percent the number of workers paying all payroll taxes. The average formality premium is highest continued 57 Table A4. Continued 58 Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts among the least educated. Neumark, Brazil Date: 1996–2001 Negative employment Estimates provide no evidence that Cunningham, Data: Brazilian Monthly effects (-0.07). minimum wages in Brazil compress and Siga Employment Survey (Pesquisa the income distribution—lifting family (2006) Mensal do Emprego, or PME) incomes at the lower points of the in- come distribution—and, if anything, sometimes indicate that minimum wages have the opposite effect of re- ducing family incomes in the lower tail of the distribution. Montenegro and Chile Date: 1960–1998 Results suggest that both ´ s (2004) Page Data: Household surveys from minimum wages and job the University of Chile’s security regulations Economics Department reduce the employment opportunities of the young, unskilled, and particularly unskilled youth while promoting the employment rates of skilled and older workers. We have also found indications that job security regulations may force some workers, continued The World Bank Research Observer, vol. 32, no. 1 (February 2017) Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts particularly women and the unskilled, out of Bhorat et al. wage employment and into self-employment. Fang and Lin China Date: 2004–09 Minimum wage changes (2013) Data: Urban Household Survey have significant adverse (UHS) effects on employment in and minimum wage data the Eastern and Central collected at the county level regions of China and result in disemployment for females, young adults, and low-skilled workers. Youth: -0.136 to -0.156 At-risk groups: -0.265 to - 0.340. Wang and China Date: 2000–07 Negative employment Gunderson Data: China Population Statistic effects in slower-growing (2011) Yearbook regions (-0.156 to - 0.178); larger negative effects in non-state- owned organizations that tend to be more responsive to market pressures; much larger lagged effects reflecting the time needed for adjustments to occur; no adverse employment effects in the prosperous 59 and growing. continued Table A4. Continued 60 Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Bell (1997) Colombia Date: 1977–1987 Substantial Minimum wages have a strong impact Data: Colombia’s Annual disemployment effects of on wages judging by their proximity to Industrial Survey minimum wages are the average wage and time series found, where the impact estimates. is estimated at roughly 2 percent to 12 percent during 1981–87. Implied Elasticities suggest that the increase in the rela- tive value of the mini- mum wage in Colombia from 1977 to 1987 (roughly 15 percent) had the effect of reducing manufacturing employ- ment by 5 percent over this period. Manloney and Colombia Date: 1997 and 1999 A rise in the minimum The effect on real wages of a change in Nun~ ez Mendez Data: National Statistical Agency wage has a statistically the real minimum wage is high for (2004) (DANE) and National very significant impact those earning 70 percent to 90 Household Survey (ENH) on the probability of percent of the minimum wage; 87 becoming unemployed percent of the rise in minimum wages that again decreases with is communicated to wages. a rising position in the continued The World Bank Research Observer, vol. 32, no. 1 (February 2017) Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts wage distribution (- 0.161 to -0.356). Bhorat et al. El Hamidi and Costa Rica Date: 1976–1992 Increase in the minimum Terrell (2001) Data: Household Survey of wage relative to the Employment and average wage is Unemployment, Ministry of associated with an Labor Wage data increase in the level of covered sector employment by 0.56 percent, but no effect on the number of self- employed over time, and an increase in the average number of hours worked per week by 0.14 percent in the covered sector and 0.34 percent in the uncovered sector. These findings may be interpreted as supporting the monopsonistic model, which predicts that increases in wages can increase employment up to the point where the marginal cost is equal to the marginal revenue. Gindling and Costa Rica Date: 1988–2000 A 10 percent increase in Legal minimum wages have a Terrell (2005) Data: Legal minimum wage data minimum wages lowers significant positive effect on the wages from the Ministry of Labor, employment in the of workers in the covered sector (with Household Surveys for Multiple covered sector by 1.09 an elasticity of 0.10), but no effect on 61 continued Table A4. Continued 62 Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Purposes by the Costa Rican percent and decreases the wages of workers in the uncovered Institute of Statistics and average number of hours sector. Census, and industry data from worked of those who the Costa Rican Central Bank remain in the covered sector by about 0.6 percent. Despite the wide range of minimum wages, the largest impact on the wages and employment of covered sector workers is in the lower half of the distribution. Eriksson and Czech Date: 1998–2000 Negative employment and The larger the proportion of low-paid Pytlokova Republic Data: Average Earnings working hours effects of - workers in a firm, the higher the (2004) Information System 14.4 percent and 5.1 increase in the firm’s average wage. percent, respectively, in the year following the 1998 minimum wage increase. Negative employment and working hours effects of - 5.1 percent and 5.4 percent, respectively, in the year following the 1999 minimum wage increase. Jones (1997) Ghana Date: 1970–1991 Provides fairly strong Implied elasticities Data: Yearbook of Labour evidence that the suggest that a large The World Bank Research Observer, vol. 32, no. 1 (February 2017) minimum wage in proportion of the continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Statistics (ILO), International Ghana had a negative public sector Bhorat et al. Financial Statistics (IMF), impact on employment workers displaced by African Employment Report in the formal private the minimum wage (1990), Penn World Tables, sector (-0.12 elasticity) shifted into informal World Bank Social Indicators ofand formal public sector sector employment. Development (1995), World (-0.17 elasticity). Bank’s Regional Informal employment Program on Enterprise increased as a result of Development (1992) increased minimum wage. Employment of women. Gindling and Honduras Date: 1990–2004 Disemployment effects in Wage elasticity of 0.29 overall. Terrell (2007) Data: Honduras Minimum Wage the private sector (-0.46). However, the welfare—the total Decrees and the Permanent earnings—of low-paid workers in the Household Surveys for Multiple large firm–covered sector falls with Purposes higher minimum wages. Alatas and Indonesia Date: 1990–1996 Some evidence of a Cameron Data: Annual Survey of negative employment (2003) Manufacturing Firms (Survei impact for small domestic Tahunan Perusahaan Industri, firms but no employment SI) impact for large firms— foreign or domestic. Comola and De Indonesia Date: 1996–2004 Minimum wage hikes The negative and Mello (2011) Data: Indonesian Statistics destroy formal sector jobs significant Bureau’s National Labor Force (-0.056), but these jobs coefficient on Survey (Sakernas) working age losses are more than unemployment (15 to 65 years), National compensated for by the seems to suggest Economic Survey (Susenas), expansion of the informal that the decrease in Industrial Survey (Survei sector (0.074), formal-sector Industri) suggesting that employment due to 63 minimum wage a rise in the relative continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts 64 legislation is hurting, value of the instead of protecting minimum wage vulnerable workers. shifts workers from “queuing” unemployment to the inactive population of the informal sector. Del Carpio, Indonesia Date: 1993–2006 The employment effects of Minimum wages are more binding in Nguyen, and Data: Annual Manufacturing minimum wages are small firms than in large firms. Wang (2012) Survey (Survei Industri or SI) significant and negative and the National Socio- among all firms (-0.0233 Economic Survey (Susenas) to -0.0542), but more predominant in small firms and less educated workers and less among large firms and workers with high school education. Disemployment effects are stronger for nonproduction workers (À0.054) and women than for production workers (À0.023). Harrison and Indonesia Date: 1990–1996 Results suggest that the A 1 percent increase in the real value of Scorse (2010) Data: Annual Manufacturing minimum wage increases the minimum wage was associated Survey, Badan Pusat Statistik led to employment losses with a 0.675 percent increase in the for production workers real unskilled wage. across all sectors in manufacturing. The World Bank Research Observer, vol. 32, no. 1 (February 2017) Indonesia Date: 1993–2000 continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Bhorat et al. Magruder Data: Indonesian Family Life During the 1990s’ Shift from informal to The bottom quartile of the wage (2013) Survey (waves 1, 2, and 3), massive foreign formal employment distribution experiences massive wage Statistics Industry (SI) investment and rapid in local industries; gains when minimum wages grow, so economic growth, wages tradeable industry that a doubling of minimum wages is were increased and a saw no movements; associated with a 150 percent to 160 “big push” in the untradeable, percent increase in the 25th percentile economy increased nonindustrializable of the wage distribution. domestic demand. services saw a rise in Observing one out of the informal 33 districts in Indonesia, employment. formal employment increased and informal employment decreased only in local industries; tradeable manufacturing firms saw no growth in employment, and untradeable but nonindustrializable services saw an increase in informal employment. Rama (2001) Indonesia Date: 1993 Urban unemployment Average wages increased by 5 percent Data: Labor Force Survey decreased by 0 percent to to 15 percent. (Sakernas) 5 percent. The employment effects, however, varied substantially by firm size: small firms apparently 65 continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages 66 sector shifts experienced substantial decreases in employment, whereas some large firms actually saw their employment increase. Workers in these large firms, the author concludes, are the evident winners from the minimum wage hike. Suryahadi et al. Indonesia Date: 1988–2000 The imposition of Some workers who The minimum wage became more (2003) Data: National Labour Survey minimum wages has a lose jobs in the binding and impactful on the wage (Sakernas) negative and statistically formal sector and distribution in Indonesia from 1988 to significant impact on have to relocate to 2000. employment in the urban the informal sector formal sector. The face lower earnings disemployment impact is and poorer working greatest for female, conditions. young and less educated workers, while the employment prospects of white-collar workers are enhanced by increases in minimum wages (-0.112). Andalon and Kenya Date: 1998–1999 Estimates indicate that a Minimum wages were positively ´ s (2008) Page Data: Central Bureau of 10 percentage point associated with wages of low-educated Statistics’ Integrated Labour increase in the minimum workers and women in Force Survey to median wage ratio nonagricultural activities, while no could be associated with such relationship is found for workers a decline in the share of in agriculture. The World Bank Research Observer, vol. 32, no. 1 (February 2017) formal employment of continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts between 1.2 percent and 5.6 percent and an Bhorat et al. increase of between 2.7 percent and 5.9 percentage points in the share of self-employment. Bell (1997) Mexico Date: 1984–1990 Minimum wages had Minimum wages had virtually no effect Data: Mexico’s Annual Industrial virtually no effect on on wages in the formal sector. Survey, Mexican Exuesta employment in the Significant numbers of workers are Nationale de Empleo formal sector. paid at or below minimum wages. Bosch and Mexico Date: 1989–2000 A decline in the real value of the Manacorda Data: Microdata, Encuesta minimum wage explains a very (2010) Nacional de Empleo Urbano significant increase in inequality observed in Mexico between the late 1980s and late 1990s. Feliciano (1998) Mexico Date: 1970, 1980, 1990 Large reductions in the Data: Industrial Census, Monthly minimum wage were Industrial Survey found to increase employment of females aged 15 to 64 years (À0.58 to À1.25). Demanded shifts away from older skilled males toward less-skilled male workers. Castillo- Puerto Rico Date: 1956–1987 The imposition of the US- Three data sets of earnings show Freeman and Data: U.S. Department of Labor level minimum wage to remarkable spikes at the relevant Freeman Minimum Wage Industry Puerto Rico distorted the minima in each distribution, implying (1992) Studies, Census Population for Puerto Rican earnings that the minimum wage law is a major Puerto Rico, Current distribution, substantially determinant of actual wages paid. Population Survey (CPS) for reduced employment on 67 Puerto Rico the island (0.20 to - continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages 68 sector shifts 0.91), reallocated labor across industries, and affected the characteristics of migrants to the United States. Eriksson and Slovak Date: 1998–2000 Negative effect on Average wage of firms employing, Pytlokova Republic Data: Average Earnings employment in 1998 relatively, many workers from the (2004) Information System after first wage hike, but lower end of the wage distribution is no effect in 1999 after raised more as a consequence of hikes second wage hike. (Very in the minimum wage. small number of observations that may not present adequate results.) Dinkelman and South Africa Date: September 2001, March No significant effects on There was a large, significant increase Ranchhod 2002, September 2002, March employment or hours of in wages after the minimum wage was (2012) 2003, September 2003, March work for domestic increased between 18.9 and 21.7 2004 workers in the informal percent. Data: South African Labour sector. Force Surveys Bhorat, Kanbur, South Africa Date: September 2000 to Results suggest a Substantial increase Substantial increase in farmworkers’ and Stanwix September 2007 significant employment in contract coverage wages by approximately 30 percent. (2014) Data: South African Labour reduction in agriculture for farmworkers Force Surveys from the minimum inSouth Africa. The wage—and in particular number of workers a noticeable move away with a written from employment of part- employment time workers—an in- contract increased to crease in wages on aver- reach 52 percent in The World Bank Research Observer, vol. 32, no. 1 (February 2017) age and a rise in 2007. nonwage compliance. continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Overall average of hours Bhorat et al. worked fell in the postlaw period, suggesting that employers adjusted to some extent on the inten- sive margin, and it ap- pears that hours of work increased by more in areas where wages were lower in the prelaw period—driven largely by the fall in part-time employment. Hertz (2005) South Africa Date: September 2001 to Hours of work per week Average wages of those employed September 2004 decreased (-0.47 for increased by approximately 20 Data: South African Labour women and -0.28 for percent. Force Surveys men), and employment fell (between -0.19 and - 0.33) for domestic service workers. Bhorat, Kanbur, South Africa Date: 2000–07 No clear evidence that the Evidence of a significant increase in real and Mayet Data: South African Labour introduction of minimum hourly wages in the postlaw period in (2012) Force Surveys wage laws had a the retail, domestic work, and security significant impact on sectors examined. The taxi and employment in a given forestry sectors did not experience an period for the retail, increase in wages. domestic work, taxi, security, and forestry sectors. Workers in the 69 retail (-4.5 percent), security (-4.5 percent), continued Table A4. Continued 70 Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts and domestic work (-7.7 percent) sectors experienced a reduction of hours, but increases in wages outweighed these effects. Murray and van South Africa Date: January to March 2005 No large disemployment Walbeek Data: 103 semistructured effects occurred for farm (2007) interviews of large- and workers, but there was medium-scale farming some indication that employers employers substituted at the lower skills margin as well as adjusted at the intensive margin of labor to reduce weekly wage costs. On average, workers hours were reduced between 27 and 35 hours per week, as opposed to the standard 45-hour work week. Conradie (2004) South Africa Date: 2004 Disemployment effects are Data: Survey of 190 wine and found with table grape table grape farmers farm workers (-0.59) and wine farm workers (À0.33). Nattrass and South Africa Date: 2010–11 The National Bargaining Seekings Data: National Bargaining Council for the Clothing (2014) Council for the Clothing and and Manufacturing The World Bank Research Observer, vol. 32, no. 1 (February 2017) Manufacturing Industry Industry launched a continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Bhorat et al. wage compliance drive in 2010, which resulted in the closing of four factories that threatened at least 20,000 jobs in Newcastle, South Africa. Garbers (2015, South Africa Date: 1997–2007 Findings indicate that unpublished) Data: October Household Survey formal unskilled farm and Labour Force Survey employment decreased by approximately 16 percent as a result of the 2003 agricultural minimum wage regulation, of which 7.5 percent is directly attributable to higher unskilled labor costs resulting from the wage floor. There is also evidence of skill and capital intensification resulting from the minimum wage. Del Carpio, Thailand Date: 1998–2010 Minimum wage increases In spite of substantial noncompliance, Messina, and Data: Labor Force Survey and have small the minimum wage in Thailand is Sanz-de- Household Socio-Economic disemployment effects on binding, and it has a bearing on actual Survey female, elderly, and less- wage (0.36). continued 71 72 Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Galdeano educated workers and (2014) large positive effects on the wages of prime-age male workers (À0.055). As such, increases in the minimum wage are associated with increases in household consumption per capita in general, but the consumption increase is greatest among those households around the median of the distribution. In fact, rises in the minimum wage increased inequality in consumption per capita within the bottom half of the distribution. Del Carpio, The N/A Real minimum wage Margolis, and Philippines increases had negligible Okamura effects on overall (2013) employment, owing to the limited coverage of minimum wage rules and high noncompliance. The World Bank Research Observer, vol. 32, no. 1 (February 2017) continued Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts Bhorat et al. Lanzona, 2012 The N/A Sectors with high Philippines coverage and compliance experienced negative employment effects. Strobl and Trinidad and Date: 1996–1998 Both large and small Males working in large firms tended to Walsh (2003) Tabago Data: Continuous Sample Survey employers in some cases have their wage increased to at least of Population responded to the the minimum level; some females in minimum wage by laying both large and small firms experienced off workers. a wage increase due to the minimum Noncompliance by wage. employers was proven to be a substantial issue in implementing increased minimum wages. Del Carpio and Vietnam N/A Low-income or otherwise Liang (2013) vulnerable workers (including women, youth, recent labor- market entrants, the low- skilled, nonmanagerial nonproduction workers such as cleaners or guards, elderly workers, and those employed by small firms) are particularly likely to beshut out of the formal continued 73 74 Table A4. Continued Study Country Parameters of study Employment/hours worked Formal/informal Wages sector shifts labor market as a result of overly high minimum wages. Nguyen (2010) Vietnam Date: 2004 and 2006 Minimum wage increase The effect of the minimum wage Data: Vietnam Household Living reduced employment of increase on wages and expenditures of Standard Surveys low-wage workers in the workers is not statistically significant. formal sector. However, workers who lost formal sector jobs were able to find jobs in the informal sector. The World Bank Research Observer, vol. 32, no. 1 (February 2017)