PRI Working Paper Series No. 2

PRI Working Paper Series No. 2 Input Text i

Contents List of Tables... ii List of Figures... iii ABSTRACT... iv Employment, Productivity, Real Wages and Labor Markets in Bangladesh... 1 A. Overview and Introduction... 1 B. What Has Been Happening to Employment, Productivity, and Real Wages... 2 The Supply of Labor... 2 Labor Productivity... 6 Real Wages Outturn... 8 Wage-Productivity Relationship... 8 Productivity-Wage Relationship in a More Disaggregated Framework... 9 Summary of structural transformation... 11 C. The Nature of Labor Markets in Bangladesh... 11 Agricultural Labor Market... 13 Labor Market in Non-agricultural Informal Activities... 15 Labor Market in Formal Activities... 15 Labor Market Interactions... 16 D. Policy Implications... 17 Sectoral Growth Policies... 17 Labor Market Policies... 21 Bibliography... 23 Annex A: A Note on Labor Force Data... 25 Annex B: A Note on Sectoral Composition of GDP... 26 Annex C: GDP, Employment and Real Wages Data Used in this Paper... 27 List of Tables Table 1: Sectoral Distribution of Employment (Percent)... 5 Table 2: Structural Change in Output and Employment 1974-21... 6 Table 3: Wage Profile of Day laborer, 21... 1 Table 4: Average Wages of Day laborer, 21... 1 ii

Table 5: The Structure of Bangladesh Manufacturing Sector, FY1974-FY1... 19 Table 6: GDP Contribution of the Services and Others Sector FY75-FY1... 21 List of Figures Figure 1: Trend in Labor Force, 1974-21 (millions)... 2 Figure 2: Percent of Working Age Population, 1974-21... 2 Figure 3: Labor Force Participation Rates, 1974-21... 3 Figure 4: Percent Female Workforce... 3 Figure 5: Formality of Employment... 3 Figure 6: Sectoral Distribution of Employment Formality 25/6... 4 Figure 7: Education Attainment of Labor Force... 4 Figure 8: Trend in Employment, 1974-21... 4 Figure 9: Open Unemployment Rate... 5 Figure 1: Underemployment Rates 21... 5 Figure 11: Trend in Average Labor Productivity, 'Takas, 1995-96 Prices... 6 Figure 12: GDP Per Person Employed in 28 (constant 199 PPP $)... 7 Figure 13: Trend in Labor Productivity, 1981-21 (1995-96 taka)... 7 Figure 14: Trend in Real Wages 1981-21... 8 Figure 15: Wage-Productivity Relationship 198-21... 9 Figure 16: Average Productivity in Services 2-21 (1995/96 taka)... 9 Figure 17: Growth in Average Labor Productivity 2-21... 9 Figure 18: Growth in Services Employment, 2-21... 1 Figure 19: Sectoral Labor Productivity... 14 Figure 2: Turning Points in Real Agricultural Wages... 14 Figure 21: Structure of Agriculture, FY81-FY1... 17 Figure 22: International Comparison of Rice Productivity 29... 18 iii

ABSTRACT Bangladesh seeks to raise its growth rate from around 6% per year to 8% by the end of the Sixth Plan period. It also aims to reduce the incidence of poverty from about 32 percent in 21 to 22 percent by 215. Real wages, labor productivity and employment outcomes are critical determinants of the ability to achieve and sustain these growth and poverty reduction targets. The paper analyzes the functioning of labor markets in Bangladesh and identifies the emerging challenges and constraints. Based on this analysis, the paper suggests how these constraints might be addressed. The paper shows that labor markets work flexibly in Bangladesh and wages are market determined based on the forces of supply and demand. Rising productivity in recent years is causing real wages to rise across the country in all three sectors (agriculture, manufacturing and services). Nevertheless, the transformation process of a primarily agrarian economy to a modern manufacturing and service based economy has still a long way to go. Much of the labor force is still low-skilled and engaged in low-productivity agriculture, informal services and informal manufacturing. Female labor force participation is still low and the utilization of the demographic dividend of a growing labor force is way below potential. Actions are needed in both the demand and supply side of the labor market. An additional challenge is presented by the absence of proper safeguards and regulatory reforms in the formal labor market. Policy actions include those aimed at raising labor productivity, investment, exports and GDP growth, those that strengthen the quality of labor through investment in human capital, and those that provide regulatory and prudential safeguards to protect the rights and wellbeing of the workers. iv

PRI Working Paper Series No. 2 Input Text Employment, Productivity, Real Wages and Labor Markets in Bangladesh Sadiq Ahmed 1 A. Overview and Introduction Bangladesh seeks to raise its growth rate from around 6% per year to 8% by the end of the Sixth Plan period (Government of Bangladesh 211a). It also aims to reduce the incidence of poverty from about 32 percent in 21 to 22 percent by 215. Real wages, labor productivity and employment outcomes are critical determinants of the ability to achieve and sustain these growth and poverty reduction targets. A recent study of South Asian employment situation suggests a positive story, including for Bangladesh (World Bank 212). Over the past two decades South Asia has created jobs at around the same pace as the growth in the working age population, and the average productivity of labor and real wages are rising. This is indeed comforting, giving rise to cautious optimism that South Asia and Bangladesh might indeed be gaining from the ongoing demographic dividend of a growing labor force. Yet there are many concerns in Bangladesh. Average productivity of labor remains very low. So, apart from generating adequate numbers of higher productivity and higher real wage jobs for employing the growing labor force, additional jobs must be created for absorbing labor that withdraws from low productivity activities, typically in agriculture. This problem of disguised unemployment renders the concept of open unemployment rather mechanical and not very useful for policy purposes 2. Accordingly, to reduce the problem of disguised unemployment in Bangladesh, total employment must grow faster than labor force and also employment in manufacturing and services must grow faster than in agriculture. The ability to create higher productivity jobs at a pace that exceeds the growth in the supply of labour depends on profitability and investment. Over the longer term on average Bangladesh has achieved fairly rapid growth in private investment, although public investment lacks dynamism. Private investment in labor intensive and export-oriented garments sector has been particularly beneficial for economic growth and employment creation (Ahmed and Sattar 23; Ahmed and Sattar 24; Mckinsey 211). Yet total and private investments rates have stagnated in the past few years despite a healthy savings rate. The success in the ready-made garments (RMG) based exportoriented manufacturing that spurred the expansion of the manufacturing sector in the 2s with commensurate increases in the number of higher-earning jobs has not transferred to other manufacturing exports. This is partly because of the substantial anti-export bias of trade policies (Sattar and Ahmed 212) 3 but, also because of constraints in infrastructure (World Bank 212; Mckinsey 211). Additionally the tax policies favor private investment in real estate as against investment in manufacturing, infrastructure and other productive sectors (Ahmed 212). Trade policy also constrains the transformation of agriculture from a primarily peasant economy type sector to a more modern commercial sector. Over 6 percent of agricultural GDP comes from crops, which is dominated by food grain. Export ban on food grain aimed at keeping food prices low for domestic consumption limits the profitability of the crop sector. Diversification to higher value crops, horticulture, livestock and fisheries is adversely affected by inadequacies of rural infrastructure, marketing, technology, quality control and private investment (Government of Bangladesh, 211b). 1 Sadiq Ahmed is Vice Chairman of the Policy Research Institute of Bangladesh. Comments from Professor Nurul Islam and Professor A. R. Khan are gratefully acknowledged. Errors are the sole responsibility of the author. 2 The term disguised unemployment was first coined by Joan Robinson (1936) in the context of the Great Depression. Its use to describe the employment challenge in developing countries came later. See Rosenstein- Rodan (1956), Myrdal (1968), Lewis (1954), Nurkse (1953) and Streeten (1969) for a rich discussion of the meaning and relevance of the concept for developing countries. 3 The special privileges accorded to the garments sector including duty free access to imports through bonded warehouses are not available to most other exports. 1

The quality and productivity of labor force is another challenge. The education level of the work force is deplorably low (Bangladesh Bureau of Statistics 212). A whopping 4% of the workforce has no education; some 23% have undertaken only up to primary education; and less than 4% have tertiary education. Converting this workforce into a quality workforce that is needed to support the transformation of a rural-based peasant economy into a modern urban-based manufacturing and formal services economy is not an easy challenge. This will require huge investment in human capital as well as reforms of related policies, programs and institutions. Unlike in India, the gains from total factor productivity have been limited in Bangladesh (Ahmed 27; World Bank 27). One important reason for this is the very low level of direct foreign investment. It is now well recognized that direct foreign investment can be a major source of productivity gains through technology transfer. The experience from the garments sector in Bangladesh is an example of this. But this positive experience has not been replicated in other manufacturing activities. How to attract more foreign capital in the manufacturing sector is a major policy challenge. It is clear from this brief review that although the prospects for Bangladesh to cash on its ongoing demographic transition are good, this will not happen automatically. It will require major policy and institutional reforms on both the supply side and the demand side of the labor market. The objective of this paper is to look more systematically into these policy and institutional issues in the specific context of labour markets, employment, productivity and real wages outturn. The paper will review the historical context and provide an analysis of the emerging challenges and constraints. It will also suggest how these constraints might be addressed. B. What Has Been Happening to Employment, Productivity, and Real Wages The Supply of Labor The trend in labor force is shown in figure 1. On average the labor force has grown by 2.9% per year between 1974 and 21, as compared with a population growth rate of 2.1%. The faster expansion of the labor force is caused by two factors. First there is a rising share of population in the working age group of 15 plus (figure 2). And second, overall labor force participation has been increasing owing to the growing participation of female workforce (figure 3). Because of the growing female participation, the share of female labor in total labor force is rising (figure 4), expanding from a low base of only 12% in 1989 to 3% in 21. Figure 1: Trend in Labor Force, 1974-21 (millions) Labor Force size in milions Source: Bangladesh Labor Force Survey Various Years 4 Figure 2: Percent of Working Age Population, 1974-21 % Working age population 6 5 4 3 2 1 4 35 3 25 2 15 1 5 Source: Bangladesh Labor Force Survey Various Years 4 The definition of labor force is population age 15 and above and follows the usual definition that excludes unpaid family labor. Historical data is adjusted to confirm to this definition. Annex 1 provides details of labor force definition and sectoral distribution issues. 2

Yet, the level of female participation (only 36% in 21) remains low by international standards. As policy efforts are made to increase this participation, total labor force and the share of female labor force in total labor force will both rise. This is a potentially favorable factor for future growth prospects in Bangladesh. Figure 3: Labor Force Participation Rates, 1974-21 Participation rate (%) 1 9 8 7 6 5 4 3 2 1 1974 1984 1986 1996 2 23 26 Source: Bangladesh Labor Force Survey Various Years 21 Figure 4: Percent Female Workforce National Male Female of the total labor force is still located in rural areas is an indication of the magnitude of the transformation challenge facing Bangladesh. This spatial distribution is also correlated with the nature of labor markets and employment opportunities. Rather surprisingly some 88 percent of the labor force is still engaged in informal activities 5 (figure 5). Much of these activities, largely in agriculture, are located in rural areas. Yet a huge portion of the urban labor force is also engaged in informal activities, primarily in the service sector. 1 9 8 7 6 5 4 3 2 1 Percent Figure 5: Formality of Employment formal informal urban rural total Source: Bangladesh Bureau of Statistics (212). Percent of total workforce 35 3 25 2 15 1 5 1989 1991 1996 2 23 26 21 Some additional insights on the informal nature of the labor market can be obtained from a study done by Maligalig, Cuevas and Rosario (29) using 25/6 Labor Force Survey data. The sectoral distribution of formality of employment is shown in figure 6. Not surprisingly, 94% of agricultural workers belong to informal category. However, surprisingly, only 3% of manufacturing employment is formal and 7% is informal. In services, 21% is formal and 79% is informal. Within services, activities in areas such as financial intermediation, public administration and defense, education, and health and social work are mostly formal (71%) owing to the dominance of public service employment. Other services like construction, trade, transport, and other services are mostly informal (83%). Source: Bangladesh Labor Force Surveys Owing to rapid urbanization and labor migration, the share of urban labor force has grown substantially, rising from 12% in 1984 to around 23% in 21. Even so, the fact that some 77% With progress in literacy and education, some improvement has happened in the quality of the labor force (figure 7). For 5 Formality of employment is defined with regard to whether the activity is registered with a government institution and is subject to tax laws as well as compliance with government regulations regarding employment conditions and benefits. 3

example the percent of work force with no formal education has fallen from 47% in 1996 to 4% in 21. Similarly, the percent of work force with secondary or higher education levels has increased from 3% to 37% over the same period. Yet, the facts that 4% of the workforce has no education and 23% have only primary level education in 21 are indicative of a very low skilled workforce. This has major implications for productivity and income opportunities for the workforce. 1 Percent of labor force Figure 6: Sectoral Distribution of Employment Formality 25/6 8 6 4 2 Source: Maligalig, Cuevas and Rosario (29) Figure 7: Education Attainment of Labor Force 5 4 3 2 1 Formal Informal Agriculture Manufacturing Services 1996 21 no education primary secondary tertiary Source: Bangladesh Labor Surveys Various Years Demand for labor: The trend of employment is shown in figure 8. Overall employment has grown at 2.8% per year during 1974-21 as compared with 2.9% per year growth in the labor force 6. So, on aggregate, the growth of total employment has broadly kept pace with the growth in the labor force. As such the percent of open unemployment is low, although showing an upward trend owing to the slightly slower pace of growth of employment (figure 9). Millions of workers 6 5 4 3 2 1 Figure 8: Trend in Employment, 1974-21 Source: Bangladesh Labor Force Survey Various Years Looking at these aggregate numbers one might be tempted to conclude that Bangladesh has achieved a fair amount of success in attacking the employment challenge. At less than 5%, the rate of open unemployment appears manageable by international standards. Yet, this grossly understates the true employment challenge. As we will show later, average labor productivity is low economywide and much of the labor force is engaged in low-productivity agriculture. Open unemployment is not a true measure of the employment challenge. It is a problem of creating good jobs, defined as jobs that provide employment in high productivity, high earnings activities. Some additional insights on the employment challenge can be gauged by looking at the underemployment figures from the 21 Labor Force Survey. The rates of underemployment, defined as persons who worked for less than 35 hours per week, are indicated in figure 1. The rate of total underemployment is 2.3%, which is quite large. Underemployment rate is much higher for rural areas (22.7%) as compared with urban (12.4 %). In terms of gender, the underemployment rate is much higher for female (34.2%) as compared with male (14.4 %). 6 These are trend growth rates that are estimated by regressing the log of the respective variables over time. 4

Percent 5 4.5 4 3.5 3 2.5 2 1.5 1.5 Figure 9: Open Unemployment Rate was very rapid until 1996. Since then the employment share has basically fluctuated around 48%. Since much of the gain in employment has happened in the services sector, its share increased from 17% in 1974 to 4% in 21. Reflecting the pattern observed in agriculture but in a reverse order, the share of employment in services grew rapidly until 1996 and then stabilized around 4%. The employment share of manufacturing has increased, but rather slowly owing to the low base, growing from 6% in 1974 to 12.4% in 21. Source: Bangladesh Labor Force Survey Various Years Percent Figure 1: Underemployment Rates 21 4 35 3 25 2 15 1 5 Source: 21 Labor Force Survey Male Female Rural Urban Total The composition of employment by broad sectors is another indicator of the nature of demand for labor and its changing pattern. The distribution of employment by the three major sectors of agriculture, manufacturing and services and employment growth rates are shown in Tables 1 and 2. The main results are as follows: Period Agriculture 7 Manufacturing 8 Table 1: Sectoral Distribution of Employment (Percent) and Total Services others 9 1974 77. 6. 17. 1. 1981 61. 8.7 3.3 1. 1984 58.7 9. 32.3 1. 1985 57.8 9.4 32.8 1. 1986 57.2 1.1 32.7 1. 1989 56.6 1.1 33.3 1. 1991 53. 1. 37. 1. 1996 48.8 1.1 41.1 1. 2 5.8 9.5 39.7 1. 21 51. 9.7 39.3 1. 23 51.7 9.7 38.6 1. 26 48.1 11.1 4.8 1. 21 47.5 12.4 4.1 1. Source: National Census various years and Labor Force Surveys various years An interesting question is how do these changes in the employment pattern correlate with the developments on the output side? The changing structure of production in Employment grew fastest in the services sector, at a brisk pace of 5% per year, followed by manufacturing (4.6 %). In agriculture employment grew by 1.5% per year As a result of the slow pace of employment growth in agriculture in relation to total employment, its share of employment has declined substantially from a high of 77% in 1974 to 48% in 21. The rate of decline 7 1974, 1981 and 1991 data are from the corresponding Census data published by the Bangladesh Bureau of Statistics.. Other data are from the Labor Force Surveys. 8 For 1984-1991, data source is Osmani 25. Other data are from Labor Force Surveys. 9 Includes services, construction, water and electricity and mining. Employment in water, electricity and mining is negligible. For the remainder of the paper we will refer to this simply as services sector. 5

Table 2: Structural Change in Output and Employment 1974-21 1 Bangladesh in the sense of more jobs in higher productivity and higher earning enterprises? Agriculture Manufacturing Services & others Total GDP Share in valued added 1974-21 48.3 19.6 11.8 17.3 39.9 63.1 1. 1. Share in employment 77 48 6 12 17 4 1 1 1974-21 Average growth in value added, 1974-21 Average growth in employment, 1974-21 Employment elasticity, 1974-21 2.7 1.5.56 5. 4.6.92 5.3 5..94 4.5 2.9.64 Labor Productivity Figure 11 shows the trend in average labor productivity in Bangladesh while figure 12 compares average labor productivity with a number of comparators. Average labor productivity remained roughly constant between 1974 and 1991 12. It then started to grow and gained momentum during 2. Despite this progress, average labor productivity remains low by international standards (figure 12). Thus average labor productivity in Bangladesh in 28 was only 3% of the productivity in Sri Lanka, 38% of Indonesia, 4% of China, and 5% of India. The challenge for productivity improvement is obvious. Source: Bangladesh Bureau of Statistics the economy is illustrated in Table 2 11. Between 1974 and grew at a brisk pace of 5.% and 5.3% respectively per year, which was faster than the growth of total GDP (4.5%). Agriculture grew very slowly (2.7% per year). The faster growth in manufacturing and services provided the basis for the more rapid expansion of employment in these sectors. The services sector including construction has absorbed labor fairly rapidly, yielding an employment elasticity of.94. The corresponding employment elasticities in agriculture and manufacturing are.56 and.92 respectively. The rising output and employment shares of manufacturing and services suggest that there has been some progress towards a positive transformation of the Bangladeshi economy from a peasant economy to a more modern manufacturing and services economy. How significant is this transformation? Has this led to the creation of good jobs in Figure 11: Trend in Average Labor Productivity, 'Takas, 1995-96 Prices 8 7 6 5 4 3 2 1 Source: Bangladesh Bureau of Statistics 1 National accounts data are calculated on a fiscal year basis running from July to June. For ease of exposition, we show FY1973/74 as 1974 and so on. 11 Measuring structural change and sectoral growth are fraught with risks owing to the frequent and radical changes in national accounts. For this report we use the most recently revised national account (Bangladesh Bureau of Statistics 2) that goes back to 199 only. The reconciliation with earlier years is virtually impossible. So, we preserve the old structure for 1974 contained in Bangladesh Bureau of Statistics 1993 and calculate long term sectoral growth by taking a weighted average of growth during 1974-91 from the old series and growth during 1991-21 based on the new series. See Annex 2 for further details. 12 As noted the BBS adopted a new national accounts series from 1991 onwards using 1995-96 prices. To make the 1974-9 productivity data comparable, we backcast the data to 1974 using the revised structure and growth rates from the old series. 6

Labor Productivity 199 PPP $ Figure 12: GDP Per Person Employed in 28 (constant 199 PPP $) 16 14 12 1 8 6 4 2 Source: World Development Indicators, World Bank Website Before venturing into sectoral productivity estimates, it is important to make a caveat about the sectoral employment data and the underlying productivity estimates. The interpretation of sectoral productivity estimates is clouded by the varying quality of data, especially employment data. In particular, the employment data before 1991 is confusing owing to the definition of labor force that used the expansive definition of age 1+ in defining labor force and employment. Subsequently, the Bangladesh Bureau of Statistics changed the definition to the standard 15+ age group, but it did not adjust the old data. For this paper, we have carefully reviewed the literature on the Bangladesh labor force and employment and all available data sources. The definition of labor force and employment on aggregate and sectoral level has been made consistent to conform to the standard age group of 15+. While we believe the adjustments to the aggregate data are solid, we are less confident about the robustness of the sectoral employment data for 1974-1991 (see Annex 1 for details). The sectoral productivity trends tell the story of whether a meaningful progress towards the transition from a peasant economy towards a modern economy is happening or not (figure 13). The main results are as follows: positive for agriculture, but declined for manufacturing and services. It started growing for all sectors after that. Productivity in agriculture has grown at an average rate of 1.% per year during 1981-21. Productivity has grown faster on average for manufacturing and services, at an average rate of 1.6% per year. For the economy as a whole average productivity grew by 2% per year during 1981-21. This was partly due to growth of the individual sectors but also because of the changing composition of the economy whereby the low productivity agriculture lost ground and higher productivity manufacturing and services gained momentum. Figure 13: Trend in Labor Productivity, 1981-21 (1995-96 taka) Labor Productivity 12 1 8 6 4 2 Total Manufacturing Source: Bangladesh Bureau of Statistics Agriculture Services In some ways these results are consistent with the results found in other developing countries. The continued low productivity and the slow pace of productivity improvement in agriculture are symptomatic of a peasant economy that is transforming slowly. But there is one important divergence. This concerns the higher average productivity in services as against manufacturing. Typically, in developing countries the average productivity in manufacturing tends to be higher than services. Average labor productivity is highest in services, followed by manufacturing. Average productivity is low in agriculture. Average labor productivity grew very slowly between 1981 and 199. Productivity growth was slightly There are a number of ways this seemingly divergent finding can be reconciled with the conventional wisdom. One possibility is the quality and reliability of the labor force data, especially for the earlier periods as noted above. For example, the observed gap in the productivity between services and 7

manufacturing in the 198s could be the result of underestimation of employment in services. A second explanation is the aggregation of formal and informal services into one category. The differences in productivity between formal and informal services can be large and the high average productivity of formal services in areas such as banking, professional services, government and defense might be raising the average productivity of services in the base year which then carries through. A third possibility is the strong role of informal manufacturing, the micro and small manufacturing enterprises, in total manufacturing in the initial years that is dragging down its average labor productivity. A combination of all these factors is possibly the most plausible explanation. Real Wage Index 1981=1) 25 2 15 1 5 Figure 14: Trend in Real Wages 1981-21 General Manufacturing Services Agriculture Real Wages Outturn The productivity result tells one part of the story on economic transformation. The other part comes from the developments in real wages. The trend in real wages in agriculture, manufacturing and services is shown in Figure 14 13. The movements in real wages suggest the following: On average real wages have grown very slowly economy-wide and in the sectors. Thus, economywide, the real wage rate has grown by 1.7 % per annum from 198/81 to 29/1. Manufacturing sector has registered the fastest average growth in real wages, increasing by 2.6% per year, followed by agriculture (1.4 %). Surprisingly, the average growth in real wages in services has been the slowest (.8% per year). The slow growth of real wages in services has to be qualified by the fact that this really represents real wages in construction that may not be representative of wages in other services Source: Bangladesh Bureau of Statistics Wage-Productivity Relationship An important question is how does the growth in real wages compare with productivity? Figure 15 shows this relationship for the periods 198/81 to 29/1. The correlation between growth in average labor productivity and growth in real wages is positive economy-wide and for all sectors, which is reassuring. On average, productivity has grown at a somewhat faster pace (2 percent per year than real wages (1.7 percent per year). This relationship has three important implications. First that the path to faster pace of poverty reduction is to increase employment and labor productivity. Second, the labor market appears to work flexibly with wages moving in line with productivity rather than through non-market interventions. And third, the faster pace of productivity increase relative to real wages indicates the basic competitiveness of the Bangladesh economy with favorable long-term implications. We will revisit the labor market issues more fully below. 13 Nominal wage data is published by the Bangladesh Bureau of Statistics. There is no data for services sector; so the wage data for construction is used as a proxy. Real wages are estimated by deflating nominal wages by the consumer price index (CPI). Real agriculture wages are possibly better estimated by using rural CPI. However data for rural CPI is available only from 1995/96. Rural and general CPIs show a remarkable degree of correlation, almost 1, suggesting the two series are perfectly correlated. As well, the growth of real wages in agriculture is invariant with respect to the choice of price deflator (between general CPI and the rural CPI) for the period 1995/96 to 29/21. 8

Percent Per Year Figure 15: Wage-Productivity Relationship 198-21 3 2.5 2 1.5 1.5 Productivity Source: Bangladesh Bureau of Statistics Real Wages Total Agriculture Manufacturing Services The story gets more interesting when one looks at the sectoral results. Real wage in services has grown slower than productivity, but real wages in agriculture and manufacturing have grown faster than productivity, especially in manufacturing. Below we review these results more fully with disaggregated data from the various LFSs, particularly the 21 LFS. Productivity-Wage Relationship in a More Disaggregated Framework The improvement in the quality of data in the successive rounds of LFS is both an advantage as well as a problem. It is an advantage because more detailed data are available that improves the quality and depth of the analysis at a point in time. The main disadvantage is that it makes over time comparisons quite difficult, especially when definitional changes are made. The story gets more interesting when one looks at the sectoral results. Real wage in services has grown slower than productivity, but real wages in agriculture and manufacturing have grown faster than productivity, especially in manufacturing. Below we review these results more fully with disaggregated data from the various LFSs, particularly the 21 LFS. Let us first look at the disaggregation of productivity in the services sector. Comparable data are available from 2 to 21 that makes it possible to calculate disaggregated productivity numbers. The services data are broken down into five main categories: construction, trade, transport, professional services and other services. The trend in average productivity for these services sub-group is shown in figure 16, while figure 17 compares the growth in productivity. Average productivity Figure 16: Average Productivity in Services 2-21 (1995/96 taka) 2 15 1 5 2 23 26 21 Construction Trade Transport Professional Services Other Services Overall Services Source: Bangladesh Bureau of Statistics As expected there are large productivity differentials within the services category. Highest productivity is found in professional services, followed by construction, transport, other services and trade. The gap in productivity differential between professional services and trade is particularly large; the average productivity in professional services is almost three-fold higher 14. Percent Per Year Figure 17: Growth in Average Labor Productivity 2-21 5 4 3 2 1-1 -2 Source: Bangladesh Bureau of Statistics 14 The slow growth of manufacturing productivity in 2-21 is the result of the impact of global downturn on manufacturing production and exports in 21. The manufacturing sector recovered in 211 and if the end year is changed from 21 to 211, productivity growth jumps from 1.4 % to 2.5%. 9

How do these productivity results compare with wages? Comparable wage data is not available. However, some perspectives can be obtained from looking at wage distribution of day laborers in the various occupations based on data from the 21 LFS (Tables 3 and 4). One important caveat is that employment in non-agricultural activities is likely to be for a longer duration than in agriculture. As such overall earnings on a yearly basis in non-agricultural activities will likely be even higher than suggested by the wage differential between agriculture and non-agriculture. Weekly nominal wages (taka) Table 3: Wage Profile of Day laborer, 21 < =5 51-1 1-15 151-2 21+ Total Agriculture 13.8 42.2 26.1 15.3 2.6 1. Manufacturing 9.4 35.7 23.6 28.2 3.2 1. Construction 5.7 36.9 38.5 13.7 5.2 1. Trade 6.8 29.5 32.2 25.7 5.8 1. Transport 4.6 34.9 38.5 17.3 4.7 1. Professional 7.1 22.4 2.9 39.6 1.1 1. services Other services 7. 3.6 34.6 22.6 5.1 1. Source: Bangladesh Bureau of Statistics (212) Table 4: Average Wages of Day laborer, 21 Weekly Wages (taka) Nominal Ratio Agriculture 997 1. Manufacturing 1142 1.145 Construction 1116 1.119 Trade 127 1.211 Transport 1151 1.154 Professional services 1341 1.345 Other services 1178 1.182 Source: LFS 21 Against the backdrop of the above, the results are as follows: Consistent with productivity data, average wages in manufacturing and services are higher than in agriculture. Again, consistent with productivity data, on average wages in services are higher than in manufacturing. Within the services category, wages are higher in professional services as compared to other services. Wages in transport and other services are fairly similar and consistent with productivity data. But wages in trade are higher than what should be expected from productivity data. The reported wages in trade are likely biased upwards by the inclusion of Percent Per Year motor repair workers in this category. Such workers are fairly skilled on average than other workers in trade and are likely to command much higher wages. The most glaring inconsistency is between wages in construction and average productivity. The average productivity in construction is much higher than all other services except professional services, yet wages are lower than in trade, transport and other services. The most logical explanation seems to be the heavy influx of labor in construction over the past 1 years (figure 18). Employment has expanded by an amazing 9% per year, much faster than employment elsewhere as well as faster than the growth of value added in construction. This suggests that the construction sector has served as the employer of last resort for the huge influx of the migrant population that has tended to pull down the growth of wages as compared to other services. Except for construction, all categories of services show large improvements in productivity. These increases exceed productivity improvements in agriculture and manufacturing as well as the average for the economy. While time series wage data are not available for the various service categories, it would be reasonable to expect significant improvements in wages in the various service categories. This has also contributed to the gap in wages between construction and other services. Figure 18: Growth in Services Employment, 2-21 1 9 8 7 6 5 4 3 2 1 Source: Labor Force Survey Various Years 1

Summary of structural transformation It is now time to answer the question we posed at the beginning of this section about the nature and extent of the structural transformation of the Bangladesh economy and the implications for employment. The evidence clearly suggests that a slow but significant transformation of the Bangladesh economy has happened over the 4 years since independence where the sources of production and employment have changed in a noticeable way. There are two dimensions of this transformation. First, owing to much faster growth in services as compared with agriculture, the structure of production has changed significantly away from agriculture to services, which is reflected in a substantial reduction in the value-added share of agriculture in total GDP and a sharp increase in the value-added share of services. Manufacturing sector has also played an important role, but in a more limited way as compared with the experiences in developed countries. While manufacturing growth rate has increased faster than in agriculture and services, this is from a low base. As a result, the GDP share of manufacturing is still below 2%. Second, production and employment options have increased much faster in the urban areas as compared to rural areas contributing to the growing urbanization of Bangladesh. As a result of these structural changes, jobs have been created at a faster pace in services and manufacturing, reflected in increases in the employment shares of services and manufacturing. On the supply side of the labor market, some improvement has also happened in the quality of labor as indicated by the increase in the share of educated labor in total labor force. The other supply side structural improvement is the increasing participation of female labor in total labor force. The combined impacts of these changes in the demand and supply sides of the labor market are increases in average labor productivity and real wages. By and large real wage increases economy wide has tracked productivity improvements fairly closely. These results also carry over to the sectoral components. While these findings are indicative of a positive transformation of the Bangladeshi economy, the challenges moving forward are substantial. Average labor productivity is low by international standards. Productivity in agriculture is particularly low and it continues to employ almost half the labor force. Moreover, some 4% of the labor force has zero education and an additional 23% has only primary level education. Converting this labor force to one that has minimum skills to work in modern manufacturing and formal services is a tall order. C. The Nature of Labor Markets in Bangladesh The observed trends in labor force growth, employment, productivity and real wages have important implications for the nature of labor markets in Bangladesh. The most important indicator of the nature of labor markets is the overwhelming share of informal employment. A whopping 88% share of informal employment in 21 is indicative of the flexible nature of the labor market. Government interventions and collective bargaining if at all relevant play a very limited role in determining the labor market outcomes. When one looks at the spatial distribution of formal and informal employment (Figure 5), while 91% of rural workers are engaged in informal activities, it is remarkable that some 75% of the urban workforce is also engaged in informal activities. The other characteristic of the labour market is the dominance of employment in agriculture. Some 48% of the workforce is still engaged in agriculture even though the value-added share of agriculture has fallen to 2%. Despite improvements in productivity, the average productivity in agriculture remains very low and the gap with manufacturing productivity has grown further. Low productivity of agriculture combined with continued dominance of employment in agriculture may be suggestive of the Lewis (1954) and Fei-Ranis (1964) type of labor market in agriculture, characterized by surplus labor. A key empirical issue, however, is whether the wage rate in agriculture is above marginal productivity, as hypothesized in the Lewis-Fei-Ranis labor surplus model, or whether it is also determined by marginal productivity. In the later situation, the dominance of low productivity agriculture employment can be explained by 11

the skill constraints, information cost, and the costs of migration 15. A third characteristic that bears significance in terms of the description of the labor market is migration between rural and urban areas. As a result of rapid migration, the share of urban workforce has grown substantially accounting for 23% of total employment. The fact that 75% of urban employment is informal in nature has significant implications for the nature of labor market and may appear to fit in with the Harris-Todaro (197) and Fields (1975) type migration story and labor market evolution. As in agriculture, a key contentious issue is how wages are determined in the informal sector (Rosenzweig 28; Fields 211) Pulling these ideas together, the Bangladesh labor market can broadly be defined as falling into two categories: informal and formal. The bulk of the labor force (88%) transacts in labor market that is informal in nature indicating the ease of entry and exit. Government regulations are virtually absent here and collective bargaining does not happen. The other market, the formal labor market employs a small segment of the total labor force, presently accounting for only 12% of the workforce. It is formal in the sense that this workforce falls in the purview of activities that need to comply with government regulations and registration requirements. This sector may also be subject to collective bargaining and other forms of labor market regulations that prevail in Bangladesh. Within the informal category, there are two types of labor markets: agriculture and non-agricultural activities. Agriculture is located in rural areas. The non-agricultural activities in turn can be in rural areas or in urban areas. Much of the non-agricultural activities in both rural and urban areas relate to services (construction, transport, housing, trade, and personal services). There is limited employment in rural manufacturing; micro and small manufacturing enterprises belong to this employment category. In the formal sector, the dominant activity is medium and large scale manufacturing. The other activity is formal services (banking, public service, private professional services). Entry into formal employment could be restricted by a range of factors including education, skills, financing, information, and labor laws and unions (in the public sector). 15 A healthy debate on this is contained in Ranis (24), Rosenzweig (28) and Fields (29, 22, 211) As we saw earlier, wages tend to be higher than the national average for employment in services and manufacturing. Within the services category, wages are higher in professional services, as compared with informal services. While the wage data in Tables 3 and 4 do not distinguish between urban and rural, evidence from elsewhere suggest that urban real wages in these activities tend to be higher than related activities in rural areas, mostly reflecting productivity and skills differential 16. Entry barriers in terms of minimum wages and employment regulations are not binding in the private sector, although in the public sector employment is influenced by political interventions. An important question that has a bearing on wage determination mechanism in various markets is whether the data of Tables 3 and 4 have any significance for the market segmentation argument. Although more focused research is needed to be conclusive here, some broad observations can nevertheless be made. The first question is what explains the wage differential between agriculture and other activities, especially after allowance is made for the likely longer duration of employment in non-agriculture activities on a yearly basis? A related question is why wages in trade, transport and other services (informal sector) are higher than in manufacturing (formal sector)? A final question is what explains the relative differences in wages in various informal activities? Let us first consider the agriculture sector wages. It clearly sets a floor to wages economy wide. Wages in other activities must be at least higher than in agriculture to allow for transaction and migration costs (where relevant). But wage differential within urban services and between urban services and manufacturing are another matter. If there were entry barriers in the form of minimum wages and unionization in the formal sector then wages ought to be much higher in manufacturing and professional services (restricted sector) as compared with wages in informal services (no entry restrictions) 17. But the facts suggest the opposite: wages are higher in informal activities like trade, transport and other services as compared with manufacturing. Similarly, wages across informal services ought to be equalized since there is no segmentation within and across these activities. But in reality they are not 16 World Bank (27); Bangladesh Bureau of Statistics (212). 17 The only exception to this is government services where political influences and union pressures are quite dominant. According to LFS 21, these services account for less than 5% of total employment (2.5% in government, 1.6% in autonomous bodies and.7% in local governments). 12

equalized. It therefore seems plausible to argue that market segmentation is not a good explanation for the observed wage differential across where activities. On the other hand attributes like job preference, skill differential, education differential, migration and other transaction costs can better explain the observed wage differentials in both within and between the formal and informal sectors. This issue can be further examined in the context of the information provided by the Maligalig, Cuevas and Rosario (29) study on wage differentials between formal and informal sector employments. The main findings of the study are as follows: the wage differentials between formal and informal employment are large and statistically significant for all activities across the economy; the widest wage gaps are in construction; electricity, gas and water; health and social work; and mining and quarrying; the smallest wage gaps are in private household, fishing and education; but any formal employment does not necessarily provide better wages than any informal employment; thus mean weekly wages for informal employment in financial intermediation, education and public administration are higher than wages in formal employment in fishing, agriculture and private households. workers in informal employment receive the highest mean weekly wages in financial intermediation; education and public administration; informal workers earned the least in agriculture, fishing and construction; for those with formal employment, the top three mean weekly wages are found in financial intermediation; electricity, gas and water; and health and social sectors; the bottom three sectors with the lowest mean weekly wages in formal employment are private households with employed persons; fishing; and agriculture. These findings are remarkably consistent with the above conclusion regarding the flexible nature of the labor markets in Bangladesh. In particular, two results stand out: First, within the formal sector employment, wages are highest in activities that require substantial skills and knowledge (financial sector, health; and construction) and lowest where skills constraint is not likely to be an issue (agriculture, fishing, private household employment). The pattern is similar in informal employment: wages are highest in those informal activities where most skills are required (financial intermediation, education, health, public administration) and lowest where minimum education and skills are necessary (agriculture, fishing, construction, private household). Second, wages in informal employment in activities relating to financial intermediation, public administration, education, health substantially exceed the wages in formal employment in agriculture, fishing, and private household employment. So, the average wage gap between formal and informal activities cannot be explained by market segmentation and most likely reflect skill differentials, labor demand, and job preferences. Against the backdrop of the above, let us now examine the three main categories of labor markets in Bangladesh: agriculture, informal sector services and formal sector. We will specially comment on the likely wage determination mechanism in each of these markets and the implications for wage movements owing to the interactions of these markets. A key issue that will be discussed is how changes in one labor market affect productivity, employment and wages in other markets. Agricultural Labor Market In the traditional dual sector models all available labor in agriculture, mostly family type, are engaged in production with the wage rate determined institutionally, based on subsistence level, income sharing or other considerations and related to average productivity. Marginal productivity is low and lower than average productivity, but not necessarily zero 18. As such, the wage rate is higher than marginal productivity. Average productivity and the real wage rate remains constant for a fairly long period of time. However, as labor continues to move away from agriculture and demand for labor grows in agriculture owing to technology, 18 In the original Lewis formulation, marginal productivity was assumed to be zero. This came under substantial criticism. But as Ranis (24) points out this assumption is not necessary for the dual sector model. The critical assumption is that the wage rate is higher than marginal productivity due to institutional consideration. 13