report no. 13799-bd bangladesh labor market...

76
ReportNo. 13799-BD Bangladesh LaborMarket Policies for Higher Employment April 22, 1996 South Asia Country Department I Country Operations Division Document of the Worid Bank Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized

Upload: nguyenbao

Post on 19-Mar-2018

220 views

Category:

Documents


1 download

TRANSCRIPT

Report No. 13799-BD

BangladeshLabor Market Policiesfor Higher EmploymentApril 22, 1996

South Asia Country Department ICountry Operations Division

Document of the Worid Bank

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

CURRENCY EQUIVALENTSTaka (Tk)

US$1.00 = Tk. 41.50 (April 8, 1996)

FISCAL YEAR (FY)

July 1 - June 30

BANGLADESH

Table of Contents

Page No.

EXECUTIVE SUMMARY ............................................................... i

A. INTRODUCTION ............................................................... 1

B. INVESTMENT, GROWTH AND EMPLOYMENT .......................................................... 2The Labor Force and the Problem of Under-Employment .............................................. 2Investment and Growth are Key to Job Creation .......................................................... 3

C. INTERNATIONAL COMPETITIVENESS .............................................................. 5Defining Unit Labor Costs .............................................................. 5Estimating The Opportunity Cost and Exchange Rate Gaps ............................................ 7Labor Competitiveness in Garments .............................................................. 8Labor Competitiveness in the Leather Sector ............................................................ 10Labor Competitiveness in Textiles .............................................................. 12Labor Competitiveness In Public Enterprises ............................................................ 14Policy Implications .............................................................. 15

D. TOWARD EFFICIENT RESTRUCTURING REGULATIONS .......................................... 16Bangladesh's Restructuring Regulations .............................................................. 16Public Sector Practice .............................................................. 18Policy Implications .............................................................. 19

E. POLICIES AFFECTING WAGES .............................................................. 21A Description of Wage Differentials ............................................................... 21Minimum wages .............................................................. 24Public Sector Wage Policies .............................................................. 25Policy Implications .............................................................. 28

F. INDUSTRIAL RELATIONS .............................................................. 29A Fragmented Labor Movement .............................................................. 29Government-Labor Relations ..................... ......................................... 30Policy Implications .............................................................. 31

G. SPECIAL LABOR MARKET PROBLEMS: GENDER ISSUES AND CHILDLABOR .............................................................. 34Gender Issues .............................................................. 34

Discrimination Against Working Women . ..................................................... 34International Experience .............................................................. 35Policy Implications .............................................................. 37

Pagre no,

Child Labor ............................................................ 38The Problem in Bangladesh ............................................................ 38International Experience ............................................................ 38Policy Implications ............................................................ 38

H. LABOR POLICIES AND INCOME DISTRIBUTION ..................................................... 39Minimum Wages ............................................................ 39Severance Pay ............................................................ 40Food-For-Work ............................................................ 41Policy Implications ............................................................ 42

Lit of Tables

Paae no.

Table 1: Sectoral Distribution of Employment in 1989 ........................................................... 3Table 2: Under-employment Projections ..................... ....................................... 5Table 3: Opportunity Cost of Labor ............................................................ 7Table 4: Exchange Rate Comparisons ................. 8........................................... Table 5: Unit Labor Costs for T-shirts ............................................................ 9Table 6: Decomposing the Difference in Unit Labor Costs for T-shirts ....................................... 9Table 7: Bangladesh and Vietnam: Unit Labor Costs in Garmnents ................ .......................... 10Table 8: Unit Labor Costs for Finished Leather ............................................................ 11Table 9: Decomposition of the Difference in Unit Labor Costs for Leather ........... ..................... 11Table 10: Unit Labor Costs for Cotton Yarns ............................................................ 12Table 11: Decomposition of the Difference in Unit Labor Costs for Cotton Yarn ........ ................ 13Table 12: Public Sector Competitiveness ............................................................ 14Table 13: The Structure of Wages ............................ ................................ 22Table 14: Average Labor Costs and Value of Output/Worker ................................................ 26Table 15: International Comparisons of Activity Rates ......................................................... 34

This report was written by Hafez Ghanem and Syed Nizamuddin. It relied on background papersprepared by Jean Paul Azam, John Cameron, Giovanna Prennuchi, Masihur Rahman, Derek Robinson,Faham Rashid, Muneto Ozaki, Rushidan Islan and Shanta Devarajan, Hafez Ghanem and KarenThierfelder. Dipak Mazumdar supervised the analysis of unit labor costs. The report was discussed withthe Government of Bangladesh in July 1995.

We acknowledge with gratitude the ILO's contribution to this report through the preparation of threebackground papers -- on the analysis of labor force utilization, labor market issues and industrial relations-- and through its technical advice on the content of the report. The ILO does not share all the analyses,conclusions and recommendations set forth in this report.

EXECUTIVE SUMMARY

1. Labor market outcomes are unsatisfactory in Bangladesh. Under-employment isrampant and is equivalent to having more than one-fourth of the labor force unemployed. The labormarket is highly segmented and is full of inequities. The wage of an unskilled public sector worker is68 percent higher than that of a similarly unskilled agricultural worker in the relatively affluent Dhakaarea. The real wage of a rural worker in Rangpur fell by 48 percent between FY86 and FY93. Duringthe same period real wages of unskilled public sector workers rose by 10 percent. Women face severelabor market discrimination. More than 80 percent of women in the labor force are unpaid familyworkers as compared to 20 percent of men. When they do get paid, their wage is just one-third to one-half that of a man. Child labor is a serious problem which threatens children's education and futureprospects. More than 20 percent of children between the ages of 5 and 14 are at work. Labor marketsare not encouraging investment and growth. The expansion of some key industrial sectors is hamperedby low labor competitiveness. Unit labor costs in Bangladesh's cotton textiles sector may be as muchas 70 percent higher than in India and 110 percent higher than in Pakistan. Labor unrest and aperception of poor industrial relations are discouraging private investment, especially foreign directinvestment.

2. How can those outcomes be changed? This report analyses Bangladesh's labor marketswith the objective of deriving some practical policy conclusions to help improve the way they function.Its broad conclusions are summarized in three messages:

(a) Higher investment and growth are key to better outcomes for labor.(b) The government of Bangladesh needs to reduce its role in the labor market.(c) Labor policies should focus on improving conditions for rural and informal workers.

The Importance of Investment, Human Development and Growth

3. Labor market policies and poverty alleviation efforts are vitally linked. For poorhouseholds opportunities for productive employment are the key determinant of welfare and the onlyway for families in Bangladesh to achieve an acceptable quality of life. Macroeconomic simulationsindicate that if investment and growth remain at their present levels, under-employment could rise andbecome the equivalent of more than 30 percent open unemployment by year 2000. This would clearlyimply an increase in poverty and human misery. Raising investment to around 20 percent of GDP andachieving at least a 7 percent growth rate is the only way to reduce under-employment--to around theequivalent of 20 percent open unemployment by year 2000. This supports the report's first conclusion:investment and growth are the most powerful way to increase employment, improve labor marketoutcomes and raise household welfare in Bangladesh.

4. Increasing private and public investment in physical capital and infrastructure is key tohigher growth, more jobs and better working conditions. Private investment needs to nearly double, ifBangladesh is to achieve its accelerated growth target. This requires an environment that is conduciveto private sector development, and reforms that include trade liberalization, de-regulation,privatization, financial sector restructuring, and removing labor market distortions. To encouragemore private investment, particularly in export sectors, Bangladesh needs to remove the present highlevels of protection for domestic markets in favor of the more neutral protection provided by theexchange rate. Trade liberalization must be accompanied by a private sector development programcomprising a wide range of deregulatory and promotional actions, legal system reforms, and institution

-ii-

building to support them. These actions need to be complemented by privatization, which is animportant way of mobilizing private resources for investment. Moreover, privatization can helpimprove the performance of enterprises who are making losses while in the public domain. The slowimplementation of the privatization program may shed doubt on the Government's commitment to openup the economy to market forces and rely more heavily on private initiative. Clearly, this is the wrongsignal to provide to private investors. In addition to privatization, the success of any strategy toencourage private investment requires a well functioning financial system that mobilizes savings andallocates them to efficient investments. It is therefore important that financial sector reforms remain ontrack. Finally, increasing labor costs that do not reflect productivity gains and poor industrial relations,can be important constraints on industrial development and the expansion of exports. That is why labormarket reforms can be an important ingredient of any economic policy package that aims atencouraging private initiative.

5. Increasing efficient public investment through measures that would raise the publicsector's savings rate and enhance its project implementation capacity has long been a key Governmentobjective. The Three Year Rolling Investment Program which attempts to match the investmentprogram more closely with social priorities and available resources, provides a basis for progress inthis area. Implementation of this program, which calls mainly for public investment in physicalinfrastructure and human development, would provide a key ingredient of an enabling environment forprivate investment, domestic and foreign.

6. Human development, an important area of public investment, is key to better outcomesfor labor. Increasing the skills and capabilities of workers is necessary for economic success in anincreasingly integrated and competitive global economy. Investing in people can boost the livingstandards of poor households by expanding opportunities, raising productivity, attracting capitalinvestment, and increasing earning power. Better health, nutrition, and education also have value intheir own right, enabling individuals to lead more fulfilling lives. Investments in human developmentare often highly complementary. Adequate nutrition and health increase the ability of children to learn.And, better educated individuals would tend to have better nutrition and hygiene habits, necessary for

good health.

7. The Government's policy of increasing spending in the social sectors needs to becomplemented by measures to enhance the quality of services. This is evident in both the educationand health sectors. The present primary school system is based on a teacher-pupil contact time--2hours a day--which is one of the lowest in the world--international norms are around 4-5 hours. This isfurther aggravated by teacher absenteeism. Survey work indicates that often 50 percent of teachers areabsent, and cases where schools are functioning with only one teacher present for over a hundredchildren at different grade levels are not uncommon. The quality of health services is not better.Occupancy rates at the 31-bed Thana Health Centers are below 50 percent and only 15 percent oftreatment visits for current illnesses are to a government clinic. This situation is a reflection of poorservice quality: inadequate drugs, inadequate nursing care and lack of equipment.

8. The vast majority of Bangladesh's workers are in rural areas and agriculture provideswork for more than 60 percent of the labor force. Hence, Government policies affecting agriculture anddealing with rural poverty can have profound effects on worker welfare. Liberalization policies inagricultural that took place in the late 1980s and early 1990s succeeded in halting the rapid fall in real

-iii-

agricultural wages that was occurring in the 1970s and 1980s. The liberalization of the market forminor irrigation equipment and for fertilizers lead to a huge increase in the land under irrigation and infertilizer use. Naturally, this implied higher productivity and better outcomes for rural labor.

9. The quality of social services and of development efforts in rural areas could beenhanced through increased cooperation between Government and non-government organizations(NGOs). NGOs are better equipped than government to deliver certain types of services--for examplesupporting micro-enterprises in rural areas-- where their small size and flexible organization providethem with an advantage. In other cases competition between public entities and NGOs may beconducive to greater efficiency. Sometimes cooperation between Government and NGOs leads tooptimal results. This has been particularly true in the area of the non-formal delivery of primaryeducation and mobilizing society for the reduction of illiteracy.

10. The Government's policy agenda is vast and ambitious, which reflects Bangladesh'sdifficult development challenge. A discussion of this broad policy agenda is presented in the 1993 and1994 Country Economic Memoranda, Report Nos. 11569-BD and 12724-BD. This report considersonly one of the elements of this agenda--labor market reforms.

Government's Role in the Labor Market

11. The Government's interventions to determine wages and other working conditions inthe formal sector have politicized industrial relations, and have often led to inefficient and inequitableresults. Hence, the report's policy recommendations are based on the view that the Government mustreduce its role in the labor market. The analysis presented in the report has policy implications in theareas of competitiveness, industrial relations and protection for vulnerable groups--women and childworkers.

12. Labor competitiveness in Bangladesh varies considerably from one sector to another.Unprotected sectors with no public enterprise presence, such as garments and leather, tend to have lowunit labor costs. Sectors that have been used to operating behind protective tariffs and import quotas(textiles) or with important explicit and implicit subsidies (jute) and with a significant public enterprisepresence (textiles and jute) are not competitive. This seems to indicate that opening up the economy tomarket forces through trade liberalization and the elimination of explicit (through the budget) andimplicit (through the banking system) subsidies would help enhance labor competitiveness. A similarresult could be obtained by reducing the role of public enterprises through privatization or closure.There is no rationale for the continued existence of public enterprises in sectors producing purelyprivate goods, such as textiles and jute.

13. Labor competitiveness will most probably be enhanced by raising productivity ratherthan lowering wages. Unit labor cost comparisons indicate that wages in Bangladesh are equal to orlower than those prevailing in competitor countries but that productivity is lagging. This implies thatover-staffing should be reduced and then eliminated by accelerating worker retrenchment programs inthe public sector and by regulatory changes to lower the cost of private firm restructuring. Increasedinvestment would help displaced workers be absorbed into expanding efficient sectors. Providingincome security to workers during the transition--e.g., by introducing an urban public works program

-iv-

or creating a social fund--can be defended on social grounds and is probably necessary to securesupport for the reforms.

14. Industrial relations need to be de-politicized, which can only be achieved if theGovernment's role in setting wages and benefits is greatly reduced. Wages in the private sector shouldbe determined through collective bargaining, with little government intervention. Privatizing publicenterprises would also reduce the scope of the Government's wage-setting role and should beaccelerated. Decisions on wages and benefits for enterprises remaining in the public domain must bedecentralized to reduce political interference.

15. The above does not mean that the Government has no role play in the labor market.There is clearly room for government intervention to protect weak and vulnerable workers, andprovide safety nets to help those in transition. Particular attention needs to be paid to the problems ofwomen and child workers. Discrimination against women creates obvious equity problems, sincewomen tend to be poorer than men. It also leads to efficiency losses, because a large share of the laborforce is not well utilized and they and their families are not motivated to invest in education. TheGovernment's current focus on female education could help resolve some of the problems created bydiscrimination. Efforts to increase women's share of public sector jobs by implementing theGovernment's existing affirmative action (or quota) program would also help reduce access problems.Child labor is a symptom of severe poverty and deficiencies in the education system. Bans on childwork have proven to be difficult to enforce. They need to be complemented by other interventions.Poverty alleviation programs would help reduce child labor, as would efforts to improve the quality ofeducation and lower its implicit costs.

Labor Policies and Income Distribution

16. The policies recommended here should improve equity. If implemented, they wouldrender labor markets more competitive and flexible. Hence, they may hurt a few relatively better-offformal sector employees in the short run, as they lose jobs in inefficient firms, or as their real wagesdecline. However, by expanding opportunities for formal employment, reforms would benefit the vastmajority of the poor.

17. Public polices have not helped poor workers who are unaffected by regulations andprotective legislation. Government interventions in the labor market are usually based on a desire toimprove income distribution. However, they often end up having the opposite effect. Laborregulations protect less than 10 percent of the labor force who have regular formal employment, buthurt poor informal and rural workers whose chances for obtaining better jobs are reduced. Forexample, trade unions are pressing the government to introduce a national minimum wage of around Tk900/month. The analysis in this report indicates that if such a policy is adopted and enforced in themodem sector, it could lower rural wages by around 2.5 percent in real terms. This occurs becausethe higher costs would discourage the expansion of formal employment and more people would belooking for informal work in rural areas, depressing the already low rural wages.

18. Not all government interventions lead to such undesirable outcomes. The Food-for-Work program is an example of a Government labor market intervention that has a positive impact onpoverty and income distribution. Instead of trying to legislate higher wages or special privileges for

different groups of workers, this program provides work to under-employed rural laborers who areamong the poorest in society. Government should consider expanding this program, and introducingother types of programs that target the truly needy informal workers.

The Outlook For Bangladesh's Workers

19. These are revolutionary times in the global economy. The embrace of market-baseddevelopment by many developing and formerly centrally planned economies, the opening ofinternational markets, and great advances in the ease with which goods, capital and ideas flow aroundthe world are bringing new opportunities to workers in all countries. Good government policies couldhelp Bangladesh's workers seize those opportunities. Government should aim at encouraging privatesavings and investment and expanding exports. Concomitantly, it should ensure that labor marketinterventions benefit all workers and not just the lucky few who happen to have modem sector jobs.Good choices by the Government, in the domestic and international realm, can lead to advances in theincomes of all workers, can address both their rights and their concerns, and can help bring into thefold those who are unable to keep up or adjust on their own.

A. INTRODUCTION

* A landless agricultural day laborer living in the Bogra area with his wife and threechildren earns Tk. 35 (around $0.90) per workday, but his earning level is very unstable. Everymorning, the laborer goes to the village market hoping to get a job for the day, but he often returnsempty-handed. At harvest time, he can usually count on being employed six days a week but the rest ofthe year he may go for days without work, so that on average, he works only 2.5 days a week, for amean weekly income of Tk. 87.50. His wife is always looking for ways to earn a little extra money forthe household. Sometimes she helps husk rice, weed the fields and chop wood for better-off neighbors,which lets her put a bit more variety on their table. Their eldest daughter contributes by gathering fuel,edible weeds and roots.

* An unskilled construction worker and his garment worker wife, living in Chittagong, arelikewise troubled by low and variable income. The man earns about Tk. 60/day (around $1.50).Although the availability of employment is uncertain, the demand for construction has been rising. Onaverage, the man works four days a week for a weekly income of about Tk. 240 (or Tk. 960/month)-roughly 170 percent more than the rural agricultural worker. But breaking bricks on construction sites inthe heat and humidity is grueling work and the laborer, seriously weakened by a recent bout of hepatitis,worries that he will not be able to continue to produce and earn at the same rate. Fortunately, his wifeworks at one of the garment factories booming near Chittagong's port, where she makes Tk. 900/month(nearly as much as her husband). Her earnings let the family live better and help support the wife'selderly parents, who have remained in their native village. The wife's job also gives her newindependence in relation to her husband. Although the garments factory appears to be doing well, she hasheard that the enterprise's foreign partners are thinking of leaving Bangladesh for Vietnam, where costsmay be lower.

e An unemployed jute worker, out of work for slightly more than a year, lives on theoutskirts of Dhaka with his wife and two children. For 10 years, this worker was employed in one of theplants owned by the Bangladesh Jute Mills Corporation (a public enterprise). In his last year at work, heearned more than Tk 2000/month--5.7 times more than the agricultural worker and more than double theconstruction worker. Although the employee was willing to work hard to maintain his comparativelyprivileged standard of living, the jute mill had serious problems. Low investment meant that old, out-dated equipment kept breaking down; frequent electricity failures caused significant losses of productiontime; management was weak, interested mainly in extracting rents and kept hiring new workers to providejobs for relatives or to please higher-ups and politicians, so that the number of workers increased whilethe already low productivity kept declining. Eventually, the jute mill became almost bankrupt and theemployee was "separated." At first, his severance package of Tk. 24,400 ($61 1)--equal to 20 monthsbasic wage and dearness allowance--seemed a large sum to him, but it is now used up and he and hisfamily have nowhere to go.

* A teen-age girl confronts the employment discrimination against females that isrampant in Bangladesh. While the jute worker worries about how to revive his career, a femalesecondary school student in Khulna wonders if she will be allowed to have a career at all. This youngwoman would like to go to a university and her mother is very supportive, but her father, a civil servant,doubts that it makes sense to spend money on a girl's higher education, since one in every five womenwith a diploma is unemployed (compared to one in fifty males) and women who are employed earnroughly only half of what the men earn. The father thinks that his daughter should forgo her educationand get married and start a family, as most middle-class girls in Bangladesh do.

2

A ten-year old boy is representative of Bangladesh's urban child-workers. His father isa rickshaw puller in Dhaka, and his mother is a domestic servant who is currently out of work. Thefather's earnings are not enough to feed and care for this boy and his six siblings. Therefore, the boy lefthome to work at the Nilkhet Tempo Stand (a taxi cab stand) where his job is to procure customers, collecttaxi fares and clean vehicles. He earns Tk. 600 a month and sends Tk 300 to his parents. Working hoursare long, and in order to save on rent he spends the night in the taxi. The family would like the child tolive with them and attend school, but they are so poor that it needs the extra income earned by the littleboy to survive. Moreover, even though education is free, the family cannot afford to purchase clothes,books and other supplies needed at school.

B. INVESTMENT, GROWTH AND EMPLOYMENT

1. The above household examples are reflected in aggregate labor force data. Bangladesh'slabor force is growing at a very fast rate, and under-employment is rising. How serious is Bangladesh'sunder-employment problem? And, what is the best way to deal with it?

The Labor Force and the Problem of Under-Employment

2. Providing employment for the new entrants into the labor force and for the large numberof currently under-employed workers is Bangladesh's main development challenge. The 1991 LaborForce survey estimated the country's work force to be 51.2 million people. Since then, the labor forcehas been growing by around 3 percent per year, so that for 1994 it was estimated to be around 56 millionpeople. More than 1.5 million people enter Bangladesh's labor force annually.

3. The majority of workers are unaffected by labor regulations. Bangladesh has three typesof labor markets: rural informal; urban informal and formal. The formal labor market is characterizedby a contractual employment relationship and is governed by labor laws and regulations that protect theworker, such as minimum wages and allowances and limitations on the employer's ability to fireemployees. The rural and urban informal labor markets are not covered by protective labor regulations.The informal sectors dominate the labor market. In 1991, 47.2 percent of the labor force were classifiedas unpaid family workers; 15.4 percent were classified as self-employed; 13.9 percent were classified ascasual workers (day laborers); and only 11.7 percent had regular full-time wage employment--androughly one-third of those work in the public sector. About 68 percent of the work force is employed inagriculture; 20 percent of the labor force works in the services sector, and 12 percent are inmanufacturing.

3Table 1: Sectoral Distribution of Employment in 19 89a

1

Avg. Number of Hrs/yearHrs/week Workers (millions)

(thousands)

Agriculture 19 29530 29375Manufacturing 26 6528 8826Const./Transport 33 1892 3203Trade/services 38 7059 13380Other 33 1057 1769

Total 24 46067 56553

A/ Rows and columns do not add up due to rounding.

Sources Cameron, John (1993), "The Labor Force and its Use in Bangladesh in the1990s"

4. Under-employment caused by low demand for labor is the most significant problem inBangladesh's labor markets. Until recently, open unemployment was not an issue. Labor Force Surveysin the 1980's estimated unemployment at only 1.0-1.8 percent of the work force. On the other hand, the1989 Labor Force Survey estimated that 43.4 percent of the labor force was inadequately utilized orunder-employed. Measuring under-employment is difficult. The 1991 survey proposes a more realisticdefinition of unemployed, including those who were absolutely out of work plus those who worked lessthan 15 hours as unpaid family helpers during the reference week. This definition yields anunemployment rate of 18.3 percent. The survey also estimates that 42.8 percent of workers are under-employed. Since the under-employed are forced to work less than full-time, we estimate "timeunemployment"--the gap between the number of hours the labor force is willing to work and actual hoursworked--and use that as a proxy for under-employment. Based on the data in the 1989 Labor ForceSurvey, Bangladesh's time unemployment rate was approximately 23 percent in 1989 and has risen toaround 26 percent in 1994. Under-employment is currently equivalent to having a quarter of the laborforce out of work.

Investment and Growth are Key to Job Creation

5. Labor demand, and hence, under-employment and household welfare, will significantlydepend upon the level of investment and the GDP growth rate. Hence, government actions in the areas ofpublic investment, trade liberalization, privatization and competitiveness will have far-reaching effects onthe labor market. Growth-oriented policies are good for workers. This is demonstrated by the analysisand simulations presented below.

6. Projections of time unemployment rates to year 2000 were carried out under differentpolicy scenarios. Labor supply (in hours) was estimated assuming no significant change in the structureof the labor force. For year 2000 this approach yields a total labor supply figure of around 102,000million hours, which is consistent with a total labor force of 64.3 million people (37.7 million men and26.6 million women), compared with a 1991 labor force of 51.2 million people (31.1 million men and20.1 million women). This baseline estimate could be modified in at least two directions: (1) it may beappropriate to assume increasing female labor force participation rates, which would result in a laborforce larger than 64.3 million people by year 2000; and (2) the estimated labor force participation rate for

4the 15-20 age group may decline, assuming increased school attendance, which would yield lower laborsupply estimates.

Box 1: Higher Growth is Key to Worker Welfare

The importance of investment and growth for worker welfare can be demonstrated by the vastly differentexperiences of countries like Ghana and Malaysia. Since 1960 Ghana experienced prolonged economic declinefollowed by a modest recovery in the 1980s. As a result, in 1989 Ghana's labor force was distributed in roughly thesame way that it was thirty years ago. Only 19 percent of them were wage earners, and the remaining were mainlyoperating in the informal markets. Malaysia's economy took off during the same period. As a result, wageemployment which was 55 percent of the labor force in 1960 rose to 65 percent in 1989. It is also important to notethat in 1960 one in every two Malaysian wage earners worked on plantations; in 1989 one in ten did.

Growth not only affects the employment status of workers, but also their remuneration. Average real wagesin Ghana's manufacturing sector in 1984 amounted to a mere 13 percent of their level a decade earlier. Agriculturalwages display a similar pattern. By contrast, real wages in Malaysia's manufacturing sector have grown steadily at2.3 percent a year since 1970. And agricultural real wages in 1992 were more than twice the rate prevailing in 1971.

The reasons behind Malaysia's success in raising its workers' standards of living are well known. Itsinvestment rate is around 32-35 percent of GDP--compare to Bangladesh's 12-13 percent. In order to ensure thatcapital is used efficiently, key to increasing labor productivity and income, Malaysia pursued an outward orientationand depended on the market process--Bangladesh has only recently started liberalizing its economy.

7. Under a lpi growth, scenario in which reforms are implemented slowly, the GDP growth rateremains at its historical level of around 3-4 percent, with agriculture growing at 2.3 percent, manufacturingat 4 percent, other industries at 7 percent and services at 6 percent. For contrast, under a high growthscenario, rapid reforms would accelerate growth to 6-7 percent; with agriculture growing at 3 percent,manufacturing at 10 percent, other industries at 8 percent and services at 7 percent. Govenmment actionaffecting labor markets, such as minimum wages, trade union relations, hiring and firing regulations, andactions affecting relative productive input prices, particularly taxes and/or subsidies that may make someinputs cheaper (or more expensive) relative to labor--may determine whether or not growth will be relativelymore labor intensive. Here too, projections were made based on two scenarios. The low growth scenarioassumes that employment elasticities--the percentage increase in employment as a result of a one percentchange in output--are likely to be on the low side of elasticities historically observed in Bangladesh: 0.6 foragriculture, 0.4 for manufacturing, 0.5 for other industry and 0.5 for services. The high growth scenarioassumes that labor market reforms will encourage employment, particularly in the presently declining formalsector. In this case, employment elasticities would be on the high side of historical elasticity estimates: 0.9for agriculture, 0.6 for manufacturing; 0.9 for other industries and 0.7 for services.

8. The low growth scenario which assumes no acceleration of reforms, yields unacceptableresults (table 2). Continuing "business as usual" (12-13 percent investment rate, 3-4 percent growth andunchanged employment elasticities) would lead to an increase in under-employment, with timeunemployment reaching 29 percent by 1997 and 31 percent by year 2000. The fast growth scenario (6-7percent growth and an increase in employment elasticities) leads to a decline in under-employment, withtime unemployment falling to 24 percent in 1997 and 21 percent in year 2000.

5

Table 2: Under-employment Projections(time unemployment as a percent of labor supply

1993 1997 2000

Low Growth 26% 29% 31%High Growth 26% 24% 21%

Source: Bank Staff Estimates

9. The elasticities used in the above analysis are probably on the high side. Elasticitiesobserved in other developing countries indicate that maybe the low elasticity estimates used here arealready quite optimistic, and should be used in the high case scenario. If this is true, then a growth rateof 7 percent would not be sufficient to reduce the rate of under-employment. Bangladesh may need togrow at rates of around 8-10 percent to create enough employment opportunities for its growing laborforce and those who are currently under-employed. These are the GDP growth rates observed in manyEast Asian economies.

10. Regardless of the estimates for labor elasticities, the message from this analysis is clear: Inorder to bring about a significant decline in under-employment, Bangladesh must accelerate overalleconomic growth, and introduce labor market and other reforms aimed at encouraging investment andemployment creation.

C. INTERNATIONAL COMPETITIVENESS

11. Faster economic growth and the decline in under-employment are linked to enhancedcompetitiveness. Labor competitiveness is central to the success of Bangladesh's growth strategy basedon an increase in private investment and an expansion of exports. Two questions need to be addressed.How competitive is Bangladesh's labor compared to other South Asian countries? And cancompetitiveness be increased?

Defining Unit Labor Costs

12. Unit labor costs, which are used here to assess labor competitiveness, are defined as thetotal labor cost associated with producing one unit of a given good. To facilitate internationalcomparisons and competitiveness analysis, unit labor costs are measured in a common internationalcurrency, usually the dollar. Unit labor costs are calculated as the ratio of wages (expressed in dollars) toaverage labor productivity (expressed in physical output per worker). In order for a country's labor to becompetitive in the production of a good, its unit labor cost must be at least as low as that of othercountries.

13. Unit labor cost is divided into three components: the wage rate in domestic currency, theexchange rate and average labor productivity (see annex 1). Although these three components are notindependent of one another, it is useful to consider the impact of a change in each one separately. It isclear that an increase in wages would raise labor costs and reduce competitiveness. The exchange ratealso plays an important role in determining labor competitiveness, since it is used to translate domesticwages into foreign currency. Holding wages constant in domestic currency terms, an exchange rate

6depreciation will reduce their value in foreign currency and decrease unit labor costs. The impact ofproductivity on unit labor cost is also straightforward. An increase in labor productivity means that eachworker produces more goods in the same amount of time so that less labor time is spent on one unit of thegood, which reduces unit labor costs if wages remain unchanged.

14. The wage rate in any formal sector enterprise can be divided into two components: theopportunity cost of labor and a wage premium. The opportunity cost (or supply price) of labor is definedas the minimum payment that a formal sector firm needs to make in order to attract workers. Indeveloping countries like Bangladesh, with a large under-employed rural population, increases in formalemployment usually involve workers moving out of agriculture where they are usually under-employed.Unskilled rural workers are the cheapest labor that a formal firm could attract, but they will not acceptformal employment unless they are paid at least what they earn in agriculture. Therefore, the agriculturalwage rate is the most common measure of the opportunity cost of labor.

15. Formal sector labor is normally paid a premium over agricultural wages. Breaking downthe actual wages paid by exporting firms into an opportunity cost (supply price) component and a wagepremium component is particularly useful in isolating the impact on competitiveness of structural factors(the opportunity cost effect) from factors that could be modified by policy interventions (the wagepremium effect). Government policies, such as minimum wages or constraints on firing workers, onlyaffect the formal sector and cannot directly affect the opportunity cost of labor, which is determined bymarket conditions in agriculture. But they could change the wage premium paid by formal firms inexport sectors.

16. Ideally, the impact of a country's exchange rate policies should be based on some measureof deviation from an equilibrium rate. However, an estimate of the equilibrium exchange rate forBangladesh and its neighbors is difficult to obtain, since it would require making assumptions aboutequilibrium tariffs, investment and growth rates in each of the countries. Instead, the analysis presentedhere is based upon comparing official exchange rates with purchasing power parity (PPP) rates. PPPexchange rates are simply rates that would equalize the price of a basket of goods, measured in dollars, inall the countries being studied. Hence, they are not necessarily equilibrium rates.

17. It is difficult to obtain the data on physical productivity needed to calculate unit labor costs.Nearly all past work in this area used the ratio of value added to the number of workers as a proxy forlabor productivity. However, instead of measuring the dollar value of labor input per unit of output, thisapproach measures the dollar value of labor input per domestic currency unit of value added, which isdifficult to interpret. To avoid this problem, unit labor costs for the private sector presented here arebased on firm level surveys conducted as background for this report in Bangladesh, India, Pakistan, SriLanka and Vietnam (see annex 1). These surveys produced detailed information that permitted laborproductivity to be calculated in terms of physical units of output per worker. Similar information was notavailable for public enterprises, however. Instead of calculating unit labor costs for the public sector, theanalysis therefore compares the ratio of the wage bill in domestic currency to value added.

18. The results presented here must be interpreted with caution. As explained in Annex 1, theenterprise surveys covered a small number of firms in each country and there could be questions aboutthe comparability of data across countries. Every effort was taken to test the conclusions of this work byusing sensitivity analysis, and by comparing it with results of previous work. Nevertheless, analysis isonly indicative and more research in this area is needed.

:7Estimating The Opportunity Cost and Exchange Rate Gaps

19. Calculations presented in table 3 indicate that Bangladesh has the lowest opportunity cost oflabor among the large South Asian countries. The figures in the last column of table 2 (which presentsthe opportunity cost gap) imply that if the deviation of exchange rates from PPP levels, wage premia paidto formal workers and productivity levels are the same in all four countries, unit labor costs in Bangladeshwould be around 10 percent lower than in India, nearly 50 percent lower than in Pakistan and 70 percentlower than in Sri Lanka, so Bangladeshi labor should be much more competitive than that of thecomparator countries. It is interesting to note that the ranking of the four countries in terms of the supplyprice of labor is the same as their ranking in terms of per capita GDP, which is consistent with priorexpectations. It is also important to note that at official exchange rates--which are what investors actuallypay--agricultural wages in Bangladesh are slightly higher than the average for India, but lower than inneighboring West Bengal.

Table 3: Opportunity Cost of Labor(agricultural wage rate per day in 1993)

domestic official $s PPP $s Gape

Bangladesh Tk 38 $0.96 $5.37West Bengal Rp 34 $1.11 $7.24 -30%India b/ Rp 28 $0.91 $5.94 -10%Pakistan Rp 48 $1.72 $8.72 -48%Sri Lanka Rp 104 $2.15 $11.18 -73%

;' Defined as the natural logarithm of the ratio of the supply price oflabor in dollars (using the PPP exchange rate in Bangladesh) to thatof the comparator country.

- iExcluding West Bengal.

Source: Data on agricultural wages are from die four countries' statistical offices. Officialexchange rates are from the IFS; PPP exchange rates were calculated from table 30 of the1994 WDR, except for that of India which is based on Bank Staff estimates.

20. As table 4 shows, the ratio of the official to the PPP rate (defined as dollars per unit ofdomestic currency) in Bangladesh is higher than in India but lower than in Pakistan and Sri Lanka. Theseratios are used to calculate the exchange rate gap (the last row of table 4). Data in this last row indicatesthat at existing PPP rates, if the supply price of labor, the formal wage premium and productivity werethe same in the four countries and each country had the same official exchange rate as at present, unitlabor costs in Bangladesh would be some 16 percent higher than in India (less competitive), but around 10percent lower than in Pakistan and roughly 7 percent lower than in Sri Lanka (more competitive).

8

Table 4: Exchange Rate Comparisons ($/domestic currency)

Bangladesh India Pakistan Sri Lanka

Official Rate .0253 .0318 .0356 .0207PPP rate .1414 .2140 .1805 .1076

Ratio 17.9% 15.3% 19.7% 19.2%

Gapa/ 16% -10% -7%

a' Defined as the difference between the natural logarithm of the ratio ofBangladesh's official to PPP rate and the natural logarithm of the sameratio for the comparator country.

Source: International Financial Statistics, World Development Report 1994 and Datt (1993).

21. Tables 3 and 4 can be used to calculate the net impact of the opportunity cost and exchangerate gaps on relative unit labor costs in Bangladesh and its neighbors. The sum of these two effects,referred to as the "macro-gap" in the rest of this chapter, provides a measure of the impact of theeconomy's structure and of macro-economic policies on labor competitiveness. Measurements of themacro-gap indicate that Bangladesh has an advantage over its neighbors except India. The opportunity

cost gap provides it with a 10 percent advantage over India, which is more than offset by the exchangerate gap of 16 percent in India's favor, the overall macro-gap implies an advantage of around 6 percent inunit labor costs for India. However, Bangladesh seems to have a huge advantage vis- a-vis Pakistan and

Sri Lanka, with a macro-gap of some -58 percent with Pakistan and -80 percent with Sri Lanka.

Labor Competitiveness in Garments

22. The survey data indicates that Bangladesh's unit labor costs in garments are probably lowerthan in any other South Asian country. This corroborates other evidence from industrial surveys, and thefact that garment exports have been expanding rapidly, which indicate that Bangladesh is very competitivein this area. Table 5 shows that Bangladesh's unit labor costs for T-shirts are much lower than in thecomparators. Similar analysis was carried out for jeans and men's shirts and supported the sameconclusion of Bangladesh's competitive superiority in the garments sector vis a vis the other South Asiancomparators. It is important to note that the two key factors determining unit labor costs (wagesexpressed in dollars and productivity) work in Bangladesh's favor. In T-shirts, it has the lowest wage ofall countries and highest productivity among all comparators, except India.

9

Table 5: Unit Labor Costs for T-shirts

Unit Labor Cost Wages Productivity($/shirt) ($/year) (shirt/worker)

Bangladesh 0.11 290 2,536West Bengal 0.33 595 1,828India 0.26 668 2,592Pakistan 0.43 1,343 3,100Sri Lanka 0.79 570 719

Source: Staff estimates based on enterprise surveys.

23. Table 6 shows that Bangladesh's unit labor cost advantage in producing T-shirts far exceedsthe macro-gap calculated earlier. The wage premium Bangladesh garments workers receive is lower thanthat received by workers in all comparators except Sri Lanka. Moreover, productivity differences do noterode this wage advantage. If the wage premium is used as a measure of market segmentation anddistortion, the garments labor market in Bangladesh is less distorted than in all comparators except SriLanka. In Bangladesh the average wage of garments workers is some 74 percent above the agriculturalwage rate. This may appear high, but it is still better than in West Bengal, India and Pakistan. The thirdcolumn of table 6 indicates that this is a key reason for Bangladesh's unit labor cost advantage. The dataalso indicates the importance of productivity in determining competitiveness. Bangladesh's productivityadvantage in this sector is responsible for nearly a 33 percent advantage over West Bengal, and for asmuch as a 127 percent advantage over Sri Lanka.

Table 6: Decomposition of the Difference inUnit Labor Costs For T-shirts

(percent difference with Bangladesha')

Overall-gapb/ Macro-gap Wage-gap"I Prod-gapW

West Bengal -105 -14 -57 -33India -81 6 -89 2Pakistan -133 -58 -95 20Sri Lanka -194 -80 13 -127

^' Calculated as the natural logarithm of the ratio of the variable in Bangladeshto its value in the comparator. A negative figure implies a cost advantage forBangladesh. Also, note that columns (2), (3) and (4) must add up to column(1).

b/ Defined as the natural logarithm of the ratio of unit labor costs in Bangladeshand the comparator.

i! The wage premium is defined as the sector's average wage divided by theagricultural wage; the wage-gap is the natural logarithm of the ratio of thewage premium in Bangladesh to that of the comparator country.

di Defined as the natural logarithm of the ratio of labor productivity inBangladesh and the comparator.

Source: Bank staff calculations from survey data.

10

24. A new regional competitor is emerging in the garments market. Vietnam is quicklybecoming an important competitor for foreign investment in the garments sector. To avoid losing marketshare, it is important that Bangladesh ensure continued labor competitiveness vis-a-vis Vietnam. The datapresented in table 7 provide some cause for concern, since it shows that Vietnam is more competitive inproducing jeans and that Bangladesh's unit labor cost advantage in other product lines is much smallerthan its advantage vis-5-vis other South Asian countries. Unit labor costs for producing jeans inBangladesh may be as much as 21 percent higher than in Vietnam. However, Bangladesh's unit laborcosts for T-shirts and Men's shirts remain lower by some 53 percent and 37 percent, respectively.

Table 7: Bangladesh and VietnamUnit Labor Costs in Gannents

($/dozen)

Bangladesh Vietnam

T-shirtsaverage 1.37 2.50minimum 1.00 1.87maximum 2.00 3.86

Men's Shirtsaverage 2.96 4.44minimum 2.25 2.64maximum 3.60 6.63

Jeansaverage 5.94 4.81minimum 5.00 4.05maximum 8.02 7.72

Source: Staff calculations from survey data.

Labor Competitiveness in the Leather Sector

25. Bangladesh's labor is marginally competitive in leather according to the surveys carried outin preparation of this report. Again, this result is consistent with the results of industrial surveys andexplains the evolution of this industry. Table 8 shows that at an estimated $3.19/100 sq. ft. unit laborcosts for finished leather in Bangladesh are lower than in all the comparators, but that India (excludingWest Bengal) and West Bengal are very close competitors. Wages in dollar terms are higher inBangladesh than in India, while Bangladesh's productivity is lower than in West Bengal. Moreover, theconclusion that Bangladesh has the lowest unit labor cost in finished leather production is not as robust asresults concerning unit labor costs in garments. There appears to be much greater variability amongfirms in the leather sample than existed for garments. When unit labor costs of the least efficient firm arecompared, instead of a weighted average, Bangladesh's unit labor costs are $10/100 sq. ft. much higherthan West Bengal's ($4.2) or India's ($4.8). On the other hand, using data for the most efficient firmyields results that are very similar to those obtained using weighted averages.

11

Table 8: Unit Labor Costs for Finished Leather

Unit Labor Wages ProductivityCost ($/year) (100 sq.ft/

($1100 sq.ft.) worker)

Bangladesh 3.19 880 276

West Bengal 3.90 1,125 288India 3.51 587 167Pakistan 7.99 1,948 243Sri Lanka 12.85 1,071 83

Source: Staff estimates based on enterprise surveys.

26. Table 9 shows that Bangladesh's unit labor cost advantage vis-a-vis India and West Bengal ismuch lower than the macro-gap. The impact of labor market segmentation and distortions on wages inBangladesh's leather sector appears to be more important than in garments, with wages in leather being282 percent higher than agricultural wages. This wage premium is higher than that observed in all othercountries except Pakistan and implies a 35 percent loss of competitiveness vis-a-vis India and a 61 percentloss vis-a-vis Sri Lanka. The importance of productivity differentials in determining laborcompetitiveness is once again highlighted. If productivity in Bangladesh were the same as in India, itsunit labor costs would have been 41 percent higher, implying a significant advantage for Indian workers.Similarly, differences in productivity explain most of Bangladesh's unit labor cost advantage over Sri

Lanka.

Table 9: Decomposition of the Difference in Unit Labor Costs For Leather(percent difference with Bangladesh W)

Overall-gap b/ Macro-gap Wage-gaps' Prod-gap"

West Bengal -20 -14 -10 5India -9 6 35 -50Pakistan -92 -58 -21 -13Sri Lanka -139 -80 61 -120

i Calculated as the natural logarithm of the ratio of the variable in Bangladesh to itsvalue in the comparator. A negative figure implies a cost advantage for Bangladesh.Also, note that columns (2), (3) and (4) must add up to column (1).

hi Defined as the natural logarithm of the ratio of unit labor costs in Bangladesh and thecomparator.

' The wage premium is defined as the sector's average wage divided by the agriculturalwage; the wage-gap is the natural logarithm of the ratio of the wage premium inBangladesh to that of the comparator country.

g' Productivity gap is defined as the natural logarithm of the ratio of labor productivityin Bangladesh and the comparator.

Smour: Bank staff calculations from survey data.

112

Labor Competitiveness in Textiles

27. The data in table 10 indicates that Bangladesh's unit labor costs for producing cotton yarnare higher than in all comparators except Sri Lanka. This could be one of the factors explaining the weakposition of this industry. While Bangladesh's wages are not particularly high in dollar terms--averagedollar wages in Bangladesh are lower than in all comparators except West Bengal--productivity inBangladesh is much lower than in all comparators. The conclusion that Bangladesh's labor is notcompetitive in textiles appears to be very robust. When unit labor costs for producing cotton cloth werecalculated, they supported the same conclusion: at 12 cents/meter, unit labor costs for cotton clothproduced by Bangladesh firms is double that for West Bengal and Pakistan. If the most efficient firm inproducing yam is used as a comparator instead of the weighted average, the unit labor costs forBangladesh would be higher than in all comparators, including Sri Lanka. The same result is achievedwhen the comparison uses data for the least efficient firm. Using gray cloth as a benchmark instead ofyarn also does not change the conclusions significantly. The most efficient Bangladesh's firm in thesample has higher unit labor costs than the most efficient firms in India and Pakistan, but lower unit laborcosts than Sri Lanka; the same is true for the least efficient firm in the sample.

Table 10: Unit Labor Costs for Cotton Yarn

Unit Labor Cost Wages Productivity($/Kg) ($/year) (Kglworker)

Bangladesh 0.46 780 1,699Sri Lanka 0.19 611 3,192West Bengal 0.24 866 3,597India 0.15 1,622 10,475Pakistan 0.53 1,208 2,296

Source: Staff estimates based on enterprise surveys.

13

Table 11: Decomposition of the Difference in Unit Labor Costs Cotton Yarn( percent difference with Bangladesh "')

Overall-gap ht Macro-gap Wage-gaps' Prod-gap a'

West Bengal 87 -24 39 63India 65 6 -16 75Pakistan 109 -58 -15 182Sri Lanka -14 -80 37 30

a/ Calculated as the natural logarithm of the ratio of the variable in Bangladesh to itsvalue in the comparator. A negative figure implies a cost advantage for Bangladesh.Also, note that columns (2), (3) and (4) must add up to column (1).Defined as the natural logarithm of the ratio of unit labor costs in Bangladesh and thecomparator.

' The wage premnium is defined as the sector's average wage divided by the agriculturalwage; the wage-gap is the natural logarithm of the ratio of the wage premium inBangladesh to that of the comparator country.

d/ Productivity gap is defined as the natural logaritim of the ratio of labor productivityin Bangladesh and the comparator.

Source: Bank staff calculations from survey data.

28. The decomposition exercise presented in table 11 indicates that Bangladesh's competitivedisadvantage is due mostly to low productivity. Productivity in Bangladesh may be more than 180percent lower than in Pakistan, 75 percent lower than in India, 63 percent lower than in West Bengal and30 percent lower than in Sri Lanka. Wages do not adjust to offset this low productivity. Average wagecosts in Bangladesh's yarn-producing firms are 238 percent higher than agricultural wages. The wagepremia in Bangladesh's garments and leather sectors are also high, but are offset by higher laborproductivity. In contrast, in textiles, low relative productivity reinforces the negative impact of the wagepremia.

29. A more detailed comparison between wage practices in the dynamic garments sector and theproblematic cotton textiles industry, using regression analysis reinforces the conclusion that lowproductivity, not higher wages, lies at the heart of Bangladesh's competitiveness problem. Using datafrom the 1993 Industrial Survey, wages for 156 firms-45 in the textiles sector and 111 in garments--wereanalyzed. It was found that wages in textiles were not significantly higher than in garments, so that thedifference in the two sectors' performance must be due mainly to different relative productivities. Theaverage wage for employees in textiles is only slightly above the average wage in garments, Tk.1,633/month compared to Tk. 1,528. However, it appears that most of the high pay in the textiles sectorgoes to non-production staff. The average wages of production workers, male as well as female, areactually higher in garments than in textiles. Moreover, the argument that the garments sector iscompetitive because it employs a large number of low-paid women does not appear to hold. At Tk.1,436/month, the average wage of a female production worker in garments is not much different from thewage of a male production worker in textiles.

14Labor Competitiveness In Public Enterprises

30. A possible explanation for the textile sector's lack of competitiveness is that private firms inthe sector follow public enterprise practices (perhaps due to trade union and government pressures) andcould afford to be non-competitive because they were heavily protected. This would mean that thecompetitiveness problem in Bangladesh is mainly a public enterprise problem and should be tackled by acombination of privatization, restructuring and changes in public sector hiring and wage policies. In viewof this, table 12 analyzes labor competitiveness in public enterprises.

Table 12: Public Sector Competitiveness

Wage Premium '/ VA/worker"' Wage/VA($/year) (Percent) (dollar/year) (ratio)

Bangladesh 821 85% 505 1.63Textiles 1371 137% 817 1.68Engimeerinm 1683 157% 5959 0.28Chemicals'

IndiaTextiles 1346 140% 736 1.83Engineering 2412 198% 5560 0.43Chemicals' 2879 216% 12231 0.24

PakistanTextiles n.a. n.a. n.a. n.a.Engineering 2525 139% 3355 0.75Chemicalsc 4165 189% 17603 0.24

Sri LankaTextiles n.a. n.a. n.a. n.a.Engineering 1035 28% n.a. n.a.Chemicals' 2503 116% n.a. n.a.

lDefined as the natural logarithm of the ratio of average wages in the sector tothe agricultural wage rate.

' VA Value Added' Includes Fertilizers

Source: Statistical Offices and Ministries of Finance e of the four countries.

31. High dollar wages to not explain lack of competitiveness. As table 12 shows, in most casesaverage dollar wage costs in Bangladesh's public enterprises are lower than in the comparator countries.For example, average labor costs in textile mills are some 49 percent lower than in India. Average costsin engineering are 56 percent lower than in India and 61 percent lower than in Pakistan, but 28 percenthigher than Sri Lanka. Average wage costs in chemicals/fertilizers are about 54 percent lower than inIndia, 91 percent lower than Pakistan and 40 percent lower than Sri Lanka. It seems, therefore, that highdollar wages are not responsible for low labor competitiveness in Bangladesh's public enterprises, relativeto other South Asian countries. Clearly, Bangladesh's public sector would have been more competitive ifit paid its workers wages closer to the opportunity cost of labor. However, (as shown in the secondcolumn of table 12) the wage premia paid to public sector workers in Bangladesh are lower than thosepaid in India and Pakistan, although they are much higher than those paid in Sri Lanka.

1532. Bangladesh's public enterprises appear to have the lowest labor productivity in South Asia.The data in table 12 indicates that value added per worker in Bangladesh's textiles mills is around 38percent lower than in India. In engineering, valued added per worker is 192 percent lower than in Indiaand 141 percent lower than in Pakistan. In chemicals/fertilizers, value added per worker is about 72percent lower than in India and 108 percent lower than in Pakistan. Since public enterprises in India andPakistan are not particularly known for their high efficiency levels, these results reflect very poorly onlabor productivity in Bangladesh's public enterprises. It seems clear that compared to other South Asiancountries, the problem of public sector competitiveness in Bangladesh is more a reflection of lowproductivity than high wage costs in dollar terms.

Box 2: Employment and Production in Public and Private Jute Mills

A recent study compared employment and production of public sector and privatized jute mills, between1983 and 1988. Using a sample of 31 public mills and 31 privatized mills, it found that private mills reducedmanagerial, clerical and white collar employment, but did not lower manual production employment. Bylowering employment and keeping output unchanged, the privatized mills increased productivity.

Why did public sector employment expand during this period relative to private mills? One explanationis that government wanted to create more job opportunities for the poor. But this would imply greateremployment for manual laborers whose altematives are more limited, which did not occur. This brings us tothe second, less laudable, explanation. The public sector may have been used by politicians to dole out jobs inresponse to political pressures, a phenomenon known as "clientelism." Clientelism is more likely to generateemployment for white collar workers and organized (permanent) blue collar workers, who have much greaterpolitical power than the casual (badli) workers.

Percentage change in employment and output (1983-88)

Public Mill& Private Mflls Privatization Effect

Managerial 27 -5 -32 **

Clerical 20 -13 -33 **

White Collar 23 -9 -32 **Permanent Manual 0.9 -6.1 -7 *Casual Manual 12 12 24*

Output 12 2 -10

** Statistically significant at the 1 percent level.* Statistically significant at the 5 percent level.

Source: Bhaskar and Khan (1995)

Policy Implications

33. The above analysis indicates that labor competitiveness in Bangladesh varies considerablyfrom one sector to another. Unprotected sectors with no public enterprise presence, such as garments andleather, tend to have low unit labor costs. Sectors that have been used to operating behind protectivetariffs and import quotas (textiles) or with important explicit and implicit subsidies (jute) and with asignificant public enterprise presence (jute and textiles) are not competitive. Moreover, it seems clearthat lack of competitiveness in those important sectors is mainly due to low productivity, as wages inBangladesh do not appear to be high relative to its neighbors.

34. Some may be tempted to try resolving this problem by increasing public investment in theuncompetitive sectors with the objective of raising labor productivity through capital deepening. Thiswould be a mistake. Raising public investment in manufacturing firms (rather than in social sectors andinfrastructure where it is badly needed) may create more problems than it solves. There is no guaranteethat new public investments would be more efficient than previous ones. Enterprises may be losingmoney for reasons other than lack of capital. The initial investment may have been a mistake andinvesting more now would lead to a further misallocation of resources. Moreover, the management ofpublic enterprises in Bangladesh, as in many other countries, has a long history of inefficiency. Puttingmore resources into those enterprises may simply be "throwing good money after bad money." It couldbe desirable to have more private investment in some of those sectors. However, this can only beachieved by creating an environment conducive to private sector development, and not through morepublic spending.

35. Eliminating over-manning in the public sector and reducing the cost to private firms oflaying off workers would help improve productivity. This may appear contradictory. On the one hand,the objective is to increase employment; while on the other hand, it is recommended that over-manning beeliminated. In fact there is no contradiction. The reason that there is so much under-employment inBangladesh is that there is too much employment in the wrong sectors (and probably at the wrong wages).Total employment would increase if this changed. The huge losses of sectors like jute and textiles arefinanced by the banking system. Scarce resources have been used to subsidize a few thousandunproductive jobs. By eliminating those jobs resources can be channeled to other competitive sectors andrural areas, generating more productive employment. More jobs will be created than destroyed.Moreover, the restructured sectors would in time become competitive and in the future attract moreprivate investment, creating new job opportunities. It seems clear that safety net protection for displacedworkers will be needed, especially during the transition.

D. TOWARD EFFICIENT REGULATIONS AFFECTING RESTRUCTURING

36. If over-manning is the main cause of low productivity and low competitiveness in somesectors, then it is important that labor market regulations provide firms with sufficient flexibility torestructure themselves. Policies in this area have to deal with the problem of downsizing publicenterprises, as well as with the needs of private firms to continuously restructure in response to marketsignals. How can government regulations balance the necessity of maintaining labor market flexibilitywith the legitimate concerns of workers and their families for income security?

17

Bangladesh's Restructuring Regulations

37. Bangladesh's labor laws offer sufficient flexibility for firms to restructure themselves.Under the Employment of Labor (Standing Orders) Act, 1965, a worker may be removed from service byretrenchment, discharge, dismissal or termination of employment. A worker may be retrenched fromservice on the grounds of redundancy, in which case, the employer must give one month's notice inwriting, with a copy to the Chief Inspector of Factories. The employer must also pay the worker 30 dayswages for every completed year of service or gratuity for the period, if any, whichever is higher. Before1985, the cost of retrenchment was 14 days' wages for every completed year of service, but this waschanged by Ordinance No. XVI. The cost of firm restructuring thus doubled just as the Bangladeshieconomy was entering a period of rapid change, due to liberalization and privatization. The law alsoconstrains employers' ability to choose which worker to retrench and which worker to rehire if the needarises. Employers are required to retrench the last worker hired in the particular category. Exceptionsare allowed, but the employer must record in writing the reasons for retrenching a worker other than theone who was recruited last. If the firm wants to employ any person within a period of one year from thedate of retrenchment, it must first offer the job to the retrenched worker.

38. The lump sum payments workers receive when employment is terminated are over andabove what they get from their provident funds, to which employers must contribute 8.3 percent of eachworker's wage. Bangladesh's regulations concerning worker termination impose thus an additional costto employers of 16.6 percent of the worker's base wage--30 days salary per year of service plus an 8.3percent provident fund contribution.

39. Bangladesh's legal requirements appear similar to those of other developing countries, butimpose a much higher cost on employers than regulations in developed countries. For example, Turkey'sregulations are quite similar to those of Bangladesh. Turkish law requires two weeks' notice to be givento workers employed less than six months, incrementally rising to eight weeks' notice for workersemployed more than three years. It also provides that seniority severance pay equal to 30 days' wagesfor each complete year of service be paid to employees if they are terminated for other than immoral ordishonorable conduct. In addition payment is made if the worker terminates the contract in order toqualify for an old-age or disability pension from an approved organization. Although Turkey has nounemployment benefits scheme, legislation does provide for old age or retirement benefits for employedworkers on a joint contributory basis.

40. Bangladesh's regulations allow firms greater flexibility than in Sri Lanka. Sri Lankanemployers are required to contribute 12 percent of the employee's wage to the provident fund and anadditional 3 percent to the employees' trust fund. Gratuity, payable to workers who have worked for aminimum of 5 years, is half a month's pay per year of service. These termination benefits represent 19percent of wages per year of service, which is comparable to Bangladesh's regulations. However, SriLanka's laws significantly restrict employers' ability to terminate workers. Firms with more than 15workers cannot retrench on non-disciplinary grounds without written consent of the workers who will beterminated. To obtain this consent firms usually have to offer severance payments much larger thanrequired by law.

41. Retrenchment obligations in Bangladesh are more than those prevailing in developedcountries, such as the United Kingdom. The UK provides for payment of a lump sum based on length ofservice and earnings to workers who have been employed for two years (or five years if they work only 8to 16 hours a week) when they are dismissed on grounds of redundancy. The redundancy payment is 11/2 weeks' pay for each year of service in which the employee was age 41 or over, 1 weeks' pay for each

18year between the ages of 22 and 40 and half a weeks' pay for each year between the ages of 18 and 21,subject to a maximum of 20 years' service. Only complete years of service are counted. Maximnumredundancy pay in the U.K. would thus be 30 weeks' pay for those aged 61 with 20 years' service. InBangladesh, such a worker would receive at least 80 weeks' pay. However, unlike in Bangladesh,redundant workers in the UK can also claim unemployment benefits from the state national insurancescheme.

Box 3: The Negative Impact of High Retrenchment Costsin Brazil, India and Zimbabwe

Regulations that increase the cost of laying off workers could discourage the creation of formal employment. Firms may try toavoid those costs by hiring fewer workers, or by employing them on a daily (casual) basis which defeats the objective of providing incomesecurity for labor. Empirical evidence from Brazil, India and Zimbabwe provide some support to the view that high severance payregulations have a negative impact on formnal sector employment.

The Brazilian Rural Labor Statute of 1963 required agricultural labor contracts to meet some minimum standards, includingseverance pay. This may have led finns to hire temporary (casual) rather than permanent workers. A study of the Northeastern regionfound that although total employment increased in the 1960s, permanent employment declined, falling from 45 to 39 percent of peakseason employment. Thus, the legislation may have been counterproductive as it led to a significant reduction in permanent contractswhich offer workers greater protection.

India and Zimbabwe have very stringent job security regulations that require firms to obtain prior government consent beforelaying off workers. A statistical study of the two countries found that finns in the formal sector adjusted to the cost of this regulation byemploying fewer workers. The study estimated that on average, these regulations reduced formal employment in 35 Indian industries by 18percent and in 29 industries in Zimbabwe by 25 percent.

Not all studies detect a negative impact of severance pay regulations on formal employment. For example, most Malaysian finnsin an ILO survey stated that job security laws had no impact on employment. However, the available evidence has been strong enough toconvince countries like Senegal, Argentina and Chile to revise their regulations and reduce the cost of fu-m restructuring.

Public Sector Practice

42. In actual practice, firm restructuring costs in Bangladesh are often higher than those prescribedby the labor law. This is because private employers sometimes have to follow public sector practice in thisarea, or else face serious industrial relations problems. In the late 1980's the public sector decided to doublethe amount of gratuity, to 60 days wages per year of service. This now represents the absolute minimumthat a private employer must pay when retrenching workers. The Government has been avoidingretrenchment by resorting to "voluntary separation," which involves paying separated workers more than thelegally required severance package for open retrenchment. Around 20 thousand public sector workers wereseparated in FY93 and a similar number in FY94; workers in private concerns are demanding similartreatment. The most important recent cases of voluntary separation in the public sector concern railways andjute sector workers--around 15 thousand workers in FY94.

43. The Railway's goal was to eliminate 15,000 workers between FY92 and FY95--10,000through voluntary separation and 5,000 by normal attrition. This objective is being achieved, at a cost ofaround Tk. 275,000 per separated worker, which is around 6 times the normal retrenchment gratuityprescribed by law. To qualify for the Railways' special voluntary retirement benefits, applicants had to haveat least ten years qualifying service for pension purposes and be younger than age 52. Severance pay wasdetermined by a formula best described by the following example provided by Bangladesh Railways in theirdocumentation: A worker with 26 years of service (of which only 25 count toward the maximum pensionentitlement) and a monthly basic wage of Tk. 4,115 would receive a gratuity of Tk. 329,000, a lump sum ofTk. 192,582, pay against left over leave of Tk. 49,380 and a medical allowance of Tk. 17,550. The

19worker's total severance pay is Tk. 588,172, or 5.5 months of basic wage for each year of service. Inaddition, the worker would receive "leave prior to retirement," which is full pay and allowances for sixmonths and half pay and allowances for a further six months--roughly equivalent to a further 18 monthsbasic wages-- and also payments from the Provident Funds.

44. The package proposed to jute mill workers is less generous than that developed for theRailways. It includes both separation benefits and retraining. Gratuity is set at 60 days basic wage per yearof service. For a worker with 25 years of service and a Tk. 4,000/month basic wage the maximum to bepaid is Tk. 200,000. The average cost per separated worker, around Tk 55,000. The jute mills' package isexpected to affect around 16,000 workers and includes retraining workers for alternative employment. Toparticipate in the retraining programs, which rely on existing Government and NGO training facilities, theworker has to contribute Tk. 4,000. The worker contribution does not cover the programs' full costs.Workers can choose either a skill training or self-employment training program. Under the skill trainingoption, workers can select any of 20 training courses designed to help them obtain employment outside thejute sector. Workers who select the self-employment training option participate in a six-week coursecovering the preparation of business plans, basic management, accounting, finance and sales.

Policy Implications

45. Policies for labor retrenchment raise complex political economy, efficiency and equityconsiderations. In the short run the political economy issues tend to dominate. Governments of manycountries have sought to buy labor's acceptance of policy reforms through generous severance pay. In theview of those governments the long-term benefits from economic reforms far outweigh the short-runbudgetary cost. Generous payments to Bangladesh's Railways and jute workers may be an acceptable pricefor reforming those key economic sectors.

46. To be sustainable in the long run, retrenchment policies must achieve efficiency and equityobjectives. Relatively high severance payments, especially in Bangladesh's public sector, are not conduciveto efficient firm restructuring, raising productivity and greater competitiveness. They are also unfair,because they protect a small group of "insiders" who already have regular jobs, at the expense of the vastmajority of workers whose chances for getting formal employment declines as higher costs discourage firmsfrom hiring more permanent workers.

47. The medium-term objective of policy makers in Bangladesh should be to lower firmrestructuring costs. It is important to ensure that Bangladeshi firms have sufficient flexibility to achieve andmaintain competitiveness on world markets. This could by done by (1) rationalizing the benefits offered toretrenched public sector workers; (2) introducing an urban safety net aimed at supporting all retrenchedworkers; and (3) modifying the labor law's provisions specifying worker benefits upon terminatingemployment.

48. The public sector should not pay separated workers more than the 30 days' basic wages peryear of service prescribed by law. Public sector separation benefits should be reduced gradually, reachinglegally mandated levels by the time the current large scale retrenchment is completed. In order to effectivelyreduce separation costs while still providing redundant workers with an acceptable level of safety netprotection, the mode of paying separation benefits could be changed. Benefits may be paid with bonds orvouchers that would provide workers a monthly income that gives adequate protection. The boi:ds orvouchers would not be redeemed by the Government for cash, but could be used to purchase designatedtraining or as collateral with financial institutions to obtain credit. A secondary market for the vouchers

20

would probably emerge as some workers, eager to obtain immediate cash, would seek to sell them at adiscount.

49. To reduce the retrenchment costs bome by employers, it will be necessary to introduceadequate altemative safety net protection for displaced workers. Since in Bangladesh, workers have nosocial protection whatever other than severance pay, organized labor strongly resists firm restructuringefforts, except when accompanied by high severance pay, out of concern that otherwise their displacedworkers will be driven into poverty. Introducing some form of social protection should promote labor'sacceptance of retrenchment, reduce the cost of firm restructuring and enhance economic flexibility.

50. An urban safety net could be modeled along the lines of existing rual public works programs,Food for Work. Public works programs have been used in several developing countries, including ruralBangladesh, to help the unemployed poor provided they are willing to work for market wages. The lowwages act as a self-targeting mechanism because only the truly needy will accept the job. Because theseschemes utilize the time of the poor, they are a particularly appropriate tool during important transitionswhen many people lose their jobs or major recessions when job opportunities are unavailable. An attractivefeature of those schemes is that they could be used to provide protection to all workers and need not belimited to those with formal jobs.

51. Bangladesh could benefit from a review of the Chilean and Indian experiences with publicworks programs. Chile introduced large government-financed urban public works programs during therecessions of the mid-1970s and early 1980s. The program's objective was to create employment; at thebottom of the recession of the early 1980s it absorbed more than 10 percent of the labor force. ByDecember 1988, a year after Chile's vigorous economic recovery, that share had fallen to less than 0.1percent of the labor force. Despite the usefulness of public works programs, their administration ischallenging, especially when it comes to proper targeting. For example, the Employment Guarantee Schemeused in the Indian state of Maharashtra was successful as long as it paid low wages. A doubling of the wagein the late 1980s in line with changes in the statutory minimum, combined with budgetary pressures haseffectively eliminated the employment guarantee which was the scheme's main objective.

52. Alternatively, the Government may consider establishing a "Social Fund" with donor supportthat would channel a reasonable level of resources to displaced formal sector workers. The Fund'sresources could be used to develop income-generating activities and help workers establish micro-enterprises. To avoid additional bureaucracy, the programs could be managed by a select group of non-governmental organizations (NGOs). The World Bank has assisted governments of more than 20 countriesin setting up and financing Social Funds, and Bangladesh could make use of this experience and avoidmistakes committed elsewhere. In this regard, Bangladesh could benefit from the experience of Bolivia,Egypt, Guinea and Senegal, which have such funds that support displaced workers, especially displacedfrom the public sector.

53. As a final step in reforming the restructuring process, labor laws should be changed to furtherreduce the cost of firm restructuring. Lowering public sector norms and aligning them with existing legalrequirements, together with introducing some form of government-supported safety net, would reduceworker resistance to retrenchment and help reduce private firms' restructuring costs. Private firms wouldonly pay the legally mandated 30 days' basic wage for each year of service. The Government could thenstart lowering this payment to levels in competing countries, beginning with a reduction of the severancepayment requirement to its pre-1985 level of 14 days' basic wages per year of service.

21

54. There could be other approaches to reducing private finns' restructuring costs. For example,consideration may be given to the creation of a severance pay fund financed by employer and workercontributions. The fund would make a portion of the severance payment due to retrenchees. It is importantthat only partial payments are made and that employers bear directly some of the costs, in order to deterunneeded retrenchment. Bangladesh could benefit from the United Kingdom's experience with theredundancy payments fund.

E. POLICIES AFFECTING WAGES

55. Efficient restructuring regulations must be accompanied by appropriate wage policies toensure competitiveness. The level of formal sector wages has important efficiency and equityconsiderations. If set too high, for example, as a result of minimum wage legislation or public sectorwage policy, it could discourage modem firms from expanding employment and output. The cost of thispolicy will be borne by the millions of informal and rural workers who aspire to a formal job but cannotget it. Thus, this policy could help a few "insiders" with good jobs at the expense of the many lessprivilege "outsiders." This discussion gives rise to two questions. What is the structure of wages inBangladesh? How do minimum wages and public sector wage policies affect the functioning of the labormarket?

A Description of Wage Differentials

56. Wages differ depending on whether workers are in the public or private sectors, andwhether they are in urban or rural areas. Differences can be huge (table 13). An unskilled public sectorworker earns nearly three times as much as a similarly unskilled agricultural worker in Rangpur. Andthis does not take account of the fact that the agricultural worker spends many days unemployed.According to the 1989 Labor Force Survey, the average agricultural laborer works for 124 days per year,while the average worker in manufacturing is employed for 169 days. Differences exist within urbanareas. A construction worker, typically operating informally as a day laborer, earns 22 percent less thana public enterprise worker with similar skills. Wage differentiation is also due to geographic location.An agricultural worker in Chittagong earns 156 percent more than one in Rangpur.

57. The decline in real rural wages since the mid-1980s, while urban wages continue raisingmust be a serious cause for concern. The real wage of a rural worker in Rangpur fell by 48 percentbetween FY86 and 93. Even in the more affluent Chittagong and Dhaka areas, rural real wages fell by21 percent and 9 percent respectively. At the same time, real wages for unskilled workers in the publicsector or of construction workers rose by 10 percent and 33 percent. The fall in real agricultural wages isan indication that economic growth has been too low, and the demand for labor has been insufficient toabsorb the new entrants into the labor force. The increasing inequality and the rise in the premium paidto formal and informal urban workers, may be a reflection of government policies that keep urban wagesabove the supply price of labor: minimum wages and public sector pay policies.

22

Table 13: The Structure of Wages(taka/day) i'

FY86 FY93 Percentchange

(%)

Urban wagespublic sector (grade 1) 40 74 85engineering 26 66 154small & cottage ind. 29 58 100construction 26 58 123

Rural wages 50 64 28Chittagong 30 44 47Dhaka 30 25 17Rangpur

Price IndicesDhaka middle income 436 734 68Bangladesh Rural 369 592 60

"' Daily earnings for public enterprise workers wereobtained by dividing their total monthly earnings (basicwage plus allowances) by 26. Wages for private sectorare for unskilled workers, except in small and cottageindustries where only data for skilled workers wasavailable.

Source: Bangladesh Bureau of Statistics.

23

Box 4: Labor Market Integration

Does the existence of large wage differentials imply the Bangladesh's various labor markets are not linked to one another? Co-integration tests indicate that is not the case, and the labor markets appear to be well integrated. This is consistent with economic theoryand other country experiences. Labor markets in the same country can be segmented, in the sense that wage differentials may exist amongdifferent regions or different types of markets (formal vis-a-vis informal, private vis-A-vis public) but also integrated because shocks thatlead to changes in wages in one market usually result in changes in the rest of the economy--for example, as labor moves out of the lowwage or high unemployment sector. This result can be obtained under a generalized version of the Harris-Todaro labor market model,where the wage differential represents the cost of migration and/or the risk of being unemployed after moving from one market to another.

Agricultural wages across districts are co-integrated. Co-integration tests were carried out using monthly wage data for 20 districtsfrom January 1975 to December 1989. These series were collected by the Bangladesh Bureau of Statistics and published in the Yearbookof Agricultural Statistics of Bangladesh. Box table 4.1 presents a sanple of pairwise co-integration test results for wages in differentdistricts. A total of 37 tests were conducted for this report. Most of the districts involved are neighbors. However, the analysis alsoincluded less obviously linked pairs of districts, such as districts located across one of the big rivers or districts located as far apart aspossible. The results show without exception that the district wise wage rate series are co-integrated. This implies that all the variouswages are linked in the long run.

Box Table 4.1: Co-Integration Tests for Agricultural Wages Rv(Mckinnon Critical Value (5 percent): -3.37)

Pair of Districts ADF-Test

Chittagong-Noakhali -5.82Comilla-Dhaka -4.98Mymensingh-Tangail -6.47Bogra-Tangail -6.68Dhaka-Faridpur -4.99Chittagong-Sylhet -5.88Khulna-Sylhet -7.78

a/ The assumption of non-stationarity is not rejected forany of the agricultural wage series.

Source: Azam, Jean Paul (1994), "Labor Market Integrationin Bangladesh."

The industrial, construction and agricultural labor mnarkets are also well-integrated. Box table 4.2 presents the results of co-integration testsfor agricultural wages in Noakhali and Rajshahi and wages in the urban construction sector (informal) and the urban jute textiles sector (formal). Itindicates that these four wage rates are strongly co-integrated. This is an important result, since it shows that in the long run, the labor market inBangladesh is integrated both spatially and across industries.

Box table 4.2: Sample Co-Integration Tests for Urban and Rural Wagesg(Mckinnon Critical Value (5 percent): -3.37)

Pair of Sectors ADF-Tes

Noakhali agriculture-construction -4.74Noakhali agriculture-jute textiles -4.47Rajshahi agriculture-construction -5.08Rajshahi agriculture-jute textiles -5.51Construction-jute textiles -4.90Noakhali agriculture-Rajshahi agriculture 5.00

a/ The assumption of non-stationarity of the wage rates used herewas not rejected.

Source: Azam, Jean Paul (1994), "Labor Market Integration in Bangladesh."

24

Minimum wages

58. Are Bangladesh's minimum wage regulations responsible for the large wage differentials?Under the 1961 Minimum Wages Ordinance, the "National Minimum Wages Board" recommendsminimum wages for sectors or groups of workers referred to it by the Government. The ordinance statesthat minimum wages are to be set in sectors "for which no adequate machinery exists for the effectiveregulation of wages" if the Government finds that setting minima will be useful. "Adequate machinery"is interpreted to mean more than the existence of some form of collective bargaining. The Governmentmay consider the coverage of collective bargaining arrangements and their effectiveness and could decidethat although some enterprises had established collective bargaining procedures, the industry as a wholedid not have adequate machinery to efficiently determine wages. Currently 38 sectors are covered byminimum wage orders. They range from jute and cotton textiles to tea gardens and fishing trawlers. It isnot clear why those 38 sectors were referred to the Board, because many of them have collectivebargaining machinery that provides effective wage regulation.

59 The Board--which consists of a chairman, a representative of employers in the sector, aworkers' representative and an independent member--usually recommends three different minima forunskilled, semi-skilled and skilled workers, and in some cases, a fourth minimum is specified for highlyskilled workers. In addition to setting a minimum basic wage, the Board fixes allowances, whichrepresent around one-third of total minimum compensation in most sectors. The Re-Rolling Mills Orderof 1988 laid down six different occupational minima. The 1987 Cinema Order specifies four differentoccupational levels and four different classes of cinemas, with different minimum wage levels for each.For example, the lowest unskilled total compensation (basic wages plus allowances) in class D is 715 Tk.and 1024 Tk. in class A. The 1971 Order for Inland Water Transport specified only a minimum wage forunskilled workers.

60. Many of the minima are below market wages because they have not been adjusted for longperiods, during which market wages and prices were increasing. The law would appear to impose anobligation on the Board to reconsider minimum wage levels periodically, but this does not occur. Therecan be long gaps between consideration of minimum wages in a particular sector. Some Orders have notbeen revised at all--e.g., Jute Textile (Manufacturing) 1960, Cotton Textile 1964, Salt Crushing 1978,and Orders 26-38, originally introduced in 1983 or later. The Government may have taken the view thatalthough minimum wage intervention is no longer needed in these sectors, the minima fixed many yearsago and overtaken by inflation are best left undisturbed for fear of generating pressure to re-activatethem. This may be relevant when trade unions are pressing for a national minimum wage, since anymeasures to reduce existing coverage could generate protests. In the absence of any declaredGovernment policy regarding reference to the Board, it is likely that reference is the result of politicalpressures by trade unions.

61. As would be expected where under-employment is widespread and administrative capacityweak, it appears that the Government is unable to enforce minimum wage legislation in any sector. The1991 annual report of the Ministry of Labor and Manpower Inspectorate mentions only two investigationsunder the Minimum Wages Act, whereas in previous years, there had been 411 violations detected in1988, 518 in 1989 and 610 in 1990. These led to 13 prosecutions in 1990 and 3 in 1991. In many cases,employers and workers can avoid minimum wage regulations by declaring the worker an apprentice, eventhough he or she may be a fully trained operative.

62. If minimum wages are below market wages and government does not have the capacity toenforce them even when they are binding, one would not expect then to have much of an impact on actual

25wages paid. This conclusion is supported by econometric evidence which indicates that increases inminimum wages have had no significant impact on market wages (annex 2).

63. The conclusion that minimum wages in Bangladesh have little impact is consistent withsimilar work done on other countries. Household surveys indicate that 16 percent of Mexico's full-timemale workers and 66 percent of female workers were paid below the minimum wage in 1988. Industrialsurveys in 1986 in Morocco indicated that more than half of the firms paid their unskilled workers belowminimum wages. Few nations outside the OECD area have the administrative capacity to policeminimum wage regulations, especially in view of incentives for employers and workers to ignore theregulations when minimum wages are set so high as to discourage hiring. Both parties to an employmentcontract have an interest in avoiding a minimum wage so high that it lowers firms' profits andemployment levels. Employers clearly benefit from a lower minimum that reduces costs. Workers willalso gain as long as the wage being offered is above what they could earn in informal sector activities.

Public Sector Wage Policies

64. Despite a privatization program which began in the 1980s, Bangladesh's public sector is stillresponsible for around 50 percent of value added in manufacturing; and roughly one-third of all workerswho have regular formal jobs are employed in the public sector. Hence, public sector wage policies arebound to have an important impact on the competitiveness of a large part of Bangladesh's moderneconomy, as well as labor market repercussions.

65. Public enterprises (PEs) are covered by a single set of uniform pay scales and allowancesdetermined by the Government on the recommendation of the National Wages and ProductivityCommission (NWPC). The NWPC consists of representatives of workers, private employers andGovernment and meets at irregular intervals (between 5-8 years) when the Government decides thatpublic sector wages need to be revised. There are 16 different grades and pay scales that apply to allPEs, except that in the Jute and Textiles sectors Grade 8 provides a slightly higher minimum basic wagethan elsewhere. In principle, PE wages are set independently of civil service salaries. In practice,however, the Government has convened an NWPC whenever a Pay Commission was called to revise civilservants' salaries and the two commission's awards have been very similar.

66. Collective bargaining on wages and allowances is forbidden in PEs. The tri-partite nature ofthe NWPC does not provide a form of collective bargaining, because the Government retains the right toaccept, reject or modify its recommendations. For example, in July 1993, the Government announced aminimum basic wage for PEs of Tk. 950 instead of the Tk. 1000 recommended by the NWPC, andmaintained the housing allowance at 35 percent of basic pay instead of increasing it as recommended.Rather than a form of collective bargaining, the NWPC represents a form of consultation in a systemwhere pay and allowances are unilaterally determined by the Government as employer.

67. While the Government may make decisions unilaterally, it must respond to industrial andpolitical pressure by trade unions. These pressures are expressed politically and on occasions evenviolently because trade unions in the public sector are not allowed to express themselves throughcollective bargaining. The occasional opportunity to present a case to the NWPC creates expectationsthat are subsequently rebuffed, since workers are unable to negotiate the acceptance or implementation ofNWPC's recommendations. Such discussions as may take place are purely political and are not coveredby collective bargaining regulations. In fact, they usually take place between Central Government andTrade Union Federations that according to Bangladesh's Labor Law, are not allowed to represent workersin collective bargaining.

26

68. Under the present system of public sector wage determination, too much weight is given topolitical considerations and the PEs' capacity to pay is rarely even considered. As a result, over theyears, wage bills have increased steadily at a much faster rate than productivity in most PEs. This is oneof the reasons for PEs poor performance and weak financial situation.

Table 14: Average Labor Costs and Value of Output/WorkerIn Selected Public Enterprises

(Tk1000/year)

Avg. AnnualGrowth (%)

FY85 FY86 FY89 FY91 FY92 (85-92)

BTMCOutput/worker 93 81 96 125 151 7.2Avg. Labor Costs 13 17 20 25 26 10.6BJMCOutput/lworker 71 58 70 82 97 4.6Avg.labor Costs 14 19 25 30 40 16.0BSFICOutput/worker n.a. 59 81 189 162 18.3Avg.Labor Costs n.a. 18 25 44 45 16.4BSEC (wires)Output/Worker 516 574 707 521 693 4.3Avg.Labor Costs 18 37 47 26 31 8.6BSEC (Billet/Plate)Output/Worker 536 494 519 378 436 -2.9Avg.Labor Costs 29 37 56 59 45 6.5BCIC (Urea!Output/Worker 326 710 1642 1941 2209 31.4Avg. Labor Costs 24 50 82 94 118 25.8BCIC (TSP)Output/Worker 813 1605 2074 1674 2072 14.3Avg.Labor Costs 25 70 61 77 91 17.1

Source: Autonomous Bodies Wing, Ministry of Finance.

69. Table 14 shows the evolution of average labor costs and the value of output per worker,used as a proxy for productivity, between 1985 and 1992 in a sample of seven public sector units. Itindicates that in 5 of the 7 cases, increases in average labor costs occurred at a much faster rate thanincreases in the value of output per worker. For example, BSEC units producing steel billets/plates hadtheir average labor costs increase by more than 55 percent between 1]985 and 1992, while the value ofoutput per worker fell by 18.6 percent. In the case of TSP production under BCIC, average labor costsincreased by around 260 percent between 1985 and 1992, while the value of output per worker increasedby 150 percent.

70. Even in the two cases in the sample where productivity increased at a faster rate thanaverage labor costs, the improvement was not sufficient to achieve normal profits. A loss makingenterprise should aim to increase productivity at a much faster rate than labor costs until it reaches anormal return on its assets, and then have both variables grow at the same rate to maintain equilibrium.BSFIC had an operating deficit of Tk. 35.8 crores in FY86. Between FY86 and FY92, the value of

27output per worker increased at an average annual rate of 18.3 percent, while average labor costsincreased at an average annual rate of 16.4 percent. This improvement was not sufficient to render theenterprise profitable; it still had an operating deficit of Tk. 3.3 crores in FY92. In the case of ureaproduction under BCIC, the value of output per worker increased at an average annual rate of 31.4percent between FY85 and FY92, while average labor costs increased at an average annual rate of 25.8percent. This caused the unit's operating surplus to rise from Tk. 6.4 in crores in FY85 to Tk. 8.7 croresin FY92. However, the FY92 operating surplus only represents 1.5 percent of the value of total assets.

Box 5: The Problem of Public Sector Reforms

Like Bangladesh, many countries are currently re-thinking the role of their public sectors. The problemfacing most developing countries today is how to reduce the size of the public sector and enhance its quality. Thisusually implies retrenchment and privatization. Paradoxically, it also implies better pay for those who remainemployed in order to attract and retain high quality workers, and encourage good performance.

Most developing countries started out with policies similar to those of Bangladesh and paid public workerswages that were above their private sector counterparts. In Tanzania in 1971, a government worker earned about 14percent more than a private sector employee with the same schooling and work experience, while in Kenya in 1970the estimated differential ranged from 11 percent to 16 percent. The situation changed in the 1980s as dwindlingrevenues forced many governments to cut public expenditures, including public wages and salaries. In Zambia, thesalary of an undersecretary in 1986 was worth just 22 percent of its 1976 purchasing power. In El Salvador duringthe 1980s, real salaries of civil servants declined by 48 percent to 89 percent depending on rank.

While some reduction of government pay in most developing countries was warranted to eliminate rents andto reflect macroeconomic adjustments, such enormous declines undoubtedly had a negative impact on public sectorefficiency. Many governments, faced with a political tradeoff, hired more people rather than ensure adequate payand inputs. Low pay reduced the loyalty and dedication of public workers and was associated with an increase inrent-seeking. The lack of complementary inputs made it easier for individuals to excuse poor performance. 'Thequality of public workers further deteriorated because of salary compression, with the salaries of professional andskilled staff being allowed to fall more quickly in real terms than the earnings of those in lower salary grades.Tanzania provides an example. In 1969 the top public sector salary was dlirty times the lowest wage in governmentemployment. By mid-1980s this had collapsed to a ratio of six to one. This obviously had significant consequencesfor the hiring, retention and performance of senior public workers and the most skilled employees, and hence for theproductivity of those under their supervision.

71. In addition to wage structure distortions, the public sector's relatively low level of laborproductivity stems from three other factors. EiLt, past hiring practices have led to excessive increases in thesize of the labor force, and thus to decline in output per worker. Second, poor management practices haveoften resulted in an inefficient use of human resources. Ihird, aging capital stock, outdated technology andshortages in raw material and spare parts lower labor productivity.

72. The present centralized system of wage determination makes it impossible to link wages toproductivity. The calculations presented in table 14 indicate that changes in productivity vary significantlybetween different public sector units. For example, it seems that productivity increased in urea productionwhile labor productivity in the steel billets sector declined. Awarding the same wage increases to workers inboth sectors would mean that some workers' wages will increase less than is warranted by productivityimprovements, while others will see their wages rise at a much faster rate than their productivity.

2873. In addition to their impact on PE performance, public sector wage policies could also affectprivate sector wages, and thus employment and competitiveness in the rest of the economy. It is oftenargued that the public sector plays a leading role in less developed countries (LDC) urban labor markets,with all sectors following suit when public wages increase.

74. In Bangladesh, it is generally agreed that increases in public sector wages affect private wagesin sectors where PEs play a dominant role--for example, jute and cotton textiles. In those sectors, publicwage increases have led to wage demands from workers in private firms. Employers often gave in to thosedemands and their deficits were automatically financed by the nationalized commercial banks. In a sense,private firms in those sectors were managed the same way as public sector units. Wage data from the jutesector supports this conclusion. A comparison of wages paid by private mills (BJMA) and those paid bypublic mills (BJMC) during the period 1987-93 shows that they are highly correlated. The ratio of public toprivate earnings varied between 1.07 in 1987 and 0.94 in 1993, except for 1989-90, which witnessed sharpincreases in earnings, first in the private mills and then the public mills, as a result of a rise in the dearnessallowance and gratuity.

75. More important is whether public sector wage policies affect other "purely private" sectors,such as garments, cottage industries and construction. If Government wage policies have wide-rangingeffects on wages across the economy, particular care must be taken when setting public wages, especiallysince they may have a negative impact on the competitiveness and growth of some of the new export sectorswhose rapid development is key to the success of Bangladesh's export-led growth strategy. Econometrictests indicate that this may be the case. Wages in the public and private sectors appear to be closely linked(annex 3).

Policy Implications

76. Given that Bangladesh's present system of minimum wages has little or no impact on actualwages in the private sector and thus does not introduce distortions, the obvious policy conclusion is that theGovernment should not change it. In particular, pressures to increase minimum wages more often or toinstitute a "national minimum wage" must be resisted. At best, such policy changes would lead toinefficiencies and rents as employers and workers try to avoid the new regulations. At worst, they couldlead to higher wages but lower formal sector employment.

77. The Government of Bangladesh must discontinue the practice of centrally fixing wages forone-third of the country's formal labor force. Setting wages for such a large proportion of the labor forcebased on political rather than economic criteria will always lead to inefficiencies and inequities. The presentsituation where high public sector wage bills lead to poor PE performance and lack of competitiveness inpublic and private firms is clearly undesirable, because it discourages employment creation.

78. A reduction in the central Government's role in wage determination could be achieved in twoways. First, privatization programs need to be accelerated. Privatization automatically implies that theproportion of the labor force whose wages are politically determined will decline. A larger portion of wageswill reflect market conditions. Also, as the share of public enterprises in total formal employment declines,the impact of public sector policies on private wages will probably also fall. Second, the system of settingwages in enterprises that remain in the public domain needs to be changed and de-centralized. Wages needto reflect enterprise performance rather than political considerations (see section F).

29

F. INDUSTRIAL RELATIONS

79. Severance pay regulations that ensure flexibility and wage policies that avoid creating largedistortions will only result in higher investment, growth and employment if they are complemented byharmonious labor-management relations. There is a perception that the state of industrial relations inBangladesh is poor which especially discourages foreign investors. This view is not supported by datawhich shows the average number of work days lost due to strikes in the 1990s at around 30 thousand, nota very high number. However, those figures do not include the work days lost due to illegal strikes and"hartals," often nationwide, called by the National Labor Federation (SKOP). It is those massive strikesand the associated violence that give industrial relations in Bangladesh a bad name.

80. There are two special features of industrial relations in Bangladesh: the labor movement ishighly fragmented and politicized; and government plays a central role in industrial relations and mostbargaining occurs at the national (political) level between Central government and SKOP. How do thosefactors affect industrial relations? And how can the present situation be improved?

A Fragmented Labor Movement

81. Under Bangladesh's Industrial Relations Ordinance, 1969 (IRO), trade unions are requiredto be registered by the Government Registrar of Trade Unions. To be registered, a union's membershipmust include at least 30 percent of the workers in the establishment or group of establishments in which itis formed. The IRO thus prescribes the formation of trade unions at the firm level, although they havethe right to constitute federations. In establishments with more than one registered trade union, theRegistrar organizes a secret ballot every two years to determine the collective bargaining agent (CBA),who has the exclusive right to carry out collective bargaining on behalf of all the workers and to declarestrikes.

82. Bangladesh has a large number of unions that are linked to political parties but represent asmall fraction of the labor force. In 1992, there were 4,065 registered unions, with a total membership of1,648,783. Unions represent only some 3 percent of the labor force, or one-third of formal workers.Their average size was only 406 members; they are organized into roughly 700 union federations.Almost all federations have links (either temporary or permanent) with political parties. The three largestfederations, which together claim around 65 percent of total federation members, are the labor frontorganizations of the three major political parties.

83. A large number of weak, highly politicized unions appear to characterize industrial relationsin most of South Asia and may be due in part to a long tradition of ties between political parties andunions in the Sub-Continent. For example, as early as 1969 a report of the Indian National Commissionof Labor recognized that union multiplicity is a serious problem and suggested the election of a strongCBA as a remedy.

84. Union fragmentation and politicization may be a result of the predominance of publicenterprises in industries and the Government's direct involvement in wage setting. Because managementis centralized and collective bargaining restricted in the public sector, trade unions are unable to pursuetheir objectives through the normal machinery for industrial relations. In trying to achieve their goalsthey become dependent on political parties. The multiplicity of political parties is naturally reflected inthe structure of the labor movement, which is usually dominated by the unions affiliated with the rulingpolitical party of the moment, on the basis of relative efficiencies in satisfying labor demands.

30}

85. Under these circumstances, industrial relations involve interaction between political parties(particularly the ruling party) and trade unions, rather than interaction between enterprise managementand worker representatives. In countries like Malaysia and Singapore, where the weight of the privatesector is greater than in Bangladesh and management of public enterprises enjoys far greater autonomy,unions tend to be fewer but larger and are less politicized. The climate for industrial relations in thosecountries is much more conducive to investment than in Bangladesh.

86. Close collaboration between political parties and unions in South Asia has sometimes hadpositive effects, by helping achieve important political objectives. Unions played an imnportant role inIndia's independence movement. More recently, they were key partners in the coalition that restoreddemocracy in Bangladesh. There are also several international examples of the positive role thatpoliticized unions can play. "Solidarity" led Poland's pro-democracy movement, and "COSATU" playedan important role in creating a new South Africa.

87. However, excessive politicization and fragmentation of the labor movement has a negativeimpact on industrial relations and is likely to discourage investment. Inter-union rivalries often disruptpublic sector enterprises, as incidents in the Bangladesh Machine-Tool Factory in Joydebpur, textile millsof BTMC in Tongi, and Adamjee Jute Mills at Adamjee Nagar illustrate. Most of the time, the problemsstart when unions affiliated with one party try to overthrow the existing CBA affiliated with a differentparty. Enterprise management is rarely neutral and tends to support the union affiliated with whicheverparty is in power. This aggravates the violence, which often then spills over to surrounding privately-owned factories.

88. The formation of a SKOP in 1983 has not improved the situation. SKOP is not registeredunder the IRO and employers, as well as some trade unions, consider it an illegal organization. TheSKOP has mainly served as a vehicle for presenting union demands to Government. It has not beeneffective at coordinating union activities and limiting union rivalries at the plant level.

Government-Labor Relations

89. Excessive Government intervention in the labor market is made clear by its dealings withSKOP. The Government started negotiating directly with SKOP, excluding employers, in 1983. Sincethen, Government-SKOP relationships have often been tense with devastating effects for industrial peace.Partly because of the ineffectiveness of the established machinery for industrial relations, SKOP sought

to negotiate directly with the Government outside the existing legal framework. In order to forceGovernment to negotiate or to agree to concessions, it has often resorted to general strikes accompaniedby country-wide agitation, seriously damaging Bangladesh's industrial relations.

90. The first agreement between Government and SKOP was signed in May 1984 under thethreat of a 48-hour general strike. The agreement contained 29 clauses that covered a wide range ofeconomic and non-economic issues including: additional benefits for workers; constitution of a newWages and Productivity Commission; revision of labor legislation; protection for trade unions and theiroffice-bearers; constitution of an inquiry committee to look into alleged corruption in publicestablishments; and re-opening closed factories. The Government was slow in implementing theagreement, and SKOP claimed that by 1988, only 4 of 29 provisions were fully respected. Uniondemands for full implementation of the 1984 agreement have caused a great deal of strain in industrialrelations.

31191. Conflicts between Government and SKOP continue to threaten industrial peace inBangladesh. In December 1991, after a series of country-wide agitations, railway and road blockades anda threat of a 48-hour general strike, Government and SKOP signed a Memorandum of Understanding.Thereafter, labor claimed that Government was not implementing the understandings fast enough andthreatened another wave of agitations. A new agreement was reached in July 1992. However,implementation was still a problem and SKOP organized yet another wave of nation-wide general strikesin 1993 and 1994.

92. Failure of existing machinery for tripartite consultations, together with an excessiveGovernment role in the labor market, has led trade unions (represented by SKOP) to impose a bilateralprocess of negotiation with Government. Employers are excluded from this effective channel of nationallevel negotiation. It is no surprise that they have reacted strongly to SKOP-Government dealingsespecially since results affect private as well as public sector entelprises.

Policy Implications

93. Reducing central government's role in wage determination is key to improving industrialrelations. In addition, the issue of worker representation at the national level should be resolved, andprocedural reforms are needed to help de-politicize the union movement. Finally, Government needs toimprove present dispute resolution mechanism.

94. Collective bargaining is not functioning effectively due to public sector norms and policies.Bargaining takes place in private firms within the framework of the IRO and is practiced more widelythan in the public sector. However, the process of collective bargaining in the private sector is oftendisturbed by links between private and public sector unions and the influence exerted by the latter on theformer. Developing orderly collective bargaining in the public sector should help improve industrialrelations throughout the economy.

95. The system of setting public sector wages through a Wages and Productivity Commissioninstead of collective bargaining, is unsatisfactory. In industries where both the public and private sectorsoperate, private wages closely follow the trends set by public enterprises. This does not satisfy anybody:public sector workers feel alienated from a process in which they cannot effectively participate throughcollective bargaining, public sector management is frustrated by lack of autonomy: and private employersare unhappy because wages in many of their enterprises are in practice dictated by Govermnent.

96. The degree of government intervention in industrial relations in public enterprises variesfrom one country to another; but Bangladesh represents an extreme example of tight control. InMalaysia, wages in all public enterprises except public utilities are determined through collectivebargaining. However, the results are subject to approval by the relevant ministry. This is similar to theFrench system, where a distinction is mnade between public enterprises with special statutes and those runon a commercial basis. Employment conditions in the former type of enterprises are mainly deterninedthrough government decrees, while in the latter type of public enterprises employment terms aredetermined through collective bargaining. Unlike Malaysia, Government in France does not approve theresults of collective bargaining, but tries to influence them by issuing guidelines to negotiatorsrepresenting public enterprise managemnent.

32

Box 6: The Case for De-Centralized Collective Bargaining

The experience of several countries indicates that bargaining at the enterprise level can be an appropriateframework for achieving positive economic effects. At the other extreme are national-level bargaining, common inNorthern and Western Europe. Although these systems have fallen out of favor as Europeans grapple with risingunemployment and inflexible labor markets, they are associated with decades of positive industrial relations and goodwage and employment outcomes for European workers. But national-level bargaining requires that most workers becovered by union agreements. If they are not, as in nearly all developing countries, national agreements will benefitthe unionized sector at the expense of the unorganized and poorer groups in society. That has been the case in LatinAmerica, where centralized approaches to industrial relations have fared poorly.

Examples of how decentralized bargaining works and achieves good results can be found in the industrialcountries--Canada, the United States and Japan--and in the newly industrialized East Asian economies--Hong Kongand Korea after 1987. In Canada and the United States collective bargaining is mostly at the enterprise level--although there are some multi-employer arrangements. There is no tradition of involvement by central organizationsin bargaining in most sectors, and there are no central employer organizations. Enterprise unions have been heraldedas one of the pillars of industrial relations supporting Japan's economic achievements. In Japan, collective bargainingtakes place at the enterprise level, but confederations of labor unions play an important coordinating role during theannual bargaining exercises. In Hong Kong unions are also organized along enterprise lines, although as in Japanthere is some coordination by trade union federations. Hong Kong's unions operate in an enviromnent of strongcompetition in the output market, as well as in the labor market due to migration from China. This weakens theirbargaining position, and most unions are in the business of providing insurance and social assistance to theirmembers, rather than fighting for higher wages.

Recent developments in the Republic of Korea's system of industrial relations could provide useful lessons forother countries. Trade union freedom increased as part of the process of political liberalization which started in1987. The number of unions and their membership rose rapidly. In 1992, Korea had 7,676 firm-level unions andtwenty-one industrial federations. After a chaotic beginning, and very conflictual labor-management relations in theperiod 1987-89, collective bargaining is now becoming an established institutional arrangement in Korea. Bargainingis decentralized and is being conducted at the enterprise level, except in a few cases, such as textiles and banking,where bargaining occurs at the industry level or through a multi-employer arrangement. The Korean system ofindustrial relations is still evolving, but the tradition of enterprise-level bargaining has not been significantly modifiedin the new collective bargaining experiments since 1987, and a restructuring of bargaining patterns appears unlikelyin the near future.

97. Increasing the use of collective bargaining should be one goal of industrial relations policies.Expanding the private sector through the sale of public enterprises would have beneficial effects on thedevelopment of collective bargaining. The Government should also increasingly rely on de-centralizedbargaining for wage determination in remaining public enterprises, using a system similar to that in Malaysiaor France.

98. The issue of workers' representation at the national level needs to be resolved. Employersneed to be convinced of the benefits of recognizing a trade union alliance. Tripartite consultation would bemore useful if trade unions can present coordinated policy alternatives. The formation of SKOP representsan effort by national labor federations to achieve a certain degree of unity. If this or another alliancedevelops into a more mature organization, it could have beneficial effects on future industrial relations inBangladesh by helping to lower tensions between different federations and reducing inter-union rivalries atthe plant level that continue to threaten industrial peace.

3399. National trade union alliances (or congresses or confederations) play a positive role indeveloping sound industrial relations in many countries. They exist in most East Asian countries and arerecognized by governments and employers as partners in tripartite consultations. Examples of such unioncenters include: the National Trades Union Congress (NTUC) of Singapore; the Trade Union Congress ofthe Philippines (TUCP) and the Labor Advisory and Consultative Council (LACC) of the Philippines; theMalaysian Trades Union Congress (MTUC) and the Congress of Unions of Employees in the Public andCivil Services (CUEPACS) of Malaysia; the Labor Congress of Thailand (LCT), the National Congress ofThai Labor (NCTL) and the National Free Labor Union Congress (NFLUC) of Thailand.

100. Procedural reforms could also help resolve some of the problems caused by union multiplicityand politicization. Two changes in particular should be made to the IRO to strengthen CBAs and protectthem from minority union attacks that give rise to inter-union rivalries and violence. First, the list of unfairlabor practices should be expanded to include the employer's acquiescence in dealing with trade unions thatare not CBAs on matters on which the CBAs are entitled to negotiate. This would ensure that management(especially in public enterprises) will deal with the union elected by the workers to represent them, ratherthan the union affiliated with the party of management's choice. Second, the IRO should make clear that anon-CBA trade union has the right to represent its members only in individual grievances, which should besettled without resort to industrial action. At present, there is some ambiguity in the IRO with respect to therights and privileges of registered (non-CBA) trade unions, which promotes disputes between rival unions.

101. One of the Government's key roles in the area of industrial relations is to ensure a fair andefficient system for the resolution of disputes when they arise. The performance of Labor Courts is alsounsatisfactory. Delays in processing cases are far too long, with 4 or 5 years often required to settle a case.Delays in proceedings are often caused by the absence of the chairman or one of the two members, without

whom the hearing cannot take place. Sometimes delays are also due to the failure of one of the parties to thedispute, usually the defendant, to appear before the court. This problem occurs frequently, although the lawempowers Labor Courts to force the attendance of any person.

102. The Government should consider reforming the labor court system to make it an informal andefficient forum for dispute settlement, which appears to have been the original intention when the systemwas established. In October 1994, a national seminar sponsored by the International Labor Office took placein Dhaka to assess the problems of the system and to propose solutions. Representatives of Government,employers and workers participated, as well as members of the Labor Courts. Several recommendationsresulted from the meeting. In particular, the chairpersons and court members should receive training inindustrial relations, and adequate support services should be provided to assist the courts with theiradministrative work. Parties should be able to bring unfair labor practices and grievances to the courtswithout the prior authorization of the Director of Labor. Decisions should be appealable to the LaborAppellate Tribunal. The opinions of the lay members on the court should be given greater consideration. Itwas noted at the meeting that since the views of the lay members are often disregarded, they do not feelcompelled to attend, thus causing delays due to lack of quorum. Other recommendations included ensuringthat those bringing or defending cases could choose their own representative, providing the courts with thepower to grant injunctions and stay of proceedings, and enacting clear rules of procedure to be followed bythe courts.

103. Bangladesh could benefit from other country experiences in this area. In Singapore, theIndustrial Arbitration Court is empowered to prescribe its own procedures. It is not bound by the rules ofevidence, but may inform itself on any matter in such manner as it thinks fit. The Court is enjoined to actaccording to equity, good conscience and the substantial merits of the case, without regard to technicalitiesand legal forms. A party in proceedings before the Court is prohibited from being represented by an

34advocate, solicitor or paid agent, except in cases of contempt of court or in proceedings in which theAttorney-General has intervened

G. SPECIAL LABOR MARKET PROBLEMS: GENDER ISSUES AND CHILD LABOR

104. Faster growth and greater job opportunities will benefit most workers, but some may be leftbehind and require specific assistance. Two such groups come to mind: women who face labor marketdiscrimination and child laborers who are denied the opportunity for education and a better life. How cangovernment deal with those important problems?

Gender Issues

105. Discrimination Against Working Women. Women face egregious employmentdiscrimination in Bangladesh. Even within South Asia, Bangladesh has one of the lowest femaleeconomic activity rates (11 percent). In all sectors, working women in Bangladesh earn less thansimilarly situated men and are more severely affected by unemployment.

Table 15: International Comparisons ofActivity Rates ( percent) a/

Country Year Male Female

Bangladesh 1989 53.1 10.6India 1981 52.7 19.8Pakistan 1991 49.4 6.8Sri Lanka 1990 54.2 32.1Malaysia 1987 56.5 31.1Indonesia 1989 51.2 34.0Egypt 1989 47.6 6.1

a/ excludes unpaid family helpers.

Source: ILO, 1991

106. Employment discrimination in Bangladesh manifests itself perhaps most significantly in thefact that women have very little access to formal employment. More than 80 percent of all femaleworkers are unpaid family helpers, compared to only 20 percent of men. More than 90 percent of alleconomically active women are agricultural workers. These figures reflect barriers to better urbanemployment. Indeed, only 4.5 percent of working women (compared to 13.1 percent of men) havemanaged to enter the formal employment sector. Moreover, women have been unable to obtaingovernment jobs. Even though 15 percent of these formal jobs are theoretically "reserved" for them,women occupy only 6 percent of government positions. The situation is worse in public sectorcorporations, where only 2 percent of employees are female. In addition, nearly 1 in 5 women with auniversity degree is unemployed (compared to only 1 in 50 men), indicating that educated women aredenied equal access to higher paying jobs and careers that are normally available to university graduates.

107. Female workers are also paid lower wages than males. Micro studies indicate that femaleworkers in the rural and informal sectors earn between one- third to one-half of male wages. The

35situation in the urban formal sector appears much better, with the ratio of female to male wages rangingbetween 72 percent and 97 percent depending on skill level. However, econometric estimates of "earningfunctions" in the formal sector that control for differences in worker characteristics such as education andexperience indicate that in situations where a man and a woman have the same characteristics, the womanis paid a lower wage.

108. Government labor market interventions, such as minimum wages, maternity leave,restrictions of women's working at night, etc., appear to be largely ignored and to have very little impacton the welfare of female workers. Many entrepreneurs responding to surveys state that they do notprovide maternity benefits because workers do not ask for them. Surveys of the garments sector indicatethat most workers are unaware of the minimum wage and that 42 percent of women workers are paid lessthan the minimum. Where they are respected (due to union activity or foreign ownership), they lead tofewer women being employed, probably because they raise the cost of employing women, and are thuscounter productive. Most working women prefer a formal job in garments, even if they do not getmaternity leave or are paid below minimum wages, to being an unpaid helper in agriculture.

109. The experience of the garments sector where 80-85% percent of workers are women providesindications of the benefit to households and society of increased female labor force participation. Women'sparticipation in the formal labor market leads to important changes in intra-household relationships. A studyof female garments workers shows that 57 percent of them have control over their salaries--this is verydifferent from the situation in rural areas women are generally prevented from claiming a right over familyincome. Women employed as skilled workers and in better paying jobs tend to have a more importantdecision-making role. They spend a higher percentage of their income on education of family members thanmale workers in the same enterprise. Also, marriage prospects improve as girls earn an income, whichrelieves parents from a sense of urgency to "marry off" their girls. Thus, another positive impact of femaleemployment is to raise the age of marriage and reduce fertility.

110. International Experience. Labor market discrimination against women is a problem all overthe world. In almost all societies women have less power than men, receive less for their work, and haveless control over household resources, and in many countries they receive less education. Women receiveless pay per hour of work in market activities for two reasons. First, they have less access to better-payingjobs in the private and public formal sector, and are disproportionately represented among unpaid familyworkers and in insecure informal employment. In India only 7 percent of female workers are identified asemployees, and in Indonesia and Thailand only 21 percent and 30 percent. The rest are unpaid familyworkers or self employed. Second, when women are allowed to enter the same market as men, theycharacteristically are paid only 50 percent to 75 percent as much.

36

Box 7: Enterprise Characterista and the Hiring of Women

An econometric analysis of the factors affecting female employment was carried out using cross sectional enterprise data coDlected forthe industrial sector survey. A reduced form equation with the ratio of females to total employees as dependent variable was estimated. Thefollowing explanatory variables were included: percentage of production exported (EXPPC), size of employment (SE); square of SE (SQSE);value of assets (VA); square of VA (AQVA); age of enterprise (AGE), square of AGE (SQAGE); dummy for being located in an industrial area(LOCT; dummies for type of ownership (Foreign owned: FOWN; public enterprise: POWN, Joint Venture: JVEN); percentage of unskilledemployees in an enterprise (USPC); percentage of temporary workers (TMPPC); the existence of trade union (UN) and existence of coDlectivebargaining agent (CBA). Separate dummies were included for 13 sectors. The results of this analysis are presented in the table below.

Determinants of percent of Female Employment 1

Coefficient t-value

Total employment (SE) 0.009 9.64 *Square of SE (SQSE) 43xl04 -6.68 *Value of Asset (VA) -0.0028 -4.35 *Square of VA (SQVA) 84x109 2.35 *Age of enterprise (AGE) -0.41 -2.55 *Square of AGE (SQAGE) 0.0053 1.84 **Location dummy (LOCN) -5.15 -3.18 *

(Industrial area = 1)Percentage of value exported (EXPPC) 0.22 10.95 *

Dummy for foreign ownership (FOWN) -11.64 -3.86 *

Dummy for public enterprise (POWN) -24.46 -0.90Dummy for joint venture foreign and domestic (JVEN) -5.15 -3.28 *

Percentage of unskilled workers (USPC) -0.03 -1.36Percentage of temporary workers (TEMPPC) -0.03 -1.82Dunumy for labor union (UN) -2.45 -0.89Dummy for collective bargaining agent (CBA) -12.65 -3.97 *Age of owner (AGOWN) 0.09 2.17 *

iv Thirteen sectoral dummies and an intercept were included in the regression, but their coefficients are not reported here.** Implies significant at the 1 percent level.* Implies significant at the 5 percent level.

The above results indicate that trade unions, foreign ownership and public sector ownership have a negative impact on the hiring ofwomen, while export-oriented enterprises have a higher proportion of working women. Variables representing trade unionism and cotlectivebargaining have negative coefficients, although one is significant amd the other not. There are two possible explanations for this. First, tradeunions tend to keep wages high and thus discriminate against women. Second, reversing the direction of causation, it could be argued thatwomen tend not to join unions and therefore less union activity is observed in finns that hire more women. The coefficients on the dummy forforeign ownership and the dummy for joint ventures are negative and statisticaDly significant. It is possible that foreign firms and joint venturesfuly apply labor regulations, so that a woman worker costs them more than a man, as both maternity benefits and minimum wages are binding.Hence, they prefer hiring men. Export oriented enterprises have a higher proportion of working women. The coefficient on the share ofproduction that is exported is positive and statistically significant. This result is not simply driven by the garments idustry. An examination ofthe data indicates that exporters in the fish and seafood sectors, food processing and knitwear also employ a greater proportion of women thanother firms. This is probably because exporters face stiff competition and need to be very efficient. Hence, they cannot afford the cost ofdiscrimination amd hire the most efficient and lowest cost workers, regardless of their geider. The public sector is not promoting femaleemployment. Surprisingly, the coefficient on the dummy for public ownership is negative, although it is not statistically significant. Thisindicates that the Government's ostensible policy of expanding employment opportunities for women is not actually applied. If anything, itappears that it is more difficult for women to get a job in a public than a private enterprise.

37

111. In order to deal with discrimination, most countries have adopted legislation that establishequal pay for equal work and prohibit the exclusion of workers from certain activities based on gender.However, anti-discrimination standards have proven difficult to enforce. Even in industrial countries womenare still paid about 30 percent less than men for comparable work. The failure of these policies does notmean that governments have given up efforts to end discrimination in the workplace. Rather, there is anincreasing awareness that the focus of anti-discrimination policies should switch to improving access tomodem sector jobs--for example by the meritocratic hiring of more women in the public sector.

112. Nearly all countries have legislation fixing maternity leave and limiting night work for women.However, there is growing evidence that by increasing the cost of hiring women, such conditions depresstheir wages and discourages employers from hiring them. They also can have other perverse effects. Forexample, the ILO's committee of experts noted that many Austrian firns employ young women only onfixed-term contracts in order to avoid paying maternity benefits. Some Brazilian firms take more extremeprecautions, requiring women to produce medical certificates attesting to their sterilization before hiringthem.

113. Policy Implications. Women in Bangladesh need greater access and more productive jobs inthe formal sector. This is the best way to raise their income and status within the household. Increasinglabor demand by the formal sector is a prerequisite for higher female employment. The recommendationsurged by this report, aimed at increasing investment, growth and productive employment, should stronglyhelp women.

114. However, women will only be able to share fully in the benefits of faster growth if theprejudice against them is ended. The Government must play an active and strong role in stoppingdiscrimination. It should change its hiring practices to raise the proportion of women in the civil service,focusing on providing greater employment opportunities to well-educated women. In addition to setting anexample, the Government should start a public information campaign aimed at changing people's perceptionsof women's role in society, particularly in the workplace.

115. Government should also avoid introducing new regulations that increase the cost of femaleworkers and thereby discourage their employment, such as maternity leave policies and regulations againstwomen's working at night. Women have had good access to formal sector jobs in garments, whereregulations have been minimal. The kind of growth that has taken place in the garments industry, with itsgreat opportunities for women's formal employment, needs to be replicated in leather, electronics, processedfoods, toys, etc..

116. Some education is necessary to help women gain access to formal sector jobs. More than 60percent of Bangladesh's female population aged 5 and over have no education at all and are precluded fromholding formal jobs, since even the lowest skill categories require some education. Accordingly, theGovernment should continue its policy of stressing the importance of female education. Although theGovernment of course should encourage women's higher education, it should be noted that the mostimmediate effects will be obtained from primary education. The full benefits of the secondary and highereducation of women, which are needed to ensure access to formal employment, will not be obtained unlessand until the prejudice that limits access to high-paying formal sector jobs is drastically reduced.

38

Child Labor

117. The Problem in Bangladesh. As a result of poverty and the "unattractiveness" of theeducation system, child labor is widespread in Bangladesh. Although Bangladesh bans child labor, theregulations have proved unenforceable and more than 6 million children between the ages of 5 and 14 arelisted as economically active--20 percent of all children in this age group and 12 percent of the labor force.Most of these children are not mistreated in "sweat shops." Nearly two thirds of economically activechildren are unpaid helpers on family farms. However, some children do work long hours in unhealthyconditions in the urban infonnal market. And any child labor involves sacrificing the child's presenteducation, as well as future prospects that require education, for imnmediate economic gain.

118. International Experience. Child labor is a serious problem that is attracting worldwideattention. Nearly all countries have laws against child labor. Yet the ILO estimates that in 1990 there were78.5 million economically active children under the age of 15 in the world, nearly 8 million of them underthe age of ten. UNICEF's estimates of child labor are even higher. They calculate that in 1991 eightymillion children aged ten to fourteen were engaged in work so long or arduous that it interfered with theirdevelopment. As in Bangladesh, the vast majority of child workers all over the world are unpaid helpers onthe family farm. Most people would not condemn such work, provided the children attend school. Only asmall minority of child workers are engaged in casual wage work in urban areas. Nevertheless, for mostpeople the term "child labor" evokes disturbing images reminiscent of the "dark, satanic mills" of theindustrial revolution.

119. Cross-country evidence indicates that high rates of child labor are linked to poverty andunderdevelopment and to poor quality or availability of education. Children in poor families work becausethe family needs the extra income, especially if the parents' major source of revenue is uncertain. Asparents' incomes rise, they choose to send their children to school rather than to work. In Egypt a 10percent increase in mothers' wages results in a 15 percent decline in labor among children aged 12 to 14 anda 27 percent decline among children aged 6 to 11. In India the same increase would lower girls' labor forceparticipation by 9 to 10 percent. Low-cost schooling and an efficient education system which increasesearning potential would encourage households to send their children to school rather than to work.Malaysia's policies of favoring education for young Malays through scholarships and other subsidies, led toa decline in child labor. In Africa, the failure of the education system left parents with no choice but to sendtheir children to work.

120. Policy Implications. Legislation banning child labor has a symbolic value as an expression ofsociety's desire to eradicate this practice, but will be ineffective unless accompanied by measures to shift thebalance of incentives away from child labor and toward education. Such an agenda needs at least twoingredients: safety net protection for the poor and broader opportunity for quality education. Programs thatprovide income security for poor households, such as Food for Work or other public works programs, willhave beneficial effects on child labor. Similar results could flow from measures to reduce the cost of schoolattendance (subsidies, building schools closer to children homes, etc.) and improve the quality of education(changes in curricula, hiring of more and better teachers, etc.).

39121. The recent efforts aimed at eliminating child labor from the garments industry provide apositive example of what could be done to deal with this problem. They involve collaboration betweenemployers, workers, Government, NGOs and multi-lateral agencies (ILO and UNICEF). They involve aserious attempt at dealing with the root causes of child labor poverty and lack of education possibilities.Children who worked in governments are given stipends and are put in schools. Finally they involve aserious effort to enforce Bangladesh's child labor laws.

H. LABOR POLICIES AND INCOME DISTRIBUTION

122. How do the labor policies advocated here to enhance investment and growth affect the poorand the distribution of income? In order to respond to this question we analyze the impact of minimumwage, severance pay and public works programs on household incomes using a "computable generalequilibrium" model of Bangladesh's economy. An existing model, which had been used to study taxreform was updated and expanded to deal with labor market issues. This section presents keyconclusions. A key feature of the model is that it distinguishes between three types of workers: urbanformal, urban informal and rural. A more detailed description of the model, the scenarios and the resultsare in annex 4.

Minimum Wages

123. SKOP is pressuring the Government to institute a "national" minimum wage of Tk900/month to replace sectoral minima, and to make sure that it is respected. The Government has beenresisting this demand because it is worried that such a move would have a negative impact oncompetitiveness and employment. To study what the proposed policy would do to income distribution, abinding minimum wage was introduced to the urban formal sector of the model and raised by 50 percent.

124. An increase in binding minimum wages causes a deterioration in income distribution.Simulating a 50 percent rise in the minimum wage turns out to be detrimental to poverty alleviation, as itreduces wages to urban informal and rural workers to whom this minimum is not applied. The increasecauses 21.2 percent of the urban formal workers to become unemployed and move into the urbaninformal sector. This dramatic increase in the supply of urban informal labor reduces the nominal wage25.4 percent (the real wage declines 23.9 percent, given the slight decline in relative prices). Whiledemand for urban informal labor increases as producers substitute for the more expensive urban formallabor, the labor supply change dominates. The rural wage also declines slightly, by 4.5 percent (the realwage declines by 2.5 percent), as fewer job opportunities are available.

125. The minimum wage policy exacerbates differences in income distribution. While the realurban formal wage increases by 53 percent, there are fewer urban formal workers to benefit from it.Real wages and real income decline for other labor categories. Output in almost all sectors declines,reflecting the less efficient input use that results from the increased minimum wage. However, sectors thatuse rural labor intensively benefit from the slight decline in the rural wage due to the decline in demandfor rural labor in other sectors. The decline in available domestic goods makes imports more attractive,so the real exchange rate depreciates. This general equilibrium effect of the labor policy provides somebenefits to export sectors. The sectors that expand-jute (1.3 percent), cotton (6.4 percent), fisheries (0.8percent) and cloth (2.1 percent)--use rural labor intensively in production and depend heavily on exportmarkets.

40

Severance Pay

126. Severance pay is aimed at protecting vulnerable workers who may lose their jobs, but iffinanced in an inappropriate way, it could have negative consequences for the poor. The existingseverance pay system in Bangladesh requires employers to pay workers 2 months' wages per year ofservice in addition to Provident Fund contributions, which together adds up to a firing cost of about 25percent of a worker's wage earnings.

Box 8: Labor Policies and Trade Liberalization

The Govermnent has embarked on a liberalization programs to encourage private investment, expand exportsand integrate Bangladesh in the world economy. How can Bangladesh obtain the benefits from liberalization whileminimizing its short-run costs? Carrying out labor market reforms to increase flexibility, and hence, the ability oflabor to move from contracting to expanding sectors, could help raise the efficiency gains from trade liberalization.More importantly, it would shift the burden of adjustment away from poorer informal sector workers.

This point can be demonstrated with the help of our model. To explore the links between labor marketrigidities and trade liberalization, it is useful to consider trade liberalization with flexible "neoclassical" labor marketsas the reference point to indicate the efficient reallocation of resources and the consequent wage changes. The urbanformal labor intensive sectors have been traditionally heavily protected by tariffs. This suggests that in the absence ofgovernment policies, urban formal wages will decline following trade liberalization. In a version of the model inwhich urban formal wages are rigid, tariffs were eliminated and urban formal workers who become unemployedfollowing the policy shock move to the informal sector. It turns out that wage inflexibility has a significant impact onthe outcome of trade reform. In both scenarios, trade liberalization causes the currency to depreciate. As importsappear more attractive without the trade tax, consumers demand more. To maintain the current account balance, thereal exchange rate depreciates, promoting exports and offsetting some of the increased demand for imports.

Wage rigidities place the burden of adjustment on informal sector workers. When all wages are flexible, thereal urban formal wage declines by 2.2 percent while the real wage for rural and urban informal labor increases by6.5 percent and 4.3 percent, respectively. This pattern of wage changes reflects the initial distribution of tariffs,since the protected sectors use urban formal labor intensively. However, when the urban formal wage is rigid, theincome distribution changes dramatically. The real wage to urban formal workers increases by 9.2 percent,reflecting the decline in relative prices and the constant nominal wage. The real wage to urban informal workersdeclines by 3.4 percent. This reflects the increase in the supply of urban informal labor as 6 percent of the urbanformal workers become unemployed. Rural workers experience an increase in their real wage of 6 percent, slightlylower than when all wages are flexible. This reflects the links among labor markets. As the urban informal wagesdecline, some producers can substitute urban informal labor for rural labor.

In general, the more flexible ("neoclassical") labor market permits the expanding export sectors to expandmore. This is because the real wage can fall under the neoclassical specification, so that the jute sector, for instance,can become even more competitive following trade liberalization. It should be noted, however, that the differencesin output are not dramatic between the two scenarios. This is because the urban formal labor market in which wagerigidity is introduced is less than 10 percent of the labor force. All other wages in the economy are flexible.

127. To describe the impact of severance pay requirements for urban formal workers, a 25 percenttax on urban formnal labor was introduced to the model. In effect, urban formal workers become moreexpensive when firms pay a firing cost to shed labor. When wages are perfectly flexible, a wage tax to thefirm will be passed on to the worker in the form of lower wages. However, this is not an option when thereare other labor market rigidities. Since some formal wages in Bangladesh are rigid downwards due to public

41

sector wage and hiring policies and trade union activities, a wage tax was simulated under the assumption ofa floor on formal wages. Urban formal workers who become unemployed move to the urban informalsector.

128. The income distribution effects of severance pay are similar to those of the minimum wage,since both policies reduce the demand for urban formal labor. Following the wage tax, urban formalworkers become unemployed and move into the urban informal sector. The supply of urban informal laborincreases by 7.9 percent and nominal wages decline by 15.5 percent. Nominal wages for rural workersdecline 2.2 percent. Since the relative price declines slightly, the real wage declines by 14.6 percent forurban informal workers and 1.2 percent for rural workers. The urban formal workers who retain their jobsbenefit. While their nominal wage does not change due to wage rigidity, their real income increases 8.5percent, since they benefit from the lump sum government transfer of tax revenue and the slight pricedecline.

Food-For-Work

129. Food-for-Work is a labor market intervention aimed at the rural poor. It is a public worksprogram in rural areas targeted at the poorest workers. Targeting is achieved by keeping wages low andpaying them in the form of wheat, an inferior good in Bangladesh. The size of this program has declinedfrom 600 thousand tons of foodgrain in 1990 to 485 thousand in 1993. It is expected that it will shrinkfurther as donor food-aid is declining and the Government is reluctant to use its own resources to offset thedrop in foreign assistance.

130. The Food-for-Work program was modeled as a subsidy on wages paid to workers in ruralconstruction. For demonstration purposes, the subsidy was arbitrarily set at 70 percent of the wage. Twofinancing schemes were considered: a lump sum tax and foreign capital inflow. Given high rural under-employment, an endogenous supply of rural labor at a constant wage was assumed.

131. Food-for-Work, financed by Government tax revenue, has a clearly positive effect on incomedistribution. The model simulations indicate that the effect of the rural wage subsidy on income distributiondepends on the financing scheme. When the government finances the rural wage subsidy, the real wageincreases for rural and urban informal workers by 0.2 percent and 1.8 percent, respectively; the real wage tourban formal workers does not change.

132. Donor financed Food-for-Work has an ambiguous impact on wages but a stronger effect onrural employment. Urban informal and formal workers benefit from donor-financed food-for-workprograms, with real wages increasing 5.9 percent and 4.2 percent respectively. However, rural wagesdecline by 0.2 percent, reflecting the change in prices. The primary difference between the effect in theurban and rural sectors is the real exchange rate effect. When foreign capital inflow funds the wage subsidy,the real exchange rate appreciates. As a result, the price of domestic goods relative to imported goodsincreases and the consumer price index increases. Exports in all sectors decline. In sectors which exportheavily, such as jute, jute textiles and leather, output also declines. The real exchange rate change thataccompanies the inflow of foreign capital induces a larger increase in the demand for rural labor than doesthe wage subsidy itself. When the government finances the wage subsidy via a tax increase, ruralemployment increases by 3.8 percent and when foreign capital finances the wage subsidy, rural employmentincreases by 8.1 percent because the foreign inflow represents an additional source of demand in theeconomy.

42Policy Implications

133. The lesson from this general equilibrium analysis is clear: Bangladesh's policy makers mustbe careful not to provide high benefits to those who already have good modem-sector jobs ("the insiders") atthe expense of the masses of the poor who operate in the rural and urban informal sectors ("the outsiders").Concern with equity and with poverty alleviation would lead to interventions to expand the formal sector andprovide greater modern-job opportunities. This would normally require a reduction in the cost of formalsector workers, and less privileges and protection to insiders. Policies that benefit the poor will typically bethose that lead to higher labor demand in agriculture and off-farm rural activities. Bangladesh's successfulFood-For-Work program is an example of this type of policy.

Annex IPage 1 of 3

UNIT LABOR COST ANALYSIS

A. Algebraic Representation of Unit Labor Cost and its Decomposition

Unit labor cost (ULC) is defined as:

(1) ULC = ewL/Q

where e is the inverse of the exchange rate, w is the wage rate in domestic currency, L is the number ofworkers, and Q is the quantity of output. Labor productivity (prod) is defined as follows:

(2) prod = Q/L

substituting (2) into (1) ULC can be written as a function of the exchange rate, and wages andproductivity as follows:

(3) ULC -- ew/prod

Let oc be the opportunity cost of labor, and define the wage premium (prem) as:

(4) prem = w/oc

and let e* be the inverse of the PPP exchange rate. Using (4) and the definition of e*, (3) can berewritten:

(5) ULC = (e*oc) (e/e*) (prem) (1/prod)

Let ULC(1) and ULC(2) be the unit labor costs in Bangladesh and the comparator, respectively. DivideULC(1) by ULC(2) and take logs to get:

(6) log(ULC(1)/ULC(2)) = log (e*(l)oc(1)/e*(2)oc(2)) +log((e( 1 )/(e*( 1 ))/(e(2)/e *(2))) + log (prem( 1 )/prem(2)) - log(prod( 1)/prod(2))

In the text of the chapter, the term on the LHS of (6) was defined as the overall-gap, thefirst term on the RHS was defined as the opportunity cost-gap, the second term was defined as theexchange rate-gap, the sum of the first and second terms was defined as the macro-gap, the third termwas defined as the wage-gap, and the last term was defined as the productivity-gap.

B. The Data Used to Calculate Unit Labor Costs and its Limitations

While some aggregate data sources like the Annual Survey of Industries (ASI) or the Censusof Manufacturing Industries (CMI) contain data on wages, labor cost and cost of production formanufacturing sub-sectors, these sources were not used for the present comparison for a number ofreasons. First, there are problems of internal consistency in the CMI data for Bangladesh. Second, noup-to-date results are available, since the most recent year for which CMI results are available for

Annex 1Page 2 of 3

Bangladesh is 1989/90. The CMI and ASI results for other countries in South Asia are equally out ofdate. In the absence of industry-level survey or census data in any of the countries which could be usedas a reliable basis for estimating current labor costs and related indicators, there was no alternative tocommissioning some small surveys and case studies which would allow a more intensive analysis ofrelevant cost data for a sample of firms.

Six small surveys were commissioned in five countries--Bangladesh, West Bengal, India(excluding West Bengal), Pakistan, Sri Lanka and Vietnam--in order to compile firm-level product-specific estimates of unit labor cost, average wages and labor productivity in three key tradable sectors--garments, textiles and finished leather--which could be used to assess Bangladesh's labor costcompetitiveness vis-a-vis other countries. Two separate studies were commissioned for India: one forWest Bengal, and the other for India excluding West Bengal. Thus, the results for India in this chapteralways refer to India excluding West Bengal, although this is not always explicitly stated.

Data gathering was done primarily through survey of firms, based on responses to aquestionnaire and limited review of available accounts. Labor cost parameters were estimated for thefollowing products:

textiles: cotton yarn and gray (cotton) cloth;garments: men's shirt, men's jeans, and men's T-shirt; andtanning: finished and crust leather

The key indicators for which estimates were sought were:

Average physical productivity of laborAverage wage per employeeUnit labor cost (ULC), defined as labor cost per physical unit of outputUnit non-labor cost (UNLC), defined as non-labor cost per physical unit ofoutput.

The number of firms covered under the survey was rather small, with about 10 firms onaverage being surveyed for each product. The survey for Bangladesh was carried out for firms locatedmostly in Dhaka, and in the case of West Bengal only Calcutta-based firms were covered. The survey forVietnam was carried out for the Ho Chi Minh City area, which has much higher wages, and reportedlyhigher productivity levels than the northern part of Vietnam. It is possible that some bias was introducedin the results by such a regional concentration of the firms surveyed.

Labor productivity in public enterprises is proxied by value added per worker. One caveatwhich should be kept in mind is that the differences in domestic protection and pricing policies for theproducts of parastatals between countries is likely to have distorted this comparison. Ideally, value addedshould have been estimated at world market prices.

While estimates of unit labor costs for the selected products are likely to be quitecomparable between and within the countries, there are questions about the comparability of data onaverage wage and average production per employee for the garments sector. This is because although thewage cost and production data were required to be collected for an accounting year, so that annual wagesand productivity could be compared within and between countries, it is not certain whether the accounts

Annex IPage 3 of 3

and records of all firms in the garments sector permitted compilation of such annual data. If the dataprovided was not uniformly for one year for all the garments factories surveyed, estimates of averagewages and average productivity would not be comparable, but estimates of unit labor costs will continueto be robust.

Although some basic manufactured products (men's shirt, T-shirt, jeans, cotton yarn, graycloth and finished leather) were chosen for the purpose of the labor cost survey, there may have beenminor deviations from the product specifications, which could affect the comparability of unit labor costs.Many private firms had multiple product lines, the labor costs for the specific chosen products had to beestimated through apportionment of costs to products. The difficulties of cost apportionment of multi-product firms is an important caveat which should be kept in mind in using these cost estimates. Also,there have been some variations in labor contracting arrangements between countries. In the West Bengalgarment sector, for instance, job work done outside the firm is common and this can affect labor costsvis-a-vis firms with own machines and production lines.

Annex 2Page 1 of3

TESTING THE IMPACT OF MlMMM WAGES(Extracts from the background paper by Azam on Industrial Wage Dynamics)

The impact of minimum wages on actual wages paid in Bangladesh's industrial sector wastested using an econometric model for wage determination. Monthly wage data for 1980-93, coveringseven industrial sectors--cotton textiles, jute textiles, match, engineering, vegetable oil, small and cottageindustry, and construction--were used to estimate the model parameters. In the long run, wage rates shouldbe a function of output. Econometric tests indicated that real wages in cotton textiles, vegetable oil, jutetextile and engineering are co-integrated with the index of industrial production.

In the short run, real wage movements may deviate from this long-run relationship. An errorcorrection model of wage determination was used to analyze the short-term dynamics of wages in thoseseven sectors. The model was derived by following a selection process from general to simple, startingwith a fairly over-parameterized equation (with lags of up to 12 months), from which non significantvariables were progressively deleted to get a parsimonious representation. The model created in thismanner presents the percentage change in real wages as a function of their lagged values, the percentagechange in output (proxied by the industrial production index), the inflation rate and an error correctionterm. The estimated parameters are presented in the table over.

The results in the table provide some indication of the behavior of industrial wages and howthey are affected by changes in output and prices. The error correction mechanism is significant, with thecorrect sign in all the equations corresponding to sectors whose real wage rate is co-integrated with theindex of industrial production. This confirms that in the long run, increasing production will lead to higherreal wages. The short-term relationship between output and real wages is not as clear. In all sectors, eventhose where the co-integration assumption has been rejected, the short-run impact of industrial wagegrowth has been negative (although the elasticities are small in absolute terms), except in the constructionsector. This could imply that in the short run, output growth leads mainly to higher employment rather thanhigher wages, which is consistent with the view of Bangladesh as a labor surplus economy. In all sectors,the current rate of inflation has a significant negative impact on the real wage rate. However, some veryfast indexation mechanism is at work in the cotton textile, jute textile and small and cottage industrysectors, where the current elasticity is less than one. This means that workers in these sectors are able tovery quickly pass on part of the impact of inflation to their employers. In the other sectors, elasticity is notsignificantly different from one, suggesting that workers have no means of reacting to inflation in the veryshort run.

Tests for the impact of minimum wages were carried out by including dummy variables in theequations estimated above. The cotton textile sector had a minimum wage change in 1960 and 1986. Inthe jute sector, jute press and baling have also had minimum wage increases in 1960 and 1986. The latterincrease was not supposed to concern the weavers, whose wages are analyzed here. Nevertheless, acontagion effect is possible, whereby the weavers would obtain a wage increase when their colleagues inthe industry are granted a minimum wage increase. To test for an effect of these changes in the minimumwage, a dummy variable was included in the equation above, taking the value of zero up to 1986 and onethereafter. It turns out that this dummy variable has a positive and significant coefficient in both equations,implying that actual wages were affected by the change in the minima. It should be noted, however, thatthose two sectors have a very important public sector presence and that public sector wages were increased

Annex 2Page 2 of 3

around the same time that adjustment in minimum wages occurred. Hence, the dummy variable may becapturing the effect of the public sector wage adjustment rather than the change in the minimum wage.

Estimated Parameters of Wage Determination Model a/

Sector Lagged Waged Lagged Output Inflation CorrectionIncrease (%) (Increase %) II

Cotton Textile -0.23 -0.06 -0.49 -0.15(3.20) (2.16) (2.98) (3.41)

Jute Textile 0.16 -0.06 -0.53 -0.12(2.11) (1.83) (2.70) (3.20)

Match -0.33 -0.06 -1.02 --

(4.53) (1.91) (5.49)Engineering -0.21 0.10 -1.23 -0.12

(2.78) (3.09) (5.08) (3.37)Vegetable Oil -0.30 -0.07 -1.16 -0.10

(4.27) (2.01) (5.48) (2.76)Cottage Industry -0.29 -0.06 -0.74

(3.86) (1.73) (3.44)Construction -0.19 -- -1.35 --

(2.61) -- (4.59) --

a The dependent variable is the percentage change in the real wage rate.b/ The error correction terrns are the residuals from a long run level equation, where

the real wage is a function of output. The terms were not included for the match,cottage industry and construction sectors because the real wage rates were not co-integrated with output.

~' An additional three explanatory variables were included in this equation but theircoefficients are not reported here. These are the percentage change in real wageslagged two periods; the percentage change in output lagged two periods; and theinflation rate lagged 3 periods.

' An additional variable, the inflation rate lagged twelve periods, whose coefficient isnot reported here, was included in the equation.

Source: Azam, Jean Paul (1994), "Industrial Wage Dynamics in Bangladesh," mimeo.

Skepticism about the correct interpretation of the results for jute and cotton textiles isreinforced by the results for other sectors. The match sector's minimum wages changed in 1964, 1981and 1986. Therefore, dummy variables were included for 1981 and after and 1986 and after in thedynamic equation presented above. They did not turn out to be significant, with t-ratios below one,which indicates that minimum wage changes had little impact on that sector. Similarly, the minimumwage in the Automobile Workshops, which are part of the Engineering Sector, was changed in 1980 and1987, but the corresponding dummy variable was not significant when added to the sector's dynamicwage equation. In the Oil Mills and Vegetable Products industry, the minimum wage changed in 1970

Annex2Page 3 of 3

and 1980, so that the impact of changes could not be tested using a data set that starts in 1982. The tworemaining sectors are not covered by minimum wages.

Annex 3Page I of 2

EXPLORING THE RELATIONSHIP BETWEEN PRIVATE AND PUBLIC SECTORINDUSTRLAL WAGES

(Extracts from the background paper by Azam on industrial wage dynamics)

An econometric test of the relationship between public and private sector industrial wageswas carried out using the time series for wages in the seven urban sectors analyzed earlier to measure theimpact of minimum wages. Out of those seven sectors, jute and cotton textiles are dominated by the publicsector and the wage data reflects mainly wages in two large public corporations, BJMC and BTMC. Theimpact of public sector wages could thus be tested by analyzing the link between wages in those twosectors and wages in the match, engineering, vegetable oil, small and cottage industries and constructionsectors.

The econometric analysis was carried out in two stages. Eiri, a co-integration test wasconducted to see whether wages in those seven sectors were linked. The idea was to see if the public sectorwas an enclave or was connected to the rest of the country's labor markets. Once it was established that theindustrial labor markets were integrated, the second stage tested for causality to see whether the publicsector acts as a leader in the market, such that whenever it increases its wages the private sector mustfollow.

Stationarity and Co-Integration Testsof Industrial Wage Rates

Cotton Jute Match Engin. Veg. Oil Cottage Construc.

Cotton -0.78 -5.9* -3.0 -3.6* -3.9* -3.4* -5.4Jute -1.0 -2.7 -3.5* -3.5* -3.0 -5.1*Match -0.6 -4.2* -5.3* -2.1 -3.2*Engin. -0.7 -5.0* -2.6 -2.8Veg. Oil -0.2 -2.4 -3.5*Cottage -1.5 -4.7*Construc. -1.3

* significant at 10% or below

Sou rc: Azam, Jean Paul (1994)

The table presents co-integration analysis of wage rates for unskilled labor in the sevenindustrial sectors for which a long timne series for wages is available. It indicates that Bangladesh'sindustrial markets are fairly well integrated. The diagonal cells present the results of the stationarity tests,from which it is seen that the non-stationarity assumption is not rejected. The remaining cellsdemonstrate co-integration between the variable on the horizontal axis and that on the vertical axis, withan asterisk next to a figure indicating statistical significance. The results show that wages in cotton andjute textiles (the public sector representatives) are co-integrated with wages in all sectors except the matchindustry. That the match industry is integrated only with engineering, vegetable oil and construction is abit surprising, as it cannot be explained by a special feature of that sector's labor market, but may be areflection of the quality of the available wage data.

Annex 3Page 2 of 2

The fact that wages of unskilled workers in cotton textiles and jute are co-integrated withwages in small and cottage industries and in construction, respectively, is particularly important. Itindicates that far from being an enclave, labor markets in those public enterprise dominated sectors areconnected with sectors that are very close to the informal market. If changes in workers' wages in thejute or textile mills are associated with changes in the wages paid daily workers in construction or incottage industries, one would also expect them to be linked to wages in garments, leather and foodprocessing.

The short-run relationship between public and private wages was analyzed using theGranger-Sims causality test. The results of the analysis confirm the co-integration test's link betweenprivate and public sector wages. Public and private sector wages are jointly determined. As expected,wages in the public dominated sectors Granger-cause other wages in the economy. However, it also tumsout that those public sector wages are not autonomous. Some of the private wages, most notablyvegetable oil and engineering, Granger-cause wages in jute and cotton textiles. This could be becauseprivate sector representatives on the NWPC ensure that labor market conditions are at least partly takeninto account when wage awards are decided.

Annex 4Page 1 of 6

THE CGE MODEL(Extracts from the background paper by Devarajan et al.)

The CGE model in our analysis is an extension of the one developed by Dahl and Mitra(1991) to analyze tax policy in Bangladesh. There are 35 sectors in model. Each commodity is producedusing a Cobb-Douglas production function over two broad inputs--an aggregate of intermediate goods andan aggregate of value added. The intermediate aggregate is a Leontief composite of intermediate goods bysectors. The shares of intermediate goods in the aggregate are based on the 1986/87 input-outputproduction relationship. The value added component in the production function is a constant elasticity ofsubstitution aggregate of the value added by socioeconomic classes--rural, urban informal and urbanformal--that contribute to production. The elasticity of substitution is assumed to be 0.5 for eachproduction sector.

Imports are distinguishable from the domestic variety of each commodity. It is assumed thatdomestic goods are a poor substitutes for imports, and the elasticity of substitution is 0.8 for all sectors.Both households and producers consume the composite good which is a constant elasticity of substitution(CES) aggregate of the imported and the domestic variety.

There are different consumption patterns over the aggregate commodities for eachsocioeconomic class. Consumers allocate their income according to a linear expenditure consumptionfunction--each socioeconomic class has a given level of committed consumption expenditure. Theirremaining income, net of savings, is allocated to consumption goods according to fixed expenditure shares.

The government collects revenue from two sources: tariffs and an excise tax which isimposed on the sale of domestic commodities. It spends its income on goods and services, as well astransfer payments. All of the government expenditures are assumed to be fixed in real terms. Thedifference between government revenue and expenditure is government savings.

Using the small country assumption, Bangladesh has no influence on world prices for itsimports. Import demand functions depend on the relative price of domestic variety of the commodity to theexogenous world price. Export demand is mildly sensitive to domestic prices. The export demandelasticities are assumed to be -5 for all sectors except jute and tea for which the export supply elasticitiesare assumed to be -2. Foreign savings is the difference between the value of imports and the value ofexports, at world prices. The nominal exchange rate is the numeraire price and therefore is constant at one.

There are four sectors which sell output to meet investment demands: metal products, urbanhouse building, rural house building and other construction. Investment is a component of final demand.As this is a static model, investment does not enhance the capital stock. It is assumed that real investmentis constant at base year levels. Savings must adjust to ensure that the level of savings equals nominalinvestment. There are three sources of savings: consumer savings which is a fixed share of disposableincome (i.e. after accounting for the committed consumption) and these shares can vary by socioeconomicclass; government savings which is the difference between tax revenue and both transfer payments andfixed real spending; and foreign savings which is the difference between the value of imports and the valueof exports at world prices.

Two different closure rules to insure that savings equal investment are considered. Theforeign savings can be endogenous, indicating that any shortage in savings is financed externally as capital

Annex 4Page 2 of 6

flows into Bangladesh. Alternatively, foreign savings can remain constant and the savings rate for oneconsumer can vary to insure that savings equals investment. One can interpret this as a lump sum tax onthe consumer when savings from other sources, such as the government, decline.

The labor market specification of the model is expanded to allow us to evaluate labor policiesand their interaction with trade liberalization. The following variations in the labor market are considered:

(1) neoclassical labor markets: The labor supply of each class is constant in the base level. Thewage adjusts such that each labor market clears; and

(2) minimum wage in the urban formal labor market: The wage in the urban formal labor marketis exogenous. It may be held constant at the base level or the minimum wage can change asan experiment. Since producers cannot bid down wages, there can be unemployment assectors contract and others do not expand to absorb the displaced workers. The urban formalworkers who are unemployed move into the urban informal sector. As the supply of urbaninformal labor increases, their wage declines. The demand for urban informal workersresponds to the wage in the urban formal sector since the two inputs are substitutes inproduction.

(3) urban safety nets:

(a) contracting sectors compensate laid-off workers: When a sector contracts and shedsworkers, it compensates the laid-off workers by paying them 25% of their base wage.The sectors that expand do not contribute to the payments;

(b) government financed by wage tax: the government imposes a wage tax on all sectorsthat employ urban formal workers. The revenue collected is paid to the urban formalworkers who are displaced as their sector sheds labor. The wage tax is calculated suchthat all displaced workers receive 25% of their base wage; and

(c) government financed by a VAT: the government imposes a value added tax on the non-agricultural sectors to generate revenue for urban safety nets.

Rural safety nets are considered in a model with full employment and in a version of themodel with aggregate unemployment in the urban formal sector. The government subsidizes the wage torural workers in the rural house-building and the other construction sector. This can be financed internally,by increasing taxes, or externally as foreign savings increases.

Annex 4Page 3 of 6

Details of Simulation Results Using CGE

Table 1: Scenarios to evaluate the effects of labor policies

SCENARIO DESCRIPTION

Scenario 1 Increase the minimum wage to the urban formal workers by 50%. Urbanformal workers who become unemployed move into the urban informal sector

Scenario 2A Wage subsidy of 70% to rural workers in the rural house building and otherconstruction sectors. Allow an endogenous supply of rural workers and holdthe rural wage constant. Finance the subsidy by a lump sum tax on urbanformal workers.

Scenario 2B Wage subsidy of 70% to rural workers in the rural house building and otherconstruction sectors. Allow an endogenous supply of rural workers and holdthe rural wage constant. Finance the subsidy by foreign capital inflow.

Scenario 3A Wage tax of 25 % on urban formal workers.

Scenario 3B Value-Added tax on all non-agricultural production. The revenue generatedfrom the value added tax equals the revenue generated by the wage tax.

Annex 4Page 4 of 6

Scenarios to evaluate the interaction between trade liberalization and labor market policies

SCENARIO DESCRIPTION

Scenario 4 Trade liberalization with neoclassical labor markets.

Scenario 5 Trade liberalization with a minimum wage paid to urban formal workers.Urban formal workers who become unemployed move into the urbaninformal labor market.

Scenario 6A Trade liberalization with urban formal safety nets: contracting sectorsprovide the severance pay of 25 % of base wages. No minimum wage.

Scenario 6B Trade liberalization with urban formal safety nets: wage tax on all urbanformal workers to finance the severance pay of 25 % of base wages. Nominimum wage.

Scenario 7A Trade liberalization with urban formal safety nets: contracting sectors providethe severance pay of 25 % of base wages. Minimum wage for urban formalworkers. Those urban formal workers who become unemployed move intothe urban informal labor market.

Scenario 7B Trade liberalization with urban formal safety nets: wage tax on all urbanformal workers to finance the severance pay of 25 % of base wages.Minimum wage for urban formal workers. Those urban formal workers whobecome unemployed move into the urban informal labor market.

Scenario 8A Trade liberalization and a wage subsidy of 70% to rural workers in the ruralhouse building and other construction sectors. Allow an endogenous supplyof rural workers and hold the rural wage constant. Finance the subsidy by alump sum tax on urban formal workers.

Scenario 8B Wage subsidy of 70% to rural workers in the rural house building and otherconstruction sectors. Allow an endogenous supply of rural workers and holdthe rural wage constant. Finance the subsidy by foreign capital inflow.

Table 2: Impact Of Labor Market Policies

Percent change from SCENARIO 1: SCENARIO 2A: SCENARIO 2B: SCENARIO 3A: SCENARIO 3B:base level

Minimum Wage Rural wage subsidy, Rural wage subsidy, Urban safety net, 25% Urban safety net, value_ _ govt. financing foreign financing urban formal wage tax added tax

Real Wage

Rural -2.5 0.2 -0.2 -1.2 -2.1

Urban Informal -23.9 1.8 5.9 -14.6 -5.6

Urban Formal 53.0 0.0 4.2 1.0 0.4

Nominal Wage _

Rural -4.5 0.0 0.0 -2.2 -2.4

Urban Informal -25.4 1.6 6.2 -15.5 -5.9

Urban Formal 50.0 -0.3 4.4 0.0 0.0

Real Income

Rural -2.4 3.8 7.4 -1.2 -2.0

Urban Informal -12.8 1.7 5.6 -7.5 -4.2

Urban Formal 18.1 0.0 3.7 9.6 -1.3

Unemployment 21.2 0.0 0.0 12.2 1.8

Workers Displaced 21.2 0.8 1.4 12.2 1.9

Output

Sugar -2.1 2.1 5.8 -1.1 -0.5

Edible Oils -1.2 2.6 7.0 -0.6 -0.2

Basic Metals -5.0 -8.1 -7.6 -2.7 -0.5

Jute 1.3 0.2 -3.1 1.0 1.8

Cloth 2.1 1.6 3.0 1.4 -1.4

Leather -6.8 1.0 -0.6 -4.1 1.3

Table 3: Trade Liberalization and Labor Market Polices

SCENARIO SCENARIO SCENARIO SCENARIO SCENARIO SCENARIO SCENARIO SCENARIO 8B:4: 5: 6A: 6B: 7A: 7B: 8A:

Rural wage Rural wageCompetitive Minimum Urban safety Urban safety Urban safety Urban safety subsidy, govt. subsidy, foreignlabor market wage nets: firing nets: wage nets: firing nets: wage financed financed

cost tax cost, minimum tax, minimumwage wage

Real Wage

Rural 6.5 6.0 6.3 6.5 6.9 5.9 7.4 4.4

Urban Informal 4.3 -3.4 0.0 4.3 3.8 -5.0 2.1 30.8

Urban Formal -2.2 9.2 2.1 -3.3 9.7 9.2 -5.6 25.3

Nominal Wage

Rural -2.2 -2.9 -2.5 -2.2 -2.6 -3.1 0.0 0.0

Urban Informal -4.1 -11.5 -8.3 -4.1 -5.4 -13.0 -5.0 25.2

Urban Formal -10.2 0.0 -6.3 -11.2 0.0 0.0 -12.1 20.1

Real Income

Rural 6.1 5.7 6.0 6.1 6.5 5.6 7.5 34.8

Urban Informal 4.1 0.3 0.0 4.1 3.7 -0.5 2.0 29.3

Urban Formal -2.0 2.3 2.6 -2.0 8.4 3.3 -4.9 22.3

Unemployment 0.0 6.0 0.0 0.0 0.3 7.3 0.0 0.0

Workers Displaced 4.0 8.0 3.2 4.0 2.9 9.0 4.5 7.3

Output ._l

Sugar -9.9 -10.4 -10.0 -9.9 -8.8 -10.5 -10.2 15.4

Edible Oil -11.0 -11.3 -11.1 -11.0 -9.6 -11.4 -11.5 17.2

Basic Metals -6.7 -7.9 -7.7 -6.7 -6.5 -8.2 -15.0 -11.5

Jute 26.0 26.4 26.8 26.0 25.8 26.4 22.7 -2.7

Cloth 71.1 73.4 72.9 71.1 80.1 73.9 67.2 63.8

Leather 31.7 28.3 30.7 31.7 28.9 27.5 25.6 12.0

Annex 5Page 1 of 7

BIBLIOGRAPEC NOTES

Section B

Azam, Jean Paul (1994), "Labor Market Integration in Bangladesh"

Cameron, John (1993), "The Labor Force and its Use in Bangladesh in the 1990s"

Prennuchi, Giovanna (1994), "Rural Employment and Earnings in Bangladesh"

It also benefited from a large number of studies describing Bangladesh's labor markets and the process ofwage detennination, including:

Amin, A.T.M. Nurul (1986), "Urban Informal Sector: Employment Potentials and Problems," in R.Islam and M. Muqtada (eds.), Bangladesh: Selected Issues in Employment and Development,ARTEP/lLO, New Delhi.

Azam, Jean Paul (1993), "The Impact of Floods on Rural Wages in Bangladesh," BangladeshDevelopment Studies.

Boyce J. and M. Ravallion (1990), "A Dynamic Econometric Model of Agricultural Wage Determinationin Bangladesh," Oxford Bulletin of Economics and Statistics.

Hiroshima, S. and M. Muqtada (eds.) (1986), Hired Labor and Rural Labor Markets in Asian Studies onFarm-Level Data, ARTEP/ILO, Bangkok.

Majumder, Paul, Simeen Mahmud and Rita Afsar (1992), "Dynamism in Squatter Life: An Insight fromAgargaon Squatter Settlement," Bangladesh Institute of Development Studies, Dhaka, mimeo.

Rahman A. and R. Islam (1988), "Labor Use in Rural Bangladesh: An Empirical Analysis," BangladeshDevelopment Studies.

Rahman, Hossein Zillur and Mahbub Hossein (eds.) (1992), Rethinking Rural Poverty, BangladeshInstitute of Development Studies, Dhaka.

Rahman, R.I. (1993), "Employment, Unemployment and Poverty in the Bangladesh Economy," paperpresented to the National Workshop on Challenges of Rural Poverty by the Year 2010, Dhaka.

Section C

The analysis in this section was carried out by Syed Nizamuddin and Dipak Mazumdar usingsurvey data. The section also used a background paper:

Kawaura, Akihiko (1994), "Employment/Wage Comparison Between Garment and Textile Sectors,"World Bank Mimeo.

Annex 5Page 2 of 7

Other work used include:

Baskar, V. and Mushtaq Khan (1995), "Privatization and Employment: A Study of the Jute Industry inBangladesh," The American Economic Review, Vol. 85, No. 1.

Datt, Gaurav (1993), "Purchasing Power Parity and Poverty: A Note on Alternative Estimates for Indiaand China," World Bank Mimeo.

Edwards, Sebastian (1989), "Real Exchange Rates, Devaluation and Adjustment: Exchange Rate Policyin Developing Countries," Cambridge, MA: MIT Press.

Lindauer, David and Ann Velenchick (1994), "Can African Labor Compete" in D. Lindauer and M.Roemer (eds.), Asia and Africa: Legacies and Opportunities in Development (ICEG).

Mazumdar, D., L. Travis and C. Trikha (1992), "International Comparisons and GNP Measures,"Economic and Political Weekly, October 24-31: 2367-2371.

Mazumdar, Dipak (1989), "Microeconomic Issues of Labor Markets in Developing Countries: Analysisand Policy Implications," EDI Seminar Paper No. 40.

Summers, Robert and Alan Heston (1991), "The Penn World Tables (Mark 5): An Expanded Set ofInternational Comparisons, 1950-88, Quarterly Journal of Economics. 106: 327-368.

Section D

This section relied on two background papers:

Rahman, Masihur (1994), "Structural Adjustment and Employment in Bangladesh: An Assessment"

Robinson, Derek (1993), "Labor Market Issues in Bangladesh"

Other work used here include:

Blank, M. Rebecca and Richard B. Freeman (1993), "Evaluating the Connection Between SocialProtection and Economic Flexibility," NBER Working Paper No. 4338, National Bureau of EconomicResearch, Inc., Cambridge.

Carvalho, Soniya (1994), "Social Funds: Guidelines for Design and Implementation," HROWP 34, TheWorld Bank.

Fallon, Peter R. and Rober E.B. Lucas (1991), "The Impact of Changes in Job Security Regulations inIndia and Zimbabwe," The World Bank Economic Review, Vol. 5, No. 3, pp. 395-413, Washington,D.C.

Fizbein, Ariel (1992), "Labor Retrenchment and Redundancy Compensation in State Owned Enterprises:The Case of Sri Lanka," mimeo.

Annex 5Page 3 of 7

Freeman, Richard B. (1992), "Labor Market Institutions and Policies: Help or Hindrance to EconomicDevelopment?," Proceeding of the World Bank Annual Conference on Development Economics, pp. 117-145, Washington, D.C.

Rama, Martin (1994), "Flexibility in Sri Lanka's Labor Market," Policy Research Working Paper 1262,World Bank, Policy Research Department, Washington D.C.

Schaffner, Julie Anderson (1993), "Rural Labor Legislation and Permanent Agricultural Employment inNortheastern Brazil," World Development, Vol. 2 no. 5.

von Braun, Joachim (1994), "Employment for Poverty Reduction and Food Security," IntemationalFoodPolicy Research Institute, proceedings of an International Policy Workshop held at Airlile House,Virginia, October.

Section E

This section relied on these background papers:

Azam, Jean Paul (1994), "Industrial Wage Dynamics in Bangladesh"

Rashid, Faham (1989), "Correlation Between Public and Private Sector Wages in the Jute Industry: AnEvaluation"

Robinson, Derek (1993), "Labor Market Issues in Bangladesh"

Other work which was used include:

Anderson K., N. Hossain and G. Sahota (1991), "The Effect of Labor Laws and Labor Practices onEmployment and Industrialization in Bangladesh," The Bangladesh Development Studies.

Asaad, Ragui (1994), "The Effects of Public Sector Hiring and Compensation Policies on the EgyptianLabor Market." Prepared for the World Bank Development Report 1995, World Bank, Washington,D.C.

Azam, Jean-Paul (1994), "Effects of Minimum Wages in Developing Countries: An Exploration,"Universite de' Auvergne, CERDI, Clermont-Ferrand, and CSAE, Oxford.

Banerjee, A., J. Dolado, J.W. Galbaith and D.F. Hendry (1993), "Co-Integration, Error Correction, andthe Economic Analysis of Non-Stationary Data," Oxford University Press, Oxford.

Berg, E. (1969), "Wage Structure in Less Developed Countries," in A.D. Smith (ed.), Wage PolicyIssues in Economic Development, pp. 294-337, Macmillan: London.

Card, David and Alan B. Krueger (1994), "Minimum Wages and Employment: A Case Study of theFast-Food Industry in New Jersey and Pennsylvania," The American Economic Review, Vol. 84, No. 4.

AnnePage 4 of 7

ILO (International Labor Office) (1994), "Defending Values, Promoting Change: Social Justice in aGlobal Economy: An ILO Agenda," Report of the Director-General (Part I), International LaborConference, 81st Session, Geneva.

Morrisson, Christian, Henri-Bernard Solignac Lecomte and Xavier Oudin (1994), "Labor Law and LaborMarket," Micro-Enterprises and the Institutional Framework in Developinr Countries.

Robinson, Derek (1994), "Do Standards for the Workplace Help or Hurt?" University of Oxford,Institute of Economics and Statistics, Oxford.

Squire, Lyn and Sethaput Suthiwart-Naruepet (1995), "The Impact of Labor Market Regulations," WorldBank, Policy Research Department, Washington, D.C.

SectionF

This section is based on a background paper:

Ozaki, Muneto (1994), "Industrial Relations," ILO, Geneva.

Other work used include:

Basin, Jeremy (1991), "Striking Back: A History of COSATU," Johannesburg: Ravan Press andPretoria: Sigma Press (Pty) Ltd.

Bhattacherjee, Debashish (1987), "Union-Type Effects on Bargaining Outcomes in IndianManufacturing," British Journal of Industrial Relations.

Bhattacherjee, Debashish and Tamal Datta Chaudhuri (1994), "Unions, Wages and Labor Markets inIndian Industry, 1960-86," Journal of Development Studies. Vol. 30, No. 2, pp. 443465. Frank Cass,London.

Calmfors, Lars and John Driffil (1988), "Centralization of Wage Bargaining and MacroeconomicPerformance," Economic Policy.

Freeman, Richard B., and James L. Medoff (1984), "What do Unions Do?" New York: Basic Books,Inc.

Ghanem, Hafez, Shantayanan Devarajan and Ken Thierfelder (1995), "Trade Reform and Labor Unions:A General-Equilibrium Analysis Applied to Bangladesh and Indonesia," Washington, D.C.

Government of India, Ministry of Labor (1969), "Report of the National Commission of Labor," NewDelhi.

Harvard Institute for International Development (1990), "An Economic Analysis of Industrial Disputes inBangladesh," Working Paper No. 18, HIID/ESEPP Project, Bangladesh Planning Commission.

Annex 5Page 5 of 7

Katz, Harry C., Sarosh Kuruvilla and Lowell Turner (1993), "Trade Unions and Collective Bargaining,"Policy Research Working Paper 1099, World Bank, Education and Social Policy Department,Washington, D.C.

Kemp, Welch A. (translated and introduced by A. Kemp-Welch) (1993), "The Birth of Solidarity: TheGdarisk Negotiations, 1980," St. Martin Press, New York.

Krause, Lawrence B. and Fun-Koo Park, eds. (1993), "Social Issues in Korea: Korean and AmericanPerspectives," Korea Development Institute, Seoul.

Mondal, Abdul Hye (1992), "Trade Unionism, Wages and Labor Productivity in the ManufacturingSector of Bangladesh," Research Report No. 133. Bangladesh Institute of Development Studies, Dhaka.

Pencavel, John (1991), "Labor Markets under Trade Unionism: Employment, Wages, and Hours," BasiBlackwell, Inc. Cambridge.

Pencavel, John (1995), "The Role of Labor Unions in Fostering Economic Development," StanfordUniversity, Department of Economics.

Rahman, Masihur (1994), "Structural Adjustment, Support and Workers: Public Issues and Choice forBangladesh," mimeo.

SrectionG

There was one background paper for this section:

Rahman, Rushidan Islam (1994), "Formal Sector Employment Among Women in Bangladesh and GenderComposition of Industrial Workers." Bangladesh Institute of Development Studies, Dhaka.

Other work used in the preparation of this section include:

Ahmad, Alia and M. A. Quasem (1991), "Child Labor in Bangladesh," Al-Afsei Press, Dhaka.

Behram, Jere R. and Zheng Zhang (1994), "Women's Employment: Patterns Across Countries and OverTime." University of Pennsylvania, Philadelphia.

Bergmann, Barbara (1989), "Does the Market for Women's Labor Need Fixing?" Journal of EconomicPerspectives, Vol. 3, No. 1.

Bonnet, M. (1993), "Child Labor in Africa." International Labor Review, 132 (3).

Boyden, Jo and William Myers (1994), "Exploring Alternative Approaches to Combating Child Labor:Case Studies from Developing Countries." Innocent Occasional Papers, Child Rights Series, No. 8,International Labor Office, Geneva and the UNICEF International Child Development Centre, Florence.

Boyden, Jocelyn (1988). "Working Children in Lima, Peru." In Assefa Bequele and Jocelyn BoydenCombating Child Labor. Geneva: International Labor Office.

Annex 5Page 6 of 7

Chaudhuri, Salma and Pratima Paul-Majumder (1991), "The Conditions of Garment Workers inBangladesh--An Appraisal," Mimeo, Bangladesh Institute of Development Studies, Dhaka.

Chowdhury, N. (1986), "Revaluation of Women's Work in Bangladesh," The Bangladesh Journal ofAgricultural Economics.

Goldin, Claudia (1990), "Understanding the Gender Gap: an Economic History of American Women."Oxford University Press.

Grootaert, Christiaan and Ravi Kanbur (1994), "Child Labor." Background Paper for the WorldDevelopment Report 1995, World Bank, Washington D.C.

Hamid S. (1992), "Gender Dimensions of Poverty" in H. Rahman and M. Hossain (eds.) "RethinkingRural Poverty." Mimeo, Bangladesh Institute of Development Studies.

Horton, Susan, ed. (1994), "Women and Industrialization in Asia," University of Toronto, Institute forPolicy Analysis, Toronto.

Nazme, Sabina (1990), "Male-Female Wage Differentials in Bangladesh: The Case of SelectedManufacturing Industries in Dhaka City," Mimeo, Department of Economics, University of Dhaka.

Patrinos, Harry Anthony and George Psacharopoulos (1995), "Schooling and Non-Schooling Activities ofPeruvian Youth: Indigenous Background, Family Composition and Child Labor." World Bank,Washington, D.C.

Patrinos, Harry Anthony (1994), "The Costs of Discrimination in Latin America." HRO Working Paper45, World Bank, Human Resources Development and Operations Policy, Washington D.C.

Rahman, Rushidan Islam (1933), "Employment and Occupational Mobility Among Women inManufacturing Industries of Dhaka City: Finding from an Enterprise Level Survey," ARTEP/ILO, NewDelhi.

Rahiman, Rushidan Islam (1986), "Impact of Grameen Bank on the Situation of Poor Rural Women,"Working Paper No. 1, Grameen Bank Evaluation Project, Bangladesh Institute of Development Studies,Dhaka.

Siddiqi, Faraaz and Harry Anthony Patrinos (1994), "Child Labor Issues: Issues, Causes andInterventions," World Bank, Washington D.C.

Westergaard, K. (1983), "Pauperization and Rural Women in Bangladesh: A Case Study," BangladeshAcademy of Rural Development, Comilla.

Section H

This section relies heavily on a background paper:

Devarajan, Shanta, H. Ghanem and K. Thierfelder (1994), "Labor Market Policies, StructuralAdjustment and the Distribution of Income in Bangladesh."

ALex 5Page 7 of 7

A description of the model used and its application to tax policy is in:

Dahl, Henduk and Pradeep Mitra (1991), "Applying Tax Policy Models in Country Economic Work:Bangladesh, China and India," World Bank Economic Review.

I

IMAGING

Report No: 13799 BDType: SR