Report No. 32650-BA Bosnia and Herzegovina Labor Market Update the Role of Industrial Relations December 2005 Human Development Sector Unit Europe and Central Asia Region Document of the World Bank ACKNOWLEDGMENTS This report was prepared by a World Bank team led by Zafiris Tzannatos (task manager and principal author, WBIHD) and Christian Bodewig (co-author, ECSHD). Maria Laura Sanchez Puerta (ECSHD) provided extensive analysis of the survey data and contributed to the chapter on earnings, which also presents comparative wage data originally compiled by Lazar SestoviC (ECSPE). The report was prepared under the overall supervision of Arup Banerji, Sector Manager, Human Development Economics (ECSHD), and Hermann von Gersdoff, Sector Manager, Social Protection (ECSHD), and was sponsored by Orsalia Kalantzopoulos (Country Director, ECCU4). Augustina Nikolova processedthe document at various stages of production, while ImeldaMueller edited the entire report and prepared it for printing. The team is indebted to Government officials, employers and trade unions representatives in Bosnia and Herzegovina for their support, policy guidance and provision of statistical information. In particular the team would like to thank H.E.Radovan VignjeviC, FBH Minister of Labor and Social Policy, and H.E. Miodrag DeretiC, RS Minister of Labor and Veterans- Invalid Protection, as well as key officials, especially Assistant Ministers Mr. Muhibija DeliC, Ms.Diana KadribegoviC, Mr.Rajko KliCkoviC, Mr.Izet MehinagiC and Ms Mira Straiivuk; the officials at the Departments of Statistics Mr. Enes HadiiefendiC, Ms Jelena DokiC and Ms Rada GaliC; representatives of employers organizations Messrs. Tomislav Grizelj, Esad IbiSeviC, Damir MiljeviC, Borko DjuriC and Mr. Miso KruniC; trade union officials Messrs. Edhem Biber, Munir SpahiC, Hamdo Melez, Sulejman Hrle, Mehmed AvdagiC, Cedo Vola;, Danko RuSiCiC, Ms. Veljka OdiakoviC, Ms. Ranka MisiC, Ms. Fatima FazliC, Ms. Zorica KaziC and Ms. Amila Borovina; and colleagues from international organizations including Ms. Lejla Tanovic (ILO), Ms. Caroline Brearley and Mr. Almir Jazvin (USAID), and Mr. Andrew Lovegrove (DFID consultant). The team is grateful for valuable comments from Mr. Graham Slack (IMF). Particular thanks are due to the members of the Economic and Social Councils as well as the participants of a workshop on unions and collective bargaining held at the University of Sarajevo inMarch 2005. The team benefited greatly from support and guidance from Andrew Dabalen, Willem van Eeghen, Csaba Feher, Monika Huppi, Ivailo Izvorski, Vladimir KreaEiC, Arvo Kuddo, Toby Linden, Maniza Naqvi, Philip O'Keefe, Dirk Reinermann, Tarik SahoviC, Goran TinjiC, Erwin Tiongson, Milan Vodopivec and Ruslan Yemtsov. The peer reviewers were Ardo Hansson, Lead Economist (ECSPE), and Jan Rutkowski, Lead Economist (ECSHD), and their comments, along with those received by numerous colleagues on the concept paper and earlier versions of this work, greatly benefited the final report. TABLE OF CONTENTS OVERVIEW .................................................................................................................................. V Recent developments inthe labor market .................................................................................. vi The impact of labor regulations on labor market outcomes........................................................ x Mainpolicy directions for amore dynamic labor market inBosnia-Herzegovina .................... xi A. Create a leaner but more enforceableset of labor regulations.............................................. xii B.Review the systemofpayroll contributions......................................................................... xiv C. Review collective bargaining and wage determination ........................................................ xv D Facilitatelabor adjustment and stop the flow ofnew "fictitious" workers......................... . E.Improve the statistical basis for labor market analysis ...................................................... xvii xviii ... CHAPTER 1: LABOR MARKET CHARACTERISTICS, WAGES AND LABOR COSTS 1 Introduction and Summary.......................................................................................................... 1 A. Employment 3 Broad Employment and Unemployment................................................................................. Characteristics and Trends................................................................................ 3 4 Unemployment........................................................................................................................ 6 The Employment Rate and Hours of Work............................................................................. Informal Employment ............................................................................................................. 6 A Digressionon Education ..................................................................................................... 7 Employment and Unemployment: The Labor Force Participation Rate................................. 8 B. Wages and Labor Costs.......................................................................................................... 9 Aggregate Wages .................................................................................................................... 9 MinimumWages................................................................................................................... 11 The Tax Wedge: Payroll Deductions for Social Insurance and Taxes.................................. 13 CHAPTER 2: INDUSTRIAL WAGES AND INDIVIDUAL EARNINGS ............................ 18 Introduction and Summary ........................................................................................................ 18 A. Statistical Sources................................................................................................................. 20 Administrative Data.............................................................................................................. 20 B.Survey data............................................................................................................................ 21 22 The Ranking of Industries in the Two Entities..................................................................... Inter-IndustryPay Differentials ........................................................................................... 22 The Dispersion of IndustrialWages and Individuals' Earnings ........................................... 24 The Effect of the Minimum Wage ........................................................................................ 25 C.Conclusions 27 Individual Earnings............................................................................................................... ........................................................................................................................... 28 Differences Between Usual and Actual Pay ......................................................................... 28 The Low Paid........................................................................................................................ 30 Earnings Functions................................................................................................................ 31 Aggregate (Country-wide) Analysis ..................................................................................... 31 Entity and Gender-BasedAnalysis ....................................................................................... 34 CHAPTER3: THE CHANGINGINDUSTRIALRELATIONS FRAMEWORK ...............37 Introductionand Summary ........................................................................................................ 37 Findings................................................................................................................................. 37 A.Recommendations ................................................................................................................. 38 The Changing InstitutionalFramework................................................................................ 41 Labor Laws ........................................................................................................................... 41 i Collective Agreements.......................................................................................................... 44 Organization of Collective Bargaining ................................................................................. 47 Trade Unions......................................................................................................................... 48 50 B Determinationof Wages and Worker Benefits..................................................................... .Employers' Associations....................................................................................................... . . 51 MinimumWage .................................................................................................................... 51 Wage Determination............................................................................................................. 53 C.Wage 53 Employment Protection Legislation..................................................................................... and non-wage benefits ................................................................................................ 54 Termination of employment.................................................................................................. 54 Notice period in case of dismissal......................................................................................... . . 56 Severance .............................................................................................................................. 56 56 Part-time and temporary employment................................................................................... Waitlisted workers ................................................................................................................ 57 Employee Workbook ............................................................................................................ 58 Working hours and overtime provisions ............................................................................... 58 Leave and maternity leave provisions................................................................................... 58 D.Benchmarking Bosnia and Herzegovina's Employment Protection..................................... 60 Regulatory Gaps ................................................................................................................... 61 The Representativenessof Social Partners and Issues in Collective Bargaining..................61 Workbooks ............................................................................................................................ 62 Natural Disasters and Disability............................................................................................ 62 D.Layoffs .................................................................................................................................. 63 65 E Improve the Statistical Basis for Labor Market Analysis ..................................................... . Directions for Industrial Relations Reforms......................................................................... 68 REFERENCES ............................................................................................................................ 71 APPENDIX I:DESCRIPTIONOF THE LSMSAND VARIABLES ................................... 75 APPENDIX 11:TABLES ............................................................................................................. 79 APPENDIX 111: MTDSPRSP-LABORMARKET REFORMS .......................................... 96 APPENDIX IV: COMPARISONOF BRANCHCOLLECTIVE AGREEMENTSAND GENERAL COLLECTIVE AGREEMENTS ........................................................................... 98 APPENDIX V: UNIONSAND INDUSTRIALRELATIONSINAN INTERNATIONAL CONTEXT .................................................................................................................................. 103 List of Tables Table 1.1 FormalPaidEmployment and Productivity 1999-2004 ............................................. 3 Table 1.2 Main Labor Markets Indicators................................................................................... 4 Table 1.3 Employment Rate and Weekly Hours Worked ........................................................... 6 Table 1.4 7 Table 1.5a Labor Force Participation Rates by Entity and Gender, 2002-2003 ............................ Employment and Unemployment Rate by Age and Education.................................... Table 1.5b Labor Force ParticipationRates by Entityand Gender 2001-2004 ............................. 8 9 Table 1.6 Nominal Wages and Inflation ...................................................................................... 9 Table 1.7 Changes inAverage Net Monthly Wages ................................................................ 11 Table 1.8 Average net wage inEUR for SEE Countries........................................................... 11 Table 1.9 Calculation of the Tax Wedge in RS and FBH, 2004 ............................................... 15 .. 11 Table 2.1 Company Registry and Response Base for the Calculation of Wages. RS. 2004 ..... 21 Table 2.2 Comparison of Earnings between Administrative and Survey Data. 2003 .............21 Table 2.3 Correlations betweenIndustrial Average Earnings Calculated from Administrative Sources and the LSMS. 2003 .................................................................................... Table 2.4 Industries Ranked by Wages Relative to Average Wage inEachEntity..................22 23 Table 2.5 Coefficient of Variation of Earnings......................................................................... 25 Table 2.6 Share of Workers Whose Pay is Differentthan UsualPay ....................................... 29 Table 2.7 Number of Workers and Share of Workers Reporting Zero Earnings .................... 29 Table 2.8 Shares of Workers Paid and of Workers Paid DifferentWages, 2003...................... 29 Table 2.9 FBHand RS Formal Sector Employees: Positive Earnings Below Halfthe Average Table 2.10 RegressionResults, Dependent Variable Log (monthly earnings) ........................... Wage ......................................................................................................................... 30 32 Table 2.1l a RegressionResultsfor Male Employees Aged 25-55 With Positive Earnings (between KM40-KM4000), 2003.............................................................................. 35 Table 2.1l b RegressionResultsfor Female Employees Aged 25-55 With Positive Earnings 36 Indicative List of Benefits inBranch Collective Agreements................................... (between KM40-KM4000), 2003.............................................................................. Table 3.1 44 Table 3.2 Annual Growth Rates (%) of Real Average Wages, 2001-2003............................... 66 Table 3.3 Six Externalities of Decentralized Wage Setting...................................................... 67 List of Figures Figure 1.1 Share of informal sector employment has been on the rise inrecent years ................5 Figure 1.2 Informal employment is heavily concentratedinagriculture...................................... 5 Figure 1.3 Aggregate wages appear to have outgrown productivity inrecent years inboth 10 Figure 1.4 Unit Labor Costs inthe Manufacturing Sector ......................................................... Entites. although not so inthe RS manufacturing sector .......................................... 11 Figure 1.5 The FBHhas one of the highest minimum to average wage ratio inEurope and Central Asia............................................................................................................... 11 Figure 1.6 The RS and FBHtax wedges are at the lower end of those found in ECA and OECD ....................................................................................................................... 16 Figure 2.1 PercentageIncreases in Average Industry Wages (Moving average o f industries ranked from lowest to highest pay ones)................................................................... 24 Figure 2.2a KernelEstimates of the log real wage distributions inthe formal and informal sectors inBH(top panel), RS (middle panel) andFBH(lower panel): all workers (2001 to 2003) ......................................................................................................................... 26 Figure 2.2b Kernel Estimates of the log real wage distributions inthe formal and informal sectors inBH(top panel), RS (middle panel) and FBH(lower panel): employees only earning between KM40 and KM4000 (2001 to 2003) ............................................. 27 Figure 3.1 Collective agreements at all levels either confirm provisions or set more generous Figure 3.2 Structure and Membership of Trade Unions inBosnia and Herzegovina, 2005.......46 worker entitlements than the Labor Laws................................................................. 49 Figure 3.4 The FBHminimum wage is one of the highest in the region as a share of the average wage .......................................................................................................................... 52 Figure 3.5 Bosnia and Herzegovina's maternity leave duration stands out incomparison with other European and OECD countries, 1998-2002..................................................... 59 Figure 3.6 BH scores worse than regional comparators on an index of hiring difficulty, but compares better on firing rigidity and working hours............................................... 60 ... 111 ListofBoxes Box 1.1 Clarifying terms ........................................................................................................ 14 Box 1.2 payroll deductions? .................................................................................................. Social Insurance: A tax or a benefit. and should it be financed by general taxes or 17 Box 3.1 Commonalities, differences and gaps between the FBHand RS Labor Laws..........40 Selected provisions o f collective agreements that could be revisited ....................... Box 3.2 43 Box 3.3 RSDraft Civil Service Law ...................................................................................... 46 Box 3.4 Bosnia and Herzegovina's enterprise overstaffing challenge ................................... 55 Box 3.5 57 International experience on temporary layoffs.......................................................... The issue o f waiting lists -the FBHexample........................................................... Box 3.6 64 iv BOSNIA AND HERZEGOVINA LABORMARKETUPDATE OVERVIEW High wages comparedtoproductivity in theformal sector, the large and growing share of workers in the informal sector who are not covered by social insurance, and persistent unemploymentare the three key labor market challengesfacing Bosnia and Herzegovina (BH) today. These are the main findings o f this report, laying the analytical underpinnings for a medium-term labor reform agenda in BH. This overview presents a summary o f the findings and recommendations o f the report, while an assessment o f the structure o f and trends in the labor market i s included in Chapter 1. Chapter 2 examines the effects o f the different wage determination processes inthe two Entitieson earnings, and Chapter 3 reviews the regulatory and industrial relations framework. A review o f the international experience with collective bargaining is included in Annex V o f this report and can serve as the basis for future discussions among local stakeholders about reform directions for collective bargaining inBH. While itsfindings are broadly consistent withprevious analyses and recommendations, this report identijles recent developments in the labor market and, by focusing on industrial relations, provides new policy insights'. The need in BH is still for policies at the macroeconomic and sectoral levels that would enhance the prospects for growth o f the BH economy. Although there are some signs o f improvement (for example, in terms o f inflation control), they are too fragile and not sustainable unless systemic reforms are introduced and effectively implemented. The report shows that there have also been some improvements in labor market outcomes over the last few years. Nonetheless, there i s no reason for complacency as more could be achieved. In particular, changes in the legal and regulatoryfiamework could contribute further to the performance o f the labor market. Labor market analysis in Bosnia and Herzegovina remains constrained by severe data limitations and needs a strong cautionary note with respect to the quality o f available data. The analysis in this report relies on both survey (Living Standard Measurement Survey) and administrative data o f doubtful representativeness. The authorities plan to field an annual Labor Force Survey from 2006 which will allow testing the conclusions drawn in this report. However, policymakers would also benefit from surveys o f vacancies and job creation to improve the understanding o f labor demand. ' World ~ Bank (2002), "Labor Market in PostwarBH- How to EncourageBusinesses to Create Jobs and Increase Worker Mobility"; World Bank (2005), "Bosnia and Herzegovina Country Economic Memorandum"; World Bank (2005), "Poverty and the Labor Market in Bosniaand Herzegovina: 2001to 2004". V t%i la BH is experiencing continuous GDP growth BH Growth Rates 1997-2004 - which has led to employment creation, although not 40 0 sufficient to make a dent into persistently high unemployment. p 300 Although not as high as in the t !! immediate post-conflict years, annual 200 w output growth has averaged around 4- 2e,100 5 percent since 2000, with similar increases in employment (around 4 percent). However, this high 1997 1998 1999 2000 2001 2002 2003 2004 employment growth has not been Source: Staff estimates sufficient to reduce unemployment which appears to have stabilized at high 70 rates (at more than 20 percent o f the 60 labor force in both Entities). Most net 50 CI employment creation, around 8 percent 5 40 IEmployment annually in the Federation o f Bosnia and 30 Herzegovina (FBH) and 1 percent in the 20 Labor Force Republika Srpska (RS), has been driven 10 by the employment o f workers not 0 covered by social insurance and who are 2001 2002 2004 classified in this report as informal Source:Cwncalculationsand World Bank (2005) sector workers. based on LSMS While unemployment remains high and some groups Cfor example, younger workers and the unskilled) are affected more than others, BH has experienced a fast increase in the labor force participation rate and significant gains in employment and hours of work. The traditionally low labor force participation rate in BH stood at 59 percent in 2004, a significant increase from 48 percent in 2001. Hours worked have also increased, with 92 percent o f workers working more 40 hours in 2004 (compared to 88 percent in 2001). In the same period 136,000 additional jobs were created, corresponding to an increase in the employment rate from 40 to 46 percent between 2001 and 2004. Much of the net employment generation is driven by a growing Employment by sector, in percent of total employment informal sector. The share o f the 100% informal sector in total BH employment 90% increased from 37 to 42 percent between 80% 70% 2001 and 2004, jumping from 41 to 49 CI 5 60% percent in the RS and from 33 to 36 50% 40% percent in the FBH. In terms of 30% numbers, net employment in the 20% 10% informal sector increased by about 0Yo 111,000 overall - with about 70,000 2001 2004 new jobs in the informal sector in the &ncalculations andWorldBank(2005). FBH and nearly 40,000 in the RS. based on LSMS vi Informal employment in Bosnia and Herzegovina is heavily concentrated in the Distributionof Informal Sector Employment2004 agricultural sector, with many workers being self-employed and contributing family members. Construction There has been heavy Manufacturing hemorrhage of employment in the state- owned enterprises (SOEs) and the Trade privatization and restructuring process of large SOEs per se has been slow. The Hotels and Restaurant share o f workers in SOEs in total Utility employment fell from 37 to 20 percent between 2001 and 2004*. Today about 220,000 workers remain employed in SOEs in both Entities, compared-to more than 350,000 three years ago. However, there is evidence from enterprise surveys presented in this report that SOEs continue to suffer from substantial overstaffing. On the other hand, despite the substantial loss o f 120,000 SOE workers, total formal sector employment increased from 635,000 to 675,000 (Table 1) or by about 150,000 jobs (nearly 50 percent) in the non-SOEs sector between 2001 and 2004. The source o f this substantialjob creation inthe formal sector i s not well understood, supporting the hypothesis that there may be few effective barriers to employment creation at the current level o f economic growth. Still, it is reasonable to assume that more could have been achieved had the labor market been seamless (i.e., between the two Entities, or between the public and private sectors) and without some issues inthe labor regulatory framework (see below). There are significant differences in labor market outcomes between the Wage and Productivity Growth, 1SSS=lOO I two Entities: generally, the labor market performance of the RS trails that of the FBH on some key indicators. While +FBn wage analysis o f productivity developments is +FBH prod constrained by doubts about employment *RSwage pma RS data quality, it appears that, across the economy as a whole, wage growth in RS has outstripped productivity growth over 1999 2000 2001 2002 2003 2004 the past years, while in FBH this Source Staff eotimteo based on Statistical Institute paid envloymnt data i discrepancy is less pronounced. However, this Entity-wide picture conceals Wages and Productivity, Manufacturing, 1999=100 diverging patterns across industries and may be prone to measurement errors. Therefore, it is useful to look at individual +FBH wage industries. For example, productivity in +FBHprod the RS manufacturing sector appears to have grown strongly in recent years and in line with wages, while in FBH it has 1999 2000 2001 2002 2003 2004 I declined. Although labor force Source: Staff eotimter based on Statistical Institutepaid enployrent datal participation and the employment rates iemain higher in RS, there-has been an equalizing trend between the Entities recently. Between For a detailed analysisbased on LSMS dataofthe structureand evolutionof SOE employment, see World Bank (2005) vii 2001 and 2004, the labor force participation rate in the RS increased by only four percent annually, compared to 10 percent in the FBH. The employment rate inthe RS increased by only one percent annually, compared to eight percent in the FBH, with a resulting net employment increase o f only 9,000 (compared to 140,000 inthe FBH). The share o f the formal sector is much lower in the RS (51 percent o f total employment) than in the FBH (64 percent), and has also declined faster - by 5percent annually compared to only 2 percent inthe FBH. Unemployment in the RS increasedby 60 percent (from 15 to 24 percent) compared to 23 percent inthe FBH(from 17 to 21 percent). Wages are 20 percent lower in the RS than in the FBH, but have been increasing 4-5 percentage points faster per annum than in the FBH. The Entities also retain different systems o f minimum wage determination, rates o f payroll contributions and social insurance provisions. Most importantly, non-wage allowances are included inthe base for social insurance contributions inthe RS but not inthe FBH. Overall, formal sector aggregate wages relative to productivity in Bosnia and Herzegovina remain high in a regional context. BH's net wages have been high Average Net Monthly Wages in EUR, 2000 and 2004 compared to its neighbors in Southeast Europe (SEE), second only to Croatia, and L this picture has remained unchanged 600 between 2000 and 2004. Comparative I2000 wage levels in the EU accession p2 $88 ;;; I2004 candidates Bulgaria and Romania are 0 substantially lower than in BH. Although 9%$6 c" 6 ,.";"*bo$@ 9" Q real wage increases in BH have not been %$Q +++a@,@ GI0 the highest inthe region, they appear to be *%Q increasing at a constant rate over time, and Source World Bank staff calculations costs that are driven by high levels of net ECA tax wedge on labor, 2003 wages. Although the report does not attempt to assess the correct rate o f contributions for social insurance purposes 50 (in terms of costs to the employers and 40 levels o f protection afforded to the 30 workers), it notes that the rates and, more 20 10 generally, the social insurance systems are 0 not harmonized between the Entities. This can affect investment decisions between the two Entities and create a duality between the two labor markets. Unlike in Source: World Bank(2005b) and own calculation! the RS and international practices, social insurance contribution rates in the FBH are levied on a narrow base which excludes non-wage allowances from the calculations. Bosnia and Herzegovina's currency board regime implies the need for labor market flexibility andfor wage restraint. In adopting a currency board regime and fixing the exchange rate to contain inflation, the authorities have given up one option for macroeconomic adjustments. viii This implies the need for flexibility elsewhere and in particular in the labor market. Moreover, employment growth requires wage restraint, evident inwage increases not exceeding productivity increases. Overall, the BH labor market should be able to create more formal sector jobs if it addresses deficiencies outside the labor market, arising from poor investment climate and delayed restructuring and privatization. At the same time, there are issues within the labor market that create a duality between the two Entities,and between the formal/public and informal employment. These are the subject o fthis report. Table 1 BHSelected Labor Market and Economic Indicators, 2001-04 FBH RS BH Employment(,OOO) 2001 55 1 448 1000 --formal 370 265 635 2004 692 457 1149 --formal 441 234 675 IncreaseinEmployment 140 9 149 --formal 71 -31 40 -- informal 70 39 109 Unemployment (,OOO) 2001 112 81 190 2004 182 143 324 Increase 69 62 134 UnemploymentRate (YO) 2001 17 15 16 2004 21 24 22 Annual percentage changes in ...(2001-2004) Nominal wages 6 10 n.a. Inflation 1 1 n.a. Labor ForceParticipationRate 10 4 7 Employment rate 8 1 4 Share of formal sector in employment -2 -5 -3 GDP growth n.a. n.a. 5 Note: Figuressubject to roundingerrors. Source:Own calculations basedon LSMS 2001 and 2004 ix f 0%I ns t tc s Although substantial aspects of Share of formal sector employees reporting earnings below labor regulations are rigid on paper, 50 percent of the average wage particularly in the collective agreements, they do not appear to have the expected 20 negative effect on the labor market at E '5 large. Nominalwages have been growing 10 faster in the RS than in the FBH, although 5 the FBH minimum wage indexation 0 mechanism would have predicted faster 2001 2002 2003 wage growth in the former. In addition, the minimum wage in the FBH, currently Source Own calculations based on LSMSNote (a) Earnings are for workers aged 15 and older (2) Averages are calculated from the at 55 percent of the average wage, does - LSMSand are based on usual earnings X xx 0 Harmonize systems o f payroll contributions and rates between the Entities. C. Review systems of collective bargaining and wage determination 0 Separate public sector wage determination from that in other sectors; 0 Limit collective agreement coverage to those enterprises that are represented in bargaining; 0 Define criteria for representation o f social partners involved intripartite bargaining; 0 Create a State-level Economic and Social Council as a forum to discuss medium-term labor market and collective bargaining reforms. D.Facilitate labor adjustment and stop the flow of new "fictitious" workers 0 Promote enterprise restructuring and free up labor currently held up in fictitious and unproductive employment 0 Develop an alternative to workbooks (for social insurance purposes); and 0 Introduce legislationto address temporary layoffs. E.Improve the statisticalbasisfor labor market analysis 0 Introduce a Labor Force Survey; and 0 Increase capacity o f the statistical offices. This is an ambitious policy agenda which needs to have local ownership and careful sequencing. Its success will also depend on building administrative capacity to implement the reforms, addressing political economy considerations and undertaking statistically informed assessments. A. I wage in a regional context. The FBH minimum wage is set by the new General Minimumlaveragewage ratio,2002 (%) Collective Agreement adopted in August 2005 at an hourly net rate o f KM1.75, @ which as o f June 2005 corresponds to 55 \Rn percent o f the average wage in the &i! economy - the same level as under the old py %a GCA. Under the new regime, it i s to be {i indexed to the cost o f living, the retail &$J price index and general developments in R s la the economy and adjusted at least once per `@I year by Government decree with the 0 10 20 30 40 50 60 70 consensus o f the social partners. Source World Bank (2005) and own calculations xii respective average wage in each Entity is today about the same, and about 20 Unemployment by age percent of FBH formal sector employees 70% report earnings below the minimum wage. 60% The current FBH minimum wage level as 50% a share of the average wage i s one of the 400h --e2001 highest in a regional comparison. $ 30% --e2002 0 HoMever, the new GCA also introduces a 20% provision which allows "in exceptional 10% 0% circumstances" branch collective 15-18 19-24 25-34 35-44 54-54 55-64 agreements or compan) agreements to stipulate an alternative lower minimum SoJrce & n calcJ ations base0 on LSMS wage below KM 1.25 net per hour, which would represent 40 percent of the average wage as of June 2005. The introduction of such a lower minimum wage, even if limited to "exceptional circumstances", could open the \+a) for more wage flexibility especially in companies under stress and preventthe loss of formal sectorjobs and a further informalization ofthe economy. 2. Introduce a separate youth minimum wage. Youth unemployment is high, and the average age of workers in formal employment has been increasing fast. The youth in BH are finding it increasingly difficult to secure employment in the formal sector. Conventionally measured, unemployment among those in the 19-24 age group is almost 50 percent, nhile the self-reported unemployment rate among those aged 15-18 is reaching 75 percent. Of equal concern is the finding that more than a quarter ofthe 15-18 year olds are already out of school. Becausethe productivity of young workers i s evidently lower than those with experience (at the same level/quality of education and other characteristics), setting a common minimum wage for both groups is bound to lead to the exclusion of the oung from the formal labor market. This leads to a loss of output in the formal J sector (by employing fewer young workers) and depresses wages in the informal sector (because o f the relative labor supply increases in that sector). Setting a separate and lower minimum wage for the youth would facilitate their employment and will also increase total output in the economy. 3. Rationalizethe system of coefficients and similar provisions in collective agreements. The branch collective agreements prescribe a public sector-like system of "coefficients". These are provisions binding only for those voluntarily participating in branch collective bargaining, and no law mandates their continuation. The Labor Laws also allow firms in bad financial standing (a term not precisely defined) to pay lower wages than those prescribed in the branch collective agreement. As the empirical analysis of this report shows, many of the prescribed wage increments (`*coefficients`')` in collective agreements are evaded, as are various allowances. Rather than having a "comprehensive" system that is evaded or selectively applied, it is preferable to replace it by a less ambitious one that is transparent and enforced, benefiting norkers and reducing uncertainty among employers. In a modern diversified economy with varying orking conditions and lifelong learning, there is little justification for the mechanical M linking of wages to workers' characteristics. Although the majority of 40 or so detailed provisions in collective agreements can be maintained, others should be reconsidered by the Labor "coefficients" are included in Brach Collective Agreements (BCA) as proportional increases to the minimum wage to compensate for additional worker effort (e.g. heavy or complex work) or education and professional qualifications. Additional increases are granted on the basis o f tenure (the "seniority" allowance) and other supplements, some relating to work effort (such as overtime or night work) and some to norms (for example, leave for blood donation or participation in sports). ... Xlll si? upon which social insurance contributions are calculated. An additional effect o f exempting employer-provided benefits in the FBH from the social insurance base i s that the value o f the effective minimum wage jumps to around 65 percent o f the average wage (compared to the nominal 55 percent). 2. Harmonize the social insurance provisions between the two Entities There are substantial differences between the Entities' payroll contribution systems that accentuate duality between the Entities and theformaUinforma1 sector. Social insurance contributions and taxes are levied exclusively on employers inthe RS, but are shared in the FBH. Although the real effects o f this particular distinction might be insignificant, there are other differences that can have fiscal and social protection implications. One example i s the aforementioned difference with respect to the taxable/social insurance base in the FBH (which excludes the value o f allowances). Another one is an earmarked payroll deduction that finances child protection in the RS but i s absent in the FBH. Additional differences arise from the composition o f before- and after-tax earnings as well as the extent o f informality between the two Entities(36 percent in the FBHand 49 percent in the RS). The common objective is to create a labor market that i s seamless in its outcomes within each Entity and across both Entities. This does not imply that parallel or diverse regulatory provisions between the two Entities should be fully consolidated: the coexistence of separate laws does not necessarily impede the realization of a common market as the evidence from many federal systems in other parts o f the world suggests. However, each Entity can benefit from better practices inthe other and, inturn, reciprocate when its regulations are felt to be more appropriate. Such differences between the two Entities should be gradually reduced and eliminated, and both Entities should pay more attention to issues of compliance and to a better linking o f contributions to benefits. This will increase both the willingness o fworkers to participate in social insurance schemes as well as their protection. in n 1.Separate public sector wage setting from other sectors: The minimum wagesprescribed by the current collective agreements Shares of take-home payand payroll deductions and employment conditions apply equally across employers irrespectiveof in total labor cost corporate identity (such as state, cooperative, mixed or private). As a 100% IWageTax rule most countries have separate laws 80% Social Insurance on civil service employment including 60% ocontributions pay RS Take-home separate minimum wages and wage 40% grids. Labor laws and general collective 20% IFBHAllowancesnon- taxable agreements may (and should) set 0% IFBHBaseWage minimum labor standards and rights and FBH RS I obligations applicable to everyone, with Source: &n calculations xv markets. For example, the public sector is constrained by the fiscal envelope which is both an economic and political issue while the private sector i s driven by competitive forces. In many cases, a sensible private sector decision implyingclosure o f operations might neither be an option for the public sector nor even be desirable in the case o f a public good. Overall, the compensation "package" should be comparable between the public and private sectors in order to ensure that the labor market i s seamless across different sectors. This might require different specific provisions between the private and public sectors to compensate for differences in benefits (for example, in terms o f employment security and social insurance entitlements). Wage determination for civil servants might, therefore, need to be conducted in agreements separate from the rest o f the formal sector workers. However, given the high share o f total public expenditure spent on public sector wages, the Government needs to ensure that the new civil service scales achieve savings or are at least fiscally neutral. 2. Limit collective agreement coverage to those enterprises and workers who are voluntarily participating in bargaining. To preserve the voluntary principle of tripartism, the Government could refrain from exercising its authority to extend collective agreement coverage to those who have not voluntarily signed them. This holds in particular for the General Collective Agreements which are surprisingly detailed for general, economy-wide regulations and which are understood to be binding. In the FBH, the Labor Law provides that collective agreements bind only those employers and only the members o f trade unions who have been represented in the bargaining. The RS law states that agreements are "mandatory" for those who were represented inthe process but allows the same flexibility o f participation as in the FBH. Thus, both Labor Laws already prescribe that collective agreements are binding only for those parties who have directly participated or have been represented in the process o f concluding the agreements. Nevertheless, according to the current general collective agreement in the FBH, all employers, public and private, are in theory bound by the provisions o f the general collective agreement. Similar ambiguity arises in the RS from the mandatory membership o f companies in the Chamber o f Commerce which has been the official representative o f employers in the collective bargaining in the past. These ambiguities should be reconciled in a way that collective agreements do not impose obligations on those who have not agreed to be party to them. 3. Define representativenessof social partnersin tripartite bargaining. The labor legislation in both Entities does not set out clear minimum membership criteria and verijkation procedures for the representativeness of the social partners in collective bargaining. BH authorities should clarify such ambiguities by setting minimum representation standards. This would also clarify the role o f the Government in the negotiation process - as at present the Government is represented (rightly) as Government but also as a principal employer (in the public sector) and a dominant representative o f workers (the public sector still employs almost two-thirds o f the formal labor force). 4. Create a State-level Economic and Social Council as a forum to prepare medium-term labor market and collective bargainingreforms. The level at which collective bargaining occurs can be a critical determinant of economic outcomes. For example, a centralized system can adversely affect macroeconomic performance if it prescribes a high minimumwage that could risk startingan inflationary process. Similarly, decentralized collective bargaining can enhance or hinder microeconomic outcomes through its effects on productivity and profitability o f various companies or different sectors. The international experience with respect to different industrial relations systems provides only suggestive guidance as to what level and kind o f collective bargaining could be more appropriate for BH. However, Bosnia and Herzegovina's economic transition i s far from complete and xvi substantial enterprise sector adjustment still needs to take place. This suggests that the tripartite parties could be best served by allowing for more rather than less flexibility in the labor market that would promote better outcomes. This report recommends that the agreement on a future framework for collective bargaining be reached through a tripartite dialogue and based on more accurate data and careful economic analysis of the implicationsof legal and broaderpolicy options. A move from or to GCAs to or from BCAs or company-based agreements should be pursued on the basis o f collating quality data, identifying relevant national and international experience, building analytical capacity and continuing social dialogue under the social-economic councils. This can be facilitated by the preparation o f labor market strategies that can be undertaken by the Ministries o f Labor in discussion with Entity ESCs or, as recommended by this report, a State- level ESC. f ne 1. Promote enterprise restructuring and free up labor currently held up in fictitious and unproductive employment The State-owned enterprise sector remains overstaffed, and there is evidence of `ffictitious" and unproductive employment of workers. Some workers, while formally employed in SOEs, are actually working in the informal sector, while others are lingering on in their enterprise in unproductive employment. While the full extent of "fictitious" employment and overstaffing across the enterprise sector as a whole is uncertain, enterprise surveys suggest that the SOE sector remains substantially overstaffed and is seen as providing a major obstacle to restructuring. This suggests that the authorities need to proceed with enterprise restructuring, including privatization and bankruptcy procedures, and release excess employment to allow for a reallocation o f labor to more productive employment. 2. Develop an alternative to the workbooks as a means of keeping track of the workers' socialinsurancehistory. The requirement for employers to retain an employee's workbook until the point of separation creates and reinforces several rigidities in the labor market. Designed for a labor market characterized by full-time and lifetime employment, the current workbook practice - which tracks inhard copy the employee's employment history for social insurance purposes - has been proven to be a major defacto barrier to formal alternative forms o f employment including part-time and temporary work and secondary employment. Under the current system, the employer is required to retain the employee's workbook until the employment relationship ceases, at which time he has to return it to the employee duly stamped with the termination date. This practice o f workbooks has contributed to the emergence o f the "fictitious worker" problem: When employers cannot meet their obligations with respect to current wages, social insurance dues or severance payments, the workers naturally stop working for such employers. At the same time, while these employers are not in a position to meet their obligations and stamp the workbook, arrears start accumulating, creating a vicious circle and increasing their inability to settle. Alternative administrative arrangements for maintaining the social insurance records o f workers urgently need to be explored and introduced. Reform measures over the medium term may involve the introduction of an integrated electronic employee registration database. xvii However, in the short term procedures could be amended to enable employees to withdraw workbooks, while retainingtheir claims on unpaid wages and social insurance contributions. 3. Introducelegislationto addresstemporarylayoffs. BH's legislation does notprovidefor layoffs, and a worker is generally treated as if he were "in" or "out" of full-time employment. Although existing labor regulations provide for wage restraint in companies under financial stress, temporary employment adjustment remains limited in the absence o f regulations for layoffs. Ifthere were provisions in the legislation for an employee to be placed on temporary "lay off7 status, that employee would no longer be prohibited from seeking lawful employment with a "secondary" employer, while remaining subject to recall once the "primary" employer is able to do so. Such a temporarily laid-off worker would have no break in social insurance contribution history for the period o f layoff and would continue accumulating social insurance cover provided by the secondary employer. Addressing this legislative gap may prevent an employer's bankruptcy, while employees may also have a reasonable chance to return to their primary employer, if still desirable. The assessment of this report is that BH lacks the information basisfor an objective assessment of the characteristics and trends of the labor market and it is therefore recommended both that an annual Labor Force Survey be introduced and that the capacity o f the Statistical Offices be increased. The need for relevant, adequate, and timely statistics i s particularly desirable in the area o f labor markets where policies and reforms typically involve winners and losers. The diverse fate o f these two groups can be particularly acute in the short run, although adverse distributional effects o f reforms can be neutralized, and in fact be reversed, in the long run. Therefore, an adequate statistical system is a prerequisite for the informed and timely design, implementation and monitoring o f labor policies. This can avoid deadlocks in the dialogue arising from incorrect perceptions o f reality (as i s at times the situation in BH in certain areas) and can reduce industrial conflict. xviii CHAPTER 1:LABOR MARKET CHARACTERISTICS, WAGES AND LABOR COSTS 1, This chapter provides an overview o f the broad characteristics and trends inthe BHlabor market. It examines, on one hand, quantitative aspects o f employment (such as employment, unemployment and hours o f work) and, on the other hand, the level o f and changes in the aggregate wage as well as payroll deductions (the tax wedge). Although the data are far from ideal and at times provide a rather incomplete (and possibly unreliable) picture o f the macro economy and the labor market, there are some observations that may more or less describe the situation. 2. The labor market in Bosnia and Herzegovina still exhibits undesirable characteristics, for example 0 The formal sector remains dominated by the public sector that has no sustainable prospects for growth. e Most of the recent gains in employment comefrom the informal sector; whose share increased from 3 7 percent in 2001 to more than 42 percent in2004. e The wage levels are high in a comparative context, especially for the young whose employment is characterized by the highest rates o f unemployment and informality. e Unemploymentremains high (at more than 20percent of the laborforce) and is double or more that rate for younger workers. e Unemployment and informality is further exacerbated by low enrolment rates at successively higher levels of education,especially for secondary education inwhich only 7 3 percent o f the 15-18 year-olds are enrolled, 40 percent o f whom are in vocational programs. e Reported wage increases are ahead of inflation and quite uniform across the two Entities despite substantial differences inthe two economies, for example, with respect to the structure o fthe economy and the wage setting mechanisms. e Aggregate wages in RS have grown more strongly than in FBH - unlike what the more liberal minimumwage setting mechanism there would have implied. 3. This suggests that the restructuringand the transition of the economy of Bosnia and Herzegovina into a dynamic market economy is far from complete and much remains to be done at the broad economic policy level (including enterprise restructuring, privatization and business environment improvements). Specific labor market policies may, o f course, help in many different ways but their effect would be marginal until sustainable growth resumes. The big issue, therefore, seems to be the size o fthe overall productive base o fthe economy. 1 4. Despite the difficult state of the Bosnian economy and the labor market, there are some encouragingsigns of changes,although some might be rather fragile inpractice, ifnot the result o f poor quality o f data: The laborforce participation rate has increased significantly since 2001 (by more than 20 percent by 2004) and now stands at just under 60 percent (compared to 48 percent in 2001); Despite the increase in the labor force participation rate, the unemployment rate conventionally defined has been virtually unchanged since 2002; this is also the case for the self-reported unemployment which appears to have experienced a slight decline over time; The employment rate, a more reliable indicator, has shown a clearer tendency to increase (from 40 to 46 percent between 2001 and 2004) and this has provided employment for nearly 150,000 additional workers; A good sign is the increase in average number of hours worked by workers, indicating that the rise inemployment rate has not been at the expense o f full-time employment; The increase in employment and hours of work has also been accompanied by increases inproductivity; Incomparisonto the RS, formally reported wage increases in relation to inflation have been less excessive in the FBH where "spiraling inflation processes'' might have been expected due to the mechanical indexation o f the minimumwage; Despite a substantial loss of 120,000 jobs in SOEs since 2001, net formal sector employment appears to have increased overall by about 40,000jobs. Although the absolute level of wages is one of the highest among a comparable group of countries, the tax wedge in BH is in the middle range of ECA countries, while changes in wages over time have been more aligned with inflation and GDP growth in BHthan inneighboring countries. Conclusions: The above results and findings presented further inthis report suggestthat: Labor market improvements require increases in aggregate demand and continued economic growth. This calls for: (a) broad based economic reforms at the macro level, supplemented by labor markets policies, in particular aiming at wage restraint and promoting labor market flexibility; and (b) a better educated and trained labor force. The system of wage determination is probably bindingfor some sectors but less sofor an economy in which the informal sector is large and rising; still, there is a need for reforming the current method o f determination o f minimum wages in the FBH and also introducing a separate rate for the youth. Social insurance contributions are not excessively high and the more relevant issues may be: (a) harmonization o f the Entity Laws on Social Insurance (with respect to contribution rates and the coverage o f benefits); (b) broadening o f the contributions base (by treating non-wage benefits as part o f the worker's total compensation); and (c) more effective use o f revenues (which will increase the willingness o f workers to participate in social insurance schemes). Reforms in these three areas can boost labor mobility and reduce the extent o f informality inthe labor market, thereby increasing worker protection against the risks o f poor health, old age and unemployment. 2 A. Broad Employment and Unemployment 6. Despite conflicting evidence from different administrative and survey data, total formal sector employment in BH has probably increased over time (see Table 1 in the overview chapter and Table 1.1 below). Formal employment in agriculture has been stagnant but the sector employs very few workers (approximately 10,000 in each Entity). However, the formal manufacturing sector has lost about 10 percent o f its workers since the late-1990s. In the same period, formal employment in services increased by about 15 percent inthe RS, butthere were no employment gains for this sector the FBH (Appendix Table Al.1). It is likely that formal employment has contributed to some o f the net gains in the FBH; however, this does not appear to have been the case in the RS (see earlier table 1 in the Overview section). Overall, it appears that the formal sector might not have been as lethargic as previously thought: Despite a substantial loss o f 120,000 workers in SOEs since 2001, formal sector employment appears to have increased by about 40,000 jobs between 2001 and 2004. However, as mentioned, the picture i s far from being definite at this point, given data contradictions. Table 1.1 presents differing employment and productivity changes based both on officially published aggregate employment data from the Statistical Institutes and on data on paid employment. It shows how much labor market analysis in BH is complicated by data limitations. It has to be noted that paid employment data is more in line with LSMS-based estimates o f a 19 percent employment increase between 2001 and 2004 in the FBH and an 8 percent decline in the RS. The estimates based on paid employment suggest strong productivity growth in both Entities between 1999 and 2000, followed by a flattening out and decline in FBH and by further growth in RS in recent years. Lastly, the aggregate trends conceal employment variations between industries. For example, as Table 1.1 shows, productivity in RS manufacturing appears to have experienced a strong upward trend, compared to a fall in the FBH, which is driven by strong labor shedding in the RS manufacturingsector. Table 1.1Formal Paid Employment and Productivity 1999-2004 (Index 1999=100) 1999 2000 2001 2002 2003 2004 Formalofficially reported employment FBH 100 110 109 106 104 104 RS 100 103 100 106 108 107 Productivitybasedon formal officially reported employment FBH 100 108 111 131 136 154 RS 100 107 97 95 99 111 FormalPaidEmployment FBH 100 83 93 101 103 124 RS 100 91 86 91 85 91 Productivitybasedon paid employment FBH 100 132 120 126 126 118 RS 100 121 113 111 126 131 Formalpaidmanufacturingemployment FBH 100 116 129 145 145 170 RS 100 99 81 87 75 73 Productivitybased on paidmanufacturingemployment 3 FBH 100 93 95 97 101 99 RS 100 121 121 127 147 173 Source: Entity Statistical Institutes; productivity: Bank staff calculations; Notes: In2004, formal employment was 388,000 inthe FBH and 237,000 inthe RS in 2004 (total for BH: 625,000), while formal paid employmentwas 283,700 in FBH and96,600 inRS (total for BH:380,400); According to the LSMS2001 and2004, formal employment inthe FBH increased by 19% but declined by 8% inthe RS; 7. Despite recent shrinking, the public sector remains large, but its prospects are anemic. There has been a heavy hemorrhage o f employment in the state-owned enterprises, and the privatization o f large SOEsper se has been slow. The share o f employment in SOEs to total employment fell from 37 to 20 percent between 2001 and 2004. Today only about 220,000 workers are employed in SOEs in both Entities, compared to more than 350,000 three years ago.5 However, the public sector's share in total employment still stood at 42 percent in 2003, down from 52 percent in 2001. Conversely, the private sector's employment share has increased from 34 percent in 2001 to 40 percent in 2003. Thus, most o f the employment gains have come from increases inthe informal sector (see next section). 8. Labor force participationrate in BH has been on the increase. To the extent that the LSMSprovides a relatively accurate pictureofthe labor market, the traditionally low labor force participation rate in BH i s steadily increasing (Table 1.2). An encouraging feature is that the employment rate (which i s typically measured more accurately than the unemployment rate) is also increasing consistently. The unemployment rate seems either to have been stabilized (at a highrate), or to be slightly decreasing, ifthe self-reported unemployment rate is considered. Table 1.2 Main Labor Markets Indicators Inpercent 2001 2002 2003 2004 Unemployment Rate 16 22 n.a. 22 Self-reportedUnemployment Rate 33 31 30 na Employment Rate 40 42 43 46 Participation Rate (U+E) 48 53 n.a. 59 Participation Rate (self-reported U+E) 60 61 62 na Share of formal (to total) employment 63 61 na 58 Total employment (,OOO) 1,000 1,149 Source: Own calculations basedon LSMS and Poverty and the Labor Market in BH (2005) Informal Employment 9. The share of informal employment is high, and increasing. The share o f the informal sector in total BHemployment increased from 37 to 42 percent between 2001 and 2004, jumping from 41 to 49 percent in the RS and from 33 to 36 percent in the FBH. Interms o f numbers, net employment in the informal sector increased by about 111,000 overall - with about 70,000 new jobs in the informal sector in the FBH and nearly 40,000 in the RS. Figure 1.1 summarizes the aggregate picture between 2001 and 2004 (see also earlier Table 1 in the Overview section and Appendix Table Al.2). The share of women intotal (male and female) informal employment has been around one-third, and this is practically identical to the share o f women in total formal employment in 2001. Young workers, however, are disproportionately employed in the informal sector. In 2001, the share o f informal workers who were young (below 25 years) was nearly 20 See Poverty and the Labor Market inBH2001 to 2004 (World Bank, 2005) 4 percent, compared to a share o f this group in formal employment o f less than six percent. Similarly, low-skilled workers have a higher share o f informal employment compared to formal employment. The shares o f women, youth and low skilled in informal employment remained practically unchanged for the period covered by the LSMS. Figure 1.1 The share of informalsector employment has been on the rise in recent years Employment by Yetor, in percent of total employment I ~ 2001 2004 a n n c~IcuIalion8and World Bank (2005). based on LSMS 10. The majority of informal workers in 2001 were employed by private employers (52 percent). The rest were self-employed (25 percent) and contributing family members. By 2003 the share o f informal private sector employees decreased to 40 percent, which suggests that most o f the employment gains have probably come from defensive self-employment family mechanisms rather than the expansion o f small firms. Figure 1.2 Informal employment is heavilyconcentratedin agriculture - - - - Distribution of Informal Sector Employment2004 Construction 0 Manufacturing 7 Source: World Bank (2005) 11. The largest informal employer is agriculture. In2001the agriculture sector employed 40 percent o f all informal workers, followed by construction (19 percent), and manufacturingand trade (with 9 percent each). By 2004, the share o f agriculture in overall informal employment increased to 50 percent (see Figure 1.2). As for the employment status, in 2001 the majority o f informal workers were employed by private employers (52 percent), while around 25 percent were self-employed and about one-fifth were contributing family members. In2002 and 2003, the share o f informal private sector employees decreased to 42 and 40 percent respectively. The proportion o f informal self-employed increased, particularly in 2002, while the proportion o f contributing family members increased in 2003. This reinforces the previous observation about the limited role for employment creation that small firms appear to have recently played, as does 5 the fact that the share of construction and manufacturing ininformal employment decreased by 15 percent and 8 percent, respectively, over the last three years. TheEmployment Rate andHours of Work 12. The overall employment rate6 in BH increased from 40 percent to 46 percent between 2001 and 2004, and this increase has not been accompanied by a reduction in the hoursworked. As Appendix Table Al.2 shows, the only subgroup that experienced a decline in the employment rate was men in the RS where the employment went from 61 percent to 58 percent. Nonetheless, this group retains the highest employment rate. The employment rate o f men in the FBH increased from 48 percent to 55 percent, closing the gap with the employment rate o f men inthe RS. Employment rates for women increased by around 10 percent, from 23 to 26 percent in the FBH and from 33 to 37 percent in the RS. Overall, the employment rate (43 percent) is considered to be low, with even lower rates in the FBH (40%) compared to the RS (48%). A positive sign is that increases in the employment rate have not been accompanied by decreases in hours worked (Table 1.3). The percentage o f all workers reporting more than 40 hours worked per week increased steadily in the three rounds o f the LSMS (2001, 2002, 2003) from 88 percent to 90 percent and 92 percent. Table 1.3 Increases in the employment rate have not been accompanied by a decrease in hours worked - Distributionin percent of workers according to weekly hoursworked Hours 2001 2002 2003 5 1 1 1 10 2 1 1 20 5 4 3 30 4 4 3 40 33 38 43 50 40 33 31 60 16 19 18 All 100 100 100 N 2626 2803 2805 Source: Own calculationsbasedon LSMS Unemployment 13. The conventional unemployment rate' was 16 percent in 2001 but increased to 22 percent in 2002 and has remained at this level since (Table 1.2). To the extent that these figures are accurate, they represent a significant increase. The rates are higher for (a) women, (b) in the FBHcompared to the RS and (c) the youth (below 25) and older workers (above 55). It is not possible to calculate the conventional unemployment rate from the LSMS 2003. Based on information on the self-reported unemployed, the unemployment rate in BH declined from 33 percent 2001 to 31 percent in 2002 and 30 percent in 2003 (Table 1.2). These changes are contrary to those based on the conventional unemployment rate and mask the fact that the rate from men declined in the FBH(from 35 percent to 27 percent) but increased in the RS (from 26 percent to 30 percent) (Appendix Table A1.2). For women, the converse is true: self-reported unemployment rate declined in the RS (from 36 percent to 30 percent) but slightly increased in 'Theconventionaldefinition employmentrate is the share of employedworkers inthe working age population (15-64). The ofunemployment refers to those not employed and willing to work, seeking work, andbeing available for work. 6 the FBH(from 36 percent to 38 percent). These changes are, however, rather small and possibly erratic.' 14. Youth unemployment rates are exceptionally high. A more definitive conclusion relates to the unemployment o f the youth (below the age o f 25). The youth face exceptionally high unemployment rates in excess o f 50 percent on both counts o f unemployment (conventional and self-reporting). On the other hand, unemployment generally decreases with higher education attainment. It is below 10 percent for those with university education compared to 30-40 percent for those with primary education only (Table 1.4). Table 1.4 Employment and Unemployment Rate, by Age and Education Unemployment Rate Self-reported Unemployment Rate EmploymentRate in percent 2001 2002 2001 2002 2003 2001 2002 2003 Age 15-18 52.5 59.3 75.4 77.9 73.0 6.7 8.5 9.3 19-24 34.8 43.3 54.4 54.0 51.5 30.8 30.6 34.1 25-34 19.0 27.0 38.3 36.8 35.0 46.8 47.9 51.2 35-44 10.5 16.3 23.4 23.9 23.5 57.6 58.7 57.8 45-54 9.9 12.0 18.9 17.1 18.6 53.8 56.7 56.5 55-64 5.2 7.9 25.9 15.9 24.2 21.9 26.5 25.5 Education Unfinished elementary 15.3 15.5 34.2 23.7 25.2 22.8 27.3 27.6 Elementary 20.2 28.1 40.2 40.8 41.2 26.6 29.9 30.2 Vocational 17.4 22.3 33.6 33.0 30.5 53.5 55.0 56.2 High School 14.8 21.4 31.6 28.6 27.8 44.7 46.7 50.4 University (2y) 6.8 9.0 12.8 12.0 12.9 69.0 67.0 65.0 University (4y) 2.8 5.0 8.5 4.4 6.0 82.7 84.1 82.7 Source: Own calculations based on LSMS A Digression on Education 15. The high unemployment rates of the youth can be related to the low educational enrollmentrates at successively higherlevels of education. Inthis respect, it should be o f great concern that only 73 percent of those in the 15-18 year-age group are enrolled in secondary education. Most European countries have net secondary enrollment rates o f between 85-95 percent. In addition, a disproportionate share (40 percent) o f students who stay on in secondary education is in vocational programs. As shown in table 1.4, those with vocational education have higher unemployment rates and, as shown in the analysis o f earnings (next chapter), their earnings are lower than those with general secondary education. Focusing on the youth in households living below the poverty line, their enrollment rate in secondary education falls to only 57 percent. To make matters worse, children from poor families have a much greater tendency to be enrolled in vocational programs that usually lead to higher unemployment and lower earnings (see Chapter 2). In addition, whereas the net enrolment rate for the non-poor in ~~ The earlier LSMS surveys included information on "discouraged workers", that is, those unemployed who were no longer seekingwork becausethey believed that nojobs were available. For example, in2001 discouragedworkers accountedfor 37 percent ofthe unemployed. However, it is not possible to classify individuals as "discouraged" for LSMS 2003. 7 the tertiary age group (age 19-23) i s 27 percent, it is only 9 percent for youth from households with incomes below the poverty line.g Employment and Unemployment: TheLabor Force Participation Rate 16. The labor force participation rate was low in 2001 (47.8 percent), but increased to 59 percent in 2004. The labor force participation rate using self-reported unemployment can be calculated for the three waves" and is estimated to have increased from 59.7 percent in 2001 to 62.3 percent in 2003. Thus, unlike in the case o f unemployment where the conventional count and results based on self-reporting pointed in different directions, both estimates o f the labor force participation rates point reassuringly inthe same direction. Table 1.5a Labor Force Participation Rates by Entity and Gender, 2002-2003 2002 2003 BH FBH RS BH FBH RS LFPR 1 Total 53.7 47.8 61.4 na Na na Men 69.4 65.0 74.7 na Na na Women 38.4 31.6 47.6 na Na na LFPR 2 Total 61.6 56.1 68.7 62.3 58.1 67.8 Men 77.5 73.4 82.7 78.5 75.0 82.9 Women 46.0 39.9 54.3 46.5 42.2 52.2 LFPR2: Change 2002 -2003 Total 1.1% 3.6% -1.3% Men 1.3% 2.2% 0.2% Women 1.1% 5.8% -3.9% LFPRl:Based on Unemployment Conventionally Measured; LFPR2: Based on Self- reported Unemployment Source: LSMSown calculations 17. Despite these gains, the labor force participation rate in BH is low in a regional context where the corresponding rate is in excess o f 60 percent in some economies. Most o f the difference is due to the low labor force participationo f women inBH. For example, female labor participation rates are only 34 percent in BH compared with 57 percent in Romania. However, more recent evidence from 2004 (on a comparable statistical basis) suggests that female labor force participation rate rose by more than 30 percent (compared to an increase o f 18 percent for men). In the FBH (where participation rates have been traditionally low), female participation increased by 32 percent, compared to 14 percent inthe RS. CEM, 2005 loWith the absence o f informationon conventional labor force participation rate for 2003 (due to a routing- error on the questions about unemployment inthe survey) comparisons with 2003 cannot be made.. 8 Table 1.5b: Labor Force Participation Rates by Entity and Gender 2001-2004 2001 2002 2004 All 48.1 53.3 59.0 FBH 42.2 47.1 55.8 RS 55.4 60.9 63.O M e n 63.6 68.9 75.0 Women 32.9 38.0 43.1 Note: based on conventionally measured unemployment Source: Poverty and the LMin BH, 2001-2004 h s Aggregate Wages 18. The difficulty o f the labor market to generate jobs, even at the relatively low rates o f employment observed in BH, has been traditionally linked to rising labor costs and high social security contributions and labor taxes. Both are examined below at an aggregate level, while a more detailed presentation can be found the next two chapters. Table 1.6 Annual rates of growth in economy-wide nominal wages and Inflation between specific years and 2003 or 2004 Until 2004 1999104 2000104 2001104 2002104 2003104 RS Wages 14% 11% 11% 10% 11% CPI 5% 3yo 1% 2% 1% Difference 9% 9% 10% 9% 10% FBH Wages 7% 7% 6Yo 5% 1% CPI 1% 1% 1Yo 0% 0% Difference 6% 6% 6% 5% 1% Until2003 1999103 2000103 2001103 2002103 N a RS Wages 15% 12% 11% 9% N a CPI 6Yo 3% 2% 2% N a Differences 9% 8Yo 9% 7% N a FBH Wages 9% 9yo 9% 10% N a CPI 1% 1% 1% 1% N a Differences 8% 7% 8% 9% N a Source: Bank staff calculations 19. Indeed, the aggregate wage has been increasing much faster than inflation and this i s true for both Entities (Table 1.6). While the issue o f wage increases ahead o f inflation changes is uncontestable, some additional observations are listed below also with reference to the aggregate level. The choice o f period is critical inthe sense that there can be significant variation in the estimates o f wages and inflation depending on the choice o f the start year and end year. For example, by changing the final year (from 2004 to 2003), the difference between the changes inwages and changes ininflation inthe FBHincreases bynearly 50 percent. One safe conclusion 9 is, therefore, that the statistical basis in BH i s in need of urgent improvement. This point is pursued further in the next chapter when administrative data on wages are comparedwith those from surveys. Figure 1.3 Aggregate wages appear to haveoutgrown productivity in recent years in both Entities, although not so in the RS manufacturing sector Wage and Productlvily Growth, 1999=100 200 200 150 -+-FBH wage +FBH prod +FBH prod RS pmd 100 ++RSwage prod RS 50 50 1999 2000 2001 2002 2003 2004 1999 2000 2001 2002 2003 2004 Source: Staff esbmtes based on Stabstical harrmtepaid enploymnt data, Source: Staff estimates based on Statisbcalhsmtepaid enploymnt data Source: O w n calculations based on paid employment data from Statistical Institutes 20. Aggregate wages have been growing more rapidly in RS than in FBH and faster than productivity in both Entities.". The minimum wage setting mechanism in the FBH has been charged for creating a wagehnflation spiral, which should be absent in the RS where minimum wages are too low to be binding. However, as Table 1.7 indicates, the aggregate average wage has increased faster in the RS than in the FBH between 1999 and 2004. Also, as shown inthe next chapter, althoughminimumwages are set in direct relationto the average wage in the FBH, many low-paying industries in the RS have seen their wages rising faster than average, and in fact more than in the FBH (Appendix Tables A1.3 and A1.4). Moreover,while data limitationspreventdefinitiveanalysis of productivitydevelopmentsinBH, it appearsthat RS wages have outgrownproductivityacross the economy as a whole, while this has been less so in FBH. In contrast, as Figure 1.3 shows, productivity in the RS manufacturingsector appears to have risen strongly in recent years and in line with wages after 2002, while in FBHwages have outgrown stagnant productivity. Figure 1.4: Manufacturing unit labor costs have been growing stronger in FBH than in RS Unit Labor Cost in Manufacturing, 1999=100 I 1999 2000 2001 2002 2003 2004 Source: Bank staff calculations based on Statistical Institute paid employment data The reported increases inproductivity may be somewhat inflated to the extent that they fail to take into account the recent increase in average hours worked (see Table 1.3). 10 21. As a result, unit labor costs in the manufacturingsector have been rising in recent years in both Entities. However, due to stronger productivity increases in the RS, unit labor costs there have increased by less than in the FBH and appear to have actually decreased since 2002. (see Figure 1.4). Table 1.7 Changes in Average Net Monthly Wages (in KMand Index 100=1999) FBH RS Year AW AWIndex CPI AW AW Index CPI 1999 375 100 100 216 100 100 2000 413 111 102 274 127 114 2001 443 118 104 309 143 122 2002 483 129 104 347 161 124 2003 529 141 104 379 176 126 2004 533 142 105 422 196 128 Source: Own calculations based on StatisticalInstitutedata 22. Finally, in a comparative context during the last five years, BH seems to have had rather low increases in wages compared to other regional economies. Table 1.8 presents comparative average net wage data from countries in South Eastern Europe. While all the above observations are conditional on the quality of data, and despite the fact that BHdoes not stand out as an undesirable negative outlier in a comparative context, the fact remains that BH has the second highest average net wage (after Croatia). Table 1.8 Average net wage in EUR for SEE Countries 2000 2004 Rank Annual YOchange Rank Albania 99.1 169.3 6.0 14.3% 3 Bosniaand Herzegovina 191.2 258.2 2.0 7.8% 5 Bulgaria 94.9 122.6 7.0 6.6% 6 Croatia 435.4 556.6 1.o 6.3% 7 Macedonia 172.4 201.0 3.0 3.9% 8 Montenegro 96.3 195.9 4.0 19.4% 2 Romania 107.2 147.2 8.0 8.3% 4 Serbia 45.4 193.0 5.0 43.6% 1 Notes: Albania: public sector wages only are included; Bosnia and Herzegovina this simple average for data inRepublikaSrpskaand Federationof B&H; Bulgaria data for 2004 are for 4 4 only; Montenegro. * - - net wage doesn't include allowancesfor food, transport etc. Source: Bank staff calculations Minimum Wages 23. The Labor Law in both Entities delegates the determination of minimum wages to the General Collective Agreements (GCA). In the FBH, the GCA adopted in August 2005 prescribes a net hourly minimum wage o f Kh4 1.75, which corresponds to a monthly minimum wage o f KM 30812.This represents 55 percent o f the FBHaverage monthly wage in the month o f June 2005 o f KM 556.42 and excludes benefits wage allowances. This is the same level as that prescribed under the old GCA which set the minimum wage mechanically at 55 percent of the '* TheAugust 2005 FBH General Collective Agreement also introduces aprovision, according to which "in exceptionalcircumstances" branch collective agreements or company agreements can stipulate conditions, level, method andperiods of determination of aminimumwage below KM 1.25 net per hour. 11 average wage, resulting in a continuous average wage/minimum wage indexation spiral. In contrast, the new GCA stipulates that the hourly minimumwage "is adjusted with increases in the costs o f living, growth o f retail prices and overall economic development" at least once per year by a decree o f the FBHGovernment on the basis o f harmonizedpositions o f social partners inthe ESC for the territory o f the FBHyy'3. The determination o f the minimum wage is less straightforward in the RS, but comes to around 20 percent o f the gross average wage (including benefits). For more detailed analysis o f the minimum wage determination, see chapter 3 o f this report. 24. The minimum wage rate in the FBH is one of the highest among the transition countries in Europe and Central Asia. In a comparative context, the minimum wages in the FBHare second only to those of Lithuania, and are far ahead the ratio of minimum to average wages inmore than one dozen comparators (Figure 1.5).14 In contrast, the minimumwages inthe RS is among the lowest inthe region. Figure1.5: The FBHhasone of the highestminimumto averagewage ratioin EuropeandCentral Asia Minimumlaverage wage ratio, 2002 (YO) Lith FBH N h C Ukr Sloven Hung Slovak Czech Lat Pol Rom Bulg Est RS Arm Mold Russia Belar Azer 0 10 20 30 40 50 60 70 I Source: TransMonee and Bank staff calculations for ECA Labor Market Study 25. While nominal minimum wages are set at a high level in the FBH, evasion also appears to be relatively high. A caveat here is that the ratios o f minimumto average wages are often nominal, rather than real, in the sense that the book value o f minimum wages might not relate much to the actually paid minimum wages in the economy. Nevertheless, while nominal minimum wages are set at a high level in the FBH, evasion also seems to be relatively high. l3In contrast, ILO Convention 131 (Minimum Wage Fixing Convention: 1970) provides for taking into account both the needs of workers and their families, the general level of wages, the cost of living, existenceof social security benefitsand the relative living standardsof other social groups as well as economic factors, including the requirements of economic development, levels of productivity and effects on employment/unemployment. l4For comparison, the ratios of minimumto average wages for developedcountries range from 0.71 in Italy, to over 0.6 inAustria andNorway, and to below 0.45 in Spain, United Kingdom, Canada, and United States. 12 Perhaps more important is the finding that the percentage o f formal sector employees who receive less than 50 percent o f the average wage is very similar inthe two Entities (for details see Chapter 2 and Tables A1.5a and A1.5b). 26. I n conclusion, it seems that the problem in BH has more to do with a (historically) high level of wages rather than with recent changes in wages per se. The solution to addressing high levels o f wages is to increase productivity. However, given the slow gains in productivity, real wage growth will need to be moderated, while other components o f labor costs also need to be examined. The latter is done in the following section with respect to social insurance contributions and wage taxes (the tax wedge). The Tax Wedge: Payroll Deductionsfor Social Insurance and Taxes 27. Each Entity has a different system of social insurance and payroll taxation, with differing taxable bases and differing definitions of "net" and "gross" wages (see Box 1.1). The terms "gross" and "net" wages are used in a somewhat counterintuitive way in BH, whereby "net" wages include benefits in the RS while "gross" wages exclude them in the FBH. There are also additional differences between the two Entities with respect to scope o f social insurance (for example, whether maternity issues are covered or not) as well as the level o f contributions and administrative matters. These differences may produce some misunderstanding and lead to the miscalculation o f the tax wedge and are further discussed below. 28. More specifically, in the RS, social insurance contributions and taxes on wages are levied on the "net" wage, defined as the individual's take-home pay plus various supplements such as the hot meal allowance and the transport allowance. Contributions are levied only on employers. The contribution rate for pension and disability insurance i s 24 percent o f the net wage, 15 percent for health, 1 percent for unemployment insurance and 2 percent for child protection (which also funds maternity leave). The tax rate is 10 percent and levied, as in the case o f social insurance contributions, on the net wage. Thus the payroll deductions as a percentage o f the "RS net wage" i s 52 percent (Le. 24%+15%+1%+2%+10%) and corresponds to row 7 divided by row 4 in the last column o f Table 1.9. 29. I n contrast, the calculation of payroll deductions for social insurance and taxes in the FBH excludes wage allowances from the taxable base. They are levied on both the employer and the employee, although not in equal proportions. For example, the rates for pension insurance are 24 percent (7 percent from employers and 17 percent from employees), 17 percent for health (4 percent from employers and 13 percent from employees) and 2.5 percent for unemployment insurance (0.5 percent and 2 percent from employers and employees, respectively). They are all levied on the "gross" wage, which i s defined as the employee's wage plus the employee's share in the social insurance contribution but excluding allowances such as for transport or meals. On the other hand, the tax on wages i s set at 5 percent o f the employee wage (which is defined as the FBH"net" wage, without the wage allowance^)'^ and accrues to the Cantons. This rate, as expressed as based on "gross", is 3.8 percent. Thus the payroll deductions as a percentage o f the "FBH gross wage" are 47.3 percent (24%+17%+2.5%+3.8%) and as a percentage o f the "FBH net wages" 68 percent (row 7 divided by row 1 in Table 1.9). l5Thus, allowances (such as hot meal and holiday allowances) are not taxable inthe FBH. 13 Table 1.9 Calculationof the Tax Wedge in RS and FBH, 2004 FBH RS 2 x 5 x Minimum Average average average Wage wage wage wage Takehome 1 Base wage KM 299 545 1090 2725 * 2 Hot meal allowance 136 136 136 136 I 3 Transport allowance 50 50 50 50 * 4 (1+2+3) Net wage (KM+allowances) 485 731 1276 2911 485 Contributions 5 ss 191 349 697 1743 204 6 Tax 15 27 55 136 48 7 (5+6) Total deductions 206 376 752 1879 252 8 Memo item Social security base FBH [Row (1)/0,68] 440 801 1603 4007 Na Labor costs 9 (1+7) Excluding allowances 505 921 1842 4604 737 10 (4+7) Including allowances 691 1107 2028 4790 737 Tax wedge 11 (719) Excluding allowances 40% 40% 40% 40% 34% 12 (7110) Including allowances 30% 34% 37% 39% 34% Notes: * same irrespective of level of wages/allowances/laborcosts- so only one case is reported for benchmarking it to Allowances are taxable inthe RS andtherefore are not reportedseparately; also, the tax wedge inthe RS is the the case ofthe FBH. * The table shows labor costs are higher inthe FBHthan inthe RS for a given value of take home "wages and benefits" (compare column 5 to column 1) or, equivalently, that for the same labor costs to employers take home pay is lower inthe RS than inthe FBH. 1) Bold numbers indicate the reference points for subsequent calculations in the two Entities. "Netwage" is row (1) inthe FBHbutrow (4) inthe RS. ``Gross wage" is row (8) inthe FBH but row (10) inthe RS. 2) The tax rates are: 5% inthe FBH levied on row (l),and 10% inthe RS levied on row (4). 3) The social security contributions for pensions, health andunemployment (and, for the RS, children benefit) are 43.5% inthe FBH levied on row (8) and 42% inthe RS levied on row (4). 4) Allowances (such as hot meals and transport) are non-taxable inthe FBH. Therefore, the larger the share of allowances, the lower the tax wedge. For example, if other allowances(e.g. leave) are assumed to be equal in value to the value of the hot meal and transport allowances, the tax wedge in column (1) is reducedto 23%. 31. At around one-third of labor costs, the conventionally defined tax wedge in both FBH and RS does not stand out as excessive among comparator countries. For example, Figure 1.6 shows the tax wedge in a number of neighboringeconomies in ECA and also inthe OECD. The RS and the FBH occupy the 6`h and 8th lowest positions, but only three countries have substantially lowertax wedge than the two Entitiesamongthe 29 listedinthe Figure. 15 Figure 1.6 The RS and FBHtax wedges are at the lower end of those found in ECA and OECD ECA tax wedge on labor, 200312004 ~ 70 60 50 40 30 20 10 0 Sources: OECD, TaxingWages 2002103 for OECD and ECA OECD, except Slovenia(basedon 2001); Bank staff estimates for others (using2003 for ECA countries taken from World Bank ECA Labor Market Study, forthcoming, and 2004 for RSand FBH). Tax wedge calculatedfor non-agriculturalAW worker without dependents in ECA and productionAW worker without dependents in OECD. OECD groupings exclude ECA OECD: (i) low (less than40% wedge) includesAustralia, Iceland, Ireland, Japan, Korea, Mexico,NewZealand, Portugal, Switzerland, UK and USA; (ii) mid (40-50%) includes Canada, Denmark, Greece, Luxembourg, Norway, Spain and Sweden; (iii) high(51+ YO)includesAustria, Belgium, Finland, France, Germany, Italy andNetherlands.Note:tax wedge calculations are based on nominal regulatory provisions 32. However, such comparisons need to be treated with caution, given comparability of data and various tax exemptions. In particular, one should bear in mind that tax wedges in ECA countries are highrelative to their per capita income levels.'* Finally, another reason for the low rate o f the tax wedge in BHmight have to do with the genuinely high labor costs in absolute terms, an observation made earlier with reference to Table 1.8. 33. Improvements in the area of social insurance should focus on broadeningthe social insurance contributions and tax base by treating non-wage benefits as part of the worker's total compensation. The latter can be facilitated by setting minimumwages in a pragmatic way for the current economic conditions and making more effective use o f the revenues (which will increase the willingness o f workers to participate in social insurance schemes). This will increase worker protection against risks o f poor health, old age and unemployment and will reduce perception o f social insurance contributions as taxes (see Box 1.2). 34. At the same time, alternative ways to reduce the payroll burden of taxation should be explored. For example, BHis inthe process of introducingVAT, and excess proceeds, ifany, could be usedto finance a reduction inthe rates levied on social insurance contribution and wage taxes. This would require an analysis on types o f social spending with the highest returns and which will increase welfare the most. World Bank(2005), Labor Markets in Europe and CentralAsia: TowardsCreatingMore and Better Jobs 16 17 WAPTER 2: INDUSTRIAL WAGES AND INDIVIDUAL EARNINGS 35. This chapter examines the statistical sources of industrial wages (from administrative sources) and individual earnings (from the LSMS) and undertakes various analyses o f the characteristics and determinants o fworkers' pay. 36. With respect to the data sources, the findings suggest that the labor market informationsystem in BH is incomplete and, for certain issues, potentially misleading. In many cases the findings from administrative data contradict those from survey data) For example, the proposition that industrial relations have a significant bearing on actual labor market outcomes (notably with respect to minimum wages, low pay and the distribution o f pay) i s supported in some cases but negated in others. The findings from the analysis o f the LSMS are probably more reliable, although the surveys themselves have some shortcomings as well. Nevertheless, a number o fpropositions can be put forward. 3 7. The analysis supports earlier findings that there are significant differentials across Entities, gender, educational groups and, to a lesser extent, sectors of employment. It also casts additional light on the importance o f institutional factors on wage outcomes in both the RS and the FBH. Where differences between the present and previous analyses emerge, they do not necessarily contradict each other as the current analysis benefits from most recent information that was not previously available (for example, the LSMS data for 2003 and administrative data for 2004). In fact, our analysis suggests that there might have been some significant adjustments inthe way the labor market operates duringthe more recent years, most notably inthe FBH. For example, while minimum wages have been found to be quite binding and with ripple effects on the rest of the FBH economy in the past, they may be less so recently. This is compatible with the present findingthat institutional factors (such as specific provisions incollective agreements) do not generally have the expected impact on earnings. 38. Bearingin mind that data on wages and earnings are subject to biases arisingfrom the partial and irregular reporting and that the definitive proof is still to be produced, a number of summary statements can be made. a. There is some evidenceof a bindingminimumwage inthe FBH...: 0 The dispersion o f industrialwages has been rather constant inthe FBHover time. 0 The ranking o f industries according to the level o f relative earnings is quite comparable between the FBH and the RS, but only for the highest paying industries. For the lower paying industries the rankings between the Entities are rather mixed, which suggests that 18 there are different forces in operation in the low pay sectors that are subject to different procedures determining effective minimumwages. 0 Wage increases in low paying industries in the FBH have been larger than in higher paying industries. b. ...but there is also evidence suggesting that minimum wage determination in the FBH may not be fully binding or not as binding as before: Around 20 percent o f formal sector employees in the FBH report earnings less than the minimum wage, and the share of formal sector employees who receive less than 50 percent o f the average wage i s very similar in the two Entities, despite the different levels o f minimumwages. Although the dispersion o f individuals' earnings used to be relatively compressed in the FBH, the dispersion in the RS has declined sharply over time and is now comparable to that in the FBH; also, the dispersion o f industrial wages is no longer materially different inthe two Entities. While differences in the distribution o f earnings between the formal and the informal sectors inthe FBHwere initially large (2001), the differences have disappeared by 2003. Wage increases in low paying industries in the RS have, in fact, been greater than in the FBH. An additional point favoring the FBH is that industrial wage increases have been much lower than in the RS (Le. ripple effects, if applicable, have not been higher in the FBH- see also Chapter 1 on the relationship between wage increases, inflation and productivity). c. Institutional factors have little or no statisticallydetectable effect on earnings: 0 The institutionalvariables included inthe analysis (such as size o f firm, industrial sector, ownership and so on) generally failed to produce evidence that they have significant effects on earnings. 0 The explanatory power o f the regressions used in the analysis declined over time, suggestingthat systematic factors are becoming less relevant. 0 The effect o f education on earnings in both Entities is rather small and out o f line with what the relevant "coefficients" in collective agreements would have implied. 0 Similarly, the effect o f tenure or seniority on earnings cannot be statistically supported for either Entity. 0 Some o f the most unionized sectors (such as mining and manufacturing) are the lowest paying in both the FBH and the RS - although this may also reflect the stress these sectors (more generally, the public sector) have been under in more recent periods due to the needfor restructuring. d. Still, workers' earningsare characterized by some undesirable conditions such as: 0 Many workers are paid less than their usual wages; and although the share o f those receivingwage arrears has declined, it might still affect one-in-five workers. 0 Although some or many waitlisted workers might be employed in the informal sector, they do so at a significant reduction intheir earnings after controlling for other factors. 0 Together with the findings from the LSMS reported below, the analysis of industrial wages suggests a good possibility that the two Entity labor markets are not that seamless. 19 39. Analysis of the available data suggests that the wage determination process is not subject to easily identifiedsystematic forces, administrative or competitive. This can be the result o f the ongoing transition and incomplete economic adjustment and, probably, an indication o f relative flexibility in the labor market at large. For example, labor hoarding on the labor demand side i s suspected to still be sizeable, and labor supply exceeds labor demand; under such conditions, both workers and employers might want to collude for the sake o f either preserving employment or avoiding payroll deductions. Accordingly, an attempt to apply a rigid wage determination system might produce opportunistic effects in the sense that it affects only specific sectors under specific conditions - most likely in the public sector. Still, almost half o f workers inthe public sector receive pay that is different from their usual pay, and many (perhaps 100,000 or more) are "fictitious" workers. Inbrief, the current system does not have its intended effects. 40. This suggests the need for a leaner set of labor regulations which are enforceable. One obvious conclusion from this analysis is that, rather than having a rigid industrial relations system "in the books" that can potentially deter investors while creating all sorts o f uncertainties among employers and inequities among workers, a better approach would be to redesign the system in a way that will have fewer specifics (e.g. wage increments for different levels of education) but, instead, will include some more general and transparent principles to which everyone would be expected to adhere to. For example, more factors should be taken into consideration in setting the minimum wage than just the average wage (which has, until recently, beenthe case in the FBH), including differentiating between youth and adult workers (which is not the case in either Entity). In addition, the system o f "coefficients" (whose impact is undetectable in the general labor market even though they might have a restrictive effect in certain sectors and specific companies) is not appropriate in a dynamic market where multi- tasking and informal and lifelong learning are increasingly becoming more common. Finally, the separation o f public sector wage determination from that o fthe private sector i s common in many countries and desirable on economic grounds - productive sectors (inthe sense o f selling services or commodities) are subject to different forces than the public sector whose responsibility is the provision o f public goods. The application o f the same rules, regulations and collective agreements across the board might lead to significant inefficiencies when there are many different types o f company ownership (such as public, mixed, cooperative and private - and the latter can be formal or informal). A. 41. There are two main sources of data on earnings in BH: administrative ("grouped") data on average wages by industry, and survey data reporting individual earnings o f workers. AdministrativeData 42. Administrative data are collected in both Entities by their respective Statistical Offices and are based on returnscompleted by registeredcompanies. The companies report the total wage bill and also the number o f workers falling into different earnings brackets. Accordingly, the average wage for the company (subsequently, for a sector and eventually, for the Entity) is derived as a simple average o f the reported information by dividing the total wage bill by total employment. There is no systematic analysis o f the distribution o f earnings, which is, in any case, generally deemed to be biased downwards due to underreporting o f earnings 20 mainly by private employers (for example, in order to avoid paying the full value o f social security contributions). There is no information on hours o f work.'' 43. In additionto the underreportingof earnings by private employers, another source of bias arises from the fact that very few companiesreport the aforementioned information to the StatisticalOffices, and most of those that do are in the broadly defined public sector. Taking the example o f the RS, there were more than 16,000 registered companies at the end o f 2004, o f which only about 1,800 returned their forms. Still, this disappointing overall response rate (11 percent) is not random and is, therefore, not representative o f what happens in the economy - more than 1,400 returns came from government or mixed (Le. public and private) companies whose response rate was 86 percent and 66 percent, respectively. These two sectors account for 80 percent o f all responding companies. The private sector's response rate was a mere 10 percent (Table 2.1). Thus, the administrative data on earnings are heavily biasedtowards the public sectors2' Table 2.1 Company Registry and Response Base for the Calculation of Wages, RS, 2004 December No. of Registered No. of Companies Respondents % Respondents %Distributionof Code Type 1 2 211 Column2 0 NGOiInternational 3132 150 5% 8% 2 Private 11152 191 2% 10% 3 Cooperative 287 48 17% 3% 4 Mixed 914 783 86% 43% 5 Public 1006 664 66% 36% Total 16491 1836 11% 100% Source: RS Statistical Ofice Survey data 44. The second data source for labor market analysis is the household-level data from the Living Standards Measurement Survey (LSMS). An assessment o f the LSMS can be found inthe later part of this chapter and its methodology and relevant variables are explained in the Annex o f this report. Here, the focus is to compare the LSMS earnings data with those derived from administrative data. Table 2.2 indicates earnings in 2003, as reported in the LSMS and by Statistical Offices o f each Entity. Table 2.2 Comparisonof Earnings (KMhonth) BetweenAdministrative and Survey Data*, 2003 Monthly wages FBH RS Administrative DatdStatistical Offices 524 379 LSMSformal (public and private) -- public 536 343 491 304 -- private formal 608 429 LSMS-All (public andprivate, formal and informal) 53 1 330 LSMS-All (incl.those reporting zero earnings)* 477 231 * LSMSearnings are basedon employees ofall ages 15+ reportingpositive (usual) earnings, exceptthe last row, which refers to all workers. Sources: LSMS own calculations; Administrative data: Appendix Table Al.3. l9However, the Statistical Institutes have also been compiling data on paid employment, whose magnitudesare substantially lower than officially reported employment. Another problem with the administrative data is that the company registry is not updated-thus many of the listed 11,000 private companies may have ceasedoperations, being dormant or closeddown. 21 45. There are substantial differences between survey and administrative wage data in RS; however, this difference is less pronounced in FBH. The average formal earnings, as calculated from the LSMS, are approximately 10 percent below the average earnings reported by administrative data in the RS (KM343 vs. KM 379); however, in the case o f the FBH, the corresponding difference is very small (KM536 vs. KM 524). Although these differences are rather small and reassuring at face value, this is probably coincidental rather than by design since administrative data are heavily biased towards the public sector. Comparing the administrative- based average pay to the LSMS average pay only for the public sector increases the difference to 20percent for the RS and 10 percent for the FBH (KM304 vs. KM379 for the former, and KM491 vs. KM525 for the latter).2' 46. The administrative data and the LSMS relate to different parts of the labor market and probably, as is also argued below, none is very reliable for factual analysis of the labor market and subsequent design of relevant policies. The differences between results o f the two data sets are probably more than a simple comparison o f the averages might indicate. Another comparison consists o f estimating the correlations between industrial earnings as reported by the Statistical Offices and as calculated by the LSMS but for different categories o f workers. As Table 2.3 indicates, the LSMS reported average industrial wages for workers in the public sector correlate rather highly (0.70 and above) with the averages based on administrative data. The correlation drops significantly (by almost half) inthe case o f private sector earnings and becomes practically zero inthe case o f informal sector earnings. Table 2.3 CorrelationsBetweenIndustrial Average Earnings Calculated from Administrative Sources and the LSMS, 2003 Sector FBH RS Public 0.77 0.69 Private 0.43 0.26 Informal 0.05 -0.03 Source: Calculated from Appendix Table A2.1 In TheRanking of Industries in the TwoEntities 47. Ifthe two Entity labor markets were relatively integrated, one would expect to see inter- industryearnings differentials to be closely related. There are many elaborate tests that can help answer this hypothesis, and some o f the econometric work presented in this report and based on individual earnings attests to significant differences between the labor markets o f the two *' One explanationfor the underestimation of earnings by the LSMS is that respondents may be less accuratewhen they report their earnings. Also, workers may have atendency to report net (take home), rather than gross, earnings. Finally, it must be almost certain that the companies that supply information on earnings are current with respect to their financial obligations to their workers andthe payment of social insurance contributions. Conversely, while the closure of operations, it should mainly be related to unfulfilledfinancial obligations - companies may be unwilling to non-response rate may, inpart, be due to some genuinenegligence by the firms to fill inthe appropriateforms or complete the questionnairein an attempt to reduce payments (to workers or taxes and social insurance contributions) at least inthe short run. 22 Entities.22 However, the grouped data from administrative sources (that is, average earnings by industry as reported by the Statistical Offices) can be used to quickly examine whether there is some congruence between industries in the two Entities when ranked according to relative earnings - from highest to lowest, in comparison to each Entity's average wages. 48. The results suggest strong uniformity for the higher paying industries but quite a diverse situation for the lowest payingindustries. The relevant hypothesis here is whether the ranking o f industries by level o f earnings is comparable in the two Entities, and whether the relative position o f earnings are strongly or weakly correlated. As Table 2.4 indicates, the four highest paying industries inthe FBHare also the highest paying ones inthe RS and have the same ranking order. These four industries are financial intermediation, public administration, transport, storage and communication, and electricity, gas and water. The correlation between the relative position of earnings in thesefour industries in each Entity comes to a high 0.94. Table 2.4 Industries Ranked by Their Wages Relatively to Average Wage in Each Entity FBH RS Financialintermediation 191% 183% Financialintermediation Public administrationetc 128% 152% Public administrationetc Transport, storage and communication 126% 134% Transport, storage andcommunication Electricity,gas and water supply 125% 133% Electricity, gas and water supply Healthand social welfare 109% 119% Realestate, rentingandbusiness activities Other social and personalservices 99% 109% Health and social welfare Real estate, rentingandbusiness activities 98% 91% Education Agriculture, huntingand forestry 95% 87% Agriculture, huntingand forestry Education 93% 86% Other social andpersonalservices Mining 82% 75% Construction Manufacturing 75% 74% Wholesale, retail; certainrepair Wholesale, retail; certainrepair 73% 72% Fishing Catering 73% 67% Manufacturing Fishing 67% 63% Mining Construction 66% 56% Catering Source:Appendix Table A1.3 49. While there appears to be common forces operating at the high payingindustriesin both Entities, the similarities are reduced in the middle-paying industries. The group comprised o f the next five highest paying industries is also composed o f common sectors in both Entities Health and social welfare; Other social and personal services; Real estate, renting and - business activities; Agriculture, hunting and forestry; and Education. These five industries rank differently within this middle-range o f earnings, and the correlationdrops significantly to 0.45. 50. There is a large discrepancy among the lowest-paying industries. The final group consists o f the remaining six low-paying industries - Mining; Manufacturing; Wholesale and retail trade; Catering; Fishing; and Construction. Although this is also a common group in terms o f industrial composition, the correlation o f relative earnings within that group i s -0.53. This discrepancy at the lower end is associated with a lower dispersion o f earnings in the FBH compared to the RS: the lowest relative earnings in the FBH are 66 percent (for construction "TheF-testsimplysignificantstructuraldifferencesbetweenthetwolabormarketsandthisiswhytheempirical investigation is based on separateregressionsfor the RS andthe FBH and also for women andmen. 23 sector) o f the average wage while the lowest relative earnings in the RS are 56 percent (for Catering) and 63 percent (for Mining). This suggeststhat minimumwages inthe FBHprovide an effective floor for wages. 51. The spread of industrial wages is also wider in the RS than in the FBH. Excluding the financial sector, the range o f relative wages spreads from 56 percent to 152 percent in the RS, but only from 66 percent to 128 percent in the FBH. This is consistent with the hypothesis that settingthe minimumwage inthe FBHinrelation to the average wage has a compressing effect on the dispersion ofearnings inthat Entity.23 52. In recent years, earnings in the lowest paying industries experienced a greater proportionate increase than in the high-paying sectors in both Entities. Another look at changes in industrial earnings over time is provided in Figure 2.1.24 The figure shows percentage changes in industrial earnings ranked from the lowest to the highest. It shows that in recent years earnings in the lowest paying industries experienced a greater proportionate increase than in the higher paying ones, and this is true for both Entities. However, note that the spread is quite narrow in the FBH. In the RS, the highest paying industries also experienced significant increases inearnings almost comparable to the low paying industries. Figure 2.1 Percentage Increasesin Average Industry Wages (Moving average of industries ran' j from lowest to highest pay ones) RS 2002-2004 FBH 1999-2003 25 O'a 0 20 0% e,E 15 0% (Y Lc? 10 0% e g 5 co: n(Y 0 0% 100 200 300 400 500 600 100 200 300 400 500 600 700 EO[ Wage Wage Source: Owncalculations, basedon Statistical InstituteWage data TheDispersion of Industrial Wagesand Individuals' Earnings 53. Wage dispersion analysis suggests that the wage setting mechanism in the FBH might be more rigid than in the RS. To explore the dispersion o f earnings further, we examine the industrial wages as reportedby administrative data, and the individual earnings as reported by 23The industries listedin the last few rows of Table 2.4 containsome inwhich most workers are covered by collective agreements, as well some which are least covered. For example, manufacturingaccountsfor one quarter of all covered workers in the FBHandone insix workers inthe RS. However, earningsinmanufacturingare around 66-75 percent of the respective average industrialearnings in each Entity. Together with Mining(whose coverage rate is 98% inthe FBH), Manufacturing, ConstructionandAgricultureaccount for half of all coveredworkers in each Entity (Appendix Table A2.1). However, high unioncoverage is also common in the highestpaying industries. For the FBHwhere data exist on bothemployment and coverage of collective agreements, morethan 90 percentof workers in the financial sector are covered. The rates as similarly high (morethan 80 percent)in the other high paying industries such as Utilities,TransportandPublic administration. 24The linesrepresentthree-industrymovingaveragesrankedfrom the lowest to the highest. 24 the LSMS (see Table 2.5). Using the coefficient o f variation as the relevant measure of dispersion, the dispersion o f industrial wages in the FBH was slightly above that in the RS in 2001. While the dispersion increased in both Entitiesuntil 2004, the increase was greater for the RS. This suggests that the wage setting mechanism inthe FBHmight be rigid inthe sense that it does not allow for changes in relative wages to take place as freely as inthe RS. Table 2.5 Coefficientof Variation of Earnings Inpercent 2001 2002 2004 Average Wages in 14 Industries FBH 31.9 36.9 34.0* RS 31.7 35.0 39.2 Individual Earnings (LSMS) FBH 77.4 72.5 75.0 RS 103.1 86.8 75.1 * Latest figure for the FBH is for 2003 Note: LSMSfigures are for all employees reportingpositiveearnings. Sources:Industrialfigures calculatedfrom AppendixTable A2.1; 54. The more compressed distribution of pay in the FBH compared to the RS is also confirmed by the LSMS: the coefficient of variation was 77 percent for the FBH in 2001 compared to more than 100 percent in the RS. By 2003, the value o f the coefficient o f variation in the FBH declined (contrary to what the administrative data suggested), but the change was rather small. So, the inference about the rigidity o f wage determination in the FBH still applies. However, the LSMS data are quite contradictory to the administrative data in the case o f the RS: the coefficient o fvariation inthe RS declined by a massive (ifcorrectly measured) 30 percent and was equal to that inthe FBHby 2004. TheEffect of the Minimum Wage 55. Another way to examine the issue whether the minimum wage setting mechanism has an effect in the FBH i s through graphical analysis, using Kernel estimates o f the real wage distributions in the formal and informal sectors. Figures 2.2a and 2.2b present the relevant information for all workers and also for the more well-defined group o f employees only. The differences between formal and informal sector wages are shown in the higher density in the left tail o f the informal sector wage distribution, and the higher mean for the formal sector. The top panel refers to BH as a whole, while the middle panel presents RS results and the bottom panel FBHresults. The first column onthe left refers to 2001 and the last column on the rightto 2003. 56. At the country-wide (BH) level, there are practically no differences between wage distributions in the formal and informal sectors. This masks, however, differences between the two Entities. More specifically, separating the distributions across Entities reveals that most o f the differences between formal and informal sectors are attributed to the FBH (third row in Figure 2.2a). 57. While the formal and informal sector wage distributions in RS are virtually identical, the two FBH distributionsare very different, suggesting that wage inequality is much higher in the FBH informal sector. As Figure 2.2a indicates, the mode for the informal sector in the FBH occurs at a considerably lower wage but with its upper tail coinciding with the formal distribution (in real wages). Consequently, wage inequality is much larger inthe informal sector inthe FBH. 25 58. The above finding for the FBH is consistent with a binding minimum wage that is enforced in the formal, but not in the informal, sector. The RS result, meanwhile, is consistent with the minimum wage set at a very low level. This finding suggests that the minimum wage might well have an employment-reducing effect in the formal sector, and that some workers might be forced to look for jobs in the informal sector. However, the differences on the left tail o f the FBHdistribution practically collapsed by 2003. Figure 2.2a Kernel Estimates of the log real wage distributions in the formal and informal sectors in BH(top panel), RS(middle panel) and FBH(lower panel): allworkers (2001 to 2003) 2001 2002 2003 `1 n Source: Own calculations based on LSMS 59. However, the differences between formal and informal sector wage distributions in the FBHat the lower end disappear in2003, suggesting that the minimumwage has become less binding. While the previous discussion relates to all workers, a more precise way to check the possible effect o f minimum wages on the lower paid is to restrict the sample to employees only. This is done in Figure 2.2b where the lack o f visible differences in the RS i s reconfirmed. For the FBH, differences at the lower end o f the distributions between the formal sector and informal sector have become smaller over time and disappeared in2003. 26 Figure 2.2b Kernel Estimates of the log real wage distributions inthe formal and informal sectors in BH(top panel), RS(middle panel) and FBH(lower panel): employeesonly earning between KM40 and KM4000 (2001 to 2003) 2001 2002 2003 `1 Source: Own calculations based on LSMS Conclusions 60. Prior to examiningindividual earnings, it would be useful to attempt to summarize the above findings,given their diversity. 61. Administrative data are subject to biases, arising from the partial and irregular reporting of earnings, especially from companies in the private sector. Thus, although the examination of administrativedata has produced some results, these results cannot be construed as definitiveproof. 62. The rankings of industries accordingto the levelof relative earnings are comparable in the two Entities, but only for the highest paying industries. For the lower paying industries, the ranking is rather mixed, and the correlation of their relative earnings is high but negative. Although this finding is far from conclusive, it suggests that there are different forces operating in the two labor markets. Possibly, the high minimum wage in the FBH is partly responsible for the difference among low paying industries. The wage dispersion is relatively compressed and has been rather constant in the FBH over time. This is another indicationthat rigidwage setting forces may be in operation inthe FBH. 63. However, the dispersion of earnings in the RS, having declined rapidly over time, has become as compressed as in the FBH in 2003. Additionally, wages in low paying industries seem to have been increasing faster inthe RS than in the FBH. This runs contrary to 27 what one would expect, given that the minimumwage in the RS i s low and generally believed to be non-binding. 64. The wage distributions between the formal and informal sectors in the FBH were substantially different in 2001 and 2002, but less so in 2003. While the kernel density functions have indicated significant initial differences in the wage distributions between the formal and informal sectors in the FBH, the differences have disappeared by 2003. It i s not very clear whether this is because o f the quality o f data or because the system is: (a) evaded, or (b) evolving and cannot therefore produce systematic results. 65. There are substantial differences betweenthe RS and FBHlabor markets. Together with the findings from the LSMS reported below, the analysis o f industrial wages suggests that the two labor markets are showing different characteristics and dynamics and are not seamless. c,t idu rnings 66. This section examines three additional aspects o f pay: first, the share o f workers whose pay i s different from what is usually expected - for example, because o f non-payment, or the payment o f arrears; second, the share o f workers who receive less than the minimum wage and less than 50 percent o f the average wage - the low paid; and third, the determinants o f individual earnings through the use o f earnings functions. Differences Between Usual andActual Pay 67. The informationin the LSMS provides an opportunityto examine a critical feature of the labor market in Bosnia and Herzegovina - the underpaymentof workers. In theory there are three kinds o f workers: (i) who receive what their contracts provide for, and this those can be labeled "usual pay"; (ii) whose "actual" pay may be different than their usual pay those during a particular pay period this can be because o f arrears or because employers cannot meet - their obligations in the current period, and it can be positive or negative25;and (iii) those who are employed and receive no pay (Some non-payment can be justified, for example, in the case o f family workers. Other non-payment can be explained by an employer being currently unable to pay, but the expectation i s that the employment relationship will be resumed in the near future, though not necessarily with the exact current terms. This category o f workers is known as "waitlisted" or "fictitious" workers). 68. The share of workers experiencing wage arrears has been declining in recent years, while workers who receive actual pay that is different from usual pay has been rather constant over time. Table 2.6 shows the share o f workers receiving arrears. Although significant in size, an encouraging development is that these workers constitute a declining share over time. On the other hand, the share o f workers who receive actual pay that is different from usual pay has been rather constant over time. z5InBHthe differencebetween"actual" and "usual" pay usually signifies underpayment and payment arrears. 28 Table 2.6 Share of Workers Whose Pay is Different than Usual Pay In percent 2001 2002 2003 Workers receiving arrears RS 42 33 Na FBH 19 16 Na Workers receivingactual pay different from usualpay RS 19 22 23 FBH 26 27 25 Source; For arrears: CEM2004; for differences in pay LSMS. Notes: Arrears are defined as a delay in salary payment of 2 months or more 69. The share of workers with zero earnings is substantial in both Entities, particularly in RS. Table 2.7 shows the sample size o f workers for each Entity in the LSMS 2003 and the share o f those who did not report positive earnings in the month prior to the survey. The overall share is rather significant in the FBH (lo%), and three times higher in the RS (30%).26 This seems to be driven by the much larger informal sector in the RS, in which family workers and informal/casual employment and pay are more common. Table 2.7 Number of Workers in the LSMS2003 and Share of Workers Reporting Zero Earnings FBH percentwith 0 wages RS percent with 0 wages Total 1373 10.2 1331 30.1 Public 564 2.0 546 15.0 Private 349 0.0 219 1.8 Informal 460 28.0 566 55.7 Note: excludes workers in international organizations Source: Own calculations basedon LSMS2003 70. The private sector seems to be the most compliant with respect to the regular and full payment of wages. As Table 2.7 shows, practically none in the FBHand very few private sector workers in the RS reported zero wages. By comparison, there were a few public sector workers reporting zero earnings inthe FBH, but as much as 15 percent inthe RS. Table 2.8 Index of shares of workers paid and of workers paid different wages, 2003 FBH RS %paid wages diff. % paid wages diff. Index %paid than usualwages Index YOpaid than usual wages public 100 98 41 100 85 38 private 100 100 I1 100 98 19 informal 100 72 50 100 44 29 all 100 90 67 100 70 54 Source: Own calculations basedon LSMS 71. However, the share of workers receivingactual pay different from usual pay is quite high in both Entities. To the extent that the LSMS accurately reports the situation for the employed, the small number o f workers reporting zero earnings (outside the informal sector and the public sector in the RS) is encouraging. However, as Table 2.8 shows, the share o f workers receiving actual pay different from usual pay i s quite high: around two thirds in the FBH, and more than one half o f RS workers. It is not clear how reliable these estimates are. For example, the LSMS reports that only approximately 3 percent of the employed individuals declared 26This discrepancybetweenthe size o fthe informal sector inthe two Entities reinforces the view that the two labor marketsare different. 29 themselves as "waitlisted". Compared to the common belief that the number o f waitlisted workers might still be greater than 100,000, this discrepancy does not providemuch comfort as to the reliability of the LSMS. However, while a large number of waitlisted workers might officially be on the books, the majority of such workers are in some form o f employment and, when asked the relevant survey question, it is possible they preferred to respond as being employedrather thanbeingwaitlisted. TheLow Paid 72. Assessing the share of workers whose earnings are below the minimum wage or significantly below the average wage in their respective Entities gives crucial insights into the role of minimumwages. The LSMS allows suchexamination, and the results are summarized in Table 2.9. The analysis of earnings below 50 percent of the average wage are noteworthy, consideringthe minimumwage in the FBHwas set at 55 percent of the average wage, while it is substantially lower in RS. Note, however, that there are considerable variations and possibly contradictions among the figures presented in the said Table, supporting the reservations expressedinthe previous chapter onthe qualityof datacollection. Table 2.9 FBHand RS have a similar share of formal sector employees reporting positive earnings below the half the average wage YOearningbelow 50 percentof % earningbelow formal MW AW men women All men women All FBH 2001 15.6 29.3 20.2 11.5 23.9 15.6 2002 16.4 29.5 20.7 1.7 15.3 10.2 2003 15.1 23.9 18.1 7.0 13.9 9.4 RS 2001 2.0 6.1 3.1 8.1 16.5 11.0 2002 10.9 5.2 8.9 13.9 10.7 12.8 2003 4.6 2.4 3.8 8.0 6.5 7.5 Source: Own calculations basedon LSMS;Note: (a) Earningsare for workers aged 15 and older; (2) Averages are calculatedfrom the LSMSand are based on usual earnings. 73. The share of paid workers whose usual earnings were below the minimum wage in the FBH at 20 percent is high. While we see a slight decrease in the FBH, there are signs of dynamics for male workers in the RS betweenthe years. The former might be an indication of more effective enforcement of minimumwages inthe FBH. However, this might also be due to compositional effects if, for example, the workers earning below the minimum wage have switched to informalemployment - another undesirable feature of high minimumwages. Given employmentrestructuringandthe decline informal sector employment, the workers who retained their employment represent the more productiveones, and thus might be receivinghigher wages thanthose whose employmentwas terminated. 74. The share of workers reporting earnings below 50 percent of the average wage are roughlyequal in both Entities, despite the different minimum wage levels. When the critical levelof wage is calculatednot from administrativedata but from the LSMS data (for example, as 50 percent of the averagereported earnings), it is evident that the share of workers earning below that critical level has declined in both Entities, and significantly so in the FBH. The share of workers earning below 50 percent of the averagewage is now similar in bothEntities:one inten 30 workers. However, most importantly, the findings reveal that the minimumwage in the FBH is not fully binding, with a sizeable share o f workers reporting earnings below the minimumwage. Earnings Functions 75. The determinants o f earnings based on the results of earlier LSMS have been extensively analyzed beforea2' This section extends previous analyses to more recent years and offers additional insights. Particular attention is paid to the effects o f conventional human capital variables that are more or less typically expected to play a role in a labor market, regardless o f other considerations. These human capital variables include the worker's education and personal characteristics relative to his productivity (such as age, as associated with experience), as well as the worker's extent o f effort (for example, hours o f work and moonlighting). 76. Additional effects on earnings might also arise because o f gender differences or location (for example, different conditions existing between the two Entities) and, more interesting in the present context, because o f the economic and industrial relations environments inwhich the labor market operates. These effects come with additional variables, such as those relating to tenure with current employers (seniority allowance might have an effect on earnings), or those relating to type or ownership o f the enterprise (public enterprises are potentially more constrained by various collective agreements than private ones). Company size, unionization rates and enforcement is expected to increase with the visibility o f an enterprise. Another important factor for consideration i s the sector o f activity, since different industries are subjected to different competitive pressures or prospects - for example, utilities (gas, electricity and water) and the financial sector have the higher earnings compared to certain services sectors (such as retail and wholesale trade, catering, construction and so on). For comparability and summary purposes, earlier regressions that covered both Entities, both genders, and all sectors are first replicated. Then, separate regressions are run for each Entity and gender since the determinants o f earnings for these subcomponents are drastically diverse and do not call for a pooled regression. 28 Also, regressions separating the formal sector from the informal sector are done to obtain additional observations (pooling the two sectors together is not re~ommended).~~The substantial differences inthe earnings functions between the Entities reinforce the conclusion reached earlier that the two labor markets are relatively diverse. Aggregate (Country- wide) Analysis 77. Despite the econometric reservations about pooling different groups and locations together, Table 2.10 provides a summary basis for examining the determinants o f individual earnings across BH. The results conclusively show that there are: 0 gender differences in earnings, favoring menby 15-20 percent; 0 higher wage level differences in the FBH by 30-40 percent; differences thatfavor the more educated (for example, university graduates may be paid almost 60-80 percent more than primary schooled graduates); 0 significant gainsfrom additional hours of work; and '*Running 27See World Bank 2002 and World Bank 2005. ''Separate separate regressionswasjustified on the basis of stability (F) tests. regressionsfor the formal andthe informal sectors are reported inAppendix Table A2.2a and A2.2b. 31 0 premiafor workers in theprivate sector, although rather small.30 78. The results are less clear-cut with respect to: 0 the informal sector, where earnings do not appear to be much different from earnings in other sectors, after controlling for other worker and labor market characteristics; 0 firm size, where one might have expected that earnings increase consistently as company size increases; however, this effect is variable and at times in~ignificant;~~ 0 ten~re/seniority~~ -giventheseniorityallowanceincollectiveagreements; 0 the effect of vocational education (especially since it switches signs between different specifications); 0 age of the worker (although this is typically one o f the most reliable predictors o f earnings). Table 2.10 Regressionresults for the two Entities combined, dependent Variable Log (monthly earnings) World Bank CEM 2004 New Results 2001 2002 2001 2002 2003 Entity differential 0.40 0.39 0.31 0.3 1 0.34 Gender differential 0.22 0.17 0.12 0.12 0.14 Informal insignificant insignificant insignificant insignificant insignificant PrivateRublic 0.18 0.10 0.27 0.16 0.16 Firm size Variable variable variable variable insignificant Age (informal sector) 215 insignificant 1I3 1I3 insignificant Age (formal sector) 215 insignificant 1I3 insignificant insignificant Tenure 216 (wrong sign) 216 wrong sign) 416 (wrong sign) 116 (wrong sign) 116 (wrong sign) Education:primary insignificant 0.16 -0.15 -0.19 -0.10 Education: vocational 0.10 0.15 -0.08 -0.04 -0.02 Education: secondary 0.23 0.37 0.03 0.08 0.11 Education: college 0.38 0.53 0.26 0.27 0.32 Education:university 0.72 0.90 0.54 0.59 0.65 Hours 0.56 0.38 0.45 N 1271 1233 1205 Adjusted R-2 0.31 0.29 0.34 0.30 0.28 Notes: 1) Ratios (e.g. 215 for age in the informal sector) refer to the ratio of statistically significant coefficients to statistically insignificant coefficients inthe same group of dummies; 2) Wrong sing indicates that the coefficient is significant but has the opposite than expectedsign (e.g.earnings decline with tenure contrary to what one would expect given the seniority allowance); 3) The entity differential favors the FBH; 4) The gender differential favors males; 30Ingeneral, workers inthe state owned enterprises (SOEs) earn around20 percentless thanworkers inthe public administration,health and educationwhile workers inthe private sector are estimatedto earn about 10% more than government workers. 31This is probably due to the operations of or the creationof a new private sector that is reportedto be paying higher wages than the prevailing ones. 32Infact, the coefficient ontenure has the wrong signinanumber of specifications. Inany case, according to current provisions, when the seniority premium increases at 0.5% to 0.6 % ayear, it reaches 20% only after 36 to 30 years. 32 5) Differences in the two specifications include: (a) addinghours as aregressor since earnings relate to monthly pay; (b) omitting workers in international organizations since they are not subject to national market forces. The last three columns are based on employees reporting positive monthly earnings betweenKM40 and KM4000. Source: World Bank (2004) and own calculations based on the LSMS 2001,2002 and 2003. 79. The unexpected insignificance of many variables suggests that wages may in practice be set in a much more flexible way than that predicted by the industrial relations framework. The insignificanceo f these variables was perhaps expected in only one case, that o f the informal sector: this is a residual category that may include from low paid casual workers to self-employed professionals. However, the absence o f effects from the other variables i s rather unexpected inthe presumed inflexible industrial relations framework o f BHwhere wage setting i s prima facie mechanical and operates in specific ways (for example prescribing seniority increases) in identifiable segments o f the labor market (such as the bigger companies or the public sector).33 80. While higher levels of education fetch higher earnings, the difference is not that much, if adjusted for the additional years of schooling required to reach one level of education relative to another. Understanding the effect o f education in BH is particularly important because o f the insignificance o f the institutional variables.34 Previous analyses have claimed that the returns to education in BH are large and statistically significant. The results reported in Tables 2.10,2.1 l a and 2.1lb confirm the significance o f the education variable (either as a continuous variable, or as dummies reflecting level o f education, such as primary education, secondary education and so on). While we also confirm that higher levels o f education fetch higher earnings, the difference is not that much, if adjusted for the additional years o f schooling required to reach one level o f education relative to another. For example, in2001, the earnings o f workers with high school diploma were 23 percent higher than for workers with primary education, 37 percent higher for those with 2 years college education, and 72 percent higher for university graduates. However, post-secondary graduates have 8 and 10 additional years o f education (compare to primary school graduates) depending on whether they have completed a two-year college degree or a four-year university degree. 81. Taking into account the additional years of schooling required to achieve a higher levelof education reduces the returns to education significantly. When this is accounted for, the returns for education become 3.8 percent for high school graduates, 4.6 percent for two-year college graduates and around 7 percent for university graduates. An interesting feature is that the rate o f return to education increases with level o f education.35 However, these returns are generally lower than those one would expect (from the international literature) in a market economy. The relatively low returns to education in BH suggest that the "education coefficients" in collective agreements may not be that bindingafter all. This observation accords at face value the absence of noticeable effects from the institutionalvariables included inthe regression. 33Some ofthe weak relationships can be tracedto multicollinearitywhich is addressed later inthis chapter with some changes inthe results. 34The industrial relations system canbe non-binding either because it i s (a) well designedintheory but evaded inpracticeby market forces or (b) evolving and cannottherefore clear-cut effects inone direction or another. 35Recall also that the unemployment rate amonguniversity graduates is the lowest of all educational categories of workers. 33 Entity and Gender-BasedAnalysis 82. Given the substantial differences in the econometric results for the Entitiesand for male and female workers, the earnings functions were run separately for each of the four resulting groups (Tables 2.1l a and 2.1lb), while additional regressions are listed separately for each Entity, sector (formalhnformal) and gender (Appendix Tables A2.2a and A2.2b). To avoid statistical noise, the sample is confined only to employees who reported earnings between 10% and 10 times more than the average wage. It is also restricted to workers between the age o f 25 and 55 who are a more uniform group than the youth or the elderly. The results are more refined than those reported earlier when all groups were pooled together, and more consistent in many respects. For example: hours and education are consistently significant and their effect i s not affected much by the inclusion o f other variables; 0 the effects of education are rather small and similar in the two Entities: the rate o f return to education is around 4-5 percent for men and 7-8 percent for women; vocational education is associated with a negative effect on earnings o f around 20-25 percent (with the exception o f women workers inthe RS), and this finding can be useful infuture educational planning; waitlistedworkershavegenerally lower earnings (by 50-60 percent) this indicates that - at least some waitlisted workers find alternative or temporary employment, though such employment might be less rewarding and not commensurate with their qualifications, after controlling for other measurable factors; the effect o f working in the informal sector is generally negative but not always significant; theprivatdpublic differential is alsosmall and, controlling for other factors, inmost cases insignificant. 83. Institutionalvariables such as tenure, size of firm, industrialsector fail to produce significant results, as in the previous case o f the pooled regressions. This is also the case o f more elaborate regressions that have updated previous estimates and are reported in Appendix Table A2.2. An additional observation from these regressions is that over time the regressions explain less o f the variance o f earnings. This suggests that the effects o f the variables included in the analysis have become less systematic over time, and is consistent with the hypothesis that the industrial relations system in BHis in a state o f flux, an issue examined inthe next chapter. 34 Table 2.11a RegressionsResultsfor Male EmployeesAged 25-55 With Positive Earnings(between KM40-KM4000), 2003 DependentVariable: Log (Monthly Usual Earnings) FBH Constant 3.22 3.15 1.95 1.87 1.87 2.25 2.00 2.32 3.56 2.59 2.83 Ln (hours) 0.62 0.63 0.63 0.62 0.62 0.55 0.62 0.61 0.59 0.57 0.46 Education(yrs) 0.05 0.06 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.50 0.05 Vocational -0.21 -0.22 -0.21 -0.21 -0.22 -0.21 -0.21 -0.22 -0.19 -0.20 Age 0.06 0.07 0.07 0.06 0.07 0.06 0.07 insign 0.06 Age- squaredx 100 -0.08 -0.09 -0.09 -0.08 -0.09 -0.07 -0.08 insign -0.07 Waitlisted -0.70 -0.70 -0.69 -0.73 -0.72 -0.59 -0.69 -0.56 Secondjob insign insign insign insign insign insign insign insign Tenure insign insign insign insign insign insign insign Public sector insign insign (-.13) Privatesector insign insign insign Informal -0.12 insign insign Companysize insign insign insign wiindustries (sigdinsign) 12/13 1/13 R2 0.14 0.17 0.18 0.20 0.20 0.19 0.21 0.16 0.22 0.16 0.21 N 582 582 582 582 582 562 582 467 557 459 455 RS Constant 3.34 3.34 2.87 3.01 2.98 2.92 2.99 2.89 3.02 2.98 3.OO Ln(hours) 0.52 0.52 0.52 0.52 0.51 0.46 0.50 0.59 0.51 0.53 0.51 Education(yrs) 0.04 0.05 0.05 0.05 0.05 0.05 0.04 0.04 0.04 0.04 0.04 Vocational -0.23 -0.24 -0.25 -0.24 -0.24 -0.24 -0.29 -0.23 -0.27 -0.24 Age insign insign insign insign insign insign insign insign insign Age- squaredx 100 insign insign insign insign insign insign insign insign insign Waitlisted -0.53 -0.50 -0.48 -0.50 -0.44 -0.44 -0.42 -0.40 Secondjob insign insign insign insign insign insign insign insign Tenure -0.04 insign -0.04 insign -0.05 insign insign Public sector insign insign insign Privatesector 0.28 insign insign Informal insign insign -0.19 insign Company size insign insign 0.04 whndustries (sigdinsign) 0113 0113 R2 0.14 0.18 0.18 0.20 0.22 0.26 0.22 0.25 0.27 0.28 0.36 N 415 415 415 415 413 402 413 340 401 330 329 35 Table 2.11b RegressionsResultsfor Female EmployeesAged 25-55 With Positive Earnings (between KM40-KM4000), 2003 Dependent Variable: Log (Monthly Usual Earnings) FBH Constant 2.54 2.40 3.16 2.85 2.92 2.25 3.20 2.66 3.54 2.37 2.17 Ln(hours) 0.65 0.68 0.68 0.71 0.70 0.57 0.65 0.71 0.62 0.60 0.64 Education(yrs) 0.08 0.09 0.09 0.09 0.09 0.08 0.08 0.07 0.08 0.07 0.06 Vocational -0.27 -0.28 -0.26 -0.25 -0.24 -0.27 -0.32 -0.24 -0.32 -0.29 Age insign insign insign insign insign insign insign insign insign Age- squaredx 100 insign insign insign insign insign insign insign insign insign Waitlisted -0.67 -0.68 -0.74 -0.66 -0.67 -0.64 -0.66 -0.65 Secondjob insign insign insign insign insign insign insign insign Tenure insign insign insign insign insign insign insign Public sector 0.94 insign insign Privatesector 0.84 insign insign Informal -0.50 -0.32 -0.28 insign Company size insign insign insign whndustries(sigdinsign) 1/13 0113 R2 0.25 0.29 0.29 0.32 0.32 0.42 0.38 0.33 0.43 0.34 0.40 N 258 258 258 258 258 250 258 228 245 220 215 RS Constant 3.00 2.94 3.10 3.18 3.20 3.10 3.54 4.18 3.14 4.03 4.44 Ln(hours) 0.47 0.49 0.50 0.50 0.49 0.46 0.47 0.22 0.55 0.21 0.24 Education(yrs) 0.08 0.08 0.08 0.08 0.08 0.07 0.07 0.08 0.06 0.08 0.08 Vocational -0.10 -0.09 -0.08 -0.08 -0.08 insign -0.90 -0.08 insign -0.15 Age insign insign insign insign insign insign insign insign insign Age- squaredx 100 insign insign insign insign insign insign insign insign insign Waitlisted insign insign insign insign insign insign insign insign Secondjob insign insign insign insign insign insign insign insign Tenure insign insign insign insign insign insign insign Public sector insign insign -0.27 Privatesector insign insign insign Informal -0.23 insign insign insign Company size insign insign insign wlindustries (sigdinsign) 0113 1113 R2 0.17 0.18 0.18 0.19 0.20 0.21 0.22 0.22 0.33 0.22 0.44 N 245 245 245 245 244 237 244 206 237 202 201 36 CHAPTER 3: THE CHANGIN~ INDUiSTRIAL RELATIONS FRAMEWORK 84. BH has made significant progress in reforming previously rigid labor market legislationto better serve the needs of the emerging market economy. The Labor Laws inthe two Entities are fairly harmonized, and both set a comprehensive but relatively flexible broad framework that is subsequently used for negotiations between the social partners. More specific wage and employment provisions are delegated to rather detailed general, branch and company agreements. In turn, the provisions o f collective agreements can be adjusted (and become more favorable to workers) by enterprise rule books and individual contracts with additional variations across Entities, sectors and cantons. 85. At the same time, the system of industrialrelations is evolving and is far from being coherent. For example, there is considerable inflexibility in some statutory provisions while there are also some regulatory gaps. To a large extent, the current laws and regulations seem to be the consequence o f "path dependence" and fulfill the need for a symbolic basis upon which the parties would agree to interact. Findings 86. While some or many of their provisionsmight be evaded selectively or by the labor market at large, collective agreements can still have a substantial negative effect in some industries and sectors. Collective agreements are clearly enforced in some public sector enterprises, and rigid regulations might also be responsible for the increase in the size o f the informal sector. But not all collective agreement provisions are observed, and not in all parts o f the labor market - although it is generally understood that all employers are bound by the general collective agreements, systematic enforcement o f branch collective agreements outside the broadly defined public sector has been limited. Many de novo private sector companies in the formal sector comply with only parts o f general collective agreements, if at all. Likewise, many State-owned and privatized companies, while formally covered by such agreements, avoid or evade some o f their provisions due to the state o f their financial conditions. For example, the heating allowance in the RS, although included in collective agreements, no longer appears to be paid by most enterprises. 87. Privateemployers retaina substantialdegree of flexibility compared to the majority State-ownedsector. Unless the Government decrees an extension o f the provisions o f collective agreements to non-participating companies, the private sector i s relatively free to set wages, hours, and other working conditions. Unless unions are present at the individual company level 37 and demand to bargain, "this collective bargaining situation - apart from the absence of a regulatory regime to establish and police bargaining rights - is notably flexible".36 88. At the same time, there are obstacles to smooth labor adjustment driven by gaps in legislation and inadequatepractices. The labor legislation in BHhas no provisions for layoffs that can help companies respond, at least temporarily, to shocks and thus be given a chance to recover and meet their financial obligations, while allowing workers inthe meantime to engage in alternative employment. Instead, labor adjustment in BH has historically taken the form o f (a) excess workers stopping work at SOEs when such work does not exist, while (b) remaining nominally, or fictitiously, on the books o f the enterprises and accumulating wage and social insurance arrears (while working, at least some o f them, in the informal sector). This was aided by the fact that labor regulations stipulate that the employer must retain a worker's workbook until separation of employment occurs, at which time he must stamp it with the termination date (thus testifying that social insurance contributions have been paid) and return it to the worker. Therefore, the effects o f a legislative vacuum (lacking policy for layoffs) are accentuated by an administrative vacuum arising from outdated arrangements for tracking a worker's social insurance history. 89. Flexibility through non-enforcement of collective agreements is generally undesirable. Non-enforcement o f rigid regulations leads to informality, which creates rents for officials and uncertainty among employers and new investors, and leaves workers without protection (such as social insurance ~overage).~' 90. The characteristics of the industrialrelations system are nevertheless changingover time both defacto and also because existing arrangements are being renegotiated. There is an emerging group o f companies in the private sector and new employers' associations. There are also changes in the trade union structure and, together with higher private sector participation, they have initiated negotiations for new general collective agreements to replace those agreed in 2000. In the FBH, the social partners agreed a new GCA in August 2005, which introduces a new system o f minimum wage determination, but which leaves many provisions unchanged compared to the 2000 GCA. Tripartite bargaining is being strengthened through the Entity ESCs, while the creation o f a State level Council is being anticipated. Negotiations seem to be focusing on the "price of labor" (such as minimum wage determination, rate o f overtime premiums, and whether the seniority allowance that a worker accrues while he works continuously for one company is transferable, in case o f a job change, to the new employer); however, bargaining needs to take into account what is being and what can be realistically enforced. Recommendations 91. BH still faces the need for substantial enterprise restructuringand adjustments in the workforce of SOEs and in mass or voucher-privatizedcompanies (MVPs). As will be discussed in this chapter, BH's overstaffing problem is specific and grave, as evident in the emergence o f a "fictitious employment" problem, which does not appear to be solvable by applying the current statutory provisions with respect to employment termination, such as severance awards. 36Firkin,2004. 37 This is consistentwith experience from elsewhereinECA. See World Bank (2005), "EnhancingJob Opportunitiesin TransitionEconomiesof Europeand CentralAsia". 38 92. The analysis in the report points to the following four main areas of policy reforms in the labor market of Bosnia and Herzegovina: A. Create a leaner but more enforceableset of labor regulations: e Allow for a gradual reduction of the FBH minimum wage as a share o f the average wage. The FBH minimum wage, currently at 55 percent o f the average wage, is one o f the highest in a regional comparison, and a gradual reduction could open the way for more wage flexibility, especially in companies under stress, and prevent the loss o f formal sector jobs and a further informalization o fthe economy. e Introduce a separate minimum wage for the youth. Commonly practiced in many countries, this measure may help address the currently high unemployment rates among the youth o f BH. e Rationalize and simplify the system of "coefficients" and other aspects o f wages, employment and benefit determination. For example, there is no obviousjustification for supervisors to have more credentials than their supervisees, or mechanical linkages o f wages to the education o f the workers or the portability o f seniority premia to new employers (see Box 3.1) e Reduce the duration of maternity leave and pay maternity benefits through the social protection system (instead o f effectively leaving it to the employer). The one-year maternity leave duration inBHstands out ina European comparison, and possibly acts as a barrier to female employment. With female labor force participation low in BH, the authorities mightwant to review the legislation and shorten maternity leave duration. B.Review the system of payrollcontributions e Integrate non-wage benefits into the taxable base in the FBH, where workers' allowances are neither transparent ("monetized") nor included in the social contributions base. This reduces the financing o f social insurance systems and leads to inequities. Harmonize systems of payroll contributions and rates between the Entities. C. Review systems of collective bargaining and wage determination e Separatepublic sector wage determinationfiom that in theprivate sector. The minimum wages prescribed by the current collective agreements- and employment conditions apply equally across employers irrespective o f corporate identity (such as state, cooperative, mixed or private), while most countries have separate laws on civil service employment including separate minimumwages and wage grids. e Limit collective agreement coverage to those enterprises that are represented in bargaining. To preserve the voluntary principle o f tripartism, the Government and social partners involved in bargaining could refrain from exercising its authority to extend collective agreements coverage to those who have not voluntarily signed them. e Define criteria for representation of social partners involved in tripartite bargaining. The labor legislation in BH does not set out clear minimum membership criteria and verification procedures for the representation o f social partners inbargaining. e Align legislation in the RS on "occasional" or "temporary" work with similar provisions inthe FBH. 39 93. BH's system of collective bargaining traces its roots from the former Yugoslav system of self-management. Under self-management, there were few companies that were private inthe conventional use o f the term. Wage and employment conditions were set mainly by Government and trade unions, with the Government settingthe firm's wage bill oriented by social considerations. The intra-enterprise income redistribution, in terms o f individual worker salaries and benefits, was then decided collectively within the company's overall wage allocation. The 1988 Yugoslav Law on Enterprises effectively abolished self-management and introduced the right of an employer to layoff workers, albeit basically at high cost. With respect to wage determination, the Law introduced a system o f new provisions through laws, collective agreements and incomes policy to replace the self-managed system38.A Yugoslav Labor Law was adopted in 1989 and fostered the role o f collective bargaining in wage determination. Subsequently, the Entities introduced their own Labor Laws in2000 which were supplemented by General Collective Agreements (in 2000 and 2005 inthe FBH; and in 1999 but amended in 2001 in the RS). The provisions o f the Entity Laws and Collective Agreements as well as the organization o f Collective Bargaining(tripartism) are examined below. Labor Laws 94. Boththe RS and the FBHhavesubstantially revampedtheir labor lawswith the aim of introducingmore flexibility and to promotejob creation. The Labor Law o f the FBHwas published in the Official Gazette in 1999 and substantially amended in 2000. The Labor Law o f RS was published in 2000. Both laws set the minimum labor standards3' and the broad framework for obligations and rights o f both parties (employers and workers), as well as for the determination o f wages and employment conditions (including collective bargaining). Both leave a lot to be decided at the more decentralized level from general and branch collective agreements to company rulebooks and the individual employment contracts4'. 95, Compared to previous legislation, the Laws increased employment flexibility by reducingobstaclesto hiringand firing and allowing contracts of limited duration and part- time employment. They also curtailed severance obligations and stipulated a mechanism to end the status of waitlisted workers (see below). The Laws provide for collective agreements and set rules for the representation o f workers and employers. Neither law mandates a continuation o f the system o f wage determination through labor coefficients (see below), some o f which are no longer appropriate for the BHlabor market. 96. The two Entity Labor Laws are parallelin many respects but deviate in others. A detailed comparison o f the Labor Laws is presented in Box 3.2. For example, inthe area o f labor inspection, the systems o f administration and enforcement differ. While the RS requires all contracts to be in writing, the FBH law permits a contract to be oral, and, if there i s no written statement governing its duration, the contract is presumed to be (in a contestable way) for an indefinite period41. The extent o f unwritten contracts in the FBH is not known, although consideration should be given to correcting this aspect o f the FBH law, in efforts to formalize 38Vodopivec (1991), The Transition of the Yugoslav Labor Market: Crossing the Rubicon of Job Security, mimeo 39For example, both Laws protect minors and forbid discrimination on a variety of grounds, suchas union membershipor activity, race, sex, religion, political opinion, and ethnicity. 40Inthe FBHalso Cantonal legislation. 4'Oral contracts may enable an employer to deny hidher commitments, although they may also give rise to unjustified claims by the worker. 41 employment and widen the coverage o f social insurance.42 In both the FBH and RS, full time employment may be considered temporary, but only if it does not exceed two years' service. If a contract is o f indefinite duration, termination must involve a cause and will require compensation (see below). The FBH treats contracts for seasonal and temporary work as contracts o f fixed duration. The RS limits the annual length o f contracts for work o f an "occasional" or "temporary nature" to a maximum 150 hours o f work per year. Both Laws permit part-time employment. And both limitthe amount o f overtime that can be required, with a minimumo f 30% premiumfor overtime pay in the RS, while in the FBH this figure is governed by collective agreement^^^. Lastly, although such limits can be restrictive, neither Law provides caps for the use o f contracts o f limited duration or of part-time employees as a percentage o f the employer's workforce. 97. The two Entities have provisions for termination due to economic reasons and collective redundancies. For example, the FBHLaw allows for the termination of employment for "justified economic, technical or organizational reasons". The employer should, however, exhaust all possibilities for the worker(s) to be internally transferred and retrained, if needed. For collective red~ndancies:~the employer should consult with the works council inthe enterprise or, in the absence of a works council, with all trade unions representing at least 10% of employees. Similar provisions are found in the RS Labor Law. Both Laws also provide that a contract o f employment may be terminated due to an employee's breach o f obligations, non-performance, absenteeism or misconduct. 98. This suggests that the Labor Laws per se do not unduly restrict flexibility in the labor market. Unless the Government introduces legislation that would enforce collective agreements to non-participating parties in the private sector, private sector employers are free to set their wages, hours, and other working conditions, the latter subject only to the statutory minima set out in the Labor Laws. Overall, some note that the major problem in BH is not the inflexibility o fthe systemo f employmentprotectionlegislation, butbecause "the Labor Laws are, to a considerable extent, ineffective and widely ignored".45 Perhaps as much as half the labor force works in the informal sector, resulting in a significant number o f employers not paying the employment-based taxes (even current wages), some perhaps engaging in overtime work without additional compensation, and others being compensated at a rate below what the law mandates. In effect, the difficulties seem to be related to the perpetuating structure o f the economy (where the formal sector i s dominated by SOEs and where the informal sector i s large), as well as the continued reliance on prior modes o f thought (as in the case o f wage coefficients o f the BCAs) that are no longer appropriate for a market economy. 42Finkin(2004). 43The overtime premium stands at 50 percent though it is currentlysubjectto renegotiation. 44Collectiveredundancies refer to the terminationof employment of 10percent or more of workers (but no less than 5 employees) in workplacesof morethan 15 employeesover a periodof 3 months. 45Finkin2004. 42 43 CollectiveAgreements 99. In line with the Labor Laws, general and branch collective agreements set the details of the labor market framework such as mechanisms o f wage determination and adjustments, working time, worker protection and benefits, regulations governing termination of employment and conditions for trade union activities. The provisions o f collective agreements at general and branch level are extensive, as indicated in Table 3.1 (and specific details are presented in Appendix 4). In the FBH, all the Labor Law provides is that collective agreements can be territory-wide or not, and multi-employee or not; they can be bargained by one or more employers and by one or more unions; and they bind only those employers and those union members who have executed them. The RS law states that agreements are "mandatory" for those who were represented in the process but allows the same flexibility o f participation as in the FBH. Thus, both Labor Laws prescribe that collective agreements are binding only for those parties who have directly participated or have been represented in the process o f concluding the agreement^^^. Table 3.1: Indicative List of Benefitsin BranchCollectiveAgreements Benefitcategories and benefits Entity Standard Supplements 1 Seniority - increment for each additional year (transferableto new employer) both 2 Annual leavegrants both 3 Hot meal allowance both 4 Transportation to work (GCA in FS, rule book in FBH) both 5 Fundsfor heatingand winter food providing RS Hours of work (except during normal workweek) 6 Overtime both 7 Night work both 8 Work on weekends both 9 Work on state holidays FBH Away from duty station 10 Field work (ifemployeeis not coming back home) both 11 FixedSupplements both 12 Per diem - for trips both 13 Living separate from the family - for morethan2 months(basedon lastminFederalwage) FBH Conditions of work, health and safety, injury 14 Exposureto noise, dust, positionedvapor Both 15 Work where additionalsafety equipmentis required Both 16 Increaseddanger of explosion, flood etc RS 17 Dangerof gettingsevere injuries FBH Leave 18 Paidleave for holidays Both 19 Participationincultural and sportevents FBH 20 Blood donating Both 21 Seekingfor close family membermissingin the war FBH 22 Seekingof the close family member FBH Illness/Inj ury/disability 23 Sick-leavedue to injury at work (based on last month'sbasicwage (no allowances)) FBH 24 Heavy disability of the employee RS 25 Long illness or injury o fthe employee RS 26 Illness of the close family member both Marriage/ Birth/ Death 27 Marriage (own) both 46 However,the Government can extend the coverage ofthe agreements, or parts thereof, acrossthe economy. 44 Benefit categories and benefits Entity 28 Child birth both 29 Marriageo fthe child FBH 30 Deathof the employee Both 3 1 Deathof the retiredemployee(based on last reportedaverage Federalwage) FBH 32 Deathof family member (grandparent,parent, in -laws, spouse, child grandchild brother or sister) Both Other 33 Moving households Both 34 Education FBH 35 Social benefits Both 36 Special stimulations (voluntary) Both Worker Separation 37 Natural disaster Both 38 Temporary redundancy . . RS 39 Dismissal(based on worker'slast monthlywage -- foreach year of work) FBH 40 Regulatingthe rightto retirement Both Source:Select Entity branchcollectiveagreements 100. Nevertheless, according to the GCA in the FBH, all employers, public and private, are in theory bound by the provisions in the general collective agreement4'. There are differences o f opinion as to whether and the extent to which collective bargaining is binding. In the FBH, collective agreements can be territory-wide or not, and bind only those employers and those union members who have signed them. The RS law i s more cryptic, stating that agreements are "mandatory" for those who were represented in the negotiation process4'. In practice, the currently valid GCAs have an economy-wide coverage, while BCAs are largely limited to the SOEs and privatized former State-owned sectors. Inthe RS, the GCA was signed on behalf o f all employers by the Chamber o f Commerce4'. Although it is generally understood that all employers are bound by the GCAs, effective coverage o f the BCAs outside the broadly defined public sector5' has been limited, especially in the de-novo private sector5'. Only recently, steps have beentaken inthe RS to introduce a separate system to determine wages and work conditions inthe civil service (see Box 3.3). 47 According to Article 2 of the new FBH general collective agreement, "The Collective Agreement is binding and it applies to all economic companies regardlessof the ownership structure of capital in them, administrativebodies and services for administration, public institutionsand other legal subjects" 48Again, one may argue that since the Chamber of Commence inthe RS negotiatedon behalfofthe employers and membership to the Chamber is compulsory, there is potentialfor interpretingthe coverage ofthe GCA as compulsory for all. 49 The amendedArticle 81 of the RS general collective agreement, signed on behalfof employers by the RS Chamber o f Commerce, states: "This collective agreement is relatedto all employedin RepublikaSrpska regardlesson character o f ownershipand form of employer and trade unionorganizing. Includingthe civil serviceas well as publicly ownedindustrialand service sector companies. "AtthetimeofsigningtheBCAs,thesizeofthedenovoprivatesectorinBHwassignificantlysmallerthantoday. Many new privateemployers are currentlyfollowing their own company rules on wage determination and worker entitlements. 45 46 supposed to be in place for one year. However, it contained a provision that, upon its expiration, it was to remains in force until a new agreement could be reached. This has been the case until the new GCA was signed in August 2005, and which now has an open-ended duration. The current RS GCA has an open-ended duration as well. The FBH Labor Law leaves it to the collective agreement itself to specify rights o f parties to walk away from the agreement, and the currently valid GCA allows all parties to start the initiative to amend the agreement. The RS Labor Law allows each party to terminate the agreement upon a notice period o f 30 days, but defers to the agreement itselfto specify details. Organization of CollectiveBargaining 104. Until recently,the system of industrial relationsin Bosniaand Herzegovina has been almost exclusively dominated by the Government (in its dual capacity as Government and employer) and public sector trade unions. Breaking with this tradition o f Government domination, the private sector employers' association was co-signatory o f the most recent GCA was signed by the in July 2005, alongside the Governments and the trade unions. However, the previous FBH general and branch collective agreements o f 2000 were signed only by the Government and trade unions. In the RS, the Government, the Chamber o f Commerce and the trade unions were signatories o f the general and branch collective agreements in 1999 and amended in 2001. With the new FBHGCA retainingmany o f the provisions o f the old GCA, this means that effectively, the core o f the collective agreements pre-date the introduction o f the new Labor Laws in 2000. 105. Both parties can be represented by one or more trade union and employers' association, while the signatory representative needs to have the authorization of all participating parties. The FBH Labor Law has provisions which allowed the FBH Entity or Cantonal Governments to represent employers prior to the formal establishment o f employers' associations. In the RS, the parties to the collective agreements on both workers' and employers' side are those organizations with the largest membership, but with written consent o f the other associations. 106. I n legal terms, there is notable flexibility on how wages, hours, and working conditions are set by the collective agreements, company rulebooks or individual employment contracts. Wages, hours, and working conditions may be regulated (above statutory mandates) by collective agreements at national or branch (multi-employer) levels, or by collective bargaining at the enterprise level if the employees are represented by a union and demand to bargain. Therefore, where unions are present and demand to bar ain, wage setting may be done on a consensual, multi-employer basis or on an enterprise basis! Alternatively, in the absence o f a binding collective agreement, wages and working conditions may be set by the employer in its rulebook or in the individual contract. These provisions and the absence o f a regulatory regime to establish and police bargaining rights are notably flexible54. 107. Both Entities have formed their own ESCs and preparations are under way for the creationof such a body at the State level. Meanwhile the councils have emerged as the venue for tripartite collective bargaining and discussion o f economic and social trends in the labor market in BH. The ESCs convene Government, trade unions and employers associations. Tripartism i s now benefitingfrom more representative independent and private sector employers' 53Althoughdifferenttypes of collective bargainingcan leadto the same or different outcomes in the labor market, there seems to be generalconsensusthat the combination of multi-unionismand separate collectivebargainingcan lead to inferior outcomes, for example, in terms o f productivity. See Annex 1. 54Finkin (2004). 47 associations arising from the growth o f a new and mostly small and medium-sized private sector. However, while the FBH ESC meets regularly (typically on a monthly basis), the ESC inthe RS has not been meeting frequently after the breakdown o f negotiations over a new GCA in Spring of 2004. There have also beenmoves by both trade unions and employers' associations to create a State-level ESC55.While this council has not been officially created by the Council o f Ministers yet, the employers associations in both Entities and the BrCko District have jointly founded the BHAssociation of Employers5` in September 2004 and the two Entitytrade union confederations formed ajoint State-level Confederation inMay 2005. Trade Unions 108. Trade unions in both Entities mainly represent workers employed in public sector companiesand mass-privatizedenterprises as well as civil service employees. There are two separate trade union confederations in BH, one in FBH and another in the RS. In the FBH, the Confederation o f Independent Trade Unions o f Bosnia and Herzegovina (SSSBH) has about 260,000 members, which is about 68 percent o f the estimated workforce in the FBH. It i s comprised o f 23 industrial branch trade unions. The Confederation o f Trade Unions o f Republika Srpska (SSRS) has a membership o f 157,000, which represents 66 percent o f the workforce in the RS, and gathers 14 branch trade unions (See Figure 3.2). Membership has been declining as workers became either unemployed or surplus and stopped receiving wages ("fictitious" workers). This has consequently reduced the ability o f unions to collect membership fees5'. In addition, some branch unions broke out o f the confederation structure (for example, the agriculture union in the FBH and union o f metal and mine workers in the RS). As a result o f these changes, both confederations have been entertaining plans to consolidate the branch trade union system by merging affiliate branches. 55This is in linewith the reformagenda put forward in the PRSPIMTDS. See Appendix 3. 56The BHAssociation of Employers comprisesthe RS Employers Confederation,the Union of Employersin FBH and the Female Employers'Confederation UNA Zenica. 57For example, the SSRS estimates that their true membershipcomprisesan additional70,000 workers who are not receiving salaries andare not payingmembershipfees and are thereforenot includedin the total numbers. 48 a 109. The Labor Laws provide for the establishment of works councils but, as in the case of arbitration, only the bareststructures for future institutionaldevelopmenthave been laid down. As yet, it appears that few, ifany, have actually been created and, ifexperience elsewhere i s instructive, it is unlikely that employees, especially in the smaller enterprises, will feel a need to create them. In the event that the obligation o f these works councils is one o f information sharing and consultation (akin to the obligation imposed by the EU directive on European works councils), the transaction cost attendant to consultation with a works council or a union would appear to be relatively low. Thus, there might be benefits to be gained in effecting a smoother transition for those who will lose their jobs. Similarly, worker representation in firms through work councils might be an efficiency-enhancing mechanism because it helps to reduce transaction costs by improving communication and intermediation. Some research confirms this view, but others have raised doubts why work councils might be better poised to circumnavigate the problems o f incomplete contracts and opportunistic behavior. Probably, the most researched case in the area o f economic implications o f work councils is that o f Germany but, if anything, the evidence is partial and inconcl~sive.~~ Employers' Associations 110. The private sector employers' associations have been gathering strength over recent years and are now involved in collective bargaining, but their representativenessremains limited. The FBH Union o f Employers (UEFBH) represents approximately 4,100 private companies that employ around 90,000 workers mainly in Central and Western Bosnia and Herzegovina5'. The UEFBHi s geographically organized (not by sector or industrial branch), and i s a confederation o f regional small business associations (see Appendix I1 for membership organization). In the RS, the Employers Confederation (ECRS) represents the small but growing private sector, mostly from the wider Banja Luka area. It comprises mainly private small and medium enterprises (SMEs) and voucher-privatized former state-owned companies. Founded in September 2000, the ECRS i s small compared to the UEFBHand, while it represents more than 800 companies, only 178 companies are membership fee-paying members and are employing only 6,000 people6'. Its member companies are mostly active in manufacturing sectors (such as wood, paper, food and drinks, and the metal industry)as well as information technology, services, construction and wholesale sectors. While it has two collective members, the Association o f Poultry Growers RS and the Association o f Animal Food Producers, the ECRS, despite its name, i s not organized as a confederation o f constitutive associations, but o f individual member companies. Alongside the ECRS, the RS Chamber o f Commerce (RSCC) participates in collective bargaining in the RS on the employers' side, representing mainly voucher-privatized enterprises as well as companies with majority state ownership. In fact, employers in RS have previously been exclusively represented in collective bargaining by the RS Chamber o f Commerce.61 The coexistence o f the ECRS and the RSCC suggests that the consolidation o f the employers' representation in RS has still some way to go62. '*Addisonand Wagner (1994); Rogersand Streek (1994). 59 UEFBHestimate, althoughit is not clear how many companiesare actually payingmembership fees. 6o Accordingto ECRSmembershipdata. 6' WithRSCCmembership beingmandatory, there are concerns that the RSCC's representationincollective bargainingmay be out of linewith boththe RS Labor Code andILO conventions. The RSCCis not part ofthe State-levelBHEmployers'Association. 50 111. The Labor Laws provide for a broad framework for wage determination, but defer the process and rules to the collective agreements, company rulebook or the employment contract. The Laws stipulate that salaries paid cannot be lower than the ones stipulated in the collective agreements, rulebooks and employment contract. They also provide for the right o f salary increases for overtime, night work and work in difficult circumstances, however leave the determination o f levels and methods o f payment to the collective agreements and contracts. The GCAs stipulate a large number o f wage and non-wage benefits. Inboth Entities, they regulate the minimumwage or minimum price of labor as well as its indexation, but leave the determination o f individual wages to the BCAs according to a system o f complexity coefficients. Overall, 40 or more wage and non-wage/employment benefits are prescribed in the GCAs with additional provisions in branch and company agreements (See Table 3.1 and Appendix I). Minimum Wage 112. The system of minimum wage determination in the FBHis in the process of change. InAugust 2005, the social partners inthe FBHagreed on a new GCA. The new GCA stipulates a minimumwage on an hourly basis both ingross (KM2.96) and net (KM 1.75) terms, and is to be adjusted "with increases in the costs o f living, growth o f retail prices and overall economic development" at least once per year by a decree o f the FBH Government and on the basis o f harmonized positions o f the social partners in the FBH ESC. This minimum wage i s to set the base for the coefficient-driven wage determination system set out in the BCAs. The agreed net hourly minimumwage o f KM 1.75 translates into a monthly minimumwage o f KM 308, if based on a monthly working time o f 176 hours. This represents 55 percent o f the average wage paid in the FBHinJune 2005 (of KM556), 58 percent o fthe 2004 FBHaverage wage and 166 percent of the 2004 poverty line63. However, this net wage excludes wage and non-wage allowances. The new GCA also introduces a provision, according to which "in exceptional circumstances" BCAs or company agreements can stipulate conditions, level, method and periods o f determination o f an alternative lower minimumwage below KM 1.25 net per hour. A net minimumwage o f KM 1.25 per hour would translate into a monthly minimum net wage o f KM 220, representing about 40 percent o f the average wage paid inthe FBHinJune 2005. 113. Despite the replacement of the previous spiraling minimum wage indexationrule in the FBH, the level of the minimum wage remains high. The previous FBH GCA stipulated that the minimumwage was to be determined by the BCA, but that it could not be lower than 55 percent o f the average wage as published by the FBH Statistical Office and was adjusted monthly64. While the new GCA removed the spiraling average wage-linked indexation o f the minimumwage, the level o f 55 percent o f the average wage remains. Figure 3.4 presents data on the minimum/average wage ratio for selected countries in ECA and shows that the FBHhas the second highest minimum/average wage ratio (after Lithuania) within the said countries in Central, Eastern and South-Eastern Europe. 63The poverty linefor BHfor 2004 has been set at KM2223 per capita per year (Memic, 2005). 64Therefore, the minimum wage level was determinedas a share ofthe average wage level; howeverthe latter was ultimately determinedrigidly on the basis ofthe minimum wage through the coefficient system set out in the branchcollective agreements. 51 Figure3.4 The FBH minimum wage is one of the highest in the regionas a share of the average wage Minimumlaveragewage ratio,2002 (YO) 0 10 20 30 40 50 60 70 Source:World Bank(2005) and own calculations 114. Some of the BCAs stipulate even higher minimum wages. For example, minimum wages in the telecommunication and postal service sectors range between 100 and 110 percent o f the average wage inthe FBH, while they vary between 55 and 65 percent in other sectors. While the minimum wage in the textile sector is set at 55 percent of the average wage, anecdotal evidence suggest that a number o f textile companies cannot afford to pay at that level and are in fact paying lower wages, as do companies in other sectors as well65. 115. The new FBH provisions on the minimum wage are encouraging, but what matters is their implementation. The introduction o f a lower net minimum wage o f KM 1.25 per hour, even if limited to "exceptional circumstances", could open the way for more wage flexibility especially in companies under stress, which might prevent the loss o f formal sector jobs and further informalization o f the economy. Likewise, the new minimum wage adjustment mechanism allows bringingminimumwage growth more into line with the broader development of the economy instead o fjust indexing it to the average wage growth. However, much depends on the implementation o f the minimum wage adjustment: Will the Government emerge as the dominant factor, even though the decree stipulating indexation is to be based on a consensus among the social partners? Will the Government issue decrees in raising the minimumwage even ifthe social partnershavenot agreed? 116. In the RS, the minimum price of labor is set at a low level and does not constitute an effective binding minimum wage. The RS minimum price o f labor i s determined by Government decision and has been stable at around KM 80 per month since 2000.66 It is currently set at KM 82, which i s 44 percent o f the BHpoverty line and 19 percent o f the average 65FBHMinistry of Labor and Social Policy, Informationon the Application ofthe General Collective Agreement for the Territory ofFBH, July 2004 The GCA for the RS is far more detailed than for the Federation. It is executed by the Government and the trade union confederation, but also by the Chamber of Commerceand not, notably, by the employers' confederation. Article 42 requires the parties to agree upon the establishmentof the "lowest wage hour," failing which the government ofthe RS is to define it. Articles 39 and 44 through 47 provide for agreement on subsequent coefficients dependingon the employees' functions andworking conditions. (Article 50 allows employers with insufficientfunds to pay a guaranteed wage of 65 percent of the wage the employee would be entitled to under the system albeit for only a six month period. 52 wage in 2004. At such a low level, it is limited to providing the base for the coefficient- determined wages according to the provisions in the BCAs. Recent amendments to the RS Law on Social Contributions have raised the minimum social insurance taxable base from the minimum price of labor to the level o f 50 percent o f the average wage, while leaving the minimumprice of labor at Kh482. 117. Analysis of LSMS data shows that the FBHminimum wage is not fully binding. As the analysis inChapter 2 showed, a sizeable share o fworkers inthe FBHreport earnings less than the minimumwage, while the share o f formal sector workers receiving lessthan 50 percent ofthe average wage are roughly equal in both Entities. This is despite the fact that the RS and FBH have fundamentally different minimum wage levels. A binding minimum wage in the FBH, standing at 55 percent o f the average wage, would have predicted a much lower share o f formal workers reporting less than 50 percent o f the average wage than in the RS. The detailed analysis o f this is presented in Chapter 2 o f this report. WageDetermination 118. Both Entities' collective agreements retain a "coefficient-based" and public sector- oriented wage determination system. The RS GCA stipulates the system o f wage determination based on a `lowest hourly wage' which i s multipliedby coefficients between 1 and 6, depending on the complexity o f the performed work or the worker's educational attainment. The GCA specifies further that an individual's wage can be raised by a further 20 or 30 percent above the basic wage, calculated through the coefficient method, if the individual is working under "difficult work conditions" or "performs the most complex functions in the economy, social activities, state bodies and organizations". The BCAs in both Entities specify in details the criteria and corresponding coefficients for different levels o f work complexity or educational attainments. The coefficients, and the corresponding wage spread, range from 1to between 6 and 8 (see Appendix IV for more details). 119. While binding for the State-owned and mass-privatized sectors, this wage setting rule has often not been adhered to in the new private sector. A core element o f the BCAs, this coefficient-based wage determination system has remained unchanged for the last 5 years. While being eroded in the State-owned sector for reasons o f underperformance o f many enterprises, anecdotal evidence suggests that it has played a minimal role in determining actual wages in the new private sector, where companies have followed their own rules. A notable feature o f the Labor Law in the RS is that it allows employers with insufficient funds to pay a guaranteed wage o f 65 percent o f the wage the employee would be entitled to under the system, albeit for only a six month period. Wageand non-wage benefts 120. While GCAs and BCAs prescribe very detailed provisions on wage and non-wage benefits, there is anecdotalevidence that some of these benefits are no longer actually being paid. The GCAs and BCAs stipulate an extensive range o f additional wage and non-wage benefits (see Table 3.1). There are diverging patterns in the structure and content o f collective agreements in both Entities. Inthe Republika Srpska, the GCA contains very detailed provisions, and BCAs are largely synchronized with the GCA in terms o f wage and non-wage benefits. The only area that the RS GCA does not explicitly legislate is the definition o f the wage coefficients. Inthe FBH, the GCA is less detailed in its provisions than in the RS, and defers many issues to the BCAs. Moreover, the BCAs are less synchronized than in the RS, and for example provide substantially varying levels o f the minimum wage ranging from 55 to 110 percent o f the average 53 wage (See Appendix IV). While regulated in GCA and BCAs, according to anecdotal evidence some o f the allowances are not paid by many companies or are paid in a manner not prescribed in the agreements, such as the winter heating allowance and the annual leave grants. 121. Non-payment of such benefits is confirmed in the earnings analysis. The analysis in Chapter 2 reveals that many wage and non-wage benefit provisions (such as those relating to seniority/tenure as well as the minimum wage) do not appear to have (at least a statistically significant) impact on earnings. e n 122. This section reviews the major employment protection provisions in the RS and in the FBH Labor Laws and collective agreements, covering termination o f employment, severance, worker waiting lists, temporary and part-time employment, the employee workbook, working hours and overtime provisions, as well as leave and maternity leave provisions. It discusses the progress achieved with the introduction o f the new Labor Laws and identifies remaining problems, contradictions, gaps and shortcomings. Termination of employment 123, While significantly easing provisions on employment termination, the Labor Laws retain some surprising constraints. For example, workers who suffer an injury at the workplace or develop an occupational disease duringthe period of temporary incapacity to work cannot be laid off. Ifthe worker regains his or her work ability after treatment, he/she is entitled to return to the previous job or to be assigned to another appropriate job in line with remaining work capacity, whether such a position exists or not. While the RS Labor Law rules out the possibility o f layoff o f an employee based on remaining work capacity or on immediate risk o f occurrence o f disability, the FBH Law allows this, however it would require the consent o f the workers' council (which is still not functional). With respect to collective dismissals, the previous requirement for consent by trade union or workers' council to the dismissals has been removed, while maintaining the requirement for an employer to consult, rather than seek consent of, trade unions or workers councils in companies with more than 15 workers in case of collective dismissals o f more than 10 percent o f the workforce over three months. These consultations are to include information from the employer to the workers' council on the reasons for termination, the number and qualification o f affected employees, measures identified which might avoid termination (e.g. by re-assigning staff), and measures which would help employees to be recruited elsewhere6'. While this process might require additional time compared to an individual dismissal, the laws provide no obstacle to the use of collective dismissals. 67Laid-offworkers also retain a priority for re-hiring comparedto other workers with similar qualifications. 54 Noticeperiod in case of dismissal 124. The Labor Laws provide for minimumprovisions interms o f notification periods in case of dismissal, but allow the collective agreements to regulate further. The Labor Laws prescribe a prior written notification (30 days in the RS and 14 days in the FBH) by the employers to the employee which should include reasons for said termination. The notification period may be waived if the reason for termination is serious misconduct or breach o f obligations by the employee. The Laws open the possibility for collective agreements to prescribe longer notice period for staff having served longer, with the RS GCA stipulating a minimum notice period o f four months for employees with over 30 years o f service. While such an extension is standard in many countries, it may provide further obstacles to the adjustment in SOE and MVP workforces. Severance 125. The Labor Laws defer the determination of the level of severance to the collective agreements, rulebooksand employment contracts, but set a minimum rule for severance to equal not less than one third o f the average monthly salary o f the employee, as paid in the last three months before dismissal, for each full year o f employment with that particular employer. However, the RS GCA, in an amendment agreed in 2001, raises the amount to 40 percent for individuals with an employment history o f upto 10 years, 50 percent for individuals with 10 to 20 years o f work history, 60 percent for those with 20 to 30 years and 70 percent for those with more than 30 years o f service at the same employer. The substantial overstaffing in BH's enterprise sector and the fictitious worker problem implies a need to develop options to further relax severance provisions for regular and waitlisted workers. Waitlisted workers 126. BothEntity Labor Laws introduceda sunset clause for waitlisted workers; however, they have failed to resolve the waiting list issue. Worker waiting lists were an SFRY-era tool to deal with excess workers intimes of economic slump as well as those whose employment had ceased in the period between December 31, 1991 and the adoption o f the Laws in 2000 for political and ethnic discrimination reasons. According to the Laws, workers could remain in waitlisted status for no longer than 6 months (3 months in RS) from the day the Law took effect in late 1999, if the employer did not call the employee back to work before the expiry of this deadline. Affected individuals, if they had not in the meantime been employed elsewhere, had the right to a severance package if they approached the employer in order to re-establish an employment status by mid-2000 in the FBH and end-2000 in the RS. The Labor Laws provide for special severance regime limits for waitlisted workers, and limit severance pay to two-thirds o f the average wages at the level valid when the law/amendments entered force, as measured by the Statistical Institutes. This amount is multipliedby a coefficient o f 1.33 for those workers with work history o f up to 5 years, by 2.00 for those with work history between 5 and 10 years, by 2.66 for those with work history between 10 and 20 years, and by 3.00 for those with work history o f more than 20 years. The FBH legislation lumps together waitlisted workers and those who have effectively been dismissed after December 31, 1991 due to political and displacement reasons. The RS Law treats these separately - for waitlisted workers it simply stipulates a severance o f one to four times the average salary paid in the RS in the previous three months depending on length o f service, while it replicates the FBH's 1.33-3.00 coefficient solution to the `1991workers'. However, while waitlists are nowabolishedunder the laws, inrealitythey have continued in many companies (even with lower severance payments than for regular workers), and have contributed to the emergence and deepening o f the overstaffing and "fictitious" worker problem (see Text Box 3.5). 56 57 hourly minimum wage. While this is expected to help promote part-time and temporary employment, it needs to be complemented by a change to the employee workbook system. Employee Workbook 128. The Labor Laws specify the right of an employee to a workbook, which accountsfor the employee's pensionable years of service. The workbook is kept by the employer for the duration o f employment and is released to the employee at the time o f employment termination duly stamped to indicate the dates o f employment. Restingwith one employer for the duration of, employment, the workbook becomes an obstacle to part-time and multiple or parallel employment. Moreover, according to the GCAs, employment can only be terminated once all outstanding liabilities have been paid to the worker, and only at that time is the workbook released to the worker. This means that a worker i s taken hostage by the workbook in a situation where employment has effectively been terminated, but not all outstanding liabilities have been settled, and the worker cannot move on to new employment without the released workbook, such as inthe case o f "fictitious" workers. Working hours and overtimeprovisions 129. While the Labor Laws provide flexibility on working hours and overtime, the collective agreements are more rigid. Both Entity Labor Laws introduced maximum weekly working hours o f 40 for full-time employment. However, without specifying the terms, the RS GCA mandates the employer to shorten work hours to 10 hours per week for workers with physical and psychological health concerns that are potentially harmful to the said employees' health. Also, while the Laws protects minors, pregnant women, mothers o f children younger than 3 years o f age, as well as single parent o f child o f below 6 years, from overtime work, affected individuals are nonetheless allowed to work overtime if they so choose. The RS GCA, however, does not provide for this individual choice, and exempts people with disabilities (disability categories o f above 70 percent) from overtime, and sets a limit to overtime o f 10 hours. Salary increase for overtime cannot be less than 30 percent o f an individual's wage, while RS GCA raises this to 35 percent. Leave and maternity leaveprovisions 130. While the Labor Laws contain general provisions on leave entitlements, it is the collective agreements in which these entitlements are spelled out in detail and where additional entitlementsare granted. The Labor Laws specify minimum annual leave duration of 18 days, while under-aged employees are entitled to a minimum o f 24 leave days and those working in harmful conditions to 30 days. The Laws also mandate that annual leave be taken in two parts and explicitly prevent the employee from waiving annual leave. The Labor Laws also leave it to collective agreement, company rulebook or individual employment contract to stipulate more generous leave provisions. The GCAs specify the workers entitlement to an annual leave benefit. In the FBH, this entitlement is in the amount o f 50 percent o f the average wage paid in the company in the previous month, however only in case the company was profitable in the previous year, while in the RS it is three minimum wages paid in the company in the previous month. In addition, the BCAs specify a number o f additional paid leave days for marriage, child birth, death o f a family member, blood donation and other a~tivities~~. 69For details, see comparisonof generaland branchcollective agreements inAppendix IV 58 131. Both Entities provide generously for maternity leave, with mandated benefits being almost twice the commonly found levels in other counties (Figure 3.5). The Labor Laws in both Entities require employers to grant a woman during pregnancy and after childbirth (Le., for child-rearing) twelve months o f leave for each child and thereafter, with reduced hours upon returning to work until the child is two years o f age. Like other social welfare and child protection benefits, maternity benefits in the FBH are legislated to be financed by the Cantons. However, not all Cantons have passed the relevant Cantonal legislation, and yet fewer Cantons are actually paying maternity benefits. In fact, Sarajevo Canton is the only Canton paying regular child care and social welfare benefits. Sarajevo Cantonal legislation sets the maternity benefit at 60 percent o f the mother's average wage over the previous 6 months. In the RS, maternity leave i s twelve months for the first child and eighteen months for each succeeding child. Maternity pay during such leave is at full pay (calculated as the average over six-months prior to giving birth), adjusted monthly for any increase in salaries paid in the RS. The costs are paid by the Public Fund for the Protection o f Children, although the employer may bear some costs, too (for example, pay for the first four months o f leave). This reimbursement is limited to those companies who have paid the child protectioncontribution levied on the payroll. Figure 3.5 Bosnia and Herzegovina'smaternity leave duration stands out in comparisonwith other European and OECD countries,1998-2002 Maternity Leave Duration in Weeks - Bosniaand Herzegovina Norway Albania Macedonia Croatia ca:ech Republic Bulgaria Hungary M y Estonia UK Ireland Finland Denmark Greece Canada Romania Switzerland Spain Poland Netherlands Luxerbourg Austria Belgium Germany us Turkey Portugal 0 10 20 30 40 50 60 Weeks Sources: The Clearinghouseon InternationalDevelopmentsin Child, Youth andFamily Policies,ColumbiaUniversity New York, Paci (2002) and RS andFBHLabor Laws 59 Benchmarking Bosnia and Herzegovina'sEmploymentProtection 132. Cross-countrybusinessenvironment analysisshows that employers inBHface more rigidityin the hiringthan in the firing process in a regional comparison. Figure 3.6 presents rigidity indices from the 2005/2006 "Doing Business" survey for BH as well as averages for transition countries in ECA and for OECD countries. The indices measure how difficult it is to hire a new worker, how rigid the regulations are on working hours, and how difficult it is to dismiss a redundant worker. Conditions coveredby the indices include: availability of part-time and fixed-term contracts, working time requirements, minimum wage laws, and minimum conditions of employment. Each index assigns values between 0 and 100, with higher values representingmore rigid regulations. The overall Rigidity of Employment Index is an average of the three indices. For BH, the overall index is 42, down from 49 in 2004/2005, compared with the regional average of 44.3 and OECD average of 35.7. Firing costs are calculated on the basis of the number of weeks worth of salary in severance, and notification and penalties that must be paidto dismiss a worker.70 Figure 3.6 BHscores worse than regional comparators on an index of hiring difficulty, but compares better on firing rigidity and working hours Employment Rigidity in 2005 60 50 0 Bosnia and 40 herzegovina -4 X 30 rn Region C 20 0 OECD 10 0 Wficulty of Rigidity of Difficulty of Rigidity of Hiring cost Firing costs Hiring Index Hours Index Firing Index tirployment (% of salary) (weeks of Index wages) I Source: World Bank (2005), DoingBusiness2006; Highernumbers indicate greater regulation 133. The Doing Business data revealthat, despite their formal introductionin the Labor Laws, temporary and part-time employment still encounter obstacles in BH. As Figure 3.6 shows, Bosnia and Herzegovina compares unfavorably to the Europe and Central Asia regional average as well as to the OECD average in terms of difficulty of hiring7'. The high score on difficulty of hiring is in part driven by the limited availability of term contracts. Analysis in this ''Worlddifficulty ' O Bank (2005), Doing Business2006 The of hiring index measures (i) whether term contracts can be used only for temporary tasks; (ii) the maximum duration ofterm contracts; and (iii) ratio of the mandatedminimum wage (or apprenticewage, if the available) to the averagevalue added per worker. A country is assigneda score of 1 ifterm contracts can be usedonly for temporary tasks, and a score of 0 ifthey can be usedfor any task. A score of 1i s assigned ifthe maximum duration ofterm contracts is 3 years or less; 0.5 if it is between3 and 5 years; and 0 ifterm contracts can last more than 5 years. Finally, a score of 1 is assigned ifthe ratio of the minimum wage to the average value added per worker is higher than 0.75; 0.67 for aratio between0.50 and 0.75; 0.33 for aratio between 0.25 and 0.50; and 0 for a ratio less than 0.25 (World Bank (2005), Doing Business 2006). 60 report shows that while the Labor Laws introduce the concept o f temporary contracts, their use is limited to 60 days per calendar year. Moreover, the collective agreements remain without any provisions on temporary and part-time employment, and that certain practices (such as the workbook) actively undermine the use o f part-time employment. 134. The evidence on firing difficulty does not mean room for complacency, given that BH's enterprise sector downsizing is still ahead. The Doing Business findings on firing need to be treated with caution. BH's provisions appear to compare favorably with other transition countries, but not with high income OECD countries. Moreover, as discussed earlier in this Chapter, few countries face an overstaffing problem as pressing and overwhelming as BH, evident in the fictitious worker problem. This suggests that the authorities need to carefully look further at possibilities of relaxing the firing regulations so as to facilitate the much needed labor force adjustments inthe enterprise sector. 135. While the system o f industrial relations in BH covers practically all legal and regulatory aspects o f the labor market, there are some areas that are either not well-defined or totally absent. The case o f work councils and arbitration has already been mentioned. Below we examine the incomplete representation o f social partners intripartite arrangements, the partial determination o f minimum wages in the FBH, the absence of lay off regulations in both Entities, the lack of alternative to the workbooks as well as the lack o f measures that would insure employers from certain risks (such as natural disasters or inability o f employees to carry on with their duties). TheRepresentativenessof Social Partners and Issues in Collective Bargaining 136. The representativeness of the trade unions and employers' associations remains rather undefined. Labor legislation does not specify nor set out clear regulations on minimum membership and verification criteria o f the changing representativeness o f social partners.72 For example, the de novo private sector is expanding over time, while recent years have seen the erosion o f the representativeness o f the central trade union confederations. This report notes that the issue of representativeness is essential for tripartite arrangements and the credibility of collective bargaining. This suggests there i s a need for the Government and the ESCs to set minimum representativeness standards over the medium-term, and also to promote consolidation o f employers' association and trade union structures. With the situation in a flux, however, this can only happen gradually and without stifling the development o f the emerging new employers' association structures. 137. The Labor Laws do not contain detailed provisions on enterprise-level bargaining between trade unions and employers. According to the Laws, employers with more than 15 employees have to adopt a company rulebook to regulate salaries, workplace organization and other relevant issues, after mandatory consultation with the trade union(s). Moreover, an employer with more than 15 employees has to consent to the formation o f a workers council, if20 percent (FBH) or 33 percent (RS) o f the workforce so demand. Where there i s no workers' council, the trade union formation in the company has the rights and obligations o f the workers council. According to the FBHLabor Law, the role o f the workers' council i s advisory in nature, and the council can submit comments and proposals on the company rulebook to the employer. ''TheLabor Laws acknowledge a role for arbitration, although this has not yet been implemented. 61 However, the law does not have detailed provisions on the organization o f collective bargaining at the company level. The workers' councils do not fulfill the role o f worker representation in bargaining - according to the law, it i s the trade union that protects collective and individual rights o f employees who are members. 138. Most of the employers and workers currently engaged in collective bargaining in both Entities are affiliatedwith the public sector. Thus, the Government is represented "three times" at the bargainingtable -as Government, as employer and as workers. The private sector employers' associations have modestly broadened their base but, as mentioned, earlier representation has been generally low - for example, the Employers Confederation in the RS (ECRS) has only 178 membership fee-paying company members that employ only about 6,000 people. The currently valid GCA in the RS was signed in 1999 by the Chamber o f Commerce whose membership is compulsory and thus defies the notion o f voluntarism. Workbooks 139. The requirementfor employers to retain an employee's workbook until the point of separation creates and reinforcesseveral rigiditiesin the labor market. Designed for a labor market characterized by full-time and lifetime employment, the current workbook practice - which tracks in hard copy the employee's employment history for social insurance purposes - has been proven to be a major defacto barrier to formal alternative forms o f employment including part-time and temporary work and secondary employment. Under the current system, the employer is required to retain the employee's workbook until the employment relationship ceases, at which time he has to return it to the employee duly stamped. workbooks has contributed to the emergence o f the "fictitious worker" problem when employers -This practice o f cannot meet their obligations with respect to current wages, social insurance dues or severance payments, the workers naturally stop working for such employers. At the same time, since these employers are also not in a position to meet their obligations and stamp the workbook, arrears start accumulating, thus creating a vicious cycle and increasing the inability o f employers to settle. Alternative administrative arrangements for maintaining the social insurance records o f workers urgently need to be explored and introduced. Reform measures over the medium term may involve the introduction o f an integrated electronic employee registration database. However, in the short term, procedures could be amended to enable employees to withdraw workbooks, while retaining their claims on unpaid wages and social insurance contributions. Natural Disasters andDisabilig 140. Labor legislation stipulates that employers are expected to continue meeting their financial obligations even when activities cease because of a natural disaster. While this provides protection for workers, it fails to ensure the employer's ability to continue its operations and eventually, in the case o f closure, both the employer and the workers lose out. There can be mechanisms to insure employers against such events and there i s no reason why private markets cannot play a substantial role in this. But the social partners, to the extent that they feel such a need, can identify possibilities for insuring against such events. Another area o f concern is the obligation o f employers to continue employing a worker who is unable to perform hidher duties but who has been awarded an unfavorable decision to receive social insurance benefits. This problem relates to the way the authorities enforce the relevant regulations, but the fact still remains that it is an issue that should be addressed for the benefit o f both workers with disability and their employers. 62 141. BH suffers from a pronounced case of overstaffing and a country-specific "fictitious" worker problem. The grave consequences o f the creation o f waitlisted workers are widely acknowledged. The phenomenon arose because apparently neither Labor Law contemplates that an employee for whom there is no current work should be terminated. Absent termination, the employee for whom there i s no work remains an employee o f that employer, even though he or she is not being paid. Employees in this situation seek work in the informal labor market while remaining in the singular employment relationship contemplated by the law, but neither that employer nor the informal employer pay any payroll taxes. This untenable situation is driven by a conjunction o f the abstract nature o f the law with the employee's desire eventually to secure additional social insurance contributions or a significant severance sum. 142. The labor legislationin both Entities is without any provisionsfor temporary layoffs which could help prevent a fictitiousworker problem in the future. The Labor Laws should be brought into conformity with common practices. Typically, employers can adjust to economic downturns by means other than outright termination or collective dismissal. However, in lieu o f collective discharge as a response to economic conditions, employers in many industrial relations systems can resort to temporary layoffs. This is not a current option under the law in either the FBH or the RS, though in the past (under Yugoslav law) this was possible under certain conditions and usually for a limitedlayoff period. 143. A layoff can apply to an employee whose services are not currently required but might be needed in the future, He/she, therefore, remains subject to recall upon a business upturn. The laid off employee is free to work elsewhere while waiting to be recalled by the original ("primary") employer. While on layoff status, relevant payroll contributions are made by the secondary employer. Recall rights o f the workers and obligations o f the primary employers can be regulated by the law or collective agreements (for example, with respect to duration, type o fjob on reassignment and so on). 144. This report recommends exploring the introduction of legislationthat would enable an employee to be placed on "layoff' status. This will no longer prohibit that employee from seeking lawful employment with another employer, while remaining subject to recall once the primary employer's business needs allow it to expand. A recalled employee would have no break in service for the period o f layoff for the purpose of the accumulation o f seniority or benefits as the current, secondary employer would be required to pay all applicable payroll taxes for the worker. 63 145. In this final section, we provide an assessment o f the regulatory and broader industrial relations framework in BH in view o f the international evidence presented in Annex V o f this report. Should the industrial relations in BH become moreflexible? 146. On one hand, the system of industrial relations appears to be flexible. The Labor Laws prescribe only minimal provisions that can be used for subsequent arrangements to be agreed between employers and workers. Being voluntarily agreed, these agreements should reflect the positions o f the contracting parties. This is in theory. But also in practice for the period covered by the current and other recent studies, most o f the empirical analysis at the economy-wide level i s consistent with this view. On aggregate, there is little (consistent) evidence to suggest that the labor market is choked by regulatory provisions. First, BH is nowhere near a "jobless growth path", as the employment/output elasticity is near unity rather than Second, despite the relatively stagnant overall size o f the formal sector, there has been significantjob creation in the formal sector (of the order o f 23 percent since 2001) outside the ailing SOE sector, alongside an increase in informal sector employment. Third, the spectacular increase inthe labor force participation rate was translated in employment gains while unemployment remained constant - admittedly at a high level. Fourth, wage determination was statistically found to be subjected to few systematic factors that could be linked to collective bargaining and, when they did so, the link was rather weak. For example, the wage increments due to additional education or longer tenure are just what they should be (i.e. positively correlated) but not more than that. The increments are nowhere near what one would expect from the provisions o f collective agreements. 147. On the other hand, there are remainingconstraints to employment adjustment and wage rigidityincertainsectors, but this has more to do with the historyof transition, public sector policies and politicalleconomic considerations rather than the industrial relations system. In terms o f employment, although the SOEs lost one third o f their employed workforce since 2001, more remains to be done, possibly as much as another 35-40 percent reduction to bring the current employment o f 220,000 closer to 150,000. In the meantime the issue of waitlisted and fictitious workers remains unresolved, while the number o f the latter is probably still increasing. Wages appear to be rather uniformly determined in the public sector but not in the private or the informal sector. As the Table 3.2 below shows, real wage increases in the public sector are high and take place within a narrow band in both Entities and for both female and male employees. These observations have little to do with whether collective bargaining takes place at the economy-wide, branch or company level or many other aspects o f industrial relations. They relate to: (a) historical obligations o f the Government that the Government can no longer meet; and (b) current pay awards as determined by the Government (as Government and also as an employer) and public sector trade unions. The obvious approach here is for the Government and workers in the public sector to work out a feasible but also fair solution for the current deadlock, and set the future o f the public sector on a sustainable basis. 73The employment/outputelasticity is usually significantly less than one and, inthe case ofjobless growth, zero. A typical value ofthe elasticity is around0.3 to 0.6 with probably a value 0.4 being arepresentative one undernormal conditions inthe longer run. It is less than one for reasons o f higher use o f capital in production over time, rising productivity, and more effective use of labor arising from better management techniques or simply economic pressurefrom increasingcompetition. 65 Table 3.2: Annual Growth Rates (YO)of Real Average Wages, 2001-2003 Male Female Sector FBH RS FBH RS Average St. Dev. Public 8 6 10 15 9.6 3.7 Private 3 -9 18 -12 -0.2 13.8 Informal 13 2 -5 -1 2.4 7.6 *Employees with positive earnings; ** Theaverage and standard deviation are calculatedacross the rows Source: Own calculations from LSMS,Appendix Table A3.x 148. However, the industrial relations system can become more flexible by introducing legislation on layoffs, changing some provisions in collective agreements and separating public from private wage determination. This report identified as a major omission in the current system the lack of legislationfor lay 08s. It also argues that there is a need for and sound reasons why public sector wage determination should be separated fiom that in the private sector, It finally identified a series o f issues that can be usefully addressed to reduce rigidities or enhance the functioning o f the labor market. These include the determination of minimum wages as well as the introduction o f a separate minimum wagefor younger workers; the usefulness o f a mechanism for insuring employers from stoppages or closure o f operations when no fault i s involved (such asforce majeure); the need to bypass workbooks as a major determinant o f when employment ceased; and the system o f coefficients that mechanically link wages to the education or tenure o f workers that is no longer appropriate for a market economy. Should the industrial relations system become less centralized? 149. On paper, the industrial relations system appears to be centralized, but in practiceit i s decentralized and in a state of flux. In theory, industrial relations in BH are centralized with cascading provisions from the Labor Laws to GCAs and BCAs and, ultimately, company rule books and individual contracts. But in practice this hierarchy i s observed only in the public sector and at times probably ~electively~~. This fluidity is further exacerbated by the fact that labor legislation in neither Entity sets out clear minimum membership criteria and verification procedures for the representativeness o f the social partners in collective bargaining. The Government in the negotiation process i s represented (rightly) as Government, but also as a principal employer (in the public sector), and a dominant representative o f workers (the public sector still employs almost two-thirds o f the formal labor force). Yet most o f the private sector has not been party to wage-setting branch collective agreements. In practice, in the absence o f bindingcollective agreements for most o fthe labor market, wages and working conditions tendto be set by the employer in its rulebook or in individual contracts. These provisions and the lack o f clarity for collective dispute resolutions, arbitration, and criteria for representativeness o f the social partners have resulted in collective bargaining that i s partial in its coverage and its effects. 150. Still, an important lesson from the empirical literature is that centralization is probably less important for labor market outcomes than it is for labor market coordination. The international evidence shows that the important determinant o f good labor market outcomes i s whether the industrial relations system i s coordinated or not. This has to do more with what the system does than how the system i s structured (in terms o f union membership or the level at which collective bargaining takes places - such as national, regional, local levels or branch and company levels). For example, a desirable feature o f any system i s to generate aggregate wage increases that are in line with inflation and productivity changes as well as the international 74 For example, around 40 percent of public sector workers in both Entities report that their actual wages are different thantheir usualwages (Table 2.8). 66 competitiveness o f the economy (for example, through changes in the exchange rate). It is precisely at this point where theory is ambiguous and the empirical evidence tends to be dependent on country evidence and the time o f the inquiry(See Annex 1o fthis report). 151. Although collective bargaining is often seen as the main cause of high wage increases or persistent unemployment, the evidence on this link is rather inconclusive. The economic analysis o f collective bargaining separates macro and micro effects. In some cases the empirical analysis suggests that there might not even be an effect. Annex 1o f this report explains these issues in more detail. There are sound arguments and often empirical evidence confirming that coordination through pure market mechanisms can lead to good economic outcomes - and this refers primarily to bargaining at company level. However, there are also arguments and some other empirical evidence suggestingthat firm-level bargaining can provide workers with an instrument to extract firm-specific quasi-rents or can impose other forms o f externalities (see table below).75 Table 3.3: Six externalities of Decentralized Wage Setting The input price externality Decentralizedwage gains are passedon as higher product prices, thus increasing the real cost of inputs for other firms. The fiscal externality Decentralized wage gains lead to unemployment. The cost in terms of unemployment benefits is born by all tax-payers, not only those involved in wage setting. The unemploymentexternality Decentralized wage gains increase overall unemployment, making it more difficult for all unemployed workers to find a newjob. The envy externality Decentralizedwage gains create envy among other workers. The consumer price externality Decentralized wage gains are passed on as higher product prices, thus lowering the real wage of all workers. Efficiency wage externality At the decentralized level, firms have an incentive to try to increase the relative wage of their workers to increasetheir motivation. Source: Calmfors (1993): pp. 5-6 152. Still, as a theoretical proposition, delegation of collective bargaining to lower levels (e.g. firm-level) is expected to increase allocative efficiency. This refers to the microeconomic effects o f collective bargaining but i s not always confirmed in practice. In fact one o f the more fragile arrangements (in terms o f economic effects) is multi-unionism at company level. Furthermore, even if delegation o f collective agreements to lower levels i s found to increase allocative efficiency, this does not indicate whether the level o f output would be higher. This would depend on the cost and mix o f inputs (technical efficiency) and the boundaries between sectors that are often poorly defined. Another way to restate this issue i s by observing that larger firms are more likely to develop formal firm-level bargaining but firm size appears to increase the ability o f workers to increase wages. 153. In conclusion, this report has three sets of recommendations. First,even within the existing system, there is a need for change, for example to: (a) define the criteria for representativeness o f social partners involved in bargaining; (b) limit collective agreement coverage to those enterprises that are represented in bargaining; and (c) separate wage determination between the public and private sectors though avoiding duality between the two labor markets. These changes would be necessary even if the present system is fundamentally amended. 154. Second, if the system is to change, it will be critical to ensure coordination of collective bargaining at the macro level so that aggregate wage changes are in tune with ''For example, the existence of formal bargaining at the firm level has been found to increase average wages inFrance (Kramarz et al., 1994). 67 inflation, productivity,the competitivenessof the economy and the level of unemployment. What this requires is, (a) that at the general level, the Labor Laws continue to prescribe the minimum provisions under which labor is priced and allocated in the economy, and (b) at the company level, there is sufficient flexibility for individual contracting (for example, the mandatory portability o f seniority pay or the automatic wage increases due to educational credentials should have little role). The critical issue then becomes how to define the space between the labor law and individual contracts. Most countries have some provisions for filling this gap - simply for practical reasons, for example, changes inthe labor law are typically much more laborious and time consuming than changes in general or branch or regional collective agreements. 155. So long as there is sufficient flexibility for individual contracting at the company level, a policy decision in the current context of industrial relations in BH would evolve around the relative balance of the GCA and BCAs. In this respect, some observations are relevant. For example, the GCA includes sweeping provisions across the labor market and, when the bargaining process at the general level breaks down, it can provoke a generalized dispute across the whole labor market. On the other hand, the BCAs may be more in line with the conditions in the individual sectors, although historically their provisions have been more generous than those in the GCA and the extent o f homogeneity o f the sectors in times of economic transition i s questionable. Although any recommendations have been put forward in this report, the fundamental issue o f the balance between GCA and BCAs cannot be answered with certainty given the paucity o f data. However, Bosnia and Herzegovina's economic transition with substantial enterprise sector adjustment needs suggests that the tripartite parties be best served by allowing for more rather than less flexibility in the labor market to promote better outcomes. 156. Third, the report recommends that the agreement on a future framework for collective bargaining be reached through a tripartite dialogue and based on more accurate data and careful economic analysis of the implications of legal and broader policy options. A move from or to GCAs to or from BCAs or company-based agreements should be pursued on the basis o f collating quality data, identifying relevant national and international experience, buildinganalyticalcapacity and continuing social dialogue under the social-economic councils. 157. BHlacks the information basisfor an objectiveassessment of the characteristicsand trends of the labor market. As evidenced in the analysis presented in this report, the administrative data collected by the Statistical Offices are partial in coverage and often inaccurate. The offices themselves do not have the capacity to clean up the data or to follow up with the respondents. In addition, the two Entities often apply different methodologies and use terms in different ways. This prevents the emergence o f a clear picture both at Entity and State levels. Similarly, the recently introduced LSMS have been designed primarily for poverty analysis but fall sort in the terms o f reliability and clarity for labor issues. Effectively, the country lacks an information system that would create an enabling environment for an objective understanding o f the labor market (necessary for building social consensus) and for the monitoring o f various policies and reforms (necessary for the evaluation o f competing options). The design o f policies can, o f course, follow some established principles even without any 68 reference to facts. For example, measures that increase competitiveness or establish a minimum acceptable welfare level on the social front can go ahead subject to the overall economic constraints. However, more informed approaches, such as those based on reliable statistics and good analyses, can facilitate the process o f choosing between competing policy alternatives. By knowing the effects o f different scenarios in advance, one can deflate ideological or rent-seeking opposition to otherwise desirable reforms and can anticipate and prepare for the adverse distributional effects o f certain reforms. 158. Therefore, this report supports the decision to introduce an annual Labor Force Survey and recommends working on further increasing the capacity of the Statistical Offices on labor market statistics. Improvements in the enterprise databases and establishment surveys as well as the introduction o f a census will also contribute to the creation o f a more accurate statistical picture o f the economy. The need for relevant, adequate, and timely statistics i s particularly true in the area o f labor markets where policies and reforms typically involve winners and losers. The diverse fate o f these two groups can be particularly acute inthe short run though adverse distributional effects o f reforms can be neutralized, and in fact be reversed, inthe long run. Thus an adequate statistical system is a prerequisite for the informed and timely design, implementation and monitoring o f labor policies. This can avoid deadlocks in the dialogue (as is currently the situation in BHin certain areas) and even industrialconflict. 69 REFERENCES Addison, J.T. and J. Wagner (1994), On the Impact of German Work Councils: Some Conjectures and Evidence9om Establishment-Level Data, mimeo. Aidt, T. and Z. Tzannatos (2002), Unionsand Collective Bargaining: Economic Effects in a Global Environment, World Bank, Washington DC. Alba-Ramirez A., and Munoz-Bullon F., (2004) Unemployment duration, unemployment insurance and temporary layoff in Spain. Paper presented at the conference "European unemployment: Recent developments in duration analysis usingregister data", Mannheim. Betcherman, G., Olivas, and A. Dar (2004), "Impacts of Active Labor Market Programs: New Evidencefiom Evaluations, withparticular reference to Developing and Transition Countries", World Bank, Washington DC. Birks Sinclair IBHI(2003), Qualitative Study No 2. BHSkills Gaps and TrainingNeeds: Current Status and Policy Implications, DFIDProject Report, Development o f Labour and Social Polices, Sarajevo. Bosnia and Herzegovina: Statistical Bulletin(various issues), Dept o f Statistics, Sarajevo Bosnia and Herzegovina: Statistical Bulletin(various issues), Dept o f Statistics, Banja Luka Bosnia and Herzegovina(various dates): Selected legislative/industrial relations pieces such as Labor Laws and their amendments, other related laws, proposals by employers' organizations or unions, legal opinions. Calmfors, L.(1993), Centralization of WageBargaining and Macroeconomic Performance: A Survey, Seminar Paper no. 536, Stockholm University, Institute for International Studies. Carlson, Ellen (200 1). Differing labour market concepts, definitions and measurements Japan, USA, Australia. Paper presentedat the Workshop "Understanding Unemployment inAustralia, Japan and the USA: A Cross Country Analysis". Centre o f Full Employment and Equity: University o fNewcastle. 10 and 11 December. Dar, A. and Z. Tzannatos (1999), Active Labor Market Programs: A Review of the Evidencefiom Evaluations, World Bank Social Protection Discussion Paper Series, No, 9901 FeldsteinM.,(1976). Temporary layoffs inthe theory of unemployment. Journal o f Political Economy. Vol 84, pp.937-957. Feldstein,M.,(1978). The Effect o f Unemployment Insurance on Temporary Layoff Unemployment. American Economic Review, Vo1.68(5), pp. 834-46. Finkin,Matthew (2004), Labor Law Environment in Bosnia and Herzegovina, Financial Services Volunteer Corps, mimeo 71 Fischer, G., and Pichelmann, K.,(1991). Temporary LayoffUnemployment in Austria: Empirical Evidence from Administrative. Applied Economics, Vo1.23(9), pp. 1447-52 Gruber, J. (1998), HealthInsurance and the Labor Market,NBER, Cambridge, MA. Gruber, J. andA.B. Krueger(1991), TheIncidenceof mandatedEmployer-Provider Insurance: LessonsJForn Workers'Compensationinsurance,in D,. Bradford (ed.), Tax Policy and the Economy, MIT Press, Cambridge, MA. Heckman, J. and C. Pages (eds.) (2004) The Cost of Job Security in Latin America, NBER. Holmlund, B., and Storrie D.,(2002). Temporary Work inTurbulent Times, Economic Journal. Vol. 112(480), pp. 245-69. Idson, T. L.,and Valletta, R., G., (1996). Seniority, Sectoral Decline, and Employee Retention: An Analysis of LayoffUnemploymentSpells. Journal of Labor Economics, Vol. 14(4), pp. 654-76. Jansson, F. (2002). Rehires and Unemployment Duration inthe Swedish Labor Market: New Evidence on Temporary layoffs. Labour, Vol. 16(2), pp. 311-45. Kadic (2004), Analysis of OutstandingLiabilities towards Workers in Companieswhichwill be Privatized and Potential Redundancy CostsRelated to Severance Pay Programs, Consultant Report, mimeo Karni, E. (1996). Optimal UnemploymentInsurance: A Guide to the Literature. World Bank: Social Protection Discussion papers Series No. 9605. Katz L.and Meyer B., (1990). UnemploymentInsurance, Recall Expectations, and UnemploymentOutcomes. The Quarterly Journal of Economics, Vol. 105(4), pp. 973-1002. Katz, L.F., (1986). Layoffs, Recall and the Duration of Unemployment. NationalBureau of Economic ResearchWorking Paper: 1825. Kramarz, F., S. Lollivier and L.-P. Pele (1994), WageInequalities and Firm-Specific CompensationPolicies in France. INSEE-CREST, mimeo Lewis, G.H.(1986) "Union Relative Wage Effects" in0.Ashenfelter and R. Layard (eds.) Handbook of Labor Economics, Vol. 2., North-Holland, Amsterdam. Mavromaras, K.,G. and Rudolph, H.,(1998). Temporary separationsand firm size in the German labour market. Oxford Bulletinof Economics & Statistics, Vol. 60(2), pp. 215-226. Mavromaras, K., G. and Orme, C., D., (2004) Temporary Layoffs and Split PopulationModels; Journal of AppliedEconometrics, Vol. 19(I), pp.49-67. Memic (2005), Preliminary Report on Poverty in BHfor the Period 2001-2004,mimeo, BH Economic Policy and PlanningUnit, Sarajevo 72 Micevska, Maja (2004), Unemploymentand Labor Market Rigidities in SoutheastEurope. University of Bonn, ESCE Economic and Social Research Center, Cologne and Eisenstadt, mimeo Mossialos, E.and A. Dixon (2002), Funding Health Care in Europe: Weighingup the Options, in E.Mossialos, A. Dixon, J. Figueras et al. (eds.) Funding HealthCare: Options for Europe. Open University Press, BuckinghamUK. Nivorozhkin, A; (2005) Durationof Temporary Layoffs, Impact of Tenure and Enterprise Characteristics: Evidence from Sweden. Goteborg University. OECD (1999), Employment Outlook, Paris Paci, Pierella (2002), Gender in Transition, The World Bank, Washington, DC Palenzuela, D.R. and J. F. Jimeno (1996), WageDrgt in Collective Bargaining at Firm Level: Evidencefiom Spain, Annales d' economie et de statistique, No. 41/42, 187-206 Robertson, M.,(1989) Temporary Layoffs andUnemployment in Canada, IndustrialRelations, V01.28( l), pp.82-90. Rodriguez-Planas,N.,(2004) PlayingHardto Get: Theory and Evidence on Layoffs, Recalls und Unemployment, Federal Reserve BoardWorking Paper, Discussion Series No. 1999-09. Roed, K.and Nordberg, M.,(2003). Temporary Layoffs and the Duration of Unemployment, Labour Economics, Vol. 10,3 81-398. Rogers, J. and W. Streek (1994), WorkplaceRepresentationOverseas: The Works Councils Stoy, in R.B.Freeman(ed.) Working Under DifferentRules.NBER, Ruse11Sage Foundation, New York. Rutkowski, Jan (2003), TheMinimum Wage: Curse or Cure? Mimeo, April 6. Rutkowski, Jan, Stefan0 Scarpetta, Philip O'Keefe and MilanVodopivec (2004), Labor Markets In EuropeAnd Central Asia: Towards CreatingMore And Better Jobs. Draft. November. SSSBH, SSRS and SBD (2005), EmploymentandPoverty Reduction in Bosnia and Herzegovina, Recommendationsby the BHUnionsjointly preparedby Savez Samostalnih Sindikata BH (SSSBH), Savez Sindikata Republike Srpske (SSRS) iSindikat BrCko Distrikta (SBD),April 5. Stojanov( 2004), Survey andAnalysis of Debt Structure of a Sample of Bosnian Enterprises, mimeo Stewart, M.(1996), Union Differentials, Product Market Influences and the Division of Rents, The Economic Journal, Vol. 100, pp 1122-1137. Topel, R.(1983). On Layoffs and Unemployment Insurance. American Economic Review, September, pp. 541-559. Vodopivec (1991), The Transition of the Yugoslav Labor Market: Crossing the Rubicon of Job Security,mimeo 73 Vodopivec, Milanet al, (2002). IncomeSupport Systems for the Unemployed:Issues and Options. World Bank: SocialprotectionDiscussionPaper World Bank (2002), Bosnia and Herzegovina: From Aid to Dependency to Self-Reliance: A Public Expenditure Review, Washington, D.C. World Bank (2002), Labor Market in Postwar Bosnia and Herzegovina: How to Encourage Business to CreateJobs and Increase Worker Mobility. Washington, D.C. World Bank (2005), Bosnia and Herzegovina: Poverty and the Labor Market, mimeo World Bank (2005b), Doing Business 2006, Washington, D.C. World Bank (2005a), Bosnia andHerzegovina Country Economic Memorandum, Washington, D.C. World Bank (2004) Bosniaand HerzegovinaJoint IDA-IMF Staff Assessment ofthe Poverty ReductionStrategy Paper. WashingtonDC. World Bank, (2003), Bosnia and Herzegovina Poverty Assessment, Washington,D.C. World Bank (2005b), Labor Markets in Europe and Central Asia: Towards Creating More and Better Jobs, WashingtonDC. 74 APPENDIX I: DESCRIPTIONOF THE LSMS AND VARIABLES Survey description The Living Standards Measurement Survey collects data from a random sample o f households from 25 municipalities (out o f 145) inBosnia-Herzegovina o f which 14 are from the Federation Bosnia and Herzegovina(FBH) and 11 from the Republika Srpska (RS). Individual observations are subsequently weightedto provide representative estimates separately for each for each Entity and then for the State at large. A commendable feature of the survey i s that it i s available on the web -thus accessible to outside users76. The first LivingStandards Measurement Survey (LSMS) was conducted inSeptember and October 2001. The main goal o f the LSMS is to measure welfare in both monetary and non- monetary terms in Bosnia-Herzegovina. Inadditionto data consumption, several other modules cover areas such as education, health and labor markets.77 The labor markets module covers six areas: Employment Status; Main Job, Second Job, Information about persons who do not work; Information about persons looking for a job; and Status with the Employment Bureau. This report is based on panel data form Waves 1,2 and 3 undertaken in 2001,2002 and 2003. The panel survey sample is made up o f over 3,000 households drawn from the LSMS conducted in 2001. Approximately halfthe households interviewed on the LSMS were selected and carried forward into the panel survey. The households were re-interviewed for a second time in 2002 and again in September 2003. Methodolow A panel was constructed usingyears 2001,2002 and 2003. Everyperson was assigned a unique IDnumber78.The matching o f individualswas basedonthe identification number (Actpsid), gender and birth date. The universe is all persons older than 15 years old. 76Republika Srpska Institute for Statistics website: http:llwww.rzs.rs.ba/AnketeZuBHENG.htm, Federationof Bosnia-Herzegovina's FederalOffice of Statistics website: http:llwww,fzs.ba/download eng.htm 77The questionnaireconsistsof 13 modules covering Roster,Housing, Education, Labor, Migration, Health, Voucher, Credit, Social Assistance, Consumption, Non-Agricultural Activities andAgricultural Activities. 78The IDnumber is constructedby combining five variables: Municipality code, Group of Enumeration Areas, EnumerationArea or PSU, Householdnumber inenumerationarea, and Personalidentification number. There were 6,3 11respondents interviewedinall three years o fthe survey: 2,837 inRS and 3,474 inFBH.This represents around 81.2%ofthe responding wave 1sample, aretentionrate which is highcompared to many other panels around the world. Descriotion of Variables Employment: A person is classified as employed ifhe/she is 15 years or older and the employment status o f this person is non-missing. There are some workers who are classified as "employed" but did not work in the reference week. These might be (i)workers on the "waiting list" (forced administrative leave), (ii) workers who do not work because o f war and other difficulties ,and (iii)workers not working because the employer i s inthe process o f bankruptcy or liquidation. Also, some workers mightnot be classified as employed inthe cases that they did not have an agreed date for returnto work, may had been inthis status for a long time, and may not had received wages duringthe waiting period. Formalitv-Informality: This i s a constructed variable along the lines o f the WB Labor Market Study 2002.79Formality refers to employees who work in(a) public enterprises, institutions, or organizations, (b) private sector and pension contributions are paid and (c) international organizations. Informality refers to (a) private sector employees for whom pension contributions are not paid; (b) workers who are unpaid family member, farmer on own farm, or do some other activity such as sale o f agricultural and other products or provide housing, intellectua1 or other services. Unemdovment: A person is classified as unemployed ifhe/she i s (a) inthe working age population(15-64) but not employed as defined above and (b) seekingwork inthe last 4 weeks and (c) available for work duringthe reference week. This accords the ILO definition but could not be applied to the Wave 3 questionnaire because o f routing problems." However, self- reported unemployment can be calculated for each wave. Discouraged workers: If a worker i s thus classified ifhe/she was not looking for ajob inthe last 4 weeks, but wanted to work, believed that there was no suitablejob available and was available to work. Again, it is not possible to classify individuals as "discouraged" for Wave 3. Education: This question is asked only inthe first survey in2001and this information was used to infer the education level o f individuals who were present in all panels. Industry:Thirteen industries are identified: Agriculture and Fishing; Mining, Manufacturing, Utility,Construction, Trade, Hotels and Restaurants, Transport, Storage and Communication, FIRE(Finance, Insurance andReal Estate), Public Administration, Education, Health and "Others". ~ 79 According to the "The resolution concerningstatistics of employment inthe informal sector" (The Fifteenth International Conferenceof Labor Statisticians,January 1993) the defming element of informal employment is the fact that the employer is not an incorporatedbusiness. That is, aworker i s treated as informally employed ifhe (she) is either self-employed (inan unincorporatedbusiness), or an employee working for unincorporatedemployer. Note that by its nature, informaljobs are associated with different bearingofrisks and profits, andtheir job is directly affectedby their business performance; employees in the informal sector are usually more exposedto risk, becausethe riskoftheirjob termination is higher- they enjoy less protection. Note that formal sector workers may or may not be coveredby health and pensioninsurance. UKDepartmentofInternational DevelopmentReport, 2004, page34. 76 Sector: Three sectors are distinguished Public, Private and Others other/ farm/ unpaidl international).*l Workers were assigned to the Public Sector ifthey replied they worked ina public enterprise, institution or organization. The Private Sector inour sample consists o f workers inthe private formal sector. Firm Size: In2002, 2003 this question is a continuous variable. For 2001 a new variable was created for the following ranges: 1-10, 11-100, 101-500, 5011 Tenure: Years with current employer Waiting List: Individuals who did not work duringthe reference week because they were inthe waiting list. Wages: Individuals report usual wages and last paid wages (the question was "What i s the amount o f your usual monthly NET salary or earning at your mainjob?"). Previous reports based the analyses on usual wages and we follow the same procedure. These are real wages, that is wages adjusted by inflation usingthe IMF CPI indices for RS and the Federation (taking 2001 as base year). Working Age Population: Those in the 15-64 age group Labor Force Participation: The employed and unemployed as a share o f the working age population. Employment rate: The employed as a share o f the working age population Unemployment rate: The unemployed as a share o f the o f the labor force (employed and unemployed) Earning:Functions Results based on those reporting positive monthly earnings within the range o f 4 0 K M to 4,OOOKM. Workers o f international organizations are excluded. The regressions are restricted to employees only aged 25-55. Vocational: Dummy variable with value o f 1when someone has vocational education, 0 otherwise. . These individuals are also assigned the number o f years o f schooling completed under the education variable. Of those who attended vocational education, 92% have completed secondary education and 8% have completed primary. So, the coefficient of vocational captures the effect ofeducationtype, not quantity. Firmsize. Three different versions were tried. The first usedthe grouping adopted inthe World Bank (2005) whereby a company was given the value from 1to 6 depending on which bracket (1- 10, 11-20,2 1-50, 5 1-100, 101-200,20 1-500, and 501+) its number o f employees fell. The *'Note: World Bank (2002) includes international organization a5-10==10 -inthe public sector. World Bank (2005) addedthe international organizationto the "Others" category. Unless noted, we will use Bisogno-Fares definition for consistencyinthe comparisons in this report. 77 secondutilized a continuous variable as company size was originally reported inthe survey. The third was constructed from the mid-point values of the following ranges: 1-10, 11-20,21-50, 51- 100, 101-200,201-500, and 501+. All three specifications of this variable returnedmainly insignificant results with the second version returninga occasionally a marginally significant coefficient for men inthe Federation. Tenure is a continuous variable (inyears) related to the lengthof employment with the current employer. Industrywas specified as 0-1 dummies for the following 13 industries: Agriculture and Fishing; Mining,Manufacturing, Utility, Construction, Trade, Hotels and Restaurants, Transport, Storage and Communication, FIRE(Finance, Insurance, Real Estate), Public Administration, Education, Health, and Others. 78 APPENDIX 11: TABLES Appendix Table A l . l FormalEmployment 1991-2004 1991 1997 1998 1999 2000 2001 2002 2003 2004 A. Bosnia and Herzegovina Total 975 575 639 629 638 627 629 626 622 Agriculture 36 n.a. n.a. 21 20 21 21 20 19 Manufacturing 498 n.a. n.a. 257 256 248 244 234 226 Services 441 n.a. n.a. 351 362 359 364 371 371 B. FBH Total 638 373 395 408 411 407 394 387 385 Agriculture 21 n.a. 10 10 10 11 11 10 10 Manufacturing 325 n.a. 161 161 162 156 152 146 143 Services 292 n.a. 224 237 239 240 232 231 232 C.RS Total 337 202 244 221 227 220 235 238 237 Agriculture 15 n.a. n.a. 11 10 10 10 10 9 Manufacturing 173 n.a. n.a. 96 94 92 92 88 83 Services 149 n.a. n.a. 114 123 119 133 140 145 Sources:Entity Statistical Offices Appendix Table A1.2 Employment,Unemployment,and Labor Force Participation,by entity, 2001-2002-2003 Number of Observationsin LSMS 2001 2002 2003 BH FBH RS BH FBH RS BH FBH RS A. Employment (i) Total 2,582 1,266 1,316 2,725 1,355 1,370 2,785 1,443 1,342 Men 1,707 841 866 1,781 938 843 1,788 955 833 Women 875 425 450 944 417 527 997 488 509 EmploymentRate-total 40.2 35.1 46.8 42.4 37.5 48.7 43.4 40.0 47.7 Men 53.7 48.1 60.6 56.0 53.6 59.0 56.3 54.6 58.3 Women 27.0 22.8 32.5 29.1 22.4 38.1 30.7 26.2 36.8 (ii) FormalEmployment- total 1,696 861 835 1,757 926 831 1,781 967 814 Men 1,118 578 540 1,146 332 526 1,158 647 511 Women 578 283 295 611 306 305 623 320 303 FormalEmployment- rate 65.1 68.0 63.4 64.5 68.3 60.7 63.9 67.0 60.7 Men 65.5 68.7 62.4 64.3 35.4 62.4 64.8 67.7 61.3 Women 66.1 66.6 65.6 64.7 73.4 57.9 62.5 65.6 59.5 (iii) Informalemployment-total 887 406 481 989 445 544 1,006 477 529 Men 590 264 326 654 620 322 632 309 323 Women 297 142 155 335 113 222 374 168 206 Informalemployment- rate 34.4 32.1 36.6 36.3 32.8 39.7 36.1 33.1 39.4 Men 34.6 31.4 37.6 36.7 66.1 38.2 35.3 32.4 38.8 Women 33.9 33.4 34.4 35.5 27.1 42.1 37.5 34.4 40.5 B.Unemployment Total 488 257 231 739 374 365 Men 305 158 147 429 201 228 Women 183 99 84 310 173 137 Unemploymentrate-total 15.9 16.9 14.9 21.4 21.7 21.1 Men 15.2 15.8 14.5 19.5 17.7 21.4 80 Women 17.3 18.9 15.7 24.9 29.4 20.8 B'.Self reportedUnemployment Total 1,252 694 558 1,233 670 563 1,218 653 565 Men 760 45 1 309 683 345 338 707 356 351 Women 492 243 249 550 325 225 511 297 214 Self-reportedUnemployment rate 32.7 35.4 29.8 31.2 33.1 29.1 30.4 31.2 29.6 Men 30.8 34.9 26.3 27.7 26.9 28.6 28.3 27.2 29.7 Women 36.0 36.4 35.6 36.8 43.8 29.9 33.9 37.8 29.6 C. Activity Rates Labor Force Participation-total 3,070 1,523 1,547 3,45 1 1,724 1,727 Men 2,012 999 1,013 2,204 1,136 1,068 Women 1,058 524 534 1,247 588 659 Labor Force Participation-rate 47.8 42.2 55.0 53.7 47.8 61.4 Men 63.3 57.1 70.9 69.4 65.0 74.7 Women 32.6 28.2 38.6 38.4 31.6 47.6 other Labor Force Participation (employed+ self reported unemployed) 3,834 1,960 1,874 3,958 2,025 1,933 4,003 2,096 1,907 Men 2,467 1,292 1,175 2,464 1,283 1,181 2,495 1,311 1,184 Women 1,367 668 699 1,494 742 752 1,508 785 723 other Labor Force Participation (employed+ self reported unemployed) 59.7 54.3 66.6 61.6 56.1 68.7 62.3 58.1 67.8 Men 77.6 73.9 82.2 77.5 73.4 82.7 78.5 75.0 82.9 Women 42.1 35.9 50.5 46.0 39.9 54.3 46.5 42.2 52.2 Memorandum Item 82 83 81 Working age population (15-64) -total 6,424 3,610 2,814 Men 3,178 1,749 1,429 Women 3,246 1,861 1,385 81 Appendix Table A1.3 Average monthly earnings in KMand relatively to Entity Average, 2000-2004 FBH RS 00 01 02 03 04 00 01 02 03 04 InKM Total 413 443 483 524 534 347 379 422 Agriculture 414 404 453 518 515 318 338 371 Mining 380 342 395 445 433 210 245 332 Manufacturing 302 316 360 409 412 235 259 303 Electricity, gas and water supply 525 707 717 676 719 490 516 550 Construction 311 322 356 359 359 264 290 342 Trade 344 319 342 395 389 267 288 304 Catering 316 317 340 394 390 209 216 262 Transport and communications 478 601 629 685 697 461 521 567 Financial Services 816 880 973 1037 1092 543 710 865 BusinessServices 407 450 497 530 514 424 461 438 Public administration 539 599 646 694 680 565 591 658 Education 433 445 476 505 534 342 352 389 Health and social welfare 446 463 525 594 638 392 424 452 Other services 420 443 493 539 559 304 336 387 Relativeto Entity Average (%) Total 100 100 100 100 100 100 100 100 100 100 Agriculture 100 91 94 99 97 95 84 92 89 88 Mining 92 77 82 85 81 85 84 60 65 79 Manufacturing 73 71 74 78 77 70 68 68 68 72 Electricity, gas and water supply 127 159 148 129 135 141 138 141 136 130 Construction 75 73 74 69 67 81 70 16 16 81 Trade 83 72 71 75 73 80 78 77 76 12 Catering 77 71 70 15 73 69 65 60 57 62 Transport and communications 116 136 130 131 131 1I 8 128 133 137 134 Financial Services 198 199 202 198 205 176 154 I57 188 205 BusinessServices 99 102 103 101 96 123 119 122 122 104 Public administration 131 135 134 132 127 146 157 163 156 156 Education 105 100 99 96 100 118 104 99 93 92 Health and social welfare 108 105 109 113 120 125 124 113 112 107 Other services 102 100 102 103 105 100 95 88 89 92 Correlationof changes for the last three Years when information exists: 0.26 Source: Statistical offices 82 Appendix Table A1.4 Percentagechange in industrial earnings Last three* years FBH RS Financial Services 18% 59% Mining 30% 58% Construction 11% 30% Manufacturing 29% 29% Other services 22% 27% Catering 24% 25% Transport and communications 14% 23% Agriculture 28% 17% Public administration 16% 16% Health and socialwelfare 28% 15% Trade 24% 14% Education 13% 14% Electricity, gas and water supply -4% 12% Business Services 18% 3yo Average 18.3% 21.6% Note: 2001 to 2003 for the FBH; and 2002 to 2004 for the RS Source: Table AE2 83 Appendix Table A2.1 Average Earnings by Industry:Comparison Between Administrative and Survey Data, 2003 Sector Administrative LSMS FBH public private informal A Agriculture, huntingand forestry 518 453 250 B Fishing 365 449 C Mining 445 423 435 D Manufacturing 409 321 565 312 E Electricity, gas and water supply 676 516 397 499 F Construction 359 567 634 558 G Wholesale, retail; certain repair 395 300 561 477 H Catering 394 434 350 616 I Transport,storageandcommunication 685 569 919 1032 J Financial intermediation 1037 780 759 K Real estate, renting and business activities 530 636 644 299 L Public administration and defense; compulsory social security 694 671 88 M Education 505 499 584 N Health and social welfare 594 582 349 0 Other social and personal services 539 43 1 671 614 RS public private informal A Agriculture, huntingand forestry 338 309 121 423 B Fishing 282 280 C Mining 245 295 580 531 D Manufacturing 259 121 387 E Electricity, gas and water supply 516 318 221 F Construction 290 183 855 342 G Wholesale, retail; certain repair 288 199 333 188 H Catering 216 187 274 271 I Transport,storageandcommunication 521 384 501 435 J Financial intermediation 710 474 626 338 K Real estate, renting and business activities 461 97 507 290 L Public administration and defense; compulsory social security 591 370 M Education 352 322 232 N Health and social welfare 424 337 45 1 145 0 Other social and personal services 336 253 499 185 Source: Own calculation based on LSMS and Statistical Offices 84 Appendix Table A2.2a SeparateEntity Regressions: DependentvariableLog(month1yEarnings) Federationof BH LMStudy 2002 Current Study 2001 2001 2002 2003 Formal Gender differential 0.16 0.16 0.14 0.14 PrivatePublic 0.24 0.26 0.19 0.12 Firm size 1/2 (wrong sign) 113 insign insign Work Experience 116 Insign insign insign Tenure Insign 116 (wrongsign) 116 insign Waitlisted -0.40 -0.30 insign -0.51 Secondjob Insign Insign insign insign Education:primary Insign Insign insign insign Education:vocational Insign 0.19 insign insign Education:secondary 0.18 0.30 0.43 0.13 Education:college 0.39 0.63 0.74 0.35 Education:university 0.75 1.05 1.23 0.76 Industrialeffects 5/10 3/12 1/12 1/12 Hours 0.25 0.20 0.37 n 1750 538 547 572 R-2 0.34 0.42 0.36 0.26 Informal 2001 2001 2002 2003 Gender differential 0.32 0.27 (0.06) 0.28 PrivatePublic na Na na Na Firm size insign Insign 113 Insign Experience 116 (wrongsign) Insign insign Insign Tenure 114 Insign 1/6 216 Waitlisted insign Insign insign Insign Secondjob insign Insign insign Insign Education:primary -0.47 Insign insign Insign Education:vocational insign Insign insign Insign Education:secondary insign Insign 0.26 Insign Education:college insign Insign insign Insign Education:university 0.50 Insign 1.04 Insign Industrialeffects 2/10 1/12 insign 1/12 Hours 0.78 0.58 0.73 n 393 109 97 94 0.46 0.38 Notes: 1) See notes to Table 2.10. 2) Differencesinthe two specifications include:(a) addinghoursas a regressorsince earnings relate to monthly pay; (b) omitting workers ininternationalorganizations since they are not subject to nationalmarket forces; (c) restrictingthe sample to those aged25-55. The last three columns are basedon employees reporting positivemonthlyearningsbetweenKM40 and KM4000. Source:World Bank (2002) and own calculations based on the LSMS2001, 2002 and2003 85 Appendix Table A2.2b Separate Entity Regressions: Dependentvariable Log(monthlyEarnings) RepublikaSrpska LMStudy 2002 Current Study 2001 2001 2002 2003 Formal Gender differential 0.17 (0.07) 0.11 0.11 PrivateiPublic 0.56 0.61 0.39 0.45 Firm size insign insign insign insign Work Experience insign insign insign insign Tenure 114 insign insign insign Waitlisted insign insign -0.84 -0.52 Secondjob insign insign insign insign Education:primary insign insign -0.32 -0.31 Education:vocational insign insign -0.16 insign Education: secondary insign -0.14 -0.03 insign Education: college insign insign 0.07 insign Education:university 0.63 insign 0.26 0.29 Industrialeffects 7110 2112 3112 4112 Hours (0.19) 0.30 (0.23) n 1485 473 456 423 R-2 0.35 0.36 0.43 0.41 Informal Gender differential 0.26 (0.24) 0.44 (0.19) PrivateiPublic na na na na Firm size insign insign insign insign Work Experience insign insign insign insign Tenure insign 116 (wrongsign) 216 (wrongsign) insign Waitlisted insign insign insign insign Secondjob insign insign insign insign Education:primary insign insign -0.39 insign Education:vocational insign insign insign insign Education:secondary insign insign insign insign Education:college -0.12 insign insign insign Education:university -0.59 insign insign insign Industrialeffects 1/10 3112 insign insign Hours 0.59 0.56 0.47 n 530 130 121 105 R-2 0.12 0.37 0.28 0.14 Notes: 1) See notes to Table 2.10. 2) Differences in the two specifications include:(a) adding hoursas a regressor since earningsrelateto monthlypay; (b) omittingworkers in internationalorganizationssince they are not subject to nationalmarket forces; (c) restrictingthe sample to those aged 25-55. The last three columns are based on employees reporting positivemonthlyearnings betweenKM40 and KM4000. Source: World Bank(2002) and own calculations basedon the LSMS2001,2002 and2003. 86 Appendix Table A3 Distribution o f covered workers by collective agreements by industry, 2004 (constructed tablefrom possibly noisy data) FBH RS Agriculture, huntingand forestry 7.6% 5.3% Fishing 0.0% 0.0% Mining 6.1% 21.0% Manufacturing 26.3% 16.6% Electricity, gas and water supply 5.7% 0.0% Construction 7.0% 7.7% Wholesale, retail; certain repair 12.2% 0.0% Catering 0.0% 6.6% Transport, storage and communication 9.3% 17.8% Financial intermediation 2.7% 1.3% Real estate, renting and business activities 0.0% 0.0% Public administration and defense; compulsory social security 14.8% 8.2% Education 8.2% 9.3% Health and social welfare 0.0% 6.3% Other social and personal services 0.2% 0.0% Total (relative) 100.0% 100.0% Memo items Total (number) 262,780 190.790 Estimated employed population 387,381 240,000 Coverage rate 68% 79% Sources: Own estimates based on union membership data from Entity Trade Union Confederations and employment data from Statistical Offices 87 Appendix Table A4 Employment Shares by Industry : Ratio o f LSMSto Administrative Data, Federation2003 Code Industry FBH RS A Agriculture, hunting and forestry -8.7% -82.4% B Fishing 135.0% -96.1% C Mining 34.3% 51.8% D Manufacturing -3.7% 46.0% E Electricity, gas and water supply 15.6% -32.5% F Construction 126.6% 133.3% G Wholesale, retail; certain repair -1.5% 12.8% H Catering 19.5% -44.5% I Transport,storageandcommunication 3.9% -27.0% J Financial intermediation -33.0% -92.4% K Real estate, renting and business activities -23.6% -90.3% L Public administration, defense, security etc -63.4% -68.7% M Education -24.6% -2.0% N Health and social welfare -34.4% -53.7% 0 Other social and personal services 142.9% 25.0% Note: The correlations between the employment frequencies reported from administrative data and the LSMSwere 0.81 (public sector), 0.76 (Private sector) and0.16 (informal sector); LSMS data refer to all workers, notjust employees or those reporting positive earnings. Source: Own calculations from administrative and LSMS data 88 Appendix Table A5 Population in SEE Countries, in thousands 2000 2001 2002 2003 2004 Change Albania 3.113 3.132 3.150 3.169 3.188 2.4% Bosnia and Herzegovina 3.977 4.060 4.121 4.134 4.158 4.6% Bulgaria 7.827 7.913 7.868 7.824 7.424 -5.2% Croatia 4.381 4.437 4.443 4.442 4.442 1.4% Macedonia 2.026 2.035 2.020 2.031 2.044 0.9% Montenegro 0.611 0.613 0.616 0.618 0.620 1.5% Romania 22.203 22.408 22.355 21.774 21.752 -2.0% Serbia 7.747 7.727 7.504 7.504 7.504 -3.1% Total 51.885 52.326 52.077 51.495 51.132 -1.5% Average netwage in EURfor SEE Countries 2000 2001 2002 2003 2004 change Albania 99.1 116.7 129.2 134.2 169.3 71% Bosnia and Herzegovina 191.2 226.5 228.0 253.1 258.2 35% Bulgaria 94.9 100.2 111.0 115.1 122.6 29% Croatia 435.4 474.1 502.1 520.8 556.6 28% Macedonia 172.4 173.7 190.8 193.6 201.0 17% Montenegro 96.3 107.7 142.1 174.0 195.9 103% Romania 107.2 116.0 121.2 128.9 147.2 37% Serbia 45.4 98.3 151.8 176.2 193.0 325% Notes: Albania:public sector wages only are included; Bosniaand Herzegovina:simple average for datainRepublikaSrpskaand Federationof B&H; Bulgaria - datafor 2004 are for 44 only; Croatia, Macedonia, Montenegro:net wage doesn'tinclude allowances for food, transport etc.; RomaniaSerbia Source:Staff calculations 89 Appendix Table A6 Minimum wages RS FBH Amount in KM Ratio of MW to AW Amount in KM Ratio of MW to AW 1997 55.00 0.49 85.00 0.32 1998 60.00 0.35 85.00 0.26 1999 60.00 0.28 206.00 0.55 2000 80.00 0.29 227.00 0.55 2001 80.00 0.26 243.79 0.55 2002 82.00 0.24 265.49 0.55 2003 82.00 0.22 291.14 0.55 2004 82.00 0.19 293.36 0.55 RS wages 1997-98 convertedfrom YUD 90 Appendix Table A7 Industrialmonthly earnings rankedin relationto the BHaverage FBH Financialintermediation 1037 214% Rs Financial intermediation 710 147% FBH Public administrationand defense 694 143% FBH Transport, storage and communication 685 142% FBH Electricity,gas and water supply 676 140% FBH Healthand social welfare 594 123% Rs Public administration and defense 591 122% FBH Average 543 112% FBH Other social and personal services 539 111% FBH Real estate, rentingand businessactivities 530 109% Rs Transport, storage and communication 521 108% FBH Agriculture, huntingandforestry 518 107% RS EIectriciQ, gas and water supply 516 107% FBH Education 505 104% BH Average (both) 484 100% RS Real estate, renting and business activities 461 95% FBH Mining 445 92% Rs Health and social welfare 424 88% FBH Manufacturing 409 85% FBH Wholesale,retail; certainrepair 395 82% FBH Catering 394 81% RS Average 389 80% FBH Fishing 365 75% FBH Construction 359 74% Rs Education 352 73% Rs Agriculture, hunting andforestry 338 70% Rs Other social andpersonal services 336 69% Rs Construction 290 60% Rs Wholesale,retail: certain repair 288 60% Rs Fishing 282 58% Rs Manufacturing 259 54% Rs Mining 245 51% Rs Catering 216 45% Note:Monthly wages (employment)in 2003 were reportedto be 543 (387,381) in the Federation and 389 (240,000) in the Republika. Source: Calculatedfrom Appendix Table AE2 91 Appendix Table AS Trade Union Structure and Density in BH RSTrade Union Confederation Membership 1 Trade Union ofMetalworkersand Miners 17,000 2 Trade Union of Communal Services Workers 18,000 3 Trade Union of Forestry andWood-ProcessingWorkers 19,040 4 Trade Union of Textile Workers 20,000 5 Trade Union of Agricultural Workers 10,700 6 Trade Union of Tourism and Trade 6,000 7 Trade Union of Transport Workers 18,000 8 Trade Unionof Teachers 17,290 9 Trade Unionof FinancialServicesEmployees 2,000 10 Trade Unionof HealthWorkers 12,000 11 Trade Union of Policeand InteriorMinistry staff 8,300 12 Trade Union of LegalSystememployees 1,800 13 Trade Union of AdministrationStaff 5,500 14 Trade Unionof InformationTechnology and GraphicsWorkers 1,500 Total Membership 157,130 Total Number of Employed in RS 2004 236,899 Union Density 0.66 FBHConfederation of Independent Trade Unions Membership 1 Trade Union ofElectricity Workers 7,500 2 TradeUnion of Utility Workers 7,400 3 TradeUnion of Railway Workers 7,500 4 Union of IndependentPolice Workers' Associations 11,300 5 Commerce Trade Union 32,000 6 Trade Union of GraphicalWorkers 2,500 7 Trade Union of Secondary Schoolteachers 7,700 8 Trade Union of Primary Schoolteachers 13,930 9 Trade Union of Metal Workers 26,000 10 Trade Unionof ConstructionWorkers 18,300 11 Trade Union of Forestryand Wood-ProcessingWorkers 17,800 12 Trade Unionof Civil Servants 19,500 13 Trade Uniono f BHTelecom 5,000 14 Trade Unionof Chemical IndustryWorkers 4,850 15 Trade Unionof Miners 16,000 16 Trade Unionof FinancialServices Workers 7,100 17 Trade Unionof Media andPrintEmployees 2,500 18 Trade Unionof Oil and PetrochemicalIndustryWorkers 1,900 19 Trade Unionof Textile and Leather IndustryWorkers 18,500 20 Trade Unionof Transport Workers 12,000 21 Trade Unionof BHPost 3,000 22 Trade Unionof Soccer Players 500 23 Trade Unionof Agriculture andFoodProcessingWorkers 20,000 Total Membership 262,780 Total Number of Employed in FBH2004 388,100 Trade UnionDensity 0.68 92 Employers'Associations inBosniaand Herzegovina Appendix Table A9: The Union of Employers FBH Employers' Association Number of Companies Zavidovici 70 Zepce 70 Busovaca 30 Zenica 110 Drvo BH 66 Association of independent craftsmen 2,350 Kiseljak 60 Tomislavgrad 40 Fojnica 92 Association of communal companies 185 Vares 68 Sarajevo Canton 15 Tuzla Canton 10 Gorazde Canton 10 Una Sana Canton 15 Business Club Lasva Vitez 55 Stolac 75 Livno 70 Kresevo 50 Odzak 45 Vitez 98 Novi Travnik 57 Tesanj 40 Maglaj 50 Link Mostar 207 Association of SMEs Ljubuski 68 Kakanj 35 Mostar 70 TOTAL (companies) 4,111 TOTAL (workers) 90,000 Employers'ConfederationRS,see www.poslodavci.rs.ba 93 Appendix Table A10 Pensioncontributions on employee gross wage Country Employer Employee Both Belarus 35.0 1.o 36.0 Uzbekistan 31.6 2.5 34.1 Ukraine 32.0 2.0 34.0 Poland 16.3 16.3 32.5 Romania 22.0 9.5 31.5 Moldova 28.0 2.0 30.0 Albania 21.3 8.6 29.9 KYrgYz 21.3 8.6 29.9 Azerbaijan 27.0 2.0 29.0 Bulgaria 21.8 6.5 28.3 Georgia 27.0 1.o 28.0 Russia 28.0 28.0 Slovakia 21.6 6.4 28.0 Czech 19.5 6.5 26.0 Hungary 18.0 8.0 26.0 Tajikistan 25.0 1.o 26.0 Slovenia 8.9 15.5 24.4 Armenia 19.3 3.0 22.3 Estonia 20.0 2.0 22.0 Montenegro 9.6 12.0 21.6 Serbia 10.3 10.3 20.6 Croatia 20.0 20.0 Macedonia 20.0 0.0 20.0 Turkey 11.0 9.0 20.0 BH-FBH 5.3 12.6 17.9 BH-RS 15.8 15.8 Kazakhstan 10.0 10.0 Source,Bank reports.Note:provisionaldataand subject to comparabilityissues 94 Appendix Table A . l l Annual Growth Ratesin RealAverage Wages, 2001-2003 FBH RS BH Male all workers 10% 5% 10% all employees 9% 5yo 9% employees inthe public sector 8% 6% 9% employees inthe private sector 3% -9% -1% employees in the informal sector 13% 2% 10% Female all workers 8% 7% 9% all employees 8% 8% 9% employees in the public sector 10% 15% 13% employees inthe private sector 18% -12% 5?'o employees inthe informal sector -5% -1% -4% Both all workers 10% 5yo 9% all employees 8% 6% 9% employees in the public sector 9% 9% 10% employees inthe private sector 6% -12% 0% employees inthe informal sector 8% 1Yo 6% Memo (administrative data) Increase in average nominal wages 9% 11% inflation 0% 2% Real 9% 9% Note: Based on workers with positive earnings only Source: Own calculations from LSMS 95 APPENDIX Ill: MTDS/PRSP - LABOR MARKET REFORMS Medium-Term Development Strategy(MTDSFRSP) Action Plan Labor Market Reform - Activity Responsible Institution DeadIine Goal/Comment Changeentity contributionlaws FBH Government,RS Firsthalf of To harmonizeemployment insurancerates Government 2004 and unemploymentassistancecash benefits, equalize rights and increasethe mobility of labor force (the unemployment rates in bothentities are nearly identical, so should be the contributionrates). Changethe FBH Law on BHCouncil of Minister First half of To establishharmonizedcriteria for EmploymentBrokerageand Social FBH Government,RS 2004 acquiring and losing the status of an Security of the Unemployedand Government unemployedperson, to ease the conditions RS Law on Employment for acquiringright to the unemployment cash benefit, to tightenthe criteriafor acquiring the right to health insurance,to equalizethe level o f the cash benefit for BH andto stimulate an increase in mobility of the labor force. FBH Government,RS First half of To allow for subsequent acquisitionof new ~ Adopt the legislationthat will permit out-of-schooleducation Government 2004 and improving on existingqualifications. Adopt the new legal frameword on BH Councilof First half of To increase the efficiency of combating labor inspectorates Ministers,FBH 2004 informal economy. Government.RS Government FBH Government,RS First half of To accelerate the reform process. ~~ Adopt the laws or ordinancesto strengthenthe existing Government 2004 unemploymentprotectionsystem to provideamore adequate protection for workers who, owing to acceleratedreforms (privatization, bankruptcy, liquidationetc.) lose theirjobs, without their responsibility Adopt the Law on Employmentin Governmentof the Secondhalf of To unify employment bureaus inthe BrCko the BrEkoDistrict BrCkoDistrict 2004 District and introduce an unemployment protection system. Change the legislationto permit FBH Government, RS Secondhalf of Justice inuse of unemploymentbenefits. the scope ofthe right to Government 2004 unemploymentbenefits depends on the durationof pastlabor and the periodduringwhich the wage contributions were paid. Adopt the legislativeframework BH Councilof First half of To improvethe system of socialprotection. which will ensure employment o f 3 Ministers,FBH 2004 to 5% of the disabled in the civil Government, RS Government Strengthenthe Economic and FBH Government, RS Continuously To involve employers into the tripartite Social Councils Government, trade dialogue and introduce greater flexibility unions, employers' into the wage system. associations Establishthe BHEconomic and BH Council of First half of To strengthenthe tripartite dialogue at the Social Council Ministers 2005 level of BH. Reorganizeemployment bureaus FBH Government, RS Secondhalf of To increasetheir efficiency. XJRES Consistently implement [BH Council of IContinuously ITo ensureequal rights and facilitate Responsible - No. Activity Insiitution Deadline CoalKomment constitutional changes which Ministers,FBH inthe course 'eturns require all levels of government to Government, RS o f 2004 ensure that the compositionof Government public employees is in line with the ethnic composition of the - inhabitants~fromthe 1991Census 13 Review existingcollective Economic and Social Secondhalfof To reducethe informal economy and contracts Council of FBH, 2004 increaseemployment. Economic and Social - Council of RS 14 Implementthe reformof the wage Economic and Social Secondhalf of To reduce the informal economy and system: reviewthe existing level o f Council ofFBH, 2004 increaseemployment, to link the wage the minimumwage, regulations on Economic and Social system exclusivelyto labor productivity. redundancypay and on Council of RS - calculations of pastalbor 15 Equalize benefitsand rights arising BH Councilof Minister! Firsthalfof To ensure equal rightsof citizens, from unemploymentand from FBH Government,RS 2004 strengthenthe single economic space, healthinsurance Government incrase the mobility of labor force and - sustainability of returns. 16 Preciselydefine and tightenthe FBH Government,RS Firsthalf of To reducethe informal economy criteriafor acquiringand losingthe Government 2004 status of an unemployedperson, as well as for rightto unemployment benefits, and particularly for health - insurance 17 Equalizethe levelof allocations for FBH Government,RS First half of It is unsustainablethat the monthly health insuranceof the unemployed Government 2004 paymentvaries from 4 to 10KMacross the - for all cantons in the FBH cantons for the same set o frights. 18 Encourageemployers to cross from FBH Government,RS Continuously To increasethe number of employees, the informal into the formal sector, Government inthe course reduce the informal economy, financially through exemptions or reductions o f 2004 strengthenbudgetsand funds and open ofwage contributions possibilitiesfor employers to createnew - jobs. 19 EstablishIT connections between FBH Employment Inthe course To exhange available information on labor localemployment agencies Bureau,RS of 2004 demand and increase mobility ofthe labor throughoutBH and establish a Employment Bureau in BH; to assist the unemployedin finding singleand compatiblesystem; jobs. reinforce the employment agencies' role in identifying needs in local labor markets and in employment - brokerage 20 For the groups amongthe young FBH Ministry of Labor Continuously To increaseemployment amongthe special who are in aparticularlydificult and Social Policy, RS categoriesof the young. situation, such as orphans or school Ministry of Healthand drop-outs,special employment Social Protection programs mustbe developed, e. g. through subsidies to companies - which employ them. 21 Within the ongoing education FBH Employment Second half of To direct the structure of the education in reform, perform a detailed analysis Bureau, RS 2004 relationto the domestic and international ofthe demand in the labor market EmploymentBureau, labor demand. FBH Ministry of Education and Science, RS Ministry of Education and - Culture 22 Restrict administrative costs of FBH Government. RS First half of To rationalize expenses and increase the employment agencies as a share of Government 2004 share o f resources channelled into the core - their total resources tasks of the employment agencies. 23 Establish tripartite management of FBH Government. RS Second half of To ensure involvement of all interested employment agencies Government 2004 parties and implementation of obligations - (governments, unions, employers) arisingfrom the ILO conventions. 97 APPENDIX IV: COMPARISON OF BRANCH COLLECTIVE AGREEMENTS AND GENERAL COLLECTIVE AGREEMENTS ~~ Republika Srpska RS Branch Collective Agreements GCA Proposed new draft GCA The scope The IndividualCA (at the enterpriselevelsigned betweenSteeringBoardand the Enternrise'sUnions ) . should be consistentwith this BCA Signing parties BranchTrade Unionso fRS, Governmentof RS or RS Government, SSRS, RS Govt, SSRS, RSCC, relevant Ministry and, insome cases, Chamber of Chamberof Commerce ECRS CommerceRS Obligatory Enterprises at the territory of RS inthe field of the All enterprisesin the RS All enterprises in the RS parties construction and utilities (The Employer) andthe Wage Wage = basex wage base= minimum minimum price of labor minimumprice of labor BaselMinimum coefficient x ( I+ 0.5x n price of labor Wage ) + supplements;n=years o f service. not exceeding 20 I Coefficients Coefficientsvary betweenbranches, from 1to 5-7 Igroup non-qualified - I 1.5 - soecifies a svstem of ' worker coefficientsaccordingto I1group - semi-qualified 1.6 1.8 - complexity of work with worker coefficientsbetween 1 111group- medium 1.7 -2.3 and 6 complexity works, qualified worker IV group -complex and 2.4 - 2.8 diversified work, secondary education V group - complexwork 2.8 - 3.2 ,highly qualifiedworker VI group complex and - 3.3 - 3.9 specializedwork, two year highereducation VI1group -highly 4.0 - 4.8 complex and diversified work, work in organizationand management, higher education VI11group -highly 4.8 - 5.4 complex and expert work, higher education plus specialist degree (master) 98 RS Branch Collective Agreements GCA Proposednew draft GCA IX group highly - 5.4 5.9 - complex and creative work ,higher education, plus PhD Wage increase: special stimulations subjectofthe Rulebook seniority - for each year 0.5%per year of service, 0.5% per year o f service 0.3% per year and not maximum20% morethan 10% work under especially at least 20% at least 20% BCAs difficult conditions achievementin work at least20% at least 20% BCAs most complex functions at least20% at least20% BCAs inthe enterprise Night work 30% Der hour 30%uer hour 35% uer hour Overtime 35% per hour 35%per hour 35% per hour Work on weekend 30%per hour Work on Stateholidays 50% per hour 50% per hour 50% per hour Allowances Hot meal - if itis not one amount ofthe MW one amount of the MW 50%ofMW available in food Transportation to work real costs of public real costs of public real costs of public transportation transportation transportation Annual leavegrants 3 minwages in the 3 min wages in the 3 MWs, but accordingto enterprisein the last enterprise in the last BCAs month month Per diem 30% of the min. labour 30% of the min. labour 15%ofMW funds for heatingand winter food providing up to 10% ofthe MW per day wage per month 3 of hisiher last wages for 10years ofwork history40% of last history35% of last worker salary per year of worker salary per year of service, for 10-20years service, for 10-20years 50%, for 20-30 years, 40%, for 20-30 years, 6O%, for over 30 years 45%, for over 30 years 70% of last worker salary 50% of lastworker salary per year of service 3 Average Wages 4 MWs member ofthe close 2 MWs employee average wage in the averagewage inthe enterprise enterprise long ilness or injury of at least 2 amounts of the at least 2 amounts of the 2MWs the employee average wage inthe averagewage inthe enterprise enterprise long ilness or injury of 2 average wages BCAs the close family member natural disaster at least 2 amounts of the at least 2 amountsof the BCAs average wage inthe averagewage inthe enterprise enterprise Paid leave at least marriage 5 days 5 days 3 days child birth 2 days 2 days 1dav death of aclose family 4 days 3 days 3 days mernher 99 GCA deathof brother or sister 3 days 1day naturaldisaster 3 days illness of the close family 3 days 1day 2 days member moving 2 days 1 day blooddonation 2 days 2 days 2 days Guaranteed Inthe case the employerfaces financialdifficulties, he 65% for no longer than 6 wage is obligedto pay workers the guaranteedwage in the - months amount of 65-70% ofthe employee's basic wage - but not longerthan agreed. Federationof BH F Branch Collective Agreements 2000 GCA 2005 GCA Government, Trade parties Trade unions of metalworkers of BHand Government Governmentand Trade Unions,Employers' of FBH Unions Association Enterpriseswith state or majority state capital at the parties territory o f FBH in the metalprocessingindustry(The Employer) andthe membersof the Unions that signed BCA (Theemployees) All employers All employers employersand employees with special authorization, Vulnerable industriescan whose rights and responsibilities have been asubjectof temporarily opt out of separated contract minimumwage limit Wage Base wage base=minwage, set at net hourly ratenot wage base=min. wage = lower than KM 1.75, 55% ofthe averagewage adjusted annually in the FBH, last reported wage base=min. wage = according to to the "cost by FBH Statistical Office, 55% of the average wage of living, retail prices and insome branches between in the FBH, last reported developmentsin the 60 and 100% by FBHStatisticalOffice economy" Igroup simple,routine - Coefficientssubject to Coefficientssubjectto work 1 BCAs BCAs I1group less complex - work, semi-qualified worker 1.25 I11group - medium complex work, 3 year secondaryeducation 1.55 IV group - more complex, and morediversified work, 4 year secondary education, 1.7 V-I group - verycomplex qualifiedworker 2.1 V-2 group very complex - work, that include organizationof control, management,significant for the success ofthe work at other levels, highly qualifiedworker 2.3 100 IFBH 2000 GCA 2005 GCA discretionaryright of the employer O.6%, not exceeding20%, but only for the same 0.6%, not exceeding20% employer to be regulatedinBCAs to be regulatedin BCAs 30% per hour 30% per hour 50% per hour 30% per hour 50% per hour 50% per hour 20% per hour 20% per hour available in food 70% of the last reported Annual leavegrants average Federalwage 10% of the lastreported averageFederalwage ( minus 30% ifmeals and accommodationis Per diem - organized) not less than 75% of the per diem(iffree meals or Fieldwork (if employee is free accommodationis not comingback home) - organized minus 50%, and not less than 75% not less than 30 days ifbothminus80%) llast reportedaverage Deathof the employee wage in FBH Deathof the employee's 1 last reportedaverage spouse, child or parent wage in FBH 101 FBH Anniversary awards Paid leave marriageof the child 2 days death ofthe close family II member 4 days death ofthe other family member 2 days sickness of the close family member 2 days blooddonating 1 day moving 1 -2 days education 5 days participationin cultural and sport events 3 days 102 APPENDIX V: UNIONSAND INDUSTRIALRELATliONSINAN INTERNAT10NAL C0NTEXT 1, The effects o f unionism or collective bargaining vary depending on whether the economy is competitive or not and whether it is an open or closed one. Each type o f economy can give rise to different effects o f an otherwise identical configuration o f organized labor. Even for the same economy, and union characteristics, effects can be different depending on the organization (or not) o f employers who can react in a centralized or decentralized way and on how Governments perceive and apply labor and broader social policies, There are potential dozens o f combinations o f different types o f economies, unions, employers and collective bargaining arrangements and each one deserves to be examined in its own right. The BH economy i s no exception to this and the remarks below provide an overview o f the issues involved and attempt to summarize the findings from the international literature.82 WhatDo Unions Do? 2. Unions and collective bargaining can affect positively or negatively economic performance. Though much depends on the specific conditions in which the social partners operate, much depends also on what unions do. The views on what unions do can classified in three main groups. 3. First, the traditional "monopoly view" focuses on the social costs o f unionization arising from distortions when unions succeed to obtain more favorable pay and more general work conditions for their members than non-unionized workers in an assumed perfectly competitive economy (not just the labor market)83. When unions succeed in securing such benefits that would not have been possible under competitive conditions, they impose costs on society, which are called the monopoly costs o f unions. 4. A second view also considers unions to be detrimental to economic outcomes as they create rent-seeking costs. In this view, unions are seen as representatives o f special interests o f their members in collective bargaining and in the political process. Unions can promote policies that reduce competition in labor and product markets. This includes support for minimum wage legislation or trade protection. Unions support such policies ifthey increase the surplus available for sharing with the firm (the effect o f less competitive product markets) or increase the union's 82The current review is drawn from T. Aidt and Z. Tzannatos (2002) "Unions and Collective Bargaining: Economic Effects in a Global Environment" Washington DC: The World bank ((Publisher). 83This is a"strong" assumption Le. unlikely to be met inpractice inthe real world. But it serves as auseful theoretical point for benchmarking variations of economic models. bargaining power (less competition from nonunion labor markets). In addition, unions can engage in political activities and can generally involve three types o f (static) costs. First, to the extent that the union is successful in getting government regulation, an economic distortion is created, and the resulting inefficiency creates a deadweight loss to society. Second, real resources are withdrawn from production to be used in rent-seeking: To the extent that the shadow price o f these resources is positive, this constitutes another loss to society. Third, since the union's distributive success typically comes at the expense o fnonunion workers and consumers, a union's political activities may be associated with large distributional costs. In addition to these static costs, rent-seeking can lead to dynamic costs: rent-seeking can be harmful to innovations and thus hampers economic growth. 5. The third, and final, view, which is called the "organizational view" o f unions, counters the previous two views and acknowledges that unions can have economic benefits. Unions can facilitate worker-participation and worker-manager cooperation in the workplace. This can have efficiency-enhancing effects that jointly benefit workers and management and are called the participatory benefitsof unions. Such benefits can arise because, first, unions are institutions with a collective voice and can communicate worker preferences to management, as well as participate inthe establishment o f work rules and seniority provisions in the firm's internal labor market. This changes the "exit-voice" tradeoff o f workers by providing a channel through which they can express their grievances ("voice") without havingto leave the firm ("exit"). By reducing turnover ("voting with the feet"), this channel increases the incentive o f employers to provide firm-specific training, and facilitates long-term working relationships that can benefit all parties. From their side, unions may help to enforce contracts between workers and management and thus increase productivity by providing a channel through which labor can draw management's attention to changes in working methods or productiontechniques that may be beneficial to both parties. This channel also offers a mechanism by which the union can "shock" management into better practices (reduce X-inefficiency). 6. We can summarize the above discussion o f the costs and benefits o f unions using the following simple equation: Net benefit of unions =participatory and dispute resolution benefits - monopoly costs rent-seeking costs. - Alternatively, one can talk of cost and rewrite this equation as Net cost of unions = monopoly costs + rent-seeking costs -participatory and dispute resolution benefits. 7. From a theoretical perspective, the net benefitkost o f unions i s ambiguous and dependent on the relative size o f the three components. These inturn depend significantly on the economic, political, and organizational environment in which collective bargaining takes place. The economic environment affects both the monopolv costs and the participatorv benefits. J& political environment determines the rent-seekinn activities o f unions. The organizational environment (bargaining coordination, social partnership, and dispute resolution) affects all three components. Thus, judging the contribution o f unions and, more generally, collective bargaining to the achievement o f economic and social outcomes is, at the end of the day, an empirical question as unions are only one agent in the labor market which is thought to be tripartite (workers, employers, governments). Much depends also on what the other two social partners do and how all three interact in an attempt to determine wages, employment and work conditions. Eventually, all three are constrained by the state o f the national economy (competitive or not) and its relationship to the rest of the work (trade openness). In other words, it is the coordination 104 between the social partners that determine outcomes, notjust simply the membership o f unions or employers associations. WhatDo Employers' OrganizationsDo? 8. The members o f an employers' organization are individual firms, typically within a particular industry. Each employers' organization may in turn be a member o f a national employers' organization. A firm may decide to join an employers' organization to improve its bargaining position with workers (possibly organized in a union). Firms derive their bargaining power from their ability to lock out workers. The cost o f an industrial conflict from the point o f view o f an individual firm i s larger than the cost to the industry as a whole. This is because an individual firm involved in a strike i s likely to lose its market share to other firms in the industry that produce close substitutes.Accordingly, whereas each firm has an incentive to give into wage demands (called wage d@) to avoid a local conflict, the industry as a whole has less incentive to do so, and byjoining forces, it i s easier for firms to resist wage demands from unions. 9. Employers' organizations can also help reduce leapfrogging. Leap frogging occurs when individual firms increase their wage rate to extract more effort from existing workers or to attract skilled workers from other firms. When all firms engage in this kind o f behavior, the net result may well be that relative wages are unchanged but the level o f all wages has increased substantially. A strong employers' organization that coordinates the behavior o f individual firms can be helpful in internalizing this "efficiency wage externality" and preventing wage drift (as discussed above). 10. In addition, employers' organizations play an important role inproviding training: Since general training is a public good, firms are unlikely to provide much o f it unless they are subject to external pressure. A strong employers' organization can provide training facilities for firms and can impose sanctions if a firm does not pay its share o f the cost. Dispute Resolution 11. The breakdown o f negotiations between individual workers and their employers can take various forms, ranging from poor relations in the workplace (with potential costs including decreased levels o f labor productivity through poor morale) to labor turnover (the "exit" option, with the potential loss to the employer o f previously made investments in the workers' human capital). At the level o f collective contracting, the stakes are arguably much higher for both workers (and their unions) and employers, with the ultimate cost o f a negotiation breakdown being lost incomes for the workers and lost profits for employers. Giventhe potentially high level o f these costs to both contracting parties, it is likely that workers and employers have a strong incentive to achieve a solution in preference to conflict. Like all good threats, the employer's threat o f a lockout and the union's threat o f a strike are best if they ensure that an agreement i s reached before they are implemented. 12. In reality, collective bargaining does sometimes break down, and production, labor earnings, and profits are lost. It i s certainly not safe to assume that the total o f such costs is greater under the collective bargaining system than under the individual contracting system. We simply do not know whether these costs to society are greater or less than those that would arise from a breakdown in individual employer-employee pay negotiations. Indeed, given economies o f scale in the production and dissemination o f information, there are grounds for believing that the collective system, through its ability to resolve disputes, may be a less costly option from a social point o f view than individual contracting. 105 13. There is a strong presumption that when disputes do occur under collective bargaining, it is because o f asymmetries in the information possessed by the involved parties. A common example i s when the trade union "misjudges" the maximum wage that the employer i s willing or able to pay. Under such circumstances, the existence o f regulationcan prove decisive inresolving disputes through its information-gathering and -disseminating roles. 14. To understand the process, it is important to recognize the distinction between the union proper (sometimes called the official union) and its rank and file membership. Under this tripartite framework, the official union (often as a well-informed professional body) acts as an intermediary between the union membership and the employer. As such, its role is to reconcile the aspirations o fthe former against what itjudges (on the basis of its more complete knowledge o f the overall situation than that possessed by the union membership) that the employer would agree to pay. This reconciliation between worker aspirations and labor market realities may be achieved without either party having to resort to its no-trade sanction. However, should negotiations break down and a dispute occur, the role o f the official union as a purveyor o f information continues, with information being disseminated in both directions regarding concessions acceptable to each party and any new information that may materialize as the dispute progresses. This transmission o f information continues until demands fall into balance with offers, at which time a settlement is achieved. 15. Viewing a union as an information-gathering and -disseminating body suggests that governments might want to adopt policies that increase the efficacy with which unions fill this role. The introduction o f so-called cooling-off periods, duringwhich all parties take time to assess the situation fully before implementing no-trade strategies, is one example. Other such policies might require that the employer (generally seen as the party in possession of more complete information) divulge to the union and its members certain types o f information, perhaps in a standard form, to minimize the possibility that disputes will arise because workers incorrectly estimate the employer's ability to pay. 16. Some conflict is inevitable when wages and other employment conditions are set by negotiation (either collective or individual), rather than by the invisible hand o f the market. And o f course the above view that trade unions are well-informed bodies that look after the interests o f their members may not be always true: there have been examples when union leaders have genuinely misjudged the actual situation or escalated conflict more for their own reasons.84 Recognizing this fact, there are grounds for believing that a well-run centralized, union-based system o f wage bargaining may be less costly to society than an individually based negotiating system in terms o f both total transaction costs and dispute costs. Alternatively, miscalculations or mistakes in a centralized system may have more pandemic effects than those that would arise in a decentralized framework. Summary of Outcomes 17. There is no single variable capturing the "goodness" or "badness" of unions - both terms are often used to describe unions in the common press (though not in the professional literature). At the micro level economists have used many different indicators o f labor market performance such as those affecting workers (level o f wages, hours o f work, job mobility, training, benefits) or firms (profitability, productivity, employment growth, investment, research and development). 84 Suchreasons may be personalambition butneednot always be so: they may, for example, relateto the unions leaders' politicalbelief that the time is ripe for conflict inabroaderpolitical context. 106 Obviously, some good things for one party (for example, higher wages or shorter work hours for workers) are bad for the other (for example, low productivity or profitability). 18. At the macroeconomic level, economists have examined certain key indicators (such as economic growth, aggregate productivity, inflation, unemployment, wage growth, earnings inequality and employment growth) or combinations o f various indicators (such as the Okun's index - the sum o f the unemployment rate and the inflation rate, or the job quality index that compares changes in employment against wage dispersion). Others have focused more directly on effects on the aggregate labor market examining, for example, the effects on unions on aggregate real wage flexibility (to unemployment), employment adjustment speed to real wage shocks or search effectiveness (the relationship between vacancies and unemployment). 19. Inall, the international literature on the subject covers thousands o f publishedarticles and books and utilizes more than a couple o f dozen micro and macro indicators that are affected by many and diverse combinations o f collective bargaining operating under different political regimes. The first, and safest, conclusion can only be that the jury is still out. In other words, whether unions and collective bargaining are "good" or "bad" depends on what they do, what employers do, how the two parties interact inthe labor market, how the government interferes (or not) inthe economy, and what kind o f economy the social partners face (competitive or not, open or close). 20. Still, some generalizations are possible. At the micro level, empirical evidence confirms that unionized workers and those covered by collective agreements enjoy higher wages and shorter working hours and tend to receive more training than comparable non-unionized and uncovered workers. The union effects on other aspects o f the labor market are more diverse. For example, voluntary turnover is lower and job tenure i s longer in unionized firms but layoffs (particularly temporary ones) are more frequent in unionized firms. R&D tends to be lower in unionized firms which however seem to adopt new technology as fast as non-unionized ones. 'Productivity effects can be in either direction depending on the industry, country and time under investigation. 21. At the macro level, the evidence is generally more fragile though labor market coordination (not necessarily through high unionization rates) seems to lead to lower dispersion o f labor earnings and is at times associated with lower unemployment and fewer strikes. However, both coordinated and uncoordinated labor markets seem to fare comparably interms o f productivity growth and wage flexibility while there is no substantial difference between these two systems interms o f inflation and employment rates. vi 22. There is a large body o f empirical literature on the economic effects o f unions derived from microeconomic data on individual workers and establishments. It covers the uniodnonunion wage markup, both overall and also in terms o f skills, gender, occupation as well as the underlying economic conditions and the institutional environment. It also examines the effect o f unions on other economic variables such as employment growth, hours worked, productivity, job mobility, the implementation of new technology, physical investments, spending on R&D, training o f workers, profitability, fringe benefits, mode o f pay, and pension schemes. These are examined below. 107 The WageMarkup in Different Countries 23. The uniodnonunion wage markup (the "wage markup") is defined as the difference between the average wage o f unionized and non-unionized workers with similar individual and workplace characteristics divided by the average wage o f a non-unionized worker.1 The markup can be estimated in different ways. First, it can be estimated as a membership markup. The membership markup is based on information about an individual's union status and calculates the difference in wages between individual unionized and non-unionized workers. Second, it can be estimated as a recognition markup. Here workers are being categorized according to whether or not their pay i s determined by a collective agreement between a recognized union and a firm. In the latter case, individual union membership is not crucial. What matters is whether the workers' pay i s determined by a collective agreement. The distinction between the membership and the recognition markup i s important when not all the workers whose wages are determined by a collective agreement are union members. For example, when many workers are covered by collective agreements although they are not members o f a union (such as inFrance where the two figures stand at around 90 percent and 10 percent), estimates based on the membership markup underestimate the "true" markup, and it is preferable to use the recognition markup to measure the impact of unions on wages. 24. In all countries where the wage markup has been estimated, it has been found to be nonnegative (that is, it i s at times found to be statistically zero). There are, however, significant cross-country variations as well as variations o f estimates within countries and over time. There is also some evidence, albeit weak, that the wage markup is, on average, lower in high-income countries than in low- and middle-income ones. 25. The most reliable picture comes from the United States and the United Kingdom by virtue o f the many studies that have been carried out and the broad consistency o f the results. The U.S.wage markup has been estimated in more than 200 studies and is generally found to range from 12 to 22 percent. Though there i s consensus that the average markup i s approximately 15 percent this may significantly overestimate the wage gain and that the true impact o f unions on wages may be around or less than 10 percent. In the United Kingdom, more than 20 studies have estimated the markup to be inthe range of 3 to 19 percent and the average is probably also around 10 percent. 26. The evidence is sparser for other industrialized economies, but some generalizations are possible. The conventional view is that the wage markup for Australia is in the range between 7 and 17 percent. Similarly, the Canadian wage markup has been estimated to be in the range of 8 to 25 percent, although decreasing over time reaching 7- 10 percent in more recent periods. In (West) Germany, where most unions are industry unions and work and pay conditions contained in collective agreements are largely extended to non-unionized workers, the wage markup is found to be small, especially for male workers. Similarly, studies on Japan have found a small average wage markup o f about 5 percent. 27. The evidence i s even more limited for low- and middle-income economies. In the Republic o f Korea, estimates o f the markup for male workers in manufacturing industries were found to be below 4 percent in 1988 but higher later on (7 percent in 1991). The union membership wage markup in South Africa is among the highest (around 24 percent for black blue-collar workers) as i s also for Malaysia (in the range o f 15 to 20 percent depending on the type o f union involved) and both countries can be thought o f having large segments o f the labor market who, in the absence o f minimum wages, would have been quite vulnerable. For Mexico, the mark up is found to be around 10 percent but for Ghana it rises to more than 20 percent. 108 28. Some studies have estimated the membership wage markup in a cross-country context. They generally confirm that the markup is positive but often not significant (for example, in countries such as Canada, Israel, the Netherlands, Spain, and Switzerland). 29. The markup can vary depending on the percentage o f wage- and salary-earners that are unionized or covered by collective agreements. While union density per se appears to be largely unrelated to the wage markup, bargaining coverage is negatively correlated with the markup. In other words, when more workers are covered by collective agreements, the smaller, other things being equal, the wage markup appears to be. This suggests that the more workers become unionized or are covered by collective agreements, the lower is the markup that they can secure. This may be the case because the labor supply in the non-covered sector decreases when more workers become covered, pushingthe nonunion wage up. Taking this argument one step further, one may say that unions are able to secure a highmarkup only where the marginal cost to society (in terms o f impact on the macro economy) is small. In effect, unions are ultimately constrained by the wage share in the total economy: they can have wide coverage and a small markup or a highmarkup at the cost o fcoverage. 30. An interesting additional question is whether the markup is stable over time or it fluctuates with economic conditions. This question has been investigated in only two countries- the United States and the United Kingdom. The conclusion is that the wage markup inthe United States has moved procyclically but does not have a trend over the period 1983-95. For the UK the markup has declined a bit during the 1980s probably due to a fall in the wage markup for female workers. These results suggest that union power has not been curtailed significantly, despite the reduction inunion density observed over the same period inthe two countries. TheEfficiency Cost of the WageMarkup 31. When unions are successful in getting a wage markup, workers tend to be displaced from the unionized sectors to nonunion sectors. This creates a deadweight loss. A number o f studies have estimated this deadweight loss and found it, somewhat surprisingly, to be quite small. For the US estimates of the welfare loss is found to be less than 0.5 percent o f GDP (that i s less than one percentage point) . The simulated welfare loss associated with a 25 percent uniodnonunion wage markup is similarly found to be no larger than 0.2 percent o f GDP. Interestingly, the U.S. results are similar to those for Australia, where the average markup is 7 to 17 percent and where 80 percent o f the work force i s covered by collective agreements (compared to 15 percent in the US). TheDifference in the Markupfor Womenand Men 32. Unions are just one o f many determinants o f the gender wage gap. The gender wage gap i s the percentage difference between the wage o f a female worker and a male worker who otherwise have the same personal and workplace characteristics. The effect o f unions arises in three ways: first, from different unionization rates among men and women; second, from the ability o f unions to influence wages in some sectors or workplaces but not in others; and third, from differences in the wage markup for men and women. In principle, the net effect o f unions upon female wages relative to male wages is uncertain: though a higher wage markup for women than for men can reduce the gender wage gap, it can also decrease the wages o f non-unionized women to such an extent that the gender wage gap actually increases. 33. Inthe US.literature, there is very little, ifany, difference betweenthe markup for female and male workers. The same result emerges from Australian studies though most studies in 109 Britain show that the impact o f unions on women's wages is greater than that on men's wages: A typical estimate is that the markup for women is 4 to 6 percentage points larger than that for men though a few studies find the opposite result. In any case, taking into account the fact that women workers are less likely to be unionized than men workers, the net effect on the average gender wage gap is likely to be small. Evidence from other OECD and middle-income countries unambiguously supports the view that the wage markup is greater for women (for example, in Canada, Japan, West Germany, and Mexico). Differences in the Markup by Ethnic Group 34. Discrimination among workers with different ethnic backgrounds but otherwise similar productivity characteristics can lead to a wage differential. In the United States, it is not clear whether there i s a substantial difference between the wage markup for white and nonwhite workers. Some studies fail to find any difference, whereas others find that the markup i s 5 to 10 percentage points higher for blacks than for whites. In the United Kingdom, the sparse available evidence shows that a nonwhite unionized worker gets a higher markup than a similar white worker. The Private versus the Public Sector 35. It is unclear ifpublic sector workers are ina weaker position than private sector workers to exert wage pressure. Historically, they have been restricted from forming unions in many industrial countries. Even when public sector unions are legal, they are often legally barred from striking.This suggests that the average wage markup inthe public sector may be smaller than the corresponding markup in the private sector. On the other hand, public goods and publicly provided private goods are produced in an environment with no or little competition. Moreover, producers inthe public sector are not motivated by a profit concern; rather, they have a politically imposed budget constraint. The lack o f competitive pressure and soft budget constraints makes it easier to pass the costs o f high wages and overstaffing on to taxpayers. Finally, public sector unions may be able to influence employers' behavior through the political process. In many developing countries, unionization is concentrated in the public sector, and there are few legal constraints on the kind o f behavior that these unions may engage in. In such an environment, unions are able to exercise substantial political pressure. This may contribute significantly to the rent-seeking cost o f unions in addition to the impact it may have on the wage markup. 36. There is a number o f specific problems associated with estimating the public sector wage markup. First, in the private sector, wage and working conditions o f unionized workers are determined by a union-negotiated contract, whereas nonunion workers are typically excluded from the benefits o f the union contract. In the public sector where the "wage-comparability" criterion is extensively used, it i s not uncommon for both union and nonunion workers to get the same wage and working conditions. This makes it more difficult to classify workers according to how their wage is being determined and makes it preferable to use the recognition markup rather than a membership markup. Second, the work force mix differs between the private and public sectors. For example, the work force inthe public sector generally consists o f a disproportionately large share o f white-collar workers. Since the markup for white-collar workers tends to be smaller than that for blue-collar workers, failing to take the work force mix into account can underestimate the markup in the public sector. Third, workers inthe two sectors receive different amounts o f fringe benefits. 37. There are more than 75 US.studies that have estimated the wage markup for the public sector at large or for some specific groups o f workers within the public sector (such as teachers). 110 The conclusion here is that the public-sector gap averages about 8 to 10 percent. Within the US public sector, the wage markup is lowest for federal employees and highest for employees o f local governments. In fact, the average markup for workers employed by local governments is slightly higher than the economy-wide average. The difference inthe wage markup for workers in the public and private sectors in the United Kingdom is smaller for both manual and non-manual workers in the public sector than in the private sector. Other studies, including for Canada, conclude that the wage markup in the private and public sectors is very similar or that the difference is small, about 2 percentage points infavor o f the private sector. TheMarkupfor Workers with Different Skills 38. In both the United States and the UnitedKingdom, manual workers get a larger markup than non-manual workers. Likewise, semiskilled workers get a larger markup than skilled workers. Similar results have been obtained in Canada while in other countries (such as in South Africa) the wage markup for workers with different skills varies between different ethnic groups. Unions, WageDispersion, and the Return to Schooling 39. The wage mark-up for unionized workers and concentration o f unionization among low- paid workers suggest that unions reduce the wage dispersion across an economy. Moreover, the impact o f unions on the wage dispersion within the unionized sector can also contribute to low overall wage dispersion. For example, the differences between the wage mark-up for different skill groups (see above) indicate that unions contribute to the compression o f wages within the unionized sectors o f the economy. 40. There are many reasons why unions may be keen to promote a compressed wage scale across different groups o f workers employed within the unionized sectors o f the economy. One reason may be that they have egalitarian wage goals. Egalitarian wage goals can arise if productivity differs among union members and if the median member has low (compared to average) productivity. Under these circumstances, a democratic union tends to enter wage contracts that compress the wage structure. The evidence from different studies generally confirms that unions reduce wage dispersion significantly between industries, between (similar) f i r m s within an industry, and among workers within a firm. In fact, the decline in unionization in the United Kingdom in the 1980s and early 1990s contributed significantly to the rise o f wage inequality observed over the period. This suggests that unions play an important role in propping up wage levels at the bottom o f the wage structure. 41. A side effect o f the fact that unions tend to reduce the wage dispersion realates to the decisions people make about their children's and their own education. These decisions depend on a number o f factors. One important factor is the return to education in terms o f higher (future) wages. If unions reduce the return to schooling, say, by compressing the wage differential between workers with different skills, they can have an adverse impact on the formation o f human capital. On the other hand, when the relative wage o f unskilled workers increases, firms substitute away from unskilled workers. To avoid being unemployed, (unskilled) workers have to acquire more skills, so the compression o f the wage distribution may induce more, rather than less, human capital formation. The empirical evidence suggests that the wage markup i s usually higher for less educated workers or, in other words, the wage markup decreases as the education level o f workers increases. 111 The Economic Environment 42. The economic realities facing firms can make it difficult for unions to get a high wage markup. Competitive pressure from both the product market and the nonunion labor market can be particularly effective in serving this role. A number o f studies have investigated the effect o f competitive pressure on the wage markup. Most o f these studies use industry concentration (the small number o f firms controlling most o f the output in a given industry) as a proxy for a firm's market power. 43. Inthe United States, the majority of studies find a negative correlation between industry concentration and the wage markup in manufacturing industries. This is also the case in the United Kingdom though in Canada some o f the evidence suggests that industry characteristics such as concentration, import penetration, and labor substitutability have little impact on the wage markup. 44. These results do not support the theoretical predictions. However, the relationship between monopoly power and the wage markup can be masked in these estimates if wages are highinconcentrated industries even inthe absence of unions. For example, this would be the case if firms in these industries wish to forestall unionization. Another reason is that firms in concentrated sectors would like to escape possible enforcement o f competitive laws or they want to avoid the bad press associated with high profits and low wages. These reasons may be more apparent than real. For example, the fear o f provoking the response o f the competitive authorities i s hypothetical in many countries where there i s little faith in competition laws and little effort i s spent on enforcing them. Also, firms may receive better press coverage by spending part of their excess profits on health and safety improvements in the workplace or by donating to charities rather than by paying higher wages. One interpretation o f the evidence i s that firms in concentrated industries use their monopoly rent to withstand the wage demands o f unions. This may induce workers to be content with greater job security and other non-monetary benefits as a substitute for high wages. Finally, the industry concentration ratio may not be a good proxy for a firm's monopoly power inthe product market. 45. This suspicion seems to be confirmed by a few studies that have used indicators other than industry concentration to measure the monopoly power o f firms. These studies find that the wage markup is larger in firms with monopoly power than in those without it. For example, the use o f a mixture o f industry concentration ratios and a subjective measure o f entry barriers to the industry as a proxy for monopoly power in a sample o f unionized U.S. manufacturing firms suggested that the wage markup is significantly higher in noncompetitive industries than in competitive ones. In the UK the use of subjective indicators o f competition (such as asking management o f firms about the number o f competitors that they are facing in the product market) suggests that competition in the product market significantly reduces the average wage markup. In particular, in firms that operate in a competitive product market, the wage markup is, on average, zero. On the other hand, firms that have little or no competition in the product market grant a wage markup in the range o f 8 to 10 percent. Moreover, unions are unable to create a wage markup in firms that primarily operate in international markets. In short, unions are able to create a markup only in industries that are sheltered from foreign competition and the whole industry is heavily unionized. TheDesign of CollectiveBargaining 46. The evidence generally suggests that workers, on average, get a wage markup ifthey are members o f a union or otherwise have their pay conditions determined by collective agreements. 112 However, the size o f the markup may depend on how collective bargaining is organized. The institutional framework can be organized infour groups: 0 The extent o f unionism (average union density inthe industryor the percentage o f firms inthe industry that recognizes a union). 0 The level at which bargaining takes place (the firm, industry, or the national level). 0 Multi-unionism (when more than one union represents similar workers inthe same workplace). 0 Closed shops (a worker can obtain or retain a particularjob only ifhe or she is a member o f a particular union). 47. The extent of unionism. There seems to be a strong relationship between the extent o f unionism in an industry (or occupation) and the wage markup. In industrieswhere unionization is low in terms o f either density or the percentage o f firms that recognize a union, unions generally have little impact on wages. This is because attempts to raise the wages paid by a few unionized employers (above what their competitors pay) put union employers at a severe disadvantage in the product market. This increases employers' resistance to union wage pressure and encourages the union to moderate its wage demands. On the other hand, in industrieswhere almost all firms are unionized, unions will have more bargaining power and will therefore be able to secure a higher wage markup. This is knownas the "extent o f unionism" effect and inthe US increases the membership markup, although there is substantial disagreement about the magnitude o f this effect. A few studies for Canada confirm the positive effect. In the UK the wage markup is larger in industries with more than the 70 percent threshold for union density. 48. In the United States, there is no significant difference between union density and coverage o f collective agreements (both come to about 15 percent o f the workforce). In the United Kingdom, on the other hand, a large number o f workers have their pay conditions determined by collective agreements without actually being members o f a union. Therefore, focusing on the membership wage markup may bias not only the estimate o f the wage markup itself but also the estimate o f the extent o f the unionism effect. More recent British studies take this into account and analyze the relationship between the recognition wage markup and union density at the establishment level. The recognition wage markup (for semiskilled manual workers) is significant only for firms where more than 95 percent o f the work force i s unionized: The markup i s in the range o f 7 to 10 percent. In firms where a smaller fraction o f semiskilled workers are union members, the wage markup is insignificant. This suggests that workers in a workplace where management does recognize a union benefit in terms o f higher wages only if almost all the workers in the workplace are actually organized inunions 49. The level at which collective bargaining takes place. The level at which collective bargaining takes place affects how workers and employers interact, and this has implications for the size of the wage markup. One would expect the wage markup to be higher when collective bargaining is at the industry level rather than at the firm level. This prediction fails to be supported by evidence for the UK. In fact, some studies find that the biggest gains in the wage markup arise when collective agreements are fixed at the local level rather than at the national level. However, for the US, some early studies found that the wage markup was higher in industriesthat used centralized bargainingthan inindustries that used firm-level bargaining. 50. In Malaysia, the average wage markup paid by a firm that deals with an industrial union i s 20 percent, compared to 15 percent in firms that deal with a company union. In India, members o f independent plant-based unions (unions run and managed by workers employed in the plant) 113 get significantly higher wages and bonuses than workers affiliated with external unions (unions that are explicitly affiliated with a trade union federation). 5 1. Multi-unionism. The prime source of information on the effect o f multi-unionism is the United Kingdom where multi-unionism plays an important role. Inthe 1980s, about 30 percent o f all unionizedplants inthe private sector recognized more than one union for collective bargaining purposes. The 1998 Workplace Employee Relations Survey reports that still about 38 percent o f all workplaces have multiple unions. Under multi-unionism, the unions may bargain together (multiple bargaining) or separately (separate bargaining) with management. Multi-unionism is generally associated with a higher wage markup. Some studies have refined this result and show that it i s not multi-unionism per se that i s associated with the additional wage markup: it is the combination of multi-unionism and separate collective bargaining that produces the additional wage markup. Ifall the unions that represent workers at a given workplace bargain together, then the wage markup is no larger than infirms where workers (of the same type) are represented by a single union. 52. Closed shops. A closed shop exists when an employee can obtain or retain a particular job only if he or she is a member o f a particular union. The closed shop can be either pre- or post- entry. A pre-entry closed shop requires that the employee i s accepted as a member o f the relevant union ("holds a union card") before he or she can be employed in the particular trade. Historically, craft unions have managed to run a pre-entry closed shop. One example i s the International Typographical Union in the United States, which at its peak required that all individuals hired for the composing room must already have union cards. A post-entry closed shop requires that the employee joins the union upon getting a specific job. In the United Kingdom, post-entry closed shops used to be important in industries such as metal engineering, transport, and communications with 23 percent o f their workers being covered by a closed shop arrangement o f all workplaces in 1980 (and 88 percent in nationalized industries). But the importance o f such arrangements has declined significantly over time: By 1998 merely 2 percent of all workplaces had a closed shop arrangement. From a theoretical point o f view, a closed shop increases a union's control over labor supply and as a result its bargaining power. The question, therefore, is whether the presence o f a closed shop increases the wage markup over and above the basic recognition or membership effect. Post-entry closed shop does not increase the wage markup above what it would have been had the majority o f the firm's workers been unionized. On the other hand, pre-entry closed shops can increase the wage markup by as much as 100 percent. However, subsequent research finds that the premium associated with the pre-entry closed shop has been reduced and is roughly the same as that found in firms where management recommends union membership. UnionsandEmployment 53. The wage markup reduces total employment as long as the demand curve o f labor in the unionized sector o f the economy is sloping downward and the management o f unionized firms retains the right to manage (that is, management independently decides on employment after wages have been agreed with the union). However, the adverse employment effect o f an increase in wages can be reduced and even be reversed if (a) unions and firms bargain over wages and employment and agree on what is called "efficient contracts" or (b) firms have monopsony power in the absence o f collective bargaining. Some empirical evidence on whether unions and firms bargain over employment suggests that this is rarely the case in either the United Kingdom or the United States. In the United States, many contracts explicitly state that the right to determine employment remains with the management. While this is not true in the United Kingdom, U.K. unions do not generally bargain over employment. 114 54. However, although the employment level per se is not subject to formal bargaining, recruitment, staffing norms, redundancy pay, and deployment can be included in formal bargaining, and this can have indirect effects on employment. It is possible to test econometrically the right to manage model (unions push up wages and reduce employment) against the eficient bargaining model (unions pushup wages and employment). The results often reject both models and one i s tempted to conclude that "on the whole neither theory seems to be able to account satisfactorily for the data on negotiated wages and their associated employment levels." Although it is tempting to argue that the truth should lie somewhere in the middle, the clear answer may be constrained by data limitations and flawed econometric procedures. 55. Another way to assess the impact o f unions on employment is to look at employment growth noting that permanent employment growth differences between unionized and non- unionized firms are unlikely to represent long-run equilibrium positions. With this in mind, the available evidence from Canada, Jamaica, Malaysia, the United Kingdom, and the United States suggests that employment grows more slowly in unionized firms than in non-unionized ones. Studies from Canada, the United Kingdom, and the United States typically find a growth differential in the range o f 3 to 5 percentage points per year in favor o f non-unionized firms. There are a number o f possible explanations for the observed employment growth differential: Ittakes time and effort to organize a union. Consequently, at a given point intime, old firms are more likely to be covered by unions than newer firms are. Ifnewer firms expand faster than old firms, we would expect to observe higher employment growth in the newer, non-unionizedfirms. Unions are more likely to be concentrated in sectors that enjoy large rents. Ifthese sectors are less dynamic because o f monopoly inefficiencies and their activities are limitedby the size o f the domestic market, employment would tend to grow more slowly inthese sectors. 0 Unions may encourage labor hoardingby increasing hiringand firing costs. This would make unionized firms more reluctant to hire new workers duringa boom, thus reducing employment growth over the cycle. 0 Labor costs may grow faster in unionized firms than innon-unionized ones. 0 Productivity grows slower in unionized firms than innon-unionized ones. Voluntary Turnover, Layoffs, and Job Tenure 56. The evidence from Australia, Malaysia, the Republic o f Korea, the United Kingdom, and the United States unanimously shows that voluntary turnover (measured by the "quit" rate) is lower and job tenure is longer in unionized firms than in non-unionized ones. Perhaps the most convincing recent evidence that unions reduce job separations comes from Japan: the job separation rate is significantly lower in unionized firms than in non-unionized ones. Importantly, the Japanese experience has been attributed to mechanisms that provide "voice" for the employees and reduce separations in unionized firms only. All this gives support to the "collective voice" view o f unions and points to one o f the key channels through which unions can add to workplace productivity, namely by increasing the lengthofjob tenure. 57. The welfare gain associated with a reduction in labor turnover has been found to be equivalent to a 0.2 to 0.3 percent increase in GDP inthe United States inthe 1980s.For unionized firms, the gain i s estimated to be equivalent to a 1 to 2 percent reduction in costs. While these estimates are usually crude, it is interesting to notice that the welfare gain associated with participatory benefits o f this kind i s o f the same order o f magnitude as the estimated monopoly 115 cost o f unions. However, the participatory benefit accrues to organized workers (and firms) only, whereas the monopoly cost o f unions is borne by society at large. 58. While voluntary turnover tends to be lower in unionized firms than elsewhere, unions may increase the use o f layoffs, particularly temporary layoffs. Inan early study o f layoff patterns in U.S.manufacturing firms inthe 1960s and 1970s, unions were found to significantly alter the firm's choice between layoffs, wages, and hours worked in response to business cycle fluctuations. Unionized firms adjust by making temporary layoffs rather than by reducing weekly hours (work sharing) or wages. In particular, unionized (blue-collar) workers are much more likely to be laid off temporarily than non-unionized workers. One explanation for this may be the fact that junior workers can be laid off more easily. Also, senior workers typically have more influence on the union's policy thanjunior workers. Faced with the choice between a reduction in their earnings or a temporary layoffofjunior workers, unions are likely to prefer layoffs. Another explanation i s that the cost o f temporary layoffs can be shifted onto the unemployment benefit system, As long as there is less than a 100 percent experience rating (in other words, as long as the amount that firms contribute to unemployment benefits is less than the costs of the unemployment that they generate), those firms with above-average layoffs are subsidized at the expense o f firms with below-average layoffs. However, while it is true that the existence o f unions increases layoffs in the private sector, the opposite is true in the public sector. Public sector unions do not increase the wages o f their members as much as private sector unions do. Instead, they reduce layoffs and protect employment. Unionsand Hours Worked 59. The effect o f unions on the total number o f hours worked by their members (compared to non-unionized workers) i s not a priori clear. On the one hand, unions typically demand lower normal hours, more holidays, and so on. Conversely, they may be able to secure overtime work at higher rates o fpay. 60. Overall, the finding is that unions reduce the total number o f hours worked. In particular, the evidence suggests that workers in unionized firms work fewer normal hours. Moreover, unions reduce the number o f unpaid overtime hours and, in some cases, increase the amount of paid overtime work. Furthermore, unions increase the likelihood that workers receive paid holidays and unionized workers get, on average, additional weeks o f holidays compared to non- unionized workers. While these results are relatively robust and the unionhonunion total hours differential for in OECD countries i s negative, union workers seem to work more in Spain and Switzerland. The estimates range from a one- to two-hour differential per week in the United States and the UnitedKingdom to a four- to six-hour differential per week in Austria and Ireland. Unions and Profiability 61. It is a commonly held view that unions reduce the profitability o f firms because they appropriate part o f the rent that would otherwise have been available to shareholders. The fact that unions are able to get a wage markup supports this view. However, it i s unwise to deduce the effect o f unions on profitability by looking at the wage markup alone. This i s because the union's ability to extract rent from a firm depends on the bargainingpower of the union and on the size o f the rent. The bargaining power and the size of the rent in turn depend on a mixture of factors, including the structure o f collective bargaining, the structure of the product market, the production technology used, and so on. In addition, by improving morale and job satisfaction among workers and by facilitating worker-employer cooperation, unions can contribute positively 116 to profitability. Therefore, instead o f trying to capture a given rent, unions may help to create profits from which they can achieve wage gains. 62. A large number o f studies have estimated the impact of unions on profitability. These studies use a number o f different measures o f profitability such as price/cost margins, net (of wages) return to capital, Tobin's q (the market value o f the firm relative to the replacement costs o f the firm's assets), and subjective profitability judgments by management. Such studies estimate the union impact using industry, firm, or stock market data. The empirical evidence from the 1980s and early 1990s is clear: most studies find that financial performance as measured by the indicators mentioned above is better in nonunion than in union workplaces and firms. The impact tends to be larger in industries or firms that have some monopoly power in the product market. Some o f the evidence suggests that the unions' share o f monopoly profits may be as large as between 47 and 77 percent. While these figures may not be very representative, they do show that under specific circumstances unions are able to appropriate a substantial share o f monopoly profits. 63. While a few studies for the UK find that unions haves no impact on profitability, the generally accepted view is that unions have a significant negative impact on profitability in Britishmanufacturingfirms. In Japan unions were found to reduce the rate o f return on equity by 20 to 25 percent. The ratio o f profits to sales is reduced by about 40 percent. This is confirmed when profits are measured in comparison to labor's share o f income: Union presence is associated with a higher faction o f income going to workers. Likewise, the unionization o f firms in the Republic o f Korea has a negative, but statistically insignificant, impact on operating profits. 64. While all this evidence point in the same direction, the wind may be changing: some studies looking at the late 1990s in the United States and United Kingdom find that unions can have positive effects on profits. The most strikingstudy i s one which looks at 464 entrepreneurial firms in the United States at the time o f their initial offering in 1993 and follows their subsequent financial performance. The study finds that union presence is associated with better financial performance. It argues that this surprising result can be due to a number o f factors that have affected the impact o f unions on profitability in this cohort o f "new" firms coming into maturity in the mid 1990s. These include more intense product market competition, a lower degree of labor-management conflict, the adaptation o f high performance work systems, and a general fall inunion power. 65. These themes are also found in the most recent vintage o f British studies that does not find evidence o f a negative relationship between subjective performance measures and unionization using data from the latest British Workplace Employment Relations Survey. These results have been confirmed in a number o f other studies. On balance, the evidence therefore supports the notion that the negative influence o f unions on profits identified in the 1980s has diminishedinthe 1990s. 66. However, important differences still exist. First, multi-unionism (but not union recognition as such) is associated with lower (self-reported) firm profitability in the United Kingdom. A similar result is found in a study o f Australian workplaces. Second, it i s mainly those unionized firms that face little product market competition that are adversely affected by union presence. This suggests that unions mainly share supernormal profits rather than cut into normal profits. 117 67. In conclusion, it seems that unions do have a negative impact on firms' financial performance, but that product market competition and/or development o f better and more effective labor-manager practices can reduce and perhaps even eliminate this negative effect. Productivity Differentials 68. Unions can contribute positively to labor productivity by improving work morale, facilitating cooperation with management, reducing grievances (through their "collective voice" function), and so on. These participatory benefits can, however, be countered if management's ability to adjust to changing economic circumstances is reduced, for example, when unions impose restrictive practices (such as overstaffing or guaranteed overtime). Moreover, one would expect that the positive "collective voice" effects be more pronounced in countries like Japan and Germany where industrial relations are based more on cooperation through works councils and enterprise unions. Multi-unionism and other adversarial industrial relations practices o f the type that used to be more common in the United Kingdom and Australia can result innegative effects. Likewise, one would expect that more intense product market competition induces unions and management to move towards industrial relations systems which would enhance the positive effects o f unions and reduce the negative ones. Thus, it is not clear from theoretical considerations how unions will affect productivity o f firms, and it would seem that the nature o f industrial relations matters a lot. 69. The uniodnonunion productivity differential is typically estimated econometrically from a production function model. Productivity, defined either as labor productivity or total factor productivity, i s explained by the input mix (employment, capital, and hours worked), a vector o f observed firm and industry characteristics (for example, industry concentration), a union dummy variable, and other control variables (such as business cycle indicators or the level o f union coverage in the industry). The production function approach is problematic for a number o f reasons. First, measured productivity in unionized firms can be higher than in non-unionized firms without implying that unionized firms are more efficient. This is because the wage markup, other things being equal, reduces employment inunionizedfirms. As a consequence, the marginal product o f labor would be higher in unionized firms than in non-unionized ones. Second, unionized firms are likely to change their input mix in response to the wage markup. Hence, the inputmix cannot be considered an exogenous determinant o f productivity, and a simultaneity bias can develop. A third problem arises because management's role is largely ignored. Since the interaction between management and unions affects productivity levels, lack o f knowledge o f what management does can give a biased view o f the impact o f unions. This problem is more generally related to unobserved heterogeneity and can best be dealt with by estimating productivity growth models instead o f productivity level models. An alternative to the production function approach is to use subjective measures o f productivity. This is done in a number o f recent studies from the United Kingdom, Japan, and Australia where management was asked to assess their establishment's labor productivity performance relative to other establishments in the same industry. The answers range from "a lot below average", "below average", "about average", "better than average", and "a lot better than average." It is clear that this approach is also problematic and one can question ifmanagers have the required information to make reliable estimates o f relative performance. 70. With these methodological issues in mind, we now review studies that estimate the productivity level and growth differential. We start with the evidence related to the productivity level differential. 118 71. The productivity level differential. Evidence on union productivity level differentials derives mainly from United States, United Kingdom, Germany, Japan, Australia, and Canada. The pattern o f results is not clear-cut and the average differentials hides a lot o f important variation. 72. For the US the conclusion is a qualified one, as there is considerable variation across studies. For example, inthose industries in which firms are subject to substantial product market competition, unionized firms tend to have higher productivity levels than non-unionized ones. The quality o f industrial relations is also important. The "quality" o f industrial relations can be proxied by the number o f grievances filed, the number o f unresolved grievances, the number o f strikes and quits, and the use o f long-term collective agreements. Firms with high-quality industrial relations are associated with higher productivity levels and higher product quality than firms with low-quality industrial relations. On the other hand, the significantly higher absenteeism among union workers than among nonunion workers can have a negative impact on productivity; some studies find that absenteeism i s 30 percent higher among unionized workers than among non-unionized ones. There is some evidence that unionized establishments that have adopted industrial relations practices that promotejoint decision making coupled with incentive- based compensation have higher productivity than similar non-union plant. In contrast, those establishments that are unionized but maintain traditional labor management relations have lower productivity. This strongly suggests that unions have positive productivitv effects only when industrial relations are `kood." 73. The evidence from the United Kingdom is more mixed, although some patterns can be identified. Britishunions appear to have a negative impact on the level o f productivity. However, this conclusion is far from robust, and the average estimates hide a lot of variation. First, it is noteworthy that the results are also affected by how unionism i s measured. For example, those studies that use union density as an indicator of unionism find a negative productivity effect. However, studies that use strikes as an indicator tend to find positive or insignificant effects. This suggests that the adverse impact o f unions on the productivity level is not due to industrial conflict. 74. Second, the productivity effect o f unions varies over time. British unions had no impact on productivity levels before 1979, but in the Thatcher era in the early 1980s, unions appear to have had a negative impact on productivity. More recent studies usingthe subjective productivity measure from the 1998 British Workplace Employee Relations Survey show that by the end o f the 1990s, unionism per se is no longer associated with poor productivity performance. The negative impact persists only in those establishments with multi-unionism and separate bargaining, i.e., the practice where many different unions compete to organize employees in the same establishment and bargain separately with the management. Establishments with either a single union or multiple unions that bargain together are as likely as those without any unions to report that productivity is better than average. The state o f industrial relations is just one among many "environment" variables that affect the relationship between unionism and productivity. 75. Also, as mentioned earlier, product market competition is an important factor. In particular, unionization i s found to have a small negative impact on productivity on average, but this effect is driven by the non-competitive sectors of the economy. Importantly, the relevant estimates suggest that, when there are only few competitors in the product market, the probability o f reporting being above the industry average i s 14% lower for a unionized workplace than for one without unions, but when there are more than six competitors the difference is insignificant. 119 76. InJapan, unions are enterprise basedand concentrated in larger firms, and the attitude of Japanese unions is often viewed as cooperative with management. Hence, Japanese unions seem like an obvious place to look for the "collective voice" effect o f unions. Nevertheless, empirical studies from Japan find that unions have mixed effects on productivity. Some note that unions had a positive impact on productivity levels in the 1970s when technology and labor-quality variables are held constant, while others find that productivity in unionized firms was 15 percent lower than in similar non-unionized firms. The latter finding is confirmed when subjective productivity measures are used that ask managers to rank the productivity performance o f their firm relative to other firms in the same industry. However, the presence of full-time union officials in the workplace is also found to have a positive impact o f productivity. This suggests that such officials can assist management with implementation o f procedures that enhance workplace efficiency. Other studies provide further evidence on the impact o f unions in Japan based on objective productivity measures (value added per worker) in a sample o f more than 400 listedmanufacturingfirms in the 1990s: While unions have a ceteris paribus negative impact on productivity, it i s interesting to notice that this effect is mitigated by the fact that workers in unionized firms have longer tenure. By reducing job separation, unions appear to encourage cooperative behavior that raises employees' work incentive and skill formation. Overall, the evidence probably suggests that Japanese unions have an indirect positive impact productivity. 77. In Germany, unions appear to have a negative, but quantitatively small impact on productivity. This may be related to the fact that German workplaces have works councils that provide the collective voice function o f unions, even in non-unionized firms. The evidence on the impact o f works councils suggests that such councils have a positive impact on productivity but only for larger firms. A similar result is found in the Republic o f Korea where unions seem to have no impact on labor productivity in manufacturing firms, while the presence o f mandatory works councils has a positive impact. In Australia and Canada the limited evidence available has found a negative effect o f unions on productivity. 78. Malaysia i s the only middle-income economy for which evidence on productivity (level) differentials is available. One study uses the value o f total sales relative to the total work force to proxy productivity. It finds that unionized firms have higher productivity levels than non- unionized firms and that the positive productivity differential i s primarily associated with industrial rather than company unions. While the study argues that this i s prima facie evidence that unions have been associated with dynamic efficiency effects in Malaysia, it i s somewhat puzzling why the strongest productivity effects are associated with industrial unions. Although industrial unions typically have shop-level facilities, one would expect that company-based unions would bejust as good and perhaps in even a better position than industry-based unions to provide a "voice" and other efficiency-enhancing services 79. The productivity growth differential. The U.S. evidence on the uniodnonunion productivity growth differential from the 1980s and the early 1990s suggests that while unions decrease productivity growth in some cases there i s no significant difference between unionized and non-unionized firms in others. And while the available evidence indicates that Britishunions may have had a negative impact on productivity level in the 1980s, the evidence regarding productivity growth is mixed. Some studies, for example, suggest that unionized firms have higher productivity growth than non-unionized firms duringthe period the 1970sand early 1980s. Then, many firms derecognized unions and repudiated closed shop arrangements, and differences between productivity growth in unionized and non-unionized firms disappeared between 1984 and 1987. Firms that experienced a change in union arrangements in the late 1980s had higher productivity growth than both unionized firms with constant union arrangements and non- unionized firms. These results indicate that the weakening o f British unions i s one factor that 120 explains the high productivity growth in the 1980s in the United Kingdom. In a country comparative context a study o f 19 OECD countries for the period 1950-80 concluded that union densitydoes not have a statistically significant impact on productivity growth. 80. In evaluating these effects it should be kept in mind that it is highly unlikely that permanent productivity differences between union and nonunion workplaces can exist. If they did, it would imply that the gap between the two types o f workplaces would be ever expanding. Thus, the productivity differentials observed in particular for the 1980s can best be thought of as short term effects that cannot be readily extrapolated into the far future. Unions and Implementation of New Technology 8 1. Unions' attitude toward new technology (for example, computers and new machinery) is unclear. On the one hand, unions may resist technological changes because they fear immediate short-run employment losses. On the other hand, they may take a long-run view and welcome new technology that increases productivity and the prospect for future increases inwages. 82. The available empirical evidence suggests that new technology is adopted as fast in unionized firms as in non-unionized ones and that unions have little impact on technological innovation in firms. Studies o f unions and technological change in the United States, the United Kingdom, and Canada in the 1980s indicate that unions "have no effect on firms' use o f advanced manufacturing and microelectronic technologies" and that "in most cases unions welcome technological modernization; sometimes encouraging it, most often accepting it, infrequently opposing it but usually seeking to protect their members". A similar conclusion on the impact o f unions on technological change is reached for Canada though unions do have an impact on the way in which technological change i s im~lernented.~~ In particular, unionized firms are more likely to introduce technological changes than non-unionized firms for cost-cutting or production control reasons. Inthe UK unions are found to have a small positive impact on the introduction o f new microelectronic equipment inU.K.firms inthe mid-1980s. Unions, Physical Investments, and Research andDevelopment 83. The reviewed evidence on the wage markup and the effect o f unions on profitability shows that unions share rents with firms. Besides the static impact on the functional distribution o f income, this can have significant dynamic efficiency effects. These arise when firms realize that workers are going to appropriate part o f the profits associated with investments in physical capital and research and development (R&D). Consequently, a unionized firm can be expected to invest less than a similar firm operating in a competitive labor market because o f the resulting "hold-up problem". 84. This "rent-seeking" view o f unions is in contrast to the more traditional view that holds that firms will substitute away from labor toward capital in response to an increase inthe relative cost o f labor. This will increase investment and so. It i s not clear apriori how unions will effect investments in physical capital and R&D. The empirical evidence does however suggest that unions tend to reduce investment. 85. A number o f studies have looked into the issue o f under-investment by unionized firms, usingfirm- or industry-level data from the United States and the United Kingdom. They find that unionization has a negative impact on investment in physical capital. For example, in the UK, 85Betcherman, 1991. 121 holding wages and productivity constant, the rate o f investment in firms that recognize a union and have an average union density is, on average, 23 percent lower than in other firms. However, the impact i s reduced as union density is increased. In the US unionization is found to reduce investment in physical capital by about 20% in a typical firm (inthe 1970s). A more recent study from Canada finds similar results. In Germany, on the other hand, the impact o f unions on investments has been found to be negative but somewhat smaller than in North America and Britain. 86. Likewise, the available evidence suggests that unionization can reduce spending on R&D: firm-level innovations in unionizedBritishmanufacturing firms are associated with higher wages for up to seven years. This suggests that unions do share in the surplus from innovation and may explain why the spending on R&D is lower in unionized firms than in non-unionized ones. Overall, North American and German evidence suggests that unionization reduces investment by around one fifth compared to the investment rate in a non-union workplace. Importantly, inboth Canada and the USA this effect is felt at even low levels o f unionization. The UK evidence is mixed: union recognition depresses investments but the adverse effect is off-set as density rises. IN constast, in Japan union recognition seem to go hand-in-hand with greater capital intensity. Unions and Training 87. Unions are likely to affect the amount and quality o f training that employees receive in the workplace through a number o f channels. First, unions might bargain over these issues with employers and demand that training takes place. Second, the fact that unions decrease turnover can have a positive impact on the amount o f firm-specific human capital workers are willing to invest in. The empirical evidence from the United States on the relationship between training and unionism is mixed. Some studies find that the amount o f work-related and on-the-job training that workers receive in unionized firms i s higher in non-unionized ones. Others cannot find any differences with regard to specific program such as computer literacy, numeracy, and sales training. 88. In the United Kingdom, the evidence is rather clear-cut: unionized worker receive more training and benefit more from participating in such programs than non-unionized workers (as measured by post-training wages relative to pre-training wages). These effects can partly off-set the negative impact on investments inphysicalcapital and R&D. Unions, Fringe Benefits, and Health and Safe0 Regulations 89. Unions do significantly increase wages. While this can be interpreted as evidence that unionized workers earn substantial rents, some have argued that as much as two-fifths o f the wage markup is compensation for an inflexible and employer-controlled work environment. In addition to their monthly paycheck, however, unionized workers may be concerned with other issues, such as bonuses, severance pay, health and safety regulations, and paid sick leave. 90. The evidence suggests that workers in unionized firms are more likely to receive these benefits than workers in non-unionized firms. Firms with unions are more likely to provide paid sick leave, retirement benefits, cheap loans, and transportation. Unions do increase the likelihood o f improved health and safety measures inthe United Kingdom. And in Japan unions increase the use o f severance pay and the size o f the yearly bonus. 122 91. Some o f these benefits obviously contribute to increasing labor costs in general and turnover costs in particular. On the other hand, cheap loans, free transportation, paid sick leave, and safety regulations may improve worker motivation and pay o f f in terms o f higher productivity. Moreover, to the extent that inadequate safety and health provisions generate a suboptimal allocation o f labor, union-sponsored (as well as government-sponsored) safety and health regulations may increase not only individual worker's welfare but also aggregate welfare. 92. Some studies have considered the issue o f disclosure o f information inthe context o f the risk attributes o f different jobs. Commodities produced in different sectors use production technologies that expose workers to different levels o f physical risks, such as exposure to toxins and industrial accidents. Safety is desirable from the point o f view o f workers. Therefore, safety has an opportunity cost, and safer jobs pay a lower wage. In an unregulated labor market, workers may be unable to appreciate the dangers inherent in different jobs. As a consequence, firms would not be required to compensate workers fully for the hidden risks involved in their jobs. This would lead to an inefficient allocation o f labor across sectors, with too many workers doingjobs that are too dangerous. A labor market reform that induces full disclosure o f safety levels would remove this distortion. An estimate for the Mexican economy o f the welfare effects of a labor market reform o f this type found that that the well-being o f workers would increase by 0.5 percent o f baseline GDP per year. Moreover, the real income o f the owners o f the firms would increase as well because the reform increases the demand for capital for risk-abatement purposes. The total gain is estimated to be 0.6 percent o f baseline GDP per year. This is a substantial gain and is o fthe same order o f magnitude as the estimated monopoly cost o f unions. Individual Performance Pay and Seniority 93. Individual performance pay is much more prevalent among non-unionized firms in the United States than among unionized firms by as much as 16 to 23 percent. Similar results are found in Britain where the most significant difference between unionized and non-unionized firms i s that individual performance is important in wage determination only in non-unionized firms. This can have an adverse effect on productivity if individual performance pay is usedas an incentive to increase workers' efforts. However, there are counterarguments. For example, the fact that unions are able to reduce the use o f individual performance pay could be seen as evidence that the presence o f unions reduces the need for this control instrument. Seniority-based wages can be interpreted as an efficiency wage that is designed to motivate workers to stay with the same firm for a longer period o f time. Therefore, unions can increase productivity by extendingseniority-based systems to smaller firms. Unions and Pensions 94. Evidence from many countries shows that unions increase the likelihood that workers are enrolled in pension schemes. If union-sponsored pension plans do not replace private saving, national saving can increase. At the macroeconomic level, the implied reduction in the real interest rate will increase investment demand and may even have a (temporary) impact on economic growth. 95. The macroeconomic impact o f collective bargaining is hard to disentangle from other determinants o f economic performance. While the available evidence from comparative studies o f 123 the OECD countries is fragile, two general features should be emphasized. First, the impact o f collective bargaining on various aspects o f macroeconomic performance depends on the economic, legal, and political environment in which collective bargaining takes place and can vary over time. Second, important complementarities exist between key aspects o f the bargaining system. Therefore, the impact o f individual aspects such as union density or centralization o f bargaining cannot be assessed in isolation. It is the package o f institutions and their capacity to propagate shocks that matters. 96. The impact o f collective bargaining on macroeconomic performance has usually been assessed through comparative studies where the performance o f countries with (very) different bargaining systems is systematically compared. Most studies look at the economic performance of the OECD countries during the period from 1960 till today and ask how the framework o f collective bargaining affects a large number o f macroeconomic performance indicators (such as unemployment and inflation) and labor market flexibility indicators (such as real wage flexibility) in an environment inwhich workers' rights can be taken as granted. It should be pointed out that this approach does not allow us to address the question o f causality. At best cross country studies o f this type can help us identify important correlations between measures o f key aspects o f labor market institutions and economic outcomes. 97. The importance o f collective bargaining as opposed to other ways o f organizing contracting inthe labor market can be measured by union density (the proportion o f workers who are union members) or bargaining coverage (the proportion o f the work force that is covered by a collective agreement). With respect to these indicators o f collective bargaining, the evidence suggests that: 1. Union density per se has a very weak association, or perhaps no association, with economic performance indicators such as the unemployment rate, inflation, the employment rate, real compensation growth, labor supply, adjustment speed to wage shocks, real wage flexibility, and labor and total factor productivity. There is, however, one significant exception: union density correlates negatively with labor earnings inequality and wage dispersion. 2. Bargaining coverage tends to be associated with higher real wage growth (with no impact on productivity growth), lower employment rates, higher unemployment rates, and higher inflation. As with union density, bargaining coverage correlates negatively with labor earnings inequality and wage dispersion. 98. Collective bargaining i s potentially a powerful means to facilitate bargaining coordination irrespective o f union density and bargaining coverage. Various types o f coordination are shown in Table 1 and each can affect differently wage setting and other aspects o f industrial relations (for example, working conditions, holidays and leave provisions and so on). Bargaining coordination i s potentially an influential determinant o f labor market and macroeconomic performance. For example, the Japanese system o f wage setting i s decentralized (firm-based) but coordinated inthe sense that it follows company rules based on seniority (hence, they are transparent) rather than individual contracting. In this system, workers are not paid wages equal to their individual reservation wage (that is, the wage level below which the worker will not supply his or her labor), as would have been the case under individual contracting, but this difference does not necessarily affect efficiency adversely. The Netherlands and Germany also have coordinated systems through strong employer organizations, coordination among giant companies or across industries, and coordination among unions. In France the government provides coordination in the form o f exercising control over critical sectors such as public services, utilities, and large nationalized industries. In Italy, there i s informal employer 124 coordination (via the big firms and regional employers' associations) and between some union confederations. Finally, Sweden has a centralized employers' organization as well as centralized union confederations. Table 1 Aspects of BargainingCoordination A. Union centralization IIThe capacity of the national union confederation to influence wage I levelsipatternsacross the economy. B.Unionconcentration I Union concentration is high if "few" unions at the relevant level of bargaining are representingworkers. C. Employer The capacity of the national employers' confederation to influence wage centralization levelsipatternsacross the economy. D.Level ofBargaining Collective bargainingtakes place at different levels: the firm level, the industry level, and the regionalhational level. E.Informalcoordination 1) Informal consultations at the industry, regional, or national level among unions and firms. 2) Pattern bargaining (an agreement in a dominant sector is mimicked by others). F. Corporatism A combination of 1) High union density and bargaining coverage and high degree of union and employer centralizatiodconcentration and 2) Social partnership betweennational workers' and employers' organizations and government. G.Other aspects This include different types of dispute resolution procedures, the proportion of unionized workers employed in sectors that are subject to international competition, and union density. 99. The comparative literature focuses on two hypotheses about the relationship between bargaining coordination and economic performance. The first hypothesis postulates that coordinated collective bargaining leads to better economic outcomes compared to semi- coordinated collective bargaining, which, in turn, performs better than uncoordinated collective bargaining. The second hypothesis (the "hump" hypothesis) postulates that semi-coordinated collective bargaining leads to worse economic outcomes than both coordinated and uncoordinated collective bargaining. 100. The evidence suggests that bargaining coordination did have a beneficial impact on macroeconomic performance in the 1970s and 1980s, but the evidence is fragile and inthe 1990s the impact seemed to have disappeared. More specifically, countries with highly coordinated collective bargaining tend to be associated with lower and less persistent unemployment, less earnings inequality and wage dispersion, and fewer and shorter strikes comparedto countries with semi-coordinated (for example, industry-level bargaining) or uncoordinated (for example, firm- level bargaining or individual contracting) collective bargaining. 101. In terms o f productivity growth and real wage flexibility, countries with highly coordinated collective bargaining tend to perform slightly better than countries with semi- coordinated collective bargaining but may not perform differently than countries with uncoordinated collective bargaining. This lends some support to the first hypothesis but only for the 1970s and 1980s.For most economic indicators, the differences disappear inthe 1990s.Two 125 exceptions are earnings inequality and wage dispersion. These indicators are comparatively low incountries with highlycoordinated collective bargainingthroughout the whole period. 102. Although countries with either uncoordinated or coordinated collective bargaining tend to be associated with lower and less persistent unemployment and higher productivity growth than semi-coordinated collective bargaining during the period 1960 to 1990, the evidence in favor o f the hump hypothesis is, ingeneral, very weak, particularlyfor the 1990s.Interms of inflationand the employment rate, there seems to be little difference between coordinated, semi-coordinated, and uncoordinated collective bargaining. 103. These conclusions refer to one dimension o f industrial relations and take other dimensions as given (either by controlling for them or by inappropriately ignoring them). This ignores the possibility o f complementarities between union density/bargaining coverage and bargaining coordination. Such complementarities are important for the impact o f collective bargaining on economic performance, and it can therefore be misleading to focus on one particular aspect in isolation. However, one can make some generalizations though with a wide margin o f error: 0 Highunion density andbargaining coverage do not contribute to poor unemployment performance so long as they are complemented by highbargaining coordination (particularly among employers). 0 Informal coordination o f wage bargaining (informal consultations between firms and/or unions or pattern bargaining) tends to mitigate the potential disadvantage (interms o f relative high unemployment) associated with semi-coordinated (such as industry-level) wage bargaining, and can arise in countries with relatively low union density and bargaining coverage. 0 Coordination among employers tends to be more important inproducing low unemployment than coordination among employees. This suggests that employers' organizations are more effective in controlling wage drift than union confederations. 0 Countries that have competing unions and many different union confederations (multi- unionism) tend to perform worse (interms o f unemployment and inflation) than other countries. 0 The effects o f coordination can be compromised or accentuated depending on the political orientation o f the government. "Good" economic outcomes (interms o f economic growth) may arise either when strong, centralized unions are pairedwith a strong left-wing government or when weak, decentralized unions are paired with a right- wing government. A mismatch (weak unions paired with a strong left-wing government or strong unions pairedwith a right-winggovernment) can lead to poor economic outcomes. 104. These complications are further illuminated in Table 2 where an relatively obvious proposition (to most economists) is shown to have potential a number o f negative externalities. The obvious proposition here i s that more decentralized wage setting is likely to reflect better the conditions, needs and wishes o f workers and employers in specific establishments. Notice that these are o f course potential externalities and not ones that have been shown by the empirical literature to be present inall countries or at all times. 126 Table 2 Six Externalities Associatedwith DecentralizedWage Setting. The inputprice externality Decentralized wage gains are passed on as higher product prices, thus increasing the real cost of inputs for other firms. The fiscal externality Decentralized wage gains lead to unemployment. The cost in terms o f unemployment benefits i s born by all tax-payers, not only those involved inwage setting. The unemployment Decentralized wage gains increase overall unemployment, externality making it more difficult for all unemployedworkers to find a new The envy externality Decentralized wage gains create envy among other workers. The consumer price Decentralized wage gains are passed on as higher product prices, externality thus loweringthe realwage of all workers. Efficiency wage externality At the decentralized level, firms have an incentive to try to increase the relative wage o f their workers to increase their motivation. 105. The evidence summarized above focuses on the relationship between aspects o f collective bargaining and specific economic outcomes and provides us with important information as to whether or not particular bargaining institutions are systematically associated with good economic outcomes. However, by essentially taking a static view o f the nexus between collective bargaining and economic performance perhaps one o f the most important benefits o f bargaining coordination i s ignored, namely the capacity o f the bargaining system to help the economy to adjust to shocks in an effective way. In this regard, the empirical evidence suggests that bargaining coordination can reduce the adverse impact o f economic shocks on unemployment levels. 106. This concludes the long list o f specific macroeconomic findings.Although some patterns emerge, the evidence is generally too weak and fragile to warrant grand generalizations about the performance o f specific labor market institutions. Instead, it suggests that the relationship between collective bargaining and economic performance cannot be fully understood unless the general economic and political environment in which bargaining takes place is taken into account. One should therefore be careful not to infer that institutional forms that work well in one environment would also work well in other-often very different-environments. Bargaining Coordination andMonetary Policy 107. The organization o f collective bargaining can affect economic outcomes in ways other than through the direct impact on wages and employment. An important indirect link is the interaction between the bargaining structure and monetary policy. The establishment o f the European Monetary Union and the move towards central bank independence in many OECD countries inthe last 10 years have spurred an interest inthis link. It i s clear that under a regime o f decentralized collective bargaining, unions are unlikely to take into account how wage settlements might affect monetary policy. However, when collective bargaining is coordinated, unions can act strategically and take into account how the central bank might react to different wage settlements. Broadly speaking, this can affect economic outcomes through two channels. 108. The first approach takes as its starting point the inflation bias in monetary policy when the central bank is tempted to print money to expand aggregate demand once the private sector 127 has locked itself into nominal wage contracts. If unions anticipate this, all attempts to reduce unemployment below the equilibrium level will fail and inflation will be sub-optimally high. Importantly, the higher the equilibrium level of unemployment is, the higher the inflation bias would be because this makes the temptation to create surprise inflation so much bigger. Now, under a regime o f coordinated bargaining unions realize that their wage strategy affects the level of equilibriumunemployment and thus the size o f the inflation bias. Ifunions care about inflation per se (independently of their concern for real wages and employment), they have an incentive to moderate their wage demands in order to reduce the equilibrium level o f unemployment and the inflation bias. Thus, coordinate bargaining may lead to lower unemployment and inflation. 109. The second approach focuses on the interaction between the degree o f bargaining coordination and the monetary policy regime. It i s argued that equilibrium unemployment is lower, the fewer the number o f unions and the more non-accommodating the monetary regime is. The logic is that unions might realize that their wage settlements will lead to a larger reduction in real demand when the central bank i s committed to a fixed nominal money supply than when it is anticipated that the central bank will accommodate (by increasing the money supply) whatever wage settlement i s reached in the labor market. Consequently, unions have an incentive to lower their wage demands when monetary policy is non-accommodating and this leads to highlevels of (equilibrium) employment. This mechanism i s not operating ifcollective bargaining i s completely uncoordinated because in that case the price effect is too small to be internalized by the unions. Nor does the monetary regime matter, if bargaining i s completely centralized or fully coordinated: in this case, the unions can coordinate their wage policies perfectly and choose full employment irrespectively o f the monetary regime. Thus, the argument is that semi-coordinated collective bargaining combined with non-accommodating monetary policy is likely to lead to higher levels o f employment than semi-coordinated bargaining with accommodating monetary policy. If true, this suggests that semi-coordinated collective bargaining i s not associated with poor economic outcomes, as otherwise suggestedby the reasoning behind the hump hypothesis. 128