62739 Employment Recovery in Europe and Central Asia Johannes Koettl, Isil Oral and Indhira Santos1 GDP contraction since World War II), in ECA the average Key Messages contraction that year was above 5%. With the exception of  Despite high unemployment in most Eastern Albania, Belarus, Poland, and Tajikistan, all countries in Europe and Central Asia (ECA)2 countries, ECA have experienced a contraction in GDP at some point people have not withdrawn from the labor since the second half of 2008 (Table 1). market but continue to actively look for jobs. Figure 1: Output is Recovering Everywhere (Albeit  Unemployment increased significantly in ECA Unevenly) countries during the crisis, particularly among 250% youth. However, young people are also the ones 444% Relative recovery in GDP until Q3 2010 benefiting most from the recovery. 200%  Labor market recovery remained sluggish up to 150% the third quarter of 2010. Many countries have seen only a slight recovery in unemployment 100% rates, although output is recovering everywhere. 50%  Up to the third quarter of 2010, the GDP 0% upturn in most ECA countries appeared to be driven by increases in productivity and hours worked; however, these are still below pre-crisis Note: Relative recovery refers to the increase in GDP experienced during the levels. This suggests that there is room in most recovery as a share of the decline in GDP experienced during the recession. countries for further increases in productivity Recession and recovery are country-specific. The data used refers to quarter-on- quarter growth rates and uses a common methodology comparable across and hours worked, which could delay the countries. National data, especially if not seasonally adjusted, and year-on-year data might yield different results. A recession is defined as at least two recovery in employment. consecutive quarters of real GDP decline. Source: Authors’ calculations based on data from European Bank of Reconstruction and Development (EBRD, 2010). Countries are Recovering from the Recession In fact, most ECA countries experienced a recession Eastern Europe and Central Asia (ECA) is arguably the (defined here as two or more consecutive quarters of real region most affected by the global financial crisis. While GDP declines). The Baltic countries experienced the global GDP shrank by about 2.2% in 2009 (first global largest shock to output (19-30% GDP decline). These countries also experienced the longest recessions (seven months compared to the ECA average of five months). In 1 contrast, the Western Balkan and Central Asian economies We are grateful to David Newhouse and Kaspar Richter for peer reviewing and to Borko Handjiski, Sanja Madzarevic-Sujste, and Evgenij Najdov, as well as to experienced the smallest GDP contractions. Jennifer Keller, Ulrich Zachau and other colleagues for additional comments. 2 The ECA region encompasses the transition countries of Eastern Europe, the Balkans, the Community of Independent States, and Turkey. This Knowledge The economic recovery is advancing but at an uneven pace. Brief, however, does not systematically cover all ECA countries because of data Up to the third quarter of 2010, all countries, except availability. The period covered is from the first quarter of 2008 (Q1 2008) to the Romania, had started to recover in terms of GDP. In most third quarter of 2010 (Q3 2010) and the data has been seasonally adjusted. ECA Knowledge Brief countries, recovery started in the second and third quarters recovery has been particularly strong in Kazakhstan, of 2009. In the Baltic countries, where the recession was Moldova and Turkey which have all fully recovered from the most severe, recovery started only in the first half of the recession (Figure 1). However, recovery has been weak 2010. Partly due to this timing issue, the extent to which in other countries--for example, Latvia and Lithuania have output has recovered varies across countries. Output recovered just over one-tenth of the initial drop in GDP. Table 1: Timing, Duration and Intensity of Recession and Recovery Length of the GDP Percentage Drop Start of Start of GDP Length of the Recession Recovery (Number of in GDP During Recession Recovery (Number of Quarters) Quarters) up to third Recession ECA Countries quarter of 2010 CIS Armenia ARM Q4 2008 Q1 2010 30% 5 3 Belarus BLR No recession Kazakhstan KAZ Q3 2008 Q2 2009 4% 3 6 Moldova MDA Q3 2008 Q3 2009 10% 4 5 Russian Fed. RUS Q3 2008 Q3 2009 12% 4 5 Tajikistan TJK No recession Ukraine UKR Q2 2008 Q2 2009 24% 4 6 Western Balkans and Turkey Albania ALB No recession Macedonia MKD No recession Serbia SRB Q2 2008 Q2 2009 5% 4 6 Turkey TUR Q2 2008 Q2 2009 15% 4 6 EU New Member States and Croatia Bulgaria BGR Q4 2008 Q2 2010 10% 6 2 Croatia HRV Q3 2008 Q2 2010 11% 3 2 Czech Republic CZE Q4 2008 Q3 2009 5% 3 5 Estonia EST Q2 2008 Q4 2009 23% 6 4 Hungary HUN Q2 2008 Q1 2010 9% 7 3 Latvia LAT Q2 2008 Q1 2010 30% 7 3 Lithuania LTU Q2 2008 Q2 2010 19% 8 2 Poland POL No recession Romania ROM Q3 2008 12% 10 0 Slovakia SVK No recession Slovenia SVN Q4 2008 Q3 2009 11% 3 5 Note: Q = Quarter. The data used refers to quarter-on-quarter growth rates and uses a common methodology comparable across countries. National data, especially if not seasonally adjusted, and year-on-year data might yield different results. A recession is defined as at least two consecutive quarters of real GDP decline. Macedonia and Slovakia did see their GDP contract on some quarters during this period. Macedonia is a borderline case with GDP decline during Q4 2008 and Q1 2009, followed by three quarters of growth and another quarter of decline, but overall relative stagnant growth patterns. Source: EBRD. Despite High Unemployment, People Have Not not changing by more than 5% in any country between 2008 Withdrawn from Labor Markets and 2010 (Figure 2). For the most part, changes have been small compared to historical trends. This is significant Labor force participation or activity rates (refers to those because it means that people have not given up looking for who have jobs or are actively looking for one) stayed fairly jobs but continue to actively search for employment. In constant during the recession as well as the recovery period, several countries, however, changes in activity rates do ECA Knowledge Brief represent important departures from historical trends. In crisis).3 In other countries, the trends in activity rates did not Turkey and Lithuania activity rates fell between 2000 and differ by gender or were driven by changes in male activity 2007 (by 9% and 6%, respectively); yet, during the crisis rates, like in Latvia.4 (2008-2010) activity rates actually increased (by 4.2% and 3.3%, respectively). At the other extreme are Latvia and The Magnitude of the Employment Shock Varied Bulgaria where activity rates fell by 4% during the crisis, Across ECA while from 2000-2007 they increased (by 8% and 0.4%, respectively). ECA countries experienced large shocks in their labor markets. Labor markets were, in fact, the main channels Figure 2: Activity Rates in ECA Show Limited Fluctuation through which the economic crisis was transmitted to (2008-2010) households.5 While activity rates – includes the employed and the unemployed - have remained stable, employment fell 5% significantly in most countries during the recession. This was Change in activity rates between Q1 4% 3% especially the case in the Baltic countries, with employment 2008 and Q3 2010 (%) 2% falling by more than 10% during this period in Lithuania and 1% Estonia and 18.3% in Latvia (Figure 4). 0% -1% -2% Figure 4: The Employment Shock Was Large During the -3% Recession, Especially in Baltic Countries -4% -5% Change in employment during 0% country-specific recession (%) Source: Authors’ calculations based on data from ILO (2011) and UN (2009). -4% Figure 3: In Some Countries, the Past Three Years Have Seen Important Gender Differences in Activity Rate Trends -8% -12% 15% Change in activity rates between Q1 2008 and Male Female -16% 10% -20% Source: Authors’ calculations based on ILO (2011). 5% Q3 2010 (%) Only Kazakhstan and Turkey have seen increases in 0% employment over the past three years. In Turkey, the increase in employment is mainly driven by increases in the -5% number of women employed (16.8%), following increases in labor force participation. This suggests an important added- -10% worker effect in Turkey where the recession led to more TUR LTU HUN RUS EST ROM LVA CZE SVN BGR MDA women taking up employment to supplement household Source: Authors’ calculations based on ILO (2011) and UN (2009). income, often as unpaid family workers in agriculture. In a few countries, changes in activity rates differed across With activity rates fairly constant and employment numbers gender. In Turkey and Lithuania, for instance, higher activity depressed, unemployment rates spiked almost everywhere rates are being driven by an increase in the share of women between early 2008 and the third quarter of 2010, increasing joining the labor force, possibly to supplement household even in countries that did not experience a recession: income in response to the crisis (14.2% in Turkey and 6.6% in Lithuania) - see Figure 3. The extent to which these changes are permanent is not clear yet, and little is known 3 International Labour Organization (2010), “Crisis and Turkey: Impact Analysis of about the nature of these increases in participation (for Crisis Response Measures,” ILO Office for Turkey – Ankara: ILO, 2010-12-20. 4 In Latvia, changes could have been driven by sectoral adjustments, like, for example, women could have been increasingly employed as example, through large layoffs of male workers in the construction sector. unpaid family workers in subsistence farming during the 5 World Bank (2011), “The Job Crisis: Household and Government Responses to the Great Recession in Eastern Europe and Central Asia”, Washington, DC. ECA Knowledge Brief Albania (7%), Poland (31%) and Slovakia (44%).6 Among the output shock, the effects on labor markets were relatively countries that experienced a recession, only Kazakhstan did small (above the regression line) - for instance, in Ukraine, not see an increase in unemployment rates. However, large Turkey and Romania. In other countries - Bulgaria, variations were observed across countries in terms of the size Moldova, Latvia and Lithuania - the effects on labor markets of the unemployment shock, measured as the percent were larger. Policies may explain part of this variation in increase in the unemployment rate from early 2008 (first employment elasticity across countries.7 quarter) to its peak (Figure 5). The unemployment shock was especially large in the Baltic countries where the output The governments in the region have been active in trying to shock was also the largest. limit the effects of the crisis on labor markets through a wide range of employment-related programs. Examples include Figure 5: The Magnitude of the Unemployment Shock (Until wage subsidy programs in the Czech Republic, Estonia, Its Peak) Varied Significantly across Countries Hungary, Latvia, Slovakia, and Ukraine, and support for short-time work in Bulgaria, Czech Republic, Hungary, 400% Poland, Romania, and Slovenia.8 Change in unemployment rates between Q1 2008 and peak (%) 350% 300% Figure 7: Unemployment Rates Remain High (Third Quarter, 250% 2010) 200% 150% 50 100% 50% 40 0% Unemployment rate (%), Q3 2010 HRV RUS LTU LVA BIH* TUR CZE BGR MNE HUN MDA EST KAZ UKR ROM ARM SVN 30 Note: * denotes source of data are national unemployment registries and not LFS Source: Authors’ calculations based on data from ILO (2011). 20 Figure 6: There Was Variation across Countries on How Employment Reacted to a Given Output Shock during 10 Recession -30% -25% -20% -15% -10% -5% 0% 0 KAZ CZE UKR POL LTU MNE MKD EST TJK* HUN TUR ALB BIH* ROM ARM SRB RUS LVA SVN MDA BGR BLR* HRV SVK 0% ROM Change in employment during country- TUR SVN KAZ CZE UKR RUS HUN Note: * denotes source of data are national unemployment registries and not LFS. HRV For Montenegro and Tajikistan, data correspond to Second Quarter, 2010. specific recession (%) -5% Source: Authors’ calculations based on ILO (2011). MDA R² = 0.56 BGR However, despite these efforts, unemployment rates in ECA EST -10% LTU remain high (Figure 7). Unemployment rates are the highest in the Western Balkans, especially in Bosnia and -15% Herzegovina (43.3%) and Macedonia (32.1%). On the other LVA hand, (official) unemployment rates are the lowest in Belarus Change in GDP during country-specific recession (%) (0.4%), Tajikistan (4.4%), and Kazakhstan (5.8%). The -20% Source: Authors’ calculations based on EBRD (2010) and ILO (2011). Baltic countries, the most affected by the crisis, have now unemployment rates above the average for the region as a There was also variation on how employment reacted to a result of the shock. given output shock. Figure 6 shows how changes in employment related to changes in GDP during the recession. It indicates that countries that experienced the largest shocks 7 to output also experienced the largest relative declines in Aside from crisis-related policies, of course, the structure of labor markets (including labor regulation issues) and the sectoral composition of GDP and employment. However, in some countries, given the size of employment are important determinants of the output elasticity of employment. 8 Kuddo, A. (2010), “Employment Related Mitigation Measures in Europe and Central Asia Countries”, Europe and Central Asia Knowledge Brief, World Bank, 6 Washington, DC. In Albania, this means an increase in unemployment rates from 12.6 to 13.6%; in Poland, from 7.2 to 9.4%; and in Slovakia, from 9.9 to 14.3%. ECA Knowledge Brief Unemployment Is Particularly High Among Youth However, in the Baltic countries, Bulgaria, Croatia, Moldova, and Slovenia, employment continued to fall even The unemployment shock was very large for youth. The as output started to recover (Figure 9). With the exception of unemployment rate among the population aged 15-24 was Kazakhstan and Turkey, up to the third quarter of 2010, already 2.5 times higher than that for adults in ECA before those countries that have seen the largest improvements in the crisis, but the recession led to a further surge in GDP have actually seen the smallest improvements in unemployment among youth. Between 2007 and 2009, the employment (Figure 10). unemployment rate among youth more than tripled in Latvia and Lithuania, more than doubled in Estonia, and increased Figure 9: During the Recovery, Employment Has Continued to Fall in Many Countries by more than 50% in the Czech Republic. Contrast these figures with an average change in the youth unemployment 8% rate of 7.4% worldwide over the same period. Change in employment during GDP 6% recovery until Q3 2010 (%) Youth, in fact, was disproportionately affected by the crisis. The large increase in youth unemployment in ECA was not 4% only the result of an overall increase in unemployment; unemployment rates increased relatively more for youth than 2% for the rest of the population in ECA (Figure 8) and these 0% differences were especially dramatic in the Czech Republic and Slovakia. The younger segments of the work force often -2% have more flexible labor contracts and less experience, making it likely that they are the first ones to lose their jobs -4% during economic downturns. However, this flexibility can also mean that they are the first ones to be rehired as the economy recovers. Source: Authors’ calculations based on ILO (2011). Figure 8: Young People (Aged 15-24) Were Disproportionately Figure 10: So Far, Countries with the Largest Increases in Affected by the Crisis GDP during Recovery Have Seen the Smallest Improvements in Employment 0.6 8% Absolute change in the ratio of youth unemployment rate, 2007-2009 0.4 Percentage change in employment during 6% unemployment to overall TUR 0.2 4% recovery until Q3 2010 KAZ 0.0 2% RUS HUN LVA -0.2 0% CZE -0.4 -2% SVN UKR LTU HRV BGR -0.6 -4% MDA EST -6% 0% 4% 8% 12% 16% 20% Note: Kazakhstan, Macedonia, and Serbia – not depicted in the graph - are three Percentage change in GDP during recovery until Q3 2010 countries where the ratio of youth to adult unemployment also worsened during this period. Note: Time periods differ from country to country, depending on when the recovery Source: Authors’ calculations based on ILO (2011) and UN (2009). started. Source: Authors’ calculations based on EBRD (2011) and ILO (2011). Similarly, up to the third quarter of 2010, unemployment Up to the Third Quarter of 2010, Employment Had rates were yet to recover in the same way as output. This is Started to Recover, But Only in Some Countries the case even for countries that had fully recovered from the output shock (Moldova, Turkey). Overall, Turkey, Ukraine Employment growth has been particularly strong in Turkey and Russia are the three countries that had been able to and Kazakhstan during the recovery (Figure 9) - the only reduce unemployment rates the most relative to the initial two countries that have seen net employment gains overall in shock experienced. But even these countries had only the past three years. recovered less than 60% of the initial increase in ECA Knowledge Brief unemployment rates by late 2010. Many countries had seen had started to recover by the third quarter of 2010, it had no recovery at all or only a slight recovery in unemployment taken, on average, an additional two quarters for rates. This is the case for Armenia, Bulgaria, Croatia and unemployment rates to fall after GDP had started to rise.9 Lithuania. Others have experienced reductions in unemployment rates of less than 10% of the increases Figure 12: In the Recovery, Youth Unemployment Rates Are experienced during the crisis: Serbia (1%), Hungary (3%), Falling Faster Than for the Rest of the Workforce Slovenia (4%), Romania (9%) and Slovakia (9%). Interestingly, only Ukraine, Latvia, and Romania had labor 0.1 unemployment to overall unemployment, markets that experienced a stronger recovery than that seen -1E-15 Absolute change in the ratio of youth in output in the period covered in this Brief (above the 45 -0.1 degree line in Figure 11). This is consistent with Q3 2009-Q3 2010 employment trends, which suggest that up to the third -0.2 quarter of 2010, the economic recovery had not been -0.3 accompanied by a recovery on the extensive margin - that is, -0.4 in employment levels of firms. -0.5 Figure 11: Recovery in Unemployment Rates Has Been Slower -0.6 Than That in Output 140% Source: Authors’ calculations based on ILO (2011). 120% How much have unemployemnt rates 45° line Figure 13: There Is A Time Lag between the Start of GDP recovered as of Q3 2010? 100% Recovery and Unemployment Rate Recovery 80% 5 unemployment recovery after start of 4 Number of quarters of delay of 60% GDP recovery as of Q3 2010 UKR TUR 3 RUS 40% 2 1 20% EST CZE ROM LVA 0 MDA LTU SVN HUN ARM BGR SRB -1 0% HRV -2 0% 20% 40% 60% 80% 100% 120% 140% -3 How much has GDP recovered as of Q3 2010? -4 Note: GDP recovery is measured from its trough, unemployment rate recovery is -5 measured from its peak between Q1 2008 and Q3 2010. CZE MDA BGR* HUN KAZ UKR TUR HRV* ROM RUS ARM LTU SVN LVA SRB EST Source: Authors’ calculations based on EBRD (2011) and ILO (2011). Some of the hardest hit groups, like youth, are also the ones Note: * denotes that unemployment rates had not started to decline by Q3 2010. benefiting the most from the recovery. While young people Source: Authors’ calculations based on EBRD (2011) and ILO (2011). saw the largest relative increases in unemployment during the crisis, they had also experienced the strongest recovery up to the third quarter of 2010. It is only in Turkey, Although it might be too early to tell, this time lag seems Romania, the Czech Republic, and Slovakia that recovery shorter than that observed historically around the world. has been relatively weaker among youth in terms of declines Evidence from previous recessions shows that employment in unemployment rates (Figure 12). (and unemployment) take quite a bit longer to recover than GDP. Using a sample of countries from around the world Recovery, in terms of employment and unemployment, takes and more than three decades of data, the IMF estimates that, time. The nature of the recent recession - economic across all types of recessions, it typically takes three quarters slowdown combined with a financial crisis, high level of after output has started to recover for employment to financial stress for individual households and governments, similarly grow and five quarters for unemployment rates to and high uncertainty - means that the private sector may take some time to rehire workers even as production starts to pick up. In fact, in ECA, unemployment rates peaked in most countries several quarters after the GDP started to recover 9 (Figure 13). For those countries where unemployment rates This average does not include Romania, where the unemployment rate peaked before GDP started to recover. ECA Knowledge Brief peak.10 Lags are longer in the case of financial crises, Figure 15: So Far, Recovery Is Driven by Increases in Average suggesting that in those countries where the nature of the Output per Worker Rather Than by Hiring of New Workers crisis had an important finance component–like in the Baltics - the recovery may be slower (Figure 14). 120 Average output per worker (index) Employment usually lags behind output because firms wait 115 to see if a recovery is permanent before committing to hiring 110 new people. This could especially be the case in this 105 recession, where the number of hours worked fell significantly in many countries, which suggests that there is 100 potential to adjust on the intensive margin before hiring new 95 workers. 90 HUN LVA LTU RUS HRV CZE SVN UKR KAZ BGR TUR EST MDA Figure 14: How Long Does It Usually Take for Labor Markets Start of recession Q3 2010 Quarter of GDP trough=100 to Recover? Source: Authors’ calculations based on ILO (2011) and UN (2009). Unemployment This also indicates that during the economic recovery, firms may be adjusting on the intensive margin - that is, by increasing the number of hours worked per employee rather Employment than by hiring new workers. This would be a mirror strategy of the one largely followed during the recession where many 0 1 2 3 4 5 6 firms cut the number of hours worked before instead of Median number of quarters before employment (unemployment) laying off workers. In fact, by the third quarter of 2010, the reaches its trough (peak) after the end of the recession) number of hours worked had increased in Estonia, Hungary, Recessions associated with financial crises All recessions Latvia, Lithuania, Macedonia and Slovenia (Figure 16). Source: IMF (2010). Again, as with output per worker, all countries are still well below the number of hours worked before the recession, signaling that there is still significant adjustment to be done Up to the Third Quarter of 2010, Recovery Had on the intensive margin. Been Driven by Productivity Figure 16: In Some Countries, Increases in Output per Worker The evidence indicates that output recovery has been driven Are Driven by Increases in Hours Worked by improvements in labor productivity rather than increases in employment. In particular, GDP recovery has been 102 accompanied by increases in average output per worker. 100 Average hours worked (index) 98 Figure 15 shows that after falls in average output per worker 96 across the board during the recession, average labor 94 productivity in all countries was higher in Q3 2010 than at 92 90 the end of the recession (when GDP hit its lowest level in 88 each country). Average labor productivity increased by 86 almost 15% in Moldova and by more than 8% in Estonia and 84 82 Turkey. This indicates a “V-shaped” behavior in output per TUR CZE LTU MKD EST SVN BGR HUN SVK LVA POL HRV ROM worker. In a number of countries, output per worker still has Q3 2009 Q3 2010 Q3 2008=100 significant room to recover compared to its level at the start of the recession. Source: Authors’ calculations based on Eurostat (2011). Together with productivity, wages are rising in most countries. In countries where average output per worker is increasing but hours and employment are not, actual labor 10 International Monetary Fund (2010), “World Economic Outlook (WEO): productivity (output per hour) should be rising and this Rebalancing Growth,” April 2010, Washington, DC. should be reflected in higher wages. Figure 17 shows that by the third quarter of 2010, wages have started to pick up in all ECA Knowledge Brief countries11, except in Croatia and the Baltic countries. For Countries like Turkey, where the recession ended early on, the latter, the decreases are the result of an explicit policy of have had more time to recover and for labor market internal devaluation. Timing, duration, and intensity of the indicators to improve. What we see instead for most recession and recovery were also different in these countries, countries are adjustments on the intensive side. There is a with Croatia, Latvia, and Lithuania still in recession in the recovery in terms of average labor productivity (measured as first quarter of 2010 and Estonia in recession until the end of output per worker), in some countries it is driven by an 2009. increase in hours worked (a reversal of the reduction in hours worked that took place during the recession), and in Figure 17: Wages Are Increasing in Most Countries others by actual increases in productivity. The question that still remains, however, is: How long will the recovery of 140 employment and unemployment take and has the crisis had some structural effects that may delay the recovery more Average wage per worker (index) 120 100 than usual? 80 60 References 40 Eurostat (2011), 20 (http://epp.eurostat.ec.europa.eu/portal/page/portal/eurostat/home/). 0 European Bank for Reconstruction and Development (2010), “Growth in Real LTU LVA EST HRV SVN MKD MDA UKR KAZ ROM BGR BLR KYR GDP” EBRD: October, 2010. Q1 2009 Q1 2010 Q1 2008=100 International Labour Organization (2010), “Crisis and Turkey: Impact Analysis of Crisis Response Measures,” ILO Office for Turkey – Ankara: ILO, 2010-12-20. Note: Includes agricultural sector wages for all countries except for Latvia, Romania and Ukraine International Labour Organization (2011), “ILO Department of Statistics - Short Source: Authors’ calculations based on ILO (2011). Term Indicators of the Labour Market” (http://laborsta.ilo.org/sti/sti_E.html). International Monetary Fund (2010), “World Economic Outlook (WEO): In Conclusion, Are Labor Markets Recovering? Rebalancing Growth,” April 2010, Washington, DC. Kuddo, A. (2010), “Employment Related Mitigation Measures in Europe and Countries in ECA have started to recover from the output Central Asia Countries”, Europe and Central Asia Knowledge Brief, World Bank, shock associated with the financial and economic crises; Washington, DC. however, the recovery in labor markets on the extensive United Nations (2009), “World Population Prospects: The 2008 Revision,” New margin - in terms of increased employment and decreased York. unemployment - has been sluggish, at least up to the third World Bank (2011), “The Job Crisis: Household and Government Responses to the quarter of 2010. While people have remained in the labor Great Recession in Eastern Europe and Central Asia”, Washington, DC. force, employment is still lower and unemployment is still higher than before the recession. In part, this lag is not unusual as recovery in employment and unemployment lags About the Authors output recovery by around a year. This Knowledge Brief was prepared by Johannes Koettl, Economist, Isil Oral, Junior Professional Associate, and Indhira Santos, Young Professional of the Human Development Sector Unit, Europe and Central Asia Region of the World Bank. Correspondence email is isantos@worldbank.org. ________________________ 11 In some of these countries, the increase is partly due to increases in public sector wages. “ECA Knowledge Brief” is a regular series of notes highlighting recent analyses, good practices and lessons learned from the development work program of the World Bank’s Europe and Central Asia Region http://www.worldbank.org/eca