Job Dynamics in Albania A note profiling Albania’s labor market May 2018 Contents Acknowledgments .................................................................................................................................. iv Abbreviations and Acronyms ................................................................................................................... v I. Introduction and context ................................................................................................................. 1 II. Job dynamics from the labor demand side ....................................................................................... 4 The 2015 firm registry: A snapshot of the employment stock............................................................... 4 Job dynamics in Albania 2002–2015: Trends and flows ........................................................................ 6 Job creation is positive, though concentrated in a few firms and sectors .......................................... 7 Some firm types perform better than others................................................................................... 14 Gross job flows – adding up and netting out .................................................................................. 18 Section II summary ............................................................................................................................ 22 III. Labor supply: who works, in what kind of jobs, and who lacks a job? ......................................... 23 A snapshot of the Albanian labor force .............................................................................................. 23 Labor productivity is low and may continue to decline with changes in demographics ....................... 25 Jobs are not inclusive– especially for women, youth, the poor, and those with little education ......... 26 Job quality is a concern ...................................................................................................................... 29 Section III summary ........................................................................................................................... 32 IV. Conclusions................................................................................................................................ 33 Demand-side recommended policy actions........................................................................................ 33 Supply-side recommended policy actions .......................................................................................... 34 Annex I: Structural Business Statistics Data and Definitions ................................................................... 36 Annex II: Results from Firm-Level Regression Analysis ........................................................................... 40 References ............................................................................................................................................ 44 ii Boxes Box 1: Gross job flows glossary ................................................................................................................ 7 Box 2: The National Employment Service and Data Monitoring and Management ................................... 8 Figures Figure 1: Per capita income growth has recovered since 2014 ................................................................. 1 Figure 2: Services provide the highest contribution to economic growth ................................................. 2 Figure 3: Jobs distribution by firm characteristics (size, region, sector) .................................................... 5 Figure 4: Jobs distribution by firm characteristics (age, productivity) ....................................................... 6 Figure 5: Gross job flows, 2002–2015 ...................................................................................................... 8 Figure 6: Jobs trends by firm size ............................................................................................................. 9 Figure 7: Entry and exit rates of firms .................................................................................................... 10 Figure 8: Jobs trends by firm sector ....................................................................................................... 11 Figure 9: Jobs trends by region .............................................................................................................. 12 Figure 10: Jobs trends by firm age and productivity ............................................................................... 13 Figure 11: Distribution of firms by productivity level and size category .................................................. 15 Figure 12: Distribution of firms by productivity level and sector ............................................................ 16 Figure 13: Net job-creation rates by productivity quintile ...................................................................... 17 Figure 14: Where is churning highest (2013–2015)? .............................................................................. 20 Figure 15: Churning: Excess job reallocation rates, by age and productivity quintile............................... 21 Figure 16: Working-age population in 2016: A snapshot ........................................................................ 23 Figure 17: Albania’s employment rate is the highest in the Western Balkans ......................................... 24 Figure 18: Employment strengthened and unemployment fell post-2014 .............................................. 25 Figure 19: Labor productivity is low in Albania ....................................................................................... 25 Figure 20: Access to employment differs by age, gender, and education level ....................................... 26 Figure 21: Women, youth, and workers close to retirement are more likely to be out of a job ............... 27 Figure 22: Is there a “housewife” trap in Albania? ................................................................................. 28 Figure 23: Fewer good job opportunities for the poor............................................................................ 28 Figure 24: Unemployment spells last long.............................................................................................. 29 Figure 25: Type of employment in 2016 ................................................................................................. 30 Figure 26: Both women and men work mostly in low-productivity sectors ............................................. 30 Figure 27: Low-productivity jobs ............................................................................................................ 31 Figure 28: Most jobs are for adults and for low-medium-skilled workers ............................................... 32 Tables Table 1: Within-firm employment change since inception ..................................................................... 13 Table 2: “Gazelles” and their key characteristics in Albania .................................................................... 18 iii Acknowledgments This note was prepared by Maddalena Honorati and Sara Johansson de Silva with inputs from Olga Kupets and Sara Berger. The team is grateful to the Government of Albania for sharing feedback and views on findings at an early stage of the study. In particular, the team is extremely grateful to Deputy Minister of Finance and Economy Dajna Sorensen for her insights and guidance throughout the process. The team is indebted to Delina Ibrahimaj, Head of INSTAT, for the openness to access the Structural Business Survey data. The team would also like to thank representatives of the National Employment Service, the National Vocational Education and Training Qualification agency, RisiAlbania, Swiss Contact, the European Union, the United Nations Development Programme, and GIZ (Deutsche Gesellschaft für Internationale Zusammenarbeit) and all participants in the workshop discussing the preliminary findings of the analysis (held in Tirana on November 6, 2017) for their valuable suggestions and comments. The team wishes to acknowledge the leadership of Cem Mete (Practice Manager), Maryam Salim (Country Manager-Albania), and Timothy Johnson (Program Leader). The team is grateful to Victoria Strokova, who kindly peer-reviewed the study. The team is grateful for the administrative support provided by Dung Ngoc Tran in Washington, D.C., and Elda Hafizi in Albania. This study benefited from the excellent editorial work of Amy Gautam. iv Abbreviations and Acronyms ERJ Excess job reallocation EU European Union GDP Gross domestic product GJC Gross job creation GJD Gross job destruction GJR Gross job reallocation ICT Information and communications technology INSTAT Institute of Statistics LFS Labor Force Survey NES National Employment Service NESS National Employment and Skills Strategy NJC Net job creation OECD Organization for Economic Co-operation and Development SBS Structural Business Survey v I. Introduction and context 1. Since the collapse of isolationist communism in the early 1990s, Albania has taken steps toward a functioning market economy with the aim of promoting sustained economic growth . In the early years following the collapse, Albania experienced rapid economic growth, averaging 10 percent per year from 1993–1995, and inflation fell to single digits. Governance shortcomings, most notably the pyramid schemes, jeopardized these gains from 1995–1997; however, the economy bounced back with rapid growth of nearly 6 percent per year, rising into the ranks of middle-income countries by 2008 (World Bank 2015). The rapid pace of growth helped the country narrow the per capita income gap with the rest of Europe – from 18 percent of average European Union (EU) incomes in 1998 to 30 percent by 2012 – and fueled its aspirations to join the EU. Additionally, rapid growth successfully halved poverty from 25.2 percent in 2002 to 12.5 percent in 2008 (World Bank 2015). The global and Eurozone financial crisis of 2008, though, brought Albania’s growth to a near standstill. Between 2009 and 2012, gross domestic product (GDP) growth halved to less than 3 percent, and then declined to around 1.4 percent in 2013 (World Bank 2015). More recently, economic growth has been positive, albeit modest. After near stagnant growth, per capita income growth reached nearly 3 percent in 2015 and 2016 (Figure 1), thanks to higher domestic demand, private investment, and the recovery of EU trading partners (IMF 2017). Figure 1: Per capita income growth has recovered since 2014 GNI per capita growth (% per year) 10.0 8.0 6.0 4.0 2.0 0.0 -2.0 -4.0 -6.0 -8.0 -10.0 2008 2009 2010 2011 2012 2013 2014 2015 2016 Albania WB5 Balt3 Source: Estimates based on World Development Indicators. Note: WB5=Bosnia and Herzegovina, FYR Macedonia, Kosovo, Montenegro, and Serbia. Balt3 = Estonia, Latvia, and Lithuania. 2. Emerging political stability in the last decade was important for the support of policies aimed at accelerating growth, creating jobs, and furthering progress toward EU accession . The year 2005 witnessed the smooth transition of power between opposing political parties. This was followed more recently by free and fair elections in 2013, where a coalition government with a strong parliamentary majority took office with the aim of tackling fiscal consolidation and public financial management (World Bank 2015). As the country continues to consolidate its democratic systems and develop a more modern economy, Albanians are eager to continue down the path of accession of the EU (O’Brien, Nedelkoska, 1 and Frasher 2017) – in 2014, 85 percent of Albanians favored Albania’s membership in the EU (World Bank 2015). 3. Economic and political developments over the past few years, together with regional and global trends, sharply changed the structure of Albania’s economy . Following the collapse of communism in Albania, private sector development was limited to only a few sectors, namely agriculture and mining (O’Brien, Nedelkoska, and Frasher 2017). With fewer opportunities in Albania, migration became a jobs strategy for households in the early 2000s, overwhelmingly to a number of EU countries. This, in turn, funneled remittances back to Albania, averaging 15.2 percent between 2000 and 2008, and contributed to a boom in construction and services (O’Brien, Nedelkoska, and Frasher 2017). By 2012, Albania’s net migration rate was 3.2 percent, significantly higher than that of most countries in Eastern Europe, lagging behind only Bosnia and Herzegovina. The structural transformation of the economy away from agriculture also contributed to make the services sector one of the most important employers in the country. Services currently comprise the largest sector of the economy (45 percent of GDP), and over the last few years provided the highest contribution to economic growth, with higher-than-average growth rates (over 4 percent). The contribution of agriculture (which comprises nearly 30 percent of GDP) to economic growth has shrunk over time; since 2014, manufacturing and other industries have increased their contribution to growth (Figure 2). Figure 2: Services provide the highest contribution to economic growth Contribution to value added growth (percentage points) 5.0 average growth in value added in period 4.0 3.0 2.0 1.0 0.0 -1.0 -2.0 2008-2011 2011-2014 2014-2016 Agriculture Manufacturing Other industry Services Source: Estimates based on World Development Indicators. 4. Structural and cyclical factors, as well as the desire for EU accession, have led policy makers in Albania to confront the question of how to improve jobs outcomes more squarely and urgently. The challenge, no doubt, varies from region to region, and subpopulation to subpopulation. While the challenge is often about creating more jobs, it is also about increasing productivity; raising the quality of jobs; and increasing participation of underrepresented groups in the labor market. Despite variations in the jobs challenge, the World Bank’s 2013 World Development Report (WDR) argues that jobs are the key to people working their way out of poverty and hardship (World Bank 2012). The report argues that jobs are key because they enable poor people to use their most abundant asset, namely their labor, to generate income. This income sometimes comes from wage employment in the formal sector, but may 2 also come from wage employment or self-employment in the informal sector. These earnings streams are often sustainable avenues out of poverty. In addition, the WDR argues that jobs provide more than a paycheck to workers, contributing to workers’ skills acquisition and thus enhanced productivity; women’s empowerment; enhanced security through productive engagement of youth; and social stability in conflict and post-conflict societies (World Bank 2012). 5. The Government of Albania is motivated to improve jobs outcomes. The objectives outlined in the National Employment and Skills Strategy (NESS) 2014–2020 have significant direct and indirect impacts on jobs outcomes in the country. The NESS recognizes the importance of jobs for ensuring sustainable poverty reduction and shared prosperity. More specifically, the NESS promotes quality jobs and skills opportunities for all Albanian women and men by: (i) fostering decent job opportunities through effective labor market policies; (ii) offering quality vocational education and training to youth and adults; (iii) promoting social inclusion and territorial cohesion; and (iv) strengthening the governance of the labor market and qualification systems (Republic of Albania 2017). In its 2016 Annual Progress Report, the government indicated that 48 out of the 52 actions identified in the NESS had been met. More recently, a midterm review of the NESS is taking place to inform further implementation and, going forward, the development of a broader and more comprehensive jobs action plan. 6. What is needed to address Albania’s jobs challenge? On one hand, demand for workers is needed from a dynamic private sector that can provide productive jobs. On the other hand, the population needs the assets, education, and incentives to take up the jobs that are offered. Understanding the nature and job-creation potential of firms (labor demand side) and the characteristics and constraints of actual and potential workers (labor supply side) is a first step toward formulating a jobs action plan. 7. This note provides a brief, updated analysis of jobs dynamics in Albania, providing insights into where constraints to improving jobs outcomes remain and opportunities for addressing such challenges. Results-based policy making requires timely information to identify problems, design potential solutions, and evaluate policy initiatives. Using the most recent data available on Albanian labor markets from the perspectives of labor demand (firms) and labor supply (individuals), this note provides some key insights into the current situation and important dynamics over time and across firms and workers with different characteristics. The note is complemented by two other reports that look at (i) skills development challenges from the demand (employer) side, and (ii) the role and effectiveness of the National Employment Service (NES) in reducing unemployment. This note and the aforementioned reports will serve as inputs to a jobs framework and action plan for the Republic of Albania. 8. This note is divided into three additional sections. Following this introduction, the second section provides a profile of labor demand in Albania, looking primarily at job creation and job productivity. The third section presents a profile of labor supply, specifically who is working, what types of jobs they are employed in, and who is not working. This section looks particularly at job quality and job inclusiveness. The final section concludes with an overall brief summary of the analysis and questions to further guide the development of a jobs action plan. 3 II. Job dynamics from the labor demand side 9. This section looks at jobs and job creation from the labor demand side, focusing on the formal sector. In the transition to modern, market-based economies, the formal wage sector generally increases and ultimately accounts for a majority of employment. The analysis below draws on the Institute of Statistics’ (INSTAT) Structural Business Survey (SBS), whose sample is drawn on the Albania firm registry (see Annex I for more information and definitions). As such, all results refer to employment in formal/registered firms. In 2015, this corresponded to 416,680 jobs in 104,536 registered firms (based on weighted estimates from SBS 2015. This represents roughly one-half of the total paid employment captured in Albania’s Labor Force Survey (LFS) (737,700 is the sum of employers, wage employees, and single proprietor/own-account workers in 2015, based on the LFS). As discussed below, this discrepancy points to an informal sector that still accounts for a large share of employment in Albania. 10. The analysis gives a snapshot of jobs in registered firms in 2015 1 and an overview of jobs dynamics between 2002 and 2015 in the formal wage sector, with a focus on the period after the financial crisis (2013–2015). The 2015 firm registry: A snapshot of the employment stock 11. Most formal employment is in either micro firms or large firms . The median size of firms in Albania is 1 employee. Micro firms (1–4 employees) represent 90 percent of the total number of registered firms and account for 29 percent of jobs. Taken together, firms with less than 10 employees account for 38 percent of all employment – a high share compared to Organization for Economic Co- operation and Development (OECD) countries, where they account for around 20 percent (Criscuolo, Gal, and Menon 2014). Taken together with the sizeable informal sector, where firms are smaller than in the formal sector on average, the share of employment in micro firms in Albania is likely very high. On the other hand, large formal firms (here defined as more than 100 employees) represent less than one-half of 1 percent of the total number of firms, but account for one-third of all jobs (Figure 3). 12. Employment in the formal private sector is concentrated in Tirana region, which accounts for more than one-half (52 percent) of all jobs. The employment distribution by region overall reflects the geographical disparities in the distribution of firms. According to the LFS (which should cover both formal and informal employment), 28 percent of the working-age population lives in Tirana and 25 percent of all employment (formal and informal) is based there. However, Tirana, which hosts 35 percent of all registered firms, accounts for more than one-half of total formal sector employment, and larger firms are also concentrated there. Durres and Fier, which host about 10 percent of registered firms each, account for 12 percent and 7 percent of jobs, respectively (Figure 3b). 13. One out of four jobs are in the low-productivity trade and repair sector, while another 25 percent are in the industry sector (excluding construction and agroprocessing). Remaining jobs are mostly in business services (14 percent of jobs), tourism (12 percent), and other services (11 percent) (Figure 3c). While the vast majority of registered firms operate in trade and repair (43 percent) and tourism (22 percent), they are smaller in size on average. One-quarter of all jobs are in young firms (5 years or less in business), which represent one-half of all firms, and another one-quarter are in mature 1 2016 data will be released in March 2018. 4 firms (between 6 and 10 years old, representing about one-quarter of all firms) (Figure 4a). This reflects a somewhat younger firm population than in OECD countries, where 60 percent of jobs are in firms with more than 10 years in business (Criscuolo, Gal, and Menon 2014). Finally, a high share of jobs – 37 percent – are in the 20 percent most productive firms (Figure 4b). Figure 3: Jobs distribution by firm characteristics (size, region, sector) a. By size (number of employees), % 35 32 29 30 25 21 20 15 9 10 10 5 0 1-4 5-9 10-49 50-99 100+ b. By region, % 60 52 50 40 30 20 12 7 5 5 5 4 10 3 3 2 2 1 0 c. By economic sector, % 30 25 24 25 20 14 15 12 11 9 10 5 3 2 0 Trade and Industry (excl. Business Tourism Other services Construction Agr. and agro- ICT repair manuf of food) services processing Source: Estimates based on SBS 2015. 5 Figure 4: Jobs distribution by firm characteristics (age, productivity) 1. By age, % b. By productivity quintile 35 40 37 32 30 35 25 26 25 30 25 23 20 20 15 15 12 15 15 10 6 9 10 5 5 0 0-5 years 6-10 years 11-20 21 years Unknown 0 years and more Lowest 2nd 3rd 4th Highest Source: Estimates based on SBS 2015. Note: Productivity is defined as sales per worker. Job dynamics in Albania 2002–2015: Trends and flows 14. A one-year snapshot of job creation is limited from two perspectives. First, like other countries in Eastern Europe, Albania experienced the economic impact of the financial crisis between 2007 and 2012 (roughly). Trends in job creation in the formal sector are important to understand the current landscape. 15. Second, net job creation from one year to another can hide significant dynamics as jobs are created and destroyed across sectors and firms. The concepts involved in gross job flows analysis – which is based on firm registry data over several years – are explained in Error! Reference source not found.. A s uccessful transition process means that jobs become more productive – either because workers become more productive in the job that they are doing, or because low-productivity jobs are destroyed (as firms downsize or close down entirely) and higher-productivity jobs are created instead (by firms created and/or recruiting). In Eastern and Central Europe, the economic restructuring process generally started off with high job destruction in sectors that were important prior to transition, especially heavy industry. With time, new sectors took off, and gross job creation increased. As a result, excess job reallocation (the additional jobs that need to be created and destroyed to accommodate a net change in employment) also fell. Over time, the role for firm entry and exit in overall job turnover also diminished (Alam et al. 2008). 6 Box 1: Gross job flows glossary Gross job creation (GJC) in one year (in a particular sector) is the sum of all employment gains that year in firms (in that sector) that start up or expand during the year. Gross job destruction (GJD) is the sum of all employment losses in firms that contract or shut down (in that sector) during the year. Net job creation (NJC) is the difference between gross job creation and gross job destruction. Gross job reallocation (GJR) is the sum of gross job creation and gross job destruction, and characterizes the dynamics (extent of job creation and job destruction) of the labor market. Thus it is quite possible to have high gross job reallocation but low net employment growth (if high creation and destruction cancel out). Such dynamics, while not leading to more jobs, can lead to better/worse jobs, with higher/lower productivity and related earnings possibilities. Excess job reallocation (ERJ) is the difference between the gross job reallocation and the absolute value of the net employment growth. This measure captures the amount of churning by firms; i.e., how much actual job reallocation exceeds what would be necessary to accommodate the net change in employment. These indicators are expressed in thousands of jobs, or as rates by dividing them by total employment numbers (in a particular sector). NJC = GJC-GJD GJR= GJC+GJD EJR = GJC+GJD – Absolute value of [NJC]. Job creation is positive, though concentrated in a few firms and sectors 16. Net job creation – the difference between jobs created and jobs destroyed – has been increasing since the financial crisis. Net job creation in formal private firms fell significantly during the financial crisis (2008–2012) but increased successively between 2013–2015, reflecting increasing gross job creation, and falling gross job destruction (Figure 5). Between 2014 and 2015, 38,700 new jobs were created in registered firms and 19,050 destroyed, resulting in a net employment change of 19,650 jobs. (It is worth pointing out that this number of jobs is in fact below the number of jobs placements undertaken by the NES in 2015,2 according to the institution’s administrative records. ) Clearly, not all jobs in Albania are created through the NES (a minority, more likely), and at any rate NES jobs placements cannot exceed total jobs created (Box 2). 2 In 2015, the NES placed 18,762 jobseekers into jobs through intermediation services and 1,961 through employment promotion (EP) programs. 7 Box 2: The National Employment Service and Data Monitoring and Management Over the past few years, under the National Employment and Skills Strategy (NESS), the National Employment Service (NES) has made important steps to transform itself. It now operates under a new service model in at least six reconstructed offices that includes counseling services and mediation to better match the unemployed with the correct employers (Republic of Albania 2017). Critical investments were made in the operability of its information and communications technology (ICT) system and links with other national databases to support job matching (Republic of Albania 2017). While improvements have been made, inconsistencies between NES jobs placement numbers and the actual number of jobs created in 2015 point to the need to improve data monitoring and management. As pointed out in the Demand for Skills in Albania Report (World Bank 2018 forthcoming), upgrading the NES’s Labor Market Information System (LMIS) would be an important first step to improving not only data management, but also management of the various programs and interventions carried out under the NES. Figure 5: Gross job flows, 2002–2015 35 30 25 20 15 10 5 0 -5 -10 -15 -20 2008-09 2011-12 2002-03 2003-04 2004-05 2005-06 2006-07 2007-08 2009-10 2010-11 2012-13 2013-14 Gross job creation rate Gross job destruction rate 2014-2015 Net job creation rate (=employment change) Gross job reallocation rate Excess job reallocation rate Source: Estimates based on SBS 2002–2015. 17. Large firms contributed most to net employment growth. While there are few large firms in total, large firms contributed much more significantly to net job creation in the total period between 2002 and 2015, and especially in the post-crisis period (2013–2015), when net job creation peaked (Error! R eference source not found.6). The average annual number of jobs created between 2013 and 2015 was about 15,000, as opposed to virtually none during the crisis period between 2008 and 2012, and to 2,400 in the pre-crisis period (2002–2007). Seventy-five percent of net employment was created in large firms during the last period; small firms contributed negatively to job creation instead. Firms with less than 10 employees accounted on average for only about 8 percent of jobs created but for 20 percent of jobs destroyed during the same period (2013–2015). 8 18. This shift toward larger firms is consistent with a significant increase in the share of medium and large firms among new entrants. The significant reduction in entry rates of micro firms (with less than 5 employees) explains why the contribution of small firms to job creation has become more negative since 2002. The share of micro firms among new entrants decreased from 95 percent in 2008 to 70 percent in 2015, while the share of small- (10–49 employees) and medium-size (50–99 employees) firms increased the most. This may reflect the government’s anti-informality campaign, which featured more tax audits and higher penalties for noncompliance. To the contrary, while firms with less than 10 employees represent the highest share of exit firms, their share of total exit firms decreased over time while the share of firms with more than 10 employees among exit firms increased significantly over time.3 As a result, Albania’s firm entry rates are now approaching those of OECD countries, but its exit rates are considerably higher (Figure 7a). Understanding whether exit rates are so high because of competitive pressures or because of a dysfunctional business climate requires further analysis. 19. Tourism accounts for the majority of new entrant firms but also for most firms that exit . Looking at the composition of entering firms by sector, tourism, industry, and business services account for the largest share of entrant firms (37 percent, 19 percent, and 14 percent, respectively, in 2015), with the entry of business services firms growing more rapidly (Figure 7b). One-third of exit firms operate in trade and repair, and 18 percent in tourism. Tirana, Durres, and Vlore are more dynamic, accounting for the highest shares of new entrant firms (38 percent, 15 percent, and 9 percent, respectively) but also of exit firms (32 percent, 12 percent, and 8 percent, respectively). Lezhe stands as the region with the fastest growing share of new entrant firms, accounting for 1.6 percent of new entrant firms in 2005, 11 percent in 2014, and 8 percent in 2015. Figure 6: Jobs trends by firm size a. Percentage contribution to total (%) b. Number of net jobs created 100% 20,000 80% 60% 15,000 40% 20% 10,000 0% -20% 5,000 -40% -60% 0 Gross Job Gross Job Net Job Gross Job Gross Job Net Job Creation Destruction Creation Creation Destruction Creation -5,000 2008-2012 2013-2015 2002-2007 2008-2012 2013-2015 1-4 5-9 10-49 50-99 100+ 1-4 5-9 10-49 50-99 100+ Source: Estimates based on SBS 2002–2015. 3 Note that that entry and exit rates are computed based on the survey (SBS), not the registry of firms itself. However, the SBS includes the exhaustive census of enterprises with 10 and more employed, hence the entry/exit rates reflect the entire universe of firms. Since the SBS surveys a sample of enterprises with 1–9 employees (including proprietors), the derived entry/exit rates may be problematic. Using the complete registry of firms for such analysis would be preferable. See Annex I for more details. 9 Figure 7: Entry and exit rates of firms a. Entry and exit rates, based on firm inception b. Share of new entrant firms, by sector (%) date Source: Estimates based on SBS 2002–2015. Note: Entry rate: entering firms in a given year based on the firm inception date, as % of all firms excluding one-year firms in a given year. Exit rate: exiting firms (based on the last year they are observed in SBS) in a given year, % of all firms excluding one- year firms in a given year. 20. Business and other nontrade services sectors increased their contribution to job creation over time and contributed the highest share of net jobs created between 2013 and 2015. Within services, more productive business and other nontrade services accounted for most of the employment change (35 percent of net job creation, equivalent to about 5,400 new jobs created) during the post-crisis period 2013–2015, increasing by more than four times their contribution to job creation before the crisis. Tourism has the third highest average net job creation rate in recent years; on average 821 net jobs were created between 2013 and 2015, representing 6 percent of the overall employment change in the same period. Industry, mostly manufacturing excluding manufacturing of food and beverages, experienced downsizing and closures during the crisis years. Since then, however, the industrial sector has recovered, accounting for 30 percent of net job creation in 2013–2015, or about 4,400 jobs. The construction sector followed a similar pattern (Figure 8). 10 Figure 8: Jobs trends by firm sector Contribution to net job creation by firm sector (% contribution to total) ICT 1 Ag. and agro-processing 4 Tourism 6 2013-2015 Construction 9 Other services 14 Trade and repair 14 Business services 22 Industry (excl. agro-processing) 30 Industry (excl. agro-processing) -179 Construction -141 Tourism 3 2008-2012 ICT 31 Ag. and agro-processing 36 Other services 55 Business services 105 Trade and repair 190 ICT -5 Business services 1 Ag. and agro-processing 2 2002-2007 Other services 8 Tourism 10 Construction 23 Industry (excl. agro-processing) 26 Trade and repair 35 Source: Estimates based on SBS 2002–2015. 21. Job-creation patterns reflect spatial disparities in the distribution of firms and employment. Sixty percent of the average net jobs created between the period 2013–2015 (about 9,000) are in Tirana region. However, net job-creation rates during the same period were higher in Kukes and Lezhe than in Tirana (Figure 9). 22. Up until recently, young firms accounted for a disproportionate share of job creation, but older firms have increased their contribution to net job creation . In 2013–2015, unlike in the crisis or pre-crisis period, young, mature, and old firms accounted for roughly the same proportion of net job creation (Figure 9a). Simple regression analysis with random effects during the 2002–2015 period based on a few observable firm characteristics (such as sector, region, age, and firm size) confirm that larger, younger, and firms operating in trade and repair grew faster (see Annex II). 23. More productive firms have accounted for more job creation recently . The 20 percent most productive firms in the formal sector saw the most drastic dynamics in the period 2002–2015: after a period of net contribution to employment prior to the crisis, the more productive firms lost jobs in response to the economic crisis. This is consistent with evidence from the Europe and Central Asia region, where small, young, and more productive firms suffered more during the economic crisis (Arias et al. 2014). After the crisis, more productive firms appear to have rebounded more significantly, accounting for nearly 40 percent of net job creation (Figure 10b). 11 Figure 9: Jobs trends by region a. Contribution to net job creation by region (% contribution to total) 120 100 80 60 40 20 0 -20 -40 2002-2007 2008-2012 2013-2015 b. Average number of net jobs created 10,000 8,000 6,000 4,000 2,000 0 -2,000 2002-2007 2008-2012 2013-2015 Source: Estimates based on SBS 2002–2015. 12 Figure 10: Jobs trends by firm age and productivity a. Number of jobs, by age b. Number of jobs, by productivity quintile 20000 20000 15000 15000 10000 10000 5000 5000 0 0 -5000 -5000 -10000 -10000 0-5 years 6-10 years 11-20 years 21 years and Lowest Second Third Fourth Highest more quintile quintile 2002-2007 2008-2012 2013-2015 2002-2007 2008-2012 2013-2015 Source: Estimates based on SBS 2002–2015. 24. Firms in Albania grow over time during their lifecycle, especially firms that start small. Table 1 shows that within-firm growth in Albania occurred between firm size at inception (proxied by the year the firm first appears in the SBS panel) and its size as of 2015. Conditional on survival, one-third of firms grew in employment over their lifecycle as of 2015, about one-half of firms stayed the same, and 17 percent shrank.4 Employment growth rates are higher for micro and small firms but also for firms that start already large, indicating that starting size may play a key role in firm dynamics. The table also shows that as firms get older, they get larger, consistent with theory that surviving and more productive firms scale up. Substantial within-firm employment growth occurred in construction, business services, industry, and agroprocessing (about one-half of firms operating in these sectors grew over time). However, the firm dynamics patterns differed across these sectors: while a substantial share of firms in construction shrank (about 37 percent), fewer firms downsized in the agroprocessing and business services sectors. The highest share of growing firms is found in Kukes, perhaps linked to higher foreign direct investment, and less surprisingly in Tirana and Durres regions. Table 1: Within-firm employment change since inception Same size Grew (-5%<= change <= (change>+5% Shrank (change< -5%) +5%) ) Total Size (initial= first year in the panel) Micro 1 (1–4) 13.1 57.5 29.5 100 4 The result depends on whether employment is compared only in the first and last year the firm is observed in the panel or if continuously growing firms (with positive growth every year) are used. For the total economy, the share of job-creating firms (weighted) is one-third if using the first approach (reported in the table); using the second approach, the average 2002–2015 share of growing firms is 4 percent (firms in the weighted sample observed in the panel for at least two years); in 2015, 4.6 percent of firms registered positive annual employment growth. 13 Micro 2 (5–9) 40.0 10.6 49.4 100 Small (10–49) 47.2 6.3 46.5 100 Medium (50–99) 55.0 9.7 35.3 100 Large (100+) 51.0 9.8 39.2 100 Age (initial= first year in the panel) 0–5 years 16.7 51.6 31.8 100 6–10 years 18.4 48.1 33.5 100 11–20 years 14.9 55.9 29.2 100 21 years and more 32.7 19.1 48.3 100 Sector (initial= first year in the panel) Agriculture and agroprocessing 22.3 40.4 37.4 100 Industry excl. agroprocessing 22.3 39.8 38.0 100 Construction 36.7 15.5 47.9 100 Trade and repair 12.7 55.2 32.1 100 Tourism 18.9 55.1 26.1 100 ICT 25.6 38.6 35.8 100 Business services 20.2 39.1 40.7 100 Other services 11.8 64.0 24.2 100 Region (initial= first year in the panel) Berat 13.7 65.9 20.4 100 Diber 18.8 48.7 32.5 100 Durres 19.0 44.7 36.3 100 Elbasan 20.0 55.2 24.8 100 Fier 9.3 67.1 23.7 100 Gjirokaster 16.7 58.4 24.9 100 Korce 12.6 63.2 24.2 100 Kukes 19.6 25.1 55.3 100 Lezhe 15.8 46.5 37.7 100 Shkoder 13.3 55.9 30.7 100 Tirana 20.3 40.8 38.9 100 Vlore 16.7 55.4 27.9 100 Total 16.9 51.3 31.8 100 Note: Size change = employment growth is the difference in the number of employees when the firm first appeared in the panel and in the latest year observed (in 2002–2015) divided by the number of employees plus one proprietor in the first year (revised employment is used to avoid losing many observations with zero employment in the first year). Numbers refer to the percentage of firms that shrank, remained the same size, or grew with respect to their employment at inception (i.e., first year in the panel). Some firm types perform better than others 25. An increase in productive employment opportunities is largely a question of growing demand for labor from more productive firms. Understanding the characteristics and dynamics of such firms is thus of particular importance to unlock their growth potential. More productive firms increased their share of job creation, and now account for a disproportionate share of employment (i.e., large compared to their share of firms). However, large firms are not necessarily more productive than smaller firms. Figure 11 shows that the distribution of firms by productivity level is not necessarily to the right in the chart for larger firms (which would indicate generally higher levels of productivity). This is an important 14 finding as under competitive market conditions, the most productive firms should be more competitive and as such should be able to expand production and employment. The fact that productive firms remain “stunted” suggests that some factor is impeding competitiveness. Figure 11: Distribution of firms by productivity level and size category Source: Estimates based on SBS 2015. 26. Productivity levels widened over time for sectors outside of services and for trade. Comparing 2015 to 2002, the variance of firms in agriculture, agroprocessing, other industry, construction, and trade increased: the distance between low-productivity and high-productivity firms increased, as did the differences between sectors. Moreover, industry, construction, and business services became more productive, with construction the most productive sector of all. In the services sector, the trend was the reverse. Differences in productivity levels fell both within and between subsectors (Figure 12). 15 Figure 12: Distribution of firms by productivity level and sector a. Sectors outside services, and trade b. Services sectors, except trade Source: Estimates based on SBS 2002 and 2015. 27. In Albania, no clear cluster of high-growing firms exists. What are the key characteristics of job- creating firms? In more advanced economies, so-called “gazelles” – a set of young, innovative, and competitive firms – often grow more rapidly in terms of employment than other firms (Birch and Medoff 1994). Albania has no such gazelles, however. Ranking firms by their annual job-creation rate between 2014 and 2015, the top 20 percent of firms accounted for 26 percent of total job creation. The top 10 percent and 5 percent accounted for 15 percent and 10 percent, respectively, suggesting that only by narrowing down to 5 percent does a different set of firms emerge. 28. The relationship between job creation and productivity has been neither linear nor always positive, pointing to labor and product market allocative inefficiencies . Over all firms (conditional on survival), job creation has been higheramong more productive firms before 2007; during the years after the crisis until 2014, the correlation between job creation and productivity was weak. The highest job- creation rates were not among more productive firms, pointing to inefficiencies in the allocation of labor 16 and potential market segmentations, possibly across firm size and sector. In 2015, the relationship become positive and linear, with more productive firms accounting for the majority of net jobs created (Figure 13). Simple regression analysis during the 2002–2015 period based on a few observable firm characteristics (such as sector, region, age, and firm size) shows that less productive firms tend to grow faster. However, the predictive power of the statistical model is not strong, as many important variables explaining either firms’ employment growth rate or labor productivity are omitted (see Tables AII.1 and AII.2 in Annex II for estimation results). More analysis is needed to estimate the magnitude and statistical relevance of the correlation between productivity and job creation to assess the economic significance of increases in productivity, which is the percentage of job-creation growth associated with each percentage increase in labor productivity. The aim is to provide evidence on (i) whether factor and product markets are working efficiently by allocating jobs toward more productive firms, and (ii) whether workers are benefiting from productivity gains in their sectors in terms of higher wages. Figure 13: Net job-creation rates by productivity quintile 20.0 15.0 10.0 Lowest quintile 5.0 Second 0.0 Third -5.0 Fourth -10.0 Highest quintile -15.0 -20.0 2002-03 2003-04 2004-05 2005-06 2006-07 2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 Source: Estimates based on SBS 2002 and 2015. Note: Firms are ranked by quintiles of labor productivity from lowest (least productive firms) to highest. Labor productivity is measured as sales per worker. 29. The productivity of the firms with the highest job creation (the “gazelles”) is higher than that of the average Albanian firm. The narrower the definition of “gazelles,” the higher the productivity level compared to the average firm. However, the most productive firms do not create the most jobs – in 2014– 2015, the average productivity level of all expanding firms was higher than the top 5 percent (Table 2). Over time, some shifting occurred in the sector distribution among the top 5 percent job-creating firms. In 2014–2015, business services, industry, and tourism accounted for two out of three “gazelles,” although they accounted for just over one-third of all firms. In 2002–2003, by contrast, construction and trade had a much larger share than business services and tourism among “gazelles.” Both business services and the industry sector are significantly overrepresented among the “gazelles” compared to their share of total firms. From a regional perspective, Durres in particular is overrepresented, accounting for 27 percent of “gazelles” but only 11 percent of all firms. 17 Table 2: “Gazelles” and their key characteristics in Albania 2002– 2006– 2011– 2014–2015 2003 2007 2012 Top 5% Top 5% Top 5% Top 5% Expanding All firms firms Key characteristics Share of gross job creation 10 15 10 10 100 -- Median size 10 9 8 8 16 1 % with less than 50 employees 97 89 92 85 83 99 Median age 7 5 10 7 9 6 Mean productivity 213 156 176 229 265 100 (All firms =100=) Sectoral distribution Agriculture and agroprocessing 5 0 1 3 5 3 Industry excl. agroprocessing 20 28 23 21 23 6 Construction 29 28 39 8 11 4 Trade and repair 19 18 17 13 27 43 Tourism 17 15 7 20 11 22 ICT 3 1 1 2 2 1 Business services 2 6 7 25 10 8 Other services 5 4 5 8 12 12 Regional distribution Berat 8 3 2 3 3 4 Diber 2 0 9 1 2 2 Durres 9 25 31 27 15 11 Elbasan 3 3 8 3 5 8 Fier 11 5 5 9 7 11 Gjirokaster 7 7 2 4 3 3 Korce 7 2 4 1 5 6 Kukes 7 2 5 1 1 1 Lezhe 0 5 0 4 4 4 Shkoder 3 6 5 4 6 6 Tirana 37 31 26 37 41 35 Vlore 7 10 3 7 6 8 Source: Estimates based on SBS 2002–2015. Note: Gazelles are defined in terms of the year-to-year job-creation rate (i.e., number of jobs created divided by previous-year employment). Expanding firms are those firms with a positive job-creation rate between 2014 and 2015. Gross job flows – adding up and netting out 30. Job turnover (churning) remains higher in Albania than in more advanced reformers. Significant differences arise in gross job flows patterns across transition countries, which in turn are related to how far the transition process has advanced (Arias et al. 2014). Advanced modernizers, such as Estonia and Poland, experienced strong gross job creation and some job destruction before the financial crisis. 18 Countries that were less advanced on the transition process were still stuck with higher levels of excess job reallocation. In Albania, high destruction and high creation of jobs still coexist. For a given change in net employment, more jobs are simultaneously destroyed and created in Albania than in advanced reformers. For example, excess job reallocation rates ranged just below 10 percent in Estonia prior to the financial crisis: as seen above (Figure 5), they still hover around 20 percent in Albania. 31. These churning rates are highest for smaller firms. Over the period 2013–2015, job-creation rates were lower for small firms (1–10) and job destruction was dramatically higher in smaller firms than in others. As a result, small firms with less than 10 employees saw dramatic reductions in net employment, whereas net employment increased, on average, in other firms. Churning was consequently tremendously higher in smaller firms, where letting go of just one person makes a big percentage difference in total employment, than in larger firms (Figure 14a). It was also higher in construction, trade and repair, and tourism (Figure 14b). Among regions, job turnover was highest in Kukes, whereas Tirana had more moderate levels of churning (Figure 14c). 19 Figure 14: Where is churning highest (2013–2015)? a. By firm size 80 60 40 20 0 1-4 5-9 10-49 50-99 100+ -20 -40 -60 Gross job creation rate Gross job destruction rate Net job creation rate b. By sector 20 c. By region Source: Estimates based on SBS 2013–2015. 32. Excess job reallocation has increased since the pre-crisis period for both younger firms and the oldest. Similarly, the most productive and the least productive firms are the ones with highest job turnover relative to the jobs they create (Figure 5). Again, these charts show no sign of increased efficiency in job creation in the formal sector – churning remains high. Figure 15: Churning: Excess job reallocation rates, by age and productivity quintile a. By age b. By productivity quintile 40 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 2002-03 2006-07 2011-12 2014-15 2002-03 2006-07 2011-12 2014-15 0-5 years 6-10 years Lowest quintile Second Third 11-20 years 21 years and more Fourth Highest quintile Source: Estimates based on SBS 2002–2015. 21 Section II summary 33. Available firm data present a picture of jobs in formal private sector establishments, their job creation, and their propensity for productivity in Albania. The key messages are as follows: • While formalization increased over time, formal private sector employment represents one-half of total paid employment, pointing to a still high informal sector share. • Most formal private sector employment is in micro (less than 5 employees) or large (more than 100 employees) firms, located in Tirana region, in low-productivity sectors, and in the top 20 percent productive firms. • Net job creation, the difference between jobs created and jobs destroyed, has increased since the financial crisis and can largely be attributed to job creation in large firms within the business and other nontrade services sectors. • In 2015, only 4.6 percent of firms were job-creating (net), compared to firms that shrank or stayed the same over their lifecycle. The highest share of job-creating firms is in construction and business services, in Kukes and among old (more than 21 years of operation) and small firms (with less than fifty employees). • The relationship between job creation and firm productivity has been neither linear nor always positive, pointing to inefficiencies in the allocation of labor; in the absence of major distortions in the business climate, resources (in terms of sales and workers) would efficiently flow to more productive firms. Only recently (2015) did more productive firms start to account for greater job creation; however, the most productive firms (top 5 percent, the ”gazelles”) did not create the most jobs. • Larger firms are not necessarily more productive than smaller firms, suggesting that some factor is impeding competitiveness. Construction, industry, and business services experienced an increase in productivity between 2002 and 2015 and as of 2015 were the most productive sectors. • Job turnover is still high in Albania compared to other advanced reformers (Estonia, Poland), especially among smaller firms. Coupled with the increasing exit rates observed in the SBS, the finding may indicate that barriers to entry are not that high, but the probability of survival for micro and small firms is low, pointing to specific constraints among these smaller firms. 22 III. Labor supply: who works, in what kind of jobs, and who lacks a job? 34. This section presents a profile of the Albanian labor market situation from the labor supply (individual) side, using information from the Albanian Labor Force Surveys (LFS) from 2014, 2015, and 2016, with a focus on the most recent survey.5 A snapshot of the Albanian labor force 35. In 2016, just over one-half of the Albanian working-age population was employed and one-third was neither in school nor working. The working-age population comprised just over 2 million people, of which 1.3 million were active, and 0.7 million inactive, translating into a labor force participation rate of 66 percent of the population. Some 1.1 million (56 percent) were employed, and 0.9 million were jobless. Of those outside employment, some 0.2 million – 16 percent of the active population – were unemployed (Figure 16). Together, those that were neither employed (but actively looking for jobs or the unemployed) nor in school made up 0.6 million, or one-third of the working-age population. Figure 16: Working-age population in 2016: A snapshot Employed: 1120 Jobless: 885 Source: Estimates based on LFS 2016. 5 A change in the LFS sampling methodology occurred in 2012, undermining the comparability of data before and after 2012. 23 36. Albania’s employment rate is low by European standards but higher than in other Western Balkan countries (Figure 17). Men’s employment-to-population ratio (the share of men employed in the working-age population) is higher in Albania than in other Western Balkan countries, but is second lowest among a set of more advanced European comparators, higher only than that of Greece. Women’s employment-to-population ratios, at 50 percent, are also substantially higher in Albania than in neighboring Balkan countries (but below the EU28 average of 61 percent). 37. Access to employment has increased since 2014, but unemployment remains greatest among the most highly educated. Overall, from 2014 to 2016, the employment-to-population ratio increased quite significantly, from 51 percent to 56 percent for men, women, youth (aged 15–29), and older workers. The only group for which employment stagnated was the tertiary educated, which already had the highest unemployment rate (in 2016, the overall unemployment rate was almost 16 percent, versus 17 percent for the tertiary educated and 14 percent for those with a primary school education). Between 2014 and 2016, unemployment rates fell (again with the exception of the tertiary educated), from 18 percent to 16 percent. The unemployment rate remains very high for young people (aged 15–29), however, at 29 percent in 2016 (Figure 18). Figure 17: Albania’s employment rate is the highest in the Western Balkans Source: SEE Jobs Gateway 2017 (World Bank and WiiW). Note: BA= Bosnia and Herzegovina. 24 Figure 18: Employment strengthened and unemployment fell post-2014 Employment-to-population ratios Unemployment rates 80% 35% Share of population in age group (%) 70% 30% Share of active population (%) 60% 25% 50% 20% 40% 15% 30% 10% 20% 10% 5% 0% 0% Male 15-29 30-64 Secondary Total Male Female Primary 15-29 30-64 Secondary Primary or less Tertiary Total Female Tertiary or less 2014 2015 2016 2014 2015 2016 Source: Estimates based on LFS 2016. Labor productivity is low and may continue to decline with changes in demographics 38. Labor productivity levels are low in Albania – lower, in fact, than in any other country in Eastern Europe (Figure 19). Labor productivity, measured as GDP output per worker, was about US$31,000 per worker in Albania in 2016, while it was US$60,000 per worker in Lithuania, US$57,000 in Estonia, and US$48,000 in Romania. Figure 19: Labor productivity is low in Albania 80 GDP per person employed (in 1000 70 Labor Productivity 60 (GDP per person employed) constant 2011 PPP $) 50 40 30 20 10 0 Romania Hungary Western Balkans -6 Slovak Republic Albania Estonia Slovenia Lithuania Croatia Portugal Cyprus Latvia EU 27 Czech Republic Poland Bulgaria Source: World Development Indicators and LFS 2016. 39. A potential concern is that Albania’s population is aging, which over the medium and long term will put pressure on productivity growth. Albania’s demographic profile is somewhat younger than that of the rest of the Western Balkans: the population under 30 comprises 41 percent of the total population, compared to 36 percent for neighboring countries. However, average age is increasing. Between 1980 and 2015, the share of population under 30 fell, whereas the share of population above 30 increased. 40. Migration abroad has become an important jobs strategy; however, it could also negatively impact productivity growth in the sending country in the long run. High outmigration implies that part of the productive workforce is abroad – often the most educated, although no recent data exist on this. 25 Clear benefits arise from migration, and the fact that people choose to migrate is evidence of these benefits to individuals and households. Nonetheless, the impact on economic growth and welfare in the sending country can be negative if remittances do not make up for the loss in workforce. 41. The number of Albanian emigrants increased more than threefold over recent decades: from 0.2 million in 1990 to 1.05 million in 2017, representing almost 40 percent of the resident population (UN Statistics 2018). This figure is the highest in the Western Balkans after Bosnia and Herzegovina. Albania experienced significant outflows of skilled workers (about 40 percent of their highly educated workforce had emigrated to OECD countries by 2010) ; in parallel Albania had also seen slight domestic wages increase with high unemployment and inactivity. Skilled emigrants may not be easily replaced by remaining workers; in addition, although remittances contribute to poverty reduction, they may also increase reservation wages (i.e., the lowest wage for which a person would be prepared to work) (IMF 2016). Finally, the impact of return migration (the transfer of knowledge and human capital obtained abroad) and financial investments in the diaspora likely generate economic and productivity gains (World Bank 2017). Jobs are not inclusive– especially for women, youth, the poor, and those with little education 42. Women, youth, and those with little education are more excluded from jobs. First, for every age group and every level of education, women are less likely to work. Young people are also much less likely to work than older adults, partly, as will be seen, because they are in education. For men, the differences disappear by age 30, but women’s employment access peaks between 40–44 years of age (Figure 20). The gender gap is highest for young women of childbearing age, and second highest for women approaching retirement. Educated women are much more likely to work than those with less education; in fact, the gender gap is highest for women with a secondary education, among whom only 44 percent work, compared to 65 percent of men with a secondary education. Figure 20: Access to employment differs by age, gender, and education level Employment-to-population ratio, by age and gender Employment-to-population ratio, by education and gender 60-64 64 24 40 55-59 70 55 14 Tertiary 69 64 5 50-54 79 62 17 45-49 81 67 14 40-44 80 71 9 Secondary 65 44 21 35-39 76 69 7 30-34 78 59 19 25-29 65 52 14 20-24 37 28 9 Primary or less 57 48 9 15-19 11 5 6 100 80 60 40 20 0 20 40 60 80 100 80 60 40 20 0 20 40 60 80 Share of employed in age group (%) Share of employed in population (education group) Female Gender gap Male Female Gender gap Male Source: Estimates based on LFS 2016. 43. Those not working are either in school, unemployed, or inactive for reasons other than education (largest share). Enrolment in education should be a valuable investment to improve future job prospects; as such, this is a good thing. In what follows the jobless are therefore defined as those who are neither in school nor working. Nearly 650,000 people are unemployed or inactive not in school. The 26 inverse of access to jobs, joblessness affects mostly women and youth. Young people, and especially young men, are more likely to be unemployed than others: one in four unemployed is a young man aged 15–29. Overall, the other inactive – those who are neither working nor looking for work – make up the largest share of the jobless. 44. Many young people are neither working nor building skills . In Albania, the share of young people who are neither in school nor employment (NEET) ranges between 30–40 percent, depending on age group and gender (Figure 21). These shares are about twice as high as the EU average. This is problematic from two perspectives: as shown below, much of this unemployment is long-term and structural in nature. A young person who does not find a job after school risks remaining without one. Second, given the low productivity and demographic pressures facing Albania, the high share of young people neither improving their future productivity through schooling nor contributing to productive work is an additional drag on economic growth and development. Figure 21: Women, youth, and workers close to retirement are more likely to be out of a job Inactive not in school and unemployed Youth not in Employment, Education or Training (NEET) EU(28) 60-64 20-34 14 23 55-59 15-34 12 19 50-54 45-49 20-34 31 41 40-44 15-34 27 36 35-39 Albania 30-34 22 40 30-34 25-29 30 46 25-29 20-24 40 38 20-24 15-19 19 20 15-19 60 40 20 0 20 40 60 60 40 20 0 20 40 60 80 NEET as % of population in age group Male Female Thousand people Female Male Unemployed Other inactive Source: Estimates based on LFS 2016. 45. Family responsibilities and lack of hope of finding a job hold back women’s labor market participation. For the age group 25–34, family responsibilities are the main reason explaining the share of inactive women not in education; the share of men engaged in family responsibilities is negligible (Figure 22). However, women are also more likely than men to be “discouraged workers” – those who would actually like to work, but have given up looking because they have not succeeded in finding a job. Finally, women are more likely than men to enter into retirement early: more than one-half of inactive women aged 55–64 are retired. The lack of hope is not without foundation. Unemployed women are most likely to be unsuccessfully looking for a job after a period of inactivity due to family responsibilities, whereas men are likely to have been employed before they became unemployed. The influence of social norms and traditional gender roles in women’s employment outcomes mirrors the conclusions of a recent analysis of labor demand and skills development in Albania, which found that employers considered women’s family responsibilities an important obstacle to hiring them (World Bank 2018 forthcoming). 27 Figure 22: Is there a “housewife” trap in Albania? Reasons for inactivity (excluding education) Unemployed by previous status (%) 55-64 Other 45-54 Inactive because of family responsibilities 35-44 Migration period abroad 25-34 Full-time education 15-24 Employment 100 50 0 50 100 (incl. apprenticeship Male thousand people Female and training) Family responsibilities Illness or disability 0% 10% 20% 30% 40% Discouraged Other Retirement Female Male Source: Estimates based on LFS 2016. 46. The poor also have poor labor market outcomes, reflecting lower education levels, lack of job- relevant skills, lower mobility, and other constraints to accessing better job opportunities. As elsewhere in the world, the poor are disadvantaged in Albania’s labor markets. As a result, they have lower labor force participation rates, higher unemployment and inactivity rates, and worse labor market outcomes, as they are generally engaged in unpaid jobs or low-productivity (and low-paid) jobs in the informal sector as self-employed and wage employees. More than one-half of working-age poor people are either unemployed or inactive not in school. Almost one-half of workers in the poorest quintile are self- employed versus one-third of workers in the richest quintile (Figure 23). Figure 23: Fewer good job opportunities for the poor b. Occupational status by consumption a. Labor market status by consumption quintile quintile Source: Household Budget Survey 2014. 28 47. Overall, unemployment in Albania is largely long-term and structural. Long-term unemployment (more than one year) characterizes 66 percent of the unemployed in Albania, compared to the average of 72 percent among Western Balkan countries. Nonetheless, the share of long-term unemployed among women is higher than among men (68 percent versus 65 percent). The share of unemployed who had been unemployed for less than one year is about the same (one-third) (Figure 24). In fact, unemployment is largely structural, with a majority of the unemployed having gone without work for a minimum of two years, and one-third for more than four years. Long durations of unemployment may explain the high shares of discouraged workers among Albania’s inactive population. Figure 24: Unemployment spells last long a. Share of unemployment by duration and b. Long-term unemployed, share of total gender unemployment, by gender Unemployment by duration (%) Female Male 0% 20% 40% 60% 80% 100% <1 year 1-2 years 2-4 years > 4 years Source: Estimates based on LFS 2016 for Albania. Panel b based on SEE Jobs Gateway 2017 (World Bank and WiiW). Note: BiH= Bosnia and Herzegovina. Job quality is a concern 48. Most of the employed do not work for a wage, but for themselves or their families. Among the 1.1 million employed Albanians, about 0.6 million, or 58 percent, are nonwage workers: they are not employees in a firm, but are active as heads of household enterprises or as unpaid contributing workers in such enterprises (Figure 25). Among the wage employed, about 20 percent are informally employed, meaning that they do not benefit from social security, paid annual leave, or paid sick leave. Additionally, over the last three years, the share of public wage workers decreased by 4 percentage points; nevertheless, the public sector still represents an important part of wage employment (36 percent). Hence in total, only one-third of Albania’s employed population and less than one-fifth of the total working-age population are in formal wage employment. The structure of employment is a central explanation for the low-productivity levels of jobs in Albania. 29 Figure 25: Type of employment in 2016 Employed: 1120 Source: Estimates based on LFS 2016. 49. Women have less access to “better” jobs, but no striking gender differences arise in the structure of employment. Almost one-third of employed women are unpaid family workers compared to one-fifth of men. Taken together with their low access to work, this implies that only one-third of women aged 15–64 have some form of paid employment, including nonwage work (Figure 26). Men are more likely to be employers and own-account workers. Women are slightly more likely than men to be formal wage workers, and comprise almost one-half of the formal wage workforce. The high share of nonwage work is mirrored in the high share of agricultural employment: 34 percent of all men’s jobs and 44 percent of all women’s jobs are in agriculture. With the exception of construction workers, who are almost exclusively men and agriculture, the share of women and men working in other sectors is similar. Figure 26: Both women and men work mostly in low-productivity sectors Employment by occupational status and gender Employment by economic sector 700 700 Total employed by occupation (thousand people) Total employed by sector (thousand people) 600 600 4%, 24 30%, 186 19%, 119 500 500 1%, 7 Employer 31%, 151 Other services 400 12%, 75 400 31%, 155 36%, 227 Unpaid family/unspec 12%, 72 10%, 50 Trade 300 1%, 3 300 24%, 118 Own account worker 12%, 75 14%, 72 Construction 11%, 67 200 200 7%, 35 Informal wage employee Industry 100 100 34%, 215 44%, 220 30%, 187 37%, 181 Formal wage employee Agriculture 0 0 Male Female Male Female Source: Estimates based on LFS 2016. 30 50. Albania has seen job creation in the past few years, but mostly in self-employment and in less productive jobs. In fact, the increase in employment-to-population ratios between 2014 and 2016 was accompanied by a downward trend in overall labor productivity (Figure 27). In other words, although new jobs were created (employment grew by 11 percent in the period, or by 114,000 jobs), these new jobs were, on average, less productive than existing jobs. Indeed, nonwage work increased most – especially informal own-account work (not captured in the registered business survey analysis in section II). Albania’s wage level is lowest among Western Balkan countries, and is significantly lower than the Austria. In 2016, average monthly gross wages in Albania were about 25 percent of those in Austria, while Montenegro’s wages were 54 percent of the Austrian level. On the positive side, informal wage work, as well as unpaid family work, fell as well between 2014 and 2016. Services accounted for most of the new jobs on a net basis (consistent with the firm registry data), whereas employment growth in agriculture was very low. Between 2015 and 2016, manufacturing accounted for the largest single increase in employment (27 percent), followed by agriculture (18 percent), trade (19 percent), business services (14 percent), ICT (10 percent), and tourism (7 percent).6 Figure 27: Low-productivity jobs a. GDP per employed person and employment b. Average monthly gross wages, Austria=100 (PPP €-based) GDP per employed person vs employment 800 1150 750 1100 Thousand constant Albanian Leke 700 Employment (thsd people) 1050 650 1000 600 950 550 900 500 450 850 400 800 2014 2015 2016 GDP per employed person Employment Source: Panel a is based on estimates based on WDI, Albania LFS 2014–2016. Panel b is based on SEE Jobs Gateway 2017 (World Bank and WiiW). Note: BA = Bosnia and Herzegovina. PPP = . 51. The new jobs went primarily to low-medium-educated adults. One in 10 jobs created went to a person aged 15–29 – for young women, the share in new job creation was only 4 percent. This may partly reflect a reduction in the youth population in the same period (2014-2016). A majority of new jobs were taken by those with a secondary education or less, while only 20 percent of jobs went to persons with a tertiary education. These job-creation shares are more or less proportional to the share of persons with low levels of education in Albania’s adult population. However, they are also a sign of the high share of low-skilled jobs that dominate the Albanian economy: 40 percent of jobs went to persons with primary levels of education or less (Figure 28). The Albania STEP Employer Survey conducted in 2017 (Honorati 6 Because of a break in the definition of different sectors between 2014 and 2015, it is not possible to conduct a detailed analysis of sector developments between 2014 and 2016. 31 and Johansson 2018 forthcoming) provides similar evidence of much higher demand for low- and medium- skilled occupations from formal firms. Figure 28: Most jobs are for adults and for low-medium-skilled workers Share of total job creation 2014-2016 Share of total job creation 2014-2016 Unpaid Female Adults (30-64) Youth (15-29) family Own account Male Adults (30-64) Youth (15-29) Employer Wage (formal) Wage Tertiary (informal) Secondary Services Primary or less Industry 0% 20% 40% 60% Agriculture -100% -50% 0% 50% 100% Source: Estimates based on WDI and LFS 2014–2016. Section III summary In summary, the employment situation has improved in Albania, but the creation of more inclusive and better-quality jobs still faces many challenges. The main points from this section are as follows: • In 2016, over one-half of the Albanian working-age population was employed, while one-third was neither in school nor working. • Access to employment has increased in Albania (the highest among Western Balkan countries), but remains low compared to such access in more advanced European comparators. • Labor productivity is the lowest in the region and is further threatened by Albania’s aging population and increasing outmigration. • Women, youth, and those with little education are most excluded from jobs. Family responsibilities and the lack of hope of finding a job are two of the greatest barriers keeping women from actively participating in the workforce. • Most of the employed do not work for a wage, but for themselves, meaning that they do not benefit from social security, paid annual leave, or paid sick leave. Women in particular are disproportionately represented among those who are unpaid family workers. • New jobs created between 2014 and 2016 were, on average, less productive than existing jobs, mostly driven by self-employment and wage employment in less productive jobs. Recent job creation has largely benefited adults and low-medium-skilled workers, with fewer opportunities for those with higher levels of education. 32 IV. Conclusions 52. This note presents an analysis of both the demand and supply side of jobs dynamics in the Republic of Albania. The paper presents evidence of the current jobs outcomes in Albania. Among these outcomes include: the lack of job creation among Albania’s most productive firms; low job productivity; increasing informality; and poor job inclusiveness for youth, women, and the poor. 53. Drawing on the evidence presented, this note contributes to the discussion on the creation of a jobs action plan in Albania. This section identifies a few priorities for policy action organized around demand-side and supply-side issues. These policy actions are by no means conclusive as they do not capture reforms in the business environment. Nonetheless, they are an important first step to improving jobs outcomes in Albania. Demand-side recommended policy actions 54. Upgrade technology and improve financing policies for micro and small firms. Firms with less than 10 employees are less productive than medium and large firms and seem to face greater constraints to expansion and survival. Applying a more targeted approach to upgrading firms’ existing technology can improve competitiveness, innovation, and connectivity to local and foreign value chains. In many cases, local business conditions may differ greatly from the national average (including practices of the informal sector, corruption, and access to finance). Firms may also need direct advisory services and technological support, especially in areas outside the main commercial and financial centers in Tirana and Durres. 55. Design sector-specific policies that support “high -skill” job creation. For sectors with higher potential for “high-skill” job creation, in particular, business services and industry, the provision of fiscal incentives for research and development, training, and exports, and better access to financing capital for small- and medium-sized firms can potentially ease existing constraints in these sectors. 56. Foster productive self-employment and entrepreneurship. The share of nonwage workers (own- account, employers, and unpaid workers) has been increasing, representing 58 percent of employment in 2016. Compared with larger firms, the self-employed have less access to information on training (including on assessment of training needs) and face significant opportunity costs in attending affordable training (because they need to work at the same time). To address these issues and ultimately increase productivity growth among the self-employed, it is critical to upgrade their skills and business practices, as well as broaden their access to markets and value chains. Possible interventions include: (i) providing government-sponsored business development services and training to entrepreneurs; (ii) expanding access to standards/certification systems; and (iii) reforming the tax and regulatory system to reduce incentives for informal work and increase awareness of the benefits of formalization. 57. Improve the productivity of agricultural workers and employers. To date, agriculture remains the largest employer in Albania, but the sector’s productivity remains low. This is compounded by a lack of dynamic private sector investment in regions of the country that are largely rural. As a result, making agricultural jobs more productive will hinge on commercialization, both by helping smallholder farmers enter value chains and by facilitating agribusiness investments — themselves a source of wage jobs and potential positive spillover effects in the associated logistics services. Agribusiness is still currently underdeveloped. Its potential could be untapped by upgrading technology, including digital technology, 33 to increase the productivity of agriculture. However, investments in technology needs to be complemented by strong investments in physical infrastructure (permitting the transfer of knowledge through the Internet, and the physical distribution of produce, for example), as well as skills development (including both basic education and training) that can help farmers appreciate, understand, and use technology appropriately. Supply-side recommended policy actions 58. Facilitate the transition from school to work. The share of young Albanians who are neither in school nor employment nor training (NEET) ranges between 30–40 percent, depending on age group and gender. These shares of youth are about twice as high as the EU average. A young person who does not find a job after school risks remaining without one. Increasing the role of the private sector, through internship and apprenticeships, is an important means of facilitating the transition from school to work. The apprenticeship model piloted by the Swiss is one example that could be scaled up, and/or other forms of internships could be promoted, especially for first-time labor market entrants.7 59. Enhance the provision of childcare services and enact more flexible labor regulation to reduce the gender gap in labor force participation. Albania could do more to provide care support so that women are better able to balance family responsibilities and work. Interventions could include: (i) relaxing women’s competing time demands from family duties by financing an increased supply of affordable early childcare services; and (ii) making maternity benefits8 more flexible to allow them to be split with fathers so that mothers can go back to work earlier. Additional research is needed in this area to unpack which family duties are keeping women from entering the labor force. Finally, additional policies to be considered include requiring employers to provide paid leave to take care of sick relatives and making childcare payments tax deductible for both men and women; these have been introduced in high-income countries and the available evidence seems to show that they improve women’s work–life balance. 60. Increase Albania’s potential workforce. As Albania’s population is shrinking, policies to bring fertility back to replacement level to increase the working-age population in the long run could be considered. Birth grants are adopted by several countries, both universal and targeted, though the evidence on their impact on fertility is mixed. 61. Strengthen the capacity of the NES and its labor market information system (LMIS). Data monitoring and management need to be enhanced. To date, Albania has a web-based LMIS, but there is no evidence on the extent to which it is used by students and jobseekers, and it does not capture the active jobseeking behavior of the registered unemployed. An improved LMIS (including advanced statistical profiling techniques) and increased capacity of the NES (including better counseling, career guidance, and job matching with employers) would contribute to increasing the NES’s cost-efficiency and to directing its scarce resources more effectively. The NES’s employment promotion (EP) programs could be reformed and redesigned to better target vulnerable groups that are more likely to be jobless (youth, women, the lower-educated). To the authors’ knowledge, no rigorous impact evaluation has been conducted on any of the seven EP programs administered by the NES; evidence based on monitoring data 7 Generally, reforms are needed in tertiary education and in the vocational education and training (VET) system to equip students and workers with the skills demanded by employers, especially in the context of changing content/type of skills needed due to technological progress (see Honorati and Johansson 2018 forthcoming). 8 It should be noted that Albania has one of the most generous maternity leave benefits in the world. 34 shows that these programs could be better targeted to vulnerable jobseekers and to firms that are more in need of job search assistance. However, no information is available to track placement rates by social and demographic profiles of beneficiaries or to understand which program worked best for whom in terms of job placement. A standardized profiling approach, as used in many OECD and EU countries, could be introduced to improve the targeting and cost-efficiency of the NES’s EP programs and job intermediation services. 62. Maximize the benefits of emigration. Luring qualified people to return to Albania to contribute to “brain gain” has proven to be more easily said than done. Return decisions are complex and skills acquired abroad may be less transferable, thus causing a loss in human capital among return. In the short run, more value could be added by creating virtual professional networks to allow for transfer of know- how. Additional policy options are to allow transferable pensions and to develop straightforward, transparent migration regulations. Finally, for those emigrants who do return, Albania can establish policies that better facilitate their reinsertion into the labor market. 35 Annex I: Structural Business Statistics Data and Definitions The analysis is based on the panel of Albanian firms in 2002–2015 that was constructed by appending the annual data from the Structural Business Statistics (SBS). The SBS comprises active enterprises in Albania of all legal forms. The population consists of all enterprises that, according to the Statistical Business Register, were active in December of the reference year. Enterprises with 1–9 employees are surveyed by sample survey. Enterprises with 10 and more employees are surveyed exhaustively. The data are collected directly from enterprises with interviewers. The printed questionnaire is filled in at the moment of interview in enterprises (INSTAT 2017). A unit of observation is an enterprise defined by INSTAT as “the smallest combination of legal units that is an organizational unit producing goods or services which benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current resources. An enterprise carries out one or more activities at one or more locations (local unit).” The relationship between an enterprise and a legal unit is therefore defined as: “the enterprise corresponds either to a legal unit or to a combination of legal units.” The classification of economic activities is done according to the European industrial activity classification NACE. The original data have a break in series in the year 2010 because of the implementation of NACE Rev. 2 and changes in the coverage of SBS. From 2010 Fishing (code 05 of NACE Rev. 1.1) and Activities of membership organizations n.e.c. (code 91 of NACE Rev. 1.1) were excluded whereas private health and education (codes 85, 86, 87, and 88 of NACE Rev. 2) were added. For years 2002–2009 NACE Rev.1.1 is reported; for years 2010–2014 there is a double coding in NACE Rev 1.1 and NACE Rev. 2; and in 2015 only NACE Rev.2 is reported. In the panel data for 2002–2015, a 3-digit NACE Rev. 2 code was attached to each firm-year using: 1) The firm’s NACE Rev. 2 code reported in 2010–2015, if the firm appeared in the sample in both periods 2002–2009 and 2010–2015; 2) NACE Rev. 2 code, which is the mode among all firms with a given NACE Rev. 1.1 code in the period 2010–2014, if the firm appeared in the sample before 2010 when NACE Rev.2 was implemented. Then eight sectors were created using the following correspondence with 3-digit NACE Rev.2 codes (Table AI.1). Table AI.1: Economic sectors classification used in the analysis Sector used in the analysis 3-digit NACE Rev. 2 code Agriculture and agroprocessing 14-32, 101-110 Industry excl. manufacture of food and beverages 51-99, 120, 131-390 Construction 411-439 Trade and repair 451-479 Tourism 493, 551-563, 791-799 ICT 611-639 Business services 581-602, 641-783, 801-829 Other services Remaining codes 36 Region refers to 12 prefectures. This variable was created on the basis of district codes from 1 to 36 (variable location). In the original SBS data for 2013, the location variable has zero values for all firms. Using a firm’s location reported in earlier or later years, its location in 2013 was recovered for the firms observed in the sample at least for two years. However, there are many missing values for one-year firms included only in the 2013 sample. Firm’s size group is constructed on the basis of adjusted employment; i.e., annual average number of employees provided in the dataset plus one proprietor. The “annual average number of employees” provided in the SBS does not include proprietors, but the “adjusted employment” variable does (by construction it is equal to “annual average number of employees” provided +1). Firms with zero employees (sole proprietors) represent one-third of the sample on average (Table AI.2). Table AI.2: Distribution of firms by “annual average number of employees,” unweighted Year Emp=0 Emp>0 Total 2002 2,414 3,722 6,136 2003 1,662 3,637 5,299 2004 2,616 4,562 7,178 2005 2,303 4,760 7,063 2006 1,950 4,869 6,819 2007 1,818 4,792 6,610 2008 2,335 4,504 6,839 2009 2,583 4,972 7,555 2010 4,125 5,854 9,979 2011 3,773 5,795 9,568 2012 3,351 6,848 10,199 2013 2,904 7,326 10,230 2014 2,812 7,408 10,220 2015 4,234 8,790 13,024 Source: SBS, INSTAT. Firm’s age is based on the date of enterprise’s creation (the original source of data is the Statistical Business Register). If creation date was missing in some year(s), it was recovered with the use of firm’s creation date reported in earlier or later years. Ownership. The SBS data received for the analysis do not include any ownership variable, so it is not possible to define state-owned enterprises (SOEs) and foreign-owned firms. Labor productivity is defined as turnover per worker. Turnover comprises the total amount invoiced by the observation unit during the reference period, and this corresponds to market sales of goods or services supplied to third parties (INSTAT 2017). Employment used in the denominator refers to adjusted employment; i.e., annual average number of employees plus one proprietor. Based on the distribution of firms by turnover per worker in each year, firms are divided into five productivity quintiles, where one is the lowest productivity quintile and five is the highest. Value added at basic prices is calculated in the data as difference between production value and intermediate consumption. It was not used for this analysis but can used for further analysis. 37 Investments during the reference period include goods, whether bought from third parties or produced for own use, having a useful life of more than one year, including nonproduced tangible goods such as land. Wages and salaries per employees corresponds to the annual average wages and salaries paid from enterprise per employees. The job flow concepts follow the definitions of Davis and Haltiwanger (1992, 1999): Gross job creation in subcategory s in year t equals the sum of all employment gains in firms in subcategory s that expand or start up between t and t-1: JC st   (emp est emp es (t 1) ) , e S  where empest denotes the average number of employees in firm e in subcategory s in year t,9and S+ stands for the set of all expanding firms in the relevant subcategory. Likewise, gross job destruction in subcategory s in year t equals the sum of all employment losses in firms in subcategory s that contract or shut down between t and t-1: JD st   emp est  emp es (t 1) , e S  where S– stands for the set of all contracting firms in the corresponding subcategory. The sum of these two measures yields a measure for gross job reallocation (GROSS) and their difference gives the net employment growth (NET): GROSS st  JC st  JD st , NET st  JC st  JD st . To capture the amount of “churning” by firms, i.e., job reallocation in excess of the amount required to accommodate net employment change, a measure of excess job reallocation, equal to the difference between the gross job reallocation and the absolute value of the net employment growth, is widely used: EXCESS st  GROSS st  NETst . All of these job flows are converted into rates by dividing by average employment across the two years. For example, the job creation rate can be written as JC st JCR st  * 100%, X st 9 For the analysis of job flows, average number of employees provided in the data is used as a measure of employment. 38 where X st   (emp est e S emp es (t 1) ) / 2 , i.e., total employment of all firms in a subcategory averaged over the two years. All job flow measures are calculated for the firms that have nonmissing employment in at least two consecutive years. Gazelles (i.e., the fastest-growing firms in terms of created job opportunities) are defined on the basis of distribution of all expanding firms by year-to-year job-creation rate. Here, the job-creation rate is calculated as employment gains between t and t-1 divided by employment in year t (if employment was zero, 1 was added to both current and previous employment). If the average employment over the two years was used in the denominator of the job-creation rate, many years would have 0 firms defined in the top 5 percent or 10 percent of firms in terms of the job-creation rate. To calculate entry and exit rates, all firms in the pooled sample for 2002–2015 were first divided into four types:10 1) Entering firms: Firms entering a register (i.e., appearing for the first time) in a given year, excluding one-year firms. Alternative definition of entering firms is based on the year of creation; 2) Exiting firms: Firms that are observed in the register over 2002–2015 for the last time, excluding one-year firms; 3) One-year firms: Firms present in the register for only one year; 4) Continuing firms: Firms that were in the register in a given year, as well as in the previous and subsequent year. For firms that appear in the register with gaps (e.g., in 2002, 2005, 2006, 2015), only one entry (in the first year) and one exit (in the very last year) is assumed. The entry rate is defined as the number of entering firms divided by the total number of firms excluding one-year firms in a given year. The exit rate is defined as the number of exiting firms divided by the total number of firms excluding one- year firms in a given year. Firm survival is defined as the number of continuing and exiting firms by birth-year (excluding one-year firms). Given the fixed time-span, both left and right censoring occurs. The data quality was verified and cross-checked with INSTAT publications on the full business registry. There are no issues of missing data with respect to the “adjusted employment“ variable used to generate job growth rates. The variable turnover or sales also does not have missing values (but has zero values in 3012 observations during 2002–2015). 10 Adapted definitions from Bartelsman, Scarpetta, and Schivardi (2005). 39 Annex II: Results from Firm-Level Regression Analysis Table AII.1: Dependent variable: Annual employment growth rate (number of jobs created/destroyed divided by the average firm size in 2 consecutive years), 2003–2015 Model 1 Model 2 Pooled OLS RE Pooled OLS RE Log(size) 0.032*** 0.043*** 0.028*** 0.037*** (0.003) (0.005) (0.003) (0.005) Firm’s age -0.006*** -0.010*** -0.006*** -0.011*** (0.001) (0.001) (0.001) (0.001) Log(real turnover per -0.030*** -0.086*** worker) (0.003) (0.005) Industry excl. -0.021 0.013 -0.022 0.022 manufacture of food (0.014) (0.030) (0.015) (0.031) and beverages Construction -0.070*** -0.069** -0.048*** -0.015 (0.014) (0.030) (0.014) (0.030) Trade and repair 0.009 0.021 0.047*** 0.112*** (0.015) (0.030) (0.016) (0.031) Tourism 0.004 0.005 -0.016 -0.033 (0.017) (0.033) (0.018) (0.034) ICT 0.068** 0.066 0.069** 0.072 (0.033) (0.056) (0.033) (0.056) Business services -0.026 0.004 -0.046*** -0.015 (0.018) (0.034) (0.018) (0.034) Other services 0.003 0.039 -0.014 0.003 (0.018) (0.034) (0.018) (0.034) Diber 0.040* 0.001 0.051** 0.038 (0.022) (0.042) (0.024) (0.043) Durres 0.007 0.013 0.021 0.037 (0.016) (0.031) (0.017) (0.031) Elbasan -0.044** -0.069* -0.033* -0.043 (0.018) (0.035) (0.019) (0.036) Fier 0.004 0.015 0.020 0.059* (0.017) (0.033) (0.018) (0.034) Gjirokaster -0.022 -0.020 -0.021 -0.005 (0.019) (0.038) (0.020) (0.038) Korce -0.044** -0.045 -0.038** -0.033 (0.018) (0.037) (0.019) (0.037) Kukes 0.067** 0.100* 0.065** 0.090* (0.029) (0.052) (0.029) (0.053) Lezhe 0.033 0.043 0.033 0.054 (0.021) (0.042) (0.021) (0.043) Shkoder 0.001 0.006 0.006 0.018 (0.019) (0.037) (0.019) (0.037) Tirana -0.016 -0.019 0.010 0.035 (0.015) (0.029) (0.016) (0.030) Vlore -0.022 -0.006 -0.006 0.034 (0.018) (0.034) (0.018) (0.034) 2003 . -0.013 -0.106*** -0.020 40 . (0.018) (0.022) (0.018) 2004 0.109*** 0.066*** . 0.051*** (0.022) (0.018) . (0.018) 2005 0.035* -0.024 -0.071*** -0.034** (0.019) (0.016) (0.020) (0.016) 2006 0.038** -0.026 -0.066*** -0.032** (0.019) (0.016) (0.019) (0.016) 2007 0.099*** 0.035** -0.004 0.027* (0.019) (0.015) (0.019) (0.015) 2008 0.042** -0.019 -0.057*** -0.024 (0.019) (0.015) (0.019) (0.015) 2009 -0.008 -0.078*** -0.103*** -0.081*** (0.019) (0.015) (0.019) (0.015) 2010 0.064*** -0.010 -0.033* -0.018 (0.019) (0.014) (0.019) (0.014) 2011 0.023 -0.055*** -0.049*** -0.038*** (0.019) (0.014) (0.019) (0.013) 2012 0.036** -0.060*** -0.065*** -0.071*** (0.018) (0.013) (0.018) (0.013) 2013 0.037** -0.054*** -0.065*** -0.070*** (0.018) (0.012) (0.018) (0.012) 2014 0.066*** -0.040*** -0.021 -0.042*** (0.018) (0.012) (0.018) (0.012) 2015 0.123*** . 0.033* . (0.018) . (0.018) . Constant -0.017 0.042 0.094*** 0.061 (0.026) (0.044) (0.026) (0.045) N 44309 44309 43706 43706 R2 0.010 0.013 AIC 90238.424 . 87681.952 . BIC 90525.490 . 87977.250 . Note: Robust standard errors (clustered at firm’s id) in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. Model 2 includes potentially endogenous variables - Log(real turnover per worker). Breusch and Pagan Lagrangian multiplier test for random effects shows that RE model is preferred for both specifications. Agriculture and agroprocessing is the reference sector. Berat is the reference prefecture. Annual GDP deflator (retrieved from WDI in October 2017) is used to get real turnover per worker. 41 Table AII.2: Dependent variable: Logarithm of real turnover per worker, 2003 –2015 Model 1 Model 2 Pooled OLS RE Pooled OLS RE Log(size) 0.043*** 0.013 0.046*** 0.017* (0.010) (0.009) (0.010) (0.009) Firm’s age -0.003 -0.008*** -0.003 -0.010*** (0.002) (0.002) (0.002) (0.002) Employment growth -0.097*** -0.152*** rate (0.008) (0.007) Industry excl. -0.255*** 0.011 -0.256*** 0.020 manufacture of food (0.062) (0.052) (0.062) (0.052) and beverages Construction 0.392*** 0.447*** 0.386*** 0.442*** (0.060) (0.053) (0.060) (0.054) Trade and repair 1.362*** 0.955*** 1.363*** 0.954*** (0.061) (0.051) (0.062) (0.051) Tourism -0.496*** -0.340*** -0.496*** -0.339*** (0.060) (0.052) (0.060) (0.052) ICT 0.269** 0.133* 0.275** 0.145* (0.108) (0.075) (0.107) (0.076) Business services -0.643*** -0.175*** -0.646*** -0.167*** (0.070) (0.062) (0.070) (0.063) Other services -0.319*** -0.223*** -0.320*** -0.214*** (0.068) (0.055) (0.068) (0.056) Diber 0.346*** 0.325*** 0.350*** 0.326*** (0.094) (0.081) (0.094) (0.082) Durres 0.308*** 0.180*** 0.310*** 0.182*** (0.068) (0.055) (0.068) (0.055) Elbasan 0.180** 0.162** 0.176** 0.154** (0.077) (0.067) (0.077) (0.067) Fier 0.364*** 0.401*** 0.365*** 0.406*** (0.074) (0.061) (0.074) (0.061) Gjirokaster 0.064 0.185** 0.062 0.183** (0.089) (0.072) (0.089) (0.073) Korce 0.030 0.085 0.026 0.079 (0.074) (0.063) (0.074) (0.063) Kukes -0.184* -0.226*** -0.178* -0.210** (0.098) (0.087) (0.099) (0.088) Lezhe 0.096 0.185** 0.099 0.193** (0.083) (0.079) (0.083) (0.080) Shkoder 0.036 0.050 0.037 0.053 (0.077) (0.064) (0.077) (0.064) Tirana 0.564*** 0.455*** 0.563*** 0.452*** (0.064) (0.052) (0.064) (0.052) Vlore 0.287*** 0.322*** 0.285*** 0.324*** (0.070) (0.062) (0.070) (0.063) 2003 0.076*** 0.183*** 0.066*** 0.184*** (0.024) (0.028) (0.023) (0.029) 2004 . 0.107*** . 0.118*** . (0.027) . (0.027) 2005 0.100*** 0.139*** 0.093*** 0.135*** (0.022) (0.025) (0.022) (0.025) 42 2006 0.136*** 0.175*** 0.129*** 0.171*** (0.024) (0.024) (0.024) (0.024) 2007 0.143*** 0.153*** 0.142*** 0.159*** (0.026) (0.023) (0.026) (0.023) 2008 0.166*** 0.156*** 0.160*** 0.154*** (0.027) (0.022) (0.027) (0.022) 2009 0.176*** 0.131*** 0.166*** 0.120*** (0.028) (0.021) (0.028) (0.021) 2010 0.152*** 0.083*** 0.149*** 0.082*** (0.031) (0.020) (0.031) (0.020) 2011 0.093*** 0.078*** 0.088*** 0.074*** (0.031) (0.018) (0.031) (0.018) 2012 0.018 0.000 0.011 -0.008 (0.031) (0.017) (0.031) (0.017) 2013 -0.013 -0.036** -0.020 -0.044*** (0.031) (0.016) (0.031) (0.016) 2014 -0.001 -0.063*** -0.003 -0.067*** (0.031) (0.013) (0.031) (0.013) 2015 0.088*** . 0.091*** . (0.032) . (0.032) . Constant -0.066 -0.063 -0.056 -0.056 (0.081) (0.071) (0.081) (0.071) N 43706 43706 43706 43706 R2 0.282 0.284 AIC 138573.805 . 138446.828 . BIC 138860.418 . 138742.126 . Note: Robust standard errors (clustered at firm’s id) in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. Model 2 includes potentially endogenous variables – empgrowth. Breusch and Pagan Lagrangian multiplier test for random effects shows that RE model is preferred for both specifications. Agriculture and agroprocessing is the reference sector. Berat is the reference prefecture. Annual GDP deflator (retrieved from WDI in October 2017) is used to get real turnover per worker. 43 References Alam, Anós Casero, Khan, and Udomsaph. 2008. “Unleashing Prosperity: Productivity Growth in Eastern Europe and the Former Soviet Union.” Washington, DC: World Bank. Arias, Sanchez-Paramo, Davalos, Santos, Tiongson, Gruen, de Andrade Falcao, Saiovici, and Cancho. 2014. “Back to Work: Growing with Jobs in Europe and Central Asia.” Washington, DC: World Bank. Bartelsman E., S. Scarpetta, and F. Schivardi. 2005. "Comparative Analysis of Firm Demographics and Survival: Micro-level Evidence for the OECD Countries.” Birch, and Medoff. 1994. “Gazelles.” In Labor Markets, Employment Policy and Job Creation , ed. Solmon and Levenson, pp. 159-167. Boulder, CO and London: Westview Press. Criscuolo, Gal, and Menon. 2014. “The Dynamics of Employment Growth: New Evidence from 18 Countries.” OECD Science, Technology and Industry Policy Papers, 14, Paris, OECD Publishing. Davis, Steven J, and J. Haltiwanger. 1992. “Gross Job Creation, Gross Job Destruction, and Employment Reallocation”. The Quarterly Journal of Economics, Vol. 107, No. 3 (Aug., 1992), pp. 819-863 Published by: The MIT Press. Davis, Steven J. and John Haltiwanger. 1999. "Sectoral Job Creation and Destruction Responses To Oil Prices Changes”. Journal of Monetary Economics, 2001, v48(3,Dec), 465-512. Honorati, and Johansson. 2018. “Demand for Skills in Albania: An Analysis of the Skills Toward Employment and Productivity Survey”, the World Bank, Washington, DC. International Monetary Fund (IMF). 2016. “Emigration and Its Economic Impact on Eastern Europe.” IMF Staff Discussion Notes, SDN 16/07, IMF, Washington, DC. IMF. 2017. “Albania: Staff Report for the 2017 Article IV Consultation. Key Issues.” IMF, Washington, DC. Institute of Statistics (INSTAT). 2017. “Results of Structural Survey Of Economic Enterprises 2015.” O’Brien, Tim, Ljubica Nedelkoska, and Ermal Frasher. 2017. “What is the Binding Constraint to Growth in Albania?” Center for International Development at Harvard University. Boston, Massachusetts. Republic of Albania. 2017. “National Employment and Skills Strategy 2014-2020: Annual Progress Report 2016.” Tirana, Albania. UN Statistics (2018), “International migrant stock: the 2017 revision,”. http://www.un.org/en/development/desa/population/migration/data/estimates2/estimates17.shtml. World Bank. 2012. “Jobs.” World Development Report. Washington, DC. World Bank. 2015. “Country Partnership Strategy for Albania: 2015-2019.” Report No. 98254. World Bank, Washington, DC. 44