Indonesian Economic Transformation and Employment Policy input for an Indonesia Jobs Strategy 1 Report No: AUS13186 . Republic of Indonesia Economic Transformation and Employment Policy input for an Indonesia Jobs Strategy June 2016 GTC02 EAST ASIA AND PACIFIC Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. 2 This report is part of the World Bank’s support to the Government on an Indonesia Jobs Strategy. This report is prepared by Claire H. Hollweg, Maria M. Wihardja, Massimiliano Calì, Milan Nedeljkovic, Julian L. Clarke, and Brasukra G. Sudjana. It also includes data and inputs by Ali Zafar, Arif Khan, Ahsan Tariq Butt, Agnesia A. Hasmand, Daim Syukriah, Bagus Arya Wirapati, and Michaelino Mervisiano. It also benefited from inputs and guidance from Vivi Alatas (Lead Economist, Poverty GP), Truman Packard (Lead Economist, Social Protection & Labor), Ndiame Diop (Lead Economist, Macro Economics & Fiscal Management), Mona Haddad (Practice Manager, Trade & Competitiveness), and Tatiana Nenova (Program Leader, World Bank Office Jakarta), as well as comments from peer reviewers: Elizabeth Ruppert Bulmer (Lead Economist, Jobs CCSA) and Marc Tobias Schiffbauer (Senior Economist, Macro Economics and Fiscal Management). The team acknowledges the inputs provided by a number of government officials, private sector businesses, and labor union activists. 3 Contents EXECUTIVE SUMMARY .............................................................................................................................. 6 1. INTRODUCTION: INDONESIA’S ECONOMIC TRANSFORMATION........................................................................... 8 THE JOBS OBJECTIVE: SHIFTING TOWARDS QUALITY AND PRODUCTIVE JOBS ........................................................... 8 INDONESIA’S COMMODITY BOOM… ............................................................................................................. 8 …CREATED A TWENTY-FIRST CENTURY DUTCH DISEASE… .................................................................................. 9 …WITH ADVERSE IMPACT ON JOBS............................................................................................................. 10 IS THERE AN OPPORTUNITY FOR ANOTHER STRUCTURAL TRANSFORMATION? ....................................................... 11 2. LABOR TRANSITION TRENDS: RESULTS FROM THE IFLS................................................................................... 13 SERVICES SECTORS CONTINUED TO ABSORB LARGE SHARES OF WORKERS ............................................................ 13 LARGE COUNTRY, LOW MOBILITY ............................................................................................................... 15 HAVING THE RIGHT SKILLS AND INITIAL ENTRY INTO FORMAL/INFORMAL EMPLOYMENT MATTER .............................. 16 ADVANTAGE: YOUNG AND MALE ............................................................................................................... 17 3. DRIVERS OF LABOR TRANSITION: RESULTS FROM LABOR MOBILITY COST AND EMPLOYMENT ELASTICITY ANALYSES .... 18 LABOR MOBILITY COSTS IN INDONESIA ARE RELATIVELY HIGH............................................................................ 19 …BUT, HAVE BEEN DECLINING ACROSS SECTORS AND REGIONS ......................................................................... 21 THE RIGHT SKILLS MIX AND INFORMALITY REMAIN IMPORTANT ......................................................................... 22 DO REAL WAGES MATTER? ....................................................................................................................... 25 EVIDENCE FROM EMPLOYMENT ELASTICITY .................................................................................................. 26 4. LABOR CONTENT OF EXPORTS: POTENTIAL SOURCES OF DEMAND FOR LABOR? ................................................... 34 DO EXPORTS SUPPORT EMPLOYMENT?........................................................................................................ 34 WHICH SUB-SECTORS CONTRIBUTE SIGNIFICANTLY TO WAGES AND EMPLOYMENT?............................................... 35 5. POLICY IMPLICATIONS............................................................................................................................. 38 ANNEX 1 – LABOR MOBILITY COST METHODOLOGY .......................................................................................... 42 METHODOLOGY..................................................................................................................................... 42 DATA .................................................................................................................................................. 44 ANNEX 2 – LABOR CONTENT OF EXPORTS METHODOLOGY ................................................................................. 46 4 Figures FIGURE 1 – SECTORAL CONTRIBUTION TO GDP (%) ............................................................................................................. 8 FIGURE 2 – SECTORAL CONTRIBUTION TO EMPLOYMENT (%).................................................................................................. 8 FIGURE 3 – SHARES OF TOTAL EXPORTS, 1989-2013 .......................................................................................................... 9 FIGURE 4 – INDONESIA CPI RER VIS-À-VIS USD, JAN 2002-DEC 2015 ................................................................................. 10 FIGURE 5 – RER DECOMPOSITION - ANNUAL CONTRIBUTIONS TO INDONESIAN RER CPI CHANGE (VS. USD): TRADABLE PRICES BASED ON EXPORT AND IMPORT INDICES, 2002-2015............................................................................................................. 11 FIGURE 6 – RER DECOMPOSITION – ANNUAL CONTRIBUTIONS TO INDONESIAN RER CHANGE (VS. USD): FOOD VS. NON FOOD PRICES, 2002-2015 ................................................................................................................................................... 12 FIGURE 7 – LABOR SHARES (%), AGED 15 AND ABOVE, ACROSS SECTORS AND REGIONS, 1997-2007 ............................................ 13 FIGURE 8 – LABOR MOBILITY COSTS VS. GDP PER CAPITA IN INDONESIA AND OTHER TPP COUNTRIES ............................................. 20 FIGURE 9 – LABOR MOBILITY COSTS, WAGES, AND EMPLOYMENT, ACROSS SECTORS AND ACROSS REGIONS, 1997-2007..................... 22 FIGURE 10 – LABOR MOBILITY COSTS ACROSS SECTORS BY WORKER TYPES, 2000-2007 ............................................................. 24 FIGURE 11 – LABOR MOBILITY COSTS ACROSS SECTORS BY JOB TYPE, 2000-2007 ..................................................................... 25 FIGURE 12 – REAL WAGES ACROSS SECTORS, 1997-2007 .................................................................................................. 25 FIGURE 13 – REAL WAGES ACROSS REGIONS, 1997-2007 .................................................................................................. 26 FIGURE 14 – EMPLOYMENT BY GENDER AND BY 3-SECTOR OF EMPLOYMENT, 1990-2015 .......................................................... 30 FIGURE 15 – EMPLOYMENT BY AGE GROUPS AND 3-SECTOR OF EMPLOYMENT, 1990-2015........................................................ 31 FIGURE 16 – EMPLOYMENT BY FORMALITY AND 3-SECTOR, 1990-2015 ................................................................................ 33 FIGURE 17 – DIRECT AND TOTAL LABOR VALUE ADDED OF EXPORTS, 1995-2011 ..................................................................... 34 FIGURE 18 – TOTAL LABOR VALUE ADDED OF EXPORT SHARE, SELECT COUNTRIES, 1995-2011..................................................... 34 FIGURE 19 – TOTAL NUMBER OF JOBS (DIRECT AND INDIRECT) SUPPORTED BY EXPORTS (IN ‘000), 2007 ........................................ 35 FIGURE 20 – DIRECT AND INDIRECT LABOR VALUE ADDED OF EXPORTS, 2011........................................................................... 36 FIGURE 21 – DIRECT AND INDIRECT LABOR VALUE ADDED OF EXPORTS (FORWARD LINKAGES), 2011 .............................................. 36 FIGURE 22 – NUMBER OF JOBS IN EXPORTS ACROSS MORE REFINED SECTORS, 1997 VS. 2011 ..................................................... 37 Tables TABLE 1 – TRANSITIONS ACROSS SECTORS AND INTO AND OUT OF LABOR FORCE STATUS (%), 2000-2007 ...................................... 15 TABLE 2 – TRANSITIONS ACROSS SECTORS AND INTO AND OUT OF LABOR FORCE STATUS (%), 1997-2000 ...................................... 15 TABLE 3 – TRANSITIONS ACROSS REGIONS (%), 2000-2007 ............................................................................................... 16 TABLE 4 – TRANSITIONS ACROSS AGGREGATE SECTORS BY SKILL LEVEL (%), 2000-2007 ............................................................. 16 TABLE 5 – TRANSITIONS ACROSS AGGREGATE SECTORS BY FORMALITY (%), 2000-2007 ............................................................. 17 TABLE 6 – SHARE OF EMPLOYMENT (%), 1990-2015 ....................................................................................................... 26 TABLE 7 – SECTORAL EMPLOYMENT ELASTICITY, 1993-2006 VS. 2007-2015......................................................................... 28 TABLE 8 – LABOR PRODUCTIVITY GROWTH RATE/GDP GROWTH RATE, 1993-2006 VS. 2007-2015 ............................................ 28 TABLE 9 – SECTORAL EMPLOYMENT ELASTICITY, BY GENDER, 1993-2006 VS. 2007-2015 ......................................................... 29 TABLE 10 – SECTORAL EMPLOYMENT ELASTICITY, BY AGE GROUP, 1993-2006 VS. 2007-2015 .................................................. 31 TABLE 11 – SECTORAL EMPLOYMENT ELASTICITY, BY FORMALITY/INFORMALITY OF JOBS, 1993-2006 VS. 2007-2015...................... 32 TABLE 12 – REGIONAL EMPLOYMENT ELASTICITY, 1993-2006 VS. 2007-2015 ...................................................................... 33 Boxes BOX 1 – DEFINING LABOR MOBILITY COSTS ...................................................................................................................... 18 BOX 2 – EMPLOYMENT ELASTICITY METHODOLOGY NOTE .................................................................................................... 27 5 Executive summary The commodities boom in the early 2000s reversed the expansion of Indonesia’s industrialization that had begun in the 1980s. The commodities boom led to an increase in the export share of agriculture; mining and mineral commodities expanded at the expense of manufactured exports and services. This was accompanied by an appreciation of the real exchange rate (RER), which reduced the relative competitiveness of Indonesia’s tradeable sectors and contributed to sluggish performance in non- commodity exports. This reversal in economic transformation had an adverse impact on jobs. Most jobs created during the commodities boom were in trade and retail as well as social and personal services. These sectors exhibit low productivity jobs. Today, around 70 percent of workers are employed in agriculture and these low- end services. Meanwhile, increasing labor productivity is crucial to boosting growth (World Bank, 2014). The end of the commodities boom and recent depreciation in the nominal value of the Rupiah suggest the possibility that Indonesia could retrace the path back to manufacturing and high productivity activities. Stickiness in the depreciation of the Rupiah relative to other commodity exporters, however, implies that this transformation will only be possible if accompanied by policy reforms that target local retail prices and food price inflation. Domestic prices have been rising faster than border prices relative to US and Asian peers, due to increased protectionism in the form of higher tariff and non-tariff barriers as well as increased logistics and distribution costs. While a depreciation of the Rupiah will enable firms to regain competitiveness and thus create jobs, it will not be enough—labor mobility costs will also have to be reduced. Labor mobility costs reflect what a worker perceives to be his or her welfare cost of moving between industries to find alternative employment, and capture the various costs that explain why workers do not move into higher-wage sectors. To unleash the economic transformation that is waiting to happen, labor needs to be able to move from the sectors/firms with low productivity to those with higher productivity. This labor mobility is typically hindered by several factors; typical impediments to job switching are skills mismatches (wages forgone because of lower productivity), limitations to geographic mobility (administrative procedures for internal migration and direct relocation costs), and severance and hiring costs (including those imposed by laws or regulations). Other factors may be location preferences, job search costs, and even the psychological costs of changing jobs. Looking at how workers transition from one sector to the other sectors of the economy, there is greater fluidity in Indonesia’s labor market, but it is defined by transitions towards low productivity services. From 1997 and up until 2007, services sectors have been the largest absorbers of workers that changed sectors of employment, consistent with services being the most important for job creation. The labor absorption of the manufacturing sector continued to be limited in comparison. There has also been little movement of workers across regions. The right skills mix appears to be an important factor behind job matching in the Indonesian labor market. Over the past two decades labor mobility costs to move between manufacturing sectors were higher in Indonesia compared with the average of 47 countries worldwide. Workers in Indonesia also face higher welfare losses from job changes across manufacturing sectors than other countries in Asia-Pacific. Within Indonesia, the labor mobility cost to move into manufacturing is lower in than to move into other sectors (including social services and agriculture), except in trade, retail and accommodation. 6 The Government has been addressing the competitiveness of manufacturing firms by reducing import licenses, developing infrastructure and increasing the efficiency of ports. To the extent that these reforms lead to lower distribution costs in the future, Indonesia’s manufacturing competitiveness is likely to improve over time, as will the demand for labor in the manufacturing sector. Manufacturing could be one of the drivers of future employment, especially through exports, and outside services employment. In 2011, exports generated US$60 billion in direct (production of exported goods) and indirect (production of inputs for exported goods) wages in Indonesia. But the number of jobs generated by exports has been on the decline since 2001, in line with the commodity boom—it rose from 18 million in 1997 to 29 million in 2001, and is now back to 19 million. With the new potential for growth in manufacturing, a similar pattern of job creation from exports can be triggered—nearly 3 million jobs per year. There are also new manufacturing sub-sectors with higher sources of wages that are on the rise, including chemical, rubber, and plastic products. Besides manufacturing, services sectors are likely to remain the biggest source of employment. Between 2001 and 2011, 17 million out of 20 million new jobs were created in services sectors. Services sectors also provide inputs into manufacturing production, as well as exports. High-end services such as finance and transport and logistics, are becoming more important in supporting other sectors as well as employment. For example, out of the 16 million jobs supported by exports in manufacturing, three quarters are supported through backward linkages, which are services supplied to the manufacturing sector. In order to adjust to the end of the commodity boom and allow manufacturing and high-end services sectors with competitive potential to emerge, policy makers will need to facilitate labor mobility and reduce its cost. Given that manufacturing and high-end services sectors will boost economic growth and help the economy to structurally transform into high-productivity economy, policy makers will need to ensure that worker skills match the requirements of available jobs. Movement of workers into higher productivity jobs within a sector, in particular agriculture where youth are exiting the sector due to low wages relative to wages in other sectors of the economy, should also be facilitated. To accelerate the transition of labor, skills training and upgrading should be a priority. Assessing the labor regulatory framework to identify measures that reduce the costs of transitioning between sectors and between geographic areas will also be needed. 7 1. Introduction: Indonesia’s economic transformation The jobs objective: shifting towards quality and productive jobs The current downturn in commodity prices provides an opportunity for Indonesia to shift away from its dependence on commodity-driven growth towards higher value-added activities in manufacturing and services. However, Indonesia faces both global and structural challenges in making this transition. Global challenges include competition from regional trade agreements, especially the TPP, but also from structurally lower global trade growth. In addition, Indonesia’s manufacturing se ctors have also been losing competitiveness to regional competitors, while most job creation in the 2000s took place in low productivity sectors. This report aims to show the patterns of economic transformation in Indonesia in the past decade and a half, especially in terms of jobs and employment. The report highlights barriers to labor movement and macroeconomic sources of demand for labor. The report seeks to contribute to the design of a jobs strategy that emphasizes the transition of workers from low to high productivity sectors. While Indonesia has, so far, relied on job creation in low-productivity, and even vulnerable, employment, future challenges would require the country to shift to higher productivity and quality jobs. Indonesia’s commodity boom… Indonesia underwent a structural transformation from agriculture to industry prior to the Asian crisis in late 1990s. The contribution to GDP of agriculture fell sharply in the 1970s, while industry’s share, including manufacturing, rose sharply and continued into the 1980s and 1990s. However, in terms of employment, industrial share of workers peaked at around 20 percent since the mid-1990s. And, while services increasingly employ a larger share of workers, it wasn’t until 2008 that services sectors overt ook agriculture (Figure 1 and Figure 2). Figure 1 – Sectoral contribution to GDP (%) Figure 2 – Sectoral contribution to employment (%) 60 60 50 50 % of total employment 40 % of GDP 40 30 20 30 10 20 0 10 1964 1988 2012 1960 1968 1972 1976 1980 1984 1992 1996 2000 2004 2008 0 Agriculture Industry 1989 1991 1980 1985 1987 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 Manufacturing Services Agriculture Industry Services Source: World Development Indicators Source: World Development Indicators However, since recovering from the Asian economic crisis in the late 1990s , Indonesia reversed its pattern of industrialization, and increasingly relied on agricultural and mining commodities . By 2010, the share of agricultural commodities in total exports has increased to 20 percent, from 10 percent in 2000. The share of mining and mineral commodities has also increased from 8 percent in 2000 to 22 8 percent in 2010.1 In fact the total share of raw commodities exports overtook manufactured products in 2008 (Figure 3). Indonesia’s share of ASEAN manufactured exports has also fallen from 25 percent in 1980s to 10 percent by 2015.2 Figure 3 – Shares of total exports, 1989-2013 60 50 40 % of total export 30 20 10 0 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 Raw commodities Processed commodities Manufacturing Source: World Development Indicators …Created a twenty-first century Dutch disease… One of the most prominent effects of commodity booms is the appreciation of the Real Exchange Rate (RER), which decreases the relative competitiveness of the non-booming tradable sector. The experience of Indonesia during the first decade of 2000s is no exception, with the Indonesian Rupiah RER (vis-à-vis the US) appreciating over 80 percent between 2002 and 2011(Figure 4)3. This pattern has also been in line with other commodity exporting countries as Brazil and South Africa. 4 This RER appreciation played a major role in the relative decline of Indonesian non-commodity exports, manufacturing in particular, during that period. The RER appreciation has only been partly reversed during the dramatic decline in commodity prices over the past 4-5 years, and it has been more sticky downwards than other commodity exporting countries. This has contributed to the sluggish performance of the non-commodity exports, which have yet not bounced back to the pre-boom period. 1 World Bank, 2012. 2 Data source: WDI. 3 Calì and Nedeljkovic, 2016. 4 Calì and Nedeljkovic (2016) compares Indonesia’s RER to Brazil, South Africa, China, India, Malaysia, Philippines, Thailand, and Vietnam. 9 Figure 4 – Indonesia CPI RER vis-à-vis USD, Jan 2002-Dec 2015 240 220 200 180 160 142 140 120 100 80 60 jan.02 jan.09 dec.15 Source: World Bank staff’s calculations based on IMF and IFS data …With adverse impact on jobs Most of the jobs created during the commodity boom were in informal and casual work. Informality is defined as own-account work (self-employed with workers), self-employed with unpaid family workers or temporary workers; casual worker in agriculture or non-agriculture; and unpaid family work. Formality is defined as self-employed with permanent worker, government workers, or private workers. Out of the 20 million jobs created between 2001 and 2011, 82 percent of these were in non-tradable services. Most of the jobs created were in social and personal services, trade and retail, and construction. Manufacturing created 4 million jobs, while agriculture shed 860,000 jobs during the same period. This trend was also happening with new labor market entrants – the share of new workers who found employment in manufacturing declined from 25 percent in 1997 to 17 percent in 2007. Wholesale/retail trade and personal services’ share of young workers increased from 20 to 26 percent in the same period.5 This ‘wrong’ type of specialization has contributed to low growth of labor productivity and stagnant real wages. Between 1990 and 2009, Indonesian manufacturing experienced a 20 percent cumulative increase in average TFP index.6 However, labor productivity of manufacturing grew by 2.9 percent between 2001 and 2012; while labor productivities of transportation, trade, and agriculture grew by 21.5 percent, 4.8 percent, and 4.5 percent, respectively. During this time, real wages of both formal and informal workers were stagnant or declining for most sectors. 7 Indonesia’s real wages are also among the lowest in the region.8 This specialization has also been associated with declining returns to schooling over time, which helps to perpetuate the lack of incentives for younger cohorts to acquire skills. Coxhead (2014) found that, in 2007, the returns to schooling for younger workers (15-28 years old) in the formal sector were between 4 and 7 percent for each additional year of schooling, and between 1.5 and 3.9 percent in the informal 5 World Bank, 2014. 6 When further disaggregated, sectors such as machinery and instrument and textiles, clothing and footwear experienced higher TFP growth, while resource-based sectors experienced flat TFP growth (Fitriani et al., 2012). 7 Tadjoeddin (2016) presented real wages for self-employed, casual workers in agriculture and non-agriculture, and regular employees, based on Sakernas classifications. 8 Diop (2016) compares Indonesian real wages to China, Malaysia, the Philippines, Thailand, and Vietnam. 10 sector. These were lower than returns to schooling for older workers and lower than average returns in 1997 for all workers, which were between 7 and 9.3 percent. These declining returns to schooling over time helped to perpetuate the lack of incentives to acquire skills for younger cohorts. 9 Is there an opportunity for another structural transformation? Recent drops in commodity prices and nominal Rupiah depreciation seem to suggest that Indonesia has another shot at a structural transformation toward manufacturing or other higher productivity activities. But, as mentioned earlier, RER depreciation over the past four to five years has been stickier downwards compared to other commodity exporters and relative to the size of the nominal depreciation of the Rupiah. Calì and Nedeljkovic (2016) investigate the reasons behind this stickiness in the Indonesian RER by performing a novel decomposition of the bilateral RER across two dimensions. These analyses show that Indonesia’s RER stickiness in 2014-2015 is mainly due to local retail prices rising faster than border prices, relative to the US and Asian peers (Figure 5). This could be due to increases in tariffs and non-tariff barriers as well as in distribution costs, which in turn is consistent with increased protectionism vis-à-vis foreign investments, and with the rise in domestic fuel prices associated with the reductions in fuel subsidy. Another decomposition shows that the food prices component of the RER is also the one putting most upward pressure to the RER (Figure 6). Among comparator countries, Indonesia had the second highest average annual relative food price inflation over 2003-2015 and this food price inflation has determined a slower pace of RER depreciation over the entire period, including the most recent one. This confirms the role of protectionist food trade policy in undermining the price competitiveness of Indonesian exports. On the other hand their analysis suggests that changes in the relative unit labor cost have not played any significant role in the slow rate of the recent Indonesian RER depreciation.10 Figure 5 – RER decomposition - Annual contributions to Indonesian RER CPI change (vs. USD): tradable prices based on export and import indices, 2002-2015 0.20 RER USA_INDONESIA 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 2003 2005 2007 2009 2011 2013 2015 rer_n rer_t rer_d rer _cpi Legend: rer_n: non-tradeables; rer_t: tradeables; rer_d: distribution wedge; rer_cpi: CPI RER 9 Declining returns to schooling could also be caused by low labor force growth. Between 1997 and 2007, Indonesia faced declining employment growth. Even during a job recovery period in 2003-2007, the employment ratio fell by 0.1 percent per year. At the same time, higher shares of workers are educated. In 2007, 21 percent of workers graduated from high school, compared to only 11.5 percent in 1990 (World Bank, 2010). 10 Indonesia’s unit labor cost in 2014 was higher than Malaysia, the Philippines, and Vietnam (Diop, 2016). However, the quarterly trend in Calì and Nedeljkovic (2016) is decreasing since 2012. 11 Source: World Bank staff’s calculations, based on IMF and IFS data Figure 6 – RER decomposition – Annual contributions to Indonesian RER change (vs. USD): food vs. non food prices, 2002-2015 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20 2002 2004 2006 2008 2010 2012 2014 ner cpi_food cpi_nonfood rer Legend: ner: nominal exchange rate; cpi_food: food CPI; cpi_nonfood: non-food CPI; RER: real exchange rate Source: World Bank staff’s calculations, based on IMF and IFS data This lack of further depreciation also hinders manufacturing firms to regain competitiveness and, therefore, their employment potential. However, since the middle of 2015 the Government has been addressing some of the sources of the distribution wedge, by reducing some import licenses and by pushing on infrastructure development and port efficiencies. Should further reforms continue, their impact in lowering distribution costs are likely to improve the competitiveness of the manufacturing sector and, eventually, its demand for labor. On the labor supply side, the ability of workers to transition across sectors is important to make this process of structural transformation unfold. Workers need to be able to relocate across sectors of the economy in response to these changing opportunities. Indonesia is a country that is geographically divided across remote islands with high transportation costs, as well as spatially divided in terms of sectoral locations. This makes labor mobility all the more difficult for workers to move from agriculture to manufacturing, for example. Thus there are reasons to believe that the labor supply side may present a significant challenge for Indonesia, and should also be the focus of policy. The rest of the paper looks at the labor transition trends in the past two decades. It then looks at a number of possible explanations across both the supply and demand side of labor. The analyses draw on a number of background work, including labor mobility cost, employment elasticity, and labor content of export. 12 2. Labor transition trends: results from the IFLS How does labor reallocate across sectors of the economy in Indonesia in response to changing incentives? Using the World Bank’s Trade and Labor Adjustment Costs toolkit, the implicit costs that workers face when moving across sectors and regions of the Indonesian economy are measured, by type of worker, and over time. High labor mobility costs can slow down the process of structural transformation. This section is based on the Indonesia Family Life Survey (IFLS), with comparisons between 1997-2000 and 2000-2007 (waves 2, 3, and 4).11 Services sectors continued to absorb large shares of workers Between 1997 and 2007, the share of workers in low-productivity services increased in Indonesia since the start of the commodity boom. Figure 7 presents recent employment trends in Indonesia across sectors (top) as well as geographic areas (bottom). The most notable shift in employment trends is the increase in the share of workers in social services, as well as trade and accommodation services. On the other hand, the share of workers in manufacturing declined. Despite these changes in sectoral labor shares, there has been relatively little change in labor shares across islands in Indonesia. Sumatra has seen a slight increase in the share of employed persons aged 15+ between 1997 and 2007 and Java slight decreases, but other provincial shares have changed by less than 1 percent. There is also a large variation in employment within sectors across regions. Figure 7 – Labor shares (%), aged 15 and above, across sectors and regions, 1997-2007 100 90 Social services 80 Finance, other business 70 Transport, communications 60 Trade, accommodation 50 Construction 40 Utility supply 30 Manufacturing 20 Mining, quarrying 10 Agriculture, forestry, fishing 0 1997 2000 2007 11 Wave 1 (1993) is not included due to different parameters, making it incomparable to the subsequent waves. Wave 5 (2014) is not yet included since the results of this wave still need to be verified. 13 100 90 80 Sulawesi 70 Kalimantan 60 Lesser Sunda Islands 50 East Java Central Java 40 West Java 30 Jakarta 20 Sumatra 10 0 1997 2000 2007 Source: World Bank staff’s calculations using data from Waves 3 and 4 of the IFLS. Between 2000 and 2007, as well as between 1997 and 2000, there was high incidence of workers changing sectors of employment in Indonesia. Transition matrices provide the basis for estimating labor mobility costs, and we estimate worker transitions between sectors using panel data from Waves 2 and 3 as well as Waves 3 and 4 from the Indonesia Family Life Survey (IFLS) on gross labor flows between sectors and different states of work (including unemployed and out of the labor market). We distinguish the effects for different worker-types, disaggregating by skill level, age and gender, and we include the entire labor force including self-employed and own account workers. These transitions are reported in Table 1 and Table 2. Each cell reports the share of workers transitioning from each origin sector (row) to all other destination sectors (column) between 2000 and 2007. The cells on the diagonal indicate the shares of workers remaining in their current work/sector status. The transition statistics give a sense of the fluidity of the Indonesian labor market. Mining and quarrying, manufacturing, utility supply, construction, trade and accommodation services, transport and communication services, and social services all witnessed high worker turnover rates, which in part may be driven by the long time horizon between waves of the IFLS. Services sectors have been the largest absorbers of transitioning workers, consistent with services being the most important for job creation.12 For example, between 1997 and 2000 as well as between 2000 and 2007, 38 percent of transitioning workers found employment in trade and accommodation and social services, while less than 1 percent found employment in finance and other business services.13 Jobless workers in 2000 were also more likely to have found jobs in trade and accommodation services in 2007, followed by social services and agriculture (Table 1 and Table 2). The trade and accommodation sector absorbed most unemployed or inactive workers (16 percent of those exiting unemployment or of new labor force entrants flowed into trade and accommodation services). 12 See also World Bank, 2014. 13 These statistics are calculated by excluding the ‘stayers’ on the diagonal, and summing across all rows within the column, to measure which sectors workers are entering. They cannot be re-constructed directly from the tables shown. 14 The labor absorption of the manufacturing sector continued to be subdued . The manufacturing sector absorbed 12 percent of workers between 1997 and 2000 as well as 2000 and 2007. The decline in agriculture, forestry and fishing continued: it absorbed 26 percent of transitioning workers between 1997 and 2000 and 17 percent between 2000 and 2007. It should also be noted that there was increased turnover between 1997-2000 and 2000-2007: for the most part, the share of ‘stayers’ – those workers who moved jobs within the same sectors – declined for most sectors, except finance. Table 1 – Transitions across sectors and into and out of labor force status (%), 2000-2007 2000-2007 S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S1 Agriculture, forestry, fishing 65.6 0.3 4.7 0.1 2.2 6.9 0.9 0.2 4.8 14.4 S2 Mining, quarrying 35.1 16.9 3.9 0.0 3.9 6.5 3.9 0.0 18.2 11.7 S3 Manufacturing 11.7 0.4 35.4 0.2 2.6 15.9 2.1 0.4 12.9 18.5 S4 Utility supply 7.5 0.0 7.5 27.5 7.5 5.0 0.0 0.0 40.0 5.0 S5 Construction 16.6 0.2 9.2 0.2 35.2 12.6 4.0 0.0 11.6 10.6 S6 Trade, accommodation 9.4 0.1 6.6 0.1 1.9 51.2 1.4 0.5 8.7 20.1 S7 Transport, communications 13.2 1.8 6.7 0.0 3.5 12.6 32.5 1.0 23.2 5.5 S8 Finance, other business 4.1 0.0 1.4 1.4 1.4 19.2 0.0 34.2 28.8 9.6 S9 Social services 9.2 0.5 7.7 0.2 2.9 13.1 2.8 0.8 49.1 13.7 S10 Jobless 12.3 0.2 7.1 0.2 1.8 15.9 1.1 0.6 12.6 48.1 Source: World Bank staff’s calculations using data from Waves 3 and 4 of the IFLS. Note: Each cell reports the share of workers transitioning from each origin sector (row) to all other destination sectors (column) between 2000 and 2007. The cells on the diagonal indicate the shares of workers remaining in their current work/sector status. Table 2 – Transitions across sectors and into and out of labor force status (%), 1997-2000 1997-2000 S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S1 Agriculture, forestry, fishing 73.3 0.3 4.0 0.1 1.0 5.1 0.9 0.0 3.7 11.7 S2 Mining, quarrying 19.7 32.4 9.9 0.0 4.2 4.2 5.6 0.0 21.1 2.8 S3 Manufacturing 14.4 0.2 45.4 0.1 2.7 11.6 1.8 0.2 10.8 12.9 S4 Utility supply 6.4 0.0 8.5 36.2 6.4 2.1 8.5 0.0 23.4 8.5 S5 Construction 18.6 0.9 6.8 0.4 43.2 5.9 4.5 0.5 13.2 5.9 S6 Trade, accommodation 9.3 0.0 5.5 0.1 1.6 60.1 1.3 0.5 9.4 12.1 S7 Transport, communications 11.0 0.7 4.3 0.0 3.4 10.3 49.7 0.2 14.6 5.7 S8 Finance, other business 2.5 0.0 3.8 0.0 1.3 6.3 0.0 33.8 41.3 11.3 S9 Social services 7.1 0.2 4.9 0.4 2.3 7.6 1.9 0.5 66.3 8.8 S10 Jobless 15.8 0.2 6.7 0.1 1.2 12.0 0.7 0.2 7.7 55.5 Source: World Bank staff’s calculations using data from Waves 2 and 3 of the IFLS. Note: Each cell reports the share of workers transitioning from each origin sector (row) to all other destination sectors (column) between 1997 and 2000. The cells on the diagonal indicate the shares of workers remaining in their current work/sector status. Large country, low mobility However, there had been little movement of workers across regions. Most regions (islands or island groups, except for Jakarta, in this study) retained between 90 and 99 percent of their workers, between 2000 and 2007 (Table 3). The most likely workers to move were from Jakarta, but mostly to West Java or other parts of Java. Likewise, workers moving to Jakarta were also more likely to have come from other regions in Java. The low likelihood of geographic movement may be due to lower degrees of regional inequality in the 1990s. Tadjoeddin et al. (2001) found that inter-provincial inequality of per capita RGDP (regional GDP) in 1996, would be reduced by about 60 percent if oil and gas incomes and the 13 richest districts were excluded. In addition, welfare outcomes – such as health, education, purchasing power, and human development index – were less unequal than inequality of per capita RGDP. A more recent analysis 15 found that the inter-provincial coefficient of variation (CV)14 of unemployment rates increased slightly from 0.32 in 2001 to 0.33 in 2007, before peaking at 0.41 in 2011. Table 3 – Transitions across regions (%), 2000-2007 S1 S2 S3 S4 S5 S6 S7 S8 S1 Sumatra 97.91 0.6 1.3 0.1 0.1 0.0 0.0 0.0 S2 Jakarta 0.8 90.0 6.7 1.9 0.2 0.4 0.0 0.1 S3 West Java 0.5 1.3 97.3 0.5 0.3 0.0 0.1 0.0 S4 Central Java 0.6 1.2 1.2 96.3 0.6 0.0 0.1 0.1 S5 East Java 0.1 0.2 0.4 0.4 98.5 0.1 0.3 0.0 S6 Lesser Sunda Islands 0.1 0.1 0.1 0.0 0.3 99.4 0.0 0.0 S7 Kalimantan 0.0 0.0 0.0 0.1 0.8 0.0 98.9 0.1 S8 Sulawesi 0.0 0.0 0.2 0.2 0.0 0.2 1.0 98.4 Source: World Bank staff’s calculations using data from Waves 3 and 4 of the IFLS. Note: Each cell reports the share of working age individuals transitioning from each origin region (row) to all other destination r egions (column) between 2000 and 2007. The cells on the diagonal indicate the shares of workers remaining in their current region. Having the right skills and initial entry into formal/informal employment matter The right skills mix appears to be an important factor behind job matching in the Indonesian labor market. Skilled workers defined as those having attended university were more likely to have found a job in 2007, while unskilled workers with only a secondary degree were more likely to have remained jobless (Table 4). The likelihood of exiting agriculture also increases as individuals’ skills level increases, with unskilled workers having the highest incidence of staying and skilled workers the highest incidence of exiting. Students having graduated from vocational training instead had a higher likelihood of entering manufacturing than skilled or unskilled workers, while both university and vocational graduates were able to access services jobs better than unskilled workers. Skilled workers transitioning across sectors and into services had a much higher incidence of entering social services than vocational or unskilled workers. Skilled workers also had a slightly higher incidence of entering finance than vocational, but higher than unskilled workers. There was instead a lower incidence for skilled workers to enter transport and communications and trade and accommodation. Table 4 – Transitions across aggregate sectors by skill level (%), 2000-2007 Skilled Vocational Unskilled S1 S2 S3 S4 S1 S2 S3 S4 S1 S2 S3 S4 S1 Primary 20.9 4.5 61.2 13.4 42.5 7.8 38.0 11.7 67.2 4.9 15.6 12.3 S2 Manufacturing 4.1 21.6 64.9 9.5 4.3 35.9 41.4 18.4 12.5 37.1 32.1 18.3 S3 Services 3.0 5.8 85.9 5.3 4.4 10.7 70.1 14.8 12.4 6.7 65.0 15.9 S4 Jobless 3.0 5.5 61.8 29.7 4.7 9.8 45.6 40.0 13.8 7.3 28.2 50.8 Source: World Bank staff’s calculations using data from Waves 3 and 4 of the IFLS. Moving across sectors as either formal or informal worker seemed more likely to be determined by an individual’s previous formality status. In 2007, individuals entering primary activities from any sector or from joblessness were more likely to do so as informal workers than formal workers (Table 5). The same holds for workers entering services, but not manufacturing, where workers were more likely to enter as formal than informal. Informal manufacturing, services and primary sector workers changing sectors were also more likely to remain informal than enter formal employment. Thus informality may not act as a stepping stone sector into formal employment. 14 CV is defined as the ratio of the standard deviation to the mean (Tadjoeddin, 2015). 16 Table 5 – Transitions across aggregate sectors by formality (%), 2000-2007 S1 S2 S3 S4 S5 S6 S7 S1 Primary, formal 44.5 16.5 5.9 1.1 13.7 8.1 10.2 S2 Primary, informal 1.8 77.2 0.9 1.3 2.8 7.4 8.7 S3 Manufacturing, formal 2.4 4.2 48.7 2.9 13.5 11.5 16.7 S4 Manufacturing, informal 0.7 8.7 6.2 52.3 5.1 17.4 9.6 S5 Services, formal 1.3 4.1 4.4 1.1 65.0 11.7 12.5 S6 Services, informal 0.6 6.5 1.6 2.2 7.7 71.7 9.7 S7 Jobless 2.7 14.2 9.7 3.5 26.8 24.4 18.6 Source: World Bank staff’s calculations using data from Waves 3 and 4 of the IFLS. Advantage: young and male Females were more likely to remain jobless than males, while males were more likely to stay employed within their aggregate sector (primary, manufacturing and services). This reflects a higher incidence of females exiting employment from either of the primary, manufacturing or services sectors. Both males and females were most likely to find jobs in services if entering employment. Young people exiting unemployment or idleness were also more likely to find jobs in the services sector (as opposed to the primary or manufacturing sectors) than their middle-aged or older-aged counterparts. Rural workers in manufacturing or services moving to urban areas were most likely to find jobs in the services sector. The exception is individuals who were jobless in rural areas. If transitioning to urban areas, these individual were more likely to remain unemployed than find a job in the services sector. 17 3. Drivers of labor transition: results from labor mobility cost and employment elasticity analyses In moving across sectors, geographic areas, even across sectoral-formality work arrangements, workers might face costs that hinder their mobility. Labor mobility refers to the ability of workers to move between firms and industries in search of alternative employment opportunities, such as in response to wage differences. Labor mobility costs reflect what a worker perceives to be his or her welfare cost of finding alternative employment. They are defined as the costs perceived by a worker to move to a different firm or industry, independent of the reason for the move. Typical impediments to job switching are skills mismatches (wages forgone because of lower productivity), limitations to geographic mobility (administrative procedures for internal migration and direct relocation costs), and severance and hiring costs (including those imposed by laws or regulations). Other factors may be location preferences, job search costs, and even the psychological costs of changing jobs. This section discusses impediments to labor mobility, drawing on results from labor mobility costs, based on IFLS data, and employment elasticity analyses, based on the Indonesian labor force survey ( Sakernas) data.15 Box 1 – Defining labor mobility costs It is not possible to measure labor mobility costs directly, because they are not readily observable. Instead, we use an indirect method that combines the observed worker transitions between sectors with the inter-sectoral wage gaps to estimate the “labor mobility cost” to explain why workers do not transition into higher wage sectors to the extent that wage gaps are eliminated . There may be multiple reasons why wage gaps persist, e.g., on the labor supply side (when it is physically or technically difficult due to skills mismatch to obtain a job in a certain sector), or on the labor demand side (such as a lack of new private sector job openings). Both are obstacles to matching labor supply with labor demand at a market clearing wage, and both therefore imply a high cost of transitioning to said job. The methodology of this model differs from standard models of trade, which assume no frictions in the adjustment of labor. For example, workers are assumed to move without frictions from import- competing sectors to export-expanding sectors, after an economy opens to trade. But in reality, workers respond slowly to trade-related shocks. When there are low transition rates across sectors of the economy despite high wage gaps, we interpret this to mean that it is costly for workers to move. We estimate the welfare costs for a worker to switch industries/jobs using a dynamic rational-expectations model of costly labor adjustment.16 In each period, a worker can choose to move from her current industry to another one, but must pay a cost in doing so. The decision for a worker to move depends on her expected welfare gain, net of the welfare cost of moving. So for a sector that is difficult to access, we would assign a high labor mobility cost for entering that type of job. Combining the transition data with the observed wage gaps between sectors leads to estimates of labor mobility costs for entering each sector, expressed as a ratio of the annual average wage. These transition costs are estimated for all workers, as well as for skilled and unskilled workers, and for men and women.17 The methodology used here is sensitive to the level of 15 Please see Annex I for a more detailed note on the methodology of labor mobility costs. 16 See the Trade Shocks and Labor Market Adjustments toolkit developed by the World Bank (Hollweg et al. 2014). The Annex provides details of the methodology and data. 17 The cost has a common component, which does not vary across similar type workers in an industry, and an idiosyncratic component. By solving the dynamic rational-expectations model of costly labor adjustment, it is possible to derive an 18 disaggregation: the higher degree of sectoral disaggregation, the fewer the observed transitions, the higher the estimated mobility costs. It is therefore crucial to exercise caution when interpreting the results. This type of exercise is most useful for comparing the relative mobility costs across different sectors of an economy – particularly those most closely affected by trade shocks – and different types of workers. How could labor mobility costs influence Indonesia’s economic adjustment in response to falling international commodity prices, and ultimately the country’s structural adjustment in the fu ture? Changes in prices in global markets can affect the relative prices faced by domestic firms that consume or produce traded goods, which in turn affects the demand for labor as sectors expand or contract. Indonesia, as a large commodity exporter, is even more vulnerable to global price changes of few commodities that make up the majority of its exports. These types of trade-related shocks affect the relative return to labor across industries as well. The resulting wage implications will also induce supply- side labor reallocations across sectors. Large mobility costs mean sluggish reallocation of labor, reducing the potential benefits to workers and the economy as a whole. They may be high enough to dissuade labor supply to certain sectors, but these sectors may be the most productive and with the highest wages. This has implications for how Indonesia’s economy will adjust to large external shocks, including the large decline in commodity prices that Indonesia is facing today. Labor mobility costs in Indonesia are relatively high Labor mobility costs in Indonesia are high even among other countries at a similar level of economic development. Artuc et al. (2013) measure the average labor mobility cost for workers to transition across 8 broad manufacturing sectors for 47 countries worldwide over the period 1995-2007. 18 Labor mobility costs in Indonesia’s manufacturing sector are among the highest in the sample. Richer countries tend to have lower mobility costs in manufacturing, and the cross-country correlation with GDP per capita is negative and robust (Figure 8). Yet costs in Indonesia continue to be high relative to other countries, even after accounting for Indonesia’s GDP per capita level. Workers in Indonesia also face higher welfare losses to change jobs across manufacturing sectors than other countries in the Asia-Pacific. While Chile also departs from expected levels given the country’s GDP per capita, other countries such as Japan, the United States and Singapore have among the lowest labor mobility costs in the sample, significantly below other developed countries. For countries with high labor mobility costs, changes in market access would induce greater trade opportunities in some sectors while at the same time imposing import-competing pressures on other sectors of the economy. Countries with lower labor mobility costs domestically would be better positioned to realize these opportunities in the short term. equilibrium condition that is a kind of Euler-equation, which lends itself to estimation. The structural model identifies workers’ transitions across sectors of an economy that depend on the wage gaps between those sectors and the mobility costs of entering a sector. From observed data on sectoral transitions and sectoral wage gaps, we are able to estimate the labor mobility cost using GMM-type estimations. 18 Artuc et al. (2013) use employment and wage data across 8 aggregated manufacturing sectors for the period 1995-2007 from UNIDO Industrial Statistics Database. Although transitions across manufacturing sectors exclude transitions between services sectors, for example, Artuc et al. (2013) is the only source of internationally comparable labor mobility cost estimates across a wide range of countries. This international comparison still provides useful insights for Indonesia, despite it being a specific case for manufacturing. 19 Figure 8 – Labor mobility costs vs. GDP per capita in Indonesia and other TPP countries 9.00 8.00 PER 7.00 Labor mobility cost IDN 6.00 CHL 5.00 4.00 3.00 CAN 2.00 JPN 1.00 SGPUSA 0.00 4 5 6 7 8 9 10 11 12 log GDP per capita (constant 2005 US$) Source: Adapted from Hollweg et al., (2014) calculated using data from Artuc et al. (2013) and World Bank World Development Indicator s. Note: This figure plots the correlation between the estimated aggregate labor mobility cost for each country’s manufacturing sec tor expressed as a ratio of the annual average manufacturing wage (vertical axis) and economic development (horizontal axis). Level of econ omic development is measured as the average log of GDP per capita (constant 2005 US$) for 1995-2007. Labor mobility costs are correlated with aggregate indicators related to a country’s well -being, labor market characteristics, educational attainment, and regulatory distortions. Based on cross-country labor mobility cost estimates of Artuç et al. (2013), richer countries tend to have lower mobility costs, but not because the adjustment costs of firms are lower. While there is a weak correlation between mobility costs and firing costs, the correlation with GDP per capita is quite negative and robust. These results are consistent with the assertion that distortions affecting firm labor adjustments (captured by firing costs) are not the main driver of mobility costs. There are also positive correlations with the poverty head-count and the poverty gap, but no obvious correlations with inequality. Mobility costs also tend to be lower in countries more highly specialized in non-primary sectors or with highly educated work forces. Countries with lower educational quality (a higher pupil-teacher ratio) tend to have higher mobility costs. Labor market rigidities are more prevalent in countries characterized by other types of rigidities and distortions. Labor mobility costs are positively correlated with other frictions and constraints, such as time to export. Countries where labor mobility costs are high tend to obstruct trade more than countries with more nimble labor markets. Case studies that examine labor mobility costs across aggregate sectors within countries find that firm size, informality, sector-specific knowledge, and education level also affect labor mobility costs. In Brazil, Mexico and Morocco, it is relatively less costly to move into an informal than a formal job, but also less costly to move into a formal job from an informal job in the same industry, which acts as a stepping stone for formality. Industry-specific skill requirements help explain the finding that it is always less costly to become formal in the same industry than to become formal by switching industries. Evidence from Morocco shows that finding employment in large firms is less costly than in small firms. In Lao PDR, skilled workers face lower costs to transition across sectors compared to unskilled workers, although the results vary by sector. This also suggests that employers may prefer workers with sector- specific knowledge. Understanding what is driving Indonesia’s relatively high labor mobility cost can help identify policy responses to address these costs. Could Indonesia’s relatively high costs to labor mobility internationally be the result of policy-relevant factors, such as skills mismatches, sectoral characteristics, or geographic dispersion? We measure labor mobility costs to enter different sectors and regions within 20 Indonesia to better understand why labor mobility costs appear to be higher in Indonesia than in other countries in the world, what implications this may have for economic transformation post commodity- price boom in Indonesia, and available policy levers. The main added value of this part of the analysis is to see internally the costs that workers face when moving and finding/creating jobs domestically. To disentangle the drivers of labor mobility costs in Indonesia, the analysis cuts the data along various dimensions including across worker type. …But, have been declining across sectors and regions Labor mobility costs in Indonesia are high, but have declined since 1997-2000. Between 1997 and 2000, the average labor mobility cost was estimated to be 3.3 times the average annual wage, and has since dropped to 2.8 time between 2000 and 2007. This result accords with the greater fluidity in Indonesia’s labor market, observed in the transition matrices.19 There is also wide dispersion in costs that workers face to enter different sectors in Indonesia (Figure 9, top left). Labor mobility costs in Indonesia, in 2007, range from about 5 times the average annual wage to enter the utility sectors (electricity, water, and gas), to about 1.5 times the average annual wage to enter social services. Manufacturing is also one of the lowest cost sectors for workers to enter – even lower than agriculture, forestry and fishing and not very different from low-end services of trade and social services, even though wages continue to be lower in agriculture than manufacturing. Although we cannot deduce from the labor mobility cost estimates, looking at the within-sector labor shares across skills in 2007, manufacturing has a higher share of unskilled labor than utility, finance, and social services. Low productivity services sectors, such as trade and social services, also have lower costs to entry than agriculture, forestry and fishing. Labor mobility costs also tend to be correlated with the size of the sector or the region, but not average wages (Figure 9, bottom). Leaving employment either by becoming unemployed or exiting the labor force involves the lowest transition cost. While it is still less costly for individuals to exit the labor force than it is to find alternative employment in other sectors and/or regions, there are still costs that workers face when making this transition, such as the personal costs that workers face when being unemployed. The changes in the flows of workers across sectors in Indonesia coincide with changes in the labor mobility costs over time. These results are reflective of structural adjustment in Indonesia away from agriculture towards services. Agriculture is no longer a sector with strong job creation. Instead services are the most important sector for job creation. The relatively low entry costs for services is consistent with robustly significant and strong employment elasticities for these sectors (see below). There is also wide dispersion in costs that individuals face to move across regions. Java – in particular Jakarta, West Java and Central Java – are the least costly for workers to enter. Sumatra – a region like Java where manufacturing is concentrated – is also among the least costly regions for individual mobility. Instead, moving to peripheral regions, such as the Lesser Sunda Islands and Sulawesi, are associated with significantly higher costs – three times that of Jakarta. Between 1997-2000 and 2000- 2007, the costs to enter all regions declined, except in the Lesser Sunda Islands, where the cost workers 19We include all individuals employed in a sector in the transition statistics, regardless of whether they are wage workers, own- account workers, or non-wage family workers. However, the average sectoral wage is calculated for wage earners only. This could bias the average sectoral wage either upward or downward, which may in turn bias the labor mobility cost estimate for entering the sector. Similarly, the average sectoral wage does not differentiate between the skills mix of workers in the sector. 21 face to enter the sector and region increased (Figure 9, top right). Regional differences are more likely to be driven by geographic allocation (and isolation), accessibility and connectivity, as well as (perception of) better job opportunities. For example, between 2000 and 2011, most of new manufacturing firms were established in Java and Sumatra (despite the fact that most of existing manufacturing firms are already in these regions to begin with).20 Figure 9 – Labor mobility costs, wages, and employment, across sectors and across regions, 1997-2007 Utility supply Lesser Sunda… Finance, other business Sulawesi Mining and quarrying Kalimantan Transport and communication East Java Construction Agriculture, forestry, fishing Sumatra Social services Central Java Manufacturing West Java Trade and accommodation Jakarta Jobless 0 2 4 6 0 2 4 6 8 10 Labor mobility cost Labor mobility cost 1997-2000 2000-2007 1997-2000 2000-2007 Utility supply Sulawesi Finance, other business Kalimantan Mining, quarrying Lesser Sunda Islands Transport, communications East Java Construction Central Java Agriculture, forestry, fishing West Java Social services Manufacturing Jakarta Trade, accommodation Sumatra 0 10 20 30 40 0 5 10 15 20 25 Average wage Employment share Average wage Employment share Source: World Bank staff’s calculations using data from Waves 3 and 4 of the IFLS. The right skills mix and informality remain important Having the right skills mix is important for labor mobility. Given that both manufacturing and low- productivity services sectors have low entry costs, the ability for workers to enter these sectors may come down to other worker-specific factors such as skills. Skilled workers face lower costs to transition across sectors compared to unskilled workers, although the results vary by sector. A study shows that in today’s job market, theoretical and practical knowledge of the job as well as core generic and subject- based skills seems to be on the demand (ADB 2010). There is also higher demand in the service and export-oriented sector (ADB 2010). Vocational workers often face higher costs to transition across 20 World Bank staff own calculation using Indonesia’s manufacturing census 2000 and 2011. 22 sectors than unskilled workers. This is the case in the primary sector, as well as many services sectors. This may reflect a mismatch between the skills vocational workers are taught, and those that are demanded by the private sector, for example.21 The return of vocational high school over primary school is not so much different from the return of general high school over primary school (data spans from 1994 to 2007), with general high school cost of IDR 5.3 million per year compared to vocational high school cost of IDR 6.8 million per year (ADB 2010). Unemployment rates among vocational graduates have also been increasing from 1990 to 2007 (ADB 2010). This raises questions on the quality of vocational schools. Moreover, World Bank (2014) also noted that more than half of senior secondary graduates are employed in unskilled occupations, while half of tertiary graduates are employed in occupations below their level of education, which reflects the skills mismatch issues in Indonesia’s labor market. Despite the importance of skills, the Indonesian economy is unable to provide productive jobs for skilled workers. Three quarters of skilled workers started off in either the social service sector or jobless in 2007. Vocational school graduates face low entry costs into low productivity services sectors (social services and trade). But, they face higher entry costs for higher value-added services sectors such as finance and other business sectors. Select interviews with the private sector suggested that vocational graduates were not equipped with the needed skills relative to their unskilled counterparts. Female workers face greater costs to transition across sectors compared to male workers. This is consistent with the observed low number of worker transitions by women, and generally lower employment elasticities for females than males in the period of 1993-2006, excluding 1997-1999 (see below). In general, labor mobility costs are higher for female than male workers in most sectors, except agriculture. These higher labor mobility costs for females may also be a reflection of their lower wages. A study shows that the human capital and the ability to generate higher income of men is perceived to be higher than that of women and therefore men are sent to migrate for work; women become especially immobile after marriage; women also prefer to opt for employment which is more compatible with their everyday lives (Jϋtting and de Laiglesia, 2009). Young workers tend to have lower labor mobility costs to enter sectors than other age cohorts. For example, young workers (age 15-24) have the lowest entry costs in finance, trade and accommodation, transport and communication, construction, and social services. 21 See also Alisjahbana (2008). 23 Figure 10 – Labor mobility costs across sectors by worker types, 2000-2007 Jobless Jobless Social services Social services Finance, other business Finance, other business Transport and… Transport and… Trade and accommodation Trade and… Construction Construction Utility supply Utility supply Manufacturing Manufacturing Mining and quarrying Mining and quarrying Agriculture, forestry, fishing Agriculture, forestry,… 0.00 2.00 4.00 6.00 0.00 2.00 4.00 6.00 Labor mobility cost Labor mobility cost Unskilled Vocational Skilled Male Female Jobless Social services Finance, other business Transport and… Trade and accommodation Construction Utility supply Manufacturing Mining and quarrying Agriculture, forestry, fishing 0.00 2.00 4.00 6.00 Labor mobility cost Old Middle aged Young Source: World Bank staff’s calculations using data from Waves 3 and 4 of the ILFS. Comparing across formality status as well as sectors, the results show that it is less costly for workers to enter manufacturing as formal than as informal workers. On the other hand, it is less costly for workers to enter the primary sector as informal than as formal (Figure 11) – but only for agriculture, forestry and fishing (not mining and quarrying). This is supported by the transition matrices above, whereby manufacturing or services workers entering primary activities tend to do so as informal primary workers. Within services sectors, only in electricity, gas, and water supply, and trade and accommodation do workers find it less costly to enter as informal than formal; for other services sectors it is less costly to enter as formal. There are few differences in labor mobility costs for workers to move between urban and rural locations, suggesting urbanization (unlike provincial changes) is not a barrier for job switching. 24 Figure 11 – Labor mobility costs across sectors by job type, 2000-2007 3.50 2.50 3.00 2.00 Labor mobility cost Labor mobility cost 2.50 1.50 2.00 1.50 1.00 1.00 0.50 0.50 0.00 0.00 Primary Manufacturing Services PrimaryManufacturing Services Jobless Formal Informal Urban Rural Source: World Bank staff’s calculations using data from Waves 3 and 4 of the IFLS. Do real wages matter? Real wages at the sectoral and regional levels in Indonesia did not experience major changes across the years. The IFLS data on real wages indicate that finance and utility sectors still have the highest real wages in 2007, just as they did in 1997, although both have seen real wages declining. Other sectors have seen increased real wages, except for transport, which also saw a lower real wages in 2007 compared to 1997. At the regional level, all regions saw increased real wages, between 1997 and 2007, except for Jakarta. This could partially explain the lack of regional movement of workers. Figure 12 – Real wages across sectors, 1997-2007 35 30 25 20 15 1997 10 2000 5 2007 0 Source: World Bank staff’s calculations, using data from Waves 2-4 of the IFLS. Note: Wages in millions of 2010 Rupiah adjusted for purchasing power parity across provinces. Wages include salary/wages plus ben efits and net business profit last year. 25 Figure 13 – Real wages across regions, 1997-2007 25 20 15 1997 2000 10 2007 5 0 Sumatra Jakarta West Java Central Java East Java Lesser Sunda Kalimantan Sulawesi Islands Source: World Bank staff’s calculations, using data from Waves 2-4 of the IFLS. Note: Wages in millions of 2010 Rupiah adjusted for purchasing power parity across provinces. Wages include salary/wages plus benefits and net business profit last year. The relationship between wages and employment is mixed. Tadjoeddin (2015) confirmed the negative relationship between real wages22 and employment in the manufacturing sector. However, the same study also found a positive correlation between real wages and employment in large and medium manufacturing firms, suggesting that a 1 percent increase in real wages is associated with a 0.55 percent increase in employment. Evidence from employment elasticity Employment elasticity in Indonesia is declining at both the national and sub-national levels. Analysis on employment elasticity supports the findings of the labor mobility cost analysis. The pattern over the past 20 years (1993-2015, excluding 1997-1999) shows the important role of services and manufacturing sectors in job creation and the decline of agriculture. This is in line with the trend of the employment shares of agriculture, manufacturing and services. While the share of employment in the primary sector has decreased significantly from 56 percent to 34 percent between 1990 and 2015, the share of manufacturing has increased slightly from 10 percent to 13 percent, and the share of services has increased substantially from 34 percent to 53 percent (Table 6). Table 6 – Share of employment (%), 1990-2015 Year Primary23 Manufacturing Services 1990 55.93 10.17 33.90 1991 54.15 10.40 35.45 1992 53.67 10.55 35.79 1993 50.82 11.10 38.08 1994 46.52 13.23 40.25 1995 44.24 12.64 43.12 22 Predicted real wages in the model of employment function used in Tadjoeddin (2015), which also ran a first step regression of productivity on wage. 23 Primary sector consists of agriculture and mining/quarrying. 26 1996 44.40 12.60 43.00 1997 41.76 12.89 45.35 1998 45.73 11.33 42.94 1999 44.03 12.97 43.01 2000 45.78 12.96 41.26 2001 44.81 13.31 41.88 2002 45.03 13.21 41.76 2003 47.17 12.39 40.45 2004 44.43 11.81 43.76 2005 44.93 12.72 42.35 2006 43.01 12.46 44.53 2007 42.23 12.38 45.39 2008 41.35 12.24 46.42 2009 40.78 12.24 46.98 2010 39.51 12.78 47.72 2011 37.73 13.54 48.74 2012 36.61 13.88 49.51 2013 36.05 13.27 50.69 2014 35.25 13.31 51.44 2015 34.03 13.29 52.69 Source: World Bank staff’s calculations based on Sakernas Box 2 – Employment elasticity methodology note The employment elasticity analysis is conducted at both the national and regional levels, using the Indonesian labor force survey data (Sakernas) available from BPS-Statistics Indonesia. Employment elasticity is defined as the percentage point increase in employment for one percentage point increase in GDP or GRDP. Sectoral (value-added) employment elasticity is defined as the percentage in sectoral employment for one percentage point increase in the corresponding sectoral value-added. The periods of observations are 1993 to 2015, with a cut-off period of 2007, because 2007 was the first year the Government increased the size of the labor force survey to provide stability in the overall data quality. The analysis also excludes the Asian crisis period of 1997-1999. Regional groups are: Java-Bali, Sumatera, Nusa Tenggara, Sulawesi, Kalimantan, Maluku, and Papua. These groupings are similar to the ones used for labor mobility, except for the eastern Indonesia regions of Maluku and Papua, which are excluded from the labor mobility cost analysis due to lack of data. In terms of the age groups, 15 to 24 year olds are defined as young workers; 25-64, adult; and 65 and over, old. Definition of formal is based on the simplified definition from BPS-Statistics Indonesia, in which formal workers include employees and employers assisted by permanent workers. This is a less restrictive definition than the international definition of formality that includes labor contract and the provision of social security benefits. Log-on- log regression model is used, controlling for regional dummies and interaction terms, following Kapsos (2005). Wald tests are used to test for the significance on the employment elasticity coefficients. At the national level, employment elasticity declined from 0.51 in 1993-2006 (excluding 1997-1999) to 0.32 in 2007-2015, but employment elasticity declined very significantly at the end of the commodity boom from 0.39 in 2007-2012 to 0.16 in 2013-2015. This indicates fewer job creations in percentage 27 terms per one percent GDP growth in the recent period. However, this may also point to some labor productivity increases.24 In fact, this decline in the employment elasticity is consistent with the increase in the labor productivity growth divided by GDP growth25 that increased from 0.55 in 1993-2006 to 0.68 in 2007-2015. Nationally, this indicates that the economy has become less labor intensive. Between the two periods, transportation and agriculture experienced the biggest drop in elasticity, from positive to negative elasticities, while the biggest increases took place in mining and social services (Table 7). Table 8 shows the sectors that have become less labor intensive are agriculture, manufacturing, utility, trade and accommodation, transport and communication, and finance and other business services. Sectors that have become more labor intensive are social services. Manufacturing experienced a reduction in elasticity from 0.75 to 0.56. However, this could also point to some labor productivity improvement. Labor productivity growth over GDP growth of manufacturing sector increased from 0.22 in 1993-2006 to 0.50 in 2007-2015. Meanwhile, manufacturing sector creates about 1.2 million informal jobs and 5.6 million formal jobs between 1993 and 2015. Table 7 – Sectoral employment elasticity, 1993-2006 vs. 2007-2015 1993-2006 (excl 2007-2015 1997-1999) National 0.51 0.32 Agriculture, forestry, fishing 0.38 -0.29 Mining and quarrying 0.53 1.94 Manufacturing 0.75 0.56 Utility -0.28 -0.18 Construction 0.59 0.72 Trade, accommodation 0.71 0.38 Transportation, communication 0.79 -0.30 Finance, other business 1.95 1.77 Social services -0.06 0.84 Source: World Bank staff’s calculations based on Sakernas Table 8 – Labor productivity growth rate/GDP growth rate, 1993-2006 vs. 2007-2015 1993-2006 2007-2015 National 0.55 0.68 Agriculture, forestry, fishing 0.90 1,28 Mining and quarrying -1.51 -0.62 Manufacturing 0.22 0.50 Utility 0.88 1.36 Construction -0.75 0.20 Trade, accommodation 0.25 0.57 Transportation, communication 0.21 1.22 Finance, other business -1 -0.52 Social services 0.86 0.22 Source: World Bank staff’s calculations based on Sakernas 24 Tadjoeddin, 2015. 25 Note that for small changes in output: ∆/ = ∆/ + ∆/, where ∆/ is the growth rate of output, ∆/ is the growth rate of labour, and ∆/ is the growth rate of labour productivity. Divided both the LHS and the RHS with ∆/, we get: 1 = ∆ ∆ + / , where is employment elasticity. We calculate the compound annual growth rate (CAGR) for both labor productivity and GDP for 1993-2005 and 2007-2015 and divided the CAGR for labor productivity by the CAGR for GDP. 28 Services sectors most robustly create jobs throughout the observed periods and across regions. Employment elasticities for trade and accommodation and finance and business services are robustly significant and strong for the 1993-2006 and 2007-2015 periods, and across all regions. This is in line with the finding that labor mobility costs for low-productivity services were relatively low during the 2000-2007 period. The high employment elasticity in finance and business services seems to be due, not to low labor mobility costs, but to the ‘base effect’ with a very low share of employment (o ut of total employment) of 0.73 percent in 1993. The higher-than-one employment elasticity in finance and business services may also be due to negative labor productivity growth since, as shown above, real wages in the sector has declined between 1997 and 2007, assuming a positive correlation between productivity and real wages. In fact, value-added per worker in finance and business services have gone down from 475 million IDR per worker to 234 million IDR per worker. Labor productivity growth divided by GDP growth for finance and business services were indeed negative: -1 in 1993-2006 and -0.52 in 2007-2015 (Table 8). Services sectors are also important for female job creation. Female employment elasticities between 1993 and 2015 were higher than male employment elasticities in utility, trade, transportation, finance, and social services (Table 9). Meanwhile, male employment elasticities between the same periods were higher in agriculture, mining, manufacturing, and construction. Table 9 – Sectoral employment elasticity, by gender, 1993-2006 vs. 2007-2015 Gender 1993-2015 1993-2006 2007-2015 (excl 1997- (excl 1997- 1999) 1999) Agriculture, forestry, fishing Female .0002 0.10 -0.39 Male 0.17 0.54 -0.23 Mining and quarrying Female 0.17 -0.94 -0.37 Male 2.85 0.79 2.20 Manufacturing Female 0.49 0.43 0.38 Male 0.68 0.99 0.69 Utility Female 0.03 0.02 -0.21 Male -0.03 0.03 -0.17 Construction Female 0.54 -0.02 0.87 Male 0.82 0.61 0.71 Trade, accommodation Female 0.51 0.50 0.42 Male 0.49 0.91 0.34 Transportation, communication Female 0.95 1.29 -0.83 Male 0.23 0.77 -0.25 Finance, other business Female 1.68 2.01 1.69 Male 1.66 1.92 1.80 Social services Female 0.77 0.32 1.02 Male 0.34 -0.31 0.69 Source: World Bank staff’s calculations based on Sakernas It also confirms the labor mobility costs finding that male workers have an advantage in finding a job in the period prior to 2007; employment elasticities for male workers are higher than female workers in that period. But, this trend was reversed post-2007. This is also evident in the stronger average annual growth rate of female employment after 2007. Average annual growth rate of female employment is 2.71 percent between 1993 and 2006 and 4.38 percent between 2007 and 2015. Meanwhile, average annual growth rate of male employment between 1993 and 2006 is 2.99 percent and 3.18 percent 29 between 2007 and 2015. For the whole period of 1993-2015, female employment average annual growth rate is 3.64 percent, creating about 12.4 million jobs, while male employment average annual growth rate is 3.17, creating about 18.5 million jobs. Most female employment post-2007 were created in the services sector (Figure 14). Although female employment elasticities were higher than male in some services sub-sectors in 1993- 2015 period and across all regions in 2007-2015 period, there is no evidence that this is the result of higher labor force participation (LFP) rate for female workers . Between 1993 and 2015, LFP for women increased by 14 million for female and 27 million for male, with an average annual growth rate of 1.78 percent for female and 2.05 percent for male. The higher employment elasticities for women in the later period were due to the rise of female employment. Figure 14 – Employment by gender and by 3-sector of employment, 1990-2015 40,000,000 35,000,000 30,000,000 25,000,000 20,000,000 15,000,000 10,000,000 5,000,000 0 1991 2009 2014 1990 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2010 2011 2012 2013 2015 Primary - female Manufacturing - female Services - female Primary - male Manufacturing - male Services - male Source: World Bank staff’s calculations based on Sakernas Contrary to the lower mobility cost for young workers, their employment elasticities are only strong in some services sub-sectors. Young workers face lower employment elasticities compared to older workers across all years and all sectors (Table 10). There are a few important sectors with high elasticities for young workers include mining, construction, trade, finance and social services. But, even in these sectors, their elasticities are lower than older workers. More youths are going out of the agriculture sector. Moreover, workers do not stop working at the age of 64, and those aged 65 and over are more likely to enter agriculture and some services sub-sectors. This implies that there are factors behind the low employment elasticity for youth in many sectors that are not explained by the low labor mobility costs . This prompts a study of what these factors are. Figure 15 shows the significant decline of youth employment in the primary sector. 30 Table 10 – Sectoral employment elasticity, by age group, 1993-2006 vs. 2007-2015 Age group 1993-2015 1993-2006 2007-2015 (excl 1997- (excl 1997- 1999) 1999) Agriculture Youth -0.65 -0.55 -1.37 Adult 0.20 0.53 -0.14 Old 0.70 1.14 -0.01 Mining and Quarrying Youth 1.91 -0.09 1.29 Adult 3.12 -0.17 2.05 Old 2.50 1.53 1.57 Manufacturing Youth -0.02 -0.21 0.04 Adult 0.82 1.14 0.71 Old 0.58 0.37 0.53 Utility Youth -0.14 -0.18 -0.27 Adult -0.18 -0.30 -0.14 Old -0.10 -0.18 0.40 Construction Youth 0.39 0.62 0.35 Adult 0.92 0.68 0.77 Old 0.70 -0.07 1.41 Trade, accommodation Youth 0.45 0.40 0.27 Adult 0.52 0.79 0.40 Old 0.50 0.52 0.40 Transportation, communication Youth 0.08 0.59 -0.69 Adult 0.32 0.80 -0.23 Old 0.65 1.10 -0.23 Finance, other business Youth 1.76 1.18 1.80 Adult 1.88 2.17 1.77 Old 1.72 1.77 1.30 Social services Youth 0.32 -0.64 0.50 Adult 0.56 0.05 0.89 Old 0.52 -0.21 0.97 Source: World Bank staff’s calculations based on Sakernas Figure 15 – Employment by age groups and 3-sector of employment, 1990-2015 60,000,000 50,000,000 40,000,000 30,000,000 20,000,000 10,000,000 - 1995 2003 2011 1990 1991 1992 1993 1994 1996 1997 1998 1999 2000 2001 2002 2004 2005 2006 2007 2008 2009 2010 2012 2013 2014 2015 Primary - youth Manufacturing - youth Services - youth Primary - adult Manufacturing - adult Services - adult Primary - old Manufacturing - old Services - old Source: World Bank staff’s calculations based on Sakernas 31 There is also strong evidence of formalization in most sectors . Most sectors, except trade and utility, saw increases in formal employment elasticity between 1993-2006 (excluding 1997-1999) and 2007- 2015 (Table 11). These increases were also accompanied by decreases in informal employment elasticity between the two periods, except in social services. From 1993-2015, mining, manufacturing, trade, and transportation are sectors with higher employment elasticities in the formal than the informal jobs. 26 This is in line with the employment data , which shows that the average annual growth rate of formal employment has increased from 1.36 percent to 5.80 percent between the period of 1993-2006 and the period of 2007-2015. Meanwhile, the average annual growth rate of informal employment has declined from 1.81 percent to -0.48 percent between the same periods of time. Informal employment has created about 14.3 million jobs while formal employment has created about 23.7 million jobs between 1993 and 2015. Services sector created the most formal employment relative to manufacturing and primary sector (Figure 16). Table 11 – Sectoral employment elasticity, by formality/informality of jobs, 1993-2006 vs. 2007-2015 Informal/formal 1993-2015 1993-2006 2007-2015 (excl 1997- (excl 1997- 1999) 1999) Agriculture, forestry, fishing Informal 0,17 0.60 -0.47 Formal -0.34 -1.50 1.30 Mining and quarrying Informal 2.07 1.19 0.07 Formal 2.86 -0.16 3.65 Manufacturing Informal 0.40 0.46 -0.41 Formal 0.69 0.90 1.10 Utility Informal 0.14 0.09 0.16 Formal -0.05 0.04 -0.23 Construction Informal 1.81 1.13 0.63 Formal 0.23 0.15 0.85 Trade, accommodation Informal 0.31 0.48 0.07 Formal 1.13 1.72 1.21 Transportation, communication Informal 0.25 1.11 -0.74 Formal 0.29 0.22 0.40 Finance, other business Informal 2.59 3.34 1.54 Formal 1.57 1.79 1.81 Social services Informal 0.69 0.38 0.55 Formal 0.47 -0.17 0.92 Source: World Bank staff’s calculations based on Sakernas 26 Note that GDP data only measures the formal GDP and does not includes the informal GDP. 32 Figure 16 – Employment by formality and 3-sector, 1990-2015 45,000,000 40,000,000 35,000,000 30,000,000 25,000,000 20,000,000 15,000,000 10,000,000 5,000,000 - 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Primary - informal Manufacturing - informal Services - informal Primary - formal Manufacturing - formal Services - formal Source: World Bank staff’s calculations based on Sakernas Employment creation seems to increase, between the 1993-2006 and 2007-2015, in eastern Indonesia than in the more saturated markets of western Indonesia. The employment elasticities in Java-Bali, Sumatera, Nusa Tenggara and Sulawesi decreased during the two periods, while those in Kalimantan, Maluku, and Papua increased (Table 12). This could indicate improved productivity in the more economically active western regions of Indonesia. But increases in employment elasticities in those regions are also partly driven by sharp increases in elasticities in select sectors: agriculture in Papua, mining in Maluku, utility in Kalimantan, and finance and business services and social services in all three regions. Table 12 – Regional employment elasticity, 1993-2006 vs. 2007-2015 Region 1993-2006 2007-2015 (excl 1997- 1999) Java-Bali 0.43 0.25 Sumatera 0.62 0.44 Nusa Tenggara 0.36 0.22 Kalimantan 0.53 0.54 Sulawesi 0.43 0.27 Maluku -0.22 0.60 Papua 0.69 1.26 Source: World Bank staff’s calculations based on Sakernas 33 4. Labor content of exports: Potential sources of demand for labor? Do exports support employment? If manufacturing labor mobility cost is one of the lowest, could the sector be a driver of future employment? One of the important ways that manufacturing could play a bigger role in employment generation is through exports.27 Exports contribute to both direct and indirect wages,28 as well as to the number of jobs created. This section draws on the newly developed World Bank Labor Content of Exports database (LACEX) to explore possible implications of trade for labor-market outcomes in Indonesia. The database comprises information on the labor value-added and the jobs content of exports at the aggregate level, the sectoral level, and exports associated to global value chains. The database has been computed on the basis of a panel of global input-output, national accounts and other aggregate data spanning intermittent years from 1995 to 2011 from the Global Trade Analysis Project (GTAP) and employment data from the International Labor Organization (ILO). We also extend the database for Indonesia using labor force survey data from Sakernas, which are available for more sectors than the ILO data. 29 In 2011 exports generated USD 60 billion in direct and indirect wages in Indonesia, growing four-fold since 2001. Of these just over half were wages paid to people working directly on the exports, while the rest was paid to workers of firms providing inputs to the export sector (Figure 17). Every USD 1000 of Indonesian exports support around USD 324 in wages, split between those who contribute directly (USD 198 for direct wages) and indirectly (USD 126 for indirect wages). This labor intensity of exports is higher than most country comparators, including Malaysia, the Philippines, Thailand and Vietnam and similar to India and South Africa (Figure 18). It is, however, lower than China and Brazil, which are the developing countries with the highest share of wages in exports. Figure 17 – Direct and total labor value added of exports, Figure 18 – Total labor value added of export share, select 1995-2011 countries, 1995-2011 10000 20000 30000 40000 50000 60000 45 40 Total LVAX Share (%) 35 US$ Million 30 25 20 1995 2000 2005 2010 1995 2000 2005 2010 Year Year IDN BRA CHN Direct LVAX Total LVAX IND ZAF Source: World Bank staff elaboration based on LACEX Source: World Bank staff’s elaboration based on LACEX . Note: IDN: Indonesia 27Diop (2016) also argues for a bigger role of manufacturing. 28Labor value added comprises direct and indirect labor value added. Direct LVA are wages paid to domestic employees producing the inputs for the exported products; indirect LVA are wages paid to domestic employees producing the inputs for the exported products. 29 These data are different to those of the previous section to compute the labor mobility costs for Indonesia, which are based on the IFLS. 34 Manufacturing also supported most jobs related to exports . Out of the 21 million jobs supported by exports (directly and indirectly) in 2007, 16 million were supported by manufacturing, almost three quarter of which via backward linkages (Figure 19). In 2007, every USD 1 million of Indonesian exports supported around 163 jobs in 2007, of which 65 are direct and 97 are indirect. This job intensity of exports is higher than that of any comparator countries, except Vietnam. However, it has been declining over time. While each export consignment is associated with a large number of jobs, around one job in five is supported by exports in Indonesia, considerably less than most other East Asian countries. This low share is a by-product of the relative closeness of the Indonesian economy and of the increased importance of relatively unproductive low-skilled jobs in non-tradable sectors de-linked from the export economy. These types of jobs are those that have been generated during the commodity boom of the 2000s.30 However, the number of export jobs has been in decline since 2001. Between 1997 and 2001, the total number of export jobs in Indonesia’s economy increased significantly from 17.8 to 29.5 million, but have been declining since 2001. In 2011, the number of export jobs was only slightly above that in 1997 by about 1 million. Figure 19 – Total number of jobs (direct and indirect) supported by exports (in ‘000), 2007 Manufacturing Agriculture Mining and Energy Transport and communication Trade, sales and accomodation Other commercial services Public services Construction Financial services Electricity, gas and water Real estate, renting 0 5,000 10,000 15,000 Thousands Direct JOBX Indirect JOBX Source: Elaboration on World Bank (forthcoming), Labor Content of Exports database Which sub-sectors contribute significantly to wages and employment? New manufacturing sub-sectors are the sources of labor value added of exports. For example, chemical, rubber and plastic (CRP) products are the largest source of wages among export sectors with over USD 11 billion in wages, followed by processed food (Figure 20). The former category includes rubber, chemical wood pulp and tires exports, while the latter is dominated by palm oil. It is worth noting that over half of the labor value added supported by processed food is actually indirect, a lot of which comes from agriculture, which accounts for 61 percent of the indirect labor inputs to processed foods exports. On the other hand, traditionally important export sectors, including minerals, metals, and energy, lag behind in terms of labor contributions mainly on account of the relatively low labor intensity. 30 World Bank, 2014. 35 Figure 20 – Direct and indirect labor value added of exports, 2011 Chemical, Rubber, Plastic Products Processed Foods Machinery and Equipment nec Minerals nec Metals nec Wood Products Wearing Apparel Textiles Leather Products Paper Products, Publishing Trade and Transport Services Energy Extraction Agr, Forestry, Fisheries Transport Equipment Other Private Services Manufactures nec Metal Products PubAdmin/Defence/Health/Educat Ferrous Metals Mineral Products nec Beverages and Tobacco Products Construction Electricity, Gas, Water Dwellings 0 5,000 10,000 15,000 US$ Millions Direct LVAX Indirect LVAX Source: World Bank staff’s elaboration based on LACEX. The sector most intensely used to provide inputs to exports is trade and transport services, confirming the importance of services for goods’ export competitiveness, and manufacturing in particular .31 USD 8.7 billion in wages in trade and transport services are due to services provided for exports in other sectors, compared to USD 1.6 billion in wages paid directly for trade and transport services exports (Figure 21). The by-product of this is that any restrictions on these services sectors may affect the competitiveness of the downstream manufacturing sectors using these services so intensely. Figure 21 – Direct and indirect labor value added of exports (forward linkages), 2011 Trade and Transport Services Chemical, Rubber, Plastic Products Agr, Forestry, Fisheries Processed Foods Minerals nec Other Private Services Textiles Machinery and Equipment nec Wood Products Energy Extraction Transport Equipment Wearing Apparel Leather Products Metals nec Paper Products, Publishing PubAdmin/Defence/Health/Educat Metal Products Mineral Products nec Construction Manufactures nec Ferrous Metals Electricity, Gas, Water Beverages and Tobacco Products Dwellings 0 2,000 4,000 6,000 8,000 10,000 US$ Millions Direct LVAX Indirect LVAX Source: World Bank staff’s elaboration based on LACEX 31 Duggan et al., 2015. 36 Processed food and CRP, as well as apparel, also support large number of workers . While CRP and processed food have similar overall labor value added in exports, the latter supports a much larger number of export jobs than the former – 4.7 million vs. 2.1 million (Figure 22). This is mainly due to the low-wage jobs supported in agriculture. Both sectors have seen the number of jobs supported by exports more than double between 1997 and 2011, thus compensating the drop in wood products, trade and transport services and agriculture, forestry and fisheries. Figure 22 – Number of jobs in exports across more refined sectors, 1997 vs. 2011 1997 Wood Products Trade and Transport Services Agr, Forestry, Fisheries Processed Foods Wearing Apparel Manufactures nec Chemical, Rubber, Plastic Products Textiles Machinery and Equipment nec Leather Products Energy Extraction Paper Products, Publishing Other Private Services Minerals nec PubAdmin/Defence/Health/Educat Metal Products Metals nec Mineral Products nec Transport Equipment Ferrous Metals Beverages and Tobacco Products Construction Electricity, Gas, Water Dwellings 0 1,000 2,000 3,000 4,000 Thousands Direct JOBX Indirect JOBX 2011 Processed Foods Chemical, Rubber, Plastic Products Wearing Apparel Agr, Forestry, Fisheries Machinery and Equipment nec Wood Products Manufactures nec Paper Products, Publishing Trade and Transport Services Textiles Other Private Services Energy Extraction Leather Products Minerals nec Transport Equipment Metals nec Metal Products Ferrous Metals PubAdmin/Defence/Health/Educat Beverages and Tobacco Products Construction Mineral Products nec Electricity, Gas, Water Dwellings 0 1,000 2,000 3,000 4,000 5,000 Thousands Direct JOBX Indirect JOBX Source: World Bank staff’s elaboration based on LACEX and Sakernas. 37 5. Policy implications Indonesia faces a difficult challenge: it is still dependent on job creation in low-productivity sectors, but its future growth will require a shift toward higher value-added activities in manufacturing and services. Across the three main sectors, the Indonesian economy is shifting toward higher value-added activities. In manufacturing, new sectors, such as food processing, chemicals, machinery, are supporting job creation. These manufactured exports are also supported by services sectors such as transport, utility, and finance. But, labor mobility costs for these skill-intensive sectors are still high. These sectors require different sets of skills than the low value-added services that have absorbed most jobs in the past two decades. Supporting this economic transformation and preparing the labor force to adjust to the new opportunities will entail deep and comprehensive reforms. However, two necessary areas for action are highlighted: improving the provision and quality of skills; and improving the investment climate in poorer and more spatially isolated areas. Firstly, improving the provision and quality of skills is crucial for Indonesian firms and workers to undertake higher-productivity activities. Interviews with firms, as well as evidence quoted in World Bank (2014), confirmed the mismatch between skills of vocational, secondary, and tertiary graduates, and the competences needed in manufacturing industry. However, Indonesian private sector firms also need to invest more in skills development. In 2009, Indonesian firms of all sizes offered fewer training opportunities to their workers, relative to other countries in East Asia Pacific and the world. For example, less than 40 percent of large Indonesian firms provide training opportunities to their employees, as opposed to about 70 percent in East Asia.32 Specific recommendations that have been offered in the past and which are also relevant for reducing labor mobility costs through improved skills development, include33:  Improving the provision of basic skills at the general and vocational high schools. Basic skills, such as mathematics, reading, and science, have been found wanting in graduates of both general and vocational schools. But, improving basic education at the primary level will take years to take effect in the labor force. Therefore, a shorter-term solution is to improve the provision of these skills at the secondary level.  Incentivize firms to provide training opportunities to their workers . Firms are better provider of job- and task-related trainings because they know what are needed from their workers. But, some firms, especially smaller ones, face operational constraints (for example, if they have to stop production) when sending their workers on a training program. Incentives for firms could partly cover these costs and could take the form of subsidies that are partially funded from public training funds.  Subsidize targeted groups for specific trainings. Workers may face financial constraints – transportation costs or loss of income – in accessing privately provided trainings and accreditations. Direct, targeted subsidies for workers could also compensate workers for their travel costs, course fees, as well as possible loss of income. 32 World Bank, 2014. 33 World Bank, 2014. 38 Secondly, there is a need to improve the investment climate, especially in the outer islands (Sulawesi, Maluku, Papua, Nusa Tenggara). This is to address the issue of low labor mobility across the regions. Instead of encouraging workers to move away from their regions, the alternative is to bring investment, and jobs, to those regions. However, due to their lower population density, lack of connectivity, and lack of infrastructure, it remains uneconomical to bring in manufacturing investment to these regions. At the national level, through its economic reform packages, the Government has tried to improve the investment climate, especially by simplifying the processing of investment licenses, reducing the number of licenses, by undertaking Doing Business reforms (ie. for signaling effect). It could build on this momentum to affect further reforms at the sub-national level, by:  Streamlining local business licenses by providing standards for faster application processes for the general trading license (SIUP) and company registration (TDP). Different implementations of local business licenses in the more than 500 regencies and municipalities in Indonesia contribute to making the local investment climate uncertain. Making the application processes for these licenses standardized could reduce this uncertainty.  Focus on transport and logistics infrastructure investment. Although there are other factors that impede the cross-regional mobility of Indonesian workers, improving transport and logistics infrastructure could reduce labor mobility costs. At about 24 percent of GDP, Indonesia’s logistics cost is higher than Thailand (16 percent). 34 However, because infrastructure investments in the above regions tend to be less commercially viable compared to those in Java or Sumatra, the Government needs to provide targeted public incentives. It could, for example, target the use of its Viability Gap Fund to subsidize transport infrastructure investments in eastern Indonesia. Thirdly, assess the labor regulatory framework to ensure that bottlenecks to labor mobility are removed and to allow more fluidity in the labor market. Regulations affecting hiring and firing of workers, severance pay, and wage adjustments, can all have an adverse impact on labor mobility. These policy issues are well known to the Government . In a way, these challenges are also structural in nature and have been difficult to address in the past. However, with the current Government’s focus on regional development, economic liberalization, and infrastructure development, the likelihood that these challenges will be tackled in the near future has increased. There are, however, remaining knowledge gaps. We do not know, for example, what specific ‘mobility costs’ are inherent in labor mobility costs. Because the possibilities are wide – opportunity cost of moving across sectors and regions, networking cost, access to training, transport cost for rural workers, family obligations, lack of apprenticeship opportunity for young workers – it is difficult to come up with specific reform considerations at this point. This study is also unable to address issues related to skills mismatches. 34 World Bank (2015) has detailed recommendations on reducing Indonesia’s logistics cost. 39 References Asia Development Bank. 2010. “Indonesia Skills Report. Trends in Skills Demand, Gaps, and Supply in Indonsia.” Alisjahbana, Armida. 2008. “Education and Skills Mismatch.” Mimeo. Jakarta: World Bank. Artuç, E., D. Lederman, and G. Porto. 2013. “A Mapping of Labor Mobility Costs in the Developing World.” Policy Research Paper No. 6556. Washington, D.C.: World Bank. Calì, Massimiliano, and Milan Nedeljkovic. 2016 (forthcoming). “RER competitiveness and the cost of protectionism: The case of Indonesia.” Jakarta: World Bank. Calì, M., J. Francois, M. Manchin, D. Oberdabernig, H. Rojas-Romagosa and P. Tomberger. 2015. “The Labor Content of Exports.” Mimeo. World Bank. Coxhead, Ian. 2014. “Did Indonesia’s boom leave its poor behind? Adverse labor market trends in the post-crisis era.” University of Wisconsin-Madison. Diop, Ndiame. 2016. “Making Indonesian manufacturing an engine of growth again: Now or never?” Mimeo. Fitriani, Fitria, Leonardo Iacovone, Beata Javorcik, and Gonzalo Varela. 2012. “Policy Note 5 – Productivity Performance in Indonesia’s Manufacturing Sector,” in Rahardja et al., Picking up the Pace: Reviving Growth in Indonesia’s Manufacturing Sector . Jakarta: World Bank. Hollweg, Claire H. 2016 (forthcoming). “Labor Mobility and Structural Transformation in Indonesia.” Jakarta: World Bank. Hollweg, Claire H., Daniel Lederman, Diego Rojas, and Elizabeth Ruppert Bulmer. 2014. Sticky Feet: How Labor Market Frictions Shape the Impact of International Trade on Jobs and Wages. Directions in Development. Washington, D.C.: World Bank. Jϋtting, Johannes P. and Juan R. de Laiglesia. 2009. “Is Informal Normal. Towards More and Better Job s in Developing Countries,” OECD. Tadjoeddin, M.Z. 2015. “Spatial dimensions of key employment outcomes in Indonesia.” Journal of Comparative Asian Development, 14:3, 466-509. Tadjoeddin, M.Z., Widjajanti I. Suharyo, and Satish Mishra. 2001. “Regional disparity and vertical conflicts in Indonesia.” Journal of the Asia Pacific Economy, 6(3), 283-304. World Bank. 2010. Indonesia Jobs Report: Towards Better Jobs and Security for All . Jakarta: World Bank. World Bank. 2012. Policy Note 1: Why the Manufacturing Sector Still Matters for Growth and Development in Indonesia. Jakarta: World Bank. World Bank. 2014. Indonesia: Avoiding the Trap. Development Policy Review . Jakarta: World Bank. World Bank. 2015a. Improving Indonesia’s Freight Logistics System: A Plan of Action . Jakarta: World Bank. 40 World Bank. 2015b. Policy Note 5. Revealing the Impact of Relaxing Service Sector FDI Restrictions on Productivity in Indonesian Manufacturing. Jakarta: World Bank. 41 Annex 1 – Labor mobility cost methodology Methodology The Trade and Labor Adjustment Costs toolkit was designed by the World Bank’s Trade Unit to analyze how labor markets in developing countries adjust to permanent trade-related shocks in the presence of costly labor mobility (see Hollweg et al. 2014). Examples of such shocks are changes in trade policy, whether at home or abroad, and enduring changes in international trade patterns that affect prices in global markets. These are distinguished from transitory shocks and other short-run business cycle fluctuations in that the shocks that result from economy-wide reallocations of labor are not temporary but permanent. As a first step, we estimate labor mobility costs. As a second step, we simulate the economy-wide response of trade shocks and measure labor adjustment costs. See Box A1 for details on the economic model underpinning the Trade and Labor Adjustment Costs toolkit. Box A1: Economic model underpinning the Trade and Labor Adjustment Costs toolkit Labor Mobility Costs are estimated based on observed worker transitions between sectors in response to differences in sectoral wages. Using a structural model of workers’ choice of employment sector, a worker employed in sector chooses to remain employed in sector or move to sector but by incurring a cost (for simplicity we assume the economy has only two sectors). This cost has a common component (average mobility cost caused by labor market frictions that does not vary across similar type workers in an industry), and a worker-specific component , (the differences in the idiosyncratic benefits of being employed in a particular sector relative to sector ) that captures personal circumstances such as family constraints or other preferences. The worker’s expected welfare in sector , , is the present discounted value of her real wage, a sector-specific fixed non-pecuniary benefit, and an option value reflecting the possibility of moving to a different sector with a higher wage. If the wage in sector rises, a worker in sector will experience an increase in welfare due to the higher option value even if she never actually moves. None of the components (wage, sector-specific non-pecuniary benefit, and option value) is specific to the worker, only the sector, whereas the idiosyncratic moving cost is particular to the worker. In each period, the worker decides whether or not to move from his/her current sector to another sector or work status (unemployed or out of the labor force), but incurs a cost to move. The decision to move is based on which sector offers a higher expected welfare net of moving costs. Each worker makes the choice in each period that maximizes his/her lifetime expected utility. The expected welfare benefit of moving from sector i to sector j, ( − ), depends on the wage differential between sectors. The worker will move from sector to sector if the expected welfare benefit of moving ( − ) exceeds the cost of doing so ( + , ), namely if: − ≥ + , . By solving this dynamic rational-expectations model of costly labor adjustment, it is possible to derive an equilibrium condition that is a kind of Euler-equation, which lends itself to estimation. The structural model of sectoral employment choices generates flows of workers across sectors of the economy that depend on the wage gaps between those sectors and the mobility costs of entering a sector. The solution to the model is the employment allocation. The flows of workers across sectors depend on the model’s parameters, inclusive of the mobility costs . It is then possible to estimate these parameters 42 by matching the predicted flows of workers simulated by the model with the observed flows of workers in the data. Labor mobility costs can be estimated from the model using data on observed employment flows and wage differentials between sectors or into and out of employment over time. Employment flows and wage differentials come from panel data on workers’ sector of employment, average sectoral wages, and individual worker characteristics. From observed data on sectoral transitions and sectoral wage gaps, we are able to estimate the labor mobility cost using GMM-type estimations. Simulating the dynamic adjustment paths to the new equilibrium employment-wage outcomes in the affected sector and the remaining sectors of the economy following an exogenous trade-related shock is the next step. The resulting mobility cost estimates represent a key input variable for these simulations. The market-clearing employment and wage path solutions reflect workers’ optimization of their utility dependent on expected wages and costs to change sectors. The analysis uses an equilibrium model in which the structure of the economy is specified using assumptions about the production function in aggregated sectors as well as demand functions and their parameters. Parameters are calculated using data on relative wages, labor and consumption shares, and labor’s share of output across sectors of the economy. The analysis also allows for movements in and out of unemployment or the labor force. The production and demand functions are then used to calibrate the initial steady state of the economy. Labor Adjustment Costs are estimated for each country facing a hypothetical trade-related sectoral shock, and are calculated as the difference in workers’ welfare between the potential post-shock equilibrium outcome with zero labor mobility costs and the actual post-shock equilibrium in the presence of labor mobility costs. The change in relative prices and real wages following the shock will induce some workers to reallocate their labor across sectors. The magnitude of this reallocation depends on the size of the labor mobility costs. The new resulting equilibrium welfare of a worker, , is compared to her initial pre-shock welfare, , and her potential maximum welfare, , if mobility costs were zero. The maximum potential gains to trade (PG) are therefore − , and actual gains to trade (G) are − . Labor adjustment costs (LAC) representing the forgone welfare gains to trade due to labor mobility costs are therefore: LAC = PG – G = Vmax – V. Source: Artuç et al. (2013) 43 Data Data to estimate labor mobility costs in Indonesia comes from the Indonesia Family Life Survey (IFLS). The IFLS is an ongoing longitudinal survey in Indonesia. The sample is representative of about 83 percent of the Indonesian population and contains over 30,000 individuals. The first wave of the IFLS was conducted in 1993/94 (IFLS 1), the second wave in 1997 (IFLS 2), the third wave in 2000 (IFLS 3), and the fourth wave in 2007/08 (IFLS 4). To measure an individual’s sector of employment we concentrate only on primary occupation of the individual. The IFLS divides sectors of employment into 9 aggregate sectors based on ISIC rev. 3. We consider only individuals aged 15 years old and above. The sectors are listed as follows: i. Agriculture, forestry and fishing ii. Mining and quarrying iii. Manufacturing iv. Electricity, gas and water v. Construction vi. Trade, restaurants and accommodation services vii. Transport, storage and communication viii. Financial institution, real estate, leasing and services company ix. Services community, social and individual Sector of employment was further aggregated into primary (agriculture, forestry and fishing; mining and quarrying), manufacturing and services. After determining sectors of employment, we created a residual category, called Unemployed/Out-of- labor force. It consists of individuals who are at least 15 years old and did not have a job in the past week. Skill level was determined by using an individual’s higher level of education attended, grouped into three categories: unskilled, vocation and skilled. Vocational are individuals who completed senior high vocational or college (D1, D2 or D3). Skilled are individuals who completed university (Bachelor’s, Master or Doctorate) or open university. Other individual-level characteristics include gender and age. An individual was considered young if between the ages of 15 and 30, middle-age if between the ages of 31 and 49, and old if 50 and over. We also looked at transitions across 8 geographic regions. IFLS 1 and IFLS 2 originally sampled in 13 provinces in Indonesia, which was increased to 21 provinces by IFLS 4. These were grouped into: Sumatra (Aceh, North, West and South Sumatra, Riau, Lampung, Bangka-Belitung Islands), Jakarta, West Java (West Java and Banten), Central Java (Central Java and Yogyakarta), East Java, Lesser Sunda Islands (Bali and West Nusa Tenggara), Kalimantan (West, Central, South, East Kalimantan), and Sulawesi North Sulawesi and South Sulawesi (North and South Sulawesi). We also looked at transitions across sectors as well as into/out of urban or rural areas and formal or informal jobs. Formal workers are those that are self-employed with permanent workers, government workers or private workers. Informal workers are own account or self-employed with unpaid family workers, casual workers, or unpaid family workers. 44 Income was also based on the individual’s primary occupation. Income was calculated as the individual’s salary and wages during the last year including the value of all benefits, or business profit last year (after taking out all expenses). Income was then deflated by purchasing power parties across rural and urban areas and provinces, based on the national poverty line within the year. For the simulations we need data on average sectoral wages, share of labor in each sector, sectoral consumption shares, and the wage bill in the sector’s value added or output. Average sectoral wages and labor shares were calculated form the IFLS. The wage of unemployed/out of the labor force was assumed to be zero. The sector’s share in household consumption and the wage bill in the sector’s value added comes from the social accounting matrices of the Global Trade Analysis Project (GTAP). 45 Annex 2 – Labor content of exports methodology The LACEX database has been computed by Calì et al. (2015) on the basis of a panel of global input- output and other aggregate data spanning intermittent years from 1995 to 2011 from the Global Trade Analysis Project (GTAP) and employment data from the International Labor Organization (ILO). 35 The input-output tables in the GTAP dataset allows one to exploit a form of social accounting data – a variation on the social accounting matrix (SAM) where incomes are shown in the rows of the SAM and expenditures are shown in the columns (see Hertel (2013) and McDougal (2001)). The structure of the underlying social accounting data provides a comprehensive and consistent record of national income accounting relationships between different sectors, including intermediate and final demand linkages. These are used to construct country-specific measures of the direct and indirect contribution of labor to the value added contained in a given country’s domestic production and exports. 36 The resulting dataset covers 24 sectors (6 services sectors, 3 primary sectors and 15 manufacturing sectors), over 100 countries, and intermittent years between 1995 and 2011. Specifically, in order to obtain these labor value added measures, two intermediate multiplier matrixes need to be calculated from the social accounting and other aggregate data. The first is the Leontief inverse matrix, which measures the inputs contained in a unit of final output, and includes both direct and indirect inputs across all sectors of the economy. The second is a diagonal matrix with elements equal to the compensation of employees’ shares of the sector’s total output. Using these two intermediate matrixes as multipliers of a diagonal matrix with elements equal to either the sector’s domestic production or exports, one can obtain the compensation of employees’ shar es of final outputs or exports. Using the same method, we can isolate the jobs content of exports by replacing the underlying labor value added share in production with the number of jobs contained in production, which can be derived from ILO employment data or country-specific data sources.37 ILO employment statistics contain data for a number of countries for 11 macro sectors. In order to get a more refined sectoral disaggregation, we also match the GTAP data with employment statistics from the labor force surveys, resulting in 16 sectors. However, due to issues with matching sectors, we cannot separate mining from some manufacturing activates, including metals and ferrous metals from machinery and metal products. With this methodology we can split the total contribution of labor (either the labor value added or the number of jobs) to final output and exports into its direct and indirect components based on the linkages with the rest of the economy. The direct component measures a sector’s domestic labor contri bution embodied in its own domestic production and exports. It is the wages paid or number of jobs a sector uses to produce its own output or exports directly. The indirect component measures a sector’s domestic labor contribution embodied as inputs to (considering forward linkages) or from (considering 35 GTAP represents a massive combined effort of international institutions and universities. Over time, the dataset has grown to include more countries and more sectors. The latest version, GTAP 9, has data on 129 countries/regions and 57 sectors (Narayanan, 2012). To maintain backward compatibility, Calì et al. (2016) start with the 1997 structure of regions and sectors, and carry this forward in aggregation of more recent iterations of the dataset. The GTAP website also provides extensive documentation on the underlying data structure, its sources, and the GTAP model structure for each release (www.gtap.org). 36 In the remainder of the paper we refer interchangeably to total labor value added content, wages and compensation to employees. 37 More formally, in terms of the model of Appendix 1, one needs to replace the B ̂ matrix, which is a diagonal ̂ matrix with the J matrix indexed over i, j with diagonal elements equal to the number of jobs in output Z. 46 backward linkages) other sectors’ domestic production and exports. It is the wages paid or number of jobs a sector contributes to other sectors’ output or exports – if measured by forward linkages – or the wages paid or number of jobs a sector uses from other sectors to produce its output or exports – if measured by backward linkages. These contributions can be further split between skilled and unskilled employees. Forward linkages are the indirect contribution of a sector when considering the contribution of that particular sector as an input to other sectors’ domestic production or exports. This treats the particular sector as an upstream activity, and measures how much labor a particular sector is used by all other sectors. Backward linkages are the indirect contribution of a sector when considering the contribution of all other sectors to that particular sector’s value added. This treats the particular sectors as a downstream activity, and measures how much labor a particular sector uses from all other sectors. The labor value added of exports is measured as nominal values in US$ (denoted LVAX), and as a share of gross exports (LVAX share), which is a measure of the labor intensity of a country’s or sector’s exports. The jobs content of exports is measured as the number of jobs (denoted JobX), and relative to gross exports (JobX share), which is a measure of the job intensity of a country’s or sector’s exports. We also use measures of the labor value added and job content of domestic production (denoted LVAD and JobD) for comparisons. 47