1 Structural Transformation and Labor Productivity in Indonesia: where are all the good jobs? Frederico Gil Sander and Pui Shen Yoong1 This version: February 14, 2020. Abstract This paper focuses on the role of structural transformation in creating ‘good jobs’ that provide middle- class wages in Indonesia. Labor productivity growth results from improvements in productivity within sectors of the economy, as well as from structural transformation – the shift of capital and labor to more productive sectors. Building on McMillan and Rodrik (2011) and McMillan, Rodrik and Sepulveda (2017), we disaggregate labor productivity growth to assess the contributions of within-sector productivity growth and of structural transformation. We then further distinguish between the sources of structural transformation from the relative growth of employment in more productive sectors, and from the productivity differences between sectors that gain and lose jobs. We find that in Indonesia’s case, labor productivity growth has mostly been driven by within-sector improvements, rather than structural change. This is because most new jobs have been in low value-added services, which are not much more productive compared to sectors shedding jobs (largely agriculture). Both demand-side and supply-side interventions are needed to boost labor productivity so that more Indonesians can have middle-class jobs. JEL Classification: E24, J23, O11, O40 Keywords: Structural change, structural transformation, labor productivity, labor demand 1 Authors and affiliations: Frederico Gil Sander (Lead Economist) and Pui Shen Yoong (Economist), World Bank Macroeconomics, Trade and Investment Global Practice. This paper is written as a background paper for Indonesia’s Jobs Report, Pathways to Middle-Class Jobs in Indonesia (2021). We would like to thank Maria Monica Wihardja (World Bank, Poverty and Equity Global Practice), Wendy Cunningham (World Bank, Social Protection and Jobs Global Practice) and Massimiliano Calì (World Bank MTI Global Practice) for inputs, and Mercoledi Nasir and Abror Perdana for research assistance. The views expressed in this paper are of the authors only and should not be attributed to the World Bank, its Executive Directors, or the countries they represent. 1 2 1. Introduction By some measures, the Indonesian labor market has never looked better. Underpinned by sound macroeconomic policies, steady economic growth of about 5 percent per annum over the past decade was associated with strong job creation. Between 2008-2018, the economy created 2.1 million jobs annually on average, and 240,000 Indonesians ceased to be unemployed each year. As a result, the employment rate climbed to 64 percent of the working-age population (defined as those aged 15 years and above) while unemployment has declined from 11.2 percent in 2005 to 5.4 percent (Figure 1). Yet most Indonesians do not have ‘good jobs’ that provide an entry point to the middle class (Figure 2 and Box 1). Although most Indonesians have a job, the majority do not earn – and hence do not consume – enough to be considered part of the middle-class2, a group that enjoys economic security from poverty and vulnerability. For example, in the trade, hotel and restaurant sector, which accounts for a quarter of all employment, less than a tenth of all workers earn a ‘middle-class wage’ – defined as at least IDR 3.7 million per month (US$260). More broadly, only 40 percent of workers are salaried or ‘formal’3 – a slow evolution from 30 percent in the early 2000s. Even then, half of all salaried employees do not have a written contract, and employer compliance with severance pay and social security benefits is limited4. This paper focuses on one driver of the quality of jobs in Indonesia: labor productivity growth and, in particular, the (limited) contribution of structural transformation. It shows how structural change – here defined as the reallocation of workers from low- to high-productivity economic activities – has contributed only a small share of labor productivity growth in the recent two decades. The main reason is that most 2 Following World Bank (2019), middle-class households are those that consume 3.5-17 times the national poverty line per month. 3 This follows the simplified definition of ‘formal’ jobs by Statistics Indonesia, i.e. employers with permanent workers and wage employees. The share of formal jobs is higher (50 percent of workers) following the official definition, which uses a combination of work status and occupation to define these terms. 4 Data from 2007-2008 suggest that workers receive only 10–14 percent of the severance payment due (Alatas et al. 2011; World Bank 2010). 2 3 new jobs created are in in low-Value Added (VA) services, which are not much more productive than those in agriculture, the sector that has been shedding jobs. The main takeaway is that Indonesia need not worry as much about the quantity of jobs as the quality of those jobs. Both demand-side and supply-side interventions are needed to boost labor productivity so that more Indonesians can have middle-class jobs. Figure 1: Most Indonesians are employed… Unemployment rate and GDP growth, percent; Employment rate, percent (secondary axis) 14 employment rate (right-hand side axis) 70 12 60 10 50 8 unemployment rate 40 6 30 4 20 2 gdp growth 10 0 0 Source: BPS, SAKERNAS (August 2018) Figure 2: …but not many have a good job Share of all workers with a middle-class wage, % 50 45 Share of all workers with at least a middle-class wage 40 35 Sector share of all employment 30 25 20 15 10 5 0 Source: World Bank staff calculations from SAKERNAS (August 2018) Note: See Annex for more details and middle-class wage definition. 3 4 Box 1: What is a good job? Definitions of a good job vary depending on the country context, but there are some general principles. Rodrik and Sabel (2019), for example, define a good job as “stable, formal-sector employment that comes with core labor protections such as safe working conditions, collective bargaining rights, and regulations against arbitrary dismissal�. Some definitions are broader than others; the International Labor Organization’s ‘Decent Work’ agenda urges the creation of “opportunities for women and men to obtain decent and productive work in conditions of freedom, equity, security and human dignity�. The OECD’s Good Jobs Strategy (OECD, 2018) similarly looks at indicators of labor market inclusiveness, but also adaptability and resilience to shocks and new opportunities. These definitions range in their scope because it depends, in part, on the question: good jobs for who/what? While good jobs may simply be “those that provide greater well-being to the people that hold them�, the societal value of a job is much broader (World Bank, 2013). For practical reasons, this note follows a narrower, more ‘individualistic’ definition that focuses solely on formality and wages. The premise is that formal, higher-paying jobs tend to facilitate ascension to the middle class and provide greater worker protection than informal, lower-paying jobs. It is recognized, however, that this is not always the case, and that other non-monetary measures of employment quality should also be evaluated. Source: Authors 2. The evolution of work and workers over the past two decades Jobs have played a central role in Indonesia’s ascension from a poor to a lower middle-income country. The drastic decline in the poverty rate5 – from 64 percent of the population in 1990 to 5.7 percent in 2017 – was largely driven by the creation of jobs in industry and services sectors, which boosted household 5 Using the international poverty line of US$1.90 a day (2011 PPP). 4 5 incomes by paying more than agricultural work. The diversification of livelihoods away from agriculture towards industry and services corresponded to a shift in the structure of the economy (Figure 3 and Figure 4). Today, agriculture only accounts for 13 percent of output and about a third of employment, far less than it did in 1995 (Table 1). The services sector has displaced agriculture as the main economic engine, generating about half of output and employment today. Meanwhile, the shares of output and employment in industry have remained constant, as increases in construction and other industrial activity compensated for the decline in manufacturing over 1995-2018. Figure 3: The economy diversified away from Figure 4: …but shifts in labor have been more agriculture into industry and services… dramatic than shifts in output Share of total value added, percent Share of total employment, percent 100 100 Public sector Public sector 90 90 80 80 High VA High VA 70 sevices services 70 60 Low VA Low VA 60 services 50 services 50 Other industry 40 Other industry 40 30 30 Manufacturing 20 Manufacturing 20 Agriculture 10 10 0 Agriculture 0 Source: BPS via CEIC, World Bank staff calculations Source: SAKERNAS, World Bank staff calculations Note: High value-added services are financial, real estate, business & insurance services (henceforth abbreviated as “financial services�), transport & storage, information and communications (henceforth abbreviated as “transport and communications�). Low value-added services are wholesale and retail trade/motor vehicle repairs, accommodation & food services (“trade, hotels and restaurants� for short), and community, personal and other services. ‘Other industry’ comprises mining, utilities and construction. 5 6 Table 1: Low-end services has displaced agriculture as the main source of jobs Share of total value added, percent Share of total employment, percent 1995 2018 1995 2018 Agriculture 17 13 44 28 Industry 42 42 25 23 …o/w manufacturing 24 21 19 15 …o/w other industry 18 21 6 8 Services 41 45 31 49 …o/w low value-added 20 23 23 36 …o/w high value-added 15 19 5 8 …o/w public administration 6 4 3 5 Source: BPS and Sakernas, World Bank staff calculations See note above. Over the past two decades, shifts in labor have been more dramatic than shifts in output. The share of workers employed in agriculture declined by about 16 percentage points between 1995-2018 – four times more than the fall in the share of agricultural output over the same period (Table 1). This reflects an increase in agricultural labor productivity, or real value added per worker, of 3 percent per annum on average. Meanwhile, the share of workers in low-VA services increased by 13 percentage points over the same period. This was far more than the increase in its share of output, suggesting declining productivity in the sector. Overall, these low VA services sectors created 46 percent of all new jobs over 2000-2018. In contrast, the share of Indonesian workers employed in high VA services sectors only increased by 3 percentage points over the same period6. Nevertheless, because the productivity level in low-VA services is higher than in agriculture (Figure 5), this relative shift still helped boost family incomes. 6 This is the case in spite of high employment elasticity. For every 1 percent growth in the output of financial services, employment in the sector grows by 1.5 percent, above the economy-wide average elasticity of 0.4 percent. Source: Wihardja (2017), unpublished. 6 7 Sectors with higher levels of labor productivity tend to pay higher wages, but only a minority of Indonesians work in these sectors. As Figure 5Figure 6 shows, there is a positive association between the relative productivity of a sector (measured as the log of the ratio between the sector’s labor productivity and average labor productivity of the economy) and median monthly wages. A simple bivariate regression indicates that a one percent increase in the relative productivity of the sector is associated with a 0.24 percent increase in median monthly wages. However, the most productive sectors – mining and quarrying, and financial services – employ only 4 percent of all Indonesian workers. Most Indonesian workers – 64 percent – are engaged in agriculture and low-VA services. This has important implications for the quality of jobs, as employment in these sectors tends to be informal and less likely to pay middle-class wages. Figure 5: More productive sectors tend to pay better wages, but employ a minority of workers 2.5 Real median monthly wage in2018, community, personal, financial services and other services 2.0 utilities IDR millions 1.5 transport and mining and quarrying communications 1.0 agriculture manufacturing construction 0.5 trade, hotels, restaurants 0.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 Log of sector productivity as a share of total productivity, 2018 Source: Sakernas, August 2018, World Bank staff calculations Note: Public sector is included in community, personal and other services. Size of the bubble denotes the sector’s share of total employment in 2018. Labor productivity is calculated as real output (value added) per worker. The national median monthly wage is IDR1.2 million. 7 8 The supply of skills has increased over the past two decades but the share of the workforce with tertiary education remains small. As the Government intensified efforts to increase primary and secondary enrollment rates, the domestic workforce has become more educated. 60 percent of the labor force had completed at least some secondary education in 2018, compared to 40 percent in 2000. Forty percent of the labor force had at least an upper secondary or vocational education in 2018, significantly higher than 24 percent in 2000. The share of the labor force with some form of tertiary qualification (diploma I-IV) also increased to 12 percent, compared to only 5 percent in the early 2000s (Figure 6). Nonetheless, the share of the labor force with tertiary education remains low compared to many other emerging and developing economies (Figure 7). Figure 6: The Indonesian workforce has become more educated on average… Share of labor force aged 15+, percent 100.0 90.0 80.0 Tertiary 70.0 60.0 Upper secondary 50.0 40.0 Lower secondary 30.0 20.0 10.0 Primary 0.0 No schooling Source: Sakernas, various years Figure 7: …but the share of the workforce with a tertiary education is still low Share of labor force aged 15+ with tertiary education, percent (latest available) 35 28.6 29.9 30 28.3 25.5 25 20.7 20 15.9 16.4 15 12.2 12.5 10 5 0 Indonesia Viet Nam South Africa Thailand Brazil Philippines Malaysia Colombia Peru Source: ILOSTAT, World Bank staff calculations 8 9 Although they are few, highly educated workers tend to have good jobs. Workers with a tertiary education tend to have, on average, favorable labor market outcomes. Over 80 percent of those who are working are engaged in formal, salaried jobs, mostly as employees in the services sector. These workers tend to be employed in high value-added services (especially financial and business services) and public sector jobs, which are knowledge-intensive sectors more likely to pay middle-class wages. They are also more likely to work in high-skilled jobs, i.e. as professional, managerial or technical workers. Indeed, the average monthly wages of those with tertiary education (Diploma I-IV or higher) put them firmly in the middle class (Figure 8), as they earn on average 1.5 times as much as those with only a high school education. For wage employees, the return to an additional year of schooling is 8.7 percent on wages7. Figure 8: Wages of tertiary-educated workers place them well into the middle class Average monthly real wages of workers by highest certificate, IDR millions Diploma IV/university or higher 4.6 Diploma III 3.7 Diploma I/II 3.6 Vocational high school 2.7 General high school 2.7 Middle school 2.0 Elementary school 1.8 Did not complete/not yet completed elementary… 1.5 No schooling 1.2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Source: Sakernas, August 2018, World Bank staff calculations The labor market outcomes of those with only a secondary education are much more mixed. Although most of the secondary-educated labor force can find jobs, the unemployment rate for this group is relatively high (8 percent) compared to other educational groups and to the national unemployment rate8. Even among those who are employed, many are working in precarious, non-salaried jobs. Only about half 7Mincer regression estimates by Wihardja and Alatas (2018, unpublished). 85.6 percent of the tertiary-educated labor force and 1.9 percent of the primary-educated labor force are unemployed. Only 1.5 percent of the uneducated labor force is unemployed. 9 10 have formal jobs, mostly as employees in the services sector and manufacturing; the remainder are about equally divided between those who are self-employed or in casual/unpaid family work. These statistics suggest that the skills acquired from secondary education are insufficient to gain higher quality, better- paid employment, especially for those who only complete the lower secondary years. There may also be an issue of low demand for mid-level skills that could be performed by secondary-educated workers, given Indonesia’s limited reliance on sectors such as labor-intensive manufacturing for exports and tourism. Many secondary-educated Indonesians – especially women – have opted out of the labor force entirely. Overall, 22 percent of working-age Indonesians with a secondary education are not working nor in school (Figure 9), a share that has increased by 1.8 percentage points since 2010. About 2/3 of this increase can be attributed to secondary-educated women residing in urban areas, especially younger women aged 20- 44. This group contributes significantly to the low female labor force participation (LFP) in Indonesia, which has remained at about 50 percent over the past three decades – below the regional average of 61 percent (Figure 10). The gap between LFP rates of men and women is highest during these prime working years (35 percent), when individuals are likely to be most productive. There are many potential reasons for the rise in inactive, relatively educated women in urban areas, for example, migration to urban areas where the supply of childcare services is more limited9; increases in wealth that make it less critical for women to contribute to household income; few jobs available for secondary-educated women that pay enough to offset costs of childcare; and evolving gender norms. Further research is needed to understand the reasons why relatively educated women are not seeking employment, especially in urban areas. 9In rural areas, women may be more likely to live with extended families who provide childcare. Improved access to childcare may increase female LFPR in Indonesia by as much as 13 percentage points (Halim et al., 2019). 10 11 Figure 9: A quarter of those with only a secondary education are inactive/unemployed Share of population aged 15+, percent 100 90 student 80 70 60 not studying/not working 50 40 30 unemployed 20 10 0 employed secondary tertiary Source: Sakernas, 2017 Figure 10: Only half of all women are in the labor force; a share that has not changed since 2008 Share of female population aged 15+, percent 80 Indonesia Malaysia Thailand 75 Vietnam Philippines Developing EAP 70 65 60 55 50 45 40 Source: WDI, World Bank staff calculations 11 12 Taken together, these trends suggest that there are both demand- and supply-side constraints to improving the quality of jobs. On the demand side, sectors that are absorbing the most labor are the least productive. On the supply side, most workers have at best a lower secondary education. In addition, 56 percent of today’s labor force is estimated to have been stunted as children, which impairs cognitive development10 and further constrains the creation of domestic jobs in high-value added services sectors. Both these trends dampen the growth of labor productivity in Indonesia, eroding the quality of jobs and growth in GDP per capita. To understand how these trends can be adjusted, the next section delves deeper into the ‘demand’ side of the problem, looking at the contribution of structural transformation to labor productivity over the past two decades. 3. Higher labor productivity: pre-requisite for good jobs As argued earlier, labor productivity is a key determinant of wages and the quality of jobs overall. At the aggregate level, increases in labor productivity can occur in two ways: (i) within-sector productivity improvements, which occur as workers and firms in a given sector become more efficient. This can occur through capital deepening or upgrading, technological change, or the reduction of misallocation of resources, i.e. as less profitable firms or plants exit and are replaced by more efficient ones. For example, aggregate labor productivity increases when labor productivity in manufacturing rises due to the entrance of more innovative firms or the use of more efficient machines and equipment; 10 See Galasso and Wagstaff (2017). 12 13 (ii) structural change, which occur as workers (and other inputs of production) move from lower- to higher-productivity sectors of the economy. For example, aggregate labor productivity rises when workers move from agriculture to manufacturing, which generates more value added per worker. This is also called the ‘intersectoral’ effect. Mathematically, this can be expressed by Equation 1, following McMillan and Rodrik (2011): ∆𝑃1−0 = ∑𝑛 𝑖 𝑖 𝑖 𝑛 𝑖 𝑖 𝑖 𝑖=1(�1 − �0 ) ∗ 𝜃0 + ∑𝑖=1(𝜃1 − 𝜃0 ) ∗ �1 (Eq. 1) 𝑖 where P is the economy-wide labor productivity level, �𝑡 is the sectoral real labor productivity level of 𝑖 sector i in period t=0,1, 𝜃𝑡 is the share of total employment in sector i in period t=0,1. The first term – changes in productivity across time, weighted by initial employment shares – is the within- sector effect, which is positive as firms in the sector become more efficient on average. The second – changes in employment shares across time, holding productivity constant in the end period – is the structural change effect, which is positive when changes in employment shares are positively correlated with productivity levels (i.e. when workers move, on average, to sectors that are more productive). The structural change component can be further disaggregated into ‘between’ and ‘cross’ effects11 (see McMillan, Rodrik and Sepulveda, 2017) to distinguish the contributions of changes in employment and changes in productivity to aggregate labor productivity growth. With no change to the within-sector term, this can be expressed by modifying Equation 1 as follows: ∆𝑃1−0 = ∑𝑛 𝑖 𝑖 𝑖 𝑛 𝑖 𝑖 𝑖 𝑛 𝑖 𝑖 𝑖 𝑖 𝑖=1(�1 − �0 ) ∗ 𝜃0 + ∑𝑖=1(𝜃1 − 𝜃0 ) ∗ �0 + ∑𝑖=1(𝜃1 − 𝜃0 ) ∗ (�1 − �0 ) (Eq.2) The three terms on the right-hand side of Equation 2 are the within, between and cross-sector components respectively. The ‘between’ term reflects shifts in workers across sectors, weighted by sector 11 de Vries, Timmer and de Vries (2015) call these the static and dynamic components of structural change, respectively. 13 14 productivity in the initial period. At the sector level, when this term is positive (negative), it means that workers are moving into (out of) the sector. When aggregated across all sectors, this term will be positive if on average workers moved into sectors that were more productive at the outset compared to sectors that lost workers. At the sector level, the cross term will be positive when the sector has gained workers and productivity grew, or lost workers and productivity declined. When the term is negative, it could mean that a sector is shedding workers and experiencing increases in labor productivity at the same time, or that a sector is increasing its share of employment but experiencing decreases in labor productivity over time. As McMillan, Rodrik and Sepulveda (2017) note, the former is not necessarily a ‘bad’ development, especially if the sector in question is agriculture. At the aggregate level, the ‘cross’ term simultaneously reflects changes in the shares of employment and changes in productivity. When this term is positive, it indicates that workers have, on average, moved to sectors with higher productivity growth over the time period (hence the alternative terminology of ‘dynamic’ effect). The decomposition of the structural transformation term into between and cross components therefore allows us to see whether structural transformation was driven primarily by dynamic sectors where productivity growth was faster. It may also be useful to ask whether structural transformation is being driven by the relative growth of employment in a particular sector, or by the difference in productivity levels between the sectors that gain and lose jobs. The magnitude of the structural transformation component can also be understood as a function of the share of workers that move from low- to higher-productivity sectors (the larger the number of workers that move, the larger the contribution from structural transformation) and the difference in productivity levels between sectors (the larger the productivity gap between sectors losing and gaining workers, the larger the contribution of structural transformation for a given shift of workers). To better understand these drivers of structural transformation, we modify the between and cross terms in Equation 2. We replace the temporal dimension (i.e. productivity in the beginning of the period) with the cross-sectional dimension (i.e. the sector’s productivity relative to that of agriculture in the final 14 15 period; agriculture is chosen as a benchmark sector given that structural transformation tends to involve switches out of agriculture). With no change to the within-sector component, the equation can be rewritten as follows: ∆𝑃1−0 = ∑𝑛 𝑖 𝑖 𝑖 𝑛 𝑖 𝑖 𝐴 𝑛 𝑖 𝑖 𝑖 𝐴 𝑖=1(�1 − �0 ) ∗ 𝜃0 + ∑𝑖=1(𝜃1 − 𝜃0 ) ∗ �1 + ∑𝑖=1(𝜃1 − 𝜃0 ) ∗ (�1 − �1 ) (Eq. 3) 𝑖 𝑖 Because the weights (𝜃1 − 𝜃0 ) sum to zero and the level of productivity in the reference sector is fixed, the aggregate ‘between’ effect across all sectors is also zero. However, for any given sector, the ‘between’ term gives an indication of the extent the sector gained workers, which can then be compared with the contribution from the ‘cross’ term, which indicates the extent the sector is more productive than agriculture. We call this modified cross term the ‘productivity difference’. These modified terms now enable us to understand the relative contributions of (i) the direction and magnitude of labor reallocation, and (ii) the productivity gains that occur when workers move across sectors – which solely depends on the productivity differential of the destination sector relative to agriculture. For example, as shown in Table 3a below, both the manufacturing sector and the trade, hotels and restaurant sector (THR) made the same contribution to structural transformation in Indonesia over 2000-2018. However, the larger, ‘between’ component for THR reflects the fact that relatively more workers moved to the THR sector, while the larger ‘productivity difference’’ component in manufacturing reveals that productivity gains from moving to manufacturing was higher. Applying these decompositions to Indonesian data, we find that labor productivity growth has mostly been driven by within-sector improvements, rather than the reallocation of resources from low- to high- productivity sectors (‘structural change’). These results are broadly consistent with what is observed in other developing economies. Between 2000-2018, structural change only contributed less than a third of 15 16 overall labor productivity growth, or 1.0 out of 3.4 percentage points per annum12. This was a broadly similar proportion to the contribution of structural change in Thailand and the Philippines over the same period (Figure 11), and in line with all middle-income countries on average13, but less than in Vietnam, where nearly half of total labor productivity growth was due to structural change. Figure 11: Within-sector changes have driven most of Indonesia’s labor productivity growth, similar to other countries in the region Contribution to labor productivity growth, annual average from 2000-2017, percentage points Within Sector Structural change 120 100 11.7 28.7 23.0 80 47.5 38.6 60 40 20 0 -6.8 Indonesia Malaysia Vietnam China Thailand Philippines -20 Source: World Bank staff calculations using data from WDI, BPS and Sakernas (for Indonesia) Labor productivity improvements in the services sector have driven Indonesia’s labor productivity growth over the last two decades. Efficiency gains in the services sector contributed about 37 percent of total labor productivity growth over 2000-2018 or 1.2 percentage points per annum (Table 2). The contribution to growth was larger in the first half of that period (2000-2010) than in the second half (2010- 2018), perhaps indicative of gains seen in non-tradable sectors during the commodity boom. The agriculture sector has also experienced productivity gains, contributing 25 percent of total labor productivity growth or 0.8 percentage points per annum over the same period. Its contribution was not 12 Rogerson (2017), ADB (2018) and Kim et al (2018) also arrive at similar results for Indonesia, albeit for slightly different time periods. 13 Moving across sectors contributes almost two-thirds of productivity growth for low income countries, but closer to one third for middle income countries. See Merotto, Weber and Aterido (2018). 16 17 as much as in China and Vietnam, but more than in Philippines, Malaysia and Thailand (Structural change contributed positively to labor productivity growth (and hence to overall economic growth) in Indonesia, but not as much as in some other East Asian countries. Structural change in the industry and services sectors contributed equally (about 22 percent respectively) to overall labor productivity growth over 2000-2018. This reflects the movement of workers away from agriculture into industry and services, which have relatively higher average labor productivity. In overall magnitude, the contribution of structural change to labor productivity growth is smaller in Indonesia than in Thailand and Vietnam but higher than in Malaysia and Philippines. There are two reasons why this is the case. First, Vietnam moved a larger proportion of workers out of agriculture into other sectors: from 2000-2017, the share of employment in agriculture shrunk by 1.9 percentage points respectively per annum, more than the 0.9 percentage points in Indonesia over the same period. Second, average productivity levels are higher in Vietnam’s industry sector and Thailand’s services sector compared to the corresponding sectors in Indonesia (Error! Not a valid bookmark self-reference.). Figure 12: Structural change does not contribute as much to labor productivity in Indonesia as in regional peers, due to smaller reallocations away from agriculture and lower productivity of industry and services ). Finally, within-sector industry gains only made a small contribution to overall labor productivity growth (10 percent or 0.3 percentage points per annum), especially compared to most other countries in the region (Structural change contributed positively to labor productivity growth (and hence to overall economic growth) in Indonesia, but not as much as in some other East Asian countries. Structural change in the industry and services sectors contributed equally (about 22 percent respectively) to overall labor productivity growth over 2000-2018. This reflects the movement of workers away from agriculture into industry and services, which have relatively higher average labor productivity. In overall magnitude, the contribution of structural change to labor productivity growth is smaller in Indonesia than in Thailand and Vietnam but higher than in Malaysia and Philippines. There are two reasons why this is the case. First, Vietnam moved a larger proportion of workers out of agriculture into other sectors: from 2000-2017, the 17 18 share of employment in agriculture shrunk by 1.9 percentage points respectively per annum, more than the 0.9 percentage points in Indonesia over the same period. Second, average productivity levels are higher in Vietnam’s industry sector and Thailand’s services sector compared to the corresponding sectors in Indonesia (Error! Not a valid bookmark self-reference.). Figure 12: Structural change does not contribute as much to labor productivity in Indonesia as in regional peers, due to smaller reallocations away from agriculture and lower productivity of industry and services ). As shown in Table 3a below, firms operating in extractives and utilities became less efficient in their use of inputs over the last two decades. In fact, within-sector changes in industry have barely made a dent in labor productivity growth in recent years, contributing only 0.1 percentage points per annum over 2010- 2018. Table 2: Efficiency gains in services have contributed the most to labor productivity growth 2000-2018 2000-2010 2010-2018 Average annual labor productivity growth 3.4 3.4 3.4 Within-sector 2.4 2.6 2.2 …o/w agriculture 0.8 0.6 0.9 …o/w industry 0.3 0.6 0.1 …o/w services 1.2 1.5 1.2 Structural change: 1.0 0.8 1.2 …o/w agriculture -0.6 -0.3 -0.6 …o/w industry 0.8 0.5 1.0 …o/w services 0.8 0.6 0.8 Source: World Bank staff calculations using data from WDI, BPS and Sakernas (for Indonesia) Structural change contributed positively to labor productivity growth (and hence to overall economic growth) in Indonesia, but not as much as in some other East Asian countries. Structural change in the industry and services sectors contributed equally (about 22 percent respectively) to overall labor 18 19 productivity growth over 2000-2018. This reflects the movement of workers away from agriculture into industry and services, which have relatively higher average labor productivity. In overall magnitude, the contribution of structural change to labor productivity growth is smaller in Indonesia than in Thailand and Vietnam but higher than in Malaysia and Philippines. There are two reasons why this is the case. First, Vietnam moved a larger proportion of workers out of agriculture into other sectors: from 2000-2017, the share of employment in agriculture shrunk by 1.9 percentage points respectively per annum, more than the 0.9 percentage points in Indonesia over the same period. Second, average productivity levels are higher in Vietnam’s industry sector and Thailand’s services sector compared to the corresponding sectors in Indonesia (Error! Not a valid bookmark self-reference.). Figure 12: Structural change does not contribute as much to labor productivity in Indonesia as in regional peers, due to smaller reallocations away from agriculture and lower productivity of industry and services Annual change, 2000-2017, percentage points Philippines 2000-2017… Thailand 2000-2017… China 2000-2017… Vietnam 2000-2017… Malaysia 2000-2017… Indonesia 2000-2017… -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Annual Change (percentage points) Agriculture, within Industry, within Services, within Agriculture, structural Industry, structural Services, structural Source: World Bank staff calculations using data from BPS and SAKERNAS, WDI 19 20 One reason why structural change has contributed relatively little to overall labor productivity growth in Indonesia is that most workers moved from agriculture to low-end services, which are not much more productive. Column B of Table 3a shows that between 2000-2018, workers moved out of agriculture mostly into trade, hotels and restaurants, as well as into ‘other services’. These are sectors that are not much more productive than agriculture, as seen in Column C (the ‘productivity difference’ or modified ‘cross’ term). A typical worker in these sectors only produces 1.4 to 2 times as much output as a typical worker in agriculture ( b, last column). In contrast, workers moving into financial services have the biggest impact on productivity growth, followed by mining and quarrying (see column A+B). These are the sectors with the biggest labor productivity gaps vis-à-vis agriculture: the average worker in financial services and in extractives is 7 and 16 times more productive, respectively (Table 3a). Overall, although labor productivity gaps have decreased over the past two decades ( b)14, they remain sizeable, indicating the need to accelerate growth-enhancing structural change. Table 3a: Workers have mostly moved into low VA services that are not much more productive than agriculture, limiting the contribution of structural change to overall labor productivity growth (C) Productivity (B+C) Structural (2000-2018) (A) Within (B) Between difference change Agriculture 0.206 -0.137 0.000 -0.137 Mining & quarrying -0.099 0.006 0.078 0.083 Manufacturing 0.112 0.015 0.033 0.048 Utilities -0.006 0.005 0.014 0.019 Construction 0.035 0.023 0.057 0.080 Trade, hotels & restaurants 0.104 0.035 0.018 0.052 Transport & communication 0.144 0.001 0.002 0.002 14The only exception to the overall trend is transport, storage and communications services sector, which experienced high labor productivity growth from the mid-2000’s. 20 21 Financial services -0.012 0.017 0.093 0.110 Public sector 0.005 0.005 0.005 0.010 Other 0.034 0.031 0.007 0.038 Table 3b: Labor productivity gaps have declined, but remain significant Ratio of sector’s real output per worker vis-à-vis agriculture 2000-2002 2016-2018 2000-2018 Agriculture 1.0 1.0 1.0 Mining & quarrying 60.7 15.8 30.8 Manufacturing 4.8 3.6 4.5 Utilities 18.9 5.0 10.5 Construction 4.8 3.5 4.3 Trade, hotels & restaurants 2.0 1.6 2.0 Transport & communication 1.8 4.4 3.2 Financial services 16.0 6.8 12.9 Public sector 4.0 2.0 3.2 Other services 1.4 1.3 1.4 Source: World Bank staff calculations using data from BPS and SAKERNAS. Columns refer to period averages. Structural change could have contributed more to labor productivity growth if not for the decline of the manufacturing sector in the 2000s. Although medium and large manufacturing firms tend to create stable and formal employment, as seen in the high share of waged workers in the sector (80 percent), the share of workers in manufacturing steadily declined after the Asian Financial Crisis and has hovered around 14- 15 percent in recent years. Such ‘premature deindustrialization’15 (Figure 13) has dampened the 15 This phenomenon refers to declining shares of output and employment in manufacturing at a lower income per capita level than seen in advanced economies. 21 22 movement of workers from agriculture into manufacturing. Overall, the manufacturing sector has created few jobs in Indonesia compared to its neighbors, and most of these jobs have limited value-addition or knowledge intensity16. Although manufacturing job creation picked up in 2017-2018, accounting for a third of all jobs created over the period, most gains have been in coal/oil and gas refinery and food processing activities. The latter, which has experienced strong output growth recently, tends to create low- to mid-skilled jobs. Figure 13: Premature deindustrialization has hindered the reallocation of workers from agriculture to manufacturing Source: World Bank (2019) based on Diop (2016) Accelerating structural change is critical for faster economic growth and poverty reduction in Indonesia. As in other East Asian countries, labor productivity is the main driver of economic growth and poverty reduction in Indonesia, contributing 88 percent of its GDP per capita growth between 2000-2017. Changes in the share of the working-age population (“demographic change�) and the employment rate contributed the remainder, while changes in the labor force participation rate detracted from growth (primarily because more young adults were in school). Within-sector productivity improvements drove the bulk of 16 The share of tertiary-educated workers in the manufacturing sector is only 7 percent. 22 23 economic growth, as in other countries; however, the contribution of structural change to growth was smaller in Indonesia than in China and Vietnam – countries that experienced faster GDP per capita growth over the period (Figure 14). This suggests that accelerating worker flows from lower to higher-productivity economic activities can not only have a direct impact on the quality of jobs (through higher wages), but also an indirect impact through faster economic growth. Figure 14: Labor productivity is the main driver of growth in Indonesia and other countries Contribution to value-added per capita growth over 2000-2017, percentage points Within-sector productivity Structural change Employment rate Participation rate Demographic change Philippines 2000-2017 2.2 0.7 Total=3.5% Thailand 2000-2017 2.1 1.3 Total=3.4% China 2000-2017 7.9 1.1 Total=8.7% Vietnam 2000-2017 2.4 2.1 Total=5.3% Malaysia 2000-2017 -0.1 2.1 Total=3.0% Indonesia 2000-2017 2.5 1.0 Total=3.9% -4 -2 0 2 4 6 8 10 Source: World Bank staff calculations using data from BPS and SAKERNAS, WDI Regardless of where and how productivity improvements occur, they will become even more critical to Indonesia’s economic wellbeing in the future. Although there is still ample room to boost female labor force participation in Indonesia, the growth of the working age population has been slowing and demographic changes and the employment rate are unlikely to contribute much more to economic growth in the future, as Indonesia nears the end of its demographic dividend. The next section outlines 23 24 the policies that are needed to accelerate labor productivity growth, specifically through faster and higher-quality structural transformation, as a strategy to create better jobs. 4. How can Indonesia create more middle-class jobs – faster? The preceding analysis suggests two broad pathways for creating better jobs in Indonesia: first, make firms more productive, and second, help workers transition more quickly from less to more productive activities. While the former applies to all sectors of the economy, the sectors that have the most potential to create middle-class jobs are manufacturing and high-end services17. Policies to help firms operating within these sectors become more efficient and moving workers into these sectors more quickly would therefore have the greatest impact in boosting overall labor productivity, and hence economic growth. A policy agenda for ‘good jobs’ is heavily intertwined with a policy agenda for higher productivity and consists of three sets of policy interventions. First, interventions to help firms become more competitive are critical for the creation of good jobs. This can be achieved by making it easier for firms to access high- quality equipment and raw materials to boost production and exports, as well as helping firms adopt new technologies Second, investments in human capital are needed to help nurture a local talent pool that is attractive to innovative foreign firms looking to expand in Indonesia, as well as to foster the growth of domestic firms that will create high-skilled jobs. Third, continuous investments in physical and digital connectivity are critical to help workers reallocate more easily across sectors and regions of Indonesia, thus accelerating the pace of structural change. 1. Increase export-oriented foreign direct investment (FDI)18 17 This is not to say that agricultural productivity is not important, but it is not the focus of this piece. 18 This section draws heavily on the work of Massimiliano Calì and others for Part B of the World Bank Indonesia Economic Quarterly December 2018, “Strengthening Competitiveness�. 24 25 Even an in environment of trade tensions, manufacturing can still create good jobs for Indonesians and provide an ‘escalator’ to the middle class. Manufacturing generates spillovers to other sectors (such as services), creates more formal employment and promotes opportunities to close the gender gap in labor force participation (see Rahardja and Winkler, 2012). It is also intensive in mid-skilled labor and routine tasks, making it easier for workers to switch from agriculture and low-end services to manufacturing jobs. While there are a range of interventions that can help the manufacturing sector, boosting export-oriented FDI is the most critical to creating better jobs. This is because medium and large manufacturing plants that are foreign-owned and/or export tend to generate more jobs and pay higher wages on average (Figure 15 and Figure 16), reflecting higher levels of underlying productivity (Wihardja, forthcoming). Although foreign-owned manufacturing plants made up only 9 percent of all plants in 2015, they are responsible for around half of Indonesia’s manufacturing exports. Apart from the direct potential contribution to employment, foreign-owned plants can also generate technology spillovers to other firms and generate greater competition, which can lead to innovation and lower costs. Figure 15: Foreign-owned and/or exporting Figure 16: …and they create more jobs plants pay higher wages on average… Average annual wages by manufacturing plant type, IDR Average employment by manufacturing plant type, IDR million million 35 32.3 700 30 600 25.7 25 22.6 23.1 500 20 400 15 300 10 200 5 100 0 0 Domestic Foreign Non- Exporter Domestic Foreign Non- Exporter firm firm exporter firm firm exporter Source: World Bank staff calculations from Statistik Industri, 2015 25 26 The Government would need to lift wide-ranging restrictions on firms’ productive capacity if it is to boost manufacturing exports and FDI. At present, firms face barriers in importing some types of capital goods and raw materials and are subject to minimum local content requirements, which increase the costs of production and/or reduce the quality of production inputs (Rahardja and Varela, 2014). Eliminating (i) tariffs on key inputs for manufacturing, (ii) letters of recommendation for importing industrial inputs and (iii) pre-shipment inspections would make it much easier for firms to export and hence create more good jobs in the sector. Similarly, lifting foreign equity limits would help attract more export-oriented manufacturing FDI. Analysis by Calì, Doarest and Presidente (2019) shows that Indonesia’s Daftar Negatif Investasi (Negative Investment List, or DNI) reduces the entry of foreign manufacturing plants, increasing the costs of key inputs to production and reducing the profitability of downstream industries. Lifting the DNI would drive improvements in the average plants’ performance (as measured by the probability that the plant would invest), labor productivity, average wages paid and the adoption of new production technologies. Removing foreign equity limits would also boost job-creating investments in high VA services sectors. According to World Bank analysis, the incidence of foreign equity restrictions in the DNI tends to be higher in services, particularly in transport and communications, education, and financial services, and real estate and health. These restrictions reduce the quality and increase the costs of services domestically, inhibiting the productivity of firms in other sectors of the economy (Duggan et al. 2013) – and suppressing average wages paid by firms. For example, restrictions on the entry of foreign labor inhibit knowledge sharing that leads firms to become more innovative, and overall results in a less attractive talent pool for companies looking to invest. In health, for example, only 41 out of 200,000 practicing doctors are foreign, and they are not allowed to visit patients (only allowed to train doctors). This results in a very limited base of skilled health professionals, which deters job-creating foreign investments in the health sector. Overall, removing 26 27 these foreign equity limits, as well as sectoral reservations for small and medium-enterprises could increase average wages by 15 percent (World Bank, 2018). 2. Invest in human capital – both of today’s and tomorrow’s workforce Creating the conditions for more Indonesians to enter and complete tertiary education would encourage domestic and foreign firms to create more high-skilled jobs, especially in high VA services sectors. Currently, the low share of the workforce with a tertiary degree – among other factors – hinders the creation of more ‘high-skilled’ jobs (i.e. managers, technical and professional staff). Firms located in Indonesia frequently cite the lack of high-skilled talent as a hindrance to business expansion – more so than in neighboring countries such as Malaysia, Thailand and the Philippines (Table 4). New firms, such as start-ups in the booming digital economy, also end up having to offshore certain skills or hire foreign talent19. As a result, many major technology companies and multinationals prefer to locate their investments in neighboring countries. Expanding the talent pool locally by ensuring more young Indonesians are prepared for and have the means to access formal tertiary education would help these firms create more high-skilled positions that can be filled by locals. The Government’s recently-launched Kartu Indonesia Pintar-Kuliah program20 to provide more university scholarships to poor students is a step in the right direction, assuming it is well-targeted. 19 The ride-hailing firm Go-Jek, for instance, has set up an engineering hub and acquired a recruitment platform in Bangalore, India to hire ‘top’ developer talent (Tech Crunch, 2019). However, the digital talent shortage is not unique to Indonesia: 90 percent of respondents in a survey of Southeast Asia’s startups believe that the skills gap is a major constraint to their firms’ growth (see Nikkei Asia, 2019). 20 KIP Kuliah is a scaled-up version of an existing program, Bidik Misi. The Government’s target is to almost double the number of beneficiaries from 420,000 to 818,000 recipients. 27 28 Providing more on-the-job training (OJT) to existing workers and firms could also help boost productivity, particularly within the services sector. While investments in human capital can help tomorrow’s workforce, today’s workers would benefit from re-skilling and training opportunities that would help them become more productive within their existing jobs or reallocate to higher-skilled jobs. Firms could offer formal programs on management, interpersonal and communication skills to employees to increase firm productivity. Such training could help wage employees boost their earnings: in Indonesia specifically, training is estimated to increase the returns to one year of experience by a further 2.4 percentage points21. Currently, only about 12.5 percent of firms offer on-the-job training, one of the lowest among middle-income countries22 (Gomez-Mera and Hollweg, 2018). Moreover, a wide range of evidence has shown that training firm owners on better business and management practices can boost productivity, employment and wages (Bloom et al., 2013)23. The Government could therefore work with the private sector and vocational institutes to increase the provision of OJT, for example by designing its new superdeduction for job training so that firms can cover training costs and making training of the domestic workforce a requirement for foreign worker visas. Table 4: A higher share of firms find it difficult to hire high-skilled workers in Indonesia compared to neighboring countries Percent of firms that cited inadequate skills as the key barrier to hiring each type of worker Type of worker Indonesia Malaysia Thailand Philippines (2009-15) (2015-16) (2015-16) (2015-16) Managers 76.7 30.2 75 34.2 Non-production technicians, 67.3 50 86.7 55.6 associate professionals, and sales workers Skilled production workers 55.1 39.5 46 69.3 21 Mincerian regression estimates by M. Wihardja and H. Alatas using data from Sakernas 2016-2018. The returns to an additional year of experience is estimated at 3 percent on wages. 22 This estimate is from the World Bank Enterprise Survey (2015), which only captures formal training in firms. 23 Research by Dalton et al. (2019) shows that advising small Indonesian retailers on good business practices through a handbook and experiential learning modules helped to boost their profits 35 percent. 28 29 Unskilled non-production workers 43.4 25.6 57.1 38.9 Unskilled production workers 21.8 38.5 25 48.1 Source: Gomez-Mera and Hollweg (2018) based on World Bank Enterprise Survey data. 3. Continue to build physical and digital infrastructure Finally, the Government should continue to invest in connectivity infrastructure to make it easier for workers to reallocate across sectors. Research has shown that workers in Indonesia face high barriers to transitioning to jobs in new sectors and/or in other regions (see Calì, Hidayat and Hollweg, 2018). These barriers range from physical to psychological and monetary constraints. To reduce some of these mobility costs, the Government needs to continue investing more and better in transport infrastructure – both between rural and urban areas and within metropolitan areas – that would connect workers to better employment opportunities elsewhere. Toll roads – where the Government has invested heavily in recent years – are important, but so are arterial roads that provide last-mile connectivity. Making it easier for workers to move would allow human capital and resources to flow to the most efficient sectors and firms, boosting labor productivity and job creation.24 Moreover, continuing to invest in infrastructure would support growth and job creation in the construction sector, which, as previously noted, has been a source of structural change – and one that has low formal education requirements. Expanding access to digital infrastructure could also boost access to good jobs, especially for women and youth. Technology may enable more Indonesian women to participate in the labor force by making it easier to work from home, which, given the low female LFPR, would result in productivity gains. More available information about labor market opportunities, for example through online job portals, could 24Bryan and Morten (2019) estimate that removing all barriers to internal labor migration in Indonesia would result in a 22 percent increase in labor productivity. 29 30 also reduce the costs of searching for employment and matching with a suitable job – both from the perspectives of firms and employees. Structural transformation will continue to occur in Indonesia, but policymakers need to recognize that it is not an automatic process that always enhances growth. Without any changes to the status quo, a greater share of Indonesians will continue to work in low-skilled, low-paid jobs, mostly in low VA services sectors, making it difficult for them to enter the middle class. As more and more tasks become automated, net job losses in such sectors may occur – or, at the least, these jobs will continue to receive very low wages. Since high-skilled jobs – jobs with a high share of non-routine, cognitive tasks – are less likely to be destroyed and may in fact increase on a net basis, automation may exacerbate labor market inequality. The good news is that Indonesia has high potential to increase labor productivity in the short term by opening up the economy to attract investments that create good jobs, and in the medium term by investing in human capital and connectivity. 30 31 Annex: Most Indonesians work in sectors where less than 10 percent of workers receive middle-class wages Share of all Share of Share of all workers Share of waged employ- waged with at least a workers with at least ment workers middle-class wage a middle-class wage Agriculture, forestry, hunting and fishery 28.5 40.3 1.4 3.5 Mining and quarrying 1.2 88.5 25.8 29.2 Manufacturing 14.6 79.7 12.1 15.2 Electricity and gas 0.3 95.4 24.1 25.3 Water supply, sewerage, waste 0.4 87.1 11.4 13.1 management, remediation Construction 6.6 93.8 6.9 7.4 Wholesale and retail trade, vehicle repair 18.4 62.5 7.2 11.6 Transportation and storage 4.3 96.2 14.7 15.2 Accommodation and food service activities 6.1 57.5 5.8 10.0 Information and communication 0.7 85.6 28.0 32.7 Financial and insurance activities 1.4 99.1 42.9 43.3 Real estate activities 0.3 94.1 28.6 30.4 Business activities 1.3 86.4 25.6 29.7 Public administration, defense and 4.7 100.0 42.2 42.2 compulsory social security Education activities 4.8 98.2 28.5 29.0 Human health and social work activities 1.5 97.9 26.2 26.8 Other service activities 4.8 91.5 5.6 6.1 Source: Sakernas 2018, World Bank staff calculations Note: This estimation follows the World Bank (2019) definition of the middle class as monthly household per capita consumption of 3.5-17 times the local poverty line, which was IDR401,220 (USD 28) in March 2018. Assuming a household of two working adults and two children, this works out to about a total consumption of IDR 5.62 million (3.5*4*IDR401,220) a month. Assuming that there are 1.5 working adults per household and a marginal propensity to consume of 1.2, each working adult must earn IDR 3.8 million (IDR 5.62 million / 1.5) per month to qualify as a middle-class household. 31 32 References Bryan, G. and M. Morten. 2019. “The Aggregate Productivity Effects of Internal Migration: Evidence from Indonesia�. Journal of Political Economy, 127:5 (October 2019), 2229-2268. Calì, M., Doarest, A. and G. Presidente. 2019. “The Impact of Foreign Entry and Competition in a Developing Country: Plant-level Evidence from Indonesia.� mimeo. Calì, M., Hidayat, T. and C.H. Hollweg. 2018. “What is Behind Labor Mobility Costs? Evidence from Indonesia�. Background Paper for “Time to ACT: Realizing Indonesia’s Urban Potential�. Washington, D.C.: World Bank. de Vries, G., Timmer, M. and K. de Vries. 2015. “Structural Transformation in Africa: Static Gains, Dynamic Losses�. The Journal of Development Studies, 51:6, 674-688. Diop, N. 2016. “Making Indonesian Manufacturing an Engine of Growth Again: Now or Never?� Mimeo. World Bank. Dalton, P., Ruschenpohler, J., Uras, B. and B. Zia. 2019. “Experimental Evidence from Small Retailers in Indonesia.� World Bank Policy Research Working Paper 8933. Washington, D.C.: World Bank. Duggan, V., S. Rahardja, and G. Varela. 2013. “Service Sector Reform and Manufacturing Productivity: Evidence from Indonesia.� Policy Research Working Paper 6349. Washington, D.C.: World Bank. Galasso, E. and A. Wagstaff. 2017. “The Economic Costs of Stunting and How to Reduce Them�. World Bank Group Policy Research Note. Washington, D.C.: World Bank. Gomez-Mera, L. and C. Hollweg. 2018. Firm performance and constraints in Indonesia. Mimeo. World Bank. 32 33 Halim, D., Johnson, H. and E. Perova. 2019. “Preschool Availability and Female Labor Force Participation.� Policy Research Working Paper 8915, June 2019. Washington, D.C.: World Bank. McMillan, M., and D. Rodrik. 2011. “Globalization, Structural Change, and Productivity Growth.� In Making Globalization Socially Sustainable. Geneva: International Labour Organization and World Trade Organization. Merotto, D., Weber, M. and R. Aterido. 2018. “Pathways to Better Jobs in IDA Countries: Findings from Jobs Diagnostics�, Jobs Working Paper No. 14. Washington, D.C.: World Bank. OECD (2018), Good Jobs for All in a Changing World of Work: The OECD Jobs Strategy, OECD Publishing, Paris, https://doi.org/10.1787/9789264308817-en. Rahardja, S. and D. Winkler. 2012. “Why the Manufacturing Sector Still Matters for Growth and Development in Indonesia.� Jakarta, Indonesia: World Bank. Rodrik, D. 2016. “Premature Deindustrialization�. Journal of Economic Growth, 21(1), 1–33. Rodrik, D. and C. Sabel. 2019. “Building a Good Jobs Economy�. Draft dated April 2019. Working Paper. Available at https://j.mp/2G5tnvX World Bank. 2013. World Development Report: Jobs. Washington, D.C.: World Bank. World Bank. 2018. “Indonesia Economic Quarterly, December 2018: Strengthening Competitiveness.� Jakarta, Indonesia: World Bank. 33