Economic Growth and Employment in South Asia

Abstract

The South Asian countries, it is often said, have been achieving fairly rapid economic growth but this growth has been largely jobless so that they now face a daunting employment challenge. The story is not in fact quite so simple. Not all the countries of South Asia have been achieving rapid economic growth and economic growth has not been jobless in any of the countries. What can justifiably be said is that economic growth in these countries should have improved employment conditions much more than it actually did.

1 Introduction

The South Asian countries, it is often said, have been achieving fairly rapid economic growth but this growth has been largely jobless so that they now face a daunting employment challenge. The story is not in fact quite so simple. Not all the countries of South Asia have been achieving rapid economic growth and economic growth has not been jobless in any of the countries. What can justifiably be said is that economic growth in these countries should have improved employment conditions much more than it actually did.

But what does this mean? Employment growth, as we shall see below, actually equalled labour force growth in all the countries. But this does not imply anything about improvement or deterioration in employment conditions. To understand why, we need to recognise that the economies of South Asia are dual economies à la Lewis with a formal or organised sector and an informal or unorganised sector.¹ The formal sector offers good jobs but employs only a small proportion of the workforce. The bulk of the workforce is in the informal sector, which holds large stocks of surplus labour in the form of underemployment and very low-productivity employment of many workers. Labour force participation also tends to be low as many people do not even look for work because they see this as a futile exercise. In this setting, economic growth improves employment conditions only when it reduces the stock of surplus labour (including the potentially available labour of many who are currently out of the labour force) through a process of transfer of those in underemployment and low-productivity employment in the informal sector to more productive jobs either in the formal sector or within the informal sector itself. It is generating such a process of labour transfer that economic growth in South Asian countries has performed rather poorly.

But why has even rapid economic growth failed to rapidly improve employment conditions? The answer is to be found in the inappropriateness of the pattern of growth. Services have always been extraordinarily important (accounting for an unusually high share in GDP) in South Asian economies, which are still at early stages of development. Quite extraordinarily, moreover, economic growth in most of these countries (the only exception being Bangladesh) have also been services led. To add to the extraordinariness, the employment intensity of services has been and remains very low, a fact that stands in sharp contrast with what is observed in other countries of the world.

The growth process in South Asian economies, therefore, has brought too little structural change in employment. The movement of workers from agriculture to non-agriculture has been much too small and even this small movement has often been into low-productivity informal employment in non-agriculture.

Undoubtedly, the employment challenge that confronts the South Asian countries is formidable. The argument of this paper is that ‘business-as-usual’ growth, even if this can be sustained (there are reasons to think that services-led growth will be difficult to sustain), will do little to enable these countries to meet the employment challenge. The countries’ own past experience makes this clear. A reorientation of the growth strategy is needed. Meeting the employment challenge will require rapid manufacturing-led growth.

The paper is organised as follows. In the section that follows, the growth experience of South Asian economies during the past decade and a half is scrutinised to see how growth affected employment conditions. In the third section, the employment challenge that these countries face is outlined. In the final section, the argument that rapid manufacturing-led growth is what is required to meet the employment challenge is sketched.

The statistical data used in the paper have been assembled from a variety of sources: SARNET database (built by collecting data from national sources) maintained at the Institute for Human Development (New Delhi), the World Bank (WDI database), the International Labour Organisation (KILM database) and the Asian Development Bank (Key Indicators database). Simple interpolations have sometimes been used to construct a dataset for five South Asian countries (Bangladesh, India, Nepal, Pakistan and Sri Lanka) covering the same period. Some data for a set of comparator countries—China, Indonesia, Korea, Taiwan and Thailand—have also been extracted from the same sources: The World Bank, the ILO and the ADB.

2 Economic Growth and Employment in Five South Asian Economies, 2000–2015

2.1 Characteristics of Employment

While economic growth during the past decade and a half cannot be said to have been poor in any of the five South Asian countries under consideration, it was reasonably rapid only in three of these—Bangladesh, India and Sri Lanka (Table 1). In Nepal and Pakistan, GDP growth was slow, just 4% per annum. The other more interesting fact is that no particular relation between GDP growth and employment growth is observed across the countries; India’s GDP growth of 7% was associated with employment growth of only 1.5% while Pakistan’s GDP growth of 4% was associated with employment growth of 3.3%. On the other hand, employment growth closely followed, indeed virtually equalled, the labour force growth in each of the countries.

Table 1 Growth rates (percent per annum), 2000–2015

These are the features that tell us that South Asian economies are labour-surplus dual economies. GDP growth is independent of employment growth in such economies; labour is in excess supply so that no ‘labour constraint’ on growth exists. On the other hand, most people do not have access to any kind of institutionalised social security and must work to survive. This means that most of those who are in the labour force are also employed. But many engage in work sharing (in self-employment and in casual wage employment) and many others engage in very low-productivity work. Very few remain unemployed and these few are educated (hence looking for good jobs in the formal sector) and generally belong to relatively well-off households (so that they can afford to remain unemployed). The employment problem manifests itself in poverty (which results from underemployment and low-productivity employment) and not in unemployment (which does not imply poverty). Employment growth reflects labour force growth and tells us little about growth in the demand for labour associated with economic growth.

Some of these characteristics of employment and unemployment are empirically discernible (Table 2a). Except in Sri Lanka, the formal sector employs a very small proportion of the workforce; a large majority (between 80 and 90%) of the workers work in the informal sector, either as self-employed or as casual wage employees. Even in Sri Lanka, where the formal sector employs 37% of all workers, the majority is still in the informal sector.

Table 2 (a) Structure of employment, 2015. (b) Underemployment and unemployment, 2015

It is in self-employment and casual wage employment that there is much scope for work sharing; a given amount of work can be performed by a flexible number of workers. In the case of self-employment, the working members of a family share both the work and the income. In the absence of investment and technological change, an increase in the number of working members in families results in increased work sharing (i.e. increased underemployment). In the case of casual wage employment, not all the workers find employment on any given day but no worker fails to find employment on all days. In short, an increase in the number of persons seeking work in the informal sector results in increased employment (and not in increased unemployment) but also in increased underemployment. Given that the bulk of the labour force is in the informal sector, the same hold for the economy as a whole: labour force growth means employment growth but employment growth is often associated with rising underemployment.

On a snapshot view, therefore, underemployment, not unemployment, shows up to be the major problem (Table 2b). It should be said that the estimates of underemployment are not robust. In Nepal, for example, underemployment has almost certainly been seriously underestimated; in a situation, where 83% of the workers are in self-employment, underemployment has to be much higher than 6%. In Sri Lanka, too, underemployment appears to have been underestimated. Moreover, the estimates, even where they are reliable, are not comparable across countries as the definition is not uniform. They are presented here only to illustrate the point that underemployment of the employed is more significant than unemployment in South Asian economies. And unemployment, as noted above, is essentially confined to educated youth from relatively well-off households waiting in the queue for good jobs.²

Change in employment conditions, therefore, cannot be discerned from employment growth or change in unemployment. Employment growth in the formal sector and change in underemployment in the informal sector can tell us much about the change in employment conditions, but time-series data on these indicators are in general unavailable. Only the generally available information on change in the structures (sector shares) of output and employment associated with economic growth can be analysed to see if and to what extent the employment conditions have been improving.

Of the sectors, agriculture is almost wholly informal (except perhaps in Sri Lanka, which has a large plantation sector) while the non-agricultural sectors are themselves dualistic with formal and informal segments. Movement from agricultural employment to non-agricultural employment, therefore, is not equivalent to movement from informal employment to formal employment. However, even informal employment in non-agriculture is, as a rule, more productive and remunerative than agricultural employment. Thus, movement of workers from agricultural employment to non-agricultural employment usually does mean improvement in employment conditions. Such movement, when large, leads to a decline in employment in agriculture and thus to a rise in labour productivity in agriculture, which also means improvement in the conditions of agricultural employment (since it implies a decline in underemployment and in very low-productivity employment).

2.2 Economic Growth, Structural Change and Employment

Historically, economic growth has been associated with persistent decline of agriculture together with rise of manufacturing at an early phase (when the income level is low) and with rise of services at a later phase (when the income level is high). In today’s developed countries, the share of manufacturing in GDP increased for a long period before beginning to decline, which was also when the share of services began to increase rapidly.³ The share in total employment in the economy followed the same trajectory as the share in GDP for both manufacturing and services. However, because technological change was important in manufacturing but not in services, growth of labour productivity was generally significant in manufacturing but quite insignificant in services. So, the share of manufacturing in total employment was always lower than its share in output while the share of services in employment tended to equal its share in output. In more recent periods, the same pattern of structural change has also been observed in the rapidly growing economies of East Asia (Table 3). Indeed, no country has yet attained even middle-income status without industrialisation.⁴

Table 3 Growth of GDP and employment in selected East Asian economies

The growth experience of South Asian economies, which still are in early stages of development, defies the historically observed pattern. Economic growth in these economies during the last decade and a half has been services led (Table 4). And labour productivity growth has been much higher in services than in manufacturing, which is to say that the employment intensity of growth has been much lower in services than in manufacturing. Bangladesh stands out as the sole exception; here growth was manufacturing led and the employment intensity of growth was higher in services than in manufacturing.

Table 4 Growth of GDP and employment in South Asian economies, 2000–2015

The dominance of services in the economy is a feature that the South Asian countries inherited from their colonial past. In British India in 1946, services accounted for 38% of GDP while manufacturing accounted for 17%.⁵ High share of services in GDP was thus a part of the initial conditions in the South Asian economies. The kind of growth that they experienced only increased the dominance of services. In 2000, the share of services in GDP ranged from 44% in India to 62% in Sri Lanka (Table 5a). It continued to grow in all countries except Bangladesh throughout the period 2000–2015. The manufacturing-led growth in Bangladesh did lead to a reduction of the share of GDP in services, but the share still remained high as it was already very high in 2000.

Table 5 (a) Sector shares in GDP and in employment, 2000. (b) Sector shares in GDP and in employment, 2015

A comparison of the structural characteristics of South Asian economies with those of East Asian economies is illuminating. In 2015, the share of services in GDP ranged from 55% in Nepal to 66% in Sri Lanka (Table 5b). On the other hand, the share of manufacturing in GDP ranged from just 6% in Nepal to 19% in Bangladesh. In East Asian economies (some of which already counted as developed), around 2012, the share of services in GDP ranged from 40% in China to 66% in Taiwan (Table 6). The share of manufacturing in GDP, on the other hand, ranged from 25% in Indonesia to 36% in China and Thailand. Around the time, the lowest income East Asian country was Vietnam, the per capita GDP (in current international Dollars) of which was the same as that of India (the second richest country in South Asia) and the highest income country in South Asia was Sri Lanka, the per capita GDP of which was just above that of Indonesia (the second poorest country in East Asia).

Table 6 Sector shares in GDP and in employment, some East Asian economies

The other contrasts between South Asia and East Asia relate to employment intensity of manufacturing and of services. In South Asian economies, the share of manufacturing in employment tends to be close to its share in GDP while the share of ‘services’ in employment is much lower than its share in GDP.⁶ In East Asian economies, the opposite is true; the share of manufacturing in employment is much lower than its share in GDP, while the share of ‘services’ in employment is close to its share in GDP. Thus, labour productivity in manufacturing is much lower than that in services in South Asia while the reverse is the case in East Asia. The data in Tables 5b and 6 suggest that labour productivity in manufacturing was 2–2.5 times that in services in East Asia but 0.2–0.9 times that in services in South Asia.

The pattern of growth in East Asia was far more effective in improving employment conditions than the pattern of growth in South Asia (Tables 3 and 4). This can be seen most clearly from a comparison of India’s experience during 2000–2015 with Korea’s during 1963–1990 (the period during which Korea’s economy underwent transformation from a low-income economy to a high-income economy). The growth rates of GDP in the two countries were not radically different: 7.2% in India and 7.5% in Korea and the rate of growth of agriculture was virtually the same: 3% in India and 2.9% in Korea. The rate of growth of non-agriculture was exactly the same: 8.2% in both countries. The big difference was in the composition of non-agricultural growth. This was driven by the growth of manufacturing (14.8% per annum) in Korea and by the growth of services (8.9% per annum) in India. Manufacturing recorded a growth of 7.5% in India and services recorded a growth of 6.6% in Korea.

The differences in the pace and pattern of employment growth were dramatic. The growth of employment was much slower in India (1.5% per annum) than in Korea (3.3% per annum). The growth of employment in non-agriculture (6.3% per annum) was very rapid in Korea so that employment in agriculture was declining fairly rapidly (at 1.5% per annum). In India, the growth of non-agricultural employment (3.2% per annum) was not rapid enough to turn the growth of agricultural employment negative. The far more rapid growth of non-agricultural employment in Korea is explained by two factors: the far higher growth of manufacturing and the far higher employment intensity of growth in services. In Korea, output growth of 14.8% per annum was associated with employment growth of 8% per annum in manufacturing; in India, output growth of 7.5% per annum in manufacturing was associated with employment growth of 2.3% per annum. So, even in manufacturing, the employment elasticity (i.e. the ratio of employment growth to output growth) was actually higher in Korea (0.54) than in India (0.31). In services, output growth of 6.6% in Korea was associated with employment growth of 5.6% (implying an employment elasticity of 0.85), while in India, output growth of 8.9% was associated with employment growth of only 2.4% (implying an employment elasticity of 0.27).

We can draw two conclusions. First, the employment effect of growth depends on both the pace and the pattern of growth. In South Asian countries, growth did not significantly improve employment conditions both because it was not rapid enough and because its pattern was wrong. Only India had rapid economic growth but the particular pattern of this growth blunted its effectiveness in improving the employment conditions. Growth was slower in Nepal, Pakistan and Sri Lanka and the pattern of growth was similar to India’s; the improvement in employment conditions was naturally less significant in these countries than in India. Bangladesh did have manufacturing-led growth with better effects on employment but the growth was not rapid enough.

A second conclusion is that services growth is much more employment intensive when manufacturing leads the growth process than when services lead it. In East Asian economies, where manufacturing led the growth process, services growth was far more employment intensive than in South Asian economies (except Bangladesh), where services led the growth process. Even in Bangladesh (a South Asian economy), where manufacturing led the growth process during 2000–15, services growth was more employment intensive than in other South Asian economies. The possible reasons are as follows. Services that lead the growth process are high skill and hence have low employment intensity; services that grow as complementary to growth of manufacturing are relatively low skill and have high employment intensity. These very features also imply that services-led growth engenders higher income inequality than does manufacturing-led growth. Higher income inequality, in turn, generates demand for skill-intensive services.⁷

3 The Employment Challenge in South Asia

The employment challenge that each of the South Asian economies faces is truly formidable. This can be seen from the results of an illustrative exercise reported in Table 7. Before discussing these results, however, we need to say something about the method used to derive the estimates of disguised unemployment, which relate exclusively to females. We have assumed that the currently low-female labour force participation rate in South Asian countries is explained by the fact that there are many ‘discouraged workers’ among women; many remain out of the labour force because they judge the prospects of finding work to be extremely poor. As employment conditions improve, therefore, the female labour force participation rate in South Asian countries will rise and eventually be around 50% (this roughly is the observed global average). Except in Nepal, the participation rate currently is significantly below 50% and the difference between the number implied by a participation rate of 50% and the number implied by the current participation rate gives the number in disguised unemployment (i.e. the number that can be expected to enter the labour force as employment conditions improve).

Table 7 The employment challenge

In Nepal, the number in disguised unemployment turns out to be negative. The reason is that the current female participation rate is 80%, which is much too high and reflects widespread poverty and distress participation (such high female participation rates are also observed in the countries of sub-Saharan Africa). Here, the female participation rate is expected to decline as employment conditions improve and poverty declines. We should also recall that underemployment is most likely to have been seriously underestimated in both Nepal and Sri Lanka. For these reasons, we have chosen to leave these two countries out in carrying out the illustrative exercise.

The employment challenge is equally daunting in the three countries considered—Bangladesh, India and Pakistan—and we can focus on the group of three, i.e. SA-3, in outlining it. Between now and 2025, in SA-3, 10.4 million persons are expected to enter the labour force every year. If 10.4 million full-time jobs can be created every year, the number unemployed (25.2 million) and the number underemployed (82.5 million) will remain unchanged but the number in disguised unemployment will increase (by around 2 million per year). If underemployment and disguised unemployment are to be eliminated (so that the Lewis Turning Point is reached) by 2030, 23.6 million full-time jobs will need to be created every year (note that the number unemployed will still remain unchanged). The enormity of the task can be appreciated when we note that, between 2000 and 2015, employment increased by 8.8 million per year and only a part of the incremental employment was full time.

The point of the exercise is to underline the fact that growth-as-usual will not lead to a resolution of South Asia’s huge employment problem in the foreseeable future. Both the pace and the pattern of economic growth will need to change; growth will need to be significantly faster and the growth of non-agriculture will need to be much more employment intensive. To meet these requirements, the South Asian countries will need to achieve rapid manufacturing-led growth as the East Asian countries have done. That manufacturing-led growth will be much more employment intensive than services-led growth has been is clear from the analysis in the preceding section. There are very good reasons to think that manufacturing-led growth will also be faster than services-led growth and we discuss these in the concluding section that follows.

4 The Need for Manufacturing-Led Growth: Some Concluding Observations

The empirical finding that, in historical experience, growth has always been manufacturing led at early stages of development and services led at later stages inspired the formulation of Kaldor’s well-known ‘growth laws’.⁸ These can be summarised in the form of the following propositions. First, increasing returns to scale are highly significant in manufacturing so that output growth itself causes productivity growth.⁹ Second, growth of manufacturing has important spill over effects on other production sectors and causes economy-wide productivity growth through reallocation of labour and other resources from agriculture and traditional services to the dynamic manufacturing sector.¹⁰ Third, growth of manufacturing induces growth of modern services as complementary and ancillary to manufacturing. The upshot is that the greater is the excess of manufacturing growth over GDP growth, the faster is GDP growth. Manufacturing-led growth is also the fastest achievable growth.

There are demand-side factors that also lend strong support to manufacturing-led growth at early stages of development. At low levels of per capita income, the income elasticity of demand for manufactures tends to be higher than that for services so that domestic demand for manufactures grows faster than that for services. Moreover, manufactures are tradable goods par excellence and external demand can play an important role in supporting rapid growth of manufacturing.

Recently, in light of India’s recent experience of rapid services-led growth, some economists have argued that certain services have now acquired the characteristics of manufacturing (as noted above), and hence can lead the growth process even in a low-income economy.¹¹ The prime example of such services is information technology (IT) and related services. India’s rapid services-led growth in the 2000s, therefore, is not perhaps so surprising. Indeed, it is perfectly possible that other low-income countries will also achieve rapid services-led growth in the twenty-first century.

As it happens, India still remains the sole exception to the rule; it is the only country to have achieved rapid services-led growth (at least for a period) at an early stage of development.¹² More importantly, even a cursory look into India’s growth process during 2005–12, the period of fastest growth, reveals that growth was not in fact led by the ‘dynamic’ services, the services that arguably have acquired the characteristics of manufacturing (Table 8). IT and communication services did grow very rapidly but these accounted for rather small shares of services and of GDP. On the other hand, financial services, which accounted for larger shares of services and of GDP, also recorded very rapid growth. We also know that the high-end segments of transport (e.g. air travel), trade (e.g. malls and supermarkets) and ‘hotels and restaurants’ recorded rapid growth. What explains the rapid growth of these ‘non-dynamic’ services? These services are not exported. The rapid growth of IT services, which are exported, could not possibly have induced such rapid growth of the ‘non-dynamic’ services. Here is a possible explanation: the rapid growth of services, and hence the rapid services-led growth was actually sustained by large inflows of foreign finance (inflows of foreign capital and remittances amounted to 6–7% of GDP on average during the period) rather than by ‘dynamic’ services. India’s rapid services-led growth has in reality been foreign-finance-led growth. This, of course, is a hypothesis and must remain so here. The point is that it is wrong to interpret India’s services-led growth to have been driven by the services that have now acquired the characteristics of manufacturing.

Table 8 India’s services-led growth

There also are reasons to think that services-led growth cannot be sustained for long, not just in India but also in other South Asian countries. This kind of growth brings about a serious imbalance between the structure of domestic absorption (consumption plus investment) and that of domestic production. In India today, goods account for around 65% of domestic absorption but just over 40% of domestic production.¹³ This means a high trade deficit. Continued services-led growth can only mean growing imbalance, and hence rising trade deficit, which can only be sustained if large inflows of foreign finance are sustained. Services-led growth is fragile.

South Asian countries need rapid employment-intensive growth to meet the employment challenge. And this means that they need manufacturing-led growth, which will be both rapid and employment intensive.

Appendix

Statistical Tables

See Tables 1, 2, 3, 4, 5, 6, 7 and 8.