FIRST DRAFT
Towards a Microeconomics of Growth* Robin Burgess LSE and CEPR Anthony J. Venables LSE and CEPR . Abstract: Growth and development typically involve the creation of new economic activities. New productive sectors develop, and old activities come to be performed in radically different ways. Evidence from both economic history and economic development suggest that this process of structural change is central to increasing growth and raising standards of living.
Indeed many early development theories viewed development as a process of
transformation from agriculture to manufacturing. The focus on structural change as being the driving force of economic development led a majority of low income countries to pursue of policies of inwards looking and frequently state-led industrialisation. The experiences of different countries and regions has been extremely mixed. In many regions centrally planned industrialization generally failed to deliver the increases in productivity and improvements in living standards which were originally expected (e.g sub-Saharan Africa). By contrast, the Asian miracle saw dramatic increases in living standards associated with state-led structural change, the share of the labour force in manufacturing rising, rising from around 10% in many countries in the early 1960s to 30% by the late 1990s. Despite these experiences, not much is known about the likely pattern of structural change in future development. What sorts of new activities will take place in the fast growing countries of the 21st century? Are there new ‘patterns of development’, analogous to those identified by Chenery et al in the 1960s? And most importantly, what factors are conducive to the fostering of new activities? Our goal in this paper is to explore some of the issues surrounding the development of new activities in low-income countries.
Our central thesis is that this process is frequently
‘lumpy’, manifesting itself in rapid growth of particular regions or sectors.
Spatial
inequalities tend to increase during periods of rapid economic development. At the sectoral 1
level, many fast growing cities, regions and countries have export specialisations in a very narrow range of activities. Recognition of these facts requires a reorientation of the analytical frameworks and empirical approaches that are used to investigate growth. We find it useful to divide the factors that are important in determining a location’s growth performance into two groups, that we term ‘1st advantage’ and ‘2nd advantage’ (terminology that we adapt from economic geographers’ notions of first-nature and second-nature geographies). ‘1st advantage’ refers to the conditions that need to be met to provide the environment in which new activities can most profitably be developed. They include most of the factors that traditional theory has focussed on, such as access to inputs (labour and capital), access to markets, and provision of basic infrastructure and institutional environment. ‘2nd advantage’ factors comprise the skills, technical know-how, networks and market linkages involved in efficient modern production. The distinguishing criterion is that 2nd advantage factors have, inherently, some element of increasing returns to scale. Furthermore, these increasing returns are often external to the firm, so associated with market failure. It is increasing returns that underlies the ‘lumpiness’ of development. We draw out the way in which 1st and 2nd advantages interact to shape the pattern of development, and argue that development policy needs to take account of the quite different natures of these advantages. On the research side, empirical work also needs to capture this micro-level heterogeneity and diversity. We use recent empirical work on India to illustrate the importance of sectoral and spatial heterogeneity and to demonstrate the role of microbased empirics. JEL classification no: Keywords:. Produced as part of the Globalization programme of the UK ESRC funded Centre for Economic Performance at the LSE Addresses: R. Burgess Dept of Economics London School of Economics Houghton Street London WC2A 2AE, UK
A.J. Venables Dept of Economics London School of Economics Houghton Street London WC2A 2AE, UK
[email protected] http://econ.lse.ac.uk/staff/rburgess/
[email protected] http://econ.lse.ac.uk/staff/ajv/
2
1. Introduction Growth and development typically involve the creation of new economic activities. New productive sectors develop, and old activities come to be performed in radically different ways. Evidence from both economic history and economic development suggest that this process of structural change is central to increasing growth and raising standards of living (Schumpeter, 1959; Rostow, 1960; Hirshman, 1958; Kaldor, 1969; Sokoloff, 1986). Indeed many early development theories viewed development as a process of transformation from agriculture to manufacturing (Lewis, 1954). Labour productivity is higher in the newly developing sectors, and the reallocation of labour also pulls up productivity in traditional sectors. The focus on structural change as being the driving force of economic development led a majority of low income countries to pursue of policies of inwards looking and frequently state-led industrialisation.
The experiences of different countries and regions has been
extremely mixed (see Rodrik, 2003). In many regions centrally planned industrialization generally failed to deliver the increases in productivity and improvements in living standards which were originally expected (e.g sub-Saharan Africa). By contrast, the Asian miracle saw dramatic increases in living standards associated with state-led structural change, the share of the labour force in manufacturing rising, rising from around 10% in many countries in the early 1960s to 30% by the late 1990s. Despite these experiences, not much is known about the likely pattern of structural change in future development. What sorts of new activities will take place in the fast growing countries of the 21st century? Are there new ‘patterns of development’, analogous to those identified by Chenery et al in the 1960s? And most importantly, what factors are conducive to fostering new activities? Our goal in this paper is to explore some of the issues surrounding the development of new activities in low-income countries.
Our central thesis is that this process is frequently
‘lumpy’, manifesting itself in rapid growth of particular sectors or regions. Locations that are in many respects similar follow different paths, specialising in different activities. Recognition of these facts requires a reorientation of the analytical frameworks and empirical approaches that are used to investigate growth. We find it useful to divide the factors that are important in determining a locations’ growth 3
performance into two groups, that we term ‘1st advantage’ and ‘2nd advantage’ (terminology adapted from economic geographers’ notions of first-nature and second-nature geographies). ‘1st advantage’ refers to the conditions that need to be met to provide the environment in which new activities can most profitably be developed. ‘2nd advantage’ refers to the fact that many aspects of growth are self-reinforcing. The lumpiness of the growth process suggests the presence of increasing returns, so that a location’s advantage in an activity derives, in part, from the very fact that it has a presence in the activity. It is the interaction of 1st and 2nd advantages that yields the patterns we observe, and that makes it so difficult to pin down the causes of rapid economic growth. Of course, increasing returns and cumulative causation have featured in many stories about economic development, from the development economists of the 1950s through to Murphy, Shleifer and Vishny (1989). We think it important to recognise that key aspects of returns to scale are often found at the micro-level – within very narrow sectors of production, or small districts of a city. A microeconomics of growth is needed to capture these sectorally and spatially concentrated processes Empirical work needs to capture this micro-level heterogeneity and diversity. Cross-country comparisons, though useful in identifying some of the important 1st advantages cannot provide enough detail to be really insightful about the growth process. Instead a more bottom up approach is required, where microeconomic studies are used to build up an evidence base on what works in different countries, regions and sectors. The justifications for this microeconomic approach are manifold. First it is clear that the evidence base which can be exploited to evaluate economic policies has dramatically expanded in recent years. Household, firm and sub-national data are now available for a wide range of countries. This makes broad brush comparisons between countries and cross-country regression less relevant as a tool for identifying what works. This type of data also allows us to evaluate the microeconomic impact of macroeconomic policies. Second it is becoming transparent that the institutional context in which policies are implement and accumulation decisions are made really matter for outcomes (see Djankov et al, 2003; Rodrik, 2003). The resurgence of interest in the political economy of development is a reflection of this realization. This makes it natural to study single countries and makes it less clear what one learns from cross-country regressions where these factors cannot typically be controlled for. Third it is clear the actors in development work are becoming more diverse. Policy and institutional innovations are happening all the time and are being implemented by central government, local government, non-government organisations and local communities. Evaluating these innovations requires 4
careful microeconomic work at the subnational level. Our route for investigating these issues is to start (section 2) with a selective overview of the facts concerning the growth of new productive activities in developing countries. We then develop a framework of analysis that can be used to think about the determinants of structural change and growth, developing our 1st and 2nd advantage distinction (section 3). Section 4 goes into more detail, outlining what we know about some of the key elements of 1st and 2nd advantage. Section 5 turns to India, as an application of some of our thinking. Section 6 concludes. 2. Aspects of modern economic growth. Much of the recent work on the determinants of growth has focussed on country aggregate growth performance.
The most popular examples of this are cross-country growth
regressions, typically regressing growth performance on a number of explanatory variables. While these regressions reveal something of the importance of 1st advantages – such as human capital, good policy and institutions and open trade regimes -- many of their results are not particularly robust (see Temple 1999; Djankov et al 2003; Rodrik, 2003). And importantly, these studies are generally much too broad-brush to be of much use to policy makers. Our approach is different, we will highlight some aspects of the modern growth process, and then draw out the implications of these aspects for our understanding of the determinants of growth. In particular, we think that an aggregate approach overlooks two of the most important aspects of the performance of many developing countries; the uneven performance of different regions within each country and the different performance across sectors.
Spatial concentration For many countries development is associated with increasing spatial inequality. The finding of rising then declining spatial inequality during development dates back at least to Williamson (1965) and has been confirmed in many studies since. This increase in spatial inequality arises partly from spatial concentration in the development of manufacturing. We see this most clearly in data for large countries. In Section 5, for example, we demonstrate 5
how states in southern India have come to prominence in manufacturing.
In Mexico
manufacturing has become highly concentrated in regions that border the US, and spatial variation in per capita incomes has increased dramatically since the mid 1980s (Cikurel 2002). In China, Demurger et al (2002) chart increasing spatial inequalities in per capita GDP from the mid 1980s. Coastal provinces experienced the greatest decline in the share of agriculture in employment and output, and the fastest growth of per capita income. The spatial concentration of new activities also occurs at a much finer level of disaggregation than suggested by the state or province level data above. Bangalore is a good example; the city is estimated to account for 25% of India’s software exports, with some 100,000 workers in the city producing 3% of India’s total exports (Aranya 2002). The polarization of activity in a number of ‘city regions’ in Asia is documented by Aranya. Within cities, we know that particular sectors frequently cluster in particular districts. Spatial concentration is most dramatically demonstrated by the role of urbanisation – and of mega-cities – in development. The number of cities in the world with a population of more than 1 million went from 115 in 1960 to 416 in 2000; for cities of more than 4 million the increase was from 18 to 53, and for more than 12 million from 1 to 11 (Henderson 2003). Henderson (1999) finds that national urban concentration (the dominance of the largest city) rises with growth from low-income levels, peaks at low-middle levels (per capita 1987 PPP income of around $2500) and then declines.
This tells us that, despite the massive
diseconomies associated with developing country mega-cities, there are even more powerful economies of scale making it worthwhile for firms to continue to locate in these cities. Urbanisation is evidently one of the clearest features of the development of manufacturing and service activity in developing countries, yet discussion of urbanisation is strangely absent from the usual economic analyses of growth and development.
Sectoral concentration There is a substantial but rather old development economics literature on the changes in countries’ production structures typically associated with development, deriving from Chenery’s ‘normal’ pattern of sectoral change (Chenery 1960, Chenery and Taylor 1968). Using a trade framework, the specialisation of countries has been researched by a number of 6
authors, for example Leamer (1984, 1987) and Schott (2003), who find ‘development paths’ linking the structure of countries’ production to their factor endowments. As well as being quite dated, these studies are based on sectorally aggregate data, and seem to miss key aspects of modern specialisation. What do we know about specialisation during development in the 21st century? First, globalisation has created the possibility of trade in new products and services, and of a much finer pattern of specialisation. The impact of new technologies is perhaps best illustrated by the experience, already referred to, of India (and Bangalore in particular), which is able to export labour services embodied in bytes. The finer pattern of specialisation arises with the growth of production networks. Component parts and semi-finished goods can cross borders multiple times, and countries are able to engage in ‘vertical specialisation’, just producing one very narrowly defined part of the product. Data on these activities can be hard to obtain, as they do not correspond to the standard commodity classifications of trade. However, one of the fastest growing elements of world trade has been trade in parts and components, now accounting for around 30% of world trade in manufactures (Yeats 1998). A striking feature of growth has been the fact that many countries have done well in a few extremely narrow product segments. Once again, India’s success in software products is an example. Rodrik and Hausman (2002) point to the fact that Bangladesh is successful in exporting shirts, trousers and hats (but not bed-linen or footballs), while Pakistan does well in bed linen and footballs.
Pairing up similar countries (Bangladesh/ Pakistan, Honduras/
Dominican Republic, Korea/ Taiwan) Rodrik and Hausmann demonstrate the low level of overlap in the exports of these country pairs. For each of these countries more than 30% of exports to the US are accounted for by the top 4 product lines, measured at the 6-digit level (eg ‘hats and other headgear knitted or from textile material not in strips’). The data therefore reveals that not only is success frequently concentrated spatially, it is also concentrated sectorally.
Export growth Narrow sectoral specialisation can develop only if output is exported from the city, region or country, and rapid growth of exports is a further aspect of successful development. Overall 7
confirmation of the importance of export growth in the modern growth process is provided by Dollar and Kray (2000), who identified a set of developing countries they term the ‘globalizers’. They rank developing countries according to the decline in their tariff rates between the 1980 and the late 1990s, and the increase in their trade to GDP ratio, and select countries that are in the top 40 of both lists.1 Table 1 (first two columns) indicates how very much more open these countries became relative to the non-globalizers (all other developing countries), and gives per capita growth rates for the last four decades. The striking point is that while these countries fared worse than others in the 1960s and 70s, their performance was dramatically better during the 1980s and 1990s, with per capita growth of 5.3% pa compared to -0.8% pa for the non-globalizers. It is also noteworthy that these countries exports became increasingly oriented to high-income countries. The share of their exports going to a given set of high-income countries rose from 69% in 1980 to 78% in 1997. Nonglobalizing middle and low income countries actually started off with a higher share going to high income countries, but the share fell from 76% in 1980 to 72% in 1997. Table 1: Growth and trade performance of the globalizers. % fall in tariffs, 1980 late 90s
% increase in trade / GDP, 1980 late 90s
Annual growth pc income 1960s
Annual growth pc income 1970s
Annual growth pc income 1980s
Annual growth pc income 1990s
Globalizer s
64%
92%
1
1.7
2.6
5.3
Nonglobalizers
29%
1%
2.2
2.8
0.2
-0.8
50%
4.5
3.4
2.5
1.9
High income Source: Dollar and Kray (2000).
These findings make a convincing case that full participation in the world economy is an inherent part of modern economic growth. The point is made somewhat differently by Lindert and Williamson (2001). 1
There were 16 such countries, and the two African countries that came closest to the criterion were added, giving: Argentina, Bangladesh, Bolivia, Brazil, China, Costa Rica, Ghana, India, Malaysia, Mexico, Nepal, Philippines, Poland, El Salvador, Thailand, Uganda, Uruguay, Vietnam. The early trade liberalizers were Chile, Turkey, Hong Kong, Singapore, 8
...the empty set contains those countries that chose to be less open to trade and factor flows in the 1990s than in the 1960s and rose in the global living-standard ranks at the same time. As far as we can tell there are no anti-global victories to report for the postwar Third World. However, these findings do not establish a causal relationship between trade and growth, and still less do they identify particular trade policies as direct causes of growth. The fact that export orientation is an aspect of growth does not imply that an open trade policy is a sufficient condition for growth.
Our analytical approach will draw out the distinction
between conditions that may perhaps be necessary for growth, although are not sufficient to ensure it takes place. We now turn to this, with a simple analytical framework that can capture the aspects of growth outlined in this section.
3. Analytical framework: 1st and 2nd advantage. To understand the determinants of this growth process it is necessary to divide the driving factors into two sets, operating in distinct ways. One set we label first advantage. These are the factors that can be viewed as pre-conditions for the development of new activities; they are – rather loosely – necessary but not sufficient conditions for growth. They include most of the factors that traditional theory has focussed on, such as access to inputs (labour and capital), access to markets, and provision of basic infrastructure and institutional environment. The other group of factors we call second advantage; these comprise the skills, technical know-how, networks and market linkages involved in efficient modern production. The distinguishing criterion is that these are all factors that inherently have some element of increasing returns to scale. Furthermore, these increasing returns are often external to the firm, so associated with market failure. It is increasing returns that underlies the lumpy performance described in the previous section. Before looking at elements of 1st and 2nd advantage in detail, we spend the rest of this section outlining some theory on how they interact to shape the growth process. We do this by drawing on work on sub-national inequalities (Venables 2003). However, the rather broad messages that we draw out have wider applicability, to international and intersectoral South Korea and Taiwan are in the rich country group, not in this list of globalizers. 9
developments (see for example Puga and Venables 1999 and Fujita et al 1999 for more fully developed models). Suppose an activity in a particular location – location s – has average labour productivity p(s)a(n(s)). The first part, p(s), measures the 1st advantage of location s, capturing all aspects of the environment that, while specific to the location, are exogenous for the firm and sector under study. (We discuss the components of this in detail in the next section). The second part is a function a(.) of the level of activity in location s, which can be measured by employment, n(s). This function measures 2nd advantage, and we assume that it is initially increasing and then decreasing, capturing increasing returns to scale that become progressively exhausted. In contrast to 1st advantage which is location specific, 2nd advantage is represented by the same function at all locations, although this is evaluated at location specific values of the endogenous variable, n(s). The economy has many potential locations for the activity, and we rank these such that p(s) is high for low values of s. For simplicity, we also assume that total employment, N = Σs n(s), is exogenous and increasing through time. What form does growth take in this economy? Equilibrium patterns of development are illustrated in Figure 1. The horizontal axis is total employment, N, so the economy moves to the right along this axis through time. The vertical axis is employment in each location and the curves on the figure illustrate equilibrium employment in each location, n(s).2 This employment level is determined by entrepreneurs producing in any location where the value of output per worker exceeds the wage, and hiring workers to the point where output is equal to the wage, p(s)a(n(s)) = w. The wage is, in turn, a function of total employment, w(N), rising as workers are drawn out of agriculture into the new activities, w’(N) > 0; perfect migration means that the wage is the same in all locations. The first Figure, 1a, illustrates the case where there is no 2nd advantage, so a(n(s)) is a constant, the same for all locations. There is however 1st advantage heterogeneity, and this maps in a continuous way into outcomes. Better locations (those with low s, and hence high p(s)) get activity sooner and are larger than worse ones.
Thus, the curve n(1) gives
employment in the location with the best 1st advantage, n(2) in the second best, and so on.
2
The figures are generated from a simple numerical example, available on request. 10
Increasing N causes both intensive and extensive growth, enlarging existing locations and make it profitable for new locations to gain activity. Figure 1b gives the equilibrium when 2nd advantage returns to scale are present. The figure is obviously much less continuous. Once a particular location starts to attract activity increasing returns cut in and lead to rapid growth. Outcomes are determined by a combination of the 1st advantage of each location, determining the sequencing, and the 2nd advantage of increasing returns, determining the growth path of each location. We learn several things from comparison of Figures 1a and 1b. The first is that development, in the presence of 2nd advantage, is uneven; locations do not develop in parallel but in sequence. Second, development is more concentrated. For any value of total employment, N, activity is concentrated in fewer and (on average) larger locations. Third, 1st advantage determines the order of development and, at any point in time, predicts which locations are successful and which are not. However, the dependence of performance on 1st advantage is not continuous. Very small differences in 1st advantage can translate into large differences in outcomes for locations at the margin of development; indeed, if two locations were ex ante identical, it might be the case that only one has the activity, the choice of which being a matter of historical accident. Conversely, improving a location’s 1st advantage does not necessarily confer any immediate payoff in terms of attracting the activity, as the location may remain below the current threshold. And amongst the set of active locations, differences in 1st advantage have little effect on employment, as they are dominated by acquired 2nd advantages. What about the efficiency of the equilibrium development pattern? The final panel, Figure 1c, gives the socially efficient structure of activities (with the same functions p(s)a(n(s)) as Figure 1b). As is apparent, efficiency involves more locations developing sooner than is the case at the equilibrium, although reaching smaller eventual scale.
The source of the
inefficiency is essentially a coordination failure, meaning that there is a role for policy to raise welfare by accelerating the rate at which activities are initiated in new locations. In the equilibrium of Figure 1b new activity is initiated when it becomes privately profitable for an individual small agent, but this agent does not internalise the effects of his action on others in the location. Clearly, the precise policy response to this depends on the source of the increasing returns, which we discuss in the next section. Here we simply note that free entry of ‘large developers’ who take the entire surplus created by the activity in a location, will be 11
sufficient to move the equilibrium to the social optimum. We note also that these efficiency statements are merely about the distribution of activity for each given value of N. The presence of market failure makes it likely that equilibrium N will also be less than socially optimal. With this structure in mind, let us now turn to discussing some of the elements of 1st and 2nd nature advantage. 4. Sources of advantage A new agenda for growth is emerging. This is driven in part by the cross-country literature which tries to identify the policy and institutional factors which explain key differences in economic performance across countries (see Temple, 1999; Djankov et al, 2003; Rodrik, 2003). This literature is moving in a microeconomic direction where the challenge now is to identify which institutional and policy choices are appropriate in different circumstances. This new growth agenda is also being driven by a microeconomic literature which attempts to directly evaluate the vast range of policy and institutional innovations which are taking place around the world. A key challenge in this literature is assess the extent to which lessons learnt are portable to other settings. Both literatures share the common focus of trying to understand which policies and institutional reforms are appropriate for encouraging growth and reducing poverty in specific regions and countries. In this section we draw on these literatures to try and identify what might be some of the key elements of 1st and 2nd advantage identified in the previous section. Our coverage is illustrative rather than complete. To make things more concrete we focus in Section 5 on one country, India, and present some recent evidence which suggests that certain key elements if 1st and 2nd advantage have a central bearing on structural change and growth in that country.
4.1: 1st advantage: A location can be favoured by its endowments, its geography, its institutions and its policies. These are all heavily researched topics, and here we just provide a brief overview of some of what is known. 12
Labour Production decisions are taken on the basis of the value of output net of wage and other costs. The first point then is, that low wages (given labour quality) are a source of 1st advantage. This suggests the possibility of ‘leapfrogging’ (Brezis et al 1993): new activities will be deterred from going into centres of established activities in which wages (or prices of other immobile factors) have been bid up. The other side of the equation is the quality of labour. The need to promote human capital accumulation in developing countries is well established. The best estimates for developing countries, from micro-econometric studies that take endogeneity and measurement errors issues seriously, line up with those for developed countries -- each additional year of schooling is associated with a 6-10% increase in earnings (Duflo, 2001; Krueger and Lindhal, 2001). This evidence appears robust across both methods and locations. This suggests an agenda where investment in education is used as a means of encouraging growth and attacking poverty. Dreze and Sen (2002), for example, identify lack of basic education in many parts of India as a key factor constraining economic and social development. Low levels of education and literacy are identified as one of the key reasons why India has not experienced the high rates of growth experienced by countries in East Asia. This still leaves open the central question of how such an expansion in education might be achieved. New work in the area is paying much more attention to this, specifically -- the market conditions under which it is provided and the incentives faced by education providers. One strand of research focuses on policy design. Banerjee et al (2003) have used randomised experiments to look at whether NGO implemented remedial education and computer assisted learning programs in India affected learning. Randomized intervention has also been used in Western Kenya, for example, to evaluate whether improving the quality of education by increasing the supply of textbooks and improving child health affect attendance and attainment (see Glewwe, Kremer and Moulin, 2000; Kremer and Miguel, 2002). Another strand focuses on whether there needs to be changes in the organization of how policy is delivered. Political representation, decentralization and involvement of non-government organizations and the private sector are major issues here (see e.g. Chattopadhyay and Duflo, 2001; Hsieh and Urquiola, 2002). The question of how social relations between ethnic groups 13
affects delivery is also a central theme here (see Miguel and Gugerty (2002)). Policy thinking on the way to expand human capital typifies a change in the way that economists now think about development. There is general support for the idea that institutions for delivery matter and should be the object of reform. Moreover, only good micro-evidence will enlighten the processes at work. The link between the micro and macro picture can then be formed. Capital Innovative means of expanding access to credit are also attracting attention as a means of reducing poverty. The large cross-country literature on credit shows a strong correlation between financial depth and growth (see, for example, King and Levine, 1993). In the context of poverty reduction, one observation from the field is that the poor tend not to have access to banks and other formal financial institutions. Thus aggregate credit expansion may not necessarily deliver benefits to the most disadvantaged groups. A recent theoretical literature emphasizes links between inequality and development via the operation of credit markets. Even if the poor have access to investment opportunities, it may be difficult for them to exploit them (see Banerjee and Newman, 1993; Aghion and Bolton, 1997). This suggests that expansion of access to credit may be critical to encouraging growth and structural change and of getting people out of poverty traps. As in the case of schooling, the prevailing view has now shifted firmly towards an appreciation of the fact that mechanisms of expansion and delivery matter. One line of attack on this issue has been to look at the functioning of informal institutions which, to some extent, have filled the void left by market and state failure to reach the poor (see Besley, 1995). Another line has been to look at whether changing the way that formal credit institutions deliver credit makes a difference. A much heralded and recent innovation in this area has been micro-finance institutions like the Grameen Bank which target the poor and rely on peer selection and peer monitoring to overcome the need for collateral. These schemes are typically operated by non-government organizations.
But assessing their
effectiveness requires policy evaluation. In the case of the Grameen Bank, there are recent studies that shed light on the ability of credit to affect livelihoods and poverty (Pitt and Khandker, 1998, Morduch, 1998).
14
A central concern in this literature is whether changes in institutional design can overcome the problems of elite and political capture which have plagued formal credit. This emerging micro evaluation literature is capable of generating specific directions for policy and institutional reform which cannot be gleaned from the cross-country finance and development literature. Trade and market access Globalization and its effects on economic development have been the subject of an intense and passionate debate over the last decade. The optimistic view argues that trade liberalization, and the implied elimination of barriers to competition, is the right road for developing countries to promote growth and eradicate poverty (see, for example, Dollar and Kray (2001, 2002), Frankel and Romer (1999), Sachs and Warner (1995) and World Bank (2001)). Skeptics object that there can be no such progress without an active role for domestic institutions and policies to correct market failures (Rodrik and Rodriguez (2000), Rodrik (2002)). Some authors have argued that there are instances, especially in very poor economies, where liberalization is in fact detrimental to growth, by inhibiting infant industries and the local accumulation of knowledge (Krugman (1981), Haussman and Rodrik (2002), Young (1991), Stiglitz (1995, 2002). Others emphasize that, in the presence of capital market imperfections, liberalization exacerbates income inequality within countries, imposing high costs on less favored regions, social groups or sectors of activity, and with ambiguous effects on average performance (see Banerjee and Newman (2003), Trefler and Zhu (2001)). Recent work in this area is focusing on examining the microeconomic impact of recent trade liberalization episodes. Using plant level data for Chile Pavcnik (2001), for example finds evidence of within plant productivity improvements that can be attributed to liberalised trade for plants in the import competing sector. Export performance can be hampered or helped not only by trade policy, but also by geography. Evidently, much of the success of the northern border region of Mexico is a direct consequence of its proximity to the US, and coastal regions of China do well because they are on the coast. A number of studies have tried to quantify the importance of these geographical factors. Gallup and Sachs (1999) regress national per capita income on four variables; a measure of the endowment of hydro-carbons, a dummy variable for the incidence of malaria, and two measures of geography; internal geography, captured by the proportion of population within 15
100km of the coast, and external geography, as measured by shipping costs (the cif/fob ratio). They find that these four variables account for 69% of the per capita variation in incomes across countries. Redding and Venables (2003) show how a large part of the growth of countries’ exports can be accounted for by two measures of their access to markets, together with a measure of institutional quality. One is ‘external geography’, measuring where the country is relative to sources of growing import demand, and the other ‘internal geography’, measured by the proportion of land area close to coastline or navigable rivers. Each of these factors are significant in explaining export growth, although their quantitative importance varies across regions. The basic results from their study are reported in Figure 2. The first bar is the deviation in each region’s exports/GDP from the world average, and remaining bars (summing to the first bar) the contributions of external geography (or foreign market access), internal geography, institutions, and the residual (the regional dummy). Picking some examples, West Europe’s high trade is explained by high scores on all variables, while Oceania’s average performance comes from poor external geography combined with good internal geography and good institutions.
Sub-Saharan Africa’s low exports can largely be accounted for by poor
geography and institutions, with only around 30% of the poor performance left to the residual. In the context of our present argument, the key point is the contrast between these cases and SE Asia, where 80% of performance is not accounted for. Our suggestion is that this is a region where 2nd advantage forces were operating powerfully during the period, but they are simply missed by the standard methods that focus on 1st advantage. We suspect that examination of the residuals of many growth regressions would yield similar findings, pointing again to the need for research with more focus on micro mechanisms, and based on more disaggregate data. Property Rights and Contracts The importance of well-defined property rights has been a staple issue in the economic development literature. Recent cross-country papers have established a link between property rights and income per capita. Acemoglu, Johnson and Robinson (2001) instrument risk of expropriation using settler mortality and find that insecure property rights are associated with lower income per capita. Hall and Jones (1999) show that a social infrastructure variable (which captures both the quality of government institutions and openness to trade) is positively associated with income per capita. These papers suggest a strong mechanism for 16
property rights reform working through increases in aggregate output. On the other hand, how to map from these findings into concrete policy suggestions is not clear. In this respect recent subnational studies are pointing the way. There is increasing evidence, for example, that secure land rights, in particular, are an important vehicle for promoting both equity and efficiency. Banerjee, Gertler and Ghatak, (2001) show that increases in tenurial security in West Bengal also had large positive effects on agricultural output. In a similar vein, Lin (1992), for example, showed that the move from collective to household farming in China starting in 1978 led to large productivity increases in agriculture. In all cases, these reforms yielded gains for low-income groups. This is in line with Besley and Burgess (2000)'s results that there was significant redistribution effected by land reform acts passed at the state level over the 1958-1992 period. Recent literature has also begun to steady contracting in non-agricultural situations to understand the underpinnings of investment decisions in firms. Banerjee and Duflo (2000) emphasize the importance of reputations in enforcing contracts in the Indian software sector. McMillan and Woodruff (1999) use survey data on firms in Vietnam to study the importance of social networks in access to credit and investment decisions. These studies share in common an emphasis on the role of social networks in promoting business development. They provide an important context for policy measures focusing on improvements in business climate to promote investment and the development of firms. Regulation The post-war model of economic development followed in many countries saw the state as the main actor in promoting growth (see Rostow, 1960). To this end, a huge variety of regulations were put in place to influence private actors. In some countries, these regulatory structures still remain, while there has been deregulation in certain spheres in others. One view of regulation is that it is implemented by benevolent governments intent on fixing market failures. In so far as such market failures underpin the failures of the poor to access higher incomes, this could be thought of as a pro-poor agenda. However, there is increasing empirical evidence that (noble as the intentions of the architects of regulation may have been), regulation has not been either an engine of economic development nor a boon for the
17
poor.3 With many developing countries having passed through long periods of central planning, deregulation of the economy to improve the climate for investment and entrepreneurship is increasingly emphasized. Obviously claims about the impact of regulation can only be assessed on a case-by-case basis. Economic analysis is increasingly playing a role in putting some structure on this problem and in identifying specific directions for reform. This raises the specter of appropriately structured de-regulation being part of the growth promotion and anti-poverty agenda. Economic analysis is increasingly playing a role in identifying specific directions for de-regulation that help the poor. One key theme is to provide an improved climate for investment and entrepreneurship. For example, Djankov et al. (2002) collect data on the time and number of procedures an entrepreneur must complete to officially open a business in 85 developing countries. They find that heavy regulation of entry is associated with less democratic governments, greater corruption and larger unofficial economies – which supports the idea that entry regulations are not in the public interest. Besley and Burgess (2002) find that pro-worker labor regulations in India are associated with lower investment, productivity and output in registered manufacturing. Output in unregistered manufacturing, in contrast, is increased, which is in line with the Djankov et al (2002) who find that countries with many entry regulations tend to have larger informal sectors. Quantification of the different elements of these costs at the sub-national level is central if we are to move towards identifying the key building blocks of a business environment which facilitates growth and poverty reduction (Sutton, 2002). Few exercises in development economics are more important. 4.2: 2nd advantage: 2nd advantage arises because of increasing returns to scale in the sector and location under study. There are a number of different sources of such advantages, varying in mechanism and in sectoral and geographical scope.
3
This comes as no surprise to students of the political economy of regulation -- see, for example, Stigler, (1971) and Shleifer and Vishny, (1998). 18
Technological capabilities Success in a particular product line requires mastery of the technology.
This point is
particularly emphasised in the work of Sutton (2002) who argues that to compete successfully, a firm must be within a relatively small window, defined by production costs and product quality.
Acquiring capability requires fixed outlays on R&D and related
activities, so is associated with increasing returns to scale. There may also be learning by doing, amplifying such effects. For Rodrik and Hausman (2002) the process is one of experimentation, with firms having to incur upfront expenditures to learn about the comparative costs of their location. Technological spillovers: Spillovers mean that productivity depends on the activities of others, and is an evident source of market failure and under investment.
The geographical and sectoral scope of these
spillovers is crucial for 2nd advantage effects; if spillovers are global then the function a() depends just on a world value of n, so is not a source of local advantage. For Rodrik and Hausman (2002) the results of experiments become public knowledge; however, the knowledge revealed is specific to the activity/ location combination. If knowledge spillovers occur through job turnover they are also likely to be concentrated. Sutton (2002) argues that knowledge is embodied in teams of workers, and that it is difficult to move an entire team between locations. Turning to the evidence, the consensus is that information spillovers attenuate rapidly over space (for example Jaffe et al 1993 on patent citations and Rosenthal and Strange 1999 on patterns of plant births).
Face-to-face communications remain important, despite
technological changes (Gasper and Glaeser 1999, Storper and Venables 2002).
Spatial
density of activity is also found to raise productivity. For example, for the US, Ciccone and Hall (1995) find, from work on US states, that doubling employment density raises labour productivity by 6%. As for the sectoral spread of spillovers, there is a long-standing debate in the economic geography literature between the relative importance of ‘Jacob’s externalities’ and ‘MarshallArrow-Romer (MAR) externalities’. The former derive from the diversity of a city, the argument being that innovation and creativity is stimulated by serendipitous interaction in a large and diverse economic environment, such as New York City. 19
By contrast, MAR
externalities are knowledge spillovers within narrowly defined sectors, as information on how to produce is transmitted from firm to firm. Detailed empirical studies of productivity spillovers in developing countries confirm the importance of MAR externalities. The positive dependence of productivity on employment levels in the same city and industry has been found by studies for a number of countries (eg for the US and Brazil, Henderson 1988, Indonesia, Henderson and Kuncoro 1996, and Korea, Henderson et al, 1999) We should note that acquisition and spread of knowledge is important in selling as well as in production. Roberts and Tybout (1996) establish the importance of learning effects in firms export behaviour, and Evenett and Venables (2003), using a fine commodity and geographical disaggregation, show that learning to export also has a destination specific component. Market linkages A further source of 2nd advantage derives from input-output linkages between firms. If there are trade costs of any sort, so that goods cannot be shipped costlessly between locations, then firms value proximity to supplier and customer firms. Downstream firms gain from being close to their suppliers, and vice versa. These linkages constitute a pecuniary externality which raises productivity in the sector/location concerned, and gives rise to clustering of activity, just as outlined in Section 3 (see Fujita et al 1999).
5. Structural change in India In this section we use India to illustrate how our framework may be applied to generate cleaner and more robust insights into which factors drive structural change and growth. The basic pattern of structural change in India is shown in Figure 3 where we graph out real agricultural and non-agricultural output per capita over the 1960-1997 period for the sixteen main states of India. With the exception of the Punjab and Haryana agricultural output per capita has remained relatively flat over the period. Aggregate economic growth has been driven in large part by the rise in non-agricultural output. In 1960 average real state nonagricultural and agricultural output per capita were of similar magnitudes at Rs. 406 and Rs. 425 respectively. By 1997, non-agricultural output at Rs. 2814 was more than double agricultural output (Rs. 1266). And as Figure 3 illustrates the experiences of different states 20
with structural change have been strikingly different. States such as Andra Pradesh, Gujarat, Haryana, Karnataka, Kerala, Maharashtra, Punjab, Tamil Nadu and West Bengal have experienced high rates of non-agricultural output growth. These states located mainly in the south and west of India have grown quickly from low initial levels of non-agricultural production. In contrast, Assam, Bihar, Jammu & Kashmir, Madhya Pradesh, Orissa, Rajasthan and Uttar Pradesh have only experienced very limited structural change with levels of real non-agricultural per capita remaining close to those for non-agricultural output. Understanding which institutions and policies account for different rates of structural change across Indian states represents a key challenge. Our concern with structural change and growth is driven ultimately by a concern with welfare. India is almost unique amongst developing countries in having carried out household expenditure surveys on a regular basis since the 1950s. This allows consistent and comparable series of rural and urban poverty measures to be constructed (see Ozler, Datt and Ravallion, 1996). In Figure 4 we graph rural and urban poverty headcounts (which capture the percentage of the rural and urban population in poverty) for the sixteen main Indian states across the 1958-2000 period.4 Over this period 49 percent of the rural and 41 percent of the urban population are classified as poor. If we compare Figures 3 and 4 what is striking is that states which experienced rapid structural change also experienced faster rural and urban poverty reduction. Poverty tends to be both high and persistent in laggard states where nonagricultural growth failed to materialise. In short, structural change and growth appear to matter for poverty reduction.5 Simple regression analysis helps to bring this point home. In Table 2 we show regressions of poverty measures on output per capita measures using the panel data on the sixteen main Indian states for the 1958-1992 period. Column (1) shows that increases in the levels of non-agricultural output are associated with significant falls in overall poverty.
In contrast, changes in
agricultural output do not affect poverty. We find a similar pattern when we break poverty 4
We are grateful to Gaurav Datt and Martin Ravallion for providing us these state-level poverty figures (see Ozler, Datt and Ravallion (1996)). Gaurav Datt was kind enough to provide us with comparable updates which allowed us to extend the series from 1994-2000. 5 Using Indian state level data Ravallion and Datt [1996] have established a robust link between income growth and poverty reduction. They find that the elasticity of the incidence of poverty with respect to net domestic product per capita was –0.75 for the period 19581991. This figure is close to the elasticity of –0.76 which Besley and Burgess [2003] find using cross-country data for developing economies. 21
into its urban and rural components in columns (2) and (3). It is the growth of nonagricultural output which is exerting a significant effect on urban and rural poverty. In columns (4) and (5) we look at manufacturing within non-agricultural output. In India this is divided into two sectors: (i) a registered sector which cover firms with employment levels above ten with power and twenty and (ii) a unregistered or informal sector which covers firms below these thresholds.6 In columns (4) and (5) we see that increases in per capita registered manufacturing output reduce urban but not rural poverty. This makes sense as these larger firms are located mainly in urban areas. In contrast increases in output from firms in the unregistered manufacturing sector is significantly negatively associated with rural poverty but not urban poverty. Differential emergence of small businesses appears important for explaining the pattern of rural poverty reduction across Indian states.
Model Log ag output pc Log non-ag output pc Log reg manu output pc Log unreg manu output pc State effects Year effects Adjusted R2 Observations
Table 2: Poverty and Output in India (1) (2) (3) (4) Total Urban Rural Urban headcount headcount headcount headcount OLS OLS OLS OLS cluster(state) cluster(state) cluster(state) cluster(state) 4.22 1.46 4.60 1.53 (1.37) (0.48) (1.27) (0.56) -16.27*** -9.34*** -16.18*** (3.00) (2.42) (2.23) -4.51*** (3.20)
YES YES 0.83 318
YES YES 0.88 318
YES YES 0.79 318
(5) Rural headcount OLS cluster(state) 3.49 (0.91) -0.40 (0.15)
-1.91 (1.65)
-5.99*** (2.52)
YES YES 0.89 318
YES YES 0.79 318
Absolute standard errors in parenthesis. Standard errors adjusted for clustering by state. *significant 10% level, ** significant 5% level, ***significant 1% level. Sample is a panel of the 16 main Indian states for the period 1958-2000. Regressions only include years when NSS survey carried out.
These observations bring into sharp relief how important understanding the microeconomic determinants of structural change is. Below we look at some examples of work on India which look at this issue. Our focus is on manufacturing, often seen as the engine of structural change. India represents a rich laboratory for looking at institutional and policy determinants
6
The registered sector makes up about 9 percent of state output and the unregistered sector 5 percent. 22
of structural change. The fact that it is a federal democracy has meant that different states have implemented different policy and institutional reforms which can be evaluated whilst controlling for other factors. The growing availability of data at both the state, household, industry and firm level has meant that careful microeconomic analysis of policies is becoming more feasible. Our examples our chosen to demonstrate how 1st advantage factors and the interplay between 1st advantage and 2nd advantage factors determine the pattern of manufacturing growth in India. As will become apparent their remain large gaps in our understanding of the microeconomic drivers of structural change in India. The examples, however, do illustrate how the framework we have developed in this paper can be used build up an evidence base on what matters. Example [1]: 1st advantage: labor regulation and registered manufacturing Traditional views of the growth process put development of manufacturing at centre stage in the structural change accompanying economic development. A casual look at the performance of some of the more successful Asian economies after 1960 adds credence to this view. For example, between 1960 and 1995, manufacturing as a share of GDP grew from 9 percent to 24 percent of GDP in Indonesia, 8 percent to 26 percent in Malaysia and 12.5 percent to 28 percent in Thailand. All of these countries had strong overall growth performances and saw significant falls in absolute poverty. In contrast, the Indian economy did not experience a significant expansion of manufacturing as a share of national income. Manufacturing output constituted 13 percent of GDP in 1960 (ahead of the countries listed above) but grew to only 18 percent of GDP by 1995.7 India's overall growth over this period was also relatively modest and it did not exhibit the extent of declining absolute poverty experienced elsewhere in Asia. In Figure 5 we observe that registered manufacturing performance differs strikingly across states and time. Certain states: Andhra Pradesh, Gujarat, Karnataka, Tamil Nadu and Maharashtra show striking growth in registered manufacturing, while states like Assam, Bihar, Jammu and Kashmir, Uttar Pradesh and West Bengal display relative stagnation (albeit from very different base levels). The fact that states within a centrally planned economy performed so differently in terms manufacturing is striking. Much of the literature so far has 7
Figures on manufacturing shares come from various issues of the World Development Indicators, World Bank, Washington D.C. 23
focused on whether the Indian industrial strategy, which involved extensive use of industrial licensing and tariff and non-tariff barriers, led Indian manufacturing to perform poorly relative to other countries (see Singh [1964], Bhagwati and Desai [1970] and Bhagwati and Srinivasan [1975]). In particular, relative to countries in East Asia which experienced rapid manufacturing growth (World Bank [1993] and Bhagwati [1998]). Nationwide policies, however, have limited scope to explain the marked cross-state variation in manufacturing performance that we observe in Figure 5. To understand these patterns we need to focus on state specific policies and conditions. And here the notion of the ‘business’ or ‘investment’ climate which manufacturing firms face in different parts of India has taken center stage [see Stern, 2001]. Labor regulations have been identified as an important element of investment climate in India (Stern, 2001; Dollar, Iarossi and Mengitsae, 2001, Sachs et al, 1999). Besley and Burgess (2002) examine whether it an important source of 1st advantage which could help explain differences in manufacturing performance across Indian states. To do this they exploit two key facts: (i) labor regulations only apply to firms in the registered manufacturing sector (ii) the Indian constitution empowers state governments to amend central legislation. The key piece of central legislation relating to industrial disputes is the Industrial Disputes Act of 1947 which sets out the conciliation, arbitration and adjudication procedures to be followed in the case of an industrial dispute. This Act has been extensively amended by state governments during the post-Independence period. Besley and Burgess (2002) read the text of each amendment and coded each as pro-worker (+1), neutral (0) or pro-employer (-1). The pattern of change across Indian states is shown in Figure 6 where it is apparent that the direction of regulation varies across states and time. By coding these amendments they are able to ascertain whether labour regulations in a state are moving in a pro-worker or proemployer direction. Besley and Burgess (2002) then check whether the pattern of regulatory changes shown in Figure 6 the pattern of manufacturing development shown in Figure 5. Column (1) shows that moving in a pro-worker direction is associated with lower overall per capita manufacturing output levels. This effect is accounted for by the fact that pro-worker labor regulation led to less output in registered manufacturing (column (2)). Investment in this sector is lower in states with more pro-worker labor regulation. Column (3) shows that the effect goes the other way for unregistered manufacturing. That is states with more pro-worker 24
labor regulations tend to have larger informal manufacturing sectors. This makes sense as where workers are able to extract more of the rents from production in registered sector, capitalists will prefer to remain in the unregistered sector where labor has no power. In columns (4) and (5) we see that regulating in a pro-worker direction is also associated with increases in urban poverty but does not affect rural poverty. This reflects the fact that the effects of labor regulation are mainly being felt in the registered sector which is found mainly in urban areas. These results suggest that attempts to redress the balance of power between capital and labour can end up hurting the poor. Table 3: Labor Regulation and Manufacturing Output in India (1) (2) (3) (4) (5) Urban Rural Log unreg Log reg Log total headcount headcount manu output manu output manu output pc pc pc Model OLS OLS OLS OLS OLS cluster(state) cluster(state) cluster(state) cluster(state) cluster(state) -0.071** -0.178*** 0.081*** 2.17*** -0.976 Labor (2.09) (2.82) (2.41) (3.25) (0.62) regulation [t1] State effects YES YES YES YES YES Year effects YES YES YES YES YES Adjusted R2 0.92 0.93 0.75 0.88 0.80 Observations 509 508 509 547 547 Source Besley and Burgess (2002). Standard errors adjusted for clustering by state. Absolute t statistics in parenthesis.*significant 10% level, ** significant 5% level, ***significant 1% level. Sample is a panel of the 16 main Indian states for the period 1958-1992.
As Besley and Burgess (2002) show the policy choices of state governments in India as regards labor regulation have strongly affected manufacturing performance. 1st advantage factors like regulation which are in part under the control of sub-national governments will have a strong bearing on whether or not manufacturing develops in areas under their jurisdiction. And this in turn will have welfare consequences for citizens in those regions. It is important to note that the large differences in manufacturing performance were present well before liberalization in 1991.8 This suggests that countries or regions wishing to develop manufacturing must pay attention to the 1st advantage factors which affect the business climate which firms face and not only to the trade regime. The institutional environment affects the investment and location decisions that entrepreneurs make.
8
Besley and Burgess (2002) restrict their econometric analysis to the 1958-1992 period in order to better identify the impact of domestic state level policies. 25
Example [2]: 1st advantage: rural banks and unregistered manufacturing Our focus this far has been on the registered manufacturing sector. The majority of individuals in low income countries, however, do not have the choice of working in a factory. This is particularly the case for those resident in rural areas. For them the relevant choice might be between remaining in agriculture or starting a small business. These new activities are typically more productive than basic agriculture and thus encourage both growth and poverty reduction. Understanding which factors drive structural change by facilitating the emergence of small businesses is a key challenge. Burgess and Pande (2003) try to make some inroads into this issue by evaluating whether a massive rural branch expansion program in India affected structural change and growth in India. Over the 1961-2000 period over 30000 new branches were opened in rural areas. The rationale was for the program was simple. The government identified lack of access to finance as a key reason why growth was stagnant and poverty persistent in rural areas. The failure of banks to enter rural areas was seen as a brake on entrepreneurship and the emergence of new activities. To address this Indian central bank first nationalised commercial banks in 1969 and then imposed a license rule in 1977 which state that for each branch opened in a banked location (typically urban) banks had to open four branched in unbanked location (typically rural). This rule was removed in 1990 and branch building in rural areas came to a halt. As a result of the imposition of the 1:4 rule, states which had fewer banks per capita before the program in 1961 received more bank branches between 1977 and 1990 leading to an equalization in population per bank branch. Entrepreneurs, small businessmen and agriculturalists were explicitly targeted in the mandated lending practices of banks. The program is an example of government targeting a key 1st advantage factor as a means of encouraging the spread of new economic activities across rural India. To evaluate the program Burgess and Pande (2003) use these 1977 and 1990 trend breaks in the relationship between initial financial development and rural branch expansion attributable to license regime shifts as instruments for the number of branches opened in the rural unbanked locations. Some key results are contained in Table 4.
In column (1) we see that
rural branch expansion is associated with higher total manufacturing output per capita. This is in line with the main finding of the paper that the banking of rural India was associated with 26
increases in total output driven by a rise in non-agricultural production. Agricultural output, in contrast, was unaffected. Column (2) and (3) show that the rise in manufacturing was concentrated in the unregistered sector. Column (3) suggests that the arrival of banks in rural areas helped people to start and expand small businesses. By addressing a key 1st advantage factor the (albeit forced) entry of banks is seen to have been a spur for entrepreneurship and structural change. In contrast, registered manufacturing which is located mainly in urban areas was unaffected (column (2)). In column (4) we see that the structural changed engendered by rural branch expansion had positive consequences in terms of reducing rural poverty. Urban poverty, in contrast, is unaffected (column (5)). Table 4: Rural Banks and Manufacturing Output in India (1) (2) (3) (4) Rural Log unreg Log reg Log total headcount manu output manu output manu output pc pc pc Model IV IV IV IV cluster(state) cluster(state) cluster(state) cluster(state) -0.15* 0.04 0.28* -5.02** No branches (2.10) (0.57) (1.85) (2.14) opened in rural unbanked locations pc State effects YES YES YES YES Year effects YES YES YES YES Adjusted R2 0.94 0.94 0.81 0.75 Observations 579 579 579 627
(5) Urban headcount IV cluster(state) -0.82 (0.92) YES YES 0.92 627
Source Burgess and Pande (2003). Standard errors adjusted for clustering by state. Absolute t statistics in parenthesis. *significant 10% level, ** significant 5% level, ***significant 1% level. Sample is a panel of the 16 main Indian states for the period 1961-2000. Robust standard errors are reported in parentheses. The number of branches in rural unbanked locations is normalized by 1961 population. The two instruments for this variable are the number of banked locations in 1961 per capita interacted with (i) an indicator variable which equals one if the year>1976 and a post 1976 time trend (ii) an indicator variable which equals one if the year>1989 and a post 1989 time trend respectively.
Example [3]: interplay between 1st and 2nd advantage: technological capability and trade liberalization But how do 2nd advantage factors which are intrinsic to firms and industries interact with 1st advantage factors which affect the environment in which firms function? As we have argued it is this interplay of factors which is critical to understanding modern economic growth. Such interactions are difficult to study empirically as 2nd advantage factors are themselves endogenous.
27
One way in which we can get some purchase on this issue is to look at how the technological capability of firms or industries affects how they respond to a liberalization shock. That is, in line with the theory in section, we might expect firms in the same industry to respond differently to the threat of entry depending on how competitive they are with the new firms and products. How firms respond will in part depend on the technological choices they have made in the past. Firms that are close technological frontier are in a better position to compete with new firms and products. They may respond by investing in new technology to retain existing markets and to benefit from increased access to new markets. In contrast, investment incentives in laggard firms are likely to be blunted and these firms are likely to be harmed by liberalization. This makes it clear that the responses we see to the same liberalization shock are likely to be heterogeneous across geography and sectors. Heterogeniety in post-reform performance is also likely to be harmed magnified if the institutional environment in which firms are located varies. That is we can think of interactions between market access and other 1st advantage factors leading to differential performance in two firms which are have identical technological capabilities. To look at this issue of interplay between 1st and 2nd advantage factor Aghion et al (2003) exploit the fact that India experienced a massive trade liberalization in 1991. This shock was common across firms in the same industry. However, firms in the same 3-digit industry varied in terms of their level of pre-reform productivity which can be taken as proxy of technological capability. On dimensions such as labor regulation, the institutional environment in which firms function also varied across Indian states as we studied above. In Table 5 we study how interactions between the liberalization reform (captured by dummy variable which switches on in 1991), pre-reform technological capability (captured by the ratio of labor productivity in a 3-digit state-industry relative to the most productive stateindustry in India) and pre-reform state-level investment climate (captured by the labor regulation measure) affected performance in the registered manufacturing sector in India. We look at these relationships using a panel of 3-digit state-industries for the period 1980-1997. Regressions contain state-industry fixed effects which allows us to examine how output, employment and productivity in 3-digit state-industries was affected by these factors. In the first row of Table 5 we see that state-industries which were closer to the Indian productivity frontier (for that industry) saw greater increases in output, employment, labor productivity and total factor productivity. This shows that state-industries with greater technological 28
capability benefited more from liberalization. In contrast, laggard state industries which were below median productivity for India experienced below trend rates of increase in output, employment, labor productivity and total factor productivity following liberalization in 1991. Manufacturing performance is therefore a function of the interplay between a 1st advantage factor (market access) and a 2nd advantage factor (technological capability). In the second row of Table 5 we observe, in line with the results presented in Table 3, that state-industries located in states with more pro-worker labor regulation saw less growth in output, employment, labor productivity and total factor productivity for the 1980-1997 period. This makes it clear that institutional environment (as captured by 1st advantage factors such as labor regulation) in which 3-digit industries function affects how well they perform. Moreover in the third row of Table 5, where we report results for the interaction between labor regulation and liberalization, we observe that the negative impact of pro-worker regulation is magnified in post liberalization period. That is it is even more damaging in terms of output, employment, labor productivity and total factor productivity growth for state-industries to be located in a pro-worker state when market access is increased. These findings from Aghion et al (2003) emphasize that the initial level of technology and institutional context mattered for whether and to what extent industries and states in India benefited from liberalization. Table 5: Technological Capability, Liberalization and Manufacturing Performance in India (1) (2) (3) (4) Log total factor Log total reg Log reg manu Log reg man productivity manu output employment output per worker Model IV IV IV IV cluster(state) cluster(state) cluster(state) cluster(state) Pre-reform 0.574*** 0.412*** 0.162*** 0.096*** proximity*reform (8.20) (9.36) (3.68) (2.59) Labor regulation -0.076*** -0.044*** -0.032*** -0.038 (3.80) (2.75) (3.20) (2.92) Labor -0.066*** -0.054*** -0.011*** -0.041 regulation*reform (7.33) (0.007) (2.75) (8.20) State-industry fixed YES YES YES YES effects Year dummies YES YES YES YES 2 R 0.88 0.86 0.83 0.61 Observations 19623 19623 19623 18528 Source Aghion, Burgess, Redding and Zilibotti (2003). Standard errors adjusted for clustering by state-year. Absolute t statistics in parenthesis. *significant 10% level, ** significant 5% level, ***significant 1% level. Sample is a three dimensional unbalanced panel of 3digit industries in the 16 main Indian states 1980-1997.
29
These findings have interesting implications for policy. First, in line with the theory outlined in Section 3 it is clear that the level of productivity pre-reform matters for the post liberalization trajectory. The key importance of the interplay between 1st advantage and 2nd advantage factors has a critical bearing on the pattern of post liberalization growth. This of course begs the question of why level of productivity were different in different 3-digit industries in different states in the first place. That is beyond the scope of the analysis however it does point to the understanding what determines the industrial capabilities of firms as being a central area of research (see Sutton, 2002). Second, the institutional environment matters in the extent to which liberalization will be truly growth-enhancing. For instance, rigidities in the labor market may limit the positive impact of trade liberalization. The institutional and policy choices that state governments make have a central bearing on the extent to which the regions they govern will benefit or not from liberalization. Third, liberalization may have adverse effects on industries and regions that are initially less developed. This may call for complementary measures to offset the negative distributional consequences of reforms, e.g., investment in infrastructure and support for knowledge acquisition in backward areas
6. Conclusions Working out which factors which drive structural growth has always been central to development economics. As growth theory has developed and disaggregated data become available this work has been moving in a microeconomic direction. Some broad correlations are emerging in new cross-country work which tries to explain the very different growth experiences of countries. These serve as useful signposts for more detailed microeconomic work though we are at an early stage in that process. For example, though there may be some consensus that access to finance matters for entrepreneurs it is unclear how such access should be provided (e.g. private banks, development banks, microfinace). And what is appropriate may depend on the type of business and how developed the legal system and other institutions are. This microeconomic evaluation work described in this paper represents the frontier for identifying what the 1st advantage pre-conditions for growth are. This approach based on 30
careful building of microeconomic evidence has become standard in developed economies like the US and UK. In low income countries this work is at an earlier stage. However, there is a growing consensus that evidence building via evaluation of policy and institutional reforms represents the best approach for identifying what works. The 1st advantage examples we provide from India illustrate this process at work. But, as this paper makes plain, identifying sources of 1st advantage is not the whole story. The spatial/sectoral pattern of modern economic growth suggests that it is the result of a complex interaction of 1st advantage factors and what we call 2nd advantage factors. These factors which include technological capabilities, technological spillovers and market linkages are the results of investment and location decisions made by firms. Future growth will thus in part be function of past investment and location decisions made by entrepreneurs. As the example from India brings out, it is the interaction between 1st advantage and 2nd advantage which determine where and in which sectors growth takes place. This observation brings into sharp relief the fact that fact modern economic growth is the result of interactions between institutions and organizations. Understanding the factors that drive the technological, investment and location decisions of firms, which are key concerns in the industrial organization literature, are therefore central to understanding modern economic growth. These decisions will themselves be a function of institutions and other 1st advantage factors which make the empirical study of growth challenging. A few key policy implications emerge from the paper. First, where 1st advantage elements do not exist rapid structural change and growth is unlikely to take place. This backward state of affairs often does not reflect the preferences of citizens. Much recent work has therefore focussed on how institutional innovations and changes in the manner in which policy is delivered can be used to tackle the 1st advantage factors which engender backwardness. Second, ensuring that 1st advantage elements are present is no guarantee that rapid economic growth will take place in a given location or sector. As we have shown there are numerous examples of growth only taking place in very narrow sectors and of fast growing locations being adjacent to backward locations even when both locations are subject to the same institutions and policies. 2nd advantage factors are thus central to understanding economic growth in low income contexts. Whereas we have made some progress in identifying the necessary 1st advantage factors for growth to take place we have made less progress in understanding what drives 2nd advantage factors like the technological capability of firms. 31
References Acemoglu, Daron; Johnson, Simon; James A. Robinson, [2001], “The Colonial Origins of Comparative Development: An Empirical Investigation”, American Economic Review; 91(5), 1369-1401. Aranya, R. (2002) ‘Spatial disparity in the globalizing city: the case of the silicon plateau emerging in Bangalore’ processed, University of Trondheim and at http://www.wider.unu.edu/conference/conference-2003-1/conference2003-1.htm Banerjee, A and Newman, A. (2003) `Inequality, Growth and Trade Policy', MIT, mimeograph. Banerjee, A., S. Cole, E. Duflo and L. Linden, (2003 ) “Improving the Quality of Education in India: Evidence from Three Randomized Experiments” mimeo MIT. Banerjee, Abhijit and Esther Duflo, [2000], “Reputation Effects and the Limits of Contracting: A Study of the Indian Software Industry,” Quarterly Journal of Economics, 115(3), 989-1017. Banerjee, Abhijit V., Gertler, Paul J. and Maitreesh Ghatak, [2002], “Empowerment and Efficiency: Tenancy Reform in West Bengal”, Journal of Political Economy , 110(2), 23980. Besley, Timothy, [1995], “Saving, Credit and Insurance,” in Jere Behrman and T.N. Srinivasan (ed) Handbook of Development Economics , Vol. IIIa Amsterdam: North Holland. Besley, Timothy and Robin Burgess, [2000], “Land Reform Poverty and Growth: Evidence from India,” Quarterly Journal of Economics, 115(2), 389-430. Besley, Timothy and Robin Burgess, [2002], “Can Labor Regulation Hinder Economic Performance? Evidence from India” CEPR Discussion Paper No 3260. Besley, Timothy and Robin Burgess, [2003], “Halving Global Poverty” forthcoming Journal of Economic Perspectives, Spring Issue. Bhagwati, Jagdish [1998] in Alhuwalia, Isher and Ian Little (eds) India's Economic Reforms and Development: Essays for Manmohan Singh (Delhi: Oxford University Press) Bhagwati, Jagdish and Padma Desai, [1970], India: Planning for Industrialization (Delhi: Oxford University Press) Bhagwati, Jagdish and T.N. Srinivasan, [1975], Foriegn Trade Regimes and Economic Development (Delhi: McMillan) Burgess, Robin and Rohini Pande, [2003], “Do Rural Banks Matter? Evidence from the Indian Social Banking Experiment” typescript, LSE and Columbia University. 32
Brezis, Elise S., Paul R. Krugman, and Daniel Tsiddon, (1993) "Leapfrogging in International Competition: A Theory of Cycles in National Technological Leadership," American Economic Review, , 83, 1211 Chattopadhyay, Raghabendra and Esther Duflo, [2001], “Women as Policy Makers: Evidence from an India-Wide Randomized Policy Experiment,” typescript, MIT. Ciccone A. and R. Hall (1996) , ‘Productivity and the density of economic activity’ American Economic Review 86, 54-70. Cikurel, D, (2002), ‘Why Mexico’s regional convergence broke down’, processed University of California, San Diego and at, http://www.wider.unu.edu/conference/conference-20025/conference2002-5.htm. Demurger, S, J. Sachs, W.Woo, S. Bao, G. Change and A. Mellinger, ‘Geography, Economic Policy and regional development in China’ processed University of California, Davis and at http://www.wider.unu.edu/conference/conference-2002-2/conference2002-2.htm Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer, [2002], “The Regulation of Entry,” Quarterly Journal of Economics 117(1), 1-37. Djankov, S., Glaeser, E., La Porta, R., Lopex de Silanes, F., Shleifer, A. (2003) “The New Comparative Economics” CEPR Discussion Paper No 3882. Dollar, David and Art Kraay, [2000], “Growth is Good for the Poor,”typescript, the World Bank. Dreze, J. and Sen, A. (2002) “India: Development and Participation” (Oxford: Oxford University Press) Duflo, Esther. [2001], “Schooling and Labor Market Consequences of School Construction in Indonesia: Evidence from an Unusual Policy Experiment”, American Economic Review ; 91(4): 795-813. Evenett, S. and A.J. Venables (2003) ‘Export Growth in Developing Countries: Market Entry and Bilateral Trade Flows’ processed, LSE. Feenstra, R.C (1998) ‘Integration of trade and disintegration of production in the global economy’, Journal of Economic Perspectives, 12, 31-50. Frankel, J and Romer, D (1999) `Does Trade Cause Growth?', American Economic Review, 89(3), 379-99. Fujita, M., P. Krugman and A.J. Venables, (1999) ‘The spatial economy; cities, regions and international trade’, MIT press Cambridge MA. Gallup, J.L. and J. Sachs (with A.D. Mellinger), (1999) ‘Geography and economic development’, in B. Pleskovic and J.E. Stiglitz (eds), Annual World Bank Conference on Development Economics, 1998, World Bank, Washington DC. Glewwe, Paul, Michael Kremer and Sylvie Moulin, [2000], “Textbooks and Test Scores: 33
Evidence from a Prospective Evaluation in Kenya” mimeo Department of Economics, Harvard University. Hall, Robert E and Chad Jones, [1999], “Why Do Some Countries Produce So Much More Output per Worker than Others?” Quarterly Journal of Economics, Vol. 114, pp. 83-116. Henderson, J.V. (1988) ‘Urban development; theory, fact and illusion’ OUP, Oxford, Henderson, J.V. and A. Kuncoro (1996) ‘Industrial Centralization in Indonesia’, World Bank Economic Review, 10, 513-540 Henderson, J.V., T. Lee and J. Lee ‘Scale Externalities in a Developing Country, Journal of Urban Economics, 2001, 49, 479-504. Hirshman, Albert (1958) ‘The strategy of economic development’ (New Haven: Yale University Press) Hseih, Chang-Tai and Miguel Urquiola, “When Schools Compete, How Do They Compete? An Assessment of Chile's Nationwide School Voucher Program,” typescript Princeton University. Hummels, D. (1999a) ‘Have international transportation costs declined’, mimeo, Chicago. Hummels, D. (1999b) ‘Towards a geography of trade costs’, mimeo, Chicago. Hummels, D. (2000), ‘Time as a trade barrier’, mimeo Purdue University. Hummels, D., J. Ishii and K-M Yi (2001) ‘The Nature and Growth of Vertical Specialization in World Trade’ Journal of International Economics, 75-96. Kaldor, Nicholas, [1967], “Strategic Factors in Economic Development”, (Ithaca: Cornell University Press). King, Robert and Levine, R. [1993]: “Finance and Growth: Schumpeter might be Right” Quarterly Journal of Economics, CVIII, no. 3 August, 717-738. Kremer, Michael and Edward Miguel, [2001], “Worms: Education and Health Externalities in Kenya” NBER Working Paper #8481. Krueger, Alan B. and Mikael Lindhal [2001], “Education for Growth: Why and For Whom?” Journal of Economic Literature XXXIX (4), 1101-1136. Krugman, P. (1981) ``Trade, Accumulation and Uneven Development'', Journal of Development Economics 8(2), 149-161 Leamer, E. (1984) ‘Sources of comparative advantage’, MIT press, Cambridge MA Leamer, E. (1987) ‘Paths of development in the three factor n- good general equilibrium model’, Journal of Political Economy, 95. Lewis, W.A., [1954], ``Economic Development with Unlimited Supplies of Labour'' The Manchester School of Economic and Social Studies 22, 139-191 34
Limao N. and A.J. Venables (2001), ‘Infrastructure, geographical disadvantage, transport costs and trade’, World Bank Economic Review, ?????. Lin, Justin Yifu, [1992], “Rural Reforms and Agricultural Growth in China,” American Economic Review, 82(1), 34-51. McMillan, John and Christopher Woodruff, [1999], “Interfirm Relationships and Informal Credit in Vietnam,” Quarterly Journal of Economics, 114(4), 1285-1320. Miguel, Edward and Mary Kay Gugerty, [2002], “Ethnic Diversity, Social Sanctions, and Public Goods in Kenya”, typescript, Department of Economics, Berkeley Morduch, Jonathan, [1998], “Does Microfinance Really Help the Poor: Evidence from Flagship Programmes in Bangladesh” mimeo Department of Economics, Harvard University Murphy, K. A. Shleifer and R.W. Vishny (1989),’Industrialization and the big push’, Journal of Political Economy, 97, 1003-26 Ozler, B. Datt, G. and Ravallion, M., [1996], ``A Data Base on Poverty and Growth in India'' mimeo, World Bank. Pavcnik, N. (2002) “Trade Liberalization, Exit, and Productivity Improvements: Evidence from Chilean Plants,” Review of Economic Studies 69, pp. 245-76 Pitt, M. and Khandekar, S. [1998]: “The Impact of Group—Based Credit Programs on Poor Households in Bangladesh: Does the Gender of Participants Matter?” Journal of Political Economy 106 (5). Redding, S. and A.J. Venables, (2000) ‘Economic Geography and International Inequality’, CEPR DP, 2568, revised version at http://econ.lse.ac.uk/staff/ajv/winc.pdf Redding, S. and A.J. Venables, (2003) ‘Geography and Export Performance: external market access and internal supply capacity’, NBER wp 9637 and at http://econ.lse.ac.uk/staff/ajv/research_material.html#isit Rodrik, Dani (2003) “Growth Strategies” mimeo Harvard University Rodrik, D. and R. Hausmann (2002) ‘Economic development as self-discovery’, NBER wp no 8592. Rodriguez, F and Rodrik, D (2000) `Trade Policy and Economic Growth: A Skeptic's Guide to the Cross-national Evidence', Macroeconomics Annual, NBER and MIT Press. Rostow, W.W. (1960) The Stages of Economic Growth (Cambridge: Cambridge University Press) Roberts, Mark J., and James R. Tybout. 1997. “The decision to export in Colombia: an empirical model of entry with sunk costs.” American Economic Review 87: 545-64. Sachs, J and Warner, A (1995) `Economic Reform and the Process of Global Integration', 35
Brookings Papers on Economic Activity, 0(1), 1-95. Schott, P. (2003) ‘One Size Fits All? Heckscher-Ohlin Specialization in Global Production’ American Economic Review (Forthcoming) Schumpeter, J. (1934) “The Theory of Economic Development” (Cambridge: Harvard University Press) Shleifer, Andrei and Vishny, Robert, [1998], The Grabbing Hand: Government Pathologies and Their Cures (Cambridge: Harvard University Press) Singh, Manmohan, [1964], India's Export Trends (Oxford: Clarendon Press)) Sokoloff, Kenneth (1986) “Productivity Growth in Manufacturing in Early Industrialization: Evidence from the American Northeast, 1820-1860” in Stanley Engerman and Robert Gallman “Long Term Factors in American Economic Growth” (Chicago: Chicago University Press) Stern, N. (2001) ‘A strategy for development’, (Washington: World Bank) Stigler, George, [1971], “The Theory of Economic Regulation,” The Bell Journal of Economics, Spring, 3-21. Stiglitz, J (1995), Wither Socialism?, MIT Press, Cambridge, MA. Stiglitz, J (2002), Globalization and Its Discontents, WW Norton and Co Sutton, John, [2002] “Rich Trades, Scarce Capabilities: Industrial Development Revisited” The Economic and Social Review 33(1): 1-22. Temple, J. (1999) ‘The new growth evidence’, Journal of Economic Literature, XXXVII, 112-156. Trefler, D and Zhu, S (2001) `Ginis in General Equilibrium: Trade, Technology and Southern Inequality', NBER Working Paper, 8446 World Bank [1983], The East Asian Miracle, (Oxford: Oxford University Press). World Bank [2001], Globalization, Growth and Poverty: Building an Inclusive World Economy (New York: Oxford University Press) Yeats, A. (1998) ‘Just how big is global production sharing’ World Bank Policy Research Working Paper no. 1871. Young, A (1991) `Learning by Doing and the Dynamic Effects of International Trade', Quarterly Journal of Economics, 106, 369-406.
36
n(s)
n(7)
n(1) n(2)
N Figure 1a: Development without 2nd advantage
n(s)
n(1)
n(2) n(7)
N Figure 1b: Equilibrium development
n(s)
n(1) n(2)
n(7) Figure 1c: Optimal development
N
Latin America
North America
"k (V)
"k (M)
"k (t)
"k (c)
East Europe
SE Asia
West Europe
:k
S-S Africa
ME and NAfrica
Other Asia
Oceania
Figure 2: Regional export performance, 1994-97: Contributions of foreign market access "k (M), internal geography "k (t), insititutions "k (c), and residual :k
agricultural output per capita
non-agricultural output per cap
Andhra Pradesh
Assam
Bihar
Gujarat
Haryana
Jammu & Kashmir
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
3000 2000 1000 0
3000 2000 1000 0
3000 2000 1000 0
3000 2000 1000 0 1960 1970 1980 1990 2000
1960 1970 1980 1990 2000
1960 1970 1980 1990 2000
year
Figure 3: Output Per Capita In India
1960 1970 1980 1990 2000
rural poverty headcount
urban poverty headcount
Andhra Pradesh
Assam
Bihar
Gujarat
Haryana
Jammu & Kashmir
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
1960 1970 1980 1990 2000
1960 1970 1980 1990 2000
1960 1970 1980 19902000
1960 1970 1980 1990 2000
80 70 60 50 40 30 20 10 0
80 70 60 50 40 30 20 10 0
80 70 60 50 40 30 20 10 0
80 70 60 50 40 30 20 10 0
year
Figure 4: Rural and Urban Poverty in India
registered manufacturing output
unregistered manufacturing outp
Andhra Pradesh
Assam
Bihar
Gujarat
Haryana
Jammu & Kashmir
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
600 500 400 300 200 100 0
600 500 400 300 200 100 0
600 500 400 300 200 100 0
600 500 400 300 200 100 0 1960 1970 1980 1990 2000
1960 1970 1980 1990 2000
1960 1970 1980 1990 2000
year
1960 1970 1980 1990 2000
Figure 5: Manufacturing Output Per Capita in India
Andhra Pradesh
Assam
Bihar
Gujarat
Haryana
Jammu & Kashmir
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
1960 1970 1980 1990 2000
1960 1970 1980 1990 2000
1960 1970 1980 1990 2000
1960 1970 1980 1990 2000
4 2 0 -2
direction of labor regulation
-4
4 2 0 -2 -4
4 2 0 -2 -4
4 2 0 -2 -4
year
Figure 6: Labour Regulation in India
