Catalog

Walking on two legs

Even after countries begin to industrialize and workers move off the farm to find jobs in industry and services, agriculture continues to play an important role. Although agricultural productivity growth may not be able to match that in the other two sectors, modest gains provide a basis for industrialization by ensuring a steady supply of affordable agricultural produce to urban workers, as well as livelihoods for the large numbers of workers who remain. In some economies (such as Taipei,China and Korea) land reforms and policies that support rural livelihoods (e.g., the FELDA scheme in Malaysia) have played an important role both in supporting the broad expansion of agrarian and rural incomes, and in regulating the flow of workers out of agriculture. In turn, rising rural incomes have helped constitute a market base that allows industries to expand.

The transfer of agricultural land to industrial and commercial uses is also another important part of the overall process of change and growth. But as the experience of the PRC and India attests, this process can be politically and economically fraught if rights are unclear and institutions are weak or badly governed. Lifting agricultural productivity growth and ensuring an orderly and politically acceptable distribution of land are important challenges, and no state can afford to neglect them.

But from the perspective of economic catch-up and the creation of jobs, agriculture is not where developing Asia's future lies. That future lies elsewhere. Figure 3.1.17 shows the historical relationship between changes in agricultural output and employment shares and GDP growth. The historical pattern is striking. Growth is strongly and inversely correlated with agricultural output and employment shares in developing Asia and in the rest of the world. In only a handful of cases is positive growth associated with increasing agricultural shares in output and employment, and these are for countries with extensive and productive agricultural land frontiers, which is not a feature of most countries in developing Asia.

But how relevant are "old models" of industrialization and growth for understanding how developing Asia might evolve? Are service activities going to take on new significance and become the locomotive that moves developing Asia forward? Or do industry and manufacturing incubate dynamism in a way that is unique? These are some of the questions considered in this section.

Growth and structural transformation

The starting point is to consider the ways in which the evolution of economic structure has in the past been linked to economic growth. Is growth uniquely associated with the expansion of industrial or manufacturing output shares? Figure 3.1.18 shows the relationship between economic growth and changes in the shares of industrial output and employment over the past 35 years for a broad sample of countries in the international economy.

The data in the figure appear to provide compelling evidence that industry "matters. "Those countries that have increased their industry shares most have, on average, grown more quickly. Likewise, those countries where employment shares in industry have risen most have enjoyed faster GDP growth. In developing Asia, many countries have sustained growth and expanded their industrial output shares, including Cambodia, PRC, India, Indonesia, Korea, Lao PDR, Malaysia, Sri Lanka, Thailand, and Viet Nam. But others that grew had declining industrial output shares. In general, growth was slower in these economies, which include several Central Asian economies (e.g., Armenia, Kazakhstan, and Uzbekistan) and Hong Kong, China. In the case of Hong Kong, China, a hallmark of its development has been the shift to highly productive services. In Central Asia, declining industrial output shares reflect the retirement of moribund activities that were a creation of the earlier Soviet planning model.

Data for manufacturing output shares tell a similar story (employment data are unavailable). Figure 3.1.19 documents the positive correlation between the change in manufacturing output shares and overall output growth. Countries in the first quadrant with the highest increases in the manufacturing share and in the output growth rate are Cambodia, Indonesia, Korea, Lao PDR, Malaysia, and Thailand.

In the PRC, the manufacturing share in total output has been traditionally much higher than anywhere else in developing Asia, although it declined with respect to the average of the 1980s. It still accounts for over one third of total output, only matched in developing Asia by Malaysia, Tajikistan, and Thailand. The share of manufacturing employment, on the other hand, has declined from about 15% in the 1980s to 11% now, a result of the restructuring of heavy industries that were owned by the state. The share of India's manufacturing output is significantly lower than the PRC's, and over the sample period hovered around 15–16%, while the share of manufacturing employment has been at around 11%. In recent years (2005–2006), the tempo of activity in Indian manufacturing has picked up. Box 3.1.5 illustrates how the size of manufacturing industries might be measured and gauged in an international perspective.

It would seem that industry, and manufacturing in particular, has had an important role to play in growth in developing Asia. The countries that have grown most quickly also tend to be "overrepresented" in manufacturing. It is also the case that countries that have developed complex export baskets, which tend to have a high share of manufactured exports, have also grown quickly (Box 3.1.6). But it is also important to ask what the relationship between growth and service shares looks like. In the section Looking back, it was shown that resources move out of agriculture into both industry and services. Figure 3.1.20 shows the links between growth and changes in services shares in output and employment.

The relationship between services share in output and growth is negative, but not significant. Larger shares of services are, in a broad international panel, associated with slower growth. But this is not surprising, since the panel includes rich countries. These move at the pace of the frontier, where services are a big part of the economy. In developing Asia, a pattern of slowing growth is readily evident as incomes in the NIEs escalate toward the OECD frontier. There is basically no systematic relationship between growth of output and services share in employment. Most observations are clustered in the first quadrant because most countries have growth and most countries have seen their share of services employment rise.

How does developing Asia's services share stack up when measured against international norms? Box 3.1.7 reports the results of an exercise to answer this question.

Clearly, industrialization and growth of output are closely associated (as shown earlier). But has industrialization been a prerequisite for output growth? To look at this question, an event analysis is undertaken in which episodes of growth are compared with preceding and concurrent evolutions in the pattern of output.

The methodology followed is similar to that of Hausmann et al. (2005b). First, growth is defined in terms of a moving average that is calculated as the annual (exponential) growth rate over a 7-year period (i.e., from t+1 to t+7; from t+2 to t+8, etc.). Using these moving averages, growth episodes are identified. Definitions of "rapid growth," "growth acceleration," and "sustained growth" are given in Box 3.1.8.

Table 3.1.9 identifies all cases in developing Asia of "rapid growth" and "growth accelerations" since the mid-1960s (depending on data availability). The number of episodes of rapid growth in the region has been high, a total of 302, with an average growth rate of 7.3%. The countries with the highest number of rapid growth episodes are the PRC and Singapore, with 28 each. The other NIEs and the ASEAN-4 countries (except the Philippines) have more than 20 such episodes. The number of growth accelerations is obviously much smaller, but nevertheless high, a total of 34, with the average acceleration being 6.55 percentage points. Accelerations often correspond with "take-offs" in economic growth (e.g., Other Southeast Asian countries), growth recoveries (e.g., Malaysia after its 1985–86 recession), or natural resource discoveries (e.g., Azerbaijan).

Although the fastest accelerations are seen in Azerbaijan and Tajikistan, of more than 20 percentage points, some of the very high accelerations in the Central Asian republics are really "bounces" after contractions (and the one-time events surrounding the breakup of the ex-Soviet Union). This is also the case for some Pacific islands (e.g., Kiribati, Solomon Islands). Apart from these two "special cases," PRC, Malaysia, and Thailand had growth accelerations of over 4 percentage points.

The average growth acceleration for those countries whose growth before the acceleration was positive (so eliminating Bangladesh, Central Asia, and the Pacific countries with contraction before the acceleration) is 4.28 percentage points. Most countries in developing Asia have experienced at least one instance of growth acceleration in the last few decades (Kiribati with three, and PRC and Papua New Guinea with two each). Bhutan, Micronesia, Sri Lanka, and Tonga did not have any.

Of the 24 growth accelerations for which the exercise could be undertaken, 13 were of nonsustained growth, and 11 had sustained growth. Of these 11, six (the NIEs and the PRC twice) also had rapid growth during the 7-year period preceding the growth acceleration. Information on these is shown in Table 3.1.10

Have changes in the structure of output been uniquely identified with these episodes of rapid growth, accelerating growth, or sustained growth? Clues may be provided by comparing levels and changes in the shares of output before and around these episodes. The first row of Table 3.1.11 records shares of industry, manufacturing, and services around the time of the growth episodes, and the second row, the shares immediately before the episode. The third row presents t-statistics, where the null is that the shares in both periods are equal.

The results of Table 3.1.11 show that episodes of rapid growth are preceded by rising industry, manufacturing, and services shares in aggregate output. The share of industry rises by 1.3 percentage points, that of services by about 1 percentage point, and that of manufacturing by 0.5 percentage points. Though modest, these differences are statistically significant. But there is no readily detectable link between growth accelerations and changes in output shares. However, sustained growth is associated with an increase in the share of services, and not with changes in either industry or manufacturing shares.

It is difficult to draw strong conclusions from these findings. As both industry and services shares rise during episodes of rapid growth, this implies that agriculture shares fall prior to rapid growth episodes (confirming the relationship shown in Figure 3.1.1 above). The relationship between services and sustained growth probably reflects the fact that the share of services in output expanded over a wide range of per capita incomes in the NIEs during a period in which they also grew quickly.

Another way to dissect the data is to split observations into episodes: rapid growth and nonrapid growth; episodes of growth accelerations and no growth accelerations; and episodes of sustained growth and growth that was not sustained. These episodes can then be cross-tabulated with changes in the shares of industry, manufacturing, and services output. So, for example, the number of episodes of rapid growth with increasing industry shares can be compared with the number of rapid episodes where there was no increase in industry shares. Likewise, episodes in which growth was not rapid can also be split into those cases associated with expansion of industry shares and with nonexpansion of industry shares.

Table 3.1.12 provides a breakdown for episodes of rapid and nonrapid growth. In each cell, two numbers are presented. The top number is the number of counts for events identified in the corresponding row and column. So, for example, there were 73 cases of rapid growth where there was no preceding increase in industry's share in output. But there were also 174 cases of rapid growth where industry's share did rise. The numbers in italics at the bottom of each cell refer to the number of observations that would be predicted if rapid growth and changes in industry shares were (statistically) independent of each other. So, randomly, there would be 96.5 expected occurrences of rapid growth and no increase in industry's share.

By comparing the number of actual with expected observations it is possible to test whether changes in sector shares and growth are independent or not. In the case of rapid growth events, the chi-square rejects the null hypothesis of independence for industry and manufacturing output shares. Moreover, by comparing the cell counts with their expected values, it can be confirmed that rejection occurs because there is a positive association between an increase in industry's (manufacturing's) share and rapid growth. In the case of services, however, the null hypothesis cannot be rejected, suggesting that there is no systematic relationship between increases in the share of services and subsequent episodes of rapid growth.

Similar tests conducted on growth accelerations and episodes of sustained growth (for which sample sizes are much smaller) failed to reject the null of non-association for changes in industry, manufacturing, and services shares.

Finally, a probit regression was estimated. In this equation, the dependent variable is a dummy variable that takes the value of one at the time of rapid growth and zero otherwise. Dependent variables were changes in manufacturing (or industry) and services shares. The results indicate that a rise in the manufacturing share increased the probability of rapid growth by 3.7%. (The coefficient of services is negative but insignificant.)

The data show that expansion of industrial and manufacturing output shares is also positively associated with growth. The complexity and sophistication of a country's export basket, which is likely to be positively influenced by a heavy weight for manufacturing goods, is a statistically significant predictor of subsequent growth. This analysis leans to the conclusion that rapid output growth is more closely tied to expanding industry and manufacturing shares than to services shares. However, growth accelerations and sustained growth are not systematically correlated with changes in output shares at all.

The role of services would appear to have been more complicated. Services shares have risen in both slow- and fast-growing economies. Successful episodes of industrialization are likely to have been supported by the parallel development of efficient services infrastructure (see below). For slow-growing countries, services may have played an important role in mopping up surplus labor released from agriculture.

What roles might be played by industry and services in moving ahead?

Industry

Developing Asia's success in industrialization-in particular the development of a vibrant manufacturing sector that competes on a global scale-is unrivalled. Figure 3.1.22 illustrates vividly how developing Asia's manufacturing industry has ascended in global markets from the 1970s when its share was still miniscule. The PRC, NIEs, and ASEAN-4 in particular have seen significant growth. Other Southeast Asian economies are now just beginning to register on the global scene. Fears that the PRC would close opportunities for other countries have proven unfounded. Instead, its emergence has helped forge new patterns of production and specialization with East and Southeast Asia that build on complementarities, and a refined division of tasks (see the chapter, Trade and structural change in East and Southeast Asia, in Part 1). Engaging in these complex production networks requires that countries continue to look outward but build internal capabilities that will enable repositioning and rebalancing as circumstances change. Different paths are possible. Some countries may focus on the production of intermediate goods, as in Singapore, or on the development and branding of final goods, which is more akin to what Korea has done.

But not all countries have fared equally well. In South Asia, India's emergence in global manufacturing has been sedate, and it has lost ground to the NIEs and ASEAN-4 and long since been overtaken by the PRC. More generally, growth of industry and manufacturing in South Asia has been listless when measured on a global scale. Within ASEAN, the Philippines has also become bogged down, and Indonesia has lost much of its momentum following the Asian crisis.

The development of a vibrant industrial and manufacturing base is likely to be an essential ingredient in development strategies for some time to come. As before, success will pivot on acquiring those capabilities needed for continuous upgrading. Indeed, the premium on the self-adapting capabilities may increase if the life cycle of some activities is shortened, either as a result of more intensive competition in international markets, or an acceleration of technological progress. Protectionism presents a potent risk as scale economies, diversity, and technological upgrading depend critically on big markets.

Drawing on past experience, a stylized trajectory for industrialization is still likely to involve: first, establishing a narrow base in a labor-intensive manufacturing industry, such as garments or footwear; then, diversifying into new and gradually more sophisticated activities; before eventually specializing in areas where a competitive advantage has been built and consolidated. What precisely a country will produce at any particular point in time, and how it will migrate to new activities and change its basket of manufactured goods and exports, seem to depend on country-specific and idiosyncratic factors (Hausmann et al. 2005a). There is striking evidence that participating in export markets that are expanding quickly on a global scale and in which the rich industrialized countries are also participating can sustain growth (see e.g., Trade and structural change in East and Southeast Asia, in Part 1, and Hausmann et al. 2005a). But policy has an important role to play in at least two ways: removing blockages to doing business and investment, and incubating conditions in which the private sector can experiment and learn what it can do profitably. Box 3.1.9 above sets out some blockages that have hindered industrialization in India and the Philippines in the past. Prospects will depend on easing these constraints no