رشد بهره وری در تولید آسیایی: بررسی فرضیه پاداش های ساختاری

This paper examines the role of structural change in explaining aggregate productivity growth in the manufacturing sector of four Asian countries over the period 1963–1993. The conventional shift-share analysis is used to measure the impact of shifts in both labour and capital inputs. The results do not support the structural-bonus hypothesis, which states that during industrial development, factor inputs shift to more productive branches. This finding is robust, even when the conventional shift-share analysis is modified to take into account increasing returns to scale as described in Verdoorn's law. It is argued that improvements in productivity levels were widespread and depended negatively on the distance from the global technology frontier, confirming the Gerschenkronian notion of catch-up.

Processes of modern economic growth and catch up do not merely involve a significant increase in productivity levels, but also entail changes in the distribution of inputs and outputs across sectors. Kuznets stated that ‘it is impossible to attain high rates of growth of per capita or per worker product without commensurate substantial shifts in the shares of various sectors’ (Kuznets, 1979, p. 130). The hypothesis that structural change is an important source of growth and productivity improvement, is a central tenet of the growth-accounting literature (Maddison, 1987) and is derived from classical models of a dual economy (Lewis, 1954). Assuming the existence of surplus labour in some parts of the economy, a shift of labour towards modern industry will be beneficial at the aggregate level, as workers with low productivity will be put to more productive uses. Various studies have shown that this shift has been an extra source of aggregate productivity growth in addition to any sectoral productivity growth in many countries (see Syrquin, 1984, for an overview). The field recently attracted new attention in attempts to explain the East Asian miracle (Lucas, 1993; Young, 1995; Nelson and Pack, 1999). Most studies focus on the shift from agriculture to manufacturing and have little to say about the importance of shifts within the manufacturing sector. Nevertheless, the industrial development literature suggests that in the course of economic growth, labour and capital shift from less productive manufacturing branches towards more productive branches. As a consequence, aggregate productivity growth in the manufacturing sector will be boosted in addition to any intra-branch growth. We call this the ‘structural-bonus hypothesis’. Using data for 13 manufacturing branches, the hypothesis is tested in this paper for four rapid growing Asian countries: India, Indonesia, South Korea and Taiwan for the period from 1963 to 1993. Use is made of the conventional shift-share analysis introduced by Fabricant (1942) to analyse productivity effects from shifts of labour. The method was extended by Massell (1961) to incorporate shifts in labour and capital simultaneously. The shift-share methodology is still a popular tool to decompose aggregate productivity growth (see Paci and Pigliaru (1997) or Fagerberg (1999) for recent applications). One of the main criticisms of the conventional shift-share analysis is the neglect of increasing returns to scale as described in the Verdoorn Law (Verdoorn, 1949). As stressed by Kaldor (1966), increasing returns include both static returns due to processes of labour division and specialisation in growing sectors and dynamic effects as technical progress is fostered by output growth. Because they feature prominently in many endogenous growth models, the study of increasing returns to scale recently attracted renewed attention (Fingelton and McCombie, 1998; Harris and Lau, 1998). If returns to scale differ across sectors, as argued by Kaldor and inputs shift to the sectors with higher returns, the effects of structural change on productivity growth are bigger than indicated by the conventional shift-share analysis. This paper proposes a modified shift-share analysis which takes into account Verdoorn effects. Using estimates of sector-specific effects, the importance of structural change for aggregate productivity growth can be better assessed. The remainder of the paper is organised as follows. Section 2 presents arguments in favour of the structural-bonus hypothesis for the manufacturing sector based on the industrial development literature. This is followed by a discussion of the data in Section 3. The shift-share analysis is applied for labour productivity growth in Section 4. Using data on capital stock, we present an analysis of total factor productivity growth in Section 5, which takes into account not only shifts of labour, but also shifts of capital. In Section 6, a variety of possible sources for biases in the conventional shift-share analysis are discussed, focusing in particular on the implications of the Verdoorn Law. We propose a modified shift-share analysis in Section 7 that explicitly takes into the account dynamic scale effects of Verdoorn. This decomposition formula uses branch-specific estimates of the Verdoorn effect to measure the importance of structural change for aggregate productivity growth.

We set out to study the impact of shifts of factor inputs on aggregate productivity growth in the manufacturing sector. Using a conventional shift-share analysis we found that that the structural-bonus hypothesis is not supported by the evidence on Asian industrial development. Reallocation of inputs within the manufacturing sector did not provide an extra bonus to aggregate productivity growth, in addition to growth in individual branches. This holds not only for labour productivity growth but also for total factor productivity growth. We investigated the possibility of a particular bias in the shift-share analysis against finding empirical support for the structural-bonus hypothesis. In particular, we focused on the possible distorting effects of ignoring the Verdoorn effect. The shift-share analysis was modified to take into account this effect. Using estimates of branch-specific Verdoorn elasticities, it was shown that the sectoral Verdoorn coefficients do not differ significantly from each other and therefore, the inclusion of the Verdoorn effect has little influence on the decomposition results. This finding is a striking one and runs counter to the expectations derived from the industrial development literature which suggests that in the course of industrialisation, factor inputs move into more productive manufacturing industries. This is not to say that structural change is not important for economic growth. We examined one particular, albeit important, aspect of structural change, namely the impact of reallocation of factor inputs on aggregate productivity growth. We did not consider changes at the demand side of the economy, which together with developments at the supply side, determine the actual reallocations of output and inputs across industries. What the results of the shift-share analysis do show is that these reallocations did not provide an additional bonus to aggregate productivity growth in the manufacturing sector of the Asian countries. Instead, aggregate productivity growth is driven by widespread productivity improvements across all manufacturing branches. A major drawback of the shift-share analysis is the inability to account for inter-industry spillovers. Output growth in a sector with strong backward and forward linkages may promote growth of productivity in other sectors via rent- and/or knowledge spillovers. Although strong inter-industry spillover effects have been found in advanced countries, it is debatable whether similar effects are present in countries which operate far from the world technology frontier and undertake little R&D. The debate on this issue continues and further investigation into this matter, for example by using input-output tables, is desirable. Two main explanations for our finding of little support for the structural-bonus hypothesis can be given. First, the finding indicates that in rapid growing developing countries manufacturing-wide effects (Harberger's yeast effects) are more important than industry-specific effects (mushroom effects). This is in line with theories of conditional convergence, which point to the importance of general economy-wide factors for growth, such as the technological competence of a country's people, the availability of an overlay of financial and business services, the existence of an extensive physical infrastructure and high investment ratios (Abramovitz, 1989). Similarly, from a policy perspective, one can argue that policy changes in favour of macro-economic stability tend to affect most industries at the same time, rather than a limited number of industries. Hence productivity levels tend to improve in all manufacturing branches. There is another reason why this is possible. Previous research has shown that the Asian countries had very low levels of both labour and total factor productivity in all manufacturing branches from an international perspective (Timmer and Szirmai, 1999). This indicates that the technology gaps relative to the world productivity leaders are big. Whereas more developed countries are bound by global technological developments which are unevenly spread across industries, developing countries with sufficient technological capabilities have high opportunities for technology catch up and productivity growth in all manufacturing branches. This argument finds support in our finding that the initial productivity level in a branch relative to the level in the US was highly significant and negatively related to subsequent productivity growth. An explanation of productivity growth in Asian manufacturing cannot do without the Gerschenkronian notion of catch up.