"تکنولوژی مناسب" توضیح تفاوت رشد بهره وری: یک رویکرد تجربی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|11375||2005||15 صفحه PDF||سفارش دهید||6159 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Development Economics, Volume 77, Issue 2, August 2005, Pages 517–531
This paper aims at giving empirical content to the Basu and Weil (1998) [Basu, S., Weil, D.N., 1998. Appropriate technology and growth. Quarterly Journal of Economics 113, 1025–1054] theory of growth, in which localized innovation, assimilation of spillovers and differences in speeds of capital intensification yield diverse patterns of international convergence and divergence. The contributions of these sources to labor productivity growth are quantified for a sample of countries, using data envelopment analysis techniques. Regression analysis shows that the observed patterns were mainly driven by processes of creating spillover potential through capital intensification. Assimilation appears to be much slower than assumed in Basu and Weil's model.
Recently, Basu and Weil (1998) introduced a new theoretical model of international productivity growth dynamics, which could generate patterns of international productivity convergence and divergence that are more in line with reality than the results obtained from other endogenous growth models. These patterns are the net result of two opposing forces in the model. Like in the Solow (1956) model, it is assumed that new knowledge about production technologies is immediately public. These spillover effects imply tendencies towards convergence of productivity growth rates. Tendencies towards divergence are caused by the novel assumption that new knowledge is only ‘appropriate’ for countries that produce according to technologies similar to the innovator's technology. Such countries will reap the gains from innovation immediately, whereas other countries will not benefit at all. In this set-up, if innovation would take place at similar rates across technologies, the well-known Solow results would follow. Basu and Weil (1998, henceforth denoted as BW), however, assume that innovation is ‘localized’ at high-end technologies.1 Countries that operate low-end technologies could thus fall behind the world's technology leaders. The aim of this paper is twofold. First, and most prominently, it tries to give quantitative indications of the importance of localized innovation and BW's notion of appropriate technology spillovers for patterns of convergence and divergence. Second, the empirical validity of the assumption of immediate spillovers of appropriate technology will be investigated. Spillovers appear not to be immediate and arguments will be put forward that the speed of spillover assimilation should be regarded as a third determinant of labor productivity growth. The empirical analysis will be based on a recently proposed decomposition of productivity growth, which makes use of data envelopment analysis (DEA) techniques (Kumar and Russell, 2002). This methodology will be slightly modified, by adopting an intertemporal perspective suggested by Tulkens and Vanden Eeckaut (1995) that is in line with properties of the BW model. Labor productivity growth rates are decomposed into three parts, which will be interpreted as effects of the three sources of growth identified above: localized innovation, assimilation of knowledge spillovers, and creating potential for appropriate technology spillovers through investment. This methodology will be applied to Penn World Tables data on GDP, labor inputs and capital inputs for a set of 53 countries for the period 1965 to 1990. In a second stage, the decomposition results will be used in a convergence analysis based on regression techniques. Additional outlier analysis investigates how the three sources of growth contributed to the extraordinary performance of growth ‘miracles’ and ‘disasters’, i.e. countries that experienced very high or low productivity growth rates, even after correction for potentially favorable initial conditions. The rest of the paper is organized as follows. In Section 2, the BW model and the decomposition framework will be discussed in more detail and the relation between the theoretical and empirical approaches will be shown. Section 3 is devoted to a discussion of the data and the estimation of the reference production frontier, that is, the set of best-practice production processes. In Section 4, convergence and divergence of labor productivity growth rates and their three sources will be studied. Section 5 deals with an analysis of countries that emerge as outliers from the convergence regressions. Section 6 concludes.
نتیجه گیری انگلیسی
This study provides an empirical framework to study the labor productivity growth performance of countries from a perspective suggested by the model of economic growth by Basu and Weil (1998). A prominent feature of this model is the localized nature of innovation. Innovations for capital-intensive technologies do not affect the performance of capital-extensive technologies, and the other way round. Analysis of convergence processes suggests that localized innovation causes a tendency towards divergence. At low levels of capital intensity, hardly any innovation was found, whereas the global technology frontier was steadily pushed at high capital intensities. This does not bode well for developing countries that were mostly stuck within a range of technologies characterized by low capital intensities with little potential for further growth by means of spillovers. The results of a decomposition of labor productivity growth along the lines of Kumar and Russell (2002) indicate that for the original BW model to have empirical relevance, the assumption of immediate spillovers of appropriate technology needs to be relaxed. Assimilation of technologies new to a country is a costly and slow process. As a result, many countries perform at labor productivity levels that are far from the global technology frontier. We find evidence of catching up through assimilation, but the process is slow and assimilation rates are very heterogeneous across countries. In our view, the results of this study ask for a number of future research efforts. First, our analysis is based on aggregate economies. Some of them experienced rapid industrialization, whereas others had already entered the stage of tertiarization and still others generated their outputs predominantly in agricultural activity. Sector-specific analyses would add to the empirical operationalization of the appropriate technology concept. Second, it may be worthwhile to link both assimilation performances and creating spillover potential performances to social and technological capability indicators like schooling, infrastructure, openness to trade, etc. (see e.g. Verspagen, 1991 and Hall and Jones, 1999). In earlier convergence studies, such factors turned out to be critical, but it is unknown which factors relate most predominantly to knowledge assimilation, and which to creating potential. Finally, findings for a growth miracle like South Korea suggest a sequence in which countries first created opportunities for labor productivity growth by rapidly increasing capital intensities. Next, learning through the effective assimilation of new, appropriate technologies gained importance. A final step towards full development is further investment in modern equipment to be able to profit from current innovations at the global technology frontier. A natural question to ask is whether other countries should try to copy this behavior, in view of different initial endowments of social and technological capabilities. Given non-immediate assimilation, countries are basically faced with a choice between two extreme alternative policies: assimilating knowledge specific to the technology currently in use, or investing into more advanced technologies. New technology has a higher potential for productivity growth, but involves the costs of starting a new process of assimilation. Predictions about the optimal policy mix can only be based on a Basu and Weil type of model augmented with non-immediate spillovers. Construction and estimation of such a model is a challenging task ahead.