Economic Development and Premature De-industrialization

IDE Research Bulletin

SATO Hajime

March 2019

PDF (132KB)

Outline of Research Products

First, we conducted literature survey on premature de-industrialization. While the term ‘de-industrialization’ has been widely used for the experiences of advanced countries since the 1970s (Tregenna 2011), it appeared that Dasgupta and Singh (2007) seemed to be the first case which used the term for developing countries and proposed the concept of ‘premature deindustrialization’. The study was prompted by the rapid growth of India, which was characterized as service-led growth, and tested the ideas of Kaldor (1967) who considered the manufacturing sector as the engine of growth. Their main findings were that the manufacturing sector had been the engine of growth, that the service industry can be the complementary engine of growth as well, and that there were two types of premature deindustrialization, one with decreasing share of the manufacturing employment and one with decreasing shares of the manufacturing employment and output.

The concept of premature deindustrialization was later independently proposed by Rodrik (2013, 2016), who had considered the manufacturing sector as the escalator of economic growth. He refined the concept, especially what ‘premature’ means. First, the shares of manufacturing employment and output decreases at the early stage of economic development and second, the peaks of the shares are lower compared to those of advanced countries. Also, he addressed the mechanisms of why premature deindustrialization set in for developing countries especially form the 1990s, and suggested the effects of globalization. Relying on his study, several studies have surveyed the existence of premature industrialization (e.g. Amirapu and Subramanian 2015, Felipe et al. 2014), and their results basically confirmed the occurrence of premature deindustrialization, although in some points they are mixed, depending on the data and time span of each study.

Based on the survey, second, we reinvestigated the relationship between the manufacturing output/employment and the levels of income. Drawing on the Rodrik’s analysis, we used expanded samples and different classifications. Firstly, for the relationship between manufacturing shares and income, we estimated the peak levels of OECD countries as the group of developed countries and non-OECD countries as the group of developing countries. The peak level of manufacturing shares both in terms of industrial outputs and employment are higher in the former than the latter. The projected incomes at the peak level of manufacturing shares is quite higher for OECD countries than non-OECD countries. This might suggest that the premature deindustrialization takes place.

Also, we examined the other question of whether deindustrialization has been accelerating in recent years by conducting analysis for each time period. The result shows that the peak level of manufacturing shares and corresponding incomes has been dropped consistently until the 1990s for any indicators (industrial outputs and employment). However, this trend become ambiguous after 2000.

Accordingly, the analysis of the paper adds some evidence regarding the occurrence of premature deindustrialization in recent years. At the same time, our findings imply that the term ‘premature’ deindustrialization might require careful reconsideration, as the phenomena of deindustrialization in latecomers should further be carefully considered and analyzed in terms of their mechanisms and effects.