Premature Deindustrialization through The Lens of Occupations: Which Jobs, Why, and Where?

Abstract

A recent literature documents that manufacturing employment growth in developing countries has been sluggish over the past decades, and that deindustrialization has often set in at historically low levels of income. However, there is little evidence on which kind of jobs are disappearing prematurely, and some debate on whether the phenomenon is structural or transitory. In this article, I use a new data set on manufacturing employment by occupation to document four stylized facts about`premature deindustrialization': rst, it is mostly unskilled jobs that have disappeared, and also the wage premium of workers with little formal education in manufacturing relative to other industries has declined. Second, the disappearing jobs have been among the most formalboth relative to other industries, and to the manufacturing average. Third, premature deindustrialization has been driven by occupations which are intensive in tasks that are suitable to automation by information and communications technology (ICT).Fourth, the phenomenon pertains most clearly to high and middle income countries, as low income countries have been spared from premature job losses. Overall, the employment patterns are consistent with a pervasive shift of the`automation frontier' separating tasks that are automated from those which are not, and suggest a structural decrease in the ability of manufacturing to employ unskilled labor productively. JEL: O14, O33, J23

Deskilling among Manufacturing Production Workers

“…given their scarcity at most income levels (Figure 2) and their higher educational requirements (Table 1). 51 Further, the results from the companion paper Kunst (2019) suggest that transitions to industries other than manufacturing have become more frequent in recent decades, as fewer "compensating" other production jobs have been created to make up for the loss of craftsman jobs.…”

“…For US manufacturing, Acemoglu and Restrepo (2019) argue that what distinguishes the period after 1987 from the previous four decades (during which erating pace of technology adoption (cf. World Bank (2008) and Kunst (2019)) accompanied by a strong reduction of labor shares, and Bessen (2015) argues that historically, technological changes which reduced the returns to experience had a tendency to also reduce labor shares for extended periods. 57 It is interesting to compare the recent experience of developing countries with automation to the US experience between the 1950s and 1970s: in the US, labor market prospects remained favorable during this period in spite of rapid automation, as rising incomes increased labor demand across a broad range of non-tradable sectors (Autor, 2015).…”

“…Since such technologies were first adopted in high income countries, the reduction in the employment share of other production workers is already apparent for these countries. In Kunst (2019), I show that once conditioning on GDP per capita, employment in such occupations has decreased also in middle income countries when comparing the 1960-1990 to the post-1990 period. 32 An alternative test to rule out a supply-side explanation of the declining craftsman wages is to control for the average educational attainment by country-year in the relative wage-regressions and its interaction with a craftsmandummy (which requires dropping the then collinear country-period fixed effects).…”

“…12 Craftsmen correspond to major group 7 of ISCO, other production occupations include major groups 8 ("Plant and machine operators and assemblers") and 9 ("Elementary occupations"), and white collar occupations subsume major groups 1 ("Legislators, senior officials and managers"), 2 ("Professionals"), 3 ("Technicians and associate professionals") and 4 ("Clerks"). In Kunst (2019), I argue that it is insightful to further distinguish between the high skilled white collar occupations in major groups 1-3 and the medium skilled clerks in major group 4 when analyzing the recent impact of ICT on the labor demand in manufacturing. However, this distinction becomes most relevant towards the end of my sample period, and is not the focus of this paper.…”

“…This is the case especially in the earlier part of my sample period and in developing countries with technologically less advanced manufacturing industries, prior to the adoption of more advanced and digitally controlled machines (cf. Katz 1998, andKunst 2019).…”