In two previous posts on this blog, I’ve discussed the issue of premature deindustrialization and some of its possible consequences. In a recent paper, I, along with co-author Bret Anderson of Southern Oregon University, explored the potential consequences of premature deindustrialization further by examining the possible connections between premature deindustrialization and the defeminization of industrial employment. Premature deindustrialization is a situation in which the shares of manufacturing value added and employment begin to shrink at per-capita income levels much lower than those of the early industrializers, along with manufacturing employment peaking at lower levels. The scarce manufacturing jobs that do remain, however, are likely to be relatively high paying jobs that countries and workers compete for. In our work, we assessed whether premature deindustrialization is a feminizing or defeminizing force in industrial employment. By examining 62 countries from 1990 to 2013, we find that premature deindustrialization is likely to amplify the male bias of industrial upgrading.

The Evolving Nature of Deindustrialization and Defeminization of Industrial Employment
In a recent New York Times article, Joe Stiglitz stated that “the observation is uncontroversial. Global employment in manufacturing is going down because productivity increases are exceeding increases in demand for manufactured products by a significant amount.” The recent academic literature provides evidence for this trend globally; lower peaks in manufacturing employment at lower levels of GDP per capita. This trend is occurring alongside another interesting trend in industrial employment; changing patterns of women’s relative employment in industry.

The previously observed trend of a global feminization of labor, typically associated with manufacturing for export, may have begun to reverse course toward “defeminization” as industrial production become more capital intensive. This trend implies that the male bias of industrial upgrading outweighs the perceived female bias of export-orientated production. At the individual level, we can speculate as to why this negative relationship might occur; for example it may be due to reduced labor cost pressures in capital-intensive production, gender biases, or lack of job training for women. At the macroeconomic level, dramatic regional differences between Asian and Latin American feminization and deindustrialization patterns suggest that the economy-wide factors that condition the link between industrial upgrading and defeminization of labor may be relevant. The question that emerges is how the timing of deindustrialization is related to the male bias of industrial upgrading.

Competitiveness and the Deindustrial Regime Type
One issue in trying to answer this question is how to measure premature deindustrialization. We argue that the competitive position of manufacturing is an important link between premature deindustrialization and defeminization. Even in cases with continued success in exports, industrial upgrading has been defeminizing. See figure 1, which illustrates the relationship between capital deepening and defeminization of employment for the period from 1990 to 2013, below.

To study the potential male bias in industrial upgrading, we use an economy’s competitive position as a proxy for the deindustrialization regime type. Specifically, we suggest that prematurely deindustrializing economies are marked as being less competitive. To get to this position, we bring together Kaldor’s work, feminist scholarship, and structuralist critiques. In the Kaldorian framework, manufacturing is the engine of growth. This process is related to an economy’s competitive position as the growth of manufacturing output is constrained by demand. In the later phases of development, this demand must come from exports. In his work on premature deindustrialization, Rodrik finds that the decline in manufacturing employment that stems from a drop in relative prices is smaller in countries with more competitive manufacturing sectors. This result strongly suggests that an economy’s competitiveness is related to the deindustrialization regime.

Global competitiveness also appears within the structuralist critique literature in a related way. In particular, modern global production networks (GPN) are the contemporary analogue of past peripheral commodity exports popularized by Prebisch. Monopsony bargaining power from the top of GPNs, fallacies of composition related to many developing countries pursuing export-led growth simultaneously, and excess supplies in global markets, can all lead to falling terms of trade for developing countries’ manufacture exports. In short, the same processes that affected primary commodity exporters in the past century appear to be a barrier for exporters of manufactured goods that find themselves on the lower rungs of GPNs.

A large body of feminist economic literature further argues that women in the formal sector tend to lose their job faster than men, and usually have worse access to social safety nets that can ultimately impact their relative bargaining power. Further, women’s employment may be more sensitive to output fluctuations, particularly for the industrial sector.

We take a few main points from this literature. For late industrializers, relative price declines in manufacturing have outpaced relative productivity gains. For premature deindustrializers, however, the decline in manufacturing employment that stems from a drop in relative prices is smaller for more competitive manufacturing countries. Finally, since women’s employment is more responsive to output fluctuations, a drop in the relative prices of manufacturing would result in a drop in women’s share of manufacturing employment relative to men’s. Our overall take-away is that premature deindustrializers are likely to be in a less competitive position and, when relative prices of manufacturing fall, this fall will result in a drop in the ratio of women to men in manufacturing employment. This process leads to a lower bargaining power for women and amplifies the male bias of industrial upgrading.

Simulating the Male Bias of Industrial Upgrading
Figure 1, above, shows the negative correlation between industrial upgrading and the relative proportion of women in industrial employment. Our aim in this section is to find if there is a difference in that relationship depending on whether a country is in the premature deindustrialization category. The sample includes sixty-two countries and covers the years from 1990 to 2013. We use a country’s competitive position as a proxy for premature deindustrialization. This allows for a continuous, rather than discrete measure. We construct a “Balassa Index” of relative competitive advantage, for all countries and years in our sample, which is calculated by dividing the ratio between manufacturing and total exports for a given country and year by the ratio between world manufacturing and total exports.

The figure below shows the kernel density probability distribution of competitiveness, disagreggated into two groups by timing of peak of manufacturing employment, as measured by a seven-year moving average. The figure shows that countries that peaked in manufacturing employment before 1990 tend to be more competitive, and countries that peaked more recently tend to be uncompetitive. This result lends credence to the theory that premature deindustrialization — a phenemona prevalent in late peakers — is associated with competititivness.

Next, we ran a country fixed effects regression model to estimate the impact that industrial upgrading has on women’s relative employment share in industry. Our dependent variable was the female-to-male ratio in industrial employment, and the independent variables were K/L ratio, the Balassa Index, an interactive term between the two, and several control variables. We found a positive and significant relationship between the Balassa index of competitiveness and the ratio of female-to-male workers in industry. More competitive countries have higher relative rates of female industrial employment. The capital labor ratio also had the expected negative effect, suggesting that industrial upgrading is defeminizing. Perhaps most interesting for the purpose of our question is the interactive term, which has a positive and statistically significant effect. This result suggests that competitiveness mitigates the negative effect on female employment of capital deepening. Less competitive countries, the premature deindustrializers, have a more intense male bias of industrial upgrading.

Finally, the figure below shows a Monte-Carlo simulation built from the previously described regression. It illustrates the effect of the K/L ratio on the ratio of female-to-male employment in industry by deindustrial regime type. It provides another way of seeing that less competitive countries, the premature deindustrializers, have a more intense male bias of industrial upgrading.

Conclusion
The late industrializers are deindustrializing at earlier stages of development than their predecessors and the global trends in the gender composition of industrial employment are evolving. Starting from the premise that industrial upgrading has a male bias, we tested the hypothesis that premature deindustrialization is likely to amplify that bias.

For the empirical test and simulation, we used an economy’s competitive position as a proxy for the deindustrialization regime type. The results for sixty-two countries spanning the years from 1990 to 2013 support our hypothesis that premature deindustrialization is likely to amplify the male bias of industrial upgrading. We hope to add to the evidence as we continue work on this question going forward.

Joshua Greenstein is Assistant Professor of Economics at Hobart and William Smith Colleges.

Photo: NaZemi