世界发展银行-Manufacturing-in-Structural-Change-in-Africa_36页_2mb

Summary of "Manufacturing in Structural Change in Africa"

Core Content

This working paper investigates the patterns, causes, and timing of industrialization and deindustrialization in Sub-Saharan Africa (SSA) using panel data methods and data from multiple sources. The study focuses on the share of manufacturing in GDP and employment, and explores whether these trends indicate premature deindustrialization as suggested by some earlier research.

Main Findings

• Premature Deindustrialization: Recent studies suggest that manufacturing output and employment shares in SSA tend to decline at lower income levels than in advanced economies. However, this paper finds no overwhelming evidence to support the claim that SSA countries have begun to deindustrialize prematurely.

• Heterogeneity Across Subregions: The study documents significant geographic variation in manufacturing trends across SSA subregions. The Southern subregion is the only one observed to have experienced deindustrialization, but this does not appear to be premature.

• Data and Methodology: The analysis is based on data from 41 SSA countries spanning 1960–2016, making it the longest panel data set on this topic. The data includes:

  • Manufacturing value added and employment from the GGDC 10-sector Database and WDI.
  • Income and population data from the Maddison Project Database.
  • Oil revenue and exchange rates from the IMF's AFRREO database.
  • Minimum wages from the ILOSTAT database for selected oil-exporting countries.

  The paper uses both linear fixed-effects models and fixed-effects fractional logit models to account for the bounded nature of manufacturing shares (fractional response variables). The latter is more suitable for capturing structural changes in industrialization.

• Industrialization Trends: The study finds that manufacturing shares in GDP (RMVA) and employment (EMP) generally follow an inverted U-shaped trajectory across SSA, with some exceptions. For example:

  • West Africa: RMVA increases from 0.08 to 0.12 in the 1970s, then declines to 0.07 by the 2010s.
  • East Africa: RMVA remains relatively stable over time.
  • Central Africa: RMVA peaks at 0.25 in the 1980s and then declines.
  • Southern Africa: RMVA increases to 0.15 in the 2000s and then slightly declines.

• Subregional Analysis: The Southern subregion is the only one showing a decline in manufacturing shares, but the turning point of this decline does not occur at a lower income level than that of advanced economies, indicating it is not premature.

• Role of Natural Resources: The paper explores the Dutch disease and resource curse hypotheses, which suggest that resource booms can negatively impact manufacturing. By analyzing the share of manufacturing value added in non-oil GDP, the study attempts to isolate the effect of oil revenues on industrialization.

• Methodological Considerations: The paper acknowledges the challenges of analyzing fractional response variables and highlights the importance of using nonlinear models to avoid the pitfalls of linear probability models. It also notes the limitations of existing studies, such as the small sample size and inclusion/exclusion of specific countries (e.g., Mauritius), which may affect the interpretation of results.

Key Information

• The study uses 41 SSA countries and a panel data approach.
• It employs two econometric models: linear fixed-effects and fixed-effects fractional logit.
• The inverted U-shaped pattern is found for manufacturing shares in GDP, but not for employment.
• Southern Africa is the only subregion with deindustrialization, but the timing does not suggest premature decline.
• Mauritius significantly affects the results of premature deindustrialization in SSA, and its exclusion changes the interpretation of the data.
• The Dutch disease and resource curse are considered as potential explanations for the observed trends in oil-rich countries.
• The paper is part of the World Bank's effort to provide open access to research and contribute to development policy discussions.

Conclusion

The study concludes that the evidence for premature deindustrialization in SSA is not robust, and that geographic heterogeneity plays a key role in shaping industrialization trajectories. While some subregions may experience deindustrialization, the timing does not support the notion of premature decline. The findings highlight the need for country-specific and subregional analyses in understanding the complex dynamics of manufacturing in SSA.