What are the drivers of deindustrialisation and why does it matter?

Office worker in India

This week on the ESRC GPID blog, we continue a series of pieces on deindustrialisation and developing countries.

Deindustrialisation is emerging as the contemporary development trajectory for many middle-income developing countries.

In many cases though, the label ‘deindustrialisation’ is not quite the correct word, as the pattern is generally a plateauing of manufacturing shares or value-added shares rather than relative (or absolute) decline.

The label ‘deindustrialisation’ tends to be used because the alternatives are less well known to a wider audience. For example, ‘premature maturity’ from Kaldor, or ‘tertiarisation’, or Eastwood et al.‘s ‘premature de-agriculturalisation’.

Patterns of ‘deindustrialisation’ suggest a variety of types of deindustrialisation as we covered last week. In some cases, the value-added and employment shifts are towards services. In some cases, towards the ‘financialisation’ related FIRE (Finance, Insurance and Real Estate) sectors. In other cases, towards the non-FIRE service sectors. In yet others, a return to agriculture to some degree.

And in many, a mix of the above.

Each mode of deindustrialisation is likely to have its own causes and consequences, distributional dynamics and political economy.

What drives different types of deindustrialisation?

Rowthorn and Coutts list four causes of deindustrialisation in advanced countries. Specifically, (i) outsourcing and thus deindustrialisation is a statistical artefact caused by the contracting out of manufacturing jobs to services (for example, cleaning or catering); (ii) a fall in the relative prices of manufactures or a fall in the income elasticity of manufactures; (iii) international trade leading to higher competitive pressures to raise the labour intensity production to compete (or the substitution of labour with capital); (iv) decreases in the rate of investment which disproportionately affect manufacturing as most investment is in manufacturing.

Palma adds several more possible causes. First, deindustrialisation could be caused by Dutch disease due to natural-resources-led growth. Second, deindustrialisation could be caused by outsourcing. Third, changes in policy regimes in OECD countries away from Keynesianism and technological progress could be other drivers of deindustrialisation.

In contrast, Rodrik emphasises trade liberalisation over time and the impact of China’s entry into manufacturing as drivers, whilst Felipe et al. argue that premature deindustrialisation is caused by the fact that large national increases in labour productivity have been counteracted by a shift of manufacturing jobs to lower productivity economies. They argue that the average employment share in manufacturing that could be achieved has fallen over time, and countries have experienced deindustrialisation earlier than they used to. In short, the changes in supply chains and shift to lower productivity economies has spread manufacturing jobs more thinly, making it harder for individual countries to sustain high levels of manufacturing employment. They note that global employment in manufacturing and gross domestic product (GDP) shares have changed very little in the last forty years. What has happened is that international competition has spread manufacturing across more countries.

Finally, Treganna, as we discussed last week, conducted a decomposition of deindustrialisation in manufacturing employment in 48 countries. She finds that in most cases, the decline in manufacturing employment is due to a fall in labour intensity in manufacturing (i.e. a rise in productivity).

Why does deindustrialisation matter?

Many of the theories above are predicated to a greater or lesser extent on manufacturing as a ‘special’ sector. There is a long history in heterodox economics that manufacturing is special (e.g. Kaldor’s growth laws, Chenery and Syrquin, Hirschman, Kalecki, Prebisch, and Thirwall).

There are a set of growth-enhancing externalities of manufacturing not reflected in relative prices and a market-based equilibrium sectoral structure may not be optimal for growth. The intersectoral reallocation of labour and other factors may increase aggregate productivity.

Specifically, drawing from Kaldor and Hirschman, and others, backward and forward linkages (direct and indirect) are stronger in manufacturing than in other sectors; dynamic economies of scale are stronger in manufacturing than in other sectors; rises in manufacturing productivity are likely to spill over into other sectors; most technological change occurs in the manufacturing sector (a cumulative causation argument); and higher import income elasticities and relative tradability exist in manufacturing, therefore manufacturing can alleviate the balance of payments constraint on growth in the absence of stable and favourable terms of trade.

In a similar vein, Rodrik argues that most services are (i) non-tradable, and (ii) not technologically dynamic, and that (iii) some sectors are tradable and dynamic, but they do not have the capacity to absorb labour. That said, similar shortcomings can be observed about the manufacturing sector, in the sense that a significant share of manufacturing is (i) non-traded even though it is tradable, and (ii) much of manufacturing in developing countries is not technologically advanced, at least in relative terms to other modern sectors, and (iii) where some manufacturing sectors are technologically dynamic, they may not create much employment, as some service sectors do. Furthermore, returns to scale imply that as costs fall, demand rises for manufacturing foods (high-income elasticities of demand), triggering more manufacturing and higher incomes, more demand, and cost reductions in a cumulative causation process.

In sum, there are not only various forms of deindustrialisation but also a range of theories on its causes. A question follows as to who are likely to be the winners and losers (in an absolute or relative sense) from deindustrialisation. In Kuznets’ seminal work, he hypothesised that inequality would rise during industrialisation as labour moves from the more equal rural sector to the less equal urban sector. What he did not discuss – as he didn’t envisage it at that time – is what if the labour flow is from manufacturing towards services or back to agriculture?

Next week, we’ll take a look at the distribution dynamics of deindustrialisation in developing countries.

Andy Sumner is a Reader in International Development in the Department of International Development, King’s College London. He is Director of the ESRC Global Poverty & Inequality Dynamics (GPID) Research Network.