Are there good and bad varieties of deindustrialisation?

Central Bangkok, Thailand

Deindustrialisation has become the experience of many middle income developing countries. This week on the ESRC GPID blog we start a series of blogs on that topic.

We’ve covered previously the question of what is deindustrialisation and whether it matters.

To recap, deindustrialisation is typically defined as a shrinking proportion of industrial or manufacturing activity in employment and/or value-added. One could extend this to the proportion of manufactures in exports.

To date, however, much discussion of deindustrialisation has taken place with reference to the developed world. In terms of developing countries, both Gabriel Palma and Dani Rodrik have discussed such processes as ‘premature deindustrialisation’, whereby manufacturing and/or industry starts contracting at earlier levels of development (see our respective GPID briefing paper).

Deindustrialisation can take several modalities; namely manufacturing and/or industry may shrink in relative or absolute terms; GDP and/or employment shares and/or export shares accounted for by manufacturing and/or industry may also shrink. At the same time, it may be agriculture that expands, or it may be services that expand or both, and then each of these may be towards higher or lower productivity activities relative to the shrinking sector.

Treganna (2009) discusses several types of deindustrialisation based on an analysis of 48 countries and whether value added in manufacturing is growing or not in absolute and relative terms, noting that in almost all cases, employment in manufacturing does not rise. She finds that developing countries tend to fall into the category of shrinking manufacturing value added in absolute and relative terms. She conducts a decomposition of deindustrialisation in manufacturing employment in those 48 countries, decomposing the change in the level and shares of employment in manufacturing between the following components: a change in share of manufacturing GDP; labour intensity in manufacturing (i.e. productivity); economic growth and the growth of manufacturing value added. She finds that in most cases the decline in manufacturing employment is due to a fall in labour intensity in manufacturing. Her conclusion is that deindustrialisation which is associated with rising labour intensity is very different from deindustrialisation which is associated with poor performance in value added.

This resonates with the seminal work on advanced countries of Rowthorn and Wells who discussed some time ago, with reference to advanced countries, ‘positive’ and ‘negative’ deindustrialisation. In the former, unemployment does not rise as labour transferring out of manufacturing is absorbed into the service sector with new jobs. In the latter, labour is not reabsorbed, and unemployment rises. They also note a third type, when the structure of net exports shifts away from manufacturing, and thus labour and resources do so too.

Deindustrialisation – if defined as shrinking employment shares – could be experienced with an increase in the absolute number of jobs in manufacturing, as long as total employment growth is fast enough. And again, if defined as shrinking employment shares, deindustrialisation could lead to an increase in total manufacturing output if manufacturing productivity rises sufficiently quickly.

In short, both good and bad deindustrialisation are possible. Next week, we continue the discussion with looking at the drivers of good and bad deindustrialisation in developing countries, and questions about the distributional consequences of good and bad deindustrialisation.

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.