The agricultural roots of industrial development

Samuel Marden

28 December 2014



For most countries, development and industrialisation are inextricably linked. A classic literature in development argues that, for the poorest countries, improving the productivity of the agricultural sector may provide the initial spur to industrialisation (e.g. Rosenstein-Rodan 1943, Schultz 1953, Lewis 1954). This view has been influential with policymakers – according to the World Bank (2007), “there are many success stories of agriculture as an engine of growth early in the development process”.

However, while the correlation between agricultural productivity and the size of the non-agricultural sector is undeniable, the empirical evidence for increases in agricultural productivity causing industrialisation is thin. It is easy to imagine an alternative scenario where higher agricultural productivity increases the profitability of farming and crowds out the non-agricultural sector. To the extent that agricultural productivity does result in higher non-agricultural output, we know little about how it does so. Does higher farm productivity reduce the rural demand for labour, or do farmers provide an important source of demand and capital to industrialists?

China’s agricultural reforms as a natural experiment

My research addresses these questions by comparing counties in China which, by virtue of their relative suitability for cash crops and grains, benefitted differentially from China’s 1978–1984 agricultural reforms (Marden 2014). Counties suited to cash crops benefitted more from reforms, and enjoyed faster growth in agricultural output. Part of this increase in agricultural output was saved, increasing the supply of capital to local non-state firms and stimulating the growth of the non-agricultural sector. There were positive linkages from agriculture to industry in early reform-era China.

Counties suited to cash crops benefitted more from China’s agricultural reforms, because prior to the reforms the Chinese government imposed a policy of rural self-sufficiency in food. At this time, Chinese agricultural productivity was low, and most farmers faced binding subsistence constraints. Prevented from importing food from elsewhere, even farmers with land highly suitable for growing cotton or oilseeds (the main cash crops in China) had to specialise in grain production to avoid starvation. There were also political incentives to produce grain, and most farmers faced quotas for grain deliveries to the central government. By relaxing these policies, the reforms effectively liberalised the planting of cash crops.

Figure 1. Suitability for cash crops relative to grain in China

Notes: Map indicates where county level data is available, and shows suitability for cash crops defined as the ratio of the value of output of a cell’s best cash crop (of cotton, rapeseed, and peanuts) divided by the best grain (of rice, maize, wheat, and soy). Underlying measure is continuous but has been grouped into Jenks categories for readability.

The impact of the reforms on agricultural output

The first part of my analysis demonstrates that counties suited to cash crops did benefit from the freedom to plant these crops. I use pre-reform prices and a global database of theoretical crop yields to identify counties that were endowed with land more suited for cash crops than grains. Then I trace the growth of agricultural output for 561 non-metropolitan counties over forty years. Suitability for cash crops does not predict growth pre-reform. However, after the reforms, counties began to specialise in cash crops if they were suited to them, and enjoyed significantly faster growth in aggregate agricultural output. Back-of-the-envelope estimates suggest that specialisation in cash crops increased Chinese agricultural output by around 10% between 1978 and 1985 (about one-fifth of the total increase, and two-thirds of the increase Lin (1992) attributes to the contemporaneous decommunalisation of agriculture).

The link between agricultural and non-agricultural output

I then turn to whether this increase in agricultural productivity led to higher non-agricultural output. The setting of my study, China, helps me explore the link between the sectors in two ways. First, in the early 1980s, China was poor and the agricultural sector was relatively large. A large agricultural sector and underdeveloped industrial sector will tend to make linkages from agriculture to the rest of the economy more important. Second, China’s reform-era institutions resulted in well-documented geographic frictions in capital and labour markets. These frictions prevented the equalisation of wages and capital costs over space and kept local shocks local – farmers’ savings were often lent to nearby firms, and labour no longer required for farming sought work close to home.

As a result, counties suited to cash crops also enjoyed a relative increase in manufacturing and service-sector output after the reforms (they were growing at the same rate before). A 1% increase in agricultural output due to specialisation was associated with a 1.2% increase in non-agricultural output between 1978 and 1995. Counties suited to cash crops also had faster post-reform growth in savings and investment.

Discriminating between possible explanations

These results indicate that there were forward linkages, but they do not indicate why there were linkages. I consider the possibility of linkages through three classic channels. First, growing cash crops could be less labour-intensive than grains, which would free up labour for manufacturing and services. Second, richer farmers could purchase more manufactures and services, driving up demand for these goods. Third, richer farmers could save more, providing capital for local firms.

My main results – that agricultural productivity led to higher savings, investment, and non-agricultural output – are consistent with any of these channels. Other outcomes are more revealing:

  1. If growing cash crops was labour-saving, then the number of workers in agriculture should fall in places suited to cash crops. I show that the share of labour in agriculture increases. This is consistent with previous work suggesting that cash crops are more labour-intensive than grains.
  2. If the linkages were due to richer farmers demanding more non-agricultural goods, then linkages should be stronger in places where trade costs are higher and importing these goods is expensive. I show that the opposite is true – linkages are weaker in more isolated counties.
  3. If increases in the supply of capital were driving the results then the price of capital should fall. Although I don’t observe the price of capital faced by firms, I can observe firm behaviour. After the reforms, non-state firms located in counties suited to cash crops behave as if they face cheaper capital costs.

Concluding remarks

Put together, these additional results indicate that the observed linkages from agriculture to the rest of the economy were primarily due to rural saving increasing the supply of capital to local firms. This suggests that programmes which increase the productivity of agriculture may also benefit the wider economy, and that an ‘agriculture first’ development strategy could be more effective than previously thought.


Lewis, W A (1954), “Economic development with unlimited supplies of labour”, The Manchester School 22(2): 139–191.

Lin, J Y (1992), “Rural reforms and agricultural growth in China”, American Economic Review 82(1): 34–51.

Marden, S (2014), “The agricultural roots of industrial development: ‘forward linkages’ in reform-era China”, mimeo.

Rosenstein-Rodan, P N (1943), “Problems of industrialisation of Eastern and Southeastern Europe” Economic Journal 53(210/211): 202–211.

Schultz, T W (1953), The economic organization of agriculture, New York: McGraw-Hill.

World Bank (2007), World development report 2008: Agriculture for development.



Topics:  Development Economic history

Tags:  Agriculture, industrialisation, development, productivity, China, manufacturing, services, saving, capital accumulation, natural experiment

PhD Candidate in Economics, LSE; Researcher, STICERD