"Commercial agribusiness and value-chain integration, not smallholder support, drive agricultural transformation"

"ostow, 1960; Chenery, 1986; Syrquin, 1986). By calculating the changes of these variables over time, also an industrial development model is obtained and those are the two empirical models to be estimated.16 The role of agriculture in furthering industry is interesting and statistical links between the two sectors seem to be the norm rather than the exception. On the one hand, improved agricultural productivity can be viewed as releasing resources, especially labour input, to manufacturing. Jorgenson (1961) and Sachs (2008) state that without technological progress in the agricultural sector, a modern sector might not even prove viable. The argument is that only when agricultural productivity is high—implying that a farm family can feed many urban citizens so that not each resident has to feed itself—can a significant share of the population become urbanized and engage in manufacturing production. Agriculture could then be seen as pushing industrial development. However, if the migration leads to shortage in food production (forward linkages) or the two sectors’ marginal productivities converge agricultural growth can constrain manufacturing growth (Fei and Ranis, 1961). A sectoral link can also develop because manufacturing productivity exceeds that of agriculture and, therefore, pulls labour out of the latter sector. This view holds that the marginal productivity of labour in the leading modern sector (i.e., manufacturing) much higher than in the laggard one (i.e., agriculture). In fact, because of unlimited supply of labour in agriculture, the marginal productivity there is extremely low, if not negligible. Labour, therefore, has a wage incentive to migrate from agriculture to manufacturing, allowing the modern sector to further grow and develop the economy (Lewis, 1954). Whichever effect—push or pull—that dominates, the link between the sectors has to be accounted for. There are additional reasons linking the two sectors. The agricultural sector’s exports provide foreign exchange, which can be used to import material and capital goods to industry. Furthermore, with a functioning banking sector, successful agricultural savings can be channelled to and invested by industry. Redistribution of agricultural surplus can be taxed and provided as 16 Note that geography will not be explicitly accounted for, since it will be captured in the panel-data analysis by the country-specific effects. 16 support to manufacturing. Industrialization also raises demand for agricultural goods (Johnston and Mellor, 1961). Agriculture is also a client of manufacturing. For example, fertilizers are important inputs in agricultural production so backward linkages are thus important. A slow-growing agricultural sector can, therefore, act as a drag on manufacturing. The expected estimated coefficient, hence, is not unequivocally positive.17 That agricultural performance and industrial development are linked should be beyond doubt, but it is neither the purpose of this paper to sort out the causal direction of the link, nor whether that link is positive or negative. Breisinger and Diao (2008) give an example where public investment in irrigation and infrastructure supported the introduction of modern technology in agriculture, which changed farmers’ savings and investment behaviour: the Green Revolution. In the same vein, roads and distribution systems lower the costs of using technical inputs by geographically dispersed firms and households (Restuccia, Yang and Zhu, 2008). The public sector also supports technological development in different ways. It is, in particular, large projects characterized by indivisibilities, or lumpiness, which need support from the public sector. Murphy, Schleifer and Vishny (1989) even show that private investments might not occur unless the state can credibly demonstrate that is will undertake its investments. There are several reasons to expect human capital to enter with"