Rural infrastructure

The pool of available data does not permit any sophisticated hypothesis testing as to how various infrastructure variables affect nonfarm activity, but it allows us to check the plausibility of certain propositions. We regressed three indicators of agricultural output on a handful of independent variables proxying transport infrastructure, rural electrification, human capital availability, credit supply, and water management.

As a first step, individual variables were regressed on agricultural productivity. Such a specification has problems. There is bound to be omitted variable bias and bias in the covariances. Thus any inferences drawn must be treated with caution. The results highlight the importance of the credit and electrification variables: per capita bank advances in rural areas, density of bank branches, and consumption of electricity are all significant at the 5 percent level. Other proxies for electrification do less well. Primary education has little explanatory power, perhaps because the effects of education are subject to long lags. Secondary education has a stronger effect, but is only significant at the 10 percent level (tables 26 and 27).

Predictably, irrigation is closely correlated with agricultural output, as is the density of paved roads. But, total road density is not significantly different from zero, as expected-unpaved roads are virtually impassable during the prolonged rainy season, greatly undermining their economic utility for farmers.

Various combinations of the variables, whose individual significance was established in the first round of regressions, were regressed on land productivity. While the equations estimated are significant overall, that is, at least one of the explanatory variables is likely to be non-zero, standard errors tended to be large coupled with high R squares and insignificant coefficients, all of which point to multi-collinearity. For instance, the demand for credit and the willingness of banks to extend loans is greater in areas that have irrigation, electrification facilitates the use of pumps for tapping subsurface water supplies, and development of a road network stimulates the diffusion of bank branches. With this problem in mind, the independent variables were pared down to secondary education, irrigation, and paved roads. Paved roads are significant at the 5 percent level, but education has the wrong sign and is not significantly different from zero. When bank branches are substituted for education, collinearity renders both roads and credit insignificant.

Finally, the most promising variables were regressed on gross regional production at current market prices divided by the area of individual districts. Again bank branches and the road network prove to be significant. But so is education, suggesting that it has a greater influence on nonfarm activity than on agriculture. Electrification is insignificant, reinforcing some of the findings from other Asian countries.

The message from these simple tests draws attention to road infrastructure and irrigation in explaining the relatively weak effect that agriculture has had on the development of rural industry. Credit availability is also important, although we can interpret the results as saying that once a transport system is in place, the demand for credit will generate the desired supply by way of an expanding banking network.

These tests show that developing the transport infrastructure can provide a powerful stimulus to rural development. Roads are necessary for the development of markets, which induce farmers to intensify cultivation, diversify their product mix, and pursue nonfarm activities. Past research on Bangladesh has underscored the large benefits from all-weather roads. For instance, Ahmed and Hussain ( 1990) find that villages better served by transport infrastructure had a higher percentage of marketed output, especially paddy output. Use of fertilizer tended to be greater, and the price of fertilizer lower. And infrastructure development increased smaller farmers' access to institutional credit sevenfold. Finally, it influenced human capital by improving health indicators and stimulating acceptance of family planning practices. Other work by Binswanger ( 1992) and Binswanger and others (1993) on Indian districts shows that road density is significant in explaining agricultural output. For instance, the elasticity of agricultural output with respect to road density is 0.20, which is greater than the elasticity of output with respect to price, 0.13%. In addition, building roads facilitated the growth of branch banking-the elasticity of bank expansion with respect to road density was 0.8.

Similar findings have emerged from empirical analyses of rural development in Southeast Asia. In central Java the location of manufacturing activities was strongly influenced by telephone connections and nearness to a highway (Rietveld 1994). Moreover, small-scale manufacturing was able to grow in the rural areas of Taiwan (China) because of the extensive network of roads constructed by the Japanese in the first half of the century (Brautigam 1994).

Building roads and enabling water-borne transport encourage production for the 30 market and reduce the cost of moving goods. The increase in vehicular traffic can generate demand for repair facilities, locally manufactured spare parts, and metal working. Constructing and maintaining roads, culverts and dredging water courses are labor intensive activities that can generate much-needed employment. Further, construction requires materials, which can be produced by rural industries. Bangladesh either imports gravel or manufactures the necessary aggregate from highly vitrified bricks. The countryside is dotted with brick kilns. They employ mainly women workers, who first bake the bricks and then pulverize them into suitably sized aggregate (Novak 1994). Bangladesh is also poorly endowed with sand, and its collection is likewise laborintensive-divers scour the beds of the eastern rivers (Jensen and others 1989). The increase in vehicular traffic can generate demand for repair facilities, locally manufactured spare parts, and metal working.