农村：走向市场和建设道路

W_PS_ I 2 POLICY RESEARCH WORKING PAPER 2028 Access to Markets and the Improving road access to agricultural markets in Nepal Benefits of Rural Roads would confer substantial economic benefits on Hanan G. Jacob), average, much of them going to poor households. But rural road construction is more like a tide that lifts all boats than a highly effective means of reducing income inequality. The World Bank Development Research Group Rural Development December 1998  l POLICY RESEARCH WORKING PAPER 2028 Summary findings Transport infrastructure plays a central role in rural benefits from hypothetical road projects are calculated development, yet little is known about the size - or, from the predicted appreciation in value of the especially, the distribution - of benefits from road household's farmland. These predicted benefits are then investments. Among other benefits, rural roads provide related to household per-capita expenditures to assess cheaper access to both markets for agricultural output their distributional consequences. and for modern inputs. The empirical analysis, using data from Nepal, shows Jacoby develops and implements a method for large benefits from extending roads into remote rural nonparametrically estimating the benefits from road areas, much of these gains going to poorer households. projects at the household level. The idea is that since But rural road construction is not the magic bullet for these benefits get capitalized in land values, they can be poverty alleviation. The benefits are neither large enough estimated by examining how the value of farmland falls nor targeted well enough to reduce income inequality with distance from agricultural markets. Household-level appreciably. This paper-a product of Rural Development, Development Research Group-is part of a larger effort in the group to study the impact of rural roads and other forms of infrastructure on household welfare and economic growth. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Maria Fernandez, roomMC3-542, telephone 202-473-3766, fax 202-522-1151, Internetaddress mfernandez2@worldbank.org. The author may be contacted at hjacoby@worldbank.org. December 1998. (32 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center  Access to Markets and the Benefits of Rural Roads Hanan G. Jacoby* Keywords: Rural Roads, Income Distribution, Nonparametric Regression JEL Classification: 012, D31, C14 *Development Research Group, The World Bank, 1818 H Street N.W., Washington DC 20433.  I. Introduction Rural infrastructure is a major development priority (World Bank, 1994), yet little is known about the size and especially the distribution of benefits from such investments in LDCs. The distribution issue is salient, not only in the formulation of policy, but also in understanding the political constraints on the allocation of infrastructure investment. Rural roads are an. important form of public infrastructure, providing cheap access to both markets for agricultural output and for modem inputs. Given limited policy instruments for reaching the remote rural poor, road-building would seem desirable on distributional grounds. On the other hand, the benefits of infrastructure projects accrue mainly to landowners, who are generally not among the very poor. Thus, the extent to which rural road construction ameliorates income inequality is ultimately an empirical question.' In this paper, I examine the distributional consequences of rural roads using data from Nepal, a country with a largely agrarian economy, a sparse highway network, and extremely difficult terrain. To this end, I develop an empirical methodology for nonparametrically estimating the household-specific benefits from alternative road projects using information on the value of farmland and distance to agricultural markets. If land behaves like a standard asset, which is a testable assumption, then its value equals the discounted stream of maximal profits from cultivation. Hence, the income gains from lower transport costs should be capitalized in land values. With an estimate of the land value-distance relationship in hand, it is possible to describe the joint distribution of hypothetical road project benefits anrd household income. 'Howe and Richards (1984) discuss some distributional aspects of rural roads and present case studies. Also, van de Walle (1996) uses mnicro-data and a profit function approach to examine the distribution of benefits from irrigation in Viet Nam. 1  In principle, road benefits could also be estimated from the relationship between farm profits and distance to markets. However, there are several difficulties with this approach, the most nettlesome of which is that survey data rarely, if ever, provide accurate information on an essential component of profit, the cost of transporting goods and agricultural inputs to and from markets. Another difficulty is that profit (or production) functions assume a fixed technology and thus cannot easily account for potential adaptations of farmers to greater remoteness from markets, such as substitution of traditional for modern inputs or away from transport-intensive crops. The relationship between land value and distance to market is immune from such difficulties. To be sure, the idea of using land values to estimate the average benefits of infrastructure investments in a population is hardly new, though it has not to my knowledge been applied to rural transport. In any case, such estimates do not address the primary question of this paper, which is a distributional one. The innovation here is to link a household-level benefit estimate with a measure of household income. In doing so, I take a nonparametric approach. While it is true that economic theory is largely silent on the parametric form of hedonic price functions (see Stock, 1991), in practice, relaxing parametric assumptions in hedonic models is much more likely to matter for distributional questions than for questions about average benefits. Theoretically, my analysis is based on the Ellet-Walters model of rural transport (see Walters, 1968; Gersovitz, 1989), in which land rents decline with distance to markets through the influence of distance on effective prices. The model, laid out in the next section, provides a simple characterization of the potentially conflicting distributional consequences of road projects. 2  Section IiI discusses the nonparametric or, more precisely, semi-nonparametric estimation of the land value equation. Section IV describes the data, analyzes how farrner behavior is influenced by distance to market, and tests the appropriateness of a standard asset-pricing model for land. Section V presents the main empirical results, the analysis of land values and of the distribution of benefits from hypothetical road projects. Section VI concludes the paper. II . Theoretical Framework Basic Model Farmers are assumed to cultivate a single crop using x kg per hectare of a modern input, say chemical fertilizer, and / hours per hectare of labor. Crop yield y (kg per hectare) is produced with a fixed technology represented by the neoclassical production function y = f(x, 1). Let w be wage rate and v~ and p be the "effective" or farm gate prices of output and fertilizer, respectively, discussed below. Per hectare land rent, p, is defined as the maximal profit that can be earned on a hectare of land, p(w,v,p)- max{py-wI-vx} (1) I.x and can be thought of as a long-run average. Effective prices are determined by the economic geography, which is illustrated in Figure 1. A highway of arbitrary length through the countryside runs through a large city where all fertilizer is produced and output is purchased. The highway transects a series of otherwise isolated mountain valleys along which all farms are located. This is a convenient fiction, but not unlike the geography of Nepal. The total cost of transporting 3  goods between farms and the city has two components: a relatively large cost of headloading goods (i.e., using human porters) between the farm and the road and a relatively small cost of trucking goods along the road. All farmers trade agricultural output and fertilizer in a competitive market center located at the road juncture with their valley (markets at intermediate points up the valley are an inessential complication since goods must still be headloaded from the main highway). From a given farm, it takes h hours to walk to the market center and the portage cost of goods is b Rupees/(kg hours), where b is proportional to the wage. If the money prices of fertilizer and output at a particular market center are v and p, respectively, then the effective purchase price of fertilizer is v = v + bh Rupees/kg and the effective selling price of output is p = p - bh Rupees/kg. All labor can be obtained locally with zero transport costs. Unprofitable land will not be cultivated, so the limit of cultivation in terms of walking time to the market center, h , is implicitly defined by p(h ;w, p, v) = 0 . As figure 1 illustrates, h declines across valleys as one moves away from the city, because p declines and v increases; ultimately, h = 0 and all cultivation ends. As to the relationship between land rent and travel time, by the envelope theorem -b(y + x) for h < h- (2) So, the negative rent gradient is just equal in magnitude to the total transport costs per hour per hectare. Furthermore, by the convexity of the profit function in prices, > 0 for h < h (3) Thus, along any given valley, the rent function is convex. 4  Notice that convexity of the rent function does not require that farmers both purchase fertilizer and sell output at the same time (though nonparticipation in the latter market means that rents depend upon the endogenous shadow price of output). However, if one moves far enough away from a market center, farmers may stop selling output and buying fertilizer altogether, and the rent gradient would be zero beyond this point (and thus the rent function not strictly convex). Convexity is also robust to the assumption of a single crop or production technology. Figure 2 shows how the rent function along one of the valleys in Figure 1 reflects the profit maximizing choice of available crops or technologies; as travel time rises, farmers may switch away from bulkier crops or from agricultural practices that are intensive in modem inputs. Roads and Welfare Using the above framework, consider the welfare implications of building a road of given length off of the main highway into a particular valley. The local scale of the project ensures that it has no general equilibrium effects on wages or prices. To avoid specifying the source of public finance, assume that the project is funded by earmarked foreign aid. Let A denote the length of the road in foot-travel (hours walking time) equivalents. By enabling truck transport, the road effectively cuts portage costs by some fraction ,u. The new rent function, suppressing its dependence on prices and on ',u2 iS 2The parameter u reflects road quality. I do not consider the welfare cffects of variation in p because, as a practical matter, the cost of upgrading road surface (from earth to gravel or asphalt) far outweighs the small reduction in vehicle operating cost, once truck transport is feasible (Beenhakker and Lago, 1983). Although improvements in road conditions, given surface type, can substantially reeuce vehicle operating cost, these cost-savings are likely to be small compared to those of a new road. Of course, in the extreme case where an existing road is impassible to trucks, a road improvement is tantamount to a new road. 5  a(A,h) = p(u) forh < (4) =p(h-2(l-,u)) forh