NBER Working Paper No. 15863, April. Douglas Gollin and Richard Rogerson (2010).

This study argues that the poor quality of transportation infrastructure in Uganda is one of the primary explanations for high population density in rural areas and the share of labor devoted to agriculture. 

Uganda is one of the poorest countries in the world, with real per capita income of approximately $1,100 in 2003.  More than 80% of households are located in rural areas and rely on subsistence agriculture.  Rural poverty rates are high, at 34.2% in 2005-2006 compared to 13.7% in rural areas.  Uganda is landlocked and most of its food is produced on smallholder farms and consumed domestically.  Most smallholder farms raise mixed crops, with several varieties growing in the same field.  The majority of crops are grown under one kilometre from farmer households, and 37% of crops are grown over one kilometre away.  Only 1.0% of smallholder plots employ chemical fertilizer, 6.3% improved seeds, and less than 1.0% irrigation.  90% of cultivation consists of matoke (cooking bananas), beans, cassava, sweet potatoes, coffee, groundnuts, maize, millet, sorghum, and sesame.  20% of households cultivate over 5 acres and 7% over 10 acres.  Ugandan farmers also often supplement their income with livestock and nonfarm employment.  In 2001, approximately 20% owned one or more cows, 30% owned goats, and 46% owned chickens.  Approximately 40% of rural households ran informal nonfarm businesses in 2005-2006.

Though smallholder farms dominate Ugandan agriculture, there is a small yet notable commercial sector.  400 commercial farms employing 28,000 were registered with the Ugandan Bureau of Statistics as of 2006-2007.  Uganda also exports a significant amount of coffee.  Export crops of coffee, tea, cotton, and sugar total only 8% of cropped area, however.

This paper investigates rural agriculture in Uganda.  Why do so many Ugandans live in rural areas when urban standards of living are so much higher?  Additionally, why do they remain dependent on subsistence agriculture?  Gollin and Rogerson argue that the poor state of Uganda’s transport infrastructure is the answer to both questions.  Specifically, they hypothesize that “high rural-to-urban transportation costs implicitly create incentives for poor people to live close to their food sources – effectively reducing the real price of food, which is their largest single expenditure category” (9).  This may be a pressing incentive in Uganda, where households spend 50% of their expenditure on food, drink, and tobacco.  78% of Ugandans live at least two hours from a market center.  Uganda’s paved road density in 2003 approximated that of Britain in AD 350 left by the retreating Roman Empire.  Both the Ugandan national government and international institutions including the World Bank, African Development Bank, and national aid agencies have focused on infrastructure as a key target of assistance in Uganda.

An important consequence of poor infrastructure is price dispersion.  Gollin and Rogerson argue that “The poor quality of Uganda’s road network corresponds directly to an environment of high transaction costs that contributes to high dispersion of prices at a moment in time across geographic space.  …these price wedges must, in some sense, reflect underlying transportation costs, or else the pressure would be great to arbitrage away the price differences” (11-12).  Looking at average crop prices for the period July 2005 to June 2006, “nearly 300 USh/kg separated the most expensive price from the least expensive price; and on average over 100 USh/kg separated the second highest price from the second lowest” (15).  Another source of data is shipping costs as catalogued in the 2002 national Plan for the Modernization of Agriculture.  For maize, shipping costs were approximately as high as farmgate prices.  For fresh and dried cassava, transport costs could often exceed farmgate prices.  Overall, Ugandan farm to market transportation costs total an estimated four to seven times those in the United States. 

The authors develop a static general equilibrium model with two sectors, agriculture and non-agriculture.  A defining feature of the model is that transportation costs are high.  Based on this model, Gollin and Rogerson investigate “the extent to which high transport costs can partly account for the large fraction of people living in rural areas in Uganda” (2).  Additionally, they use the model to estimate the impact of four potential developments: increased agricultural productivity, increased non-agricultural productivity, reduced transportation costs, and population growth on a fixed land base.   In particular, they focus on how these changes might affect the number of people employed in agriculture and located in rural areas.

The model predicts that more labor will be allocated to agriculture in three cases: low agricultural productivity, low productivity in the intermediate goods sector providing inputs to agriculture, and high transport costs.  Fourth, if there were a ten percent increase in population (feasible in Uganda where there is high population growth), fewer people (both relatively and absolutely) will be employed in manufacturing and welfare will decrease.  An increase of 3.8% in agricultural productivity would be required just to offset this population growth.  Therefore, “in the presence of a fixed amount of land, population increases require fairly substantial improvements in agricultural productivity just to maintain a constant share of the workforce devoted to agriculture” (46).  Fifth, if total factor productivity doubled in the agricultural and non-agricultural sectors, “the share of labor devoted to agriculture is more than cut in half, and the welfare increase is roughly a factor of ten” (46-47).  If these changes were paralleled by the same improvement in transportation technology, labor devoted to agriculture would decrease by another third and the welfare gain is nearly doubled.  The authors conclude that “the consequences for development of neglecting transportation are very substantial” (47).

In summary, the authors find that “agricultural productivity improvements have a relatively large impact on the economy.  Because the non-agricultural sector is initially small, and because the economy faces a subsistence constraint that limits the expansion of the non-agricultural sector, improvements in non-agricultural TFP have relatively small positive impacts on the economy.  Reductions in transportation costs generate sizeable benefits for the economy and trigger substantial reallocations of labor across sectors.  When agricultural TFP improves at the same time that transportation costs are reduced, the welfare gains exceed those achieved from the two interventions separately, suggesting a kind of interaction effect” (3-4).

It must be noted, however, that market imperfections and transportation cartels have been proposed as alternate explanations for the high cost of transport.  For example, rural infrastructure investment in Cameroon did not decrease rural poverty.  While the policy recommendations of this study are clear, it must be noted that road construction is a very expensive enterprise.  Other studies calculate that construction such that 75% of Ugandans had an all-weather road within 2 kilometers would require spending 3.6% of GDP per year for ten years.  Reaching 50% of Ugandans with lower quality roads would require spending 2.2% of GDP per year.