Agricultural Development and Structural Transformation in a Closed Economy:

Transcription

1 Agricultural Development and Structural Transformation in a Closed Economy: Evidence from Uganda Douglas Gollin Williams College Richard Rogerson Princeton University World Bank ABCDE Conference, Paris May 30, 2011 Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

2 Outline 1 Background and Motivation 2 Models 3 Conclusions Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

3 Outline 1 Background and Motivation 2 Models 3 Conclusions Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

4 Background Structural transformation process requires models that let us think about movements of people and resources across sectors. What determines allocations of people across sectors? How would alternative policies or public investments change this allocation? We propose using simple general equilibrium models static or dynamic to think about how policies aect structural transformation and how they alter welfare. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

5 Uganda Study This presentation draws heavily on work focused on Uganda. The question motivating that study was why so many people in Uganda live and work in quasi-subsistence agriculture. The same question could apply in many other countries of sub-saharan Africa; our research focused on Uganda because (a) we went there; and (b) it seemed fairly representative of other relatively closed agriculture-dependent economies. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

6 Central Questions Why are so many people in agriculture? Within agriculture, why do people consume such a large fraction of what they produce? Explore three possible explanations: Low agricultural TFP Low non-agricultural TFP High transport costs Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

7 Uganda Background Large fractions of Uganda's population live in rural areas (85%) and work in agriculture (73%). Agricultural sector appears to have low productivity, relative to non-agriculture. Agriculture accounts for 20% of GDP, suggesting that output per worker in non-agriculture is greater by a factor of 12! (But lots of measurement issues!) Poverty is relatively concentrated in rural areas (93%). Poverty rates are much higher among rural households (34%) than urban households (14%). Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

8 Agriculture Within rural Uganda, most individuals work in agriculture, although non-farm enterprises are also important. Farms primarily produce staple foods on small plots. Major crops are cooking bananas, cassava, maize, beans. Each of these is grown by approx % of agricultural households. These four crops together account for 55% of crop area. Cash crops (coee, tea, sugar, cotton) together account for less than 8% of cropped area. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

9 Large-scale and smallholder agriculture A large-farm sector is found in tea, cotton, and sugar. Approximately 400 registered enterprises employing 28,000 workers. Smallholder farms account for almost all of agricultural employment and output. Approximately 4.2 million agricultural households employing approximately 8 million workers. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

10 Semi-subsistence farming Most farms produce similar food crops on very small scale. Large fractions of output are consumed at home. Banana (68%) Cassava (77%) Maize (48%) Beans (84%) Households trade small amounts of their output, often selling in local markets to buy non-agricultural consumption goods and intermediate inputs: salt, soap, kerosene; farm tools and chemicals. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

11 Evidence: Low agricultural productivity Uganda's crop yields lag far behind world leaders in all crops but cassava. Maize yields only 30% of world average and 11% of world leaders. Yield decits are large even relative to countries with similar agroecologies. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

12 Crop yields are low Uganda Crop Yields Relative to World Leaders Maize Sorghum Cassava Beans Coffee Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

13 Evidence: Low non-agricultural productivity Average productivity is higher than in agriculture, but it is still low by world standards. Uganda exports very limited quantities of non-agricultural goods: some minerals (gold, cobalt, petroleum); soap; hand tools; and electric current. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

14 Evidence: Remoteness and transportation Rural households are frequently remote: Three-quarters live more than two hours from a market center. One-quarter live more than ve hours from a market center. Average distance to a health clinic was 7 km (including in cities); 77 percent of people report that they walk to get to clinics. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

15 Transportation Roads are very bad, especially in rural areas. Measured transport costs are very high, and price wedges between dierent markets appear to be extremely high. The cost of moving 100 kg of agricultural goods 100 km was as high as $5.43 (trucking matoke from Mbarara to Lira), compared with $0.573 for corn in the US. Uganda's paved road density in 2003 was 16,300 km in an area of 200,000 km 2. By contrast, Britain at the time the Romans left (AD 350) had 12-15,000 km of roads in an area of 240,000 km 2. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

16 Bad roads Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

17 Transport Costs Dried Cassava Marketing Costs Ush/100 Kg Farm Gate Price 4000 Transport (Farm gate to primary market) 1000 Market dues/local tax 200 Capital costs and operating surplus 800 Primary Market Price 6000 Bagging materials 500 Labour costs (loading/unloading, sorting, weighing) 700 Transport (rural to urban market) 3500 Market dues/local tax 500 Trading Licence & security 200 Capital costs and operating surplus 1600 Wholesale Market Price Source: Plan for Modernization of Agriculture (2002), pp Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

18 Outline 1 Background and Motivation 2 Models 3 Conclusions Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

19 Modeling strategy To answer these questions, we need a general equilibrium model with at least two sectors and with explicit transportation costs. Ultimately we also need to distinguish between agriculture that is semi-subsistence and more commercially oriented agriculture. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

20 GE but Not CGE We will solve the model computationally, and we will use a general equilibrium model, but it is not a CGE model in the commonly used sense. No hidden equations; no hidden elasticities or linearities. (But less detail in terms of sectors...) Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

21 Related literature Close in spirit to Eswaran and Kotwal (1993). Some similarities to Caselli and Coleman (2001), Vollrath (2004, 2008). Also related to work on transportation and growth in Herrendorf, Schmitz, and Teixeira (2006, 2008); Adamopoulos (2005). Most similar to models of structural transformation in Gollin, Parente, and Rogerson (2004, 2007) Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

22 How to Drive an Economy into Subsistence The Food Problem Bites... Closed economy. Food is essential. Labor productivity in agriculture is low and farmers cannot produce much surplus. Large fractions of the labor force will be devoted to agriculture. Increases in agricultural productivity will have a large impact on the proportion of people in agriculture. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

23 Environment Consider a model with three regions: A city that produces non-agricultural goods Two agricultural regions that produce food. The two agricultural regions dier in that one is close to the city and the other is remote, as measured by transportation costs. A representative family that allocates its labor across regions. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

24 Schematic Representation Remote Agriculture Near Agriculture City Manufacturing a a Region 2 Region 1 Region 0 Cost q2 Cost q1 m m Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

25 Model Basics Preferences: α log(a ā) + (1 α) log(m + m) Agricultural technology (in either region): a = AaF (l, x, na) = Aal 1 θx θn x θx n θn Manufacturing technology: m = Amnm. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

26 Three-Region Model Endowments l 1 = 0.1 l 2 = 0.9 Labor is allocated endogenously. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

27 Feasibility Conditions n 0 m 0 + n 1 m 1 + x 1 1 q 1 + n 2 m 2 + x 2 (1 q 1 )(1 q 2 ) = A mn 0 n 0 a 0 (1 q 1 ) + n 1a 1 + n 2 (1 q 2 )a 2 1 θx θn = Aal1 x θx 1 nθn + 1 (1 q 1 θx θn 2)Aal2 x θx 2 nθn 2 Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

28 Equilibrium Allocations For interior solution: m 0 + m = m 1 + m (1 q 1 ) = m 2 + m (1 q 1 )(1 q 2 ) a 0 ā (1 q 1 ) = a 1 ā = (1 q 2 )(a 2 ā) Corner solutions are plausible under some specications. We solve for them and check for them computationally. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

29 Quantitative Experiment Parameter Values We loosely calibrate to micro data from Uganda. A a = A m = 1 θx =.2, θ n =.4 α =.20 m = 0 ā n 1 + n 2 = 0.80 q 1 = 0.1, q 2 = 0.6 Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

30 Benchmark Allocations Consumption Allocations: Three Region Model n 1 n 2 a 0 a 1 a 2 m 0 m 1 m Individuals in the near agricultural region consume bundles quite similar to urban residents. Individuals in the remote agricultural region seem to be in quasi-subsistence. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

31 Agriculture, Near and Far Inputs and Intensity across Regions Agriculture Production: Three Region Model l 1 /n 1 l 2 /n 2 x 1 /n 1 x 2 /n 2 ya1/n 1 ya2/n 2 ya1/l 1 ya2/l Labor intensity is greater in near region. Far greater use of intermediates in near region. Output per unit of land (yield) is 40% higher in the near region. Dierences in output per worker are not large. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

32 Equilibrium Scenarios Experiments in the Three Region Model: Consumption Allocations n 0 n 1 n 2 a 0 a 1 a 2 Benchmark A a = A m = q =.9q A a, A m, q A a, q l 1 = Pop Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

33 Manufacturing Consumption in Equilibrium Experiments in the Three Region Model: Manufacturing Consumption m 0 m 1 m 2 Benchmark A a = A m = q =.9q A a, A m, q A a, q l 1 = Pop Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

34 Agriculture in Equilibrium Intensication and Input Use Experiments in the Three Region Model: Agricultural Production ya1/n 1 ya2/n 2 ya1/l 1 ya2/l 2 Benchmark Aa = Am = q =.9q Aa, Am q Aa,q l 1 = Pop Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

35 Agriculture in Equilibrium Intensication and Input Use Experiments in the Three Region Model: Agricultural Production l 1 /n 1 l 2 /n 2 x 1 /n 1 x 2 /n 2 Benchmark Aa = Am = q =.9q Aa, Am q Aa,q l 1 = Pop Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

36 Findings Improvements in agricultural TFP have a pronounced negative impact on the number of people working in quasi-subsistence but an increase in the number of people working in commercial agriculture. Decreases in transportation cost lead to decreases in the total agricultural population and in particular the proportion of people in quasi-subsistence. Crop yields in remote areas do not change very much with interventions. There may be little feasible way to aect these yield levels, even if agricultural TFP increases, because of compensating changes in labor allocations. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

37 Outline 1 Background and Motivation 2 Models 3 Conclusions Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

38 Conclusions In an eectively closed economy, with agricultural goods displaying high income elasticities of demand, improvements in agricultural productivity necessarily matter. Agricultural productivity increases will release labor (and potentially other resources) to other sectors. High transportation costs will tend to keep people stuck in rural areas. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

39 Extensions Should be able to add a number of alternative public goods and to focus on the modeling of public investments: Trunk roads vs. rural roads Input subsidies Port facilities and access to external markets Consider the dierent implications for economies that are open to trade. Dynamics Gollin and Rogerson () Agriculture and Transformation ABCDE / 40

40 Acknowledgments Thanks to NBER's Program on African Successes. We were hosted in Uganda by the Makerere University Institute for Social Research (MISR). Thanks also to Wilberforce Kisamba-Mugerwa for intellectual and logistical support in Uganda. Gollin and Rogerson () Agriculture and Transformation ABCDE / 40