The Impact of Agricultural Price Policies on Demand and Supply, Incomes, and Imports; An Experimental Model for South Asia SWP-277 World Bank Staff Working Paper No. 277 April 1978 he views and interpretations in this document are those of the authors and ghould not be attributed to the World Bank, to its affiliated HD rganizations or to any indiviclual acting in their behalf. 144 7 .0.77 c. 2 repared by: Marc Osterrieth, Eric Verreydt and Jean Waelbroeck (consultants) Economic Analysis & Projections Dept. Copyright.@) 1978 0 i1111111 1l 1lill 1ll 1l S~~~.A. SLC002053 XI The views and interpretations in this document are those of the authors and should not be attributel to the World Bank, to its affiliated I 7 organizations or to any individual acting in their behalf. WORLD BANK Staff Working Paper No. 277 THE IMPACT OF AGRICULTURAL PRICE POLICIES ON DEMAND AND SUPPLY, INCOMES, AND IMPORTS; AN EXPERIMENTAL MODEL FOR SOUTH ASIA April 1978 This paper examines the impact of agricultural price policies on foodgrain production, consumption, and trade in South Asia. The approach is aggregative, based on a model which describes the whole of the economy of that subcontinent. In the version of the model which is used in this paper, growth in the non-agricultural sector is taken to be exogenous. Both demand for and supply of agricultural products depend on their price. The demand function is specified according to the Extended Time Expenditure System, and describes both own and the cross price elasticities of demand. The supply response reflects short run adjustments which take place as farmers adjust fertilizer input in response to prices, and long term adjustments due to the impact of prices on the rates of growth of the labor force and of the cultivated area. The model suggests that the impact of price on consumption, production, and trade is quite significant. It also suggests that appropriate price policies would be able to accelerate agricultural production sufficiently to accommodate a higher rate of growth of urban production. s;C-OR,,AL 1,1B'RAR FOR c, 'MT RECOM;SfFw AN EC Prepared by: Copyright @ 1978 The World Bank Marc Osterrieth, Eric Verreydt 1818 H Street, N.W. and Jean Waelbroeck (consultants) Washington, D.C. 20433 Economic Analysis & Projections Dept. U.S.A. TABLE OF CONTENTS Page I. INTRODUCTION 1 II. EQUATIONS FOR DEMAND AND SAVING 2 Estimated equations 3 III. THE AGRICULTURAL PRODUCTION FUNCTION 4 Estimated equations 6 IV. THE RESPONSE OF AGRICULTURAL INPUTS 7 Specification of the fertilizer input function 7 The long run response of the labor force: background and specification 8 The long run response of capital and land: background and specification 8 V. RESULTS OF MODEL SIMULATIONS 9 The base case results 10 The model's response to population growth and to growth of the urban sector 10 The model's response to prices 11 An "accelerated growth" solution 11 Table V.1: Main Results of the Simulations 13 Equations 14 Notations 15 BIBLIOGRAPHY 16 The impact of agricultural price policies on demand and supply, incomes, and imports; a experimental model f'or South Asia.* Marc Osterrieth, Eric Verreydt and Jean Waelbroeck I. INTRODUCTION 1. Though agricultural production has increased substantially in developing countries, much of the increase has been absorbed by population growth. Further, production per head has risen very slowly and has lagged behind consumption, while imports have increased. A number of recent studies have suggested that these trends will continue and that net imports of foodgrains by developing countries will approximately double between 1973-4 and 1985. This projected increase in imports would pose serious balance of payments problems for many developing countries, including some of the poorest. 1/ 2. The methodology used in making these projections is, however, very crude. The projection forecasts are extrapolations from trends. Projections of demand are built up by applying income elasticities observed over time to GNP projections. These forecasts are then marginally adjusted on the basis of special knowledge about conditions in different countries. While knowledge of investment and price policies plays a role in guiding the adjustments these variables do not appear explicitly in the equations of the projection model itself. The failure to use prices is especially a cause for concern because it may be indicative of policy-thinking which is wrong, together with difficulties in specifying adequate equations. For example, much of the literature on the food gap implicitly reflects a view of the world in which demand and supply are independent of prices and in which supply can be increased only by massive programs of government invest- ment. If such programs are not undertaken, then it is held that countries may witness an intractable rise in food imports which will absorb available foreign exchange and choke off the prospects for self-sustained growth in the future. 3. The view taken in the present paper is that what policy makers most need to understand is the trade-off between food imports and agricul- tural prices. Estimates of this trade-off should be the chief basis for / See, e.g., FAO (1974), Sandra Hadler (1976) and IFPRI (1976). * We wish to express our gratitude for the research support received from the Economic Analysis and Projections Department of the IBRD, which made this study possible. - 2 - government decisions to import food and agricultural products or to accept the unpleasant political and social consequences of an increase in agricul- tural prices. Knowledge of this trade-off should also be one of the factors in decisions to step up the public funds allocated to irrigation, rural electrification, agricultural research and other efforts to speed up the rate of growth of agricultural output. 4. In this paper we describe a projection model which takes explicit account of the relation between prices and the agricultural balance of trade. The model is simple enough to be used for projections of supply and demand in whole regions of the developing world or for the developing world as a whole. To illustrate the applicability of the approach, the model is used to assess the impact of prices on the agricultural balance of trade of South Asia. 1/ II. EQUATIONS FOR DEMAND AND SAVING 5. It is desirable to use a specification of demand in which the figure derived from adding consumption of different goods and savings is consistent with disposable income. We also felt, given the poor quality of data in many developing countries, that it was desirable to use a robust specification in which fairly strong conditions were imposed on the co- efficients estimated. An attractive specification from both these points of view is the Extended Linear Expenditure System (ELES) devised by C. Lluch (1973), which expands R. Stone's Linear Expenditures System to an intertemporal framework. The specification is (1) piqi piCi+ bi(y - Epici) (2) s = bs(y - EPJcJ) where qi = consumption per head of good i, Pi its price, y income per head, s saving, ci and bi are constants, and i and j = 1 ... n are indices charac- terizing n goods. The ci are often referred to as "committed expenditures," a term reflecting a popular interpretation of these coefficients as minimal subsistence expenditures. The expression (y - Epjcj) is then referred to as the supernumerary income, i.e., the margin between income and the cost of committed expenditures. The bi are marginal budget shares. 6. Using these interpretations, expression (1) explains expenditures as a linear function of price and supernumerary income, while (2) asserts that saving is a constant fraction of the supernumerary income. It can also readily be checked that if bs + Ebi = 1, then Epiqi + s = y, meaning that the system is additive. In addition, it can be shown that the system is homogeneous of degree zero in prices and income and satisfies the Slutzky integrability conditions. 1/ As defined in this paper, South Asia includes India, Pakistan, Bangladesh and Sri Lanka. - 3 - 7. C. Lluch, A. Powell and R. Williams (1975) have obtained time series estimates of the ELES for 27 countries, including 7 developing countries (Korea, Thailand, The Philippines, Taiwan, Jamaica, Panama and Puerto Rico). The study was based on national income expenditures data, so that agricultural products were valued at the prices paid by consumers and not the prices received by farmers. This is a drawback from the point of view of modelling, as the figure obtained is not directly comparable to agricultural production data, which are valued in national accounts at farm gate prices. 8. Lluch, Powell and Williams have also made extensive investigations of food demand on the basis of consumer budget cross-section data. An important finding confirmed that spending patterns differ substantially between town and country, with the marginal budget share for food being higher in rural areas than in cities. This situation reflects both differ- ences in patterns of living and the lesser availability of manufactured goods outside the cities. J. Mellor (1976, p. 166) provides similar evidence for India, based on his appraisal of the data available. 9. Another significant finding -- which also confirmed previous research -- was that the estimated coefficients depend on income per head: committed expenditures rise with income; the marginal budget share of food declines as income rises. The finding means that the estimated ELES functions must be considered as local approximations of a more complex, underlying demand system and should not be used for projections much beyond the range of incomes observed in the sample period. Estimated equations 10. To test the feasibility of applying the ELES to a broad group of countries, we built up data on consumption of agricultural and non-agricultural products and of saving for 27 developing countries. (The construction of the data is described in an appendix which is available on request.) It is important to note that consumption of agricultural products is described using farm gate rather than retail prices, so that the agricultural consumption series can be compared directly to data on agricultural production. 11. The ELES imposes a strong structure on the data,, and its application to regularly growing series might give a high R2 and significant coefficients, even though the specification is not appropriate. We used the Durbin Watson statistic as a heuristic test for misspecification to help determine whether the estimates were acceptable. The results led to our rejecting the estimates for two countries which have undergone a major social and political upheaval -- Bangladesh and Peru, as well as for all black African countries except Kenya and Nigeria. The difficulties experienced with the latter countries may be due to the low quality of the statistical data available. 12. Findings based on the successful estimates exhibited the same patterns as those noted in the earlier applications of the ELES and of the Linear Expenditure System, in particular the tendency for committed expen- ditures to rise with income and for the marginal budget share of food to fall as income increases. - 4 - 13. For the model described in this paper, we needed ELES demand systems for both rural and urban populations. These systems were con- structed in two stages. First we built up an ELES demand system for South Asia by averaging the coefficients obtained for India, Pakistan and Sri Lanka, using population weights to compute the averages. The coefficients for the rural and urban ELES were then obtained from that system by imposing three conditions: (1) At base year prices, food consumption per head should be the same according to the ELES obtained for the region as a whole and according to the rural and urban expenditures systems. (2) The ratio of supernumerary to total income should be the same in the base year according to the regional and the rural and urban expendi- tures systems; and (3) Marginal budget shares should be the same in the urban, rural and regional demand systems. 14. It is important to examine the price and income elasticities of demand implied by the equations. The income elasticity of demand for agri- cultural products is equal to 0.53 in the base year, the price elasticity -0.52. Both figures are in line with other quantitative studies of demand in low income countries (see, e.g. J. Mellor (1966), Chapter 4). The major study of consumer demand in developing countries -- by R. Weisskoff (1971) -- actually arrived at substantially higher income and price elasticities -- 1.1 and -0.87 respectively. 15. The reader should note that the price elasticities defined above describe the response of consumers to increased agricultural prices where their income and the prices of other goods are held constant. In the model we have constructed, an increase in agricultural prices is assumred to increase the income of.farmers, and the elasticity concept as usually defined in demand theory is therefore not considered to be meaningful. When the impact of agri- cultural price changes on the income of farmers is taken into account in the calculations, the price elasticity of demand for food is equal to -0.24. III. THE AGRICULTURAL PRODUCTION FUNCTION 16. The agricultural production function used in the model explains production as a function of changes in the physical inputs of irrigated and unirrigated land, fertilizers and labor. 17. Study of the response of crops to inputs has a long history. The classic book by E. Heady and J. Dillon (1961) provides an excellent synthesis of this type of research. In another path-breaking study, J. Mellor and U. Lele (1964) developed, on the basis of available evidence, a foodgrains production function for India, which they used to analyze the sources of - 5 - growth of production of foodgrains in that country. More recently there have been attempts to estimate the relation between inputs and aggregate production in agriculture. 1/ 18. Our own approach to the problem was inspired by the work of A. Strout (1975), who estimated a cross-section produ'ction function over two periods from data for 16 regions, Strout's specification is (3) xa/tc = a + a(ti/tc) + Yf/tc +6(f/tc)2 where xa is agricultural production, tc the land under cultivation (multi- cropped land being counted only once), ti the irrigated land, and f the input of fertilizers expressed in terms of plant nutrients. 19. As irrigated land is included in the land under cultivation, its marginal productivity is ( a+ 8). The magnitude of the coefficients implies that irrigated land is four times more productive than unirrigated land. This finding reflects the impact of a controlled supply of water both on yields and on increasing the number of crops which can be grown in a single year. 20. Strout's estimates of yields after fertilization are on the whole consistent with other cross-section estimates (see, e.g., P. Timmer (1976)). However, they exceed by some 50% the results indicated by field tests (see, e.g., the results quoted by Timmer (1974)). This discrepancy probably results from the fact that in cross-section data, fertilizer use is positively cor- related with the know-how of farmers and the natural fertility of land, so that regression estimates of these yields are biased upward. 21. The last term in equation (3) tries to capture diminishing returns in fertilizer use, a tendency confirmed both by field tests and by regression estimates. Again, experience in individual developing countries is less encouraging than is indicated by the regressions based on cross-section data which includes both developing and developed countries. According to Timmer (1975) and J. Mellor and R. Herdt (1964), this discrepancy may reflect basic differences in growing conditions in'temperate and tropical climates. What is certainly true is that new growing techniques and improved seeds are required if heavier doses of fertilizers are to be effective. Caution is therefore needed in using regression estimates of fertilizer yields. 22. An obvious omission from the specification above is labor, for which Strout was unable to estimate reasonable coefficients. As development proceeds, the increase in human and physical capital per agricultura] worker more than compensates for the migration of labor out of that sector. Intensity of cultivation therefore rises while labor input declines. Furthermore, at the latest stages of development, production tends to be concentrated on the most fertile lands, because marginal land cannot be exploited economically at the prevailing high wages. 1/ The studies we know are those of the MOIRA group H. Linneman (1975); the SARU group (1976) and A. Strout (1975). - 6- 23. The lack of good econometric estimates of the marginal produc- tivity of labor in agriculture does not mean, however, that this variable is zero. The majority of development economists today reject the once prevalent view that there exists an unproductive labor surplus in agri- culture. A classical summation of the reasons for this view is to be found in T. Schulze (1964). Using the wage rate in agriculture as an indicator of the marginal productivity of labor, Lele and Mellor (1964) estimated that the marginal productivity of labor in agriculture is 40% of its average productivity. This figure is very close to the result obtained by H. Chenery (1970) for developing economies. He tried to explain the economic performance of a sample of developing countries in terms of key growth factors (labor force, rate of saving, ratios of capital inflow and exports to GNP, and labor). His estimates imply an elasticity of 0.4 for GNP with respect to the labor force, about half the coefficient found for developed countries, but still definitely not zero. Estimated equations 24. The agricultural production function of our model is based on the specification proposed by Strout. We applied it to data for 78 countries developed by B. Choe of the World Bank, who participated in the estimation of the function presented below. While Strout's own data was aggregated over regions and covered developed as well as developing countries, Choe's data distinguished individual countries and covered the developing world only. Further, to minimize distortions due to weather, Choe's time series were aggregated into three periods: 1961-4, 1965-7, 1968-70 and 1971-4, yielding four observations per country. 25. Not surprisingly, the correlation coefficient of 0.542 obtained by Choe was substantially poorer than that reported by Strout -- 0.964. Aggregation of countries into regions improves the fit, because the inter- country differences in fertility which account for much of the regression residuals tend to average out when the data is averaged over regions. The inclusion of developed countries in Strout's data "stretched his sample" and improved the correlation coefficient. What is notable, however, is that the order of magnitude of coefficients is the same in both estimates, and the coefficients are quite significant in both. The major difference is the larger value of the 6 coefficient, in the estimate that does not include data for developed countries. It reflects the diminishing marginal productivity of fertilizers. This result is understandable in view of the remarks made above about the differences in growing conditions in tropical and temperate climates. 26. It was decided to introduce labor in the equation used in our model so that the contribution of labor to agricultural output could be reflected. The variable we introduced is the difference between actual labor input and the labor input which would normally correspond to a country's level of development. This "normal labor supply" is calculated using a function of the type used by H. Chenery and M. Syrquin in their recent book on patterns of economic growth (1975). The relation we obtained was the basis for calculating the "normal labor supply" in agriculture for each year in the base case simulation. The levels arrived at were then made exogenous, so - 7 - that the "normal labor supply" was identical in each simulation of the model. The coefficient of the (la - la) term implies that the marginal productivity of agricultural labor equalled 40% of its average produc- tivity in 1975. 27. Finally, we wanted a production function which would reproduce the 1961-75 experience. For this purpose a constant term was introduced into the equation, and coefficients other than labor were adjusted propor- tionately to ensure that the function was capable of predicting exactly 1961 and 1975 production levels. 1/ 28. The production function obtained is given in the appendix to this paper. IV. THE RESPONSE OF AGRICULTURAL INPUTS 29. It is useful to distinguish between the short and the long run responses of inputs to changes in agricultural prices. In the short run, prices can affect the amount of land put into production or the number of crops to be harvested. More careful husbandry offers latitude for increasing output by raising labor inputs. Farmers may use more or less fertilizers. 30. In the long run changes in prices can affect capital and labor supply. Through their influence on incomes, prices can affect incentives to migrate from agriculture to other sectors. The ability and willingness of farmers to improve and extend their land also depends on the level of their incomes. Specification of the fertilizer input function 31. The short run responses of inputs of labor and land to price changes are hard to study in quantitative terms because of an absence of data. With some exceptions, such as Argentina, the amount of land under cultivation in developing countries is not affected sharply, and the data necessary for accurately measuring short run changes are not available. Weather also affects the responses in terms of the feasibility and extent of multi-cropping and the area of land which can be harvested. 32. There is even less data on short run changes in labor inputs. The chief sources of information on the agricultural labor force are censuses, but these are useful only in mapping out long term trends. 33. Fertilizers are the only input for which it is possible to observe short run fluctuations. Not surprisingly, therefore, numerous studies of the response of fertilizer demand to prices are available. An excellent 1/ The proportional adjustment of the estimated coefficients also allows for differences between the domestic prices used in the model to value production, and the somewhat arbitrary prices used by both Strout and Choe to evaluate output. survey of these has been done by P. Timmer, who concluded that the short run elasticity of demand for fertilizers in response to prices is around -0.5, whereas the long run elasticity is as high as -1.5 to -2.0. In our model, the price response of fertilizer demand is assumed equal to -1.0. 34. The function used in the model also reflects long run factors which influence demand for fertilizers. One of these is the increase in cultivated and irrigated areas. The coefficients assume that fertilizer input per unit of output on irrigated and unirrigated land is equal in any given year. 1/ Fertilizer input coefficients are also markedly influenced by trends in the development of new fertilizers which provide intensive high yields and the improved understanding of farmers of the role of fertilizers in increasing output. A coefficient for the trends was chosen that assumes that demand for fertilizers will grow by 80 percent over the next 10 years if prices do not change, the situation predicted by the fore- casts of specialists for the agriculture of South Asia (see, e.g., the forecast for India by W. Hendrickx, (1975)). The long run response of the labor force: background and specification 35. Many studies, the most important of which are those of Colin Clark (1940), Simon Kuznets (1971) and H. B. Chenery (1975), have identified and measured the long 'run relation between GNP per head and the share of the labor force in agriculture. Their work has put in quantitative focus the massive urban migration of workers which is a part of the growth process. It is generally accepted that economic incentives are the main force behind this movement. The best known statement of this idea is that of J. Harris and M. Todaro (1970), which has inspired a great deal of work on individual countries. Recently Linneman, et al (1975) and Y. Mundlak (1976) have estimated, using cross-sections of countries, functions which explicitly link rural-urban migrations to income differentials, as well as to other variables. 36. Our model's equations are based on the estimates of Mundlak. We have adjusted his multiplicative constant to match migration between 1960 and 1970 to observed data. 37. The specification used is given in the appendix to this paper. The long run response of capital and land: background and specification 38. Changes in agricultural prices affect both the incentive to increase agricultural,output and the ability of farmers to purchase required equipment and materials. By influencing rural purchasing power, prices affect the ability of local authorities to finance investments in local infra- structure, which play an important role in facilitating increases in production. Prices probably influence even the decisions of the central government by making clearer the profitability of investment projects. 1/ This implies that per hectare input is 4 times as high on irrigated as on unirrigated land. - 9 - 39. We do not know of any studies specifying agricultural investment equations for developing countries, and it is doubtful that it will be possible to estimate good functions for this aggregate. This is in part due to a lack of data. The first broad set of data on agricultural invest- ment in the developing world was published in the World Tables, but the series do not go back beyond 1967 and are too short for estimations except in country cross-sections. The national accounting series are heavily influenced by large public investment projects in irrigation and rural electrification and do not give a good measure of the efforts of farmers to improve and extend their farms. Many of the farmers' efforts involve their own labor and local materials. The only "statistically visible" component -- purchases of tractors and agricultural implements -- probably represents only a fraction of actual total investment in developing countries. 4o. In the absence of good econometric data, we decided to adopt a crude but hopefully sensible representation of the response. It is based on four assumptions: a) In the central variant of the model, irrigated and non- irrigated areas were assumed to grow at rates corresponding to the projection for India made by Hendrickx (1975). b) The coefficients of the investment response function assumed that investment by farmers accounts for 80% and 40% of the increases in non- irrigated and irrigated areas respectively. They allow for government invest ment in land reclamation, extension of government canals and the contribution of rural electrification to the spread of tubewell irrigation. c) The -extension of the cultivated area due to private efforts is assumed to be proportional to the savings of farmers, as predicted by the appropriate ELES function. d) The extension due to government efforts was assumed to be exo- genous. In the base case solution it increases at the rate projected for India by Hendrickx for the increase in total cultivated and irrigated areas. 41. The equations used are given in the appendix to this paper. V. RESULTS OF MODEL SIMULATIONS 42. As 1974 was the last year for which we had complete data, the model was used to simulate the period 1974-84. Simulations were made as though the period 1974-84 was entirely in the future, so that any changes, e.g., in economic policies, would have come into force in 1975. The results are therefore to be taken as indicators of the impact of policies pursued over a decade, rather than as calculations of goals which are achievable by 1984. Table V.1 provides a summary of the main results of the simulations. - 10 - The base case results 43. The base case projection of the model, which assumed unchanged domestic agricultural prices, indicates a sharp deterioration in the agri- cultural trade balance. 44. A direct comparison with the results of other studies is not possible. The $5.3 billion deterioration predicted by our model covers all agricultural products, whereas other studies have focused on foodgrains only. However, our model does seem to give more pessimistic projections than do other studies. At 1974 prices, the predicted increase in the deficit is equivalent to 29 million tons of grains, roughly double the deterioration foreseen by other studies. 45. The discrepancy undoubtedly results in part from the inclusion in our study of all agricultural products and not just foodgrains. We suspect, however, that a more important reason is our use of a demand function which is responsive to prices, instead of the regressions of food demand on income used by others. The relative price of agricultural products had risen during the sample period (though there was a sharp fall in 1975). Price sensitive demand functions such as ours tend, then, to yield higher income elasticities than do straight regressions of consumption on income. 46. The base case simulation implies that the gap between urban and rural incomes will widen, whereas in the past the ratio between the two has been roughly constant. This reflects the assumptions that agricultural prices will cease to rise as compared with urban incomes. It can be calcu- lated that agricultural products will account for a fifth of the increase of per capita consumer spending over the coming decade. 47. It is interesting to distinguish between the contributions to output resulting from extending the cultivated area and from higher per hectare applications of fertilizers. Increases in cultivated areas would account for 53% of the increase in output, greater use of fertilizers for the balance. Diminishing returns for fertilizer use would set in as follows: the marginal yield of fertilizers would drop from the equivalent of 15.6 kilo- grams of foodgrains per kilogram of plant nutrients in 1974 to 14.1 kilograms in 1984. The model's response to population growth and to growth of the urban sector 48. We ran two simulations to test the sensitivity of the agricultural trade gap to the rate of growth of population and of the urban sector. They indicated that the gap is not very sensitive to the kind of reduction in population growth which may be hoped for in the next decade. A 0.3% drop in the rate of population growth would reduce the trade gap by $0.9 billion. This decline would result entirely from a decrease in consumption. The agricultural labor force would grow more slowly than before, and this would have a negative impact on output. That impact, however, would be entirely offset by an increase in per capita saving and investment by farmers which would occur as a result of faster growth of their per capita incomes. - 11 - 49. Given a 1% acceleration in the rate of growth of output of the non-agricultural sector, a $2.0 billion increase in the agricultural trade deficit would result. The ratio of urban to rural incomes would rise by 21% as compared with 8% in the base case, but that change would have only a small impact on migration out of agriculture. 50. It appears that the migration function used in the model is not very sensitive to income disparities. The model's response to prices 51. The trade gap would respond strongly to a 1% per annum increase in the ratio of agricultural to non-agricultural prices. Consumption would drop by $1.1 billion, production increase by $2.4 billion. As a result, the trade gap would fall by more than half. 52. The increase in farm incomes would then outstrip that of urban workers with farmers benefitting from the increase in output generated by the price increase as well as from that increase itself. The rate of growth of their nominal incomes would be 1.5% higher than in the base case. In real terms -- allowing for the higher price of the food which they consume -- the improvement would be 1.4%. 53. It is interesting to calculate the supply response of the net sales of farmers -- the much debated net supply response of subsistence agriculture (see, e.g., J. Behrman, (1968)). Higher prices would lead to a 7.7% increase in the food consumption of farmers, because of the impact on their incomes, while, as a result of reduced migration, farm population would exceed by 0.5% the figure derived in the base case. Hence slightly more than half of the 4.2% increase in agricultural output would be retained on the farm, to be consumed there. An "accelerated growth" solution 54. As a final test of the model, we simulated economic growth in South Asia using assumptions which implied a substantial improvement in economic trends in that area. The assumptions were: a) A 2% increase in the rate of growth of non-agricultural output. This could be achieved without any increase in the investment rate by reducing the incremental capital output ratio to the level achieved in many other devel- oping countries. b) A 1.6% per year increase in agricultural prices to maintain rural incomes in balance with urban incomes and to keep the agricultural trade gap from growing to unmanageable proportions. c) A gradual drop in the rate of population growth to a level in 1984 which would be 0.3% lower than that assumed in the base case. - 12 - d) Increases. of 30% in both the governments' and farmers' acti- vities to extend cultivated and irrigated areas. For government this could be achieved by better upkeep of canals and drainage and by stepping up the spread of rural electrification. The investments of farmers could be promoted by improving the rural credit network and similar measures. 55. The model suggests that, given these assumptions, there would be a balanced increase of rural and urban incomes, with per capita rates of growth of 3.3% and 3.5% per year respectively. These gains would tend to accelerate over the period. What is interesting is that the agricultural trade gap would remain manageable: it would increase by $1.3 billion, equivalent to 7 million tons of foodgrains. Table V.1: MAIN RESULTS OF THE SIMULATIONS* Base Base Low High Agric. Growth year case pbpulation urban price takeoff 11974 1984 growth growth increase 1984 ,1974________________ 1984__ 1984 1984 1984 Growth rate: agricultural prices - 0.000 0.000 0.000 0.010 O.o160 Growth rate: Aon-agric-ultural output - 0.052 0.052 0.062 0.052 0.0720 Population growth rate - 0.026 0.023 0.026 0.026 0.0245 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - _ - - - - - - - - - - - - - - - - - - - - Fertilizer consumption (mil. tons NPK) 3.69 6.58 6.0 6.60 ,6.0 6.58 ,6.0 7.40 7.2 8.43 9.0 Net arable land (millions ha) 228.10 243.30 o.6 243.40 I0.6 243.30 0o.6 244.60 0.7 251.50 1.0 Net irrigated land (millions ha) 53.20 71.10 , 2.9 71.20 ,2.9 71.20 12.9 71.90 , 3.1 80.20 4.2 Agricultural labor force (mil. workers) 198.40 237.80 1.8 230.80 ,1.5 237.10 1.8 238.70 , 1.9 234.60 ,1.7 Rural-urban migration (mil. workers) 1.37 1.94 3.5 1.83 , *2.9 2.05 14.1 1.78 , 2.7 1.81 ,2.8 Agricultural production (bil. 1974 $) 43.00 53.80 2.3 53.80 , 2.3 53.90 ,2.3 56.10 1 2.7 59.80 ,3.3 Per cap. income, farm pop. (1974 $) 84.60 87.50 0.3 90.20 0o.6 87.60 j0.4 101.50 , 1.8 117.40 ,3.3 Per cap. income, urban pop. (1974 $) 150.90 169.60 1.2 175.80 1.5 187.30 12.2 170.70 , 1.2 212.90 , 3.5 Urban labor force (mil. workers) 101.20 149.50 4.0 144.20 3.6 150.20 4.0 148.50 3.9 146.40 3.7 Per cap. consumption, agric. prod., I farm pop. (1974 $) 51.60 52.40 0.2 53.20 0.2 52.50 ,0.2 53.30 0.2 55.50 0.7 Per cap. consumption, agric. prod., I urban pop. (1974 $) 70.60 75.90 0.7 77.70 , 1.0 81.00 11.4 71.60 0.1 79.20 1.1 Per cap. saving farm pop. (1974 $) 10.10 10.60 0.5 11.10 ,0.9 10.70 ,0.5 12.50 , 2.2 14.90 ,3.9 Total consumption, agric. prod., (bil.$) 43.60 59.70, 3.2 58.90 o 3.1 61.70 ,3.5 58.60o 3.0 61.70 3.5 Agric. imports (bil. 1974 $) 0.58 5.92 5.o6 , 7.86 , 2.48 1.90 * In each column, the first number indicates the level reached by the variable in the given year, the second its 1974-84 rate of growth. 56. Equations; (Notations on these equations appear on the following a~e .1 Consumption, saving of farmers and personal incomes q- L62{n [34 04p,,+ (L269ya 34.04pa + 3.58pO3 aa + nfI40.92pa + 0.269yYa - 40.92pa - 15..V po s= O.I86r y - 34.04p + 3.58po1 a ~ a a Y= (paxa - Pff)/n y= 36399(1.052) /n Yo ~~~~~0 Production function x = I5 484.+ 3.256f + 44.050t- + I2I.Ot - O.OI2(f 2/t j+ 81.26(i - 1 a I c i c a a x 0= O.OI5 1 e t-79661ny - 0.1924(lny)] a a y= [n Y + n C/ n + ni Agricultural input response f = (pa/pf)(9.347(tc - tI) + 35.027t i)(1 + 0.055t) (0. 307 l37 (y -0.1)/y) + 0.5711n(1(I ) + 9.867ln(1+n) lmI 0.0082 l e aoa m a 1 = (1+n)l - l a a m n = 2.51 1 a a t 1 = 299(1+n) 10 1 1 la n = 2.51,10 t = 0.00022s n + 0.25844(I.0057) 1 + t- c a a c t. 0.00013s n + .29928(I.029) + ti 1 a a Agricultural import gap ma =qa xa - 15 - 57. Notations. Endogenous variablesi f Fertilizer consumption, thousands tons NPK 1a Labor force in agriculture, millions workers 1a Labor force equation, "normal level" predicted by patterns of growth equation, millions workers 10 Labor force, non-agricultural sector, millions workers lm Number of workers migrating from rural to urban areas, millions workers 1 Labor force, millions workers ma Net imports of agricultural products, millions 1974 dollars na Rural population, millions persons no Urban population, millions persons qa Consumption of agricultural products, millions 1974 dollars sa Savings of farmers, 1974 dollars tc Net cultivated area, millions hectares ti Net irrigated area, millions hectares xa Agricultura production, millions 1974 dollars y Average income per head in South Asia, 1974 dollars Ya, Income per head in agriculture in South Asia, 1974 dollars Yo. Income per head in non-agricultural sector in South Asia, 1974 dollars Exogenous variables: n Rate of growth of the population and of the labor force pa Agricultural prices pO Prices of non-agricultural products Pf Prices of fertilizers - 16 - BIBLIOGRAPHY Blakeslee, Leroy L., Earl 0. Heady and Charles F. Framingham, "World Food Production, Demand, and Trade," Iowa State University Press, 1973 Behrman Jere R., Supply Response in Traditional Agriculture: A Case Study of Four Major Annual Crops in Thailand in 1937-63, Amsterdam, North Holland, 1968 Chenery, Hollis B., and Hazel Elkington, "A Uniform Analysis of Development Patterns," Discussion Draft, July 1970 Chenery, Hollis B. and Moises Syrquin, Patterns of Development, 1950-1970, Oxford University Press, 1975 Colin Clarke, The Conditions of Economic Progress, (Second edition 1951, Third edition 1957), Macmillan, 1940 Choe Boem, Marc Osterrieth and Jean Waelbroeck, "Agricultural Prospects of Developing Countries," Discussion Draft (forthcoming in Proceedings of the Fourth Conference on Global Modeling, International Institute for Applied Systems Analysis, Vienna), 1976 Jan De Hoogh, Michiel De Keyzer, Hans Linneman and H.D.J. Van Heemst, "Food for a Growing World Population," in Gerhart Bruckman, ed., Moira: Food and Agriculture Model, Proceedings of the Third IIASA Conference on Global Modeling, September 22-25, 1975, International Institute for Applied Systems Analysis, Vienna, 1977 Food and Agriculture Organization, Assessment of the World Food Situation, Present and Future, Rome, 1974 Hadler Sandra, "Developing Countries Foodgrains Projections for 1985," IBRD Working Paper, 1976 Harris, J.R. and M.P. Todaro, "Migration, Unemployment, and Development: A Two Sector Analysis," American Economic Review, 60: pp. 126-42, 1970 Heady, Earl Q. and John L. Dillon, Agricultural Production Functions, Iowa State University Press, 1961 Hendrickx, William E., "Foodgrain Supply and Demand Situation and Prospects for Eleven Asian Nations." World Bank Agricultural and Rural Development Department Working Paper, 1975 Herdt, Robert W. and John Mellon, "The Contrasting Response of Rice to Nitrogen: India and the United States," Journal of Farm Economics, XLV, PP. 150-60, 1964 IFPRI, "Meeting Food Needs in the Developing World," Research Report No. 1, Washington, D.C., 1976 Kuznets, Simon, Economic Growth of Nations: Total Output and Production Structure, Cambridge, Bellknap Press of Harvard University, 1971 - 17 - Lele, Uma and John W. Mellor, 'Estimates of Changes and Causes of Changes in Foodgrains Production, India 1949-50 to 1960-61, " Cornell International Development Bulletin 2, Cornell University, Ithaca, 1964 Linneman, Hans, "Food for a Doubling World Population Project," Discussion Draft, 1975 Lluch, Constantino, "The Extended Linear Expenditures System," European Economic Review, April 1973, pp. 21-32 Lluch, Constantino, Alan Powell and Ross Williams, "Household Demand and Savings in Economic Development: Applications of Linear Demand Systems," Discussion Draft, 1975 Mellor, John W., The Economics of Agricultural Development, Cornell University Press, 1966 Mellor, John W., Uma Lele, Debra Biamante and Arthur Goldsmith, "Estimates of Foodgrains Production and Marketing from Input Estimates, India, 1949-50 to 1973-74, and Projections to 1983-84," Occasional Paper No. 83, Cornell University -- USAID Employment and Income Distribution Project, Department of Agricultural Economics, Cornell University, 1975 Mellor, John W., The New Economics of Growth, A Strategy for India and the Developing World, Twentieth Century Fund Study, Cornell University Press, 1976 Mundlak, Yair, "Migration In and Out of Agriculture -- Empirical Analysis Based on Country Data," Discussion Draft, International Fund Policy Research Institute, Washington, D.C., 1976 Strout, Alan, World Agricultural Potential, Evidence from the Recent Past, "Discussion Draft" 1975 Schulze, Theodoro W., Transforming Traditional Agriculture, New Haven Yale University Press, 1964 Systems Analysis Research Unit, "Systems Analysis Research Unit Model (SARUM)," Discussion Draft 1976 Timmer, Peter, "The Demand for Fertilizers in Developing Countries," Stanford Rice Project Working Paper No. 5, 1974 Timmer, Peter, "The Political Economy of Rice in Asia, Lessons and Implications," Discussion Draft, 1975 Timmer, Peter, "Fertilizers and Food Policy in Developing Countries," Discussion Draft, 1976 Weisskoff, Richard, "Demand Elasticities for a Developing Economy, An Inter- national Comparison of Consumption Patterns," in Hollis B. Chenery, ed., Studies in Developing Planning, Harvard University Press, 1971 RECENT PAPERS IN THIS SERIES: No. TITLE OF PAPER AUTHOR 250 Alternative Development Strategies of Korea S. Gupta (1976-1990) in an Input-Output Dynamic Simulation Model. 251 A Perspective on the Foodgrain Situation in S. Burki, T. Goering the Poorest Countries 25? City Size and National Spatial Strategies in W. Richardson (consultant) Developing Countries 253 Economic & Social Evaluation of Projects: J.F. Linn A Case Study of Ivory Coast 254 Valorization Charges as a Method for Financing W. Doebele (consultant) Urban Public Works: The Example of Bogota, 0. Grimes Colombia 255 The Employment Impact of Industrial Investment: J. Stern (consultant) A Preliminary Report 256 A "Stages" Approach to Comparative Advantage B. Balassa 257 Measuring the Performance of Family Planning P.S. Mohapatra Programs 258 Multi-Level Programming & Development Policy W. Candler, R. Norton 259 Alternative Concepts of Marginal Cost for J. Saunders, J. Warford Public Utility iricing: Problems of P. Mann (consultant) Application in the Water Supply Sector 260 Meeting Basic Needs in Malaysia: A Summary. J. Meerman of Findings 261 A Model for Estimating the Effects of Credit T. Husain, R. Inman Pricing on Farm-Level Employment and Income Distribution 262 Commodity Price Stabilization and the E. Brook, E. Grilli Developing Countries: The Problem Choice J. Waelbroeck 263 Industrial Policy and Development in Korea L. Westphal, Kwang Suk Lim 264 The Incidence of Urban Property Taxation in J.F. Linn Developing Countries; A theoretical and Empirical Analysis Applied to Colombia No. TITLE OF PAPER AUTHOR 265 India's Population Policy: History and Future R. Gulhati 266 Radio for Education and Development: P. Spain, D. Jamison Case Studies (Vols. I & II) E. McAnany 267 Food Insecurity: Magnitude & Remedies S. Reutlinger 268 Basic Education and Income Inequality in J. Jallade Brazil: The Long-Term View 269 A Planning Study of the Fertilizer Sector A. Choksi, A. Meeraus in Egypt A. Stoutjesdijk 270 Economic Fluctuations and Speed of Urbanization: B. Renaud A Case Study of Korea 1955-1975 271 The Nutritional and Economic Implications of L. Latham, M. Latham Ascaris Infection in Kenya S. Basta 272 A System of Monitoring and Evaluation of M. Cernea, B. Tepping of Agricultural Projects 273 The Measurement of Poverty Across Space: V. Thomas The Case of Peru 274 Economic Growth, Foreign Loans and Debt G. Feder Servicing Capacity of Developing Countries 275 Land Reform in Latin America: Bolivia, Chile S. Eckstein, G. Donald Mexico, Peru and Venezuela D. Horton, T. Carrol (consultants) 276 A Model of Agricultural Production and R. Norton Trade in Central America C. Cappi, L. Fletcher C. Pomareda, M. Wainer (consultants) 6TzT.q Sac '203 pg ;1S3CiG 4