~ 9 ‘a I f I .It- I... l. 1 ¢ x... .u... a. 3.... I ..... n... , .. n I l .1 1 Ir . «I n r h . c .f ... v .u _ u Nur 0v. 1-. -.D3 0 ut...‘ . by It. I I .{nu u . ~. 3.; . . dacn. - v . . . . I... :3... - 4Q. i. I 1.4 I It. x.o . I . . ..u .. n .A‘ ( . o ‘ fir 3 . 1 .1... .L 9...wa ,. .9...” we. .. .uxa..¢rmu.m14N k2 v41... firmwaufi.” . . 1k......x.\...:.T m. t rb 8.!- ..v. x “‘4 ‘7 I u 0' . . u . . .t ' . Giggbgz‘ "5‘ , aggfiw . . 31.nwmmunddu .. .. . ..........~l.uua~..:.. Unquautwr. niunwvw. . :uahfinuurawmfla: ‘54 . «.mcinwfi .(I 111.1. . 1n.fl4..uhww. 902.625 1965 941 1966 1,007 1967 769 1968 863 J 1969 697 1970 828 709.2 1971 555 (drought years) 1972 933 1973 532 1974 658 l 1975 863 1976 1.174 1977 393 > 855.163 1978 400 1979 910 1980 628 1981 491 J aExcluding the drought periods of 1978 and 1981 which would have made the calculated average much smaller. Source: IMF, BCEA, Ministry of Plan, Ministry of Finance. 13 attempts to "save" the Office, took the big decision to liquidate the Office and farmers' debts were forgiven. It is interesting to note that it is only during the late 19705, a few years before the dissolution of ONCAD, that farmers were allowed to pay part of their debts, vis-a-vis the Office, in the form of millet- sorghum. Soon, the Office found itself with large stocks of millet. Part of the millet was processed into millet flour to be used partly as a substitute for wheat flour in the production of bread. However, the use of millet flour in the production of bread raised heavy opposition on the part of bakers. Considering the cost of wheat flour, millet flour was relatively expensive and its use made the activities of bakers be- come less lucrative. Today, wheat flour is almost exclusively used in the production of bread. Clearly, as far as ONCAD was concerned, the production of millet- sorghum for commercial purposes was essentially discouraged. Not only was the amount of credit granted to cooperatives based on the c00pera- tive's ability or willingness to produce peanuts, but also little room was offered to cooperatives for the repayment of their debts in kind, by using millet. Essentially, all debts were paid either in kind, using peanuts or in cash. Unfortunately, it became clear in the second half of the 19705 that farmers were giving more and more priority to their food crops (essentially millet-sorghum) at the relative expense of pea- nuts. Actually, the Senegalese Sixth Development Plan notes that during the period 1961-80, average yields for millet-sorghum have increased by 1.1 percent a year, while for peanuts, the figure is -1.5 percent a year. Faced with the precariousness of peanut production as the country's main export crop, the government embarked during the 19705 into a 14 program to stimulate cotton production in the Eastern and Southern part of the country. Despite the fact that cotton production is a new activ- ity for most farmers, its performance has been relatively satisfactory. For example, between the Fourth DevelOpment Plan (1973-77) and the Fifth Development Plan (1979-81), the value of groundnut products in the country's total exports fell from 46.4 to 25.7 billion CFA, while that of cotton fell from 2.5 to about 2.3 billion CFA. Cultivated areas for cotton have expanded considerably from 6,450 hectares in 1968-69 to 39,000 hectares in 1980-81, after a peak of 43,850 hectares in 1976-77. However, the expansion of cotton production in the Southern and Eastern part of the country has led to some concerns that rice production could be discouraged because of the relative profitability of the former activ- ity. If this is so, then the growth of imports of rice can hardly be reduced, unless appropriate policies are carried out to discourage the consumption of rice. 2.2 Rice Consumption Policies During most of the era described above, the government has attempted to carry out a consumer price stabilization program for rice. Pressures on the government to carry out a price stabilization policy for rice go back to 1951, when Senegal's main supplier--Indochina--su5pended rice exports. The official retail price on the Dakar market was set at 40 CFA/kg. Because of difficulties of finding other suppliers, temporary shortages caused the domestic retail price to jump as high as 100 CFA/ kg. After import quantities returned to normal, the government faced higher CIF prices than in the past. Pressures developed and trade unions suggested the creation of a rice stabilization fund like the one that existed for peanuts. 15 In the last two decades, stabilization Operations were implemented by a variable levy on imports. Although large increases in world prices were partly reflected in the domestic official price, smaller and tempor- ary fluctuations were absorbed through government taxes and subsidies. A large part of the resources of the stabilization fund were generated from taxes on peanut exports, while taxes on rice imports were levied when world prices of rice were very low. In 1968, 14 percent of the country's total export earnings were used to purchase rice imports. In spite of such a heavy financial burden, the government stood ready to pursue its price stabilization policy for rice. These subsidies forced the marketing board to pay out well over 700 million CFA francs. This announced the first budgetary crisis of independent Senegal. From 1968 to the early 19705, world prices for peanuts increased by 119 percent. Through taxes on the export of groundnut products, benefits to the Fund reached the level of 15 billion CFA in 1974-75, while in the same year, the overall receipts from peanut exports were about 34 billion CFA. Farmers received an estimate of 12 billion CFA. However, world prices of rice increased by 235 percent in 1974 over the average level of 1970 through 1972. Nevertheless, the government continued to subsi- dize rice and by November 1974, such subsidies caused it to pay out 5.5 billion CFA. In 1975, the government raised the consumer price of rice and the producer price of millet by 67 percent and 18 percent, respectively. Possibly, this action was aimed at fostering a partial substitution of rice for millet. However, the effect of the increase in the price of millet on the supply of millet was partially mitigated by an increase in 16 peanut producer prices by 62 percent. At the same time, producer price of rice was increased by 38 percent. Fortunately, rainfall levels in 1975 were very satisfactory, reaching an average of 80 centimeters, as compared with 57 and 55 centimeters in 1974 and 1973, respectively. Rice production increased by 76 percent from the low level of 63,000 tons in 1974. In addition, peanut production increased by 47 percent, while mil- let production increased by 15 percent. Imports of rice in 1975 reached the level of 102,125 tons, the lowest since 1963. However, one major cause for the decline in imports in 1975 was due to unusually large car- ry-over stocks of rice. In 1974, imports of rice had reached 207,195 tons, the highest level ever experienced. Actually, after 1975, imports of rice have always exceeded 235,000 tons. Between 1961 and 1976, rice imports grew at an average of at least 4,000 tons a year or at an annual rate of 2.9 percent a year, while popu- lation growth was about 2.8 percent. However, during the second half of the 19705, the imports averaged about 270,845 tons from an average of 154,275 tons between 1961-75. Most of the increase in rice consumption during the second half of the 19705 was induced by the rapid urbanization the country experienced after the drought periods when urban migration accelerated. The urban population growth rate was then estimated at about 5 percent a year. As Table 2.2 shows, in 1980, approximately one- third of the country's population were living in urban areas, from less than one-quarter in the early 19605. The cereal diet of urban consumers is essentially composed of rice. The Cap-Vert region, which has 20 percent of the total population, con- sumes around 60 percent of imported rice. For example, in 1973-74, 87 percent of all imported rice went to urban areas, 61 percent of which was 17 Table 2.2 Evolution of the Urban Population in Relation to the Total.Population Year Urban Population (1) Total Population (2) Ratio (1+2) 1961 747,270 3,143,000 24% 1970 118,000 4,390,558 27% 1976 1,527,480 5,106,604 30% 1980 1,904,808 5,703,017 33% Source: Direction de la Statistique; Comptes Economiques. consumed in the Cap-Vert region where Dakar, the capital city, is locat- ed. In contrast, the cereal composition of the rural diet is essential- ly reduced to millet and sorghum. In spite of the government's efforts to improve the millet/peanut nominal price ratios from .63 in 1975 to .96 in 1979, the substitution of rice for millet in the urban consumer's diet has been very slow, if not insignificant. This reflects lagging productivity in the production of millet. During the last two decades, millet production has grown at 1.5 percent a year, while rural population increased at a pace of 1.9 percent a year. Not least important is the low priority given to the marketing of millet throughout the country. 2.3 Rice Production Policies Faced with the increasing needs for imports of rice despite govern- ment's attempts to boost the substitution of rice for millet, the govern- ment started to give greater emphasis on rice production to attain self- sufficiency. In the Fifth Development Plan, the government projects 18 self-sufficiency in rice for the country in 1985. However, given the 5 historical level of the country's self-sufficiency ratio in rice (Table 2.3), the attainment of such objective appears rather doubtful. Between 1960-64, the mean ratio was .41. After a slight decrease in 1965-69, the ratio followed a downward trend and reached the low average level of .28 between 1976-80 at the time when the country's ex- ternal debts reached over $1 billion in 1978. The rate of urbanization was estimated at about 5 percent a year in the 19705, while total popula- tion growth averaged 2.8 percent. In the Fifth Development Plan, the government recognized that most of the urban migration is due to the deterioration of the rural economy, particularly during the drought periods. Historically, plans to develop rice production in Senegal date as early as 1947 as part of government concern for agricultural diversifi- cation and the long-term goal of food self-sufficiency. By 1948, under French support, a 120 hectare experimental unit for rice cultivation was expanded by a factor of 5. Problems due to equipment delivery delays, pests and disappointing yields made the cost overruns too high. In spite of further attempts to encourage better farm practices in the Southern regions of Casamance and Sine Saloum, annual food deficits ranged between 60,000 to 80,000 tons, while local rice production in- creased by an average of 7,500 tons during the 19505. After independence, government policies toward agricultural produc- tion were implemented at two stages: at the general level, the govern- ment affects all agricultural production through its policy tools such 5The self-sufficiency ratio is defined as the ratio of domestic pro- duction to net rice availability (local production + total imports). 19 Table 2.3 Average Self-Sufficiency Ratios in Rice Year Self-Sufficiency Ratio 1960 .43 1961 .38 1962 .41 1963 .47 1964 .36 1965 .38 1966 .44 1967 .45 1968 .42 1969 .29 1970 .54 1971 .35 1972 .39 1973 .19 1974 .23 1975 .52 1976 .32 1977 .32 1978 .21 1979 .28 1980 .28 Source: The ratios are calculated using data from Ministry of Plan, BCEAO, Ministry of Finance, World Bank, IMF. 20 as, price setting, provision of infrastructure, etc. At the sector level, Land Development Agencies affect directly rice production in specific areas by providing extension services, marketing facilities, credit, etc. ‘ During the first four years of independent Senegal, little concern was voiced with respect to increasing domestic rice production. By the end of the First Four-Year Development Plan during which a large part of investments were made in social services and in infrastructure, an ex- ternal event corroborated the fragility of the Senegalese economy whose major pillar has been groundnut production. France acceded to the European Economic Community and had to drop its price support for the Senegalese peanut industry by 1967. This amounted to a loss of about one-sixth of the country's total export earnings which, obviously, had serious implications for the entire economy. Agricultural diversifica- tion themes gained in popularity. Backed by strong external assistance, the Senegalese government embarked into programs to stimulate domestic rice production by using Land Development Agencies when, at the same time, assistance was given to peanut farmers through the establishment of the Program Agricole in 1964. Curiously, most of the rice production projects followed irrigation schemes. Up until 1977, a regional LDA did not exist in the Casamance region when the Societe pour la Mise en Valeur de la Casamance, SOMIVAC, was created. Casamance is known to be the only traditional rice growing area in the country and its rainfall levels are not only the most abundant in the country, but also the most stable. For example, during the disastrous drought years in 1972 and 1973, rain- fall in the North of the country dropped by 63 and 56 percent, respec- tively, while in Casamance, the drops were by 31 and 8 percent. In 21 spite of continuous efforts to develop irrigated rice in the Northern region of the Fleuve, Casamance produces over 65 percent of domestic rice. The reasons for the government's bias in favor of rice irrigation in the North instead of rainfed schemes in the South are complex. Pos- sibly, authorities might have thought that irrigation projects for rice would be more profitable and that, since the Fleuve was not a peanut growing area in contrast to the upper Casamance and the Sine Saloun regions, the investments in land and labor in the North would not jeop- ardize production of the country's leading export crop. So far, rice production schemes under the supervision of LDAs have not met their expectations. ONCAD has been given an official monopoly on the collection, milling and distribution of domestic rice. After harvest, farmers under the supervision of a LDA pay in paddy for services received. In the supervised projects of the Fleuve, the services charged often amount to 40 percent of the harvest. The remaining paddy, aside from farmers' consumption, is paid for directly if purchased by 6 SODEFITEX or settled several months after if purchased by SAED or the PRS.7 These latter deficiencies justify the reticence of rice farmers to trade their surpluses through official channels. Between 1965 and 1975, the percent of officially commercialized rice had averaged 2.5 8 percent of local production of rice. However, the private trade of 6SODEFITEX = Societe pour 1e Developpement des fibres du Textile is a LDA operating in the Southeast of the country. 7PRS = Project Regional de Sedhiou = LDA in Casamance in Sedhiou. SAED = Societe d' Amenagement et d' Exploitation du Delta is a LDA oper- ating in the Fleuve (Northern part of the country) for irrigated cultiva- tion. 8See Table 4.1 in Chapter 4. 22 rice is more developed than is usually thought and the evidence is pro- vided by the usual gap (about 10 percent of rice production) that exists between net production retained by farmers and estimates of on-farm con- sumption. Also, official trade of paddy is markedly small in Casamance, probably reflecting the fact that Casamance farmers have fewer debts to the LDAs. In the government supervised projects, production costs are high, making adoption rates very low. At the national level, fertilizer use on rice has been low and the budgetary impact of subsidies on fertilizer has been less than a quarter of a billion CFA between 1966-67 and 1975- 76, while investment costs for irrigated rice in the North provided 4.5 billion CFA for SAED between 1975-77. Currently, SAED is facing a seri- ous implementation problem and it is feared that the failure of ONCAD to organize and promote the development of peanut production might be repeated. At the present time, the construction of the Diama Dam in the Senegal River in the North will, in principle, open thousands of hectares for cultivation of rice. The complexity of irrigated cultivation should, however, warn us that overlooking the implementation problems for rice cultivation in the North could be very detrimental. Even for Africa, as a whole, only Madagascar has a deeply rooted tradition in irrigated rice. CHAPTER 3 APPROACHES IN MODELING INTERNATIONAL GRAIN TRADE If you want to be a good shoemaker, it is not good enough to make good shoes and to know all about making good shoes; you also have to know a lot about feet. Because the aim of the shoe is to fit the foot. - George McRobie in Small is Possible. Because an important theme of this study is to analyze the effects of rice import policies during the last two decades, the successful out- come of such a task is crucially linked to accurate estimations of im- portant parameters. Up to the early 19705, questions of approaches to import demand estimation did not raise much concern and this was, in large part, due to the nature of the international markets for agricultural commodities. Since that time, many events--world food crises, increasing government intervention in trade, etc.--have profoundly changed the nature of in- ternational markets and, therefore, the old schemes using the spatial equilibrium framework may be useless, particularly when it comes to estimating import demand functions of less developed countries. The trade behavior of many countries is generally known to be by and large influenced by government policies. In the case of less developed countries, government intervention in import demand functions are even greater. However, the nature of govern- ment intervention in trade varies largely among countries. Some public interferences in trade rely on quantitative restrictions such as quotas 23 24 so that the supply of foreign goods in the domestic market is exogene- ously determined by government policy. In other cases, government inter- ventions are channeled through a public determination of domestic prices. For example, faced with a decrease in foreign exchange earnings, a govern- ment may choose to increase the domestic prices of imported food in order to discourage the consumption of imported food goods and/or decrease the prices of domestic food goods. In this case, unlike when quantitative restrictions are used, the government determines endogeneously the levels of food imports through its pricing policy. In this chapter, a large emphasis will be given to the modeling of government interventions in import demand functions. The presentation of our econometric model for the Senegalese rice import demand will fol- low in Chapter 4. 3.1 Traditional Approaches in Modeling International Grain Trade During the 19505 and 19605, the use of spatial equilibrium models either to measure or to predict the flows of grain in trade was very con- mon. However, in their analysis of supply and demand relations, spatial equilibrium models (see Bawden) usually assume perfect market competi- tion. Not surprisingly, given the constant evolution of certain struc- tural and behavioral variables that continuously affect the real world, the usefulness of such models was very questioned. As Hassler pointed out, "There is little evidence in the spatial studies that were reviewed that the analysts were very aware or concerned about the logical issues of operational usefulness of the results. Too frequently, these problems are sidestepped with the hope that someone else will follow up at a later date." Similarly, Wallace and Judge reported squared correlations of less than .5 between actual and predicted prices or price differentials 25 for each seven empirical spatial studies reviewed. This low performance of those models (for forecasting purposes) can be largely attributed to their behavior of highly simplifying the realities of studied markets. Even though any economic model must of necessity be a simplification of reality, in case where space is a minor component in face of large in- terferences due to government intervention and complex market structures, some assumptions, unfortunately very critical to spatial equilibrium models, must be altered if the performance of such models is to be improved. Today, greater government intervention characterizes world wheat markets. Unless those "distortions" are explicitly considered in order to capture their effect on domestic production, domestic consumption, imports and exports, accurate analysis of world trade would not be achieved. Similar market distortions exist in the world market for rice. Developed countries have increased their share in the world rice trade largely through increased concessional sales to some rice importers. Pressured by decreasing foreign exchange earnings, LDC rice exporters like Burma and Thailand could be expected to favor more and more govern- ment intervention in world trade for rice to lessen the impact of con- cessional sales on their export earnings. However, it is unlikely that such policies will not be welcomed by LDC's rice importers, particularly the African food deficit countries. Governments in those latter coun- tries are expected to step in the trade of food grain in an attempt to better allocate scarce foreign exchange resources between food expendi— tures and other needed imports for the pursuit of their goal of economic development. During the 19705, the economic performance of most non- oil exporting less developed countries was very unsatisfactory due to 26 slow growth in food production per capita, rising food imports, lagging foreign exchange earnings, etc. Since then, studies on agricultural policies of many LDCs were widespread, many of which focused on food im- ports by poor countries with major emphasis on government intervention. The following section reviews the literature on models used to pattern the food import demand by some selected countries. 3.2 Toward a More Explicit Recognition of Government Intervention in Trade 3.2.1 Exogenous Representation of Government Interventions In modeling food import demand by less developed countries, ex- plicit recognition of government intervention has been widespread. Typically, recognition of government intervention in food import demand is made through the use of a variable designed to capture the effects of public interference in trade. In its simplest form, the effect of government interference is represented by the use of a dummy variable. Such a scheme is used by Weisskoff in his estimation of import demand elasticities for Brazil. In addition to a dummy variable used to cap— ture the effects of changes in the Brazilian regime during the period 1948-75, the model comprises other exogenous variables such as national income, price and a time trend. He posited that the sign of the trend variable will reflect successful import substitution when it is negative and increasing import dependency when it is positive. He used several different equations. One equation modeled the aggregate imports for Brazil while a number of other equations relate different classes of imported goods to the same set of exogenous variables. For the equation estimating the import of consumer goods, the sign of the estimated co- efficient for the time trend was negative. In fact, for the period 27 under consideration, the ratio of self-sufficiency in wheat for Brazil appeared to have followed a slightly rising trend. The share of wheat imports in the Brazilian consumption had dropped to 60 percent in 1972 from 70 percent in 1948. Part of that rise in the trend of self-suf- ficiency ratio was due to the government policy through which producer prices of wheat were favorably supported (see Alain de Janvry). The limitation of models using dummy variables to pattern govern- ment intervention appears strongly when recommendations for policy al- ternatives become an important issue. More specifically, all the dummy variable can do is to attest whether change in regime shifts the import demand function or not. A downward shift of a demand for imports can have different causes: both quotas and other policies that stimulate the trading of domestic wheat through incentives to farmers can yield the same result, as far as quantities imported are concerned. There- fore, an explanation of the nature and effectiveness of government poli- cies with respect to food imports goes beyond the capacity of such models. Because the process of growth and develOpment requires on the part of a poor country a continual adjustment in imports and/or exports to make the trade gap equal the desired gap between investment and saving, a country whose foreign exchange earnings are limited will be expected to link its food import policy with other needed imports to sustain a minimum growth path or other development objectives. Such consideration has probably formed the basis of the proliferated use of a "capacity to import," which has been represented by various variables such as foreign exchange earnings, reserve stocks of gold, etc. In a study of a few 28 Asian countries'1 imports of rice, Islam used price income and foreign exchange reserves as explanatory variables. He assumed actual imports-- in contrast to desired imports--to be determined by foreign exchange revenues when the country faces a foreign exchange shortage. The re- sults of his study show that the effects of foreign reserves were sig- nificant in all countries but Malaysia. Not surprisingly, Malaysia was the only country that did not experience foreign exchange shortages dur- ing the period covered by the study. In addition, his model assumed that policies of domestic price support in those countries are only done for the mere purpose of supporting farm incomes, but not to boost output to reduce imports. Such assumption is quite surprising in that, even if in the short-run rice production is not responsive to price increase, sales of domestic rice by farmers might well increase as long as rural/ urban terms of trade are improved enough. Thus, this effect would re- inforce the effects of foreign exchange reserves on imports. Further evidence on this probable effect of producer price on sale of domestic price is provided by Joquero, Duff, Anden-Cacsina and Hayami in their study of rice in the Philippines. They found that the total price elas- ticity of marketable surplus is clearly positive and averages between .4 to .7. Similar inferences were also made by Krishna and Ehhibler for India. Another study presented by Outta attempts to estimate the im- ports demand for India. The author noted that,"In view of the active and extensive participation of the government in the economic activity of the country over the entire time period under study, the traditional practice of treating all such activities as exogenous should be 1India, Korea, Malaysia, Pakistan, Philippines and Sri Lanka. 29 reconsidered.” However, he went on saying that because government activities are based on objectives and considerations of a basic long- term character, they can be considered "given" for the model. Curiously enough, he nevertheless used a variable that sought to reflect the ef- fects of trade barriers erected by the government. As such, he selects the ratio of India's foreign exchange holdings in time t and the imports on private accounts in period t-l as a proxy for the level of the foreign exchange holdings of the country. The results of his model showed a significant effect of the latter variable on the India import demand. In spite of the relatively good performance of the model, the main short- coming of the study arises from the author's assumption, that government's actions are considered exogenous. Such an assumption is likely to lead the model to miscapture some market distortions caused endogenously by government interventions. This obviously undermines the relevancy of the model as a framework to be used in order to analyze the Indian im- port demand function. To sum up, one should note that the studies presented above deserve credit for their attempts to recognize and incorporate explicitly the effects of government interventions in their model of import demand func- tions. In spite of such credit, we saw that their attempts to include government interventions has been done through the use of numerous vari- ables such as foreign exchange earnings, foreign assets, dummy variables, etc. This leads to the conclusion that the role of government has been so far treated as exogenous to the model. Whether a decrease in foreign exchange earnings leads to an increase in producer price so that local production is stimulated or, alternatively, consumption of food imports is discouraged through increases in domestic consumer prices, the answer 30 is not provided by any of the above approaches. In other words, when government interventions are treated exogenously, the model results can only tell us what happens to import levels when foreign exchange earn- ings decrease or increase. The question of how decreases in foreign ex- change cause a decrease in imports cannot be answered. Therefore, the necessity to consider the endogenous effects of government intervention in food imports has created a good deal of concern on the part of many other researchers. This will be the main focus of the remaining part of this chapter. 3.2.2 Endogenous Representation of Government Intervention In his book entitled, Import Demand in a Small Country, Bautista noted that,"The demand function resulting from the inclusion of an exo- genous policy variable represents a mixture of an import demand function by consumers of the imported commodity and an import policy function by public policy makers setting the restrictions on imports. A more satis- factory alternative would be to represent separately the underlying structural relations." In his study of food imports in the Philippines, Bautista uses four equations; supply and demand for the imported food and supply and demand for the domestically produced one. Although the two categories contain different commodities, some goods such as rice are common. Both demand functions are assumed to depend on own prices and prices of substitutes as well as on real consumer expenditures. The supply of imports by the government is determined by the domestic price of foreign food, the price of domestic food and the capacity to import for which current and lagged export receipts are used as proxy. Therefore, his approach represents explicitly the endogenously deter- mined imports of food through market intervention by the government. 31 This latter is seen to struggle between two conflicting objectives: the necessity to minimize on expenditures on food to meet some investments required by development objectives and the political, social and economic needs to stabilize domestic food prices. The trade-off between conflicting government objectives also under- lies the study by Lattimore and Schuh on the Brazilian beef sector. In their study of the Brazilian beef industry, Lattimore and Schuh present an econometric model to examine the dynamics of beef supply and demand response over the period 1947-71. The model attempts to quantify the effects of government policies and also used a few dummy variables to single out the intervention effects brought about under the different political regimes that compose the period under study. The model is composed of three blocs of equations: policy intervention bloc, domestw: supply and demand bloc and export bloc. The first bloc includes two main equations measuring the level of intervention through price-related com- mercial policy--such as export taxes or an overvalued exchange rate--and through quantitative restrictions. For the price-related commercial policy, the process works in the following manner: if there is an in- crease in the quantity of beef demanded by foreigners, then the export price will rise, ceteris paribus. Should the government wish to prevent the domestic price from rising, the only instrument it has available will be to raise the level of trade intervention for beef. To do so, the government will lower the effective exchange rate by means of either export taxes or an overvaluation of the exchange rate for beef exporters. In the second bloc, the model distinguishes between male and female cat- tle, and postulates an investment and slaughter equation for each. The authors assumed that part of the investment in cattle occurs as a hedge 32 against inflation and the unstable rate of inflation the country exper- ienced has an impact on slaughter rate. Finally, a crop price variable is added to reflect the opportunity cost of land in other enterprises. Domestic demand was simply represented as function of the price of beef, the price of a substitute in consumption (pork) and total consumer in- come. The third bloc, denominated export bloc, consists only of an identity to determine the quantity of beef exports from Brazil given domestic supply and demand. From the empirical results, the statistical support for the maintained hypotheses was mixed. The author mentioned, however, the high level of intercorrelation among the variables and the paucity of good data. Nevertheless, there was good support for the hypotheses contained in the intervention bloc. Not all dummy variables were significant, possibly reflecting a relatively homogenous trade pol- icy under a few different regimes. Finally, the authors estimated the welfare effect of a removal of government intervention. They noted that at the mean values during the period, a complete elimination of inter- vention policies would have increased the domestic beef price by 81 per- cent. This would, after full adjustment of the supply sector, yield a net social gain of 1.001 billion cruzeiros after allowance is made for a decrease in consumers' surplus by 1.850 billion cruzeiros. Expressed another way, on the average, government intervention consisted of a tax on beef producers of l cruzeiro to save the beef consumer .45 cruzeiros. This same trade-off between competing government objectives also under- lies other studies of Brazilian export trade in soybeans, corn and COtUNL An interesting study by Turnovsky presented a model of aggregate import demand function for New Zealand over the post-war period (1947- 63) in order to derive estimates of various demand elasticities. He 33 criticizes the traditional approach in which researchers estimate import demand functions using single equation methods, and by regressing actual imports on explanatory variables--such as prices and domestic income-- and completely disregarding the export market. He first considers the relationships between imports and exports. He argues that even though capital inflows may finance a balance of trade deficit for long periods, from a long-run equilibrium point of view, a country can only purchase as much as it can afford. Hence, imports are recursively dependent on exports. He also recognizes the possibility of a simultaneous causality between imports and exports, but because New Zealand's share of world trade is small, its exports are not dependent on its imports, since the value of its imports represents a small part of its partner's budget needed to purchase its own exports. Moreover, because New Zealand is so distant from its overseas trading partners, the impact of export earn- ings on imports takes place only after some delay. Hence, the relation- ship between the country's imports and exports is both recursive and lagged. He also distinguished three concepts, namely the "desired" demand for imports, the "actual" demand for imports and the supply of imports. Desired demand for imports depends on variables determining how much pe0ple want to import; among such variables thought to reflect people's desire to purchase from abroad are prices, as well as asset levels. The actual demand for imports is introduced on the ground that people adjust their behavior to their desires only partially within the period, so that actual demand is more a short-term concept than desired demand even though they both depend basically on the same factors. On the other hand, he assumed the supply of imports to be dependent essen- tially on how much the country can afford to import, so that supply is 34 largely determined by foreign exchange availability to finance imports. He also notes the dual effect of foreign exchange on the country's trade behavior: on the demand side, foreign exchange availability reflect consumers' desire to purchase imports, while on the supply side, it re- flects the government's ability to finance the needed imports. To model the effect of government intervention, the author used as a proxy the country's net foreign exchange position, which is composed of export receipts and the country's foreign overseas assets. Obviously, as the writer notes, when net foreign exchange position is used as an instru- ment for government intervention, one is assuming that government inter- vention is carried out as a response to economic factors, and certainly, this can be misleading since other political factors can shape and even direct the government trade policy. A good example is provided by the frequent relaxation of import controls during election years. Another feature of the model is the mixed use of flow and stock concepts to trace out the stock adjustment mechanism in the process of the import demand. For further details on the flow-stock adjustment mechanism, see Nerlove. When estimating the reduced form equation of his model, Turnovsky ran into some problems due to the over-identification of the structural form of his model and the squeeze in the degrees of freedom caused by the use of too many variables for a relatively short time series. In addition, the need to satisfy all the constraints of the structural model brought about serious;computational problems, aggravated by the unavailability of a computer program to carry out the regression. In order to get around the identification problem and increase the degrees of freedom, the author abandoned the idea of stock-flow relationship in 35 the demand function. He finally used a hierarchy of models (actually three). The first was estimated by the Ordinary Least Squares (OLS), the second by Two-Stage Least Squares (TSLS) and the third by Three-Pass Least Squares (3PLS).2 The results provided by the three methods seemed to indicate a better performance of Model I, in spite of the simultan- eous equation biases that could arise from the use of OLS. Clearly a great merit of the author is his attempt to give an extensive coverage of many of the different factors that influence the import demand func- tion of many countries. In a comprehensive study of Egyptian wheat imports, Scobie used an approach very similar to the one just described above. Because Egyptian cotton production competes with the production of domestic wheat, he used what he called a "cotton export bloc" and “wheat import bloc." In the cotton export bloc, the two endogenous variables that determine foreign exchange earnings from cotton exports are taken to be the cotton export tax and the area sown. In the wheat import bloc, he used a ser- ies of equations to describe the domestic wheat demand and production, foreign aid and the formation of prices. This latter concept traces the effect of government policies on consumer prices and producer prices. Finally, he used another bloc called the balance of payments bloc. Us- ing the product of an interesting work by Hemphill, he states that the allocation of the foreign exchange budget by the government between classes of imports and reserve holdings by the central bank is a central pillar of his econometric model. He argues that Egypt, as well as many 23PLS was developed by Taylor and Wilson. It is used to estimate equations involving lagged variables and autocorrelated error terms. With this technique, consistent estimates of all regression coefficients are obtained, provided that all other explanatory variables are non- autocorrelated. 36 countries that follow a regime of a pegged exchange rate, are confronted with the problem of allocating their foreign exchange earnings between expenditures for imports and the need to satisfy a desired level of foreign exchange reserves. Therefore, while exchange receipts corre- spond to a budget constraint facing imports, fluctuations in the level of reserves make total import expenditures an endogenous variable. In other words, although he recognizes that the level of imports is con- strained by foreign exchange earnings between import expenditures and reserves, it is, in turn, endogenously determined through the balance of payments adjustment mechanism. He criticizes the "one sided use" of foreign exchange as one determinant of imports because he said, if im- port expenditures and reserve holdings are not determined independently, the estimated marginal propensity to spend foreign exchange on imports would be biased. The results of his model showed a low marginal pro- pensity to import food with respect to foreign exchange earnings. A shortfall in foreign exchange of one dollar would cause a reduction in expenditures on wheat of only $.O49. He argues that most of the adjust- ment that is needed when foreign exchange earnings decrease is provided through a squeeze of the imports of nonfood items, which presumably would cause disruption to the import of raw materials and capital goods. In addition, he describes the relationship between the cotton bloc and wheat import bloc in the following manner: a rise in foreign exchange receipts induces an increase in wheat consumption, mostly through im- ports and that tends to lower the domestic price of wheat, while the domestic cost of wheat production is higher than the world price of wheat. Because of the perceived external position, the real cotton price to producer is allowed to decline. This sets a second round of 37 reactions whereby cotton producers respond to their production dis- incentives by reducing planting for cotton in the next season so that the export earnings from which cotton eXports constitute a large part will decline, reducing the capacity to import. Domestic resources are thus encouraged to move into wheat production so that domestic wheat output will rise, reducing the need for imported wheat. Finally he notes that unlike in EEC countries where both protection and insulation policies have prevailed through the simultaneous use of the domestic farm price policy and the setting up of a variable levy on imports, in Egype, the price transmission elasticity (between world price and con- sumer price) was found to be .5. In other words, in spite of heavy government intervention, consumer prices have reflected the variations in the world price of wheat. In recommendation, the author concludes that a closer alignment of producer and consumer prices with the cost of imported wheat would relieve the budgetary pressure of the wheat sub- sidy for consumers and lessen its destabilizing impact on the importa- tion of other goods; moreover, a policy aimed at directing wheat sub- sidies to the poorest Egyptians would alleviate the social and economic effects of a wheat policy more aligned to the true costs of consuming additional wheat. To sum up, we attempted to present in this chapter the evolution of modeling approaches to import demand by LDCs. From relatively simple models in which only "traditional demand variables" such as population, income, prices, etc., are used, researchers have soon recognized the need to explicitly incorporate the effects of foreign exchange shortages in food imports. This has caused the widespread use of "import capacity variables" in food import demand functions in order to reflect foreign 38 exchange constraints faced by LDCs. Despite the merit of such a move, the resulting coefficient or elasticity of foreign exchange earnings with respect to imports cannot be fully used as a good basis for trade policy purposes. For example, a high foreign exchange elasticity with respect to food import does not tell more than that import levels are heavily influenced by foreign exchange earnings. The question of whether the government discourages the need to import food by raising domestic consumer prices or by raising the prices paid to farmers is not answered when government interventions are exogenously represented. The need to obtain a more accurate understanding of government policy in food import has then led to the building of models whereby government intervention affects endogenously the levels of imports. As presented above, this is the approach used by Bautista; Lattimore and Schuh; Turnovsky; and Scobie. A good description of this scheme is pre- sended by Abbott.3 For our econometric model in Chapter 5, the analyti- cal framework will be essentially based on this approach. 3For more detailed information about the approach presented by Abbott, see: Philip Abbott, "The Role of Government Interference in International Commodity Trade Models," American Journal of Agricultural Economics, 61, 1979, pp. 135-142. CHAPTER 4 THE ECONOMETRIC MODELING OF RICE IMPORT DEMAND IN SENEGAL Econometric theory is like an exquisitely balanced French recipe, spelling out precisely with how many turns to mix the sauce, how many carats of spice to add, and for how many milli- seconds to bake the mixture at exactly 474 degrees--but when the statistical cook turns to raw materials, he finds that hearts of cactus fruit are unavailable, so he substitutes chunks of cantaloupe; where the recipe calls for vermicelli, he uses shredded wheat.... - S. Valvanis (quoted from "A Guide to Econometrics," by Peter Kenedy) 4.1 Toward an Analytical Framework The framework used for the analysis of rice imports in Senegal is patterned after the work of Philip C. Abbott. A net import demand equa- tion will be estimated and will be defined as the gap existing between the supply of domestic rice and the level of the national demand for rice consumption. Because in Senegal imports of rice are a governmental affair, the effect of public intervention in the imports of rice will be explicitly recognized. Moreover, unlike other schemes in which public interferences are patterned directly in the import demand function through the use of a variable reflecting the country's "capacity to im- port," government's interference in rice imports are assumed here to en- dogenously affect the net import demand to be estimated. The process is described as follows: first, the import demand for rice is presented as the difference between two entities: the national demand for rice and the supply of domestic rice. Tha national demand for rice is pre- sented like a simple demand function in which variables such as 39 40 population, GNP, domestic consumer prices, appear in the right-hand side. For the supply of domestic rice, besides variables such as weather, the lagged value of domestic consumer price is used as a right- hand variable. The underlying rationale for the use of the consumer prices instead of the producer prices is that, as other studies have shown (see Tuluy and Sene), the sales of domestic rice are more respon- sive to consumer prices because most of the farmers' sales--particularly in the Southern region of Casamance--occur in private markets where con- sumer prices rather than producer prices are more likely to prevail. Available data suggest that, except in 1967, official sales of domestic rice (which are based on producer prices) have never reached 6 percent of domestic production (see Table 4.1). Instead, the gap between the production retained by farmers and their estimated consumption require- ments has averaged about at least 10 percent of total production. In other words, at least 10 percent of total production are sold without using official channels. 4.1.1 Effects of Government Intervention The approach used for the development of an import demand function for rice in Senegal considers the effects of government intervention in - rice trade as affecting endogenously the quantity of rice to be imported. Unlike alternative schemes whereby government policy variables such as foreign exchange earnings or "import capacity" are directly put together with "traditional" variables such as population, GNP, etc., in the demand function, our present approach represents the effects of government in- tervention in the following manner. 41 Table 4.1 Official Sales of Domestic Rice Official Sales of Domestic Rice Rice Production (% of Domestic Production) Year (Metric Tons) (Metric Tons) 1961 NA 1962 -- 1963 ~- 1964 -- 1965 109,000 3,173 (2.9%) 1966 125,000 6,269 (5.0%) 1967 125,000 9,642 (7.7%) 1968 135,000 7,215 (5.3%) 1969 59,000 38 (.06%) 1970 141,000 617.6 (.44%) 1971 99,000 599 (.6%) 1972 108,000 653 (.6%) 1973 44,000 427 (.1%) 1974 64,000 1,006 (1.6%) 1975 113,000 3,612 (3.19%) 1976 NA 1977 -- 1978 -- 1979 -- Source: CRED, University of Michigan, ONCAD. 42 4.1.1.1 Through World Prices The first way by which government interventions in the rice trade are provided is through public determination of domestic prices, given a level of world prices. It is assumed here that there is a price trans- mission mechanism between world prices and domestic ones. Since the~use of quotas for imports of rice has never been consistently observed, most of the government intervention in rice import could be expected to be carried out via the setting of domestic prices. Moreover, despite the government's attempts to carry out a consumer price stabilization policy for rice, available data suggest that domestic prices of rice have re- flected the ups and downs of world price movements. Nevertheless, do- mestic prices have not freely followed world prices so that domestic price adjustments to world ones reflect the extent to which government intervention in the trade of rice has taken place. An equation will be specified in which domestic prices are a func- tion of several variables, one of which is world prices. The magnitude of the coefficient of the world prices will reflect the extent of gov- ernment intervention. The closer the coefficient of world prices is to one, the more likley it is that free trade has been attempted. Con- versely, if government interventions are important, a strong divergence between world price movements and domestic ones would be expected to occur'and, thus, the coefficient of world prices would be close to zero. Note that the price transmission mechanism is one-sided, namely that because Senegal is a small country with respect to the world trade of rice, domestic prices cannot be expected to have any effect on world prices. In short, Senegal is a price-taker. 43 Changes in the domestic price of rice brought about by government intervention in reaction to changes in the world price affect the domestic demand for rice. Since quotas are not used, the extent to which government actions affect rice consumption will be closely re- lated to both incomes and domestic price elasticities of demand. On the other hand, the government intervention on trade through the price adjustment mechanism affects partly the supply of domestic rice. At least theoretically, the higher the domestic prices, the larger should be the amount of local rice sold by farmers so that, imported rice be- comes somewhat discouraged. 4.1.l.2 Through Foreign Exchange Effects As pointed out by Bautista, the direct use of an exogenous vari- able such as foreign exchange earnings in an import demand function is a mixture of an import demand function by consumers and an import pol- icy function by public policy makers. In this study, it is assumed that the effect of foreign exchange earnings on imports is an endogen- ous mechanism whereby a decrease in foreign exchange earnings should cause an autonomous increase in domestic prices, vis-a-vis world prices, so that rice consumption is discouraged and, thus, imports lowered. Meanwhile, the same decrease in foreign exchange earnings should, ceteris paribus, foster an increase in the sales of domestic rice. To sum up, the procedure used to pattern government intervention is as follows: government's interference is essentially reduced to the setting up of the domestic consumer price. This latter affects the level of rice consumption by consumers and the level of the supply of domestic rice. Together, these two components determine the level of imports. The process through which consumer prices are set is twaold. 44 There is a price transmission mechanism between world prices and domes- tic ones, through which the government indirectly determines the level of imports. On the other hand, the level of foreign exchange earnings1 reflect the government's willingness to import rice--as such, when ex- change earnings are low, domestic prices are presumably set high enough in order to discourage consumption of rice and, thus, to lessen the need to import rice. Thus, exchange earnings and world prices do not appear in the net import demand function to be estimated; those two variables are already reflected in the level of domestic prices. 4.1.2 Determination of the Net Import Demand Function The net import demand function to be estimated will be determined recursively. The model comprises three equations. The first equation presents the domestic consumer price of rice as a function of world prices and the level of foreign exchange earnings. In the second equa- tion, domestic production of rice is positioned as a function of the level of rainfall and domestic consumer price. Finally, the net import demand function to be estimated includes variables such as urban popula- tion, demand for consumption of rice, estimates of both domestic con- sumer price and rice production. The use of urban population instead of the total population is justified by the fact that, as mentioned earlier, the bulk of the consumption of rice occurs in urban areas. As for rural areas, it is only in Casamance--where most of the local rice is produced-~that the per capita consumption of rice is high. 1Foreign exchange earnings are defined as the value of the coun- try's total exports deflated by the index of import prices. 45 Moreover, since rice consumption is known to be essentially an urban phenomenon, the use of the country's GDP is not the most appropri- ate variable. Rather, an income variable for the urban citizens would be more relevant. Unfortunately, time series data on income for urban citizens are not available. Therefore, the use of the country's total GDP instead of a variable more representative of urban incomes would possibly yield an unreliable estimate of the income elasticity for the import of rice. Nevertheless, since a large part of the country's GDP comes from activities that are carried out in urban areas,2 the extent of the bias of the income elasticity may not be large. 4.2 An Econometric Model The econometric model for the Senegalese import rice demand func- tion is based on three features. First, government interventions are patterned through the setting of consumer prices. Domestic consumer prices are set as a function of world prices and foreign exchange earnings. For the latter, the lagged value of total exports deflated by the index of import prices is used as a proxy. For world prices of rice imported by Senegal, good data are clearly lacking. Although the bulk of the consumption of rice is composed of broken rice, Senegal imports different qualities of rice. Disaggregated data based on different qualities of rice are not knownto be available so that the use of one specific world price of rice would not be appropriate. Thus, we use the "implicit world price" of rice obtained by dividing the value of imports of rice by the corresponding 2Urban areas are defined as localities having more than 10,000 in- habitants. 46 quantities imported. Note that this proxy for world prices of rice underestimates the costs of imported rice since the costs of loading and unloading activities are not included. Since stabilization of domestic prices of rice has always beenaide- clared government objective, we include a third variable in the consumer price equation, namely, the lagged domestic price of rice. The magni- tude and the statistical significance of the coefficient of the latter variable can provide a clue in the assessment of the extent to which government's domestic price stabilization policy has been successful. Second, the level of rice production is determined essentially by weather variations and the annual rainfall level is used to pattern such variations. In addition, the level of the domestic consumer price of rice affects production decisions by farmers and/or the amount of rice commercialized by farmers for a given level of output. Because farmers' trading decisions are more influenced by domestic consumer prices than by the official producer prices and that agricultural1njces in Senegal are usually announced after planting, we use the lagged val- ue of consumer prices instead of the current ones. The third equation of the model is the total demand for consump- tion of rice. It is set as a function of urban population, the gross domestic product deflated by the consumer price index and the domestic consumer price. Finally, from the three equations described above, a derived net import demand can be determined as the difference between the demand for consumption of rice and the local production of rice. Table 4.2 presents the equations of the model. Table 4.3 describes the variable PD PD PW FE RC POP GDP IMP (2) (3) (4) Consumer Estimate Implicit Value of index. Domestic 47 Table 4.2 The Equations of the Model PDt = a0 + a] PDt-1 + a2 PWt + a3 FEt-l + E.l RPt = 80 + 81 Ht + 82 P0t_] + E2 RCt = 30 + a] Pot + 32 POP-+33 GDPt + E3 IMPt = 30 + a1 Pot + 32 POP + a3 oppt - RPt Table 4.3 Variable Notation price of rice (CFA francs/kilogram) deflated by CPI. of consumer price of rice from equation (1). world price of rice (CFA francs/kilogram) deflated by CPI. exports (millions of CFA) deflated by the import price consumption of rice. Urban population. Gross domestic product (billions of CFA) deflated by CPI. Average rainfall level in the country. Imports of rice. Local production of rice. Estimate of local production of rice. Time index. Time index lagged by one year. 48 names. Also, in Figure 4.1, a schematic representation of the model is provided. In equation (1), the coefficient a] should have a value between zero and one. A value of zero would indicate a complete failure of the government price stabilization policy, whereas a value close to one would mean that the government's price stabilization policy has been rather successful. Also, a2 should be between zero and one; a value close to one would indicate a free trade behavior or a constant per kg tax on rice imports, whereas a value close to zero would mean that domestic price variations are isolated from variation in world prices. As for the foreign exchange variable, the assumption that Senegal has faced foreign exchange constraints for its imports of rice leads us to predict a negative sign for the coefficient a3. The reason is that if, for example, imports of rice decrease due to a slump in foreign ex- change earnings, domestic consumer prices would have to increase in order to clear the market unless rationing--which is not used in Senega1--takes place. Moreover, stock movements which could prevent such an increase in domestic prices are not undertaken in Senegal due to the very low level of publicly held stocks of rice in the country. Thus, foreign exchange earnings and domestic prices are expected to vary inversely. In equation (2), both 81 and 82 are expected to have positive signs. Presumably, a high level of rainfall would, on the average, an- nounce good prospects for rice production. The statistical significance of the coefficient 3] would reflect the extent to which production lev- els are dependent on rainfall levels. As for the coefficient 82, tra- ditional economic theory suggests a positive relationship between 49 Figure 4.1 euem meem eeemeeoo r ~m>m4 rpmecmmm # meoucH Iii cowumpaaoa congah / .v we mueanH AJ mo xpaaam j. cowuaszmcou mama Ami! _ mmuwea emszmcoo owummeoo _ mmueem upeoz mmcwcemm mmcmguxm cmwmeou quoz oweumeocoom on» mo cowampcmmmeamm ovumssmemmwa 50 supply and prices. We anticipate that the effects of domestic prices on the supply of domestic rice would be, at best, marginal for two rea- sons. First, in Senegal, rice production is essentially a subsistence activity. Thus, prices have a limited effect on farmers' production decisions. Second, we mentioned earlier that most of the sales of rice by farmers occurs outside the official channels. Because such trans- actions are illegal, one should expect that farmers would minimize the risks involved and, thus, their incentives to produce rice for commer- cial purposes become inhibited. Nevertheless, the coefficient could indicate to what extent rice production is re5ponsive to price increases and, thus, what its impact is on import levels. For equations (3) and (4), as suggested by economic theory, the coefficients of urban population and gross domestic product should have positive signs, while that one for domestic consumer price should be negative. In the import demand function, equation (4), local produc- tion should have a negative coefficient since local rice is assumed to be a substitute for imported rice. Note that the use of rice produc- tion as a substitute for imports of rice is not very appropriate. Rather, the sale of local rice by farmers (after their subsistence needs are met) are more likely to affect the levels of imports. However, since no data series for the unofficial trade of local rice are avail- able, the production of local rice is used as a proxy for the sales of domestic rice. CHAPTER 5 RESULTS AND IMPLICATIONS The feeling that we have a better understanding of a social process--even though as is quite conceivable, it could become more unpredictable as a result--is sufficient justifica- tion for the enterprise, particularly if, as is often the case, new ways of influencing the process become available through that enhanced understanding. - Albert 0. Hirschman (quoted from Essa s in Trespassing, Cambridge University Press) 5.1 Results The model deve10ped in Chapter 4 is estimated using OLS for equa- tions (1) and (2). As for equation (3), the "Instrumental Variable Esti- mation Technique" is used; the procedure allows the equation to be re- cursively determined once the domestic consumer price is predetermined from equation (1). Equation (4) is obtained by a simple arithmetic dif- ference between equation (3) and equation (2). The empirical results of the model are presented in Table 5.1. Numbers in parentheses are t- statistics. Table 5.1 Empirical Results of the Econometric Model P0 = 22.06 + .075 PO _ + .8 PW - .00003 FE -1 Equation (1) t (3.7) (+.81) t l (9.2)t (-.83) t . A =- + . + . 5.880.121 RP, 1923; 19351 Pvt-1 1393,36": . RC = 11300 + .15 POP + 423.32 GDP - 2589.35 PO Equat'°" (3) t (.50) (3.09) (.40) (-1.27) t Equation (4) IMPt = 113000 + .15 POP + 423.32 GDP - 2589.35 P0, - 1 RPt 51 52 5.1.1 Empirical Estimates of the Equations The results of the estimation of the model are graphically presented on the following page. Figures 5.1.a and 5.1.b seem to indicate a rela- tively good fit for both equations (1) and (2) of the model. However, the performance of the model for equations (3) and (4) is less satis- factory. It is important to note here that since equation (4) is derived by subtracting equation (3) from equation (2), once the model has not performed very well in estimating rice consumption--that is, equation (3) --it could not be expected to be very successful in estimating rice im- ports from equation (4). Issues relating to the lack of performance of the model in estimating the last two equations are discussed in Appendix B of this paper. Now, we turn to the discussion of the results presented in Table 5.1. In equation (1), world prices are by far the most significant vari- able. They are statistically significant at a 99 percent confidence level. In contrast, export earnings are significant at less than an 80 percent confidence level, while the lagged consumer price variable is not significantly different from zero. For equation (2), unlike the lag- ged consumer price whose t-statistic is 0.23, weather (rainfall level)is very highly significant with a t-statistic of near 6.0. The latter equa- tion has a R2 of .70, while that one for equation (1) is .88. For equa- tion (3), urban population is significant at a 95 percent confidence level. The domestic consumer price is significant only at an 87 percent confidence level. With an estimated t-ratio of .40, the gross domestic product is not significantly different from zero. Furthermore, the pro- cedure used to estimate equation (3)--the Instrumental Variable :rxcor-o—z ZM‘O 0,0233”! ”"10 (020-1 DMZ-"'13 60 7O 53 DOMESTIC RETRIL PRICE OF RICE (1961-791 "‘*RCTUHL CONSUMER PRICE """ FITTED CONSUMER PRICE TIFWIFIIWIIIIITII 61 82 63 64 88 66 87 88 89 70 71 72 73 74 75 78 77 78 79 YEHRS Figure 5.1.a THOUSANDS RICE PRODUCTION ' (1961-19791 ""nCTURL PRODUCTION 160 - ----- FITTED PRODUCTION 140 120 100 1 1 1 I F j r I 1 1 1 1 1 r 1 1 T l 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 7B 79 YERRS Figure 5.1.b CDZO-O nun-1m: 54 THOUSBNDS RICE CONSUMPTION IN SENEGHL (1961-79) 600 460 400 350 300 250 ' 200 150 HRCTURL CONSUMPTION OF RICE """ FITTED CONSUMPTION DF RICE TIIIIIjIjTITIIIITI 61 62 63 64 65 66 67 68 69 7D 71 72 73 74 75 76 77 78 79 YEHRS Figure 5.1.c 55 Estimation Technique--does not report any R2 so that a numerical indica- tion for the fit of the equation is not provided. Finally, since equation (4) is not directly estimated, but rather is obtained from an arithmetic difference between equations (3) and (2), the goodness of its fit is very much influenced by the results of both equations (2) and (3). This relationship is well displayed by the simi- larity between Figures 5.1.c and 5.2. 5.1.2 Elasticity Estimates (at Mean Value) From equation (3) we can derive: Domestic price elasticity of rice consumption: 339*39=2589.35*——= 3P0 RC ’°38 Income elasticity of rice consumption: 205.39 _ 3110 * GDP _ 279185 ' BGDP RC = 423.32 * 31 From equation (4) we derive: Urban population elasticity of rice imports: 1201290 BIMP * POP " .15 * m7- =1.007 BPOP TMP” Rice production elasticity of rice imports: aIMP RP ._ 100368 _ -‘a—Rrr*TM—p”‘ *1788T7"-56 5.2 Implications of the Result In the presentation of the model in Chapter 4, public interventions in the rice sector were patterned through the government's domestic price determination mechanism. The latter was described as a response to changes in world prices and to fluctuations in the level of the coun- try's exports which are used as proxy for the country's foreign exchange «:04 nuaqna 56 THOUSRNDS RICE IMPORTS IN SENEDRL (1961-791 «m *“*RCTURL IMPORTS OF RICE (1961-79) °"“°ESTIMRTED IMPORTS DF RICE (1961-79) 3&1- 300 - O x “O 250 u '0’.‘. I .5 0" ‘0 0‘ 200 .. a" 7" ‘. '0' 150 .. .0. 3' o. .9. 'I.‘ o. . {I 9.. 0 a... '.”-....O 100 '8’ I T— T I I F 1 l I T T T 1 I I 1 I 1 61 62 63 64 65 66 67 66 69 7O 71 72 73 74 75 76 77 76 79 YERRS Figure 5.2 57 earnings. Certainly, the use of exports could underestimate the level of the country's foreign exchange since capital inflows are excluded. In the model presented here, it is assumed that capital movements are in the nature of financing or equilibrating transactions from the point of view of the balance of payments. In other words, the bulk of the capital transactions take place in order to help finance the country's balance of payments deficit, and, therefore, cannot be considered as a measure of the country's import capacity. Another reason for excluding such transactions as part of the country's import capacity arises from the difficulty of having consistent data on capital flows. As mentioned in Chapter 4, government actions in the rice sector are algebraically described in equation (1). Thus, the analysis of the results of equation (1) constitutes a good starting point from which the nature of the government's policy could be analyzed. From Table 5.1, the coefficients of lagged consumer prices, world prices and foreign exchange are +.O8, +.8 and -.00003, respectively. Since the coefficient of the lagged consumer price variable is smaller than .1, the model seems to indicate that the government price stabiliza- tion policy has clearly failed, even though in nominal terms, the re- verse seems to be the case. This observation is actually confirmed in Figure 5.1.a, which shows the fluctuating nature of the domestic con- sumer price. Nevertheless, it is to the credit of the government's sta- bilization policy that world price variations were higher than the domes- tic ones. The computed standard deviations for world prices and domestic prices are 8.7 and 7.4, with means of 28.9 and 39.6, respectively. For the price transmission mechanism between world prices and domes- tic ones, the coefficient of world price (.8) indicates that most of the 58 fluctuations in world prices were actually felt domestically. If world prices increase/decrease by l CFA franc per kg, domestic prices would be lifted/decreased by .8 CFA francs. Decreases in world prices are'then likely to favor the taxing of domestic consumers while world prices in- creases call for subsidies for imported rice. Again, Figure 5.3 shows the striking similarities between world and domestic price movements. Recall, however, that, as mentioned earlier, the data used for world prices do not include the cost of loading and unloading activities for imported rice. Should such costs be included, the resulting cost for imported rice would be much higher and, for periods in which domestic prices have been subsidized, the cost of imported rice would be higher than that of the consumer price of rice. For the effect of foreign exchange earnings on imports, the cor- responding elasticity--at mean values--is as low as .04. In other words, if foreign exchange earnings decrease by one billion CFA francs, imports of rice would decrease by only 82 tons as compared with an average im- port of rice of 178,817 tons! Of course, this is not to mean that for- eign exchange earnings are "abundant" in Senegal. Rather, imports of rice are given high priority for the country's allocation of import ex- penditures. For example, when the country experiences shortfalls on its foreign exchange earnings, either expenditures on other imported goods are reduced in order to forego a significant reduction on rice imports or the country's reserves of foreign currencies are drawn down or both. In the near future, it is very unlikely that the government would be able to pursue this above formula to preserve rice imports at very high levels. Already, the country's reserves of foreign currency have seri- ously deteriorated. In addition, a further squeezing up of other SSQOOP’NX ”NV “023”“ 3‘0 59 NORLD RNO DOMESTIC PRICES OF RICE (1961-79) 60 HNORLD PRICES OF RICE IN RERL TERMS .‘“.DOMESTIC PRICES OF RICE IN RERL TERMS 70- 1 1 1’ *1 1 1 1 1 1 l 1 I 1 1 1 1 1 1 BI 62 63 64 65 66 G7 68 69 7O 71 72 73 74 75 76 77 78 79 YERRS Figure 5.3 60 imported goods may be limited to a few items such as automobiles and other consumer goods since a minimum import of capital goods and raw materials will necessarily have to be met if future economic growth is to be satisfied. Thus, it is very likely that in the near future, the level of foreign exchange earnings will have a strong bearing on the amount of rice to be imported so that shortfalls in exports would cause sensible domestic price increases. In equation (3), the effects of urban population, GDP and domestic consumer prices on rice consumption are described. The empirical re- sults show that urban population is statistically the most significant variable and is followed by domestic consumer prices. The GOP is not statistically significant. This seems to indicate that the increase in rice consumption results more from the country's high urbanization rate than from an increase in consumer's income. Certainly, caution should be exercised in considering this statement. It was pointed out earlier that, on average, urban people tend to have higher incomes than their rural counterparts. Statistically, that means that the GDP and the urban population are likely to vary closely together and the more it is so,the less efficient would be the coefficients of those two variables.1 This calles for further research on consumer expenditure patterns, where- by expenditures on rice consumption by urban people would be compared with those of rural pe0p1e. The different consumer groups to be compar- ed would belong to the same income groups. The results of such a re- search would give more reliable indication as to where future government policy would focus in order to decrease the growth of imports of rice. 1Table 5.2 (on the following page) shows a correlation of .4 between real GDP and urban population. 61 Table 5.2 Correlation Matrix POPULA GNP PD RICECON POPULA 1.00000 .406378 .128099E-01 .665300 GNP .406378 1.00000 .309748 .263957 PO .128099E-01 .309748 1.00000 -.189809 RICECON .665300 .263957 -.189809 1.00000 RICEPRO .96886lE-Ol .263948 -.305067 .556636 MILPR -.762357 -.451745 - 122950 -.314651 PRIRAT .288491 .41292 .916909 -.675722E-Ol l 2 3 4 RICEPRO MILPR PRIRAT POPULA .96886lE-01 -.762357 .288491 GNP .263948 -.451745 .412922 PD -.305067 -.122950 .916909 RICECON .556636 -.314651 -.675722E-Ol RICEPRO 1.00000 -.l44l76 -.222709 MILPR -.144176 1.00000 -.502580 PRIRAT -.222709 -.502580 1.00000 5 6 7 where: RICECON = demand for consumption of rice; RICEPRO = production of local rice; MILPR = producer price of millet (in real terms); PRIRATa = price ratio between consumer prices of rice and producer prices of millet.P aNote the negative correlations between millet prices and the demand for consumption of rice. This seems to indicate the fact that even though millet has potential as a substitute for rice, for the period covered by the study, a decrease in millet prices does not appear to favor a decrease in consumption of rice. bData on consumer prices of millet are not available. However, the pro- ducer prices of millet are certainly a good proxy for the former. 62 Nevertheless, the result of the model indicates an urban population elasticity with respect to imports of about 1.007. Since the urban popu- lation has been growing at an estimated rate of 4.8 percent, imports on average have been growing at a rate of 4.8 x 1.007 = 4.83 percent. Therefore, it appears that if government's concern is to decrease the growth rate of imports of rice, much attention should be given to the urban population variable. At this point, two questions should be asked with respect to the urban population. First, why is the rate of urbanization so high as com- pared to the average growth rate of the country's population and what is needed to slow it down? Second, why urban people seem more inclined to consume rice than their rural counterparts? The answers to these two questions are inevitable parts of the tools policy makers should use if long-term food self-sufficiency is to be achieved. With respect to the first question, the data show that the rate of urbanization was highest during the 19705 when real rural income de- creased by 1 percent a year.2 That decrease in rural incomes was largely due to shortfalls in peanut production during the drought coupled with decreases in the real prices of agricultural products offered to farmers. Thus, economic difficulties in the rural sector have favored urban migra— tion. In addition, phosphate revenues in the mid 19705 stimulated the government capacity to expand its economic participation policy which has partially boosted activities in the secondary and tertiary sector.3 This further stimulation of urban based activities increased migrants' 2See the Senegalese Sixth Development Plan. 3The secondary sector includes all manufacturing activities. By tertiary sector, we mean service activities. 63 expected incomes in urban areas. However, this does not mean to imply that urban migration should be stopped by drastic measures such as an erection of tight controls against migrants. The apparent solution lies in a shift for a development strategy more oriented towards the agri- cultural sector. If phosphates and other government revenues had been put back into agriculture, the development prospects of the rural areas would have been more encouraging and the rate or urban migration would have certainly decreased. For example, in the Casamance region, de— creases in rainfall have caused heavy salt water incursion into land growing areas during the late 19705. This situation has likely reduced rice production potentials in that region, while, if a small dam had been built there, that particular problem might have been lessened. In the meantime, however, one possible way out of this rice import induced urban population is to focus on the variables that actually seem to favor the urban consumer's biases toward rice consumption. Actually, many hypotheses have been advanced, some of which relating rice consump- tion to the fact that the staple takes less time to cook, it is easily available, etc. Others even went as far as saying that rice consumption is associated with a feeling of a higher social status. As plausible as some of these arguments may sound, very few of them are actually backed with reliable empirical evidence. By not, it seems very unlikely that income levels are a major cause for the continuous increase in rice con- sumption. A 1976 study by FAO based on cross-section data found an in- come elasticity for consumption of rice of .3, the same as the one esti- mated here. Such a result indicates that, unlike what is commonly as- sessed, rice is not a luxury good in the Senegalese diet and that other traditional goods could well be substituted for it. Current knowledge 64 indicates that local cereals (millet, sorghum, corn) are much less a- vailable in urban areas than in rural areas. Now the question is: Is the growth of urban consumption of rice due to the lack of availability of local cereals or rather, local cereals are not available in urban areas simply because consumers are biased against them? Clearly, the answer to the question should not be in the form of "either/or;" rather, if substitutes are assumed to potentially exist for rice, local cereals would not only have to be available in urban areas, but also would be processed in such a way as to meet consumer preferences. This raises the question of the development of technological packages to increase agricultural productivity, the deve10pment of marketing facilities and the processing of local cereals. Again, the undertaking of these enor- mous tasks cannot be made overnight through crash program. This re- quires a new approach of economic development similar to the "New Strat- 4 This calls for a egy of Growth" advocated by John Mellor for India. long-term strategy and the intersector linkages must be carefully mas- tered before the launching of the approach. For example, in the late 19705, faced with large stocks of millet, the government embarked into a program of partial substitution of millet flour for wheat flour in the production of bread. The program came out unsuccessful mainly because the cost of millet flour exceeded that of wheat flour and, thus, the final cost of bread was higher. Had a more appropriate trade policy been undertaken, better farming practices and a more efficient marketing policy been introduced, the outcome would likely have been more satis- factory. Mountains cannot be displaced just by faith. As the Guinean 4See Bibliography. 65 leader SecKou Toure eloquently put it, "Economic development is not a 100 meter race--rather it is a real marathon.“ These above mentioned policies, particularly the promotion of local cereals, have become, in the past few years, a common topic for policy makers and are strongly reassessed in the Senegalese Sixth Development Plan. It is, therefore, h0ped that such a government concern would bring about careful research in the field of farming systems, marketing and trade policies. The third explanatory variable in equation (3), namely domestic con- sumer prices, is statistically significant at only 85 percent. The com- puted elasticity of domestic prices to rice consumption is "less" than -.4. This finding appears to indicate that pricing policy alone is un- likely to be successful as to slow down the growth of consumption of rice. In other words, if all other variables are held constant, main- taining rice consumption at a minimum level would require large increas- es in the domestic prices. Given the fact that expenditures on rice oc- cupy a large portion of urban consumers' budget, particularly among the lower income consumers, the political pressures that militate against such policy should not be overlooked. This latter remark is clearly a reassessment of the idea that, in order to tackle the steady increase in rice consumption, a broad framework is needed whereby all the different policy directions should be integrated. Equation (4) presents the import demand for rice as the excess de- mand for rice over local production. This is graphically shown in Figure 5.4. As mentioned earlier, even though price increases can re- duce import needs by encouraging supply of local rice and by discourag- ing the demand for rice, the effects of lifting prices are likely to be limited, due to the low price elasticity of demand for rice. Prices C I . Domes ~Consu Price l—v—u—n- tic mar 66 Demand Representation in the Rice Sector Supply of Domestic Rice Imports of Rice Demand for Consumption 012 Figure 5.4 Quantities 67 Thus, if imports are to be discouraged, attention should be focus- ed on two directions. 0n the one hand, the demand for consumption of rice can be moved leftward. The means to be used to achieve such a goal are the policy instruments already described in this section and include the promotion of local cereals, the improvement of marketing facilities and more committed research activities to increase productivity at the farmer level. On the other hand, the reduction of the import gap could be met by moving the supply schedule of rice rightward. Since the sup- ply of rice is presented in equation (2), the analysis of the statisti- cal results of that equation becomes at this point quite warranted. From Table 5.1, the results for equation (2) show that the level of rainfall is by far the most significant variable. Prices appear to have almost no statistical significance. In the Casamance region where most of the local rice is produced, important inputs such as fertilizer are very marginally used. This attests to the nature of rice production in Senegal as being very primary. The lack of significance of the price variable should not be inter- preted as a lack of responsiveness to price increases on the part of farmers. If we suppose, for example, that farmers sell 15 percent of their production, then an increase in prices by 10 percent would in- crease their real income by only 1.5 percent. Thus, it seems virtually certain that for the case of rice producing farmers in which a small part of production is marketed, cultivators' incomes will, on the aver- age, move directly with production changes. Incidentally, technological change provides its benefits in proportion to total production not in proportion to marketings and, thus, benefits the small farmers, as well as larger ones. 68 In the meantime, however, the pricing policy in the rice sector re- mains to be improved. The data we used show a .6 correlation coeffi- cient between peanut prices and producer prices of rice, while the cor- relation coefficient between the latter and the consumers' price of rice is .27. Not only that, but also the direct sales of local rice by farm- ers (using consumer prices) are prohibited by law. This latter situa- tion constitutes an important disincentive for the promotion of local rice. Agricultural agencies are given legal monopoly over the process- ing of the paddy produced. Since large and inefficient mills are used by the agencies, their use has produced an unnecessary increase in the cost of domestic rice. Unfortunately, it is such costs that, in part, determine the producer price to be paid to farmers.5 In spite of all the disincentives encountered by the rice produc- tion sector, the rice production elasticity computed from the results of equation (4) is -.56. This result seems to indicate that the local pro- duction of rice has a nonnegligible effect on imports. A 1 percent in- crease in local production causes a decrease in import by .56 percent. Furthermore, the competitiveness of local rice against imported rice is enhanced by the former's quality premium over the imported one, which is essentially broken. Finally, it is important to note that, given the way the import demand equation is defined, we are implicitly assuming that bgfg§g_or- ders for imports of rice are established, account is fully taken on the level of local production. Only a thorough documentation about the ad- ministrative procedures of imports of rice could shed light on whether 5In setting producer prices, marketing and processing costs are taken as given and are deducted from consumer prices; also, other ad- justments, not clearly defined, are made. 69 such an assumption is realistic or not. If it turns out that the as- sumption is not plausible, then the computed production elasticity with respect to import could be biased upward. 5.3 Projection of Demand and Import of Rice Using the results of the model, we now make a medium- and long-term projection for imports of rice in Senegal. To do that, we need to make projections for both the demand for consumption and the local production of rice. For the former, we will use the "FAO Projection Methodology." The corresponding formula for the procedure is as follows: _ t Dt - Do (1-Fd) where: 00 = base period; Dt = year for which the projection will be made; 0. 11 population growth + (income elasticity *income growth rate). For the local production of rice, we refrain from making any pre- diction. The reason is that, as many researchers seem to agree, many variables that affect the supply of agricultural commodities cannot be confidently forecasted, particularly in the case of rice in Senegal. For example, variables such as technological progress, rainfall level, the implementation rate of the rice irrigation projects, etc., do not usually provide clues from which an accurate prediction could be made for them. Nevertheless, government and many other agencies publish fig- ures on medium- and long-term demand and supply estimates for certain agricultural commodities. We will present two projection figures; one is for the import of rice in 1985 and the other is a projection of demand for rice in 1990. 70 The figure for the import of rice in 1985 would be obtained by the dif- ference between our projected demand for 1985 and the estimate of rice production for 1985 as presented in the Senegalese Sixth Development Plan. V As for the year 1990, a figure for imports of rice could not be pre- sented because we do not have access to any estimate of local production of rice for 1990. Our projection estimate will then be limited to only a consumption figure for 1990. However, on the basis of such a figure, we would be able to draw recommendations as for directions through which the need for imports of rice could be lessened. Table 5.3 presents the results of the projection estimates. Our projection estimates of rice imports in 1985 is very close to the figure presented in the Sixth Development Plan. It seems, however, that the projected figure for rice production in 1985 is likely to be too optimistic. Past experiences show that projected figures for rice production during the Third, Fourth and Fifth Deve10pment Plan have been always off the actual production and were overestimated by at least 50 percent. The 1985 figure for rice production represents an 11.3 percent growth in rice production based on an average production level of 100,600 tons during the 19705. We certainly do not mean to be too pessimistic with respect to the growth of rice production that is implicit from the Sixth Plan's figure of 191,000 tons. Rather, we feel more concerned about the costs of an underestimation of the country's needs for food imports. Rice production could certainly be increased at very high levels. However, our view is that, unfortunately, the time constraint is usually underestimated. It is hard to conceive that an extensive study of 71 Table 5.3 Projection Estimates of Rice Consumption Year Demand for Consumption of Rice Imports of Rice 6th Development Plan's Estimate Our Estimate 6th Development Our Estimatea Plan's Estimate 1985 535,256 tons ‘--— 531,000 tons 320,445 tons 1990 712,416 tons -—--#b-—-- aFor the estimation, we use an urban population growth rate of 4.8 per- cent which is the observed growth rate during the two decades covered by the data used, i.e., 1961-79. This income elasticity (.31) is given by our model. For the GDP, we use the figure forecasted in the 6th Development Plan for 1981-85, namely a 3.5 percent growth rate. Finally, our model estimate of rice consumption in 1979 is used as a base year consumption, and is equal to 379,800 tons. farmers' constraints, the finding and introduction of relevant tech- nological packages, their full implementation, the setting up of rele- vant price, marketing and storage policies, etc., could be carried out within a decade. The reduction of the country's rice import gap through domestic production increase is a topic more relevant for the next decade. For example, from Table 5.3, one can see that if rice imports for 1990 are to be limited to the 1985 levels, then local production would have to reach 391,971 tons, that is at a rate of almost 15.5 percent a year from an estimate of 191,000 tons in 1985. 72 Thus, it seems that, in the short- and medium-term, the best pros- pects for reversing the growth in rice imports lie in the development of substitutes for rice in the urban consumers' diet. So far, the results of our model show that substitutes for rice, even if they potentially exist, have been nevertheless very weak. The price and income elastici- ties computed from our model were respectively -.38 and .31. Using the Homogeneity Condition or the Slutsky-Schultz Relation,6 the cross-elas- ticities (between rice and other substitutes and complements) would be about .07. This means that an increase in the price of rice by l per- cent would, on average, increase the consumption of substitutes, say, sorghum-millet, by .07 percent. Whether such a low response on the part of urban consumers is due to a lack of availability of local cereals in urban areas or to the fact that local cereals do not fit consumers' tastes is clearly an empirical question. Nevertheless, this opens up a need for extensive research in order to promote the industrial proces- sing of local cereals to stimulate productivity increase for local cereals at the farm level. 6The Slutsky-Schultz condition states that the sum of the own and cross-price elasticities and the income elasticity for a commodity is zero. This means that the income effect of an own-price change must be consistent with the cross- and income-elasticities for the commodity. See Tomek and Robinson in Agricultural Product Prices, page 30. CHAPTER 6 SUMMARY AND CONCLUSIONS Much of the work of economists is concerned with the future, with forecasts and planning. But forecasts are triv- ial and planning is useless unless they are based on fact; and facts which are at our disposal are facts of the past.... The purpose of analysis, applied to those facts, is the ex- planation of what has happened--the explanation that is, of economic history.... So even if our business is with fore- casts of what is likely to happen or with probable results of policies to be adopted now, historical analysis comes first. - Sir John Hicks, in Causality in Economics In this study, an attempt is made to further our understanding about the nature and cause of the increasing demand for import of rice in Senegal. The demand for rice imports is presented as the gap between the demand for consumption of rice and the supply of domestic rice. Furthermore, close consideration is given to the effect of government policies, particularly in the rice sector. An econometric model was used to study the behavior of the relationships involved; the findings of the model are summarized as follows. 6.1 Demand for Consumption of Rice The model showed that most of the increase in rice consumption ob- served during the last two decades are, to a large extent, caused by the growth in urban population. Urban population is, by far, the most sig- nificant variable in the demand for consumption of rice equation. In contrast, the effects of incomes and prices were marginal. Moreover, these two variables have low elasticities with respect to the demand for 73 74 consumption of rice (.31 and -.38, respectively). The elasticities of urban p0pulation with respect to the demand for consumption of rice and the demand for imports of rice were found to be .65 and 1.004, respec- tively. The latter result indicates that, unless new actions are taken, in the near future rice imports could grow at a rate of approximately 4 to 5 percent, the same rate the urban population growth experienced dur- ing the 19705. Consequently, focusing on income and price variables to slow the demand for consumption of rice does not seem to be a viable solution. Price and income elasticities of demand are so low that only a large increase in the former and/or a large decrease in the latter could cause the demand for rice to decrease significantly. Therefore, at least from a short-term perspective, incomes and domestic prices, as well as the urban population, offer little success when used as tools to slow the demand for consumption of rice and the need to import rice. 6.2 Supply of Domestic Rice The results of the model have shown that the level of rainfall is by far the most significant variable in the determination of the supply of domestic rice. Even though a positive relationship is observed between domestic prices and the supply of local rice, statistically, the result is not significantly different from zero. This appears to corroborate the fact that rice production in Senegal is essentially at the subsistence levels, in spite of government's efforts to make the activity more commercialized. So far, rice production is mostly a prim- ary activity; that is, inputs such as fertilizer, pesticides, as well as improved tools are not significantly used. Nevertheless, potentials for improving both rainfed and irrigated rice production are not lacking. But in the short-run, prospects for a large increase in rice production 75 .aapah coma: .< an newcomoga a? apnea mvgh "oucsom mm.¢1 ~e.o- =o_uqe:mcou ese: mcvucaog use: use: pumuafiwmweh so.~ _~.¢m Lexus .mpmccagu puvuwmeo ppvz bypass «use; xucmm< uwpasa covuerF 3.: 3.3” .8553: .3255 2.8.5:. 2;. £33.. «83 .353 £33 . pasta BEE. mm.o~ mo.~m coFugsamcou use: . acwucaom can: . mcoz p~.~- . sm.em Lexus .mpmccmgu —a_8vCCo ppwz cepaam amend aocma< uwpnam maEmzm mo.~ em.em eegu=P=me~ .m_eeeeeu peeeeeeo ._Fz 8_Pe=a eases sueeo< ee_e=a .eeeeeu Laue: no.op No.mm covuae=m=ou use: m:_u=:oa u:~:. meoz ew.~p- ~o.mp Cosmo .mpmccugu pevu_»eo ppwz uvpnae waged. zucmm< u—Fnam mm.m- ~o.m_ segueeseeN .mpeeeegu Peeueeeo PFC: ue_e=a maths seeem<,ee_e=s ee_¢ esezm No.2. No.3 .8553: .mpoccgu 3.6.85 8.3:: :95 3:55 BEES 863.5 ce.mp mo.op :ovunssmcou use: mcvuczos ace: ocoz. menu~39wem< cmucrmm mp.oo- m—.¢ guano .mpuccogu pavuvweo ppwz uCFaae mags; aucmm< uwpnsa nuances ~o.~m- mp.¢ m_:og .um .mpocemgu PopuPuwo pppz uvpnaa omen; aucmo< uPPaaa anosom\mupmo mm.~m- mm.sp Lexus .mpoccmsu pmpumuwo ppm: uppnaa vocab zucmm< u¢_a=a uumaoga _u.m~- mm.~+ n.384 .um .mpmccmgu pmvuwweo .FF: uwpnae omen; zucmm< uwpaze ummmpz om.o mm.ep newnessmcou wee: mcwucaoa use: . meoz . m¢.~P- ~¢.- Lexus .m_mecm;o Pawuwmwo ppvz UCFnae magma xucmm< u¢_a=a meme—o: mm.mF- ~¢.- mesa; .em .mpaeeeeu Fewueeeo .PCz 8.Fe=a page; seeem< ue_e=a mae_pe> seen: m~.e~ m~.F~ cowunE:mcou wee: mcwucaos one: mcoz mesupsupgm< emummwggn auvpvna xuw_vna emucmu new: macpcgumh eavuuoppou ozawcsuoh -arwoee luveoea cowuanweumvo m:_PPCz covuusuoea _..eem eez oua>_ea be: —mmmcmm cw mmacwcgumh cowauzuoem muwm F.m anmp 76 are very slim. The arguments that increases in price offered to farm- ers will boost significantly the domestic supply of rice appear to be misleading. As John Mellor1 nicely puts it, "It is important to keep perspective on the role of price policies.... They are basically a pal- liative, designed to lessen the unpleasant symptoms of an underlying problem, while constructive policies for fostering a stream of technolog- ical change can be brought into play and have their effect." Thus, if the reduction of the need to import rice is to be undertaken through an increase in local production, an important element is to increase agri- cultural productivity. This calls for long-term agricultural research to improve farming practices, to make available relevant technological packages, etc. Not least important is the need for better management of LDAs which supervise a large number of rice production schemes. In the medium term, however, giving more priority to rainfed agri- culture for rice production--particularly in the Casamance region-~cou1d yield good results. Presently, seven rice production techniques exist in Senegal (see Table 6.1), out of which three are irrigated schemes. Using both private and social profitability measures, the rainfed schemes appear to perform better. This implies that more careful consideration should be given to the allocation of scarce public financial resources between rainfed and ir- rigated agriculture for rice. Despite the fact that the irrigated schemes in the North could provide greater employment opportunities than rainfed cultivation in the Casamance region, from an economic standpoint, the returns from irrigated rice do not appear to be very attractive. 1John Mellor: "The Functions of Agricultural Prices in Economic Deve10pment," Indian Journal of Agricultural Economics, 1968, pp. 23- 37. 77 Because,at least from a long-run point of view, a country can under- take only what it can afford, the social benefits accruing from increas- ed employment possibilities due to the irrigation schemes in the North should be evaluated in light of the economic returns of those projects. 6.3 Government Policies In the econometric model presented in Chapter 4, the effects of government intervention in the rice sector were patterned through the public determination of domestic prices. The results of the model show that for imported rice, most of the variation in world prices were transmitted domestically. Moreover, in real terms, the government's domestic price stabilization policy has not been successful. Also, the model has shown the foreign exchange demand elasticity of the imports of rice as low as .04. This means that a 1 percent decrease in the coun- try's export earnings would lower the imports of rice by only .04 per- cent. Thus, in case of large shortfalls in the country's foreign ex- change earnings, a relatively high priority is given to the imports or rice, possibly at the expense of other capital goods whose imports are crucial for a sustained economic growth. However, enough evidence is not found for the government's bias against the imports of capital goods, 50 that the argument that the government's import policy has hindered economic growth cannot be strongly assessed. From the results of our econometric model, the projected demand for consumption of rice in 1990 is as high as 712,416 tons. Should the government's policy in the rice sector change, this figure could be sig- nificantly reduced. In the short and medium terms, policies in the rice sector should concentrate on the promotion of potential substitutes for rice, such as 78 1nillet-sorghum and corn. Presently, the cost of imported rice for urban consumers seems to be below that of couscous.2 A study carried out by FAO in 1976 in Senegal3 had shown that 1,000 calories derived from con- sumption of broken rice and couscous cost to the urban consumers 28 CFA and 42 CFA, respectively. Moreover, the study reported that 10 grams of protein from consumption of rice and of couscous would cost 10.2 and 16.7 CFA, respectively. These findings could be a major factor explain- ing the presumed urban consumers' bias in favor of rice consumption. So far, there is no evidence that differences in taste between rice and couscous have a significant impact on consumers' preferences. Thus, the promotion of millet-sorghum and corn can have great prospects as sub- stitutes for rice. This calls for research to increase productivity at the farm level and improve marketing facilities for local cereals. As part of the policy, means should be found whereby farmers' incomes could be stabilized as much as possible.4 It is interesting to note that in most LDCs, the question of guaranteeing a minimum level of income to farmers is hardly raised; however, in international debates, one major revendication of LDCs' representatives is the stabilization (or a min- imum guarantee) of their country's export earnings! From a long-term perspective, it is likely that a gradual improve- ment of the agricultural sector could largely slow the current rates of urban migration which, from the results of our model, is the major cause of the high growth of imports of rice. 2Couscous is a traditional consumer good derived from the processhw; of millet-sorghum or corn. 3FAO, Rapport de Mission, Rome, September/October 1976. .4See David Trechter on problems and issues of programs designed to stab111ze farmers' incomes. APPENDIX A 79 Table A.l Rice Production, Rice Consumption, Imports of Rice, and Rainfall Level Rice Rice Imports Rainfall Production Consumption of Rice Level Year (Metric Tons) (Metric Tons) (Metric Tons) (Centimeters) 1960 1961 68,000 177,785 109,785 95 1962 84,000 202,137 118,137 78 1963 90,000 190,770 100,770 87 1964 106,000 290,490 184,490 86 1965 109,000 288,221 179,221 106 1966 125,000 284,288 159,288 114 1967 125,000 278,438 153,438 102 1968 135,000 320,161 185,161 100 1969 59,000 204,901 145,901 55 1970 141,000 260,237 119,237 110 1971 99,000 286,510 187,510 73 1972 108,000 277,905 169,905 90 1973 44,000 235,968 191,968 55 1974 64,000 271,195 207,195 57 1975 113,000 215,125 102,125 80 1976 116,000 360,508 244,508 60 1977 118,000 366,018 248,018 65 1978 63,000 301,996 238,996 50 1979 140,000 491,860 351,860 90 Source: Direction de la Statistique, Ministry of Plan, Central Bank, Ministry of Finances. 80 Table A.2 Price Indexes, Nominal Producer Prices of Rice and Peanut, Consumer Price of Rice, and World Price of Rice Producer Price Producer Price Consumer Import of Rice of Peanut Year Price Index Price Index (CFA/kg) (CFA/kg) 1960 .84 18 22 1961 .86 .47 18 22 1962 .90 .48 18 22 1963 .92 .47 20 21.5 1964 .96 .46 20 21.5 1965 .98 .45 21 20.58 1966 1.01 .51 21 20.58 1967 1.00 .54 21 21.68 1968 1.01 .42 21 17.60 1969 1.03 .39 21 17.95 1970 1.07 .37 21 18.40 1971 1.11 .43 21 21.17 1972 1.18 .49 21 23.1 1973 1.32 .64 25 23.1 1974 1.54 .84 30 25.5 1975 2.01 1.00 41.5 41.5 1976 2.07 .96 41.5 41.5 1977 2.28 1.00 41.5 41.5 1978 2.35 1.13 41.5 41.5 1979 2.57 1.22 41.5 41.5 Source: Service de la Statistique generale. Direction de la Statis- tique. Situation Economique du Senegal. For the import price index, data are taken from IMF, "International Financial Statistics, Yearbook, 1982.“ 81 Table A.3 Consumer Price of Rice, Implicit World Price of Rice, Gross Domestic Product (GDP), and Exports. (All Figures are in Nominal Terms) Value Consumer of Imports Implicit Price of of Rice World Price GDP Exports Rice (Millions of of Rice (Billions (Billions Year (CFA/kg) CFA) (CFA/kg) CFA) CFA) 1960 43.160 1961 30 2,697 24.56 153.58 39.97 1962 30 2,949 24.96 162.37 40.74 1963 35 2,750 27.28 175.23 29.61 1964 35 4,920 26.66 187.99 32.57 1965 40 4,476 24.97 192.31 33.81 1966 40 4,331 27.20 206.56 36.80 1967 45 5,512 35.92 200.83 33.90 1968 45 7,048 38.06 216.53 37.40 1969 45 4,674 32.03 217.36 31.90 1970 40 3,335 27.97 240.10 42.20 1971 40 4,639 27.74 247.20 34.70 1972 40 4,252 '25.02 273.60 54.40 1973 60 9,519 49.59 278.20 43.20 1974 100 18,032 87.03 338.80 94.00 1975 80 6,050 59.24 406.40 99.10 1976 80 10,676 43.66 459.30 122.80 1977 80 11,263 45.41 480.90 163.90 1978 80 13,153 55.03 438.10 105.00 1979 80 16,451 46.75 511.50 133.40 Source: Direction de la Statistique (Senegal). Comptes Economiques. BCEAD. 82 Table A.4 Population, Millet Production, and Millet Prices Producer Price Millet Urban Total of Millet Production Year Population Population (CFA/kg) (Metric Tons) 1960 1961 747,270 3,143,000 15 392,000 1962 764,980 3,211,000 16 407,000 1963 782,000 3,279,000 16 424,000 1964 911,000 3,351,000 16 478,000 1965 951,000 3,500,000 17 531,000 1966 993,000 3,576,000 17 554,000 1967 1,034,000 3,656,000 17 423,000 1968 1,082,000 3,738,000 17 655,000 1969 1,130,000 3,822,000 18 450,000 1970 1,180,000 3,956,616 17 635,000 1971 1,232,000 4,045,615 18 401,000 1972 1,286,000 4,137,168 17 583,000 1973 1,342,000 4,222,803 22 323,000 1974 1,401,000 4,430,700 22 609,000 1975 1,463,110 4,956,518 26 703,000 1976 1,527,480 5,085,388 30 621,000 1977 1,594,470 5,217,608 35 507,000 1978 1,664,860 5,381,000 35 421,000 1979 1,738,360 5,518,000 40 803,000 Source: Direction de la Statistique, Ministry of Plan. APPENDIX B 83 APPENDIX B B. Some Data Problems Uma Lele, in her study, "Food Security for Developing Countries," found a correlation coefficient of less than .5 between two data series that presumably report the same information! There is no.1 ggiggi reason to believe that data from the East African countries of her study are less accurate than data from West African nations. There- fore, the findings of this study are likely to be affected by a similar lack of accurate data. For this reason, the results should be inter— preted as reflecting Senegalese rice import policy derived from avail- able data. In this study, two major problems were encountered with respect to the data used. 8.1 Incomplete Data In Chapter V it was mentioned that the domestic consumer prices and the value of exports are deflated by the CPI and the import price index, respectively. For both deflators, the data series are not avail- able for 1961/67. For the CPI, two data series are published in Senegal. One for "European consumers" and another for "African consumers." The latter was used as a deflator in this study. The CPI for "European consumers" has a more complete data series. To derive CPI estimates for "African 84 consumers" between 1961/67, a regression was run between the two CPI for 1967/79, the CPI for "African consumers“ being the dependent variable. Using the actual values of CPI for “European consumers" between 1961/67 and the coefficients of the regression equation, estimates of CPI for “African consumers" are derived for 1961/67. For the index of import prices, a five-year moving average was used to substitute for the missing data. For example, the estimate of the import price index for 1965 was obtained by using the average import price index for the period 1966/1970. Furthermore, lack of complete data precluded the use of the country's 1 Instead, the GDP was used; because GNP as proxy for urban incomes. the latter includes incomes of foreigners, the effect of urban incomes on rice consumption might have been underestimated. Finally, the data on rice imports are not disaggregated to account for food aid. If the aid component on rice imports is important, then the coefficient estimates in the rice import equation could be biased. 8.2 Inaccurate Data Because there is not a strong basis to believe that one particular source is more reliable than another, the main criterion for choosing among data sources was to select the most complete data series. The reason is that if, for example, two or more series are mixed, the result- ing series is likely to be less accurate than each of its components. In other words, if one series is inaccurate, it might have more chance to be consistently inaccurate than a mixture of different series. 1Recall that urban areas are defined as localities having more than 10,000 inhabitants. It is certainly not clear that in all localities that satisfy this definition, rice consumption patterns are relatively important. BIBLIOGRAPHY Abbott, Philip C. 1979. "Modeling International Grain Trade with Gov- ernment Controlled Markets." American Journal of Agricultural Economics, 61. Baustista, R.M. 1978. “Import Demand in a Small Country with Trade Restrictions." Oxford Economic Papers, 30, pp. 199-216. 1978. Import Demand in a Small Country. Bawden, D.L. 1964. "A Spatial Equilibrium Model of International Trade." Journal of Farm Economics, 46, pp. 1372-1379. Carmichael, J.S. 1974. "Rice: World Situation and Outlook." Canadian Farm Economics, 4, p. 1. Club du Sahel. 1982. "Food Self-Sufficiency and Ecological Balance in the Sahel Countries." Paris, France, May. CRED-University of Michigan. 1977. "Marketing, Price Policy and Stor- age of Food Grains in the Sahel: A Survey." de Janvery, Alain. 1981. The Agrarian Qgestion and Reformism in Latin America. Johns Hopkins Press. Outta, M. 1964. "A Prototype Model of India's Foreign Sector." Inter- national Economic Review, 5, pp. 82-103. FAO. 1976. Rapport de Mission--Travaux Grogpe Mixe SenegallFAO. Rome, September/October. Farnsworth, Helen. 1958. "International Wheat Agreements and Problems, 1949-56." Quarterly Journal of Economics, 66, May, pp. 217-248. Graven, K. and A.H. Tuluy. 1981. Rice Policy in Seneggl. Stanford University Press. Hassler, James B. 1964. "An Appraisal of Spatial Studies Related to Agriculture." Journal of Farm Economics, 46, pp. 1380-1386. Hemphill, W.H. 1974. "The Effects of Foreign Exchange Receipts on Im- ports of Less Developed Countries." International Monetary Fund Staff Papers, 21, 637-677. 85 86 Hirschman, 0. Albert. 1981. Essaysgin Trespassing: Economics to Politics and Beyond. 1958. The Strategy of Economic Develggment. Yale Univer- sity Press. Ibrahima, Sene. 1980. "Farmers' Behavior Towards New Technology: The Senegalese Case." Michigan State University. Islam, 8. 1978. "Price, Income and Foreign Exchange Reserve Elasticity for Asian Rice Imports." American Journal of Agricultural Eco- nomics, 60, pp. 532-535. Jeanneret, Charles A. 1981. "L'harmonisation de la Strategie du CILSS! Club du Sahel avec 1es Plans de Developmegt Nationaux des Pays Saheliens est--elle Possible?" Universiteld'Ottawa, Canada, April. Joquero, Duff, Anden-Laesina and Hayami. 1975. "Marketable Surplus Function for a Subsistence Crop: Rice in the Philippines." American Journal of Agricultural Economics, 57, pp. 705-709. Krishna, Ray and A. Chhibber. . Policy Modeling of a Dual Grain Market: The Case of Wheat in India. IFPRI, Washington, D.C. Labonne, Michel and Bruno Legagneuse. 1977. Reflexions sur 1' Agri- culture Senegalaise. INRA, Montpellier, November. Lattimore, R. and G. Edward Schuh. 1979. "Endogenous Policy Determina- tion: The Case of the Brazilian Beef Sector." Canadian Journal of Agricultural Economics, 27, pp. 1-16. Lele, Uma and Wilfred Candler. 1981. Food Security for Developing Countries. Alberto Valdes, ed. Manderscheid, Lester. 1964. "Some Observations in Interpreting Measur- ed Demand Elasticities." Journal of Farm Economics, 46, p. 128. McCalla, Alex. 1966. "A Duopoly Model of World Wheat Pricing. Journal of Farm Economics, 48, pp. 711-727. Mellor, John. 1968. "The Functions of Agricultural Prices in Economic Development." Indian Journal of Agricultural Economics, pp. 23-27. . 1976. The New Economics of Growth--A Strategy for India and the Developing World} Cornell University Press. Mendelsohn, C. 1957. "Approaches to International Trade Under Non- price Competition." Journal of Farm Economics, 39, pp. 1724-1731. Nerlove, M. 1958. "Distributed Lags and Demand Analysis for Agri- cultural and Other Commodities." Washington, D.C. 87 Ridler, Ducan and Christopher A. Yandle. 1972. "Changes in Patterns and Policies in the International Trade in Rice." IMF Staff Paper , 19, pp. 46-86. Scobie, Grant M. 1981. "Government Policy and Food Imports: The Case of Wheat in Egypt." International Food Policy_Research Institute Research Report No. 29, Washington, D.C. Seaver, Stanley K. 1964. "Spatial Research--Measurement for Wheat." Journal of Farm Economics, 46, p. 1365. Senegal. 1977. Fifth Develgpment Plan: 1977-81. 1982. Sixth Develgpment Plan: 1981-85. Tomek, W.G. and K.L. Robinson. 1981. Agricultural Product Prices. Second Edition, Cornell University Press. Trechter, David. 1983. "The American Farm Problem." Mimeograph, Michigan State University. Tuluy, A. Hasan. 1981. Costs and Incentives in Rice Production in Senegal. Stanford University Press. Turnovsky, Stephen J. 1968. "International Trading Relations for a Small Country: The Case of New Zealand." Canadian Journal of Economics, 1, pp. 772-790. Wallace, Thomas and G.G. Judge. 1963. "The General Problem of Spatial Equilibrium: A Methodological Issue." Paper presented at the Interregional Competition Research Methods Workshop, Raleigh, North Carolina, January. Weisskoff, Richard. 1979. "Trade, Protection and Import Elasticities for Brazil." Review of Economics and Statistics, 61, pp. 58-66. World Bank. 1974. Senegal: Tradition, Diversification and Economic Development. 1117117117! (1.1.1111. III. [II Ill ‘1 El: 5'. MICHIGA 11 11111111 3129300 U NIV. LIBRRRIES llllllllWilfllllllllllll 0810899 pr .__.- .....__.._.e ‘