World Bank Reprint Series: Number Eighty-two Carlo Cappi, Lehman Fletcher, Roger Norton, Carlos Pomareda, and Molly Wainier A Model of Agricultural Production and Trade in Central America Reprinted from Economiic Integration in Cenitral America, eds. William R. Cline and Enrique Delgado (Washington, D.C.: Brookings Institution, 1978), pp. 317-70 and 647-61 CHAPTER 7 A Model of Agricultural Production and Trade in Central America Carlo Cappi, Lehman Fletcher, Roger Norton Carlos Pomareda, and Molly Wainer A. Introduction Research reported earlier in this book supports the hypothesis that eco- nomic integration was an important factor in the substantial increase in growth rates that occurred in the Central American Common Market (CACM) countries compared with their growth rates prior to integration. Inte- gration is estimated to have raised gross domestic product for the re- gion by 3 to 4 percent above levels that would have been ach!.eved in the absence of the Common Market. The gains from integration came from two major sources. First, production grew from the stimulus of importing from regional partners goods that otherwise would have been imported from outside the region. Second, a larger market encouraged increased invest- ment that further stimulated growth via the multiplier effect. The gains from integration, however, were found to be much higher in industry than in agriculture. In agriculture, the gains stemmed essentially from im- provements in efficiency through freeing trade internally in the CACM. The principal effect was the increased importation of rice and beans in- to Honduras, Nicaragua, and Costa Rica from El Salvador. This produced gains from trade-creation for the region as well as direct foreign ex- change gains for El Salvador. In the case of industry, the intercountry distribution of gains from integration has been definitely skewed. In particular, the results of other chapters show that the country with the smallest industrial base, Honduras, has had relatively slight gains from integration. These con- clusions reinfo-ce questions about the overall role of the agricultural NOTE: This chapter is based on the results of a collaborative research effort involving the Secretaria de Integracion Econ6mica de Cen- troamdrica (SIECA), the Brookings Institution, the Development Research Center of the World Bank, and FAO. Other cor.tributors to this research were Ivan Garcir%, Carlos Selva, Arnaldo Gomez, and Willy Flores--all of SIECA--and Richard Inman, Vihn Le-Si, and Scott Sirles of the World Bank. Opinions expressed here are those of the authors alone and do not reflect the official views of the sponsoring institutions. 318 sector in influencing the distribution of benefits from further integra- tion in Central America. It may well be the case that allowing intraregional agricultural trade to conform more closely to patterns of comparative a dvantaqe would give greater benefits to the poorer countries. But this hypothesis has yet to be testedz if it prnves to be the case, there are corollary ques- tions concerning the costs of incegration and who would pay them. In order to address these questions, we have constructed a linked five-country agriculturalmodel (MOCA) which describes conditions of pro- duction, demand, trade, and price and income formation, in each of the countries. It encompasses twenty-three agricultural products, processed and unprocessed, and production possibilities are specified for three farm-size groups in each country. The model was designed for approximate simulation of the actual be- havior ot markets subject to certain kinds of government policy Lnter- ventions, Its structure is based on the assumptions that (i) prices dif- fer within and between countries but are responsive to changes in aggre- gate supply and demand; (ii) farmers and marketing firms individually are price takers; tiii) the farmers with the smallest plots tend to give priority to home retentions of their maize production: and (iv) when farmers market their produce, they attempt to maximize profits.1 Trans- portation and processing activities, and their costs, are included in MOCA, as are subsidies, tariffs, and import quotas. In its mathematical structure, MOCA is an optimization model in the style of the Mexican model CHAC;2 maximization in the model is a device to insure that the solution represents the specified kind of market equi- librium. Policy goals are not maximized directly, but rather the model may be solved repeatedly under different values of policy instruments. In atteinptingto represent both supply and demand conditions in some detail, MOCA is designed to capture the main determinants of comparative advantage in the five countries. Solutions are conducted under alterna- tive assumptions about the level of institutional barriers to trade; and the responses to changes in those barriers demonstrate the basic patterns of comparative advantage. But more than tracing out comparative advantage--which is an effi- ciency consideration--MOCA has been used to explore the multiple conse- quences of steps toward greater econosmc integration. Increased trade in agricultural products will make some groups of consumers better off and others worse off. The same holds for producers, in the arjyreqate, by country, and by farm-size class. It is these price and income con- sequences--the incidence of trade impacts--that OCA is designed to analyze. While the model results presented here are provocative, and on the whole plausible, we would liketo issue a precautionary warning that they should not be taken too literally. The present version of MOCA was de- signed explicitly as a demonstration model; the work program of SIECA calls for carrying out considerable refinement of th)is version before utilizing MOCA as a policy model. TZhe main limitation of the demonstraLlfn 319 version is that product supply possibilities are treated in rather ag- gregate fashion within each country (compared to, say, the supply speci- fication inCHAC). The current set of supply activities will be replaced with a more detailed one, while retaining the existing overall framework (and matrix generating computer program) for MOCA. Given these considerations, we would consider the present results from MOCA to be only indicative. In the aggregate, the qualitative re- sults probably are reliable, particularly as they address the possibili- ties of small changes from the observed values. But we would not like to impute an unwarranted degree of exactitude to the results. The remainder of this chapter is organized as follows: section B provides some of the relevant historical background; section C describes the structure of the model; section D further discussek, the model and the processes of validating it; and section E gives the principal numerical results. Full details of model structure are given in the appendix. B. Recent History in Numbers 1. Agricultural Production in Central America During 1960 to 1974, total agricultural production in Central Amer- ica (including Panama) increased at an average annual rate of 3.9 percent (table 1). El Salvador experienced the lowest growth rate of 3 percent per year. All other countries achieved growth rates of 4 percent or bet- ter per year over this fifteen-year period. But output growth was much less impressive in percapita terms. Due to population growth rates averag- ing around 3 percent per year, per .capita agricultural production in- creased only about 1 percent per year for all of Central America. The dispersion of agricultural output growth rates by country has been larger since 1970 than it was in the 1960s. The average annual changes in agricultural sector output for 1970-74, measured at factor cost, are given in table 2. Honduras experienced the lowest growth in this recent period, while both Guatemala and Nicaragua increased produc- tion more rapidly than other countries and more rapidly than in the 1960s. Table 2 also compares agricultural output growth to overall growth in gross domestic product (GDP) for the Central American countries in 1970-74. While the dispersion of overall growth in the countries has been less than for agriculture, the overall growth rate is clearly asso- ciated with growth in agricultural output. This tendency reflects both the absolute importance of the agricultural sector and the export-oriented growth pattern of the Central American countries. Traditional agricul- tural products still account for the bulk of exports. Some selected statistics on the structure of the Central American economies are shown in table 3. Agricultural outpiit accounted for slight- ly less than a third of total output for the CACM countries in 1974. And, with the exception of Costa Rica, well over half of the popula- tion is considered rural. This means that output per person is much less in agriculture than in the other sectors. Output per person in agriculture in 1974, as a percentage of output per person in the nonagricultural economy, Table 1. Indices of Total Agricultural Production by Country, Central America, 1960-74a (1961-65 = 100) -.Z.-owth Rates Country 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1960-74 Costa Rica 102 101 103 98 93 103 114 120 124 136 141 143 153 165 155 4.1 El Salvador 81 92 948 105 110 96 100 106 104 114 115 131 119 130 142 3.0 Guatemala 77 83 96 104 103 114 102 114 118 115 122 142 149 163 165 4.8 Hondurds 85 96 97 98 99 109 115 124 129 125 125 141 136 141 139 4.4 Nicaragua 62 76 89 101 121 113 118 116 115 110 118 135 130 153 137 4.3 Panama 81 92 91 95 104 116 118 124 141 152 144 153 155 145 14G 4.3 Central America 80! 89 96 101 106 108 109 116 119 121 125 139 139 150 148 3.9 aLatin American regional producer price weights. SOURCE: Agriculture in the Americas: Stetistical Data, FOCD working paper, Economic Research Service, U.S. Department or Agriculture, April 1976. 321 Table 2. Growth in Gross Domestic Product and Agricultural Sector Product,a Central America, 1970-74 Gross Domestic Agricultural Sector Country Product Product (annual percentage change) Costa Rica 5.9 4.1 El Salvador 5.4 3.9 Guatemala 6.3 7.1 honduras 3.0 1.6 Nicaragua 4.9 6.0 aAt factor cost. SOURCE: CEPAL, on the basis of official country statistics. ranged from a low of 25.6 percent in Honduras to 41.8 percent in Costa Rica. Moreover, the countries with the largest share of population in agriculture have the smallest relative per capita output in that sector. Agricultural production for export continues to represent a large proportion of output in that sector inCentral America. In 1974, the per- centage of production for export relative to total agricultural output was: Country Percent El Salvador 50.5 Costa Rica 48.6 Guatemala 46.5 Nicaragua 41.7 Honduras 29.6 2. Evolution of Trade During the 1963-65 to 1970-72 Period The basic conditions of trade are specified in the General Treaty of Economic Integration of Central America (1960), which established the Centra' American Common Market. The agreement posited free trade for products oricinating in the region, with some important limitations (see Anexo A of tI.e Tratado de Integracion Economica de Centroamerica). A number of products were excepted from coverage under the 1960 Treaty for a transitory period of five years. In particular, imports and 322 Table 3. Agricultural Output, Agrticultural Population, and Output per Person, Cenvral America, 1974 Agricultural Agricultural Output/ Agri cultural Population as Person as Percent Output as Percent Percent of Total of Nonagricultural Country of Total GDP Population Output/Persona Costa Rica 23.4 42.2 41.8 El Salvador 28.9 55.0 33.2 Guatemala 31.0 61.1 28.7 Honduras 32.6 65.4 25.6 Nicaragua 27.3 53.7 32.5 Central America 28.8 57.3 30.2 amore precisely, agricultural output per person in rural areas, divided by nonagricultural output per person in nonrural areas. SOURCE: CEPAL/FAO. exports of basic grains were controlled in all countries under this clause. Later, in 1965, the five countries signed the Convenio de Limon to codify the regulation of trade in basic grains (defined to be maize, rice, sorghusi, and beans). Export to countries outside the region was to be authorized only in the event that no buyers within the region could be found. There have been subsequent periodic revisions of the trade agreements, but in practice quota arrangements have prevailed, often for- mally justified by plant sanitation requirements. In spite of the patchwork character of tro.de-liberalization steps, trade in aqricultural products has grown faster than incomes in the re- gion. During the 1963-72 decade, agricultural e:cports expanded at a rate of 6.7 percent per annum,3 and total exports grew by 8.3 percent annual- ly. Growth rates varied substantially by country, as shown in table 4. Agricultural exports represented approxirnately 80 percent of total ex- ports in the CACM in the period 1963-65, but because of their slaver growth compared with that of total exports, their share within the total declined over the following decade in every country (table 4). Agricul- tural imports, as a share of total imports, also declined for every coun- try during the period under consideration. 'Table 5 gives the value of agricultural exports for the periods 1963-65 and 1970-72. In spite of agriculture's declining share within total trade, it must be pointed Out that the export markets are the most dynamic source of demand for thO! region's agricultural products. Table 4. Structure and Growth of Trade, Central America, 1963-65 to 1970-724 Rates of Growth, 1963-65 to 1970-72 Percentage Share of Agriculture to Total Agricul- Agricul- Total tural Total tural Exports Exports Imports Imports Country Exports Exports Imports Imports 1963-65 1970-72 1963-65 1970-72 Guatemala 8.7a 6.2 ... 2.0 80.6 68.1 19.2 14.5 El Salvador 5.0 2.7 ... -0.3 77.6 66.5 23.9 17.2 HGnduras 9.0 8.4 ... 8.5 76.0 73.5 15.5 14.4 Nicaragua 7.1 5.6 ... 2.2 79.8 72.2 19.3 14.6 Costa Rica 12.6 11.7 ... 12.5 82.9 78.4 16.3 15.8 CACH 8.3 6.7 7.9 ... 79.4 71.5 19.2 15.3 aExports and imports measured as three-year averages of values in current U.S. dollars. SOURCE: SIECA. 324 Table 5. TradLtional and Nontraditional Agricultural Exports, As a Percentage of Total Agricultural Exports, Central America Traditional Agricultural Nontraditional Exports Agricultural Exports 8 of Total % of Total Current Agricultural Current Agricultural Country Years US$1,000 Exports US$1,000 Exports Guatemala 1963-65 126,990 94.3 7,676 5.7 1970-72 175,778 85.9 28,854 14.1 El Salvador 1963-65 127,057 94.3 7,665 5.7 1970-72 145,670 89.7 16,923 10.4 Honduras 1963-65 65,332 86.1 10,556 13.9 1970-72 123,309 92.3 10,350 7.7 Nicaragua 1963-65 89,278 88.1 12,097 11.9 1970-72 123,525 83.4 24,632 16.6 Costa Rica 1963-65 176,429 94.2 5,144 5.8 1970-72 176,429 91.6 16,173 8.4 CACM 1963-65 492,174 91.9 43,138 8a1 1970-72 745,211 88.5 96,431 11.5 SOURCE: SIECA. 3. Traditional Versus Nontraditional Exports ConsideringCentral America as a whole, traditional exports--coffee, cotton, sugar, bananas, and beef--constituted 73 percenL of total exports and 92 percent of agricultural exports during 1963-65. This picture changed little between 1963-65 and 1970-72; at the end of this perio3, exports of traditional agricultural commodities represented 89 percent of agricultural exports and 63 percent of total exports. (See tables 6 and 7.) Exports of coffee and cotton alone accounted for 70 to 90 percent of agricultural exports inGuaten3ala, El Salvador, and Nicaragua in 1963- 65, while coffee and bananas accounted for 70 to 80 percent of agricul- tural exports in Honduras and Costa Rica in the same period. The situ- ation remained essen -ially the same at the end of the period, with a slight tendency for -he shares of coffee and cotton exports to decrease and the shares of sugar and beef exports to increase. Exports of beef 325 Table 6. Traditional and Nontraditional Agricultural Exports, As a Percentage of Total Exports, Central America Traditional Nontraditional Country Years Agricultural Exports Agricultural Exports (percent of total exports) Guatemala 1963-65 75.9 4.6 1970-72 58.5 9.6 El Saivador 1963-65 73.2 4.4 1970-72 59.5 6.9 Honuras 1963-65 65.4 10.6 1970-72 67.8 5.7 Nicaragua 1963-65 70.3 9.5 1970-72 60.2 12.0 Coota Rica 1963-65 78.1 4.8 1970-72 71.8 6.6 CACM 1963-65 73.0 6.4 1970-72 63.3 8.2 SOURCE: SIECA. became fairly important for Nicaragua, where they accounted for 21 per- cent of agricultural exports in 1970-72. Banana exports also showed an increase in most countries during the period of the study, particularly in Honduras and Costa Rica (table 7). In conclusion, it is evident that traditional agricultural exports still constitute the bulk of total exports in the CACM. The slight trend of a declining contribution of traditional products to total exports has not been strong enotigh to change the basic structure of trade in these countries. When the data are put in terms of growth rates, however, it can be seen that the nontraditional exports are now much more dynamic than the traditional ones, with the exception, as noted, of Honduras (table 8). 4. Self-Sufficiency in Basic G1,ains Domestic production of basic grains--maize, beans, rice, sorghum-- did not grow very rapidly during the decade covered. In a few countries, production of grains diminished over the period. This happened in Hon- '4ur4i with production of beans and sorghum; and in Costa Rica with 326 Table 7. Traditional Agricultural Exports, As a Share of Each Country's Total Agricultural Exports, Central America Product Guatenmala El Salvador Honduras Nicaragua Costa Rica Average, 1963-65 Coffee 59.37 65.07 23. 39 21.34 52.87 Banana 4.83 0.02 53.30 1.42 29.73 Sugar 4.65 1.72 0.13 5.66 5.59 Beef 3.14 0.04 3.78 7.44 5.38 Cotton 22.31 27.46 5.49 52.20 0.63 Total 94.30 94.31 86.09 88.06 94.20 Average, 1970-72 Coffee 49.35 63.21 18.87 21.25 36.39 Banana 7.67 - 62.09 0.82 37.30 Sugar 5.81 7.04 1.21 8.25 6.26 Beef 7.75 1.02 9.51 21.04 11,56 Cotton 15.33 18.62 0.58 32.01 0.09 Total 85.91 89.89 92.26 83.43 91.60 SOURCE: SIECA production of maize, beans, and rice (table 9). Regarding the degree of self-sufficiency in grains, the following conclusions can be drawn: (i) Overall, CACM is effectively self-sufficient in the produc- tion of food grains (with some year-to-year fluctuations). (ii) El Salvador shows an increasing degree of self-sufficiency over time; except for beans, El Salvador is now a net supplier of basic grains. (iii) Guatemala was self-sufficient in rice and sorghum at the beginning of the period, but it turned into a deficit country by the end of the period. (iv) Honduras shows a trend of diminishing surplus in basic grains, but it is still self-sufficient in maize and beans. Honduras and Nicaraqua are the main suppliers of beans in the area. (v) Nicaragua has shown a slight improvement in all self-suffi- ciency indexes, particularly in beans and rice where it has generated surpluses in recent years. 327 Table 8. Average Annual Growth Rate of Traditional and Nontraditional Exports, Central America, 1963-65 to 1970-72 Traditional Nontraditional Country Agriculture Agriculture (percent per year ) Guatemala 4.75 20.82 El Salvador 1.97 11.98 Honduras 9.50 -0.28 Nicaragua 4.75 10.69 Costa Rica 11.2s 17.78 CACM 6.11 12.18 SOURCE: SIECA. (vi) Costa Rica is the only country which is experiencing a real deterioration in domestic production of basic grains. Costa Ri'a was in a deficit position in all grains at the end of the period. (vii) Given that growth in grains production is slight, it is ques- tionable whether the rapid increases in grains exports can be sustained for very long (table 10). So far, grains exports represent a miniscule portion of grains availability, but that portion is growing. 5. Trade in Agricultural Inputs Agricultural inputs still figure importantly in agricultural and total imports of the region. They accounted for 43 percent of agricultural imports of the whole area in the initial period, and 37 percent of ag- ricultural imports at the end of the period. With respect to total Imports, agricultural inputs accounted for8 percent in 1963-65 and for 5 percent in 1970-72, for the region as a whole. These imports experienced slower growth than imports of agricultural products, but they still re- main a major import category. Within agricultural inputs, fertilizers and other chemical products account for around 80 percent of total im- ports of agricultural inputs. 6. Relative Prices In spite of a decade and a half of expansion of intraregional trade, the potential for further trade remrains substantLtl. Perhaps the most direct measure of this potential is the dispersion of agriculturalprices among the five countries. significant price differentials are signals that there could be overall gains if the products were to be shipped from Table 9. Volume of Production and Rate of Growth of Production of Basic Grains, Central America, 1963-65 to 1970-72 Production Guatemala El Salvador Honduras Nicaragua Costa Rica 5 CACM (production in metric tons) Maize 1963-65 625,900 200,500 316,900 157,300 72,100 1,372,700 1970-72 778,400 325,700 309,900 199,700 35,800 1,649,500 Rate of growth (t) 3.17 7.18 -0.32 3.47 -9.53 2.66 Beans 1963-65 55,200 18,100 50,100 44,800 16,300 184,500 1970-72 62,900 30,600 47,200 44,300 7,600 188,100 Rate of growth (%) 1.88 7.80 -2.26 -0.16 -10.34 0.27 Rice 1963-65 17,000 29,500 10,300 51,100 72,100 180,100 1970-72 32,700 35,000 10,500 89,100 65,300 237,000 Rate of growth (%) 11.85 2.47 0.28 8.28 -1.41 4.00 Sorghum 1963-65 33,160 94,900 52,700 44,900 10,300 234,400 1970-72 35,500 149,800 74,600 41,600 11,400 285,900 Rate of growth (%) 1.68 6.75 -1.44 -1.09 1.46 2.88 SOURCE: SIECA, Series Estadisticas Seleccionadas de Centroamerica y Panama, December 1975. 329 Table 10. Value of Nontraditional Exports: Grains and Other Products, Central America, 1963-65 and 1970-72 Other Nontraditional Country Years Grains Products Total (thousand U.S. S) Guatemala 1963-65 277 7,399 7,676 1970-72 1,514 27,340 28.854 El Salvador 1963-65 693 6,972 7,665 1970-72 2,912 14,011 16,923 Honduras 1963-65 6,823 3,733 10,556 1970-72 2,869 7,481 10,350 Nicaragua 1963-65 956 11,141 12,097 1970-72 5,740 18,892 24,632 Costa Rica 1963-65 126 5,018 5,144 1970-72 77 16,096 16,173 Central America 1963-65 8,877 34,261 43,138 1970-72 13,112 83,319 96,431 SOURCE: SIECA the low-price country to the high-price country. Tables 11 to 15 show the 1970 or 1973 relative prices for five principal products of the Central American countries. Transportation costs no doubt account for some of the observed dif- ferentials, but those cost. should account for no more than 5 to 10 per- cent of product value, so there obviously remain unexploited trading opportunities. The tables show that, altnough the patterns vary by crop, Guatemala and Honduras are low-cost producers rather consistently while Costa Rica is, on the whole, the highest-cost producer. These differ- ences foreshadow to some extent the possibilities of trade expansion which are traced out with MOCN in section E below. 7. Labor Force and Employment In a sense, there is more pressure for labor migration than there is for movement of agricultural products among Central American countries. but, in actuality, while there is substantial seasonal and permanent mtqration within each country, very little international migration has t-en permitted. 330 Table 11. Relative Consumer Prices, Maize, 1973 Guatemala El Salvador Honduras Nicaragua Cost* Rica Guatemala 1.000 0.791 0.722 1.053 1.148 El Salvador 1.264 1.000 0.912 1.330 1.451 Honduras 1.386 1.096 1.000 1.458 1.590 Nicaragua 0.950 0.752 0.686 1.000 1.091 .4 Costa Rica 0.871 0.689 0.629 0.917 1.000 NOTE: Relative prices are computed as the column country's price divided by the row country's price. SOURCE: SIECA/Brookings Consumer Price Survey, 1973. Table 12. Relative Consumer Prices, Rice, 1973 Guatemala El Salvador Honduras Nicaragua Costa Rica Guatemala 1.000 0.963 0.773 0.898 1.027 El Salvador 1.039 1.000 0.803 0.933 1.067 Honduras 1.294 1.246 1.000 1.162 1.329 Nicaragua 1.113 1.072 0.861 1.000 1.143 Costa Rica 0.974 0.937 0.752 0.875 1.000 NOTE: Price relatives are computed as the column country's price divided by the row country's price. SOURCE: See table 11. 331 Table 13. Relative Consumer Prices, Beans, 1973 Guatemala El Salvador Honduras Nicaragua Costa Rica Guatemala 1.000 1.222 0.788 1.044 1.128 El Salvador 0.818 1.000 0.645 0.854 0.923 Honduras 1.269 1.551 1.000 1.325 1.432 Nicaragua 0.958 1.171 0.755 1.000 1.081 Costa Rica 0.887 1.084 0.698 0.925 1.000 NOTE: Relative prices are computed as the column country's price divided by the row country's price. SOURCE: See table 11. Table 14. Relative Prices at the Farm Gate Level, Sugar Cane, 1970 .uatemala El Salvador Honduras Nicaragua Costa Rica Guatemala 1.000 1.025 0.790 0.667 1.123 El Salvador 0.976 1.000 0.771 0.651 1.096 Honduras 1e266 1.297 1.000 0.844 1.422 Nicaragua 1.500 1.537 1.185 1.000 1.685 Costa Rica 0.890 0.912 0.703 0.593 1.000 NOTE: Price relatives are computed as the column country's price divided by the row country's price. SOURCE: GAFICA, "Plan Perspectivo para el Desarrollo y la Integracion de la Agricultura Centroamericana," vol 2, (Guatemala City: GAFICA, May 1974), tables H.1.1-H.1.31. 332 Table 15. Relative Prices at the Farm Gate Level, Coffee, 1970 Guatemala El Salvador Honduras Nicaragua Costa Rica Guatemala 1.000 1.420 0.899 0.992 1.203 El Salvador 0.704 1.000 0.633 0.698 0.847 Honduras 1.113 1.581 1.000 1.103 1.339 Nicaragua 1.008 1.432 0.906 1.000 1.213 Costa Rica 0.831 1.181 0.747 0.824 1.000 NOTE: Price relatives are computed as the column country's price divided by the row country's price. SOURCE: See table 14. Table 16 shows the rural population density in each couintry; the figures speak eloquently of the demographic tensions which contributed to the recent war between Honduras and El Salvador. International political circumstances are such that it is not pos- sible to contemplate significant movements of agricultural labor between countries inCentral America. The MOCA model does allow for intracountry labor movements, but to conform to reality it does not allow for move- ments across international boundaries. It does of course allow for in- ternational movement of products and, given contemporary realities, the more pertinent question is, To what extent will movement of products compensate for lack of factor mobility? Table 16. Rural Population per Hundred Acres of Land in Farms, Central America, 1965 and 1970 % Annual Rate of Country 1965 1970 Change Guatemala 93 98 1.1 El Salvador 122 130 1.3 Honduras 68 76 2.2 Nicaragua 29 31 1.3 Costa Rica 38 43 2.5 Central America 64 69 1.5 333 C. MOCA: A Spatial Equilibrium Model 1. Introduction to the Model In broad terms, the principal objective in building MOCA was to con- struct an analytic instrument that could be used to simulate the conse- quences of changes in Central American agricultural trade policies. "Pol- icies" chiefly means quotas (formal or informal), tariffs, and export subsidies, and "consequences" refers to changes in trade patterns, pro- duction levels and prices by commodity and country, incomes by farm group and country, and levels of employment and the use of other factors. More narrowly, the initialuses of MOCA have concerned an assessment of the incidenoe of benefits and costs of more liberalized agricultural trade among the CACM countries. The model solutions reported later in this chapter contain measures of the magnitude of gains and losses by country and by groups within each country. While the overall gains from expanded trade are positive, it is clear that not everyone benefits. In- deed, exactly this perception has hindered the full implementation of the freer trade regime which is codified in the CACM agreements. Although fears of unequal incidence of economic costs have been widespread, there has been no frank attempt to set out what those costs, and the corresponding gains, might be, and who will incur the costs and reap the gains. MOCA was constructed to help fill this void. A clearer appreciation of the incidence of benefits and costs perhaps can contrib- ute to the design of compensatory programs, thereby promoting expanded trade. A byproduct of this study is a quantification of patterns of com- parative advantage, over products and countries in the region. MOCA is a modified market equilibrium model, and as such it operates on the basis of comparative advantage at the margin. This last qtualification is im- portant; we consider that MOCA's validity is restricted to cases of mar- ginal change away from the observed situation. MOCA is based on cross-sectional data within each country. Supply elastics are not estimated independently and then put into the model, but rather are derived from the model's solutions.4 The representation of agricultural supply behavior is derived from an activity-analysis speci- fication of supply possibiliti'X0 RI + SRAl C. +. FB2 1+ + + - - + -+ ~ 0 PB2 + m > 0 IB2 I-a + R2 j+ C2 _ _ _ _ _ _ _ j+ FB3 + + - + + - -- PB3 + - IB3 - + R3 + I-RA3 C3 + See text for explanation of symbola. Figure 1. Schematic Tableau f or MOCA 337 of the world. Only three countries (indexed 1, 2, 3) are shown in order to save space. The subgroups of activities wit'%in each country are as follows: P - primary production activities, S = transportation and processing activities, F - factor supply activities, and D = domestic demand activities. The symbol Tij signifies trade flows from country i to country j in the region, and E and M stand for exports to and imports from the rest of the world, respectively. In the body of the mnatrix, the plus and minus symbols indicate the sign of the coefficient entries. The row designations are as follows: OBJ = objective function, FB final sales balances, PB = processing balances, IB i nput balances, R = resource availability restrictions, and C the convex combination constraints for the segmented de- mand functions.7 Figure 1 effectively represents the steps in the process of trans- forming raw materials into purchased agricultural products. First, the basic inputs enter the production process, as governed by equations IB and the resource availability restrictions, RA. The harvested products either are sold directly (shown by the + in the intersection of column P and rowFB) or they are sold to processing industries (according to equa- tions PB). Then the goods for sale, both directly from the farm and from processing factories, are allocated to three markets: domestic, foreign but within CACM, foreign but outside CACM. The process is completed by adding imports to the supply possibilities. The objective function is defined to be the maximization of producer and consumer surplus across all markets.8 This serves to replicate the actual market equilibria for all products. To implement this kind of objective function, factor costs enter the OBJ row with a negative sign and the areas under consumer demand functions enter with positive signs. Following Takayama and Judge,9 for international sales only the costs of trade activities enter the objective function. 3. The Structure of a Country Model Each country model in MOCA has its own block-diagonal structure. Primary production possibilities are subdivided accordingto whether they occur on small farms (44 hectares), medium-sized farms (4 to 35 hectares), or large farms (>35 hectares). The first group includes farmers who pro- duce mainly for subsistence, and market only their surplus; they have little access to advanced technologies; yields are low and these farmers are in a disadvantageous position to market their product. These farmers depend only on family labor and, after having fulfilled their minimum food 338 requirements, they can work for wages on medium-sized and large farms. A significant proportion of their family time is devoted to various activi- ties besides farming itself, such as wood cutting and carrying products to rather remote markets. Medium-sized farms are market-oriented although they typically re- tain a part of their production, especially grains, for family consump- tion or for animal feed. They have access to more advanced technologies, which implies the use of oxen power, fertilizer and chemical inputs. Tractor services usually are rented. These producers depend on their own family labor and they hire landless workers but only after having uti- lized their own available family labor. Their access to the market is generally better than that of small farmers. Owners of the large farms usually do not work their lands directly but rather they depend on hired labor supplied by landless laborers ans.; on a part-time basis, smallholding farmers. The large farms tend to spe- cialize more in the traditional export crops and they have access to a wider range of production technologies. These different characteristics were reflected in the FAO (GAFICA) 10 coefficients and they have been incorporated in MOCA in varying ways (which are not shown in the aggregate schema of figure 1). The main dis- tinctions are found on the side of labor markets. MOCA specifies that smallholding farmers may work their own farms and also offer themselves for employment on the largest farms. At the other extreme, on large farms there are no activities allowing for field labor on the part of the owner, but rather all labor requirements are met through hiring either smallholders or landless labor. Owners of medium-scale farms exclusive- ly work their own holdings, but they also may hire labor. The technology vectors in MOCA are defined in accordance with these differences in factor endowments. The least-mechanized tilling proce- dures are specified for the smallest farms and the most-mechanized for the largest. Hence, the present version of MOCA does not allow for di- rect capital-labor substitution at the field level. Indirect substitu- tion may occur, however, via changes inthe crop mix in the aggregate and across farm-size groups. This treatment of micro-level factor require- ments was dictated by the nature of the GAFICA set. Additional produc- tion technologies for each size class will be included in future versions. This is the principal area for which improvements are planned in the more detailed version of MOCA to be implemented by SIECA. It would be desirable to allow choice of technique at the farm level and also to specify labor requirements on a monthly basis rather than on an annual basis, in order to better capture peak and off-peak season effects. Both land and labor inputs to crop cultivation are expressed in agqreqate annual terms at present. In some cases, nevertheless, the land availability in MOCA is not exhausted in all categories because of the frequently poor economic returns obtained from further production. This result is in accord with observed experience. -Figure2 is a mini-tableau for the labor market portion of MOCA, for a single country. Table 18 shows five representative production vectors in MOCA, for Nicaragua. 339 Labor Demands in Cropping Activities Sources of Labor Supply Small Medium Large Small Medium Landless Farms Farms Farms Farms Farms Laborers 1 2 3 1 2 RHS w Small + - < 0 M Medium + - - - 0 O 3 Large + - - 0 0 e Small + + + < LS 3 Medium + c LM g Landless + + < LL Figure 2. MOCA Mini-Tableau for Rural Labor Markets, Within a Country It can be seen that technologies vary significantly by farm group. The aggregate figure for purchased inputs conceals some composition ef- fects; for example, for maize, large farms use a lower proportion of to- tal purchased inputs than do medium farms, but large farms are more in- tensive in the use of fertilizers (hence their higher yields). Yields also vary, but the relative yields do not necessarily indicate compara- tive advantage. For example, the largest farms do not necessarily have a comparative advantage in coffee production, although their yields are higher. (It is likely that the yield differences are caused in part by land-quality differences.) In the case of qrains, the larger farms take advantage of the mech- anization possibilities, and thus there is a certain amount of capital- labor substitution across farm-size groups. However, a crop like coffee is not as susceptible to mechanization. The labor input per hectare for coffee increases with farm size, in response to these factors: medium- scale farms purchase few inputs for coffee cultivation but nevertheless they tend the crops more carefully during the growing season; and large farms combine intensive cultivation practices with purchase and applica- tion of agrochemicals. Table 18. Sample Production Vectors in MOCA, for Nicaragua (all inputs given per hectare) Maize Rice Coffee Small Medium Large Small Medium Large Small Medium Large Input Unit Farms Farms Farms Farms Farms Farms Farms Farms Farms Labor Man-days 57.3 67.0 62.4 81.5 86.3 62.3 93.5 128.1 161.8 Purchased inputs S 1.2 12.3 9.3 8.6 66.9 59.6 -- -- 12.9 Yield lCg 640 870 1,000 1,740 2,340 3,200 300 340 380 0 NOTE: Purchased inputs include the total value of expenditures on machinery, animal trac- tion, fertilizer, other agrochemicals, and seeds. These figures are adapted from GAFICA data. GAFICA, "Perspectivas para el Desarrollo y la Integracion de la Agri- cultura Centroamericana," (Guatemala City: GAFICA, May 1974). 341 Another significant detail which is not revealed in the aggregate tableau of figure 1 concerns the retentions of production for home consumption. As a minimally restrictive assumption, itwas required that a subsistence quantity of maize production be retained on the smallest farms only. It is implicit that they can either retain more or supple- ment their retentions with purchases from the market. (Throughout the construction of MOCA, the spirit has been to introduce as few assumptions as possible, so that the model retains as many degrees of freedom as possible.) Four of the crops inMOCAare perennials: coffee and the three kinds of bananas. For the present version of the model, since it addresses only marginal changes away from the actual situation, only the (average) re- curring annual production costs and returns to these crops have been in- cluded. In later versions, the investment cost also will be incorpora- ted, An a full present-value formulation. Hence the present MOCA has a bias toward overstating the profitability of coffee and the bananas. In general, the subject of perennials in the context of sector programming models is one that is being studied intensively, and there is the possi- bility that new methods will evolve soon in this area. Agroprocessing industries are represented in MOCA in a simple, ab- stract manner. When products are processed, their yield coefficients are entered into the processing balances instead of the national final sales balances. Then they are discounted for the wastaqe portion which occurs in handling andprocessing, and total unit processing costs are subtracted from the objective function. The labor and other specific inputs to proc- essing are not included, as the processing sector is not part of agricul- ture proper. This is a simple but effective treatment of aqroindustries, and in this respect MOCA constitutes an improvement over the CHAC model of Mexico. 4. The MOCA Demand Structure Household demands for agricul-ural products are assumed to behave according to a set of price elasticities. The method used for incorpo- rating such deman1d functions in a linear programming model is that of Duloy and Norton. The unique aspect of demand in MOCA, as a planning model, is the procedure for deriving the price elasticities. Given that only income elasticity estimates were available, FPisch's scheme was uti-. lized to obtain corresponding price elasticities. Frisch's approach allows price elasticities to be computed from the expenditure distribution and the Engel elasticities. The main advantaqe of this approach is that the data required to estimate expenditure (or income) elasticities are more easily available than he time series data required for estimation of direct price elasticities. The expenditure distribution, that is, the per capita consumpt:ion for each one of the diEferent categories of consumption, is also available in most countries. 342 Frisch bases his analysis on certain basic assumptions: (i) The market behavior may be described by the behavior of a 'representative individual." (ii) There is "want independence among groups of commodities (discussed below). The last assumption may not be realistic if elasticities are estimated at a very disaggregated level of commodities. Nevertheless, the problem is minimized as the level of agqregation increases. The general formulation uf the Frisch approach can be summarized as follows: If X1, X, ., Xn are the quantities of commodities consumed by the representative consumer, and P1, P2, *""I Pn are the respective prices, then, p1X + p2X2 + "' + PnXn = Y is the consumer total expenditure. The traditional condition for the equi- librium of the consumer is, in terms of his utility, U: U1 U2 U (1)w, P1 P2 Pn where: au (l Uk (X1 X2, ..., Xn (k = 1, 2, ..., n). axk The common ratio w in (1) may be regarded as the marginal utility of money: w . au , all prices constant. Frisch then defined a concept which describes how w varies as total ex- penditure Y increases. He called it the 'money flexibility" coefficient or: 2w Y W = ay w ' Pi constant. He thenderived the relations between this coefficient and the usual demand elasticities: 1 - 0iEi (2) eii Ei (a i-- (3) eik = -E a (1 + w ik i k w 343 where the additional symbols are as follows: eii - own-price elasticity, good i, aik = cross-price elasticity, goodi with respect to thepra ce of goodk, Ei - Engel elasticity, and a budget share. Given values of % and E., (2) and (3) can be used to compute all the e and eik directly. The assumption implicit in this scheme is that ati pairs of goods i and k have the following property: the marginal utility derived from consuming more of good i is independent of the quantity of good k consumed. Frisch called this property "want-independence." Want-independence may not be a plausible assumption in many cases,14 but the scheme does offer an attractive short-cut for obtaining price elasticities. However, a problem is how to obtain information about rea- sonable values for the money flexibility coefficient w. Frisch himself suggested that the valueof 'wwould decreaseas income increases. Johansen was one of the first to use the Frisch scheme; he turned it around and computed values of Vfrom price and income elasticity information.15 For quite different products, he found values of w which were similar and around 2.0. De Janvry, Bieri, and NuKez16 used the same approach, relying on ex- isting demand parameters for different countries to get values of w. They then regitssed those values against levels of real per capita income and obtained the following relationship: (4) loge (- w ) = 1.5910 - 0.5205 loge p where Y* is per capita real income and P is the overall price index. The negative sign on the second parameter in equation (4) confirms Frisch's coniecture about the inverse relationship between values of w and real per capita income. More recently, in an unpublished work, Lluch and Williams17 took up Frisch's suggestion for building a series of estimates of the money flex- ibility coefficients. The authors used time series data on income and ex- penditures data for fourteen countries of income levels ranging from $129 per capita (Thailand) to $3,348 per capita (United States), and four levels of commodity aggregation. The following regression equation was obtained for values of t: (5) log10 (-u) = 1.434 - 0.331 log10Y, where Y is GNP per capita in 1969 dollars. On the basis of this result, the following interpretation was made: An economy with a GNP per capita of $1,000 has a money flexibility coefficient of about -2.7. This number declines in absolute value by about one-third of 1 percent for each GNP increase of 1 percent. 344 The money flexibility coefficient can also be obtained directly from estimation of the parameters of cardinal utility functions and from the estimation of systems of demand equations where the assumption of addi- tivityis made. DeJanvry, Bieri, and Nunez' also surveyed theliterature on values of w which were estimated using this approach and they calcu- lated the following additional regression equation of w and income: (6) loge (-w) = 1.7595 - 0.5127 loge y/P. Finally, it should be mentioned that another group of authors19 recently obtained results which appear to contradict theFrisch conjecture. Their estimates of w ranged from -0.9445 to -1.0497 for a cross-country sample. In conclusion, while there remains some confusion about Frisch's conjecture, his scheme appears to be a powerful tool and there is some econometric evidence in favor of the conjecture. We have employed all three cross-country equations (4), (5), and (6) to see how their results differwhen applied to income strata in Central American countries. Table 19 defines the income strata and associated per capita income levels, and table 20 gives the results of equations (4) to (6) in terms of values of w. It can be seen that the results do not differ notably, and so for deriving the WXCA demand parameters the average money flexibility coef- ficients (column 5 of table 20) were used. The consequent values of the direct price elasticities are given in table 21 (including some products which do not appear inthefirst version of the model). For this initial version, cross-price effects are omitted. However, some recent work20 has suggested efficient ways of including cross-price effects in linear pro- gramming models, so they probably will be added to later generations of MOCAs. Table 19. Per Capita Income by Strata (in Central American Pesos of 1970) Low Medium High Very High (50% of (30% of (15% of (5% of Country Population) Population) Population) Population) Average Guatemala 79 247 589 2,194 311 El Salvador 82 227 605 1,538 278 Honduras 59 187 458 1,540 231 Nicaragua 105 286 723 1,895 340 Costa Rica 193 466 954 3,153 537 NOTE: The original figures on per capita income were expressed in Central American pesos of 1960. Consumer price index for each country was used to express those figures in Central American pesos of 1970. SOURCE: GAFICA, 'Perspectivas para el Desarrollo y la Integracion de la Agricultura Centroamericana,w vol. 2 (Guatemala City: GAFICA, May 1974), p. 33. Table 20. Computed Money Flexibility Coefficient Values, by Income Strata (1) (2) (3) (41 (5) Income Stratum Country Per Capita income v1 w2 w3 Average w s1970 CA$) Low Income Guatemala 79 -5.5494 -6.5558 -6.3957 -6.1670 El Salvador 82 -5.4428 -6.4317 -6.3172 -6.1639 Honduras 59 -6.4600 -7.6142 -7.0445 -7.0396 Nicaragua 105 -4.7856 -5.6660 -5.8208 -5.4241 Costa Rica 193 -3.4960 -4.1470 -4.7586 -4.1306 Middle Income Guatemala 247 -3.0659 -3.6543 -4.3855 -3.7019 El Salvador 227 -3.2037 -3.8160 -4.5098 -3.8432 Honduras 187 -3.5438 -4.2147 -4.8086 -4.1890 Nicaragua 286 -3.8407 -3.3897 -4.1778 -3.4694 Costa Rica 466 -2.3033 -2.6391 -3.5544 -2.7989 High Income Guatemala 589 -1.9504 -2.3405 -3.2892 -2.5267 El Salvador 605 -1.9234 -2.3085 -3.2602 -2.4973 Honduras 458 -2.2232 -2.6627 -3.5748 -2.8202 Nicaragua 723 -1.7530 -2.1070 -3.0734 -2.3111 Costa Rica 954 -1.5174 -1.8278 -2.8039 -2.0497 Very High Income Guatemala 2,194 -0.98367 -1.1926 -2.1284 -1.4349 El Salvador 1,538 -1.1835 -1.4308 -2.3940 -1.6694 Honduras 1,540 -1.1827 -1.4299 -2.3929 -1.6685 Nicaragua 1,895 -1.0616 -1.2856 -2.2341 -1.5271 Costa Rica 3,153 -0.81447 -0.99025 -1.8877 -1.2308 Total Guatemala 311 -2.7194 -3.2472 -4.0635 -3.3434 El Salvador 278 -2.8830 -3.4394 -4.2172 -3.5132 Honduras 231 -3.1747 -3.7820 -4.4838 -3.8135 Nicaragua 340 -2.5961 -3.1021 -3.9453 -3.2145 Costa Rica 537 -2v0465 -2.4540 -3.3914 -2.6306 SOURCES: Column (1): Table 19. Columns (2)-(4): Obtained through equations (4), (5), and (6), respectively, using figures from column (1). 346 Table 21. Direct Price Elasticities for Food, Central Americaa Product Guatemala El Salvador Honduras Nicaragua Costa Rica Wheat (flour) 0.19 0.21 0.20 0.23 0.16 Sorghum (meal) 0.06 0.06 0.05 0.06 0.08 Rice 0.18 0.17 0.16 0.13 0.12 Maize 0.03 0.03 0.03 0.03 0.04 Root crops 0.15 0.14 0.13 0.16 0.08 Plantain 0.09 0.06 0.05 0.06 0.08 Guineos 0.09 0.06 0.05 0.06 0.08 Sugar 0.15 0.15 0.17 0.13 0.04 Lump molasses 0.06 0.06 0.05 0.06 0.08 Beans 0.12 0.12 0.11 0.07 0.12 Fresh vegetables 0.16 0.19 0.18 0.27 0.24 Fruit 0.21 0.30 0.20 0.18 0.18 Bananas 0.09 0.09 0.08 0.06 0.08 Beef 0.28 0.25 0.23 0.25 0.31 Pork meat 0.15 0.15 0.14 0.16 0.19 Poultry 0.30 0.2' 0.27 0.31 0.38 Eggs 0.25 0.24 0.27 0.26 0.27 Seafood 0.30 0.17 0.21 0.19 0.23 Milk and deriva- tives 0.26 0.22 0.22 0.15 0.21 Vegetable oils 0.25 0.21 0.22 0.19 0.23 Animal fats 0.12 0.14 0.11 0.16 0.19 Coffee 0.15 0.24 0.19 0.16 0.19 Alcoholic beverages 0.28 0.33 0.26 0.30 0.33 aEstimated according to the average income levels and average values of the Frisch money coefficient in table 20 and the income elasticities in table 22. Table 22 gives the expenditure elasticities that went into the compu- tations of price elasticities. There were some additional problems in obtaining base-year values of product prices and per capita income levels, but these were resolved eventually. D. Validation of MOCA The model has been designed to simulate actual (1970) behavior of the sector; therefore before using it for policy analysis, its predictive capacity must be tested. There are no formal tests of validation for mathematical programming models, but measures of goodness of fit can be used to check how closely the model predicts the levels of areas planted, production, prices, and levels of input use. 347 Table 22. Income Elasticities for Central Americaa Product Guatemala El Salvador Honduras Nicaragua Costa Rica Wlheat (flour) 0.60 0.70 0.70 0.70 0.40 Sorqhum (meal) 0.20b 0.20 0.20b 0.20 -- R.ice 0.60 0.60 0.60 0.40 0.30 Kaize 0.10 0.10 0.10 0.10 0.10 Root crops 0.50 0.50 0.50 0.50 0.20 Plantain 0.30 0.20 0.20 0.20 0.20 Guineos 0.30 0.20 0.20 0.20 0.20 Sugar 0.50 0.50 0.60 0.40 0.10 Luup molasses 0.20 0.20 0.20 0.20 0.20 Beans 0.40 0.40 0.40 0.20 0.30 rresh vegetables 0.50 0.60 0.60 0.80 0.60 Fruits 0.60 0.70 0.60 0.40 0.40 Bananas 0.30 0.30 0.30c 0.20 0.20d Beef 0.80 0.80 0.80 0.70 0.70 Pork meat 0.50 0.50 0.50 0.50 0.50 Poultry 1.00 1.00 1.00 1.00 1.00 Eggs 0.80 0.80 1.00 0.80 0.70 Seafood 1.00 0.60 0.80 0.60 0.60 Milk and deriva- tives 0.80 0.70 0.70 0.40 0.50 Vegetable oils 0.80 0.70 0.80 0.60 0.60 Animal fats 0.40 0.50 0.40 0.50 0.50 Coffee 0.50 0.80 0.70 0.50 0.50 Alcoholic beverages 0.80 1.00 0.70 0.80 0.80 Total, nonfoode 0.81 0.81 0.81 0.81 0.81 Elasticities estimated for 1965. bThe elasticity for this product was assumed to be equal to the corresponding elasticity for El Salvador. cElasticity taken from Guatemala. dElasticity taken from Nicaragua. sElasticity for nonfood expenditures estimated by Musgrove for Colombia. SOURCES: GAFICA, "Perspectivas para el Desarrollo y laIntegracio'nde la Agricultura Centroamericana,' vol. 2, (Guatemala City: GAFICA, May 1974), pp. 57-61. Philip Musgrove, "Income and Spendina of Urban Families inLatin America: The ECIEL Consumption Study," preliminary draft (Washington, D.C.: ECIEL, October 1975). 348 The model, as it has been mentioned before, is a demonstration ver- sion at this stage, and it contains a limited number of restrictions an well as behavioral specifications. Because of these limitations, the mock'). is notexpected to provide an adequate representation of the ruralsector at the level of the "representative producer.' However, because demand functions are specified on a national basis, and because foreign trade in restricted, the modelcanbe expected to provide reasonable results inthe aggregate and for each country as a whole. To recapitulate, the basic conditions under which the modelis solved for the base period are as follows: (i) each producer group has a maximum amount of land available for cultivation equal to the total area planted in the base year; (ii) each producer group can use up to a maximum amount of family labor, equal to the level available in the base year; (iii) the availability of landless workers is equal to the total actual use (hired by medium and large farmers) in the base period; (iv) there is only one technology for production of each crop, by, each group of producers; (v) no crop is restricted in its area planted; (vi) purchased inputs are available in infinitely elastic supply; and (vii) commodity demand functions are specifiedat the national level and producer and consumer prices are determined endogenously. Table 23 gives the actual values used for the 1970 land and labor restrictions. In addition to the conditions listed above, the following treatmentof tradewas adopted to reflect trade policy andexternalmarket conditions: (viii) there are upper bounds on the total imports that each country may purchase from within the region and also from outside the region, but imports by product and by CACM country are endogenous to the model; and (ix) each product shipped out of the region faces an export upper bound by country, reflecting demand conditions and quotas in importing countries; these bounds are varied in the policy experiments. MOCA was solved under these initial conditions and its predictive ability was tested in terms of areas planted, production levels, prices, and trade patterns. To measure goodness of fit, the percentage 2bsoluto deviation (PAD) between observed and predicted values was used. Table 24 gives the observed and predicted levels of production (in 1970) by crop and couatry. The percentage absolute deviation is less than 10 percent for all countries. Generally, the model overestimates production, except in Nicaragua where more crops are underestimated than overestimated. 349 Table 23. Endowments of Labor and Cultivable Land in MOCA, 1970 Guatemala El Salvador Honduras Nicaragua Costa Rica (thousand man-years) Family labor in: Group 1 farms 69.6 24.1 32.7 16.0 10.0 Group 2 farms 123.7 40.7 96.2 165.5 46.3 Landless labor 228.8 121.5 85.1 126.1 117.0 (thousand hectares) Land held by: Group 1 farms 306.7 113.0 129.8 70.9 31.5 Group 2 farms 594.2 228.2 308.6 232.3 148.2 Group 3 farms 422.0 220.8 133.0 263.8 201.2 SOURCE: GAFICA, "Perspectivas para el Desarrollo y la Integraci6n de la Argicultura Centroamericana," vol. 2 (Guatemala City: GAFICA, May 1974), tablesJ.4a-J.4d, pp. 265-69, and tables H.1.1-H.1.31. The tendency toward overproduction in MOCA appears to be a conse- quence of underestimated costs of production, and possibly overestimated yields, and hence underestimated prices (tables G-1 through G-5, appendix G). Due to the relative inelasticity of the demand curves, the price errors are greater than the quantity errors in percentage terms. Manage- ment skills and miscellaneous cost items are omitted from the MOCA pro- duction structure, as is risk. There qre known methods of incorporatinq producers' risk in this kind of model, and future versions of MOCA will include it. In any event, it should be recognized that the present MOCA has a downward bias with respect to prices and an upward bias with respect to production. Each reader can draw his own conclusions from these tables, but to the autht :;, the production fit of the model turned out to be somewhat better than expected. It is comparable to CHAC's degree of approximation to reality,23 even though the latter model had many more activities to describe the production process. On the trade side, not surprisingly, the fit is weaker, although MOCA captures the major tendencies within the region. Table 25 shows intraregional trade patterns, in MOCA and in actuality, for maize and beans. Similar comparisons for other commodities are given in appendix G. In the case of trade in maize, MOCA's pattern of exports from Hon- duras to the other countries very much resembles the actual situation, with a significant underestimation only in the case of the exports to Costa Rica. Also for maize a (vertical) comparison of imports in Guate- mala shows that the model represents very closely the trade pattern ob- served in the base period. A very accurate representation of imports is Table 24. A Comparison of the KOCA and Actual Levels of Production, 1970 (thousand metric tons) Costa Rica El Salvador Commodity Observed MOCA Deviation Commodity Observed MOCA Deviation Maize 73,400 73,926 - 526 Maize 279,000 281,874 -2,874 Rice 97,400 101,646 -4,246 Rice 37,600 34,841 2,759 Sorghum 11,500 14,792 -3,292 Sorghum 218,100 126,980 1,120 Beans 15,100 14,899 201 Beans 26,300 27,516 -1,126 Banano (export) 717,300 829,130 -111,830 Banano (export) 43,000 43,889 - 889 Banano (domestic) 117,000 113,991 3,009 Plantain 17,600 30,074 -12,474 Guineo 90,000 91,543 -1,543 Sugar vane 1,367,500 1,602,606 -235,106 Plantain 57,500 65,474 -7,974 Coffee 116,300 116,484 - 184 Sugar cane 1,644,200 1,710,743 -66,543 Cotton 127,900 128,778 - 878 Coffee 74,800 77,892 -3,092 Cotton 4,000 6,000 -2,000 Average absolute deviation 18,569 Average absolute deviation 28,611 Percentage absolute deviation 7.038 Percentage absolute deviation 12.014 Table 24 (Continued) Guatemala Honduras Commodity Observed MOCA Deviation Commodity Observed MOCA Deviation I Maize 709,200 713,613 -4,413 Maize 339,200 354,445 -15,245 Rice 26,500 26,684 184 Rice 6,500 26,057 -19,557 Sorghum 54,100 59,707 -5,607 Sorghum 47,900 48,402 - 502 Wheat 32,600 43,441 -10,841 Wheat 1,000 3,864 -2,864 Beans 55,500 54,521 979 Beans 54,700 75,259 -20,559 Banano (export) 157,300 156,437 863 Banano (export) 1,037,900 986,606 51,294 Banano (domestic) 35,500 35,500 0,000 Guineo 67,900 68,246 - 346 Guineo 284,500 298,116 -13,616 Plantain 78,200 78,608 - 408 Plantain 35,600 15,882 19,718 Sugar cane 1,240,200 1,083,066 157,134 ' Sugar cane 2,658,600 2,411,512 247,088 Coffee 32,700 29,862 2,838 Coffee 119,000 119,332 - 332 Cotton 9,200 10,413 -1,213 Cotton 151,500 152,884 -1,384 Average absolute deviation 25,419 Average absolute deviation 27,724 Percentage absolute deviation 7.061 Percentage absolute deviation 9.328 Table 24 (Continued) Nicaragua Commodity Observed MOCA Deviation Maize 231,300 228,693 2,607 Rice 123,600 106,649 16,951 Sorghum 76,900 74,307 2,593 Beans 55,700 49,410 6,290 Banano (export) 63,000 63,149 - 149 Guineo 148,800 173,783 -24,983 Plantain 60,100 68,792 -8,692 Sugar cane 1,371,700 1,502,222 -130,522 Coffee 37,100 26,471 10,629 Cotton 199,900 197,935 1,965 Average absolute deviation 20,538 Percentage absolute deviation 8.673 Table 25. Predicted and Actual Intraregional Trade in Maize and Beans, 1970 (thousand metric tons) To Costa Rica El Salvador Guatemala Honduras Nicaragua Product From Actual Model Actual Model Actual Model Actual Model Actual Model Maize Costa Rica - - - - - - - - - - El Salvador - 1.9 - - 13.51 10.1 Guatemala - - - - - - - .3 Honduras 16.7 5.0 - - 2.85 5.3 - - 3.9 3.0 Nicaragua - 9.8 - - - .2 - - - - Beans Costa Rica - - El Salvador - - - - - - - - - Guatemala - 1.0 - 2.1 - - - Honduras 16.2 4.4 6.3 - 2.6 2.6 - - 1.8 1.8 Nicaragua - 5.0 - 1.6 - .1 - - - 354 also achieved for El Salvador, Honduras, and Nicaragua. In the case of beans, import patterns for Guatemala, Honduras, and Nicaragua are repre- sented with very high precision. In the case of Costa Rica and El Salva- dor, however, the upper bounds in imports are not reached and the import patterns are more diversified than in the base periods. The final test of MOCA's validity concerns the wage rate charged for family field labor. LikeCHAC, MOCA uses an exogenous wage rate for hired labor, corresponding to prevailing practices. The question is: What percentage of that market wage is approximately charged for the use of family labor to represent the opportunity cost of their time? This per- centage is called the `rd ervation wage ratio.n This is the only parameter in the model which is estimated by "sim- ulation" (to use that much-abused word). The method, if it can be dig- nified with that term, is to solve the model under different values of the reservation wage ratio in order to see which one gives the closest fit to reality. In the papers on CHAC, there is considerable discussion of this issue and the empirical conclusion.24 Without repeatingthat here, it is worth noting that the CHAC reservation wage ratio turned out to be about 50 percent for irrigated zones and about 30 percent to 40 percent for rainfed zones. Values near 0 or 1.0 gave very poor fits. Basically, the reservation wage ratio reflects the imperfect mobil- ity of the farmer in the short run. To leave the farm for the city, say, after he has planted his crop would imply significant financial losses. This is even more true in irrigated areas, so it is reasonable to expect the reservation wage ratio to be higher there. Table 26 gives the results of the reservation wage tests. Building on the CHAC experience, it was decided to explore more carefully the range of 0.3 to 0.5. It can be seen that there is not much to choose in this range; 0.4 is chosen by the barest of margins. On the other hand, values of 0 and 1.0 definitely distort the results. The fact that the model is relatively insensitive to variations with- in the 0.3 to 0.5 range is reassuring, on the one hand, because this is a parameter whose value we cannot expect to know wit-h much precision. On the other hand, it is slightly disturbing because it indicates that the model's aggregate annual treatment of labor inputs has made it relatively unresponsive to changes in the cost of production. Again, the more dis- aggregated SIECA version should give a different picture here. S. Preliminary Policy Results As it now stands, the preliminary MOCA model appears to constitute a reasonably good static representation of Central American agriculture. We reiterate, however, that it should be regarded as a demonstration ver- sion, and that a better version is in the process of development. "Policy results" are presented in this section, not so much to pro- vide concrete recommendations as to as suggest issues and broad qualita- tive results, and to demonstrate some of the uses to which this kind of programming model can be put. Table 26. MOCA Results Under Variations in the Reservation Wage Ratio Value of the Reservation Wage Ratio 0 0.30 0.35 0.40 0.45 0.50 1.00 Country A B A B A B A B A B A B A B Costa Rica - - 13.4 7.0 13.4 7.0 13.4 7.0 13.4 7.0 13.4 7.0 - - El Salvador - - 23.9 12.0 23.9 12.2 23.9 12.2 23.9 12.2 23.9 11.9 - - Guatemala - - 19.5 7.1 19.5 7.1 19.5 7.1 19.5 7.1 19.5 7.1 - - Honduras - - 27.0 9.3 26.4 9.2 26.3 9.2 27.1 9.2 27.3 9.4 - - Nicaragua - - 17.2 8.7 17.2 8.7 17.2 8.7 17.2 8.7 17.2 8.7 NOTE: Column A: Percentage absolute deviation of consumer prices. Column B: Percentage absolute deviation of quantities produced. 356 Two sets of results are given: country-level results, for the case of Costa Rica as an example, and full regional results.- To'set up 't1e Costa Rican model for stand-alone solutions, the simple expedient of fix- ing international trade variables at their MOCA base-solution values was used. TheCosta Rican model then was addressed to two issues: measuring the supply response to price changes, and finding the price level which is necessary to attain national self-sufficiency in basic grains. The two are clearly related. 1. Country-Level Results for Costa Rica One of the most appropriate uses of a linear programmling model is the 'estimation of supply response." The model is built using cross- section microeconomic information and behavioral assumptions that effec- tively define the conditions for supply response to economic policies and investment; it is a simple matter to trace out the supply response func- tions which are implicit in the model. Given the static formulation of the model, these are equilibrium short-run supply response functions; equilibrium in the sense of allow- ing all adjustment processes to work themselves out but short run in the sense of excluding investment. They are calledsupply response functions, instead of supply functions, because when one product's price is varied, the price and quantities of other products are allowed to vary also. In MOCA, most of the responsiveness arises from substitution of land area anong crops. The procedure for tracing out the functions consists of rotating the product demand functions rightward, one at a time, b 5varying the right- hand side valueof the convex combination constraint, as shown in figure 3. In this rotation, the intercept on the price axis is held constant. The original demand function is D., and the rotated functions are denoted D1toD4. Arc elasticities of supply response are then calcLlated ex post from the price-quautity solutions (A, B, C, D, E in figure 3). The pro- cedure was carried out for maize, rice, and beans. The rotations of the demand curve were by the following percentage amounts: D1 : 20 percent down from D D2 : 20 percent up from Do' D3 : 40 percent up from Do' and D24 : 60 percent up from Do. Table 27 gives all the relevantquantity responses for changes in the rice demand curve, along with the corresponding changes in the rice price, defined as the price of polished rice. The quantities refer to the raw product at the farm gate. Figure 4 is a graphic presentation of the in- formation in table 27. Qualitatively, the following points are evident from these results: (i) the 'elasticity' of supply response is not at all constant; indeed, why should it be except to facilitate econometric curve-fitting? (ii) at low prices of rice, one set of substitute crops tends to respond more (maize, beans) and at high prices another does (sorghum). This is the 357 Price X D4 Quaotity Figure 3. Tracing out an Implicit Supply Response Function kind of multivariate information that can be extremely helpful in the course of setting price support policies. Table 28 gives some computed arc elasticities, for the widest range of points explored in each case. Both direct and cross-price elastici- ties are shown, and three product demand functions were rotated in turn. The own-price elasticities of around +0.5 generally accord with other econiometric estimates of short-run supply elasticities for grains.26 The cross-elasticities show that some crops are more likely to drop in pro- duction than others are when prices of competing crops are raised. Beans, for example, happen to be more responsive in a negative direction than other crops, and sorghum is the least sensitive when the rice price is changed. Finally, tables 29 and 30 apply this kind of analysis to the fairly typical concern of internal self-sufficiency, allowingtrade variables to respond in this instance. It can be seen that for Costa Rica, a high- cost producer of grains, self-sufficiency in maize would come very dearly indeed. To analyze the consequences of attempting to achieve maize self- sufficiency, it was assumed that prices to consumers would not be changed, 358 Table 27. Supply Response Functions for Costa Rica, Under Variations in the Ri- Price Point on the Supply Response Function A B C D E Price of Rice (CA$/kg) .1782 .1843 .1968 .2930 .4158 Quantities (000 metric tons) Rice 88.240 110.299 131.343 144.180 147.166 Maize 83.926 72.593 64.673 62.839 62.839 Beans 10.690 8.750 8.217 6.885 6.884 Sorghum 14.499 14.499 14.499 14.343 14.186 SOURCE: MOCA Solutions for Costa Rica (see text). and hence a subsidy would be implied. It can be seen that the required subsidy becomes enormous as self-sufficiency is approached, so in all likelihood consumer price rises could not be prevented. But from table 30, it can be calculated that, at total self-sufficiency, even doubling the price to consumers would not quite eliminate the net government out- lays on this program. These are, of course, illustrative figures, and they do not take into account possibilities for expanding the amount of arable land or increas- ing yields. Nevertheless, it probably is true that maize self-sufficiency is not a practical economic goal for Costa Rica. 2. Policy Results for All Five Countries The actual recent patterns of CACM agricultural trade were discussed in section B of this chapter. It will be recalled that there is a net- work of trade agreements which restricts intraregional trade, particularly in basic grains. The basic solution of the model was designed with con- straints reflecting these realities. Further steps toward economic integration in Central America are likely to depend on an adequate answer to the following question: What will be the consequences of removing some of these trade restrictions, by country and by socioeconomic groups within each country? To assist in answering this question, we have designed two regionwide "policy experi- ments' with MOCA. The first consists of a solution under a 100 percent increase in the limit on each country 's grains imports from other Central American countries. The second consists of an analysis of the incremental gains from exports to the rest of the world. A large percentage increase .4000 - i l l I II I .3600 - I \I .3200 I I .2800 0 .2400 I. .2000 Uc .1600 60 70 80 90 100 110 120 130 140 Quantity of Rice, Maize 6 7 8 9 10 11 12 13 14 Quantity of Beans, Sorghumn Figure 4. Costa Rica: Supply Response Curves with Respect to the Price of Rice Table 28. Direct and Cross-Price Supply Response Elasticities Generated by the Model, Costa Rica Quantity Response of: Rice Maize Beans Sorghum Price change for: Rice 0.501 -0.188 -0.266 -0.017 Maize -0.205 0.565 -0.490 -0.021 Beans -0.048 -0.045 0.489 -0.009 Table 29. Production, Trade, and Degree of Self-Sufficiency at Alternative Producer Prices of Maize, Costa Rica Producer Gross Net Imports (-) Domestic Coefficient of Price Production Production Exports (+) Consumption Self-Sufficiency (CA$/kg) --------------------(thousand metric tons)-------------------------- .080 62.69 59.55 -41.65 101.2 58.8 .093 75.12 72.03 -29.17 101.2 71.2 .133 95.83 91.04 -10.16 101.2 90.0 .258 115.79 110.00 +8.8 101.2 108.7 SOURCE: MOCA solutions. Table 30. Other Consequences of Self-Sufficiency in Maize, Costa Rica (0) (1) (2) (3) (4) (5) (6) (7) (a) (9) (10) (11) (12) Government Total Subsidy Revenue (+), Gross aLize Cost of Average Consumer Maize (-) or Tax Subsidy Expenditure Maize Balance Producer Income, Gross Inco.L, Import Maize Cost of Expenditure Export (+) to (-) on (-) for this Import of Price Producers Intermediaries Cost Supply Supply on Maize Revenue Consumers Elxports Policy Tax (+) Payments .080 5.02 6.49 3.33 14.84 0.147 16.19 - +1.35 - 1.35 1.35 -3.35 .093 7.00 6.53 2.33 15.86 0.157 16.19 - +0.33 - 0.33 0.33 2.33 .133 12.79 6.59 0.81 20.19 0.199 16.19 - 4.00 - -4.00 0.18 0.51 .258 29.91 7.20 - 37.11 0.337 16.19 0.71 -17.95 -2.26 -20.21 - +0.71 NOTES: Price and cost in CAS/kg and all other items in millions of t 7) - (exports) A (export price). CA5. ( 8) - (total consumption) x (cost of supply - consumer t1t) (total production) x (producers' price). price), or (8) - (4) - (6) - (7) - (9). (2) - (total production + imports) x (transformation cost ( 9) - (exports) x (cost of supply - export price), or (9) - coefficient). (4) - (6) - (7) - (8). (3) - (imports) x (import price). (10) - (8) + (9). 4) - (1) + (2) + (3i. (11) - (imports) x (import tax). 5) - (4) - t(net production) + (imports)). (12) - (exports x export price) - (imports x import price). 6) - (total consumption) x (consumers' price). SOURCE: MOCA solutions. 362 was hypothesized for intraregional trade in grains because the base is 80 small. The projected consequences of these possible policies are discussed extensively. In particular, the distribution of income between countries and within countries is seen to be affected significantly. Let us start with table 31 concerning the experiment on liberaliza- tion of intraregional trade. Liberalization leads to a significantly different allocation of production and trade although, by and large, the present tendencies toward importation and exportation are accentuated. Guatemala and Costa Rica become even greater importers than they were in the base periodst and Honduras, El Salvador, and Nicaragua increase their exports. Some exceptions: El Salvador increases its imports of beans and Nicaragua increases its imports of maize; also, Nicaragua switches from being a net importer to a net exporter of beans. Looking back at tables 11 through 13, the changes in trade tend to follow from the observed price differentials, although transportation cost differentials also affect the results. For example, Nicaragua, not Table 31. MOCA Simulation of Intraregional Trade Liberalization in Basic Grains: Production and Trade Variables (thousand metric tons) Country Maize Rice Sorghum Beans A B A B A B A B Part I. Production Costa Rica 73.9 41.4 101.6 93.8 14.8 14.9 14.9 - El Salvador 281.9 301.7 34.8 34.8 127.0 127.0 27.5 21.0 Guatemala 713,6 696.8 26.7 20.3 59.7 59.7 54.5 51.8 Honduras 354.4 390.0 26.1 - 48.2 75.3 75.3 82.8 Nicaragua 228.7 226.4 150.5 106.6 74.3 74.3 49.4 67.5 Central 1,652.5 1,656.3 297.6 297.4 324.0 324.0 221.6 223.1 America Part IT. Imports (+) and Exports C-) Rest of the -15.10 -15.1 + 4.8 + 4.8 - - - - world Costa Rica +31.80 +63.6 + 5.3 + 5.3 - - +16.2 +32.4 El Salvador -13.15 -32.0 - .2 - .2 - - + 6.3 +12.6 Guatemala +16.00 +32.0 + 4.1 + 8,2 - - + 2.6 + 5,2 Honduras -23.45 -56.2 - +15.1 - - -26.9 -34.7 Nicarague + 3.90 + 7.7 -14.0 -38.5 - - + 1.8 -15.5 NOTES: Column A: basic MOCA solution values. Column B: MOCA solutions under trade liberalization. 363 gi Salvador, increases its beans exports to Costa Rica, even though El salvador is the lower-cost producer in this case. Geographical proximity mAkes the difference. Honduras is an even-lower-cost producer, so its tX!4ns exports to neighboring El Salvador are increased. So far there is not too much that is surprising in the model; in the min it helps by attaching magnitudes to directions of changes which could be guessed by knowledgeable persons. In terms of prices (table 32), there are countervailing effects. Costa Rica's increased import bill is com- pensated for, in a sense, by lower prices to its consumers. The reverse holds forHonduras, and for the other countries this volume of additional trade is not sufficient to affect prices. From the foregoing, it is clear that Costa Rican consumers would benefit from trade liberalization. Whether farmers would gain or lose depends onwhetherprice effects are offset by quantity effects. The con- sequences for the rural income distribution are shown in table 33, and there it is evident that the lower prices to Honduran producers are more than offset by increases in production for export. Costa Rican producers suffer in net terms, and the net positions of producers in other countries are relatively unchanged (last column of table 33, part I). Within countries, the picture varies. In Costa Rica, liberalization improves the rural income distribution. The large-scale farmers, and to some extent the medium-scale ones, are the sellers of grains--which are more susceptible to mechanization and hence to larger-scale cultivation Table 32. MOCA Simulation of Intraregional Trade Liberalization in Basic Grains: Consumer Prices (CA$ per kilogram) Maize Rice Sorghum Beans Country A B A B A B A B Costa Rica .126 .106 .189 .162 .071 .060 .369 .209 El Salvador .069 .069 .150 .150 .063 .063 .255 .255 Guatemala .095 .095 .214 .214 .060 .060 .196 .196 Honduras .057 .070 .113 .119 .062 .073 .150 .171 Nicaragua .130 .130 .101 .101 .058 .058 .185 .185 NOTES: Column A: basic MOCA solution valDxes. Column B: MOCA solution under trade liberalization. Table 33. MOCA Simulation of Intraregional Trade Liberalization in Basic Grains: Income Distribution (million $CA) Landless Laborers Small Farms Medium Farms Large Farms Tbtal Country A B A B A B A B A B I. Total Group Incomes Costa Rica 35.29 30.01 5.73 5.01 19.96 16.35 5.49 1.01 66.47 52.40 El Salvador 28.95 29.19 5.57 5.57 12.62 12.62 4.41 4.42 51.54 51.79 Guatemala 27.67 26.88 14.74 14.74 27.70 27.71 10.42 10.42 80.54 79.75 Honduras 16.70 16.46 8.06 9.95 9.85 14.75 5.85 9.17 40.46 50.33 Nicaracua 30.73 30.98 4.45 4.45 8.09 8.82 9.14 9.13 52.40 53.38 Central 139.32 133.52 38.55 39.72 78.22 80.25 35.31 34.15 291.40 287.64 America II. Income Shares Costa Rica 53.1 57.3 8.6 9.6 30.0_ 31.2 8.3 1.9 100.0 100.0 El Salvador 56.2 56.4 10.8 10.8 24.5 24.3 8.5 8.5 100.0 100.0 Guatemala 34.4 33.7 18.3 18.5 34.4 34.7 12.9 13.5 100.0 100.0 Honduras 41.3 32.7 19.9 19.8 24.3 29.3 14.5 18.2 100.0 100.0 Nicaragua 58.6 58.0 8.5 8.4 15.5 16.5 17.4 17.1 100.0 100.0 Central 47.8 46.4 13.3 13.8 26.8 27.9 12.1 11.9 100.0 100.0 America NOTES: Column A: basic MOCA solution values. Column B: MOCA solution under trade liberalization. 365 than are other crops. The reduction of production on larger farms ad- versely affects the employment and incomes of the poorest, the hired la- borers, but on the whole the rural Costa Rican income distribution is made more equal. In Honduras, it is made less equal, but all farmer groups benefit in absolute terms. Although large farms, the principal employers of hired labor, expand their output considerably in Honduras, the incomes of landless laborers drop somewhat under liberalization. This consequence arises from product substitution effects: the increased Honduran exports are maize and beans which are considerably less labor intensive than the rice which they displace. Landless workers also suffer somewhat in Guatemala, but there it is because total production drops. By now it is fair to say that the model is telling us some things which we could scarcely haveknown otherwise, but nonetheless the results appear plausible. It seems evident that there are strong grounds for some compensatory schemes to accompany trade liberalization. To summarize at a more aggregate level, the costs and benefits of expanded intraregional trade are summarized in table 34. The groups which gain are (i) producers in Honduras, (ii) by a smaller margin, pro- ducers in Nicaragua, and (iii) consumers in Costa Rica. The losers are (i) consumers in Honduras, (ii) producers in Costa Rica, and (iii) land- less laborers in Costa Rica. Among countries, trade liberalization can be said to be redistributional: the poorest countries gain at the expense of the richest. But nevertheless, the poorest group in Costa Rica is one of the losers. They gain partial but not full compensation through lower consumer prices. For the final piece of analysis, the dual solution of the model was analyzed with respect to the constraints on exports to the rest of the world. The shadow price on an export upper bound signifies the gain to the objective function of one additional unit of exports of that product. MOCA's objective function is the sum of Marshallian surpluses (consumer and producer surplus), so it leads to a solution which maximizes "social welfare" in the sense of competitive equilibrium. Hence the shadow prices tell us the gains in "welfare," under this definition, from an additional ton of exports. Since tons of different products are not directly comparable, we have chosen to express the shadow prices as ratios to the export prices (table 35). For example, a shadow price/export price ratio of 40 per- cent implies that the social cost of producing another unit for export is only 60 percent of the export price. These computations take into account the fact that additional exports may mean fewer sales--and hence higher prices--on the domestic market. In table 35, the crops are grouped into four sets, and within eazh set they are listed in rank order, according to the value of the shadow price/export price ratio. It is apparent from the table that there are not many easy generalizations as far as the crops are concerned; by and large the results depend on location as well as crop. However, the following rough patterns emerge: 366 Table 34. Aggregate Impacts of Trade Liberalization Within Central America Consumer Price Country Total Producers' Income Landless Laborers' Income Index A B Index A B Index B Costa Rica 31.18 22.37 71.7 35.29 30.01 85.0 74.2 El Salvador 22.60 22.61 100.0 28.95 29.19 100.8 100.0 Guatemala 52.86 52.87 100.0 27.67 26.88 97.1 100.0 Honduras 23.76 33.87 142.6 16.70 16.46 98.6 108.5 Nicaragua 21.68 22.40 103.3 30.73 30.98 100.8 100.0 Central 152.08 156.12 101.3 139.32 133.52 94.4 - America NQOTES: Column A represents the base solution values from MOcA. Column B represents the MOCA solution under intraregional trade liberalization. The consumer price index is defined to be 100.0 for each country in the base solution. Incomes are stated in million CA$. The income index is column B divided by column A. (i) Export bananas andrice exports arehighly profitable regard- less of location (with one exception). (ii) Cotton fiber is almost as universally profitable for export. (iii) Plantain bananas are the less profitable item; in fact, in Guatemala the ratio is slightLy negative, implying that they are exported only with a slight subsidy.2 (iv) The returns to extraregional sugar exports are highly varia- ble according to country of origin, and coffee returns are variable also. (v) No country has a comparative advantage with respect to its Central American neighbors in traditional exports in general. However, by crop there are significant differences across countries. Comparing table 35 with table 7 of section B, it can be seen that the comparative advantage rankings correspond roughly, but only roughly, to historical changes in export shares by crop. Of the group I crops in table 35, six out of nine increased their export share over the 1963-65 to 1970-72 period (including rice, which is not shown in table 7). Ex- cept for cotton, six of six increased their shares. In groups II and III, shares of most crops decreased over time. It would appear that MOCA'S productioni coefficiencs are unduly hilgh, or the yields are too low, as in the case of cotton. 367 Table 35. Ranking of Crops by Comparative Advantage in Extraregional Exports Crop Country Group I. Social Returns Greater than 70 Percent cf Export Price Rice Nicaragua Cotton fiber Honduras Export bananas Honduras Sugar Guatemala Export bananas Guatemala Export bananas Costa Rica Cotton fiber Nicaragua Rice El Salvador Cotton fiber El Salvador Group II. Social Returns 40 Percent to 70 Percent of Export Price Coffee Honduras Sugar Costa Rica Cotton fiber Guatemala Coffee Costa Rica Coffee El Salvador Sugar El Salvador Cotton fiber Costa Rica Group III. Social Returns 20 Percent to 40 Percent of Export Price Coffee Nicaragua Plantain Costa Rica Coffee Guatemala Group IV. Social Returns Less than 20 Percent of Export Price Sugar Nicaragua Sugar Honduras Export bananas Nicaragua Plantain Honduras Plantain Guatemala It is natural to ask whether there is any simple explanation of the cross-country differences in table 35. Recall that we are dealing with comparative advantage at the margin, and not on the average. And that depends principally on the opportunity cost of land, the price on the domestic market, and the way in which technologies differ across farm- sLze groups. As an attempt to cut through this complex network of inter- actions, we have compiled the ratios of labor inputs per unit of output for the traditional crops listed in table 35. These ratios are entered in table 36, along with the country comparative advantage rankings (by crop) from table 35. 368 Table 36. Labor Input per Kilogram of Yield, Traditional Exports per kilogram, par Export Country Rice Cotton Bananas* Sugar Coffee Plantain Guatemala .060 .073 (4) .011 (2) .0024 (1) .320 (5) .027 (3) El Salvador .027 (2) .057 (3) n.a. .0047 (3) .172 (3) .016 Honduras .080 .052 (1) .013 (1) .0057 (5) .328 (1) .015 (2) Nicaragua .037 (1) .081 (2) .019 (4) .0057 (4) .377 (4) .013 Costa Rica .067 .076*(5) .011 (3) .0036 (2) .915 (2) .016 (1) NOTES: The fractional numbers represent labor inputs per unit of output on medium-scale farms, except where asterisked to indicate large- scale farms. The numbers in parentheses represent country rank, for that crop, in table 35 (that is, an index of comparative advantage). For rice and sugar, relative labor costs (or rather, their inverse) explain the comparative advantage rankings quite well. On the other hand, they do not explain the rankings for the other crops. It may be suggested that total unit costs would constitute a better simple index than labor input. But that route leads to difficulties: How are land and family labor valued? Their value is their opportunity cost in alternative uses, that is, in the cultivation of other crops. This effect is captured by the programming model and figures importantly in the model's comparative advantage rankings. It therefore appears that this is an area where a full-sector model provides information that can- not be obtained by simple calculation of costs of production. 369 Chapter 7 NOTES t. At a later stage, when more adequate data are available, risk aversion will be added to the list of behavioral characteristics. 2. For descriptions of CHAC and its uses, see L. M. Bassoco and R. D. Norton, "A Quantitative Approach to Agricultural Policy Planning," Annals of Economic and Social Measurement 4 (October-November 1975): pp. 571-94; and J. H. Duloy and R. D. Norton, CHAC: A Programming Model of Mexican Agriculture," in L. Goreux and A. Manne, eds. Multi-Level Plan- ning: Case Studies in Mexico, (Amsterdam: North-Holland, 1973), chapter IV-1. The equations for MOCA are given in the appendix. 3. All rates of growth in this section were calculated for the 1963-72 period, taking the 1963-65 average and the 1970-72 average as the beginning and the end of the period, respectively. Exports and imports are measured by changes in current U.S. dollar values. 4. With so many products, the number of potentially non-zero sup- ply cross-elasticities is very large, and hence the available time series data would not offer sufficient degrees of freedomtobe able to estimate the supply functions econometrically. 5. This is the central methodological aspect of MOCA. In this re- spect, it follows the Mexican model, CHAC. For a more complete discussion of methodological issues, see Duloy and Norton, "CHAC: A Programming Model'; and "Prices and Incomes in Linear Programming Models," American Journal of Agricultural Economics 57 (November 1975): pp. 591-600. 6. When the economic cost of trade barriers is included in a broad- er definition of price, then it is possible to say that the MOCA solution always equalizes prices. 7. See Duloy and Norton, 'Prices and Incomes." 8. Ibid. 9. T. Takayama and G. Judge, Spatial and Temporal Price and Alloca- tion Models, (Amsterdam: North-Holland, 1971). 10. GAFICA, "Perspectivas para el Desarrollo y la Integraci6n de la Agricultura Centroamericana," (Guatemala City: GAFICA, May 1974). 11. Duloy and Norton, "Prices and Incomes." 12. R. Frisch, "A Complete Scheme for Computing All Direct and Cross Demand Elasticities in aModel with Many Sectors," Econometrica 27 (1959). 13. In some circumstances, a slightly weaker assumption can be in- serted for some of the pairs of goods. 370 14. An evident exception is the case of clear complements, such as coffee and cream. 15. This work is cited in Frisch, "A Complete Scheme." 16. A. de Janvry, J. Bieri, and A. Nuniez, "Estimation of Demand Pa- rameters under Consumer Budgeting: An Application to Argentina," Ameri- can Journal of Agricultural Economics 54 (August 1972): pp. 422-30. 17. C. Lluch and R. Williams, "Cross-Country Patterns in Frisch's Money Flexibility Coefficient," unpublished draft (Washington, D.C.: De- velopment Research Center, World Bank, September 1973). 18. De Janvry et al., "Estimation of Demand Parameters.' 19. P. Pinstrup-Anderson, t. Ruiz de Londono, and E. Hoover, "The impact of Increasing Food Supply on Humrcn Nutrition: Implications for CommodityPriorities inAgricultural Research and Policy," American Jour- nal of Agricultural Economics 58 (May 1976): pp. 131-42. 20. W. Candler and D. Norton, "Linear Transformations in Mathemati- cal Programming," Discussion paper no. 21 (Washington, D.C.: Development Research Center, World Bank, January 1977). 21. That is: n PAD = i i i n i=i i where: Xo = observed value of the variable, product i, and i XiP = predicted value of the variable, product i. 22. P. B. R. Hazell and P. L. Scandizzo, "Competitive Demand Struc- tures under Risk in Agricultural Linear Programming Models," American Journal of Agricultural Economics 56(2) (blay 1974). 23. L. M. Bassoco and R. D. Norton, "Una Metodologia Quantitativa de la Programaci6n Agricola," Demograf(a y Economna 9 (1975): pp. 432-81. 24. See, for example, L. M. Bassoco, R. D. Norton, and J. S. Silos, Ap- praisal of Irrigation Projctts and Related Policies and Investments," Water Resources Research 19 (December 1974): p. 1073. 25. Duloy and Norton, "Prices and Incomes." 2b. See, for example, J. R. Behrman, Supply Response in Underdevel- oped Agriculture (Amsterdam: North-Holland, 1968). 27. This result may well reflect an error in nmodel data, but in qualitdtive terms it seems safe to say that plantain bananas are less profitable in export than the other crops. APPENDIX G A Model of Agricultural Production: Equations and Supplementary Tables Carlo Cappi, Lehman Fletcher, Roger Norton, Carlos Pomareda, and Molly Wainer A. MOCA Equations The 271 structural equations of MOCA are set out fully in algebraic form in this appendix. A number of strictly accounting equations are not shown here; they were included in the linear programming models but they could have been replaced as well by a report-writing subroutine. Further details on the model, including derivations of all numerical coefficients, are givern in a large mimeo notebook entitled "A Spatial EquilibriumModel for the Central American Aqricultural Sector." The reader will notice that there are a number of instances in which equations could be substituted out to reduce the rank of the linear pro- gramming matrix. However, from the viewpoints of efficient matrix gener- ator design and flexibility of model specification,1 it was preferable to generate the model in the form shown here. According to the notational convention adopted, capital Roman letters represent variables and right-hand side values, and Greek letters and small Roman letters indicate parameters. Each set of equations is given a brief title and, in most cases, a ver.bal description of each term is given in brackets after the algebraic statement. Following the title, the equation's FORTRAN name, as used in the numerical computer version, is given. When there are two symbolic names separated by a semicolon, the second one refers to the corresponding generic equation used in the schematic tableau which is discussed in the text. The number of individ- ual equations within each matrix equation is given to the right of the equation statement. A table containing descriptions of all the symbols precedes the equations. 1"Flexibilitvy here means ease of varyinq the model's structure to test differing hypotheses. This is an important attribute for an applied model. 648 All of the equations are written in inequality form. Obviously, many of them will be binding in any solution, and hence those equations Could have beenwritten as strict equalities. However, writina them as inequal- ities reduces the computer time associated with each solution, for it eliminates the need to pass through phase I of the simplex alqorithm, and by judicious use of signs we can be sure that restrictions will be bind- ing when exact equalities are desired. For readers interested in implementing sector models, it may be of interest to note that writing the algebra of this appendix was not th^ initial step in preparing the model's computer version. The initialstep was to design a symbolic nomenclature for columns and rows, assigning rer- tain fields to country indices, others to product indices, and so forth. This convention is not readily apparent in the FORTRAN equation names quoted below, becauseempty-field designations are ignored. Nevertheless, for structuring the model and for writing its matrix gernerator, it was immensely helpful to have notations established from the beginning. It wasevident beforehand that the entities which would needspecial fields were countries, farm size groups, products, inputs, and demand function segments. Also, special symbols were used for accounting balances (B) and restrictions (R). Product symbols (two characters) and all other abbreviations were decided upon at the outset, and then the equations were developed in layers, beginning with farm size class-specific equations and then moving to national and then regional balances and restrictions. 649 Table G-1. Notation for the MOCA Equations Superscripts, Symbol Description Subscripts I. Variables: FOB Country contribution to the r - Country overall objective function G - Guatemala E 1 El Salvador H - Honduras N = Nicaragua -C - Costa Rica DJ Demand curve interpolation j final (3rocessed) '9' weight variable (segment product choice variable); see g - demand curve Duloy and Norton [81 segment index XRj,r Extraregional exports MR Imports of agricultural goods j, from outside the region SMF Input of farmers' labor, own h farm size class h,r their own farms (- 1, 2, 3) smc h,r Input of labor by landless laborers SMI Labor of small farmers (class 1) on the largest farms (class 3) SINhr Value of purchased inputs, h,r excluding labor TRr Total processing costs for agricultural goods TPr Transportation costs associated with international trade in agricultural products TAr Total tariffs on agricultural imports from outside the region A few products go directly to final use, and thus they are counted both as farm-gate products (raw materials) and final products. 650 Table G-1 (Continued) Superscripts, Symbol Description Subscripts Trs Volume of intraregional trade s = country flowing from (to) country r to (from) country s, good j (Trs< 0) Qi,h,r Quantity (in tons) of raw i farm-gate product agricultural product delivered to the processing industry Dir Final sales activity for goods whose prices are fixed Sir Total annual labor available (in man-years) from size class.l farmers S2- Total annual labor available r (in man-years) from size class 2 farmers S9r Total annual labor available r (in man-years) from landless laborers Pi,h,r Hectares planted in product i Th Total arable land (in hectares), h,r farm size class h ML4 Quota on imports from other j,r countries of the region II. Parameters: Ojogo,r Area under the domestic demand j final product g function g - demand curve seg- ment index r - country pe Price for extraregional exports p m Price for imports from outside the region 651 Table G-1 (Continued) Superscripts, Symbol Description Subscripts Wr Reservation wage; valuation of the farmer's own time Wr Market wage [Wr < w- ] Ers Unit transportation cost between s = country countries r and s for shipments of good j Unit dormstic transportation costs m = imports associated with the import of good j from outside the region fe Unit domestic transportation costs e = exports associated with the export of good j to countries outside the region a. Unit tariff rate on imports from 3'r outside the region c. Unit processing costs, unprocessed i = farm-gate - zgood i product P1 Unit output of final good j from i,r the processing of farm-gate product i 9. Cumulative quantity sold at segment 3,g,r g of the demand curve for product j Yi,h,r Yield (in kg/ha) A.h Field labor requirements, in man- -i,h1r days per hectare ni Purchased input requirements, in $CA per hectare 652 The equations of MOCA are shown below; notations for the equations are in table A-1. Number of Constraints 1. Aggregate objective function [FOB] EFOB + max (1) r 2. Country-level contributions to the objective function [rFOB) ZW. D. + Z;p?XR. - p. M4R j,g,r j,g,r j j,r j J,r j,g i i -w SMF -h2Zw SMC - w SMI h=l,2 h,r h r h,r r r -2ZSINhr - TRr TP - TAr - FOBr' > 0 (5) h2 SIhr Tr Tr r r - Area under the domestic Gross revenue from demand functions for [extraregional export sales processed and other - final agricultural goods - [c.i.f. costs of imports] Reservation wage x farmers1 from outside the region labor on their own farms, farnm size groups 1 and 2 - [Market wage cost of rMarket wage cost of small-i hiring landless labor] holders (size class 1) -working on size 3 farms Total purchased input Product processing Product trans- - costs, excluding 1- costs 1 - port costs LlaborJ Li J Tariffs on imports Total contributiorn to - from outside the - the objective function > 0 L region J from country r J 653 3. Transport costs accounting rows [rRTPRI] - TP + rs Tr + 2 MR r jPsjr + . XR. < 0 (5) rproduct total transport] + intraregional transport costs, country r J costs associated with shipments to (from) coun- try s from country r. product j transport costs associated transport costs asso + with imports from outside + ciated with extra- < 0 the region regional exports - 4. Tariff accounting row [rRTARI] - TA + a. MR. < 0 (5) -Tr . 2 j,r Mj,r- rTotal tariffs, 1 + tariff rate times in- 'country r J puts from outside the < 0 region - 5. Processing costs accounting row [rRTRRI] - TRr + , ci,r Qi,h,r - i,h total processing + unit costs times Lcosts, country rJ quantities processed, < 0 raw goods i 6. Consumer-level commodity balances (endogenous-price products) [rBiBP; FBI Q + y, 0.- D. - MR. h i,r Qi,h,r 2 j,g,r j,g,r - r rs + XR .r+2:T. < 0 (64) S 654 net output of final good qyantity demanded of - j from the processing of + final good j, domes- farm gate product j - tic markets - imports from outsid + xtraregional + Net (> 0) ex- the region exports ports from country r to other countries < o of the region j 7. Consumer-level commodity balances (fixed price crops) [rBiBF; FB] Al% i + D. - MR. + XR. h i,r i,h,r j,r j,r j,r + z Trs ( 0 t20) 5 ) - 8. Convex combination constraints on the interpolation weight vari- ables for the demand functions [rRiRC; C] Z D. < 1.0 (64) g j,g,r - 9. Labor constraint for farmers of size class 1 (smallest farms) [rlRMO; R] SMF + SMI < Si (5) l,r SMr - S1 farmers' labor oni labor of class 1 farmers their own farms, + on class 3 farms, in in man-years man-years < total annual labor avail- -able (in man-years) from class 1 farmers - 10. Labor constraint for farmers of size class 2 [r2RMO; RI SMF2,r < 2 r(5) 11. Labor constraint for landless laborers [r9RMO; R] 2 SMCh < S9 (5) h hr- r 655 12. Producer-level product balances [rhBiBP; PB] Yi,h,r i,r,h + 1 i000 Qi,h,r < ° (49) yield (in kg) per hectare] + scale factor times quantity times hectares sown j produced (in tons) at the farm level 13. Land constraints (rhBTI; R] 2 i,h,r - Th,r (5) i total hectares planted 1 total arable land byl by farm size class - farm size class J 14. Labor input balances, farm size class 1 and 2 [rhBMO; IB] Z 1,h,r Pih, - 280 SMFhr - 280 SMCh < 0 (10) [labor requirements per scale factor times farmers hectare, in man-days, - labor on their own farms, times hectares planted in man-years scale factor times use 1 - of hired landless labor < 0 in man-years J 15. Labor input balances, farm size class 3 (r3BMO; IB] i,3,r Pi,3,r- 280 SMIr - 280 SMC3 < 0 (5) class 3 farms; ruse of class 1 1 use of 1 field labor - farmers in their - landless < 0 requirements L off-season laborersJ 16. Purchased input balances [rhBIN; IB] ni,h,r Pi,h,r - 1,000 SINh,r - 656 Einput coefficient per rscale factor times total 1 hectares times - value of purchased inputs, < 0 hectares sown excluding labor 17. Intraregional trade restrictions [rRiIMi rs (13) -Z2 T, ML,< s - j,r Total net imports of goodl regional import restric- j from other countries in < tion, good j, country r J , the region B. Supplementary Tables 657 Table G-2. Comparison of the MOCA Prices with Actual Prices, Costa Rica, 1970 (in CA$/kq) consumer Prices Producer Prices Coamodity Observed MOCA Deviation Commodity Observed MOCA Deviation maize .132 .126 .006 Maize .090 .084 .006 Sorghum .100 .071 .029 Rice .149 .080 .069 Beans .335 .369 -.034 Sorghum .083 .056 .027 Banano (export) .078 .047 .031 Wheat n.a. n.a. n.a. banano (domestic) .090 .114 -.024 Beans .226 .258 -.032 Guineo .100 .080 .020 Banano (export) .050 .022 .028 Plantain .100 .105 -.005 Banano (domestic) .016 .035 -.019 Coffee 1.130 .938 .192 Guineo .034 .020 .014 Rice .303 .189 .114 Plantain .022 .026 -.004 Wheat flour .350 .331 .019 Sugar cane .009 .005 .004 Sugar .160 .107 .053 Coffee .550 .360 .190 Lump molasses .200 .161 .039 Cotton .193 .189 .004 Vegetable oil .740 .688 .052 Cotton fiber .570 .572 -.002 Bran .073 .073 .000 Molasses .073 .073 .000 Cottonseed cake .073 .073 .000 Average absolute deviation .036 Average absolute deviation .036 Percentage absolute deviation 13.409 Percentage absolute deviation 27.991 NOTEt n.a. - not applicable because not produced. 658 Table G-3. Comparison of the MOCA Prices with Actual Prices, El Salvador, 1970 (in CA$/kg) Consumer Prices Producer Prices Commodity Observed MOCA Deviation Commodity Observed MOCA DeviAtion tPaize .091 .069 .022 Maize .070 .049 .021 Sorghum .080 .063 .017 Rice .126 .041 .085 Beans .363 .255 .108 Sorghum .065 .049 .016 Banano (export) n.a. n.a. n.a. Wheat n.a. n.a. n.a. Baniano (domestic) .110 .089 .021 Beans .197 0.94 .103 Guineo .110 .174 -.064 Banano (export) n.a. n.a. n.a Plantain .110 .095 .015 Banano (domestic) .034 .016 .018 Coffee 1.180 .961 .219 Guineo n.a. n.e. n.e. Rico .284 .150 .134 Plantain .021 .009 .012 Wheat flour .460 .435 .025 Sugar cane .008 .003 .005 Sugar .200 .147 .053 Coffee .450 .235 .215 Lump molasses .140 .086 .052 Cotton .196 .146 .050 Vegetable oil .510 .655 -.145 Cotton fiber .580 i 5 .175 Bran .058 .058 .000 Molasses .058 .058 .000 Cottonseed cake .058 .058 .000 Average absolute deviation .066 Average absolute deviation .058 Percentage absolute deviation 23.869 Percentage absolute deviation 44.986 NOTE: n.a. - not applicable because not produced. 659 Table G-4. Comparison of the MOCA Prices with Actual Prices, Guatemala, 1970 (in CA$/kg) Consumer Prices Producer Pricea Commodity Observed MOCA Deviation Commodity Observed MOCA Deviation Maize .115 .095 .020 Maize .066 .047 .019 Sorghum .100 .060 .040 Rice .095 .043 .052 Beans .297 .196 .101 Sorghum .078 .040 .038 Banano (export) n.a. n.a. n.e. Wheat .165 .043 .122 Banano (domestic) .130 .131 -.001 Beans .187 .090 .097 Guineo .180 .164 .016 Banano (export) .050 .014 .036 Plantain .180 .181 -.001 Banano (domestic) .016 .017 -.001 Coffee 1.650 1.509 .141 Guineo .016 .008 .008 Rice .295 .214 .061 Plantain .016 .017 -.001 Wheat flour .360 .1B9 .171 Sugar cane .008 .002 .006 Sugar IN' .000 .070 Coffee .450 .311 .139 Lump molasses .220 .139 .081 Cotton .170 .128 .042 Vegetable oil .920 .674 .246 Cotton fiber .550 .485 .065 Bran .055 .055 .000 Molasses .055 .055 .000 Cottonseed cake .055 .055 .000 Average absolute deviation .065 Average absolute deviation .047 Percentage absolute deviation 19.457 Percentage absolute deviation 42.587 NOTE: n.a. - not applicable because not produced. 660 Table G-5. Comparison of the MOCA Prices with Actual Prices, Honduras, 1970 (in CA$/kg) Consumer Prices Producer Prices Commodity Observed MOCA Deviation Commodity observed MOCA Deviation Maize .083 .057 .026 Maize .078 .049 .029 Sorghum .080 .062 .018 Rice .127 .059 .067 Beans .234 .150 .084 Sorghum .064 .047 .017 Banano (export) .180 .132 .048 Wheat .147 .073 .074 Banano (domestic) n.a. n.a. n.a. Beans .174 .094 .080 Guineo .180 .158 .022 Banano (export) .040 .013 .027 Plantain .180 .157 .023 Banano (domestic) n.a. n.a. n.a. Coffee .860 .653 .207 Guineo .028 .009 .018 Rice .228 .113 .115 Plantain .028 .008 .020 Wheat flour .340 .235 .105 Sugar cane .006 .005 .001 Sugar .190 .175 .015 Coffee .400 .197 .203 Lump molasses .220 .189 .03i Cotton .213 .097 .116 Vegetable oil .630 .698 -.066 Cotton fiber .550 .203 .347 Bran .048 .048 .000 -olasses .048 .048 .000 Cottonseed cake! .048 .048 .000 Average absolute deviation .069 Average absolute deviation .059 Percentage absoluLe deviation 27.000 Percentage absolute deviation 50.012 NOTE: n-i. - not applicable because not produced. 661 Table G-6. Comparison of the MOCA Prices with Actual Prices, Nicaragua, 1970 (in CA$/kg) Consumer Prices Produce ..ices Commodity Observed M4OCA Deviation Commodity Observed MOCA Deviation Maize .121 .130 -.009 Maize .073 .081 -.008 Sorghum .100 .058 .042 Rice .135 .044 .091 Beana .310 .185 .125 Sorghum .079 .042 .037 Banano (export) .150 .143 .007 Wheat n.a. n.a. n.e. Banano (domestic) n.a. n.a. n.a. Beans .210 .090 .120 Guineo .110 .049 .061 Banano (export) .020 .023 -.003 Plantain .110 .097 .013 Banano (domestic) n.a. n.a. n.a. Coffee 1.450 1.338 .112 Guineo .045 .015 .030 Rice .265 .101 .164 Plantain .022 .010 .011 Wheat flour .430 .453 -.023 Sugar cane .005 .003 .002 Sugar .210 .186 .024 Coffee .450 .340 .110 Lump molasses .150 .105 .045 Cotton .193 .136 .059 Vegetable oil .670 .634 .036 Cotton fiber .530 .369 .161 Bran .060 .060 .000 Molasses .060 .060 .000 Cottonseed cake .060 .060 .000 Average absolute deviation .051 AvIrage absolute deviation .047 Percentage absolute deviation 17.151 Percentage absolute deviation 38.360 NOTE: n.a. - not applicable because not produced.