The range of average landholdings across household groups is larger in Nicaragua and El Salvador than in Honduras and Guatemala . 5 In part, these differences reflect the criteria that were used to classify rural households; however, both the average landholdings and the criteria used to construct our household groups also reflect differences in access to land in the four countries. What constitutes a large holder household in Nicaragua is not the same as in Guatemala, for example.In addition to being heterogeneous, rural households exhibit diversified income sources, technologies, and demands. The same rural household commonly participates in multiple activities and receives income from various sources. Policy shocks that directly affect one activity are transmitted to others within the household as well as to other households in the rural economy. In most household groups, the share of household income from agricultural and livestock production is less than 50% and for some groups it does not reach 25%. Nearly all groups obtain around 50% of their income from wages, the majority of which are non-agricultural. Even commercial households depend heavily on wage labor for their income. Agricultural and livestock production in each household group is also diverse. For example, in Nicaragua, medium commercial households acquire a little less than one third of their total value-added from the production of basic grains, and the other four producing groups obtain between 16% and 24% from this activity. Livestock accounts for between 27% and 52%,raspberry cultivation pot depending on the household group. The shares of traditional export crops are less than 10% of total value added in all groups except large commercial households.
Production of non-traditional crops represents more than 10% of total value added in all household groups, and non-agricultural production accounts for more than 10% in all but low-education landless households and large commercial producers . Similar levels of diversification are found in the other three countries. In all groups, the average household participates in multiple income activities, including production, wage labor, and migration. There is evidence of technological diversification, reflected in differences in factor value-added shares in the same activity but across households. In general, family value added shares are smaller in the same activities for large commercial households than for subsistence producers, while market-input shares are larger for commercial producers. Technological heterogeneity across households, like differences in market access, is generally absent from aggregate economy-wide models.To construct the rural economy-wide models, we first used data from the surveys to construct a Social Accounting Matrix for each rural household group. Each of these SAMs could be viewed as generated by a single agricultural household model. The SAMs were then joined together into a rural sector-wide SAM for each country. Nearly all of the information needed to calibrate the corresponding household models and the rural economy-wide models are contained within these SAMs, as explained below. The four rural SAMs are interesting in and of themselves, because they offer a snapshot of individual groups of rural households as well as the linkages that transmit influences of policy shocks among households. The framework of the SAMs is described in appendix B. Each household SAM consists of a set of 44 production activities, 5 factors, government, 9 investment accounts, and three “rest-of world” accounts, including the rest of the rural sector of which the household is part, the rest of the country, and the rest of the world outside the country.
Unfortunately, no single data source provides all of the information necessary to estimate the SAMs. Because of this, data from diverse sources were used to construct a SAM for each rural household group in each of the four countries. The SAMs, together with econometric estimates of remittance elasticities and family value-added shares, were used to calibrate the household models that constitute each DREM. The key data sources for each country include the rural components of the El Salvador Multi-purpose Household Survey of 2003, the Guatemala National Living Standards Survey of 2000, and the Nicaragua Living Standards Survey of 2000. All three of these are nationally representative and provide information on socio-demographic variables, production and inputs, wages, migrant remittances and income from other sources, and expenditures. Anationally representative survey was not available to construct the Honduras DREM; thus, the six rural household SAMs were constructed from two data sources: a survey of rural households conducted by IFPRI-WUR-PRONADERS in 2001-2002 , and a rural household survey conducted by the University of Wisconsin and The World Bank in 2000-2001 . The IFPRI survey covered 376 households and 1066 parcels in 19 cantons, and the University of Wisconsin survey covered 850 households in 26 cantons, mostly in the northern part of the country. We primarily used the IFPRI survey to construct the Honduras household SAMs, because of its greater detail on production costs and consumption expenditures and because it is more nationally representative, as hillside zones constitute 80% of the nation’s land area. The Wisconsin data were used primarily to disaggregate family value added.
The form of each household-specific factor and consumption demand depends on technology and preferences. On the technology side, we assume Cobb-Douglas production functions for each household group and good, in which the exponents are set equal to factor shares in value added, as implied by profit maximization and available from the household SAMs. Consumption demands were modeled using a linear expenditure system with no minimum required quantities , implying that preferences of individual groups are described by a Cobb-Douglas utility function. Budget shares, like factor value-added shares, were calculated from the household expenditure columns in the SAMs. The elasticities of remittance income with respect to migration were estimated by regressing household remittances on the number of family migrants in each household. The solution to the base model for each country determines labor demands in each activity, production, full income and consumption demands for each rural household group, the agricultural wage, migration, and shadow prices of grain in subsistence household groups.This base model is the starting point for carrying out simulations to explore the impacts of CAFTA’s agricultural provisions on rural welfare.Simulations were conducted under three different scenarios designed to explore the impacts of trade policy adjustments, which are depicted in appendix A, on the rural economy of each of the four countries in the short, medium, and long run. The simulation experiments are summarized in table 4. Our simulations are founded on two propositions. The first is that domestic prices of agricultural commodities would decrease by percentage amounts equivalent to the changes in tariffs prevailing prior to CAFTA. The second is that the changes in agricultural prices would directly affect only the producers that market the good in question. That is, subsistence households would not be affected directly by trade reforms,low round pots although they may be affected indirectly, via other rural markets.The extreme or long-run scenario, in which an immediate elimination of tariffs for all agricultural goods is simulated . This scenario illustrates what might occur without transition policies, including gradual removal of tariffs, and with no change in Central American countries’ agricultural exports to the United States. Unlike NAFTA, CAFTA does not call for a reduction in tariffs for white maize. Nevertheless, we include the removal of tariffs on white maize imports in this simulation because it is intended to represent the extreme case and also because there may be some substitutability between white and yellow maize in production and consumption. The intermediate or medium run scenario simulates a case in which there is immediate elimination of tariffs for sensitive agricultural goods whose tariff-free quotasexceeded imports from the United States in recent years, and/or for which the tariff phase-out period initiates during CAFTA’s first year. How to treat maize in this scenario is complicated for various reasons. Although CAFTA differentiates between white and yellow maize, there is some degree of substitutability between the two. However, in our simulations it is not possible to differentiate between yellow and white maize. Most household production in Central America is of the white varieties, but the available data do not distinguish maize by color. Additionally, the decision of whether to include or exclude maize based on the difference between CAFTA quotas and imports depend on the period during which one measures maize imports. This intermediate scenario includes maize liberalization in El Salvador, Guatemala, and Nicaragua.
In these three countries, tariff-free quotas established for the first year of CAFTA significantly exceed maize imports from the Untied States; thus, one would expect a decrease in domestic prices in the medium but not the short run. Maize liberalization is excluded from the intermediate scenario for Honduras, where tariff-free quotas are equal or inferior to pre-CAFTA imports and are small compared with total supply . This scenario also includes the elimination of tariffs for beans and meats in each of the four countries, because a grace period was not negotiated for these products. Finally, rice was included for Honduras, where the negotiated quota exceeds current imports. Finally, the low or short-run scenario simulates a situation in which sensitive products with special safeguards and/or grace periods of 10 years or more are excluded. This scenario excludes liberalization of rice, maize, small livestock, and milk products in each of the four countries. It eliminates tariffs on large livestock and beans in Guatemala, Nicaragua, and Honduras. There are special safeguards and a 15-year phase-out of tariffs on these products in El Salvador. The phase-out period of included products initiates in year 1 and that of excluded products begins after year 5 of CAFTA’s implementation. The exception is low-quality meats in El Salvador, for which the grace period is only three years. The results of these scenarios depend on the design of the scenarios, which reflect pre-reform protection levels and the details of the agreement’s implementation in each country, the linkages among rural households and markets, which transmit the effects of the reforms through the rural economy, the mix of pre-reform production and income activities in each household group and country , and , the model parameters, which shape the responses of rural household production, consumption and migration.The extreme scenario represents a significant shock for the rural economies of all four countries. Its immediate effect is felt in the producer households that sell the affected goods prior to CAFTA. Market linkages transmit the effect from these to the other rural household groups, including landless and subsistence households. Basic grain production falls sharply in almost all cases; however, there are striking differences between countries as well as among household groups within countries . Grain production decreases by 26-30% in Guatemala, 14% in Honduras, and 8-50% in Nicaragua. Supply elasticities for each household group, which can be calculated from the simulations, reflect general-equilibrium adjustments in each country’s rural sector. For maize, these range from 0.26 to 1.15 in Nicaragua, 0.70 to 0.90 in Honduras, and 1.65 to 1.69 in Guatemala. In most cases, the largest decreases in supply are for the grains that were most heavily protected prior to the CAFTA reform: rice in Nicaragua, El Salvador and Honduras and maize in Honduras. Nevertheless, in some cases general-equilibrium effects mitigate the effects of prices changes implied by the elimination of tariffs. This is the case in El Salvador, where the price of maize decreases by 20% under the extreme scenario but maize production falls by 1.4 and 12.2 percent in small and medium commercial households, respectively. This seemingly paradoxical result is explained by the importance of livestock products in this group’s production mix and an even steeper decline in livestock products under this scenario. The changes in basic grain prices do not have a direct effect on subsistence households. However, its impact is transmitted to these households via rural markets, particularly for labor. Implicit or shadow prices of specific basic grains are almost unchanged in El Salvador, but they decrease 0.9-2.8% in Guatemala, 0.5-1.3% in Honduras and 2.1-6.7% in Nicaragua, compared to decreases in commercial prices that range from 15% to 62%. Lower shadow prices of grains accompany decreases in subsistence household incomes. Labor demands on large farms contract, causing a reduction in rural wages in all scenarios.