32952 Trade Integration and Rural Economies in Less Developed Countries: Lessons from Micro Economy-wide Models with Particular Attention to Mexico and Central America Report to the Latin America and Caribbean Regional Office (LCR) of The World Bank* J. Edward Taylor Agricultural and Resource Economics and Center on Rural Economies of the Americas and Pacific Rim University of California, Davis May 2002 Trade Integration and Rural Economies in Less Developed Countries: Lessons from Micro Economy-wide Models with Particular Attention to Mexico and Central America Summary The impacts of WTO or regional trade integration on rural production, incomes and poverty depend critically on three questions: First, to what extent are the influences of trade integration transmitted through domestic markets to economic agents in rural areas? Second, to what extent are rural producers diversified or able to diversify their economic activities in response to trade reform in order to (a) buffer themselves from negative market shocks and (b) respond to new market opportunities? And third, how do government policies facilitate or retard internal adjustments to freer trade? Answering these questions requires an in-depth understanding of the structures of rural markets and incomes, the ways in which these structures influence production and incomes in rural areas, and the policies that accompany trade integration. Impacts of trade reforms are likely to be different in diverse market and policy settings. This paper presents an analytical and empirical approach towards understanding how trade liberalization is likely to play out in rural economies, with special consideration for production, incomes, migration, and farm-nonfarm linkages. It explores the ramifications of alternative market structures, levels of activity diversification, technologies, and policy regimes in shaping outcomes of trade reforms on a local level. A simple yet flexible micro economy-wide model that includes a variety of activities, technologies, and household types of policy interest is designed and calibrated using micro survey data from Mexico and Central America. This model is then used to evaluate household and rural economy-wide impacts of policy changes in diverse rural settings. Simulations using the model are used to derive policy lessons explore the extent to which lessons from NAFTA reported in Taylor, Yúnez and Dyer (1999) and elsewhere may be relevant in alternative production, technology, and market environments characteristic of rural Mexico and Central America. Findings highlight the critical role of imperfect markets and transaction costs in shaping local economy-wide outcomes of trade policy reforms. High transaction costs and lack of access to capital and new product markets exclude poor rural households from many benefits of trade liberalization and may exacerbate poverty in the wake of trade reforms. Lacking access to rural capital and product markets, the major alternative in poor rural households is to transfer their labor to sectors of the economy with better access to credit and product markets. This is likely to entail increased wage work, including migration. Overcoming market barriers is a key to promoting more broad- based participation in the benefits of liberalization. 1 Introduction Interactions between trade liberalization and rural economies in less developed countries (LDCs) are among the least researched and understood topics in economics. This is surprising, given LDCs' reliance on primary exports, the increasing importance of poverty alleviation and broad-based growth on development agendas, and the fact that the majority of the world's population and poverty is concentrated in rural areas. Almost all economics research on trade policy impacts focuses on nations or groups of nations, offering little insight into how trade reforms play out within nations. It is through the actions of economic actors within nations that communities, regions, and countries become integrated with global markets, directly or indirectly (Taylor, 2001). Linkages among economic actors, together with the market structures and policies that influence these linkages, transmit the influences of trade reforms from directly affected actors to others inside and outside of the rural economy. Through these linkages, ever-widening circles of individuals, households, and firms become affected, directly or indirectly, by trade reforms. The consequences for rural production, incomes and migration depend upon the extent to which influences of trade integration are transmitted to different rural household populations, as well as the ability of households to adjust to changing market conditions, both positive and negative, by altering their income activities in rural areas or through migration. This paper offers an intra-national perspective on the impacts of trade reforms on rural economies. Its focus is on economic actors within nations and the linkages that transmit influences of trade reforms inside and outside the rural economy. The principal tool of this analysis is "micro economywide modeling," based on villagewide models pioneered by Taylor and Adelman (1996). This approach combines the advantages of micro models, which focus on individuals, households, and firms, with economywide models, which highlight economic linkages among economic actors but traditionally have been implemented at an aggregate (national or multi-national) level. Micro economywide models explicitly take into account the market structures that govern economic interactions and that promote or retard the spread of globalization within countries. Part 1 provides an intuitive explanation of the avenues through which trade liberalization may influence rural economies, highlighting the need for a micro economywide analytical approach to understand local impacts of policy reforms. Part 2 describes the micro economywide approach and its implementation. A stylized micro economywide model is estimated using household survey data from rural Mexico and El Salvador. It is used to explore rural economywide impacts of trade reforms and accompanying policies that directly affect distinct sectors of rural economies, including labor and capital intensive staples, cash crops, households, and migration. The Conclusion, Part 3, summarizes findings and ramifications for trade reform and rural income policies. 2 1 Micro Economywide Impacts of Policy Reforms: An Overview A major limitation of trade integration studies is their lack of an intra-nation focus. At the other extreme, there are many micro-economic studies of impacts of policy and market changes in rural areas, but they generally focus only on the households and activities that are directly affected by these changes. Research that focuses on the directly affected households and activities miss myriad ways in which trade policy changes influence rural economies and particular groups of actors within them. Impacts of trade integration within nations are complex, involving three sets of interdependent actors. First, some economic actors are affected directly, by participating in world markets either as buyers or sellers. These "global" actors include migrants who supply labor abroad while remitting income back to their home countries, firms supplying output and/or purchasing inputs from abroad, and households whose members cross national borders to purchase goods directly from foreign firms. A second group of actors are affected indirectly, through policy. Policy is one of the most important avenues through which globalization may influence economic actors. For example, opening up to world markets stimulates income growth and urbanization, which in turn create new demands for rural production and rural-to-urban migration. In Mexico, where staples have been protected by government policies, changes mandated by NAFTA lower the economic returns from staple production and encourage farmers to shift resources into other activities, possibly including internal or international migration (Yunez, 2002). The migrant-source household or the farmer who changes production in response to trade reforms is thus affected by market and policy changes that might not have occurred without trade integration. Many others potentially are affected through their market interactions with global actors or with others who interact with global actors. Farmers convert croplands into pasture to supply a growing local and urban demand for meat--possibly using remittances from children who have migrated to finance this investment. A family hires a local bricklayer to build a new house and pays him with remittances from a migrant working abroad. Both the rancher and the bricklayer are affected by trade reforms, in the sense that their activities depend upon others who are integrated, directly or indirectly, with global markets. The activities of still others may be influenced by the rancher's or bricklayer's newfound income, and so on--like ripples in a pond. Any factors influencing interactions among economic actors shape the ways in which trade reforms play out within countries, regions, and communities. These include local market structures that promote or retard demand linkages among firms and households. 3 Trade Integration and Rural Incomes: Two Extremes A well known tenet of international economics is that there are both winners and losers from trade liberalization. The impacts of trade liberalization on rural incomes are bracketed by two extremes, which we might call the "optimistic" and "pessimistic" scenarios. The optimistic scenario is that trade reforms open up new markets for goods or factors (labor) supplied by rural households, either directly or indirectly, through the market linkages described previously. The pessimistic view is that rural households, especially the poor, suffer from negative impacts of liberalization on previously protected markets, and high transaction costs exclude many rural households from new market opportunities and/or prevent them from exploiting these opportunities where they exist. Direct Market Transmission to Rural Households In order to influence rural households directly, rural markets must transmit the influences of trade reforms. High transaction costs in product and factor markets may isolate households from outside markets, limiting or blocking the transmission of policy influences. This is the motivation for recent studies of "autarkic" agricultural household economies with endogenous "shadow prices" for household nontradeables (e.g., de Janvry, Fafchamps and Sadoulet, 1991; Strauss, 1986). It is unlikely that a household will be isolated from all markets. It is more likely that markets are selectively missing for individual households (de Janvry, et al., 1991; Strauss, 1986) or for local economies of which these households are part (Taylor and Adelman, 1996). That is, household and local economies confront a mixture of tradables (goods for which there is access to outside markets) and nontradables (good for which markets are effectively missing).1 Impacts on rural incomes and poverty thus depend, in the first instance, on how trade reforms selectively affect markets for tradables and nontradables produced by different groups of rural households. The most positive scenario for rural incomes and poverty is one in which reforms enhance market opportunities for tradables supplied by rural households (e.g., raising horticultural prices for Central American farmers or wages for internal migrants). The most negative scenario is one in which households supply tradables that were protected prior to trade reforms. An example of negative impacts is the case where small commercial maize farmers in Mexico, under NAFTA, or Guatemala, under an expanded NAFTA, are adversely affected by competition from Iowa corn farmers. A less direct example is one in which rural households supplied labor to previously protected activities (e.g., to large commercial farms or, through migration, to protected urban industries). In this case, rural labor was protected indirectly, through industry protections, prior to trade reform. Competition from foreign producers depresses prices, wages and employment in previously protected activities. 1Effectively missing implies that, although the market may exist outside the household or local economy, high costs of transacting on that market preclude market participation. 4 The direct effects of trade reforms are likely to be minimal for households or local economies facing missing markets, due to high transaction costs. An example is the case of the maize farmer who produces a surplus but does not market it outside the village because of high costs of transportation, marketing, storage, and/or information. Although trade reforms reduce maize prices in commercial centers (e.g., by eliminating government price subsidies or import quotas or tariffs), many rural households and their economies do not have access to these prices. In effect, local prices are endogenous, and the transmission of price changes to local markets is, at best, imperfect. In these cases, the direct impact of trade reform may be minimal. Indirect Transmission Through Market Linkages The influences of policy changes cited above all concern households that are affected directly by the policy change, in isolation of the larger economies of which they are part. Economic interactions within the rural economy transmit the impacts of policy reforms to others inside and outside of the rural economy, like ripples in a pond. Consider the following examples. Example 1: Labor Market Linkages. A policy reform, by creating new market opportunities, increases income directly in a (e.g., non-poor rural) household, enabling it to hire labor from another (e.g., poor rural) household. This expenditure transmits benefits of the policy change beyond the directly affected household. To understand the impact on poverty (and other variables of interest), it is necessary to follow expenditure linkages to, and from, the labor-supplying household. Note that, even if the transfer is spent on consumption, it nevertheless stimulates income growth in households that sell consumption goods or services to the household benefiting from the policy. If the latter include poor households, poverty may decrease even if poor households are not influenced directly by the policy change. The reverse would occur if the policy reduced income in the directly-affected household, e.g., by liberalizing previously protected markets. Example 2: Capital Market Linkages. The household benefiting from the policy change loans money to a second household in the village, and the second household invests the funds in a new production activity--perhaps in response to new market opportunities. If the first household did not benefit from the policy, the second household, lacking access to formal credit, would not have access to funds to invest. If the borrowing/investing household is poor, then capital market linkages, combined with new market opportunities, may reduce poverty. The reverse would occur if the policy change reduced the liquidity of the directly affected household. Example 3: Missing Staple Market. Imagine a village isolated from regional staple markets by poor infrastructure; high transaction costs make villagers rely on local production to satisfy their staple consumption demands, and producers, finding it too costly to market their staples outside, seek markets for their surplus staple production within the village. These conditions create a local market for the staple; the staple price is determined by local supply and demand, not by outside markets or government policy. A household within this 5 village benefits from a policy change--say, by producing a cash crop whose price increases as a result of the policy reform. Its income increases, raising the household's demand for staples (directly or indirectly, e.g., through demand for livestock products). The local staple price increases. The higher staple price transmits benefits from the cash crop producer to local surplus staple producers. If staple producers are poor, poverty may decrease, even though the policy change did not affect staple production directly. A decrease in the cash- crop price would have the opposite effect on staple-producers' incomes in this example. If staple and cash-crop producers compete for scarce resources (land or labor), the missing staple market could limit the cash-crop supply response to the policy change (a local economywide analogue to the household with missing markets explored by de Janvry, et al., 1991). In all three of these examples, market structures are critical in shaping local economy- wide impacts of the policy change. In Example 1, without a local labor market, benefits from the policy change cannot be transmitted to the poor worker household. Example 2 depends on the existence of a local credit market to channel savings. In the third example, the local staple price is the link between the household directly affected by the policy change and the staple producer. Market imperfections are like stones that block or distort the ripples in a pond, with potentially large repercussions on rural production, incomes, migration, farm- nonfarm linkages, and poverty. Many households potentially are influenced by trade reforms indirectly, through their economic relations with households directly affected by trade policies. Many or possibly most impacts may, in fact, lie outside the households directly affected by the policy reforms. This may be particularly true for impacts of trade reforms on the rural poor, who are not as likely as wealthier households to have access to commercial markets or to the capital and other resources needed to respond to new market opportunities. Poor households nevertheless may be influenced by policies through demand linkages in markets for factors or goods, which create income multipliers and transmit impacts of migration and policy changes. Ultimately, multiplier effects of policy changes are transmitted through trade to regional commercial centers with which rural economies interact--possibly creating migrant labor opportunities there. Policy changes may unleash a variety of other general-equilibrium effects on rural economies. For example, if increased demand for labor in activities stimulated by the policy change drives up local wages, rural economies may restructure themselves around labor scarcity, shifting to less labor-intensive (and more capital-intensive) activities and production technologies. An ever-widening circle of economic actors then becomes influenced by trade reforms, even if they do not supply to markets directly affected by these reforms. Paths of influence may be even less direct than this. For example, trade liberalization leads to new construction in a regional urban center, for which migrants from rural households supply labor. Rising incomes linked directly or indirectly to trade increase the demand for meat produced on pastures where corn once grew. The migrant-sending or livestock-producing village household is affected by trade liberalization, in the sense that its activities have been reshaped by the country's integration with world markets. In this 6 scenario, the rural poor may benefit by participating in migration (and receiving remittances) and/or by entering into livestock production (financed, perhaps, by migrant remittances; see Taylor 1992). Table 1 summarizes likely influences of trade liberalization on rural households under alternative scenarios. The most positive scenario for rural incomes and poverty is given in the northwest cell of the table, depicting the case in which reforms enhance market opportunities for tradables supplied by rural households (e.g., raising coffee prices at the farm gate or wages for internal migrants). The most negative scenario is given by the southwest corner of the table, in which households supply tradables that were protected prior to trade reforms. An example of negative impacts is the case where small commercial maize farmers in Mexico, under NAFTA, are adversely affected by competition from Iowa corn farmers. A less direct example is one in which rural households supplied labor to previously protected activities (e.g., to large commercial farms or, through migration, to protected urban industries). In this case, in effect, rural labor was protected, through industry protections, prior to trade reform. Competition from foreign producers depresses prices, wages and employment in previously protected activities. The other two cells depict influences of trade reforms through markets for nontradables, i.e., goods for which markets are effectively missing for poor households or their local economies. Local nontradables may include consumer goods, intermediate goods, or factors. For example, if trade integration stimulates the demand for livestock, this may increase ranchers' demand for grain, both for human consumption and feed. If the local economy is not well integrated with outside grain markets, the increased demand by ranchers may exert upward pressure on local grain prices. This, in turn, might stimulate local grain production, while reducing real incomes of grain consumers and dampening the supply response of livestock. Influences through local factor markets depend critically on the factor intensities of goods stimulated by trade reform. For example, if the activity stimulated by trade reforms is land rather than labor intensive (the case of livestock), its expansion will tend to put upward pressure on land prices while dampening local wages. If the activity stimulated by trade reforms is labor intensive, however, local wages may increase as labor demand rises. If trade reforms decrease, rather than increasing, demand for labor- intensive agricultural goods (e.g., coffee or sugar cane in Central America), the negative effect on local wages will be amplified by the labor intensity of these activities. Wage and employment effects transfer impacts of the trade reform to worker and landholder households--and possibly to urban labor markets, through migration. These impacts may be positive or negative, as indicated in the table. Understanding the structure of local and regional markets is critical for tracing out the likely impacts of trade reforms on rural economies and on poverty. Recently, a nascent body of research has begun to explore local impacts of policy changes using "micro" economy- wide modeling techniques (Taylor, Yúnez-Naude, and Dyer, 1998; Taylor and Adelman, 1996). These studies take into account the ways in which agricultural households and firms interact in local markets, even when high transaction costs isolate them from larger, regional 7 and global markets. Micro economy-wide modeling, the basis for this paper, goes beyond both the aggregate country and micro household focus, elucidating market structures and the complex linkages that connect economic actors in the regional economies of which they are part. Part 2 presents a micro economy-wide modeling approach designed for this purpose. 2 Micro Economy-wide Models: A Graphical and Conceptual Presentation When local economies are closely integrated with outside markets, their prices are given, assuming that local actors are not influential within those markets. Local production and consumption decisions are then guided by exogenous prices. Changes in local demand affect marketed surplus available to outside markets, but they do not affect local production--with all prices given, the conditions for profit maximization do not change when local demand changes. However, when transaction costs insulate local economies from outside markets, local demand and supply are linked by endogenous, local prices. In this case, exogenous changes in local demands affect prices, and thus, production decisions. The result is a web of local economic linkages that transmit influences of policy changes among households and firms and unleash general-equilibrium effects in the local economy. Microeconomic models focusing on households, firms, or household-firms (Singh, Squire and Strauss, 1986), including those in imperfect market environments (de Janvry, et al., 1991), miss these general-equilibrium effects. Economy wide models, including computable general equilibrium (CGE) models, are designed to capture the second and higher-round feedbacks of policy changes. However, national CGE models abstract from local economies, and they do not provide the detail needed to reliably uncover the full impact of policy changes on small economies, particularly when households are simultaneously engaged in a "portfolio" of diverse activities. Our small or "micro" economy-wide modeling uses an adaptation of village-wide modeling techniques presented in Taylor and Adelman (1996). It blends microeconomic analysis with economy-wide modeling, offering an alternative to both micro (household, firm, and household-farm) and aggregate CGE models. Consider the effect of a change in an exogenous variable Z (e.g., a trade policy reform) on an endogenous variable (or vector) Y (e.g., production, income of a particular household group, or migration). Let P denote a vector of local input and output prices. The full impact of the change in Z on Y is given by: dY/dZ = Y/Z + Y/P dP/dZ (1) The first term represents direct income effects, an economy-wide analogue to the partial effects in a microeconomic model in which all prices are held constant. The second term represents the indirect, general-equilibrium effects of the exogenous shock through endogenous local prices. If all goods and factors are tradable (that is, all prices are given to the local economy by outside markets), or if supplies of all goods and services are perfectly elastic (as in a Social Accounting Matrix multiplier model), the second term vanishes. In this 8 case, a series of microeconomic models of households and firms (or, in the case of perfectly elastic supplies, a SAM multiplier model) may be sufficient to estimate local production, marketed-surplus, and income effects of the policy change. However, if some goods (e.g., labor, output) are non-tradable and supplies are not perfectly elastic, the second term in Equation (1) may be nonzero. Market linkages resulting from endogenous prices alter the effects of policy reforms in small economies. Figure 1 illustrates linkages among economic actors in local economies. Some agents are affected by trade or policy shock directly (A). Through their interactions in local economies (e.g., production, consumption, or investment expenditures), these directly affected actors transmit influences of the shock to others inside the rural economy. The indirectly affected actors, in turn, interact with others, transmitting impacts further into the rural economy. Any one of these indirectly affected actors may have linkages back to the directly affected actor. In this way, influences of the trade or policy shock swirl through the rural economy, reaching an ever widening circle of economic actors. At each round of this transmission process, part of the impact of the shock exits the rural economy, through various kinds of leakages. The most important leakages are through rural household (and household- farm) demand for manufactured inputs and especially consumption goods. These are the basis for farm-nonfarm linkages in developing country economies (e.g., Mellor, 1976). Micro economywide models are flexible and may include a large variety of economic actors. Production activity mixes, factors, and household groups reflect both the structure of the local economy and the researcher's interest. The production side of micro economywide models includes a focus on diverse activities in which rural households may be engaged, including (depending upon the context) staples, cash crops, livestock, and nonagricultural activities, including commerce and other services. Production activities purchase factor inputs explicitly or, in the case of family inputs, implicitly, from inside or outside the local economy and generate value-added. The technological relationship between factor inputs and output in each sector is nonlinear, increasing with quantity of factor inputs but at a decreasing rate, as described by sector- specific production functions. Typically, factors include labor, capital, and land--or any disaggregation of these. Prices of factors for which there are markets (e.g., hired labor) can be observed. However, by definition, markets do not exist for family factors--or, more accurately, markets for family factors exist only within households--so prices for these factors cannot be observed directly. Family-factor prices and value-added are estimated econometrically from time use information and the difference between gross value of production and the cost of all purchased inputs. The factor accounts in the model channel value-added into households, in proportion to households' shares of factor supplies. Models typically include multiple household groups, classified according to some criterion (principal income source or income level at the time of the survey, land tenure, migration experience, etc.). In addition to the endogenous accounts summarized above, micro economywide models may contain two groups of exogenous accounts, government and the rest of the 9 world, and a savings-investment account that is either exogenous or endogenous, depending upon the structure of local capital markets. The government accounts may include local, state, and federal governments. These public institutions tax rural residents and channel revenue into local public-sector activities (including some services) or else into public expenditures outside the local economy. The rest of the world typically includes the rest of the country and the world abroad. With few exceptions, the relevant rest of the world abroad for rural residents of Mexico and Central America is the United States, with which migration connections typically are strong. Savings accounts gather savings from activities and households, channeling them into investments in physical or other (e.g., human) capital (schooling) activities. Three rural capital market scenarios or "closures" are possible. First, in the unlikely case that rural households have perfect access to outside capital markets, the capital account is exogenous; local savings and investment demand can be decoupled from each other. Second, at the other extreme, rural credit markets may be missing altogether. In this case, each household's savings limit its investments--households must self-finance all projects. A third possibility is that local (e.g., informal) capital markets exist but are not integrated with outside (regional or national) capital markets. In this case, savings may flow among households in the local economy, but total savings limit total investments. In each of the latter two cases, the capital account is endogenous, because savings from outside the local economy are not available. For either a household or an entire local economy, when the supply of a particular factor exceeds demand, summed across all production activities, one of two things can happen, depending upon access to markets for the factor. The first possibility is that excess supply of the factor is marketed outside the household or local economy, at existing factor prices (e.g., fixed wages). The second possibility is that the market for the factor is imperfect, limited by high transaction costs (e.g., of monitoring workers) or by policies (e.g., limitations in land use, as under Mexico's land-reform-sector, or ejido, laws prior to recent reforms). Under this second possibility, two scenarios are possible. The first is that individual households do not have access to factor markets and thus are constrained to be self-sufficient in the factor. This case corresponds to missing markets at the household level elucidated by Strauss (1986) and de Janvry, et al. (1991). The second scenario is that households have access to local factor markets that are isolated from regional or national markets by high transaction costs. With micro economy wide models, either or both of these two scenarios can be modeled. In addition to markets for factors, markets for goods must also clear, either through interactions of supply and demand at the household or local level, or else by using outside markets to sell excess supply or satisfy excess demand for goods. The market-clearing conditions determine equilibrium quantities and prices (for each nontradable) or marketed surplus (for each tradable). A trade equation constrains the value of local "imports" or purchases of goods and services from the outside world to equal total "exports" or sales to outside markets minus net borrowing. The trade equation represents the redundant equation in these models. 10 A Stylized Rural Economy-wide (STREC) Model The economy-wide model used in the trade policy experiments below consists of separate micro agricultural household models for diverse household groups integrated into a local general-equilibrium framework. That is, the building blocks of our rural micro economy-wide model are micro models of firms and households engaged in a variety of economic activities that are of intrinsic policy interest and that may be influenced directly or indirectly by policy changes. The STREC model was designed to explore micro economywide impacts of specific market and policy changes in a diversity of rural economic contexts characteristic of Mexico and Central America. It is essentially a hybrid, selectively drawing elements from several small-economy models that have been estimated for rural Mexico and Central America over the past decade.2 Although this is a stylized model, all parameters were derived from past models and estimated using original survey data. The elements of the model include six production sectors, chosen to reflect commodities and technologies that exist in rural Mexico and Central America and are likely to be affected differentially by trade policy and market reforms. The six sectors include: · Two staple production activities. The first is labor intensive, modeled on traditional ox-and-plow maize and beans cultivation in Michoacán and Oaxaca, Mexico, and El Salvador. The second reflects more capital intensive production found in some commercial agricultural regions of Mexico and Central America. It is characterized by greater use of machinery, chemical inputs, and modern seed varieties. The production function for this activity was parameterized from surveys in Coahuila and Jalisco, Mexico. · Two cash crop activities. One is a highly labor intensive activity, modeled on chile production in the Sierra Norte region of the Mexican state of Puebla. The other is a somewhat less labor intensive activity, with production parameters estimated from coffee and sugar production in Puebla and Jalisco, respectively. This activity is capital intensive only in a relative sense; it nevertheless employs a considerable amount of labor. · Livestock production. This activity is intensive in land and capital (animal stocks). Parameters for livestock production functions were obtained from village household surveys in El Salvador, Michoacán, and Puebla. Labor shares in value-added are very low in this activity in all survey sites. ·Nonagricultural production. This is a highly heterogeneous sector comprised mostly of local services (retail, construction, etc.) together with some small-scale manufacturing. The production function for this sector uses factor value-added shares averaged from nonfarm accounts in the micro economywide models cited earlier. This sector has considerable backward linkages outside of agriculture (viz., the small store that purchases 2These include Taylor, Yúnez-Naude and Dyer (1999); Taylor, Zabin and Eckhoff, 1999; Taylor, Yúnez- Naude and Hampton, 1999; and Becerril, et al., 1996 and 1997. 11 goods from regional commercial centers). It thus represents a major leakage from the rural economy. In addition to these production activities, households may supply labor to local labor markets or to migration. Internal and international migration are included as separate activities in the model. Both may compete with local activities--especially labor intensive ones--for labor, and they generate remittance income for migrant-sending households. The chief distinction between the two migration types is that international migration usually entails remittances denominated in foreign currency and thus is affected directly by change rates. This is typically the only activity for which rural households receive direct payment in foreign currency. Factors in the model include the key inputs in rural farm and nonfarm production: labor, land, and capital. Three major household groups dominate the socioeconomic landscape of rural Mexico and Central America. They are: · Landless households, which receive value-added income only from labor activities but may supplement this by participating in domestic or international migration. These households usually have low incomes and are vulnerable on the consumption side to changes in food prices. Trade reforms have potentially conflicting impacts on landless households, first on the income side, through labor markets, and second by altering prices of consumption goods. · Commercial households typically are relatively high-income households exposed to trade policy reforms primarily on the production side. Food usually represents a small share of commercial household budgets. This means that agricultural trade reforms are likely to have a relatively small effect on commercial households through prices of consumption goods. · Subsistence households might be viewed as a mixture of the two extremes of landless and commercial households. They are smallholders for whom production and consumption are interrelated and potentially simultaneous, as depicted in agricultural household models (e.g., Singh, Squire and Strauss, 1986). As such, these households potentially are affected by trade reforms in complex ways involving both production and consumption. When prices of subsistence households' output decline, the impact on welfare is ambiguous. As producers, they lose, but as consumers they benefit. In this way, subsistence households are more akin to landless than to commercial households. Production technologies are specified as Cobb-Douglas, and consumption demands are modeled using a linear expenditure system (LES) approach. Although more complicated functional forms are possible, our experience with micro economy-wide models suggests that little is gained from the use of alternative functional forms (and necessary "guestimates" of accompanying elasticities). We have found the results of our policy experiments using similar models robust to the specification of functional forms (Taylor, Yunez and Hampton, 12 1999). This is not surprising, inasmuch as the model is always calibrated at the same point given by the survey data, and most policy experiments involve marginal changes in exogenous variables. An advantage of Cobb-Douglas production functions is that they are nonlinear yet relatively simple to implement; under the assumption of profit maximization, output elasticities are equivalent to factor value-added shares obtained directly from establishment and household survey data. The base models solve for local equilibrium prices and quantities of all goods and factors. The trade policy experiments are then run on this base. Micro computable general equilibrium models overcome the principal limitations of fixed-price, including social accounting matrix (SAM) multiplier models, by incorporating price effects, nonlinearities, and resource constraints. All parameters in the micro CGE models presented below were estimated using survey data. We view this as an advantage over aggregate (e.g., national) CGE models, which often rely on assumed parameters and outmoded data. In any general equilibrium model, results tend to be sensitive to model closure assumptions. In our experiments, we explore the sensitivity of findings to various market closure specifications. Sensitivity analysis of other market specifications can be explored. In general, the more open local markets are to the outside, the more the indirect economic impacts of policy changes are transferred outside the local economy. Parameterizing the STREC Model The most critical parameters needed for the household component of our model are value-added shares, which link household incomes to production; expenditure shares, which shape household demand linkages inside and outside the rural economy; and migrant remittance elasticities, which relate migration to remittance receipts. In this model, landless households receive value-added only from labor, and commercial households receive value- added from capital and land. Subsistence households receive a mix of value-added incomes. For each production activity, the shares of labor, capital, and land value added accruing to different household groups were obtained from past models that included the activity. Budget shares for each commodity-household combination and saving shares were obtained from past models in which both the commodity and household were included. Remittance elasticities were obtained from two-stage regressions of remittances on family time (proxied by number of migrants) allocated to internal and international migration, correcting for sample selecivity bias that may result from zero observations for some households (Taylor, 1987). Key model parameters appear in Table 2. The top panel of the Table reports average factor shares in value-added. Labor value-added shares range from a high of 0.60 (for labor intensive cash crops) to a low of 0.10 (for livestock). Value-added shares are critical in translating price and other shocks on particular production sectors into changes in different value-added concepts. They, together with shares of value-added from each production activity accruing to each of the three household groups, determine how production shocks 13 translate into changes in income of different household groups. The bottom panel of Table 2 summarizes budget shares for each household group. Changes in household incomes reverborate through the rural economy like ripples in a pond. Household expenditure patterns determine the size and direction of these ripples. Large budget shares for locally produced goods create a potential for household income changes to stimulate local production activities. For nontradables, local prices transmit changes in demand to production activities. For tradables, prices are determined in markets outside the local economy. Thus, local demand does not affect production, but it does determine the size of the net surplus available to outside markets. Tradables, Nontradables, and Local Price Transmission Goods or factors are tradables if outside prices are transmitted perfectly into the economy that is being modeled--in the present case, the rural economy represented by the STREC model. Perfect price transmission requires well functioning rural markets with low transaction costs. The extent to which outside prices are transmitted to rural economies is empirical, ranging from perfect to nil. Because of this, goods or factors that are tradable in one region (or for one household within a region; see de Janvry, Fafchamps and Sadoulet, 1991) may be nontradable in another (Taylor and Adelman, 1996). To date, there has been little effort to empirically test for price transmission in rural economies (exceptions are Larson, 2002, and Rozelle, 2002). There is evidence, however, that imperfect price transmission results in isolated markets for factors and goods in some areas (Valdez, 2002). For example, in Mexico, where the government supported prices for basic grains in the 1990s, high transaction costs prevented most farmers from benefiting by selling harvests to the government (Taylor, 2002). The extent to which price supports may have influenced local prices indurectly (e.g., through traders) is not known. The structure of our model permits us to explore the impacts of a variety of trade and market shocks on production, incomes, migration and trade in alternative market contexts. 3 Simulations of Impacts of Market and Policy Shocks The STREC model was used to explore the rural economic impacts of four types of policy and market shocks under two staple and cash crop production technologies and two staple-price transmission regimes. The policy and market shocks we explore are: 1. Changes in prices of staples produced using traditional (labor intensive) and modern (capital intensive) technologies 2. Changes in prices of cash crops cultivated under labor intensive and capital intensive technologies 3. Government direct income transfers to rural households 4. Local Currency devaluation 14 Impacts of each of these policy and market shocks are explored under two different price transmission scenarios. The first assumes perfect transmission of staple prices into the local economy. In the second scenario, there is an endogenous staple price implying minimal or no transmission of outside staple prices into the local economy. Only the first scenario (perfect price transmission) is possible for the first experiment, because the modeler (like the policy maker) is only able to change prices over which s/he has control in the model (or real world). This does not include endogenous prices of staples that are nontradable. All scenarios presented here assume an endogenous rural wage. This means that wages are determined within the model. The sensitivity of any or all of these assumptions can be explored by changing closure conditions in the model. (For implications of fixed versus endogenous wages, see "Policy Shocks and Rural Market Imperfections," below.) The results of the seven policy shock/transmission scenarios are presented in Tables 3-6. All simulations assume increases in prices or positive income transfers. Qualitatively and to a large extent quantitatively, the results of our simulations are symmetric; that is, the signs of effects reported in the tables change for price decreases or negative income transfers (taxes). The impact of trade reforms on commodity prices, of course, may be either positive or negative, depending upon the degree of protection enjoyed by the commodity prior to reforms. The findings reported here make it possible to explore the likely directions of impacts of policy or market shocks once this degree of a-priori protection is known. Perfect Market Scenarios Staple Price Shocks Impacts of North American trade reforms on staple prices are critical because of the relative efficiency of staple production in the United States and Canada and the importance of staples in production, incomes, and consumption in Mexico and Central America, especially of the rural and urban poor. Our staple price experiments explore the impacts of changes in staple prices on production, incomes, migration and trade under alternative staple- production technology assumptions in a diversified rural economy represented by the STREC model. Table 3 reports simulated results of 10-percent increases in staple prices. If trade reforms reduced rather than increased staple prices, the magnitudes of impacts would be similar but of a different sign. Column A reports results of a 10-percent increase in the price of traditional staples, Column B, the results of a 10-percent increase in the price of modern staples, and column C, a 10-percent increase in the prices of both kinds of staples. Under each technology scenario, the immediate impact of the staple price change is to increase the profitability of staple production. Producers respond by raising output. In each case, there is an increase (of 5 to 7 percent) in production of the staple whose price rises. Increasing staple output requires intermediate inputs and factors. The demand for factors used most intensively by staple producers increases disproportionately (labor, in the traditional staples case, and capital, for modern staples). This puts upward pressure on factor prices. In the three experiments depicted in Table 3, wages increase by around 0.1 to 0.4 percent in response to the 10-percent staple price increase. Higher rural wages, in turn, 15 discourage migration (by just under 0.3 to 1.1 percent) and associated remittances (by 0.2 to 0.7 percent). Factor supply constraints transmit (generally negative) impacts to other production sectors that compete with staples for factor inputs. In all staple price experiments, production of other tradables (including the other staple, cash crops, and livestock) decreases. Hardest hit are the sectors making the most intensive use of inputs demanded by the sector whose price goes up. In all three experiments, it is evident that staple production competes with labor-intensive cash crops. In Experiments A and B, output of labor-intensive cash crops falls by 0.14 to 0.56 percent. Livestock production, the least labor-intensive sector, is least adversely affected by the staple price change. Nominal incomes in households supplying factors to staple production, in turn, go up. However, higher staple prices also increase consumption costs, with negative effects on real incomes. By far the largest beneficiaries of higher staple prices are commercial households, where incomes increase in both nominal and real terms. For landless and subsistence households, nominal incomes increase, but because of the importance of staples for consumption, real incomes fall. There is a second round of demand linkages, as commercial households spend their new-found income on goods and services inside and outside the rural economy while landless and subsistence households attempt to reorient their consumption away from staples. Higher demand for tradables is transmitted outside the rural economy through various types of external market linkages. Marketed surplus increases sharply for traditional staples in Experiment 1 (given a negative base surplus) and for modern staples in Experiment 2. Higher rural production and especially consumption demand for manufactures increases external trade (by 0.09 to 0.50 percentage points). Overall, the findings in Table 3 suggest that impacts of the staple price change on production, incomes, migration, and trade are sensitive to technologies used in staple production. They also reveal income effects that are much smaller in percentage terms than the original 10 percent increases in staple prices. Relatively small impacts of price changes on rural incomes reflect the high level of diversification in the economy; all in all, with many different production (and migration) activities, a given change in staple prices has a disproportionately small impact on household incomes. Income effects are also mitigated by cross-effects of the price change on other activities, including nonstaple production and migration. Studies in the 1990s of the likely impacts of NAFTA in rural Mexico tended to focus on maize production in isolation of other rural production activities, thereby exaggerating estimated effects of NAFTA on the rural economy and on migration (see Taylor, Yúnez-Naude and Dyer, 1999). If trade integration reduces staple prices in Mexico and Central America, impacts on rural incomes will be relatively small. In fact, in real terms, many (particularly landless) households could benefit from staple price declines, as their food prices decrease. Nevertheless, these households will have to make up for lost income from staple production by shifting to new production activities and/or to migration. 16 Changes in Cash Crop Prices Impacts of trade integration on farmgate cash crop prices depend upon the extent to which cash crops are protected from outside prices prior to trade reforms as well as the extent to which prices are transmitted through rural markets. North American integration opens up new potential external markets for cash crops, as evidenced by Mexico's rising agricultural exports to the United States post-NAFTA. However, it also creates potential competitors for local cash crop producers--e.g., Central American sugar producers facing competition from Mexico under an extended NAFTA. Our cash crop price experiments explore the impacts of changes in cash crop prices under alternative market scenarios. Table 4 reports simulated impacts of 10-percent increases in cash crop prices on the rural economy represented by the model, assuming perfect price transmission in markets for cash crops, staples, and livestock. The findings in this table mirror those in Table 3. For the cash crop benefiting from the price increase, output increases, driving up local wages and pulling resources away from the production of other tradables (the other cash crop, livestock, and staples). The increase in price of the labor-intensive cash crop has a relatively large effect on wages (0.96 percent), negative impact on migration (-2.75 percent), and income benefit to landless households (just under 1 percent). The capital-intensive cash crop price increase, on the other hand, nudges wages upward only slightly (by 0.21 percent), has a small positive effect on landless household incomes (0.l6 percent), and decreases migration by less than in the labor-intensive case (0.62 percent). Predictably, labor-intensive cash crops compete most with traditional staples, whose output falls by more than 2 percent. In each of the cash-crop price experiments, the trade balance in staples, or marketed surplus, deteriorates, as the demand for staples increases (a result of rising incomes) while staple production falls. These findings illustrate the importance of technology, particularly labor intensity of cash-crop production, in shaping local economywide impacts of trade reforms. If market integration primarily affects labor-intensive cash crop production, its impacts on the incomes of landless households, local wages, and migration--whether positive or negative--are likely to be relatively large. On the other hand, if trade reforms primarily affect capital-intensive cash crop or livestock activities, the impacts on rural incomes, poverty, and migration-- whether positive or negative--will tend to be muted. Policies that encourage capital and land intensive activities at the expense of labor intensive ones could even decrease rural labor demands, depressing wages and stimulating migration. Rural Income Subsidies The negotiation of a North American Free Trade Agreement (NAFTA) in the early 1990s presented an opportunity to integrate the liberalization of Mexico's maize sector with that taking place in the rest of the economy and to harness peasant resources more efficiently in other sectors of the economy (Tellez, 1993; Bartra, 2000; Nadal, 2001). The cornerstone of the maize sector's liberalization consisted of the disappearance of support prices (then paid through the state trading agency, CONASUPO) and the simultaneous removal of trade 17 barriers, allowing imported US corn to fill the gap between domestic supply and demand. The government addressed the anticipated adverse income effects of lower staple prices (see staple price experiment, above) through its PROCAMPO program. Under PROCAMPO, rural maize-producing households receive cash transfers designed to compensate them for lower grain prices without violating the rules of current international trade agreements (i.e., NAFTA and GATT). PROCAMPO was designed to be a decoupling policy; that is, income subsidies were not expected to affect current production. We used the STREC model to explore the likely impacts of government income subsidies to rural households in a micro economywide framework. Table 5 reports the results of simulated income transfers to each of the three household groups. Under PROCAMPO, payments are made on a per-hectare basis. In our experiments, for transparency, we set income transfers equal to 10% of household base incomes. This permits us to test the extent to which direct income payments to rural households are truly "decoupled," while providing a benchmark for understanding the full impact of transfers on rural incomes, including indirect effects. Column A in Table 5 presents simulated impacts of a 10-percent income transfer to landless households; Columns B and C present impacts of transfers to the other two household groups. The results in Table 5 cast doubt on whether decoupling schemes really are decoupled in rural areas. Income transfers, other things being equal, stimulate demand and exert upward pressure on wages and prices of nontradable commodities in local economies. Wages represent a crucial factor-market link between households and production. Higher wages adversely affect production in all sectors, especially in the most labor-intensive activities. Demand linkages transmit benefits from the transfer-receiving households to others in the rural economy. For example, the transfer to landless households increases subsistence-household income by .4 percent and commercial-household incomes by nearly .5 percent. Transfers to subsistence households contribute 2.6 percent to commercial-household incomes. Because of their access to value-added from local production activities, subsistence and especially commercial households are the largest beneficiaries of indirect linkages resulting from the income transfers. The diagonal elements in the "Household Incomes" rows of Table 5 reveal that, in each case, the 10-percent income transfer results in a greater- than-10-percent increase in the recipient households' income. Income growth stimulated by the transfer reduces marketed surplus (by 1.6 to 12 percent) and increases rural-urban trade linkages (by 2.2 to 4.1 percent). These income experiments offer insight into the indirect influences of income changes in rural economies, through the linkages discussed earlier. The direct effects of the income transfers are limited, in each case, to the 10-percent increase in income for the single household group targeted by the transfer. All other impacts in Table 5 represent indirect, or "ripple," effects. Because local economies adjust both to changes in output prices and compensating income transfers, it is difficult to design an income-transfer scheme that is truly income-neutral. In the case of PROCAMPO, it is clear that subsidy payments overcompensated farmers for income lost as a result of the removal of staple price supports (e.g., see Taylor, Yúnez-Naude and Dyer, 1999). 18 Currency Devaluation In addition to altering the external market environment for participating countries, regional trade integration also has potentially far-reaching ramifications for macroeconomic policy, as illustrated by U.S. efforts to stabilize the Mexican peso in the 1990s. Most research on impacts of currency devaluation, like most trade studies, have a national rather than intra-national focus. Micro economywide models make it possible to explore the ways in which currency devaluations, like policy changes, play out within rural economies. In the STREC model, as in most Mexican and Central American villages, labor is the only export for which rural households receive payment denominated in foreign currency (typically, U.S. dollars). Devaluations increase the value of migrant remittances in local currency. This has two impacts in the model. First, it creates an immediate income transfer to remittance-receiving households. Second, it has a migration effect, increasing the attractiveness (the marginal effect of migration on remittances) of international (but not internal) migration by an amount equal to the percentage exchange-rate devaluation. This stimulates international migration, by making both local wage work and internal migration relatively unattractive. The simulation results reported in Table 6 represent the sum of transfer and migration impacts, once they have been shaped by local economywide effects. The 10-percent exchange rate devaluation stimulates international migration by 8.3 percent and remittances from abroad by 4.7 percent. This exerts upward pressure on local wages. Internal migration and output of all tradables production fall sharply. The household-income effects of remittances are significant, ranging from 1.5 percent for commercial households to 6 percent in landless households. Marketed surplus of staples decreases. The devaluation stimulates trade linkages with outside markets (by just under 2 percent). As a migration as well as an exchange-rate experiment, this simulation provides insight into how rural economies reshape themselves around migration, shifting resources out of labor-intensive tradables production and increasing their commercial ties with outside product markets. It reveals potential "Dutch disease" effects of migration on local economies, at least in the short run. In the long run, migrant remittances may have positive effects on local economies by stimulating investments in local production activities (Stark, 1991; Taylor and Martin, 2000). In rural Mexico, increased participation in migration has been accompanied by expansion of livestock over time, an activity that is both profitable (because of rising meat demands stimulated by income growth and market development) and complementary with migration (because of its minimal demands for labor; see Taylor, 1992 and Fletcher, 1999). Policy Shocks and Rural Market Imperfections Transmission of the impacts of trade policies through the rural economy depends critically upon the structure of rural markets for cash crops as well as for other goods and 19 factors of production. High transaction costs endemic to many rural economies isolate local economies from outside markets for some goods and/or factors. When this happens, the affected goods or factors become nontradable, in the sense that their prices are determined through the interaction of local supply and demand instead of being transmitted exogenously from outside (e.g., world) markets. Policy or market shocks that stimulate production (e.g., of cash crops) or increase demand (e.g., government or migrant income transfers) alter the prices of local nontradables. This transmits the influence of the policy or market shock to new production activities, and through factor markets, to new household groups. That is, nontradables create income linkages within the rural economy. Ironically, this can result in larger rural income multipliers than are evident in open rural economies, because changes in demand affect local production of nontradables directly, rather than leaking out of local economies through rural-urban trade. However, the increased local linkages come at a cost, in two forms. First, by trapping land, capital and labor in local nontradables production, high transaction costs reduce production efficiency. That is, they inhibit local economies from concentrating resources in the activities in which they enjoy a comparative advantage and thus maximizing their gains from trade. Second, external shocks that stimulate the demand for local nontradables often lead to inflation, which reduces household real incomes. The experiments presented below explore the ways in which selected rural market imperfections may alter the impacts of trade policy and market shocks on local economies in the STREC model. For each experiment conducted earlier (except for the staple-price experiment), we compare the rural economywide impacts of the policy or market change (a) when all rural prices, including wages, are exogenously transmitted from external markets, and (b) when prices of staples and labor are endogenously determined by the interplay of local supply and demand, as would be the case for economies facing high costs of transacting with outside markets. As we shall see, market imperfections have far-reaching implications for how trade reforms play out in rural economies. Table 9 compares the results of the previous experiments under these two market- structure (or in modeling jargon, closure) scenarios. Panel I compares the results of the cash- crop price simulations under the assumptions of (a) perfect and (b) missing external markets for staples and labor. Panels II and III present comparisons for the household income transfer and exchange rate experiments. The implications of market imperfections are clear by comparing columns (a) and (b) under each experiment. If capital is fixed and all inputs and outputs are perfectly tradable, with exogenous prices set by external markets, there is no mechanism to transmit the impacts of market or policy shocks from the directly affected agent (sector or household) to other production sectors in the economy. Production decisions in each activity are determined by profit maximizing behavior, based on given technologies and input and output prices. Nothing in the profit-maximization calculus changes when all prices are transmitted to local producers from external markets. This explains the zero effects of policy and market shocks on production in sectors not directly affected by the shocks in Columns (a) of Table 7. The difference between these results and the (generally negative) cross-sector results reported in 20 previous tables is due to the fact that local wages were assumed to be endogenous in the earlier experiments but set by external labor markets here.3 In the cash crop experiment, with access to perfect markets for all goods, firms and households can freely shift resources out of staple and livestock production in response to new market opportunities in cash crops. Shortages created by increased demand and/or decreased supply of staples or other goods are alleviated by trade with outside markets at given prices. This creates a high supply response in cash crops. Household incomes increase as households benefit from higher factor incomes from cash crops. The distribution of impacts across households in column (a) reflects the distribution of factor incomes across household groups. The higher cash-crop prices simulate external trade linkages, as marketed surplus of staples decreases sharply and as nonagricultural trade rises. Lacking access to external markets for labor and other goods, the local economy faces self-sufficiency constraints that create tradeoffs and competition between nontradables and cash crop production. In the absence of perfect access to outside labor, wages increase, and in the absence of access to outside staple markets, higher incomes bid up staple prices. In combination, higher wages and other prices induce households to keep resources in other production activities, dampening the cash-crop supply response. The presence of nontradables reduces production efficiency in the economy. The impact of the cash-crop price increase on household incomes is now lower, especially in real terms, constrained by the lack of access to outside markets. In one case (landless households, which have a large staple budget share), real income actually falls. Local market imperfections also shape the impacts of government (Panel II) and migrant (Panel III) income transfers in rural economies. With perfect access to outside markets (II(a)), transfers that stimulate household demand increase purchases from outside markets, without affecting local production. In contrast, when high transaction costs make some local goods and factors nontradable (II(b)), increases in demand put upward pressure on local prices and influence production. The impacts on production may be negative, if prices of key inputs (e.g., labor) rise, or else positive, if higher demand raises output prices. Endogenous local prices create local income multipliers, as well as price inflation, which tend to magnify impacts on rural economies. Comparing columns (a) and (b) in Experiment II, it is clear that direct income payments to households are "decoupled" from production only if the economy is a price taker in all markets--that is, perfectly integrated with external markets from which all prices are transmitted. When local market imperfections exist, one can no longer say that "decoupling" schemes truly are decoupled. Migration, like government income transfers, has different impacts on rural economies under alternative market structures. Under all market scenarios, the exchange rate devaluation (Panel III) creates a de-facto income transfer to migrant-sending households and stimulates new migration. When the local economy has perfect access to all markets, including labor, households can increase migration without reducing production, by hiring labor to take the place of those who migrate. In scenario (a) under Panel III, although 3Alternatively, this assumption could reflect surplus labor and an institutionally set rural wage, as in a Lewis world. 21 migration increases, there is no adverse effect on production. If households do not have perfect access to external labor markets, they must shift labor out of other activities in order to increase migration. Households can reduce production to increase migration provided that they have access to external markets to satisfy food demands. Otherwise, the need to satisfy local food demands will exacerbate the tradeoff between migration and production. In scenario (b), all production activities are affected by the currency devaluation, even though the only direct impact of devaluation is on migration. Part of the impact of migration is on food and factor price inflation, a pattern well known to students of migration and development in LDCs. These experiments explore the implications of only a few rural market imperfections. In real life, rural economic actors are likely to face imperfections in multiple markets, including labor and some commodities as well as capital, land, insurance, and information. The results in Table 7 offer a glimpse into the variety of implications that these rural market imperfections may have for production efficiency, incomes, and the potential for broad-based growth. 4 Conclusions The results of the policy experiments under alternative market scenarios, presented above, reveal difficult policy tradeoffs. Cash crop production is critical for market integration of rural economies in Mexico and Central America, because this is where most agricultural trade opportunities are likely to be found, in contrast to staples, given the relative efficiency of US and Canadian staples production which places the comparative advantage in staple production squarely to the north. Table 8 summarizes the volume and growth of Central America's agricultural exports to the United States between 2000 and 2001. It illustrates the increasing importance of horticultural exports for Central American economies: they increased 13 percent between FY 2000 and FY 2001 and constituted 30 percent of all agricultural exports from Central America to the United States in 2001. Only forest products and livestock exports grew more rapidly during this period (16 percent each). Livestock production, however, employs little labor and thus has limited potential to stimulate employment and benefit landless laborer households in rural areas. Rural economies' ability to respond to new market opportunities depends critically on their access to markets not only for cash crops, but also for other goods and factors. Lack of access to outside markets for staples and other goods limits rural economies' response to new cash-crop opportunities. Because of this, lowering transaction costs or raising productivity in other production activities may be critical to facilitate adjustment to trade and market reforms and ensure gains from trade for those engaged in cash crop production. Nevertheless, high transaction costs, by protecting some local production from outside market competition, also benefit some rural households, possibly including the rural poor if an important share of their income comes from these market and policy-protected activities. Individuals engaged in activities protected either by policies (e.g., staple price 22 supports) or by market imperfections (high transaction costs) stand to lose from trade reform and market development. This creates a rationale to implement compensatory schemes, such as Mexico's PROCAMPO income transfers to rural households. Impacts of income transfers on rural economies, like trade and market reforms, depend critically on the structure of local markets. Although local market imperfections may magnify the impacts of government transfers on local production and incomes, they also increase consumption costs and deny rural households many of the consumption-side gains from trade with new markets. Taking advantage of lower prices for consumer goods requires having access to cash income. If trade and market reforms adversely affect labor-intensive production without significantly expanding labor demand in other rural activities, securing access to cash income may require migration. The results of all of the simulations presented above reveal potentially important migration effects of trade policy changes. Exchange rates are among the most important variables directly affecting international migration. The migration response to currency devaluations is positive and large. This makes the impacts of trade and market reforms on migration potentially complex. On one hand, reforms inherently imply structural adjustments in rural economies that may displace labor and stimulate migration, at least in the short run (Martin, 1993). On the other hand, if regional market integration stabilizes exchange rates, it may reduce incentives for international (though not necessarily internal) migration. NAFTA's positive impact on macroeconomic stability in Mexico in this way may have discouraged Mexico-to-U.S. migration. Rural economywide modeling methods highlight linkages within rural economies and the ways in which these linkages, together with rural market structures, shape policy and market outcomes. Directly affected agents transmit the influences of migration to others in the rural economy. Understanding these linkages is critical to exploring how trade policies and market reforms play out within nations. 23 REFERENCES Becerril García, J., G. Dyer Leal, J. E. Taylor, and A Yúnez-Naude. 1996. "Elaboración de Matrices de Contabilidad Social para poblaciones agropecuarias: El caso de El Chante, Jalisco." Mexico City: El Colegio de Mexico, Documento de Trabajo (September). Becerril García, J., V. Evangelista and R. Martínez Villagrán. 1997. "Elaboracion De Una Matriz De Contabilidad Social Y Analisis De Multiplicadores Para La Comunidad De Naupan, Puebla." Mexico City: El Colegio de Mexico, Documento de Trabajo (September). de Janvry, A., M. Fafchamps and E. Sadoulet. 1991. "Peasant Household Behavior with Missing Markets: Some Paradoxes Explained." The Economic Journal 101:1400-1417. Cornelius, W.A. and P.L. Martin. 1993. The Uncertain Connection: Free Trade and Mexico-to- U.S. Migration. San Diego: Center for U.S.-Mexican Studies, University of California, San Diego. Fletcher, Peri L. 1999. La Casa de mis Sueños: Dreams of Home in a Transnational Migrant Community. Boulder: Westview Press. Hinojosa-Ojeda, R. and S. Robinson. 1992. "Labor Issues in a North American Free Trade Area." In N. Lustig, B. Bosworth, and R. Lawrence, eds., North American Free Trade: Assessing the Impact. Washington, DC: Brookings Institution. International Monetary Fund (IMF). 1999. Statistical Yearbook. Washington DC: IMF. Larson, Don. 2002. "Exchange rate pass-through and domestic price transmission: Methodology and application." OECD Global Forum On Agriculture: Agricultural Trade Reform, Adjustment And Poverty, Paris, 23-24 (May), CCNM/GF/AGR(2002)9 Levy, S. and S. Wijnberger (1992). Mexican Agriculture in the Free Trade Agreement: Transition Problems in Economic Reform. OECD/Gd(92) 77 Technical Paper No. 63. Paris: Organization for Economic Cooperation and Development. Lucas, Robert E.B., 1987, Emigration to South Africa's Mines, American Economic Review 77:313-30. Martin, P.L. 1993. Trade and Migration: NAFTA and Agriculture. Washington, D.C.: Institute for International Economics. Massey, D.S., Joaquin Arango, Graeme Hugo, Ali Kouaouci, Adela Pellegrino, and J. Edward Taylor. 1998. Worlds in Motion: Understanding International Migration at the End of the Millennium. Oxford: Clarendon Press. 24 Mellor, J. The New Economics of Growth. Ithaca: Cornell University Press, 1976. Rozelle, S. 2002. "The Nature of Distortions to Agricultural Incentives in China and Implications of WTO Accession." UC Davis, Department of Agricultural and Resource Economics, Working Paper. Rozelle, S., J.E. Taylor, and A. de Brauw, 1999, Migration, remittances, and productivity in China, American Economic Review 89(2), 287-91. SAGAR (Mexican Ministry of Agriculture). 1994-1997. On-line Data Bases. Singh, I., L. Squire and J. Strauss, eds. 1986. Agricultural Household Models, Extensions, Applications and Policy. The World Bank and The Johns Hopkins University Press. Stark, O. 1991. The Migration of Labor. Cambridge: Basil Blackwell. _________. "Economic Organization, Information and Development," in Chenery, H. and Srinivasan, T.N., eds., Handbook of Development Economics, Volume I. Elsevier Science Publishers, 1988. Strauss, J. "Appendix: The Theory and Comparative Statics of Agricultural Household Models: A General Approach," in Singh, Squire and Strauss (eds.), Agricultural Household Models, Extensions, Applications and Policy. World Bank and The Johns Hopkins University Press, 1986. Taylor, J.E. 1999. "The New Economics of Labour Migration and the Role of Remittances in the Development Process." International Migration 37(1):63-88. _________. 1995. Micro Economywide Models for Migration and Policy Analysis. An Application to Rural Mexico. Paris: Organisation for Economic Co-operation and Development (OECD). _________. 1992. Remittances and Inequality Reconsidered: Direct, Indirect and Intertemporal Effects, Journal of Policy Modeling 14:187-208. Taylor, J.E. and I. Adelman. 1996. Village Economies: The Design, Estimation, and Use of Villagewide Economic Models. Cambridge: Cambridge University Press. Taylor, J.E., Antonio Yúnez-Naude and Steve Hampton. 1999. "Agricultural Policy Reforms and Village Economies: A CGE Analysis from Mexico." Journal of Policy Modeling 21(4):453-480. Taylor, J.E., Antonio Yúnez-Naude and George Dyer. 1999. "Agricultural Price Policy, Employment, and Migration in a Diversified Rural Economy: A Village-Town CGE 25 Analysis from Mexico." American Journal of Agricultural Economics 81:653-662, August 1999. Valdés, Alberto. 2002. "Agricultural price distortions and price transmission in developing and transition countries." OECD Global Forum On Agriculture: Agricultural Trade Reform, Adjustment And Poverty, Paris, 23-24 (May), CCNM/GF/AGR(2002)7. Yúnez-Naude, A. 2001. "The Dismantling of CONASUPO, a Mexican State Trader in Agriculture." El Colegio de Mexico, Centro de Estudios Económicos (unpublished paper, May). 26 Table 1. Direct and Indirect Influences of Trade Reforms on Rural Incomes Protection for Local Affect of Trade Reforms Through Markets for... Production Prior to Trade Reform Tradables Nontradables Negative + +/- Positive - +/- 27 Table 2. Factor Shares in Value-Added and Household Expenditure Shares, By Activity Production Activity Basic Grains Cash Crops Nonagri- cultural Labor Capital Non- Traditional Modern Intensive Intensive Livestock tradables Factor Shares in Value-Added Labor 0.40 0.35 0.60 0.32 0.10 0.20 Animal Capital 0.04 0.05 0.05 0.03 0.27 0.00 Physical Capital 0.41 0.30 0.21 0.23 0.04 0.80 Land 0.15 0.30 0.14 0.42 0.59 0.00 Total 1.00 1.00 1.00 1.00 1.00 1.00 Household Expenditure Shares Household Group Good Landless Subsistence Commercial Basic Grains BGLI 0.14 0.05 0.02 BGKI 0.08 0.05 0.02 LVV 0.06 0.07 0.04 NONAGV 0.33 0.37 0.49 Non-local 0.31 0.36 0.33 Savings CAPITAL 0.07 0.08 0.07 HKSAV 0.01 0.02 0.03 Total Savings 0.08 0.10 0.10 Total 1.00 1.00 1.00 28 Table 3. Simulated Impacts of Changes in Staple Prices on Rural Production, Incomes, Migration, and Trade, Under Perfect Price Transmission Percentage Effects of 10% Increase in Price of... A B C Traditional Modern Staple All Staples Sector or Household Group Staple Village Production (Prices) Staples Traditional 6.65 -0.25 6.38 (10.0) (10.0) Modern -0.05 5.05 5.00 (10.0) (10.0) Cash Crops Labor-Intensive -0.14 -0.56 -0.69 Capital-intensive -0.05 -0.18 -0.22 Livestock Production -0.03 -0.09 -0.11 Nonagricultural Production -0.02 -0.09 -0.12 Wages and Migration Wages 0.09 0.37 0.47 Migration -0.27 -1.09 -1.35 Remittances -0.18 -0.72 -0.89 Household Incomes Landless 0.07(-0.53) 0.29(-0.31) 0.36(-0.83) Subsistence 0.08(-0.87) 0.44(-0.51) 0.52(-1.37) Commercial 0.58(-0.24) 0.81(0.44) 0.94(0.19) Market Linkages Marketed Surplus of Staples Traditional -15.05* 0.79* -14.31* Modern -0.34 35.04 34.71 Rural-Urban Linkages 0.09* 0.51* 0.59* (nonagricultural trade volume) * Denotes that marketed surplus is negative in the base model. A large percentage change in marketed surplus reflects a small marketed surplus in the base. 29 Table 4. Simulated Impacts of Changes in Cash Crop Prices on Rural Production, Incomes, Migration, and Trade, Under Perfect Price Transmission Percentage Effects of 10% Increase in Price of... A B C Labor-Intensive Capital-Intensive All Cash Crops Sector or Household Group Cash Crops Cash Crops Village Production (Prices) Staples Traditional -2.26 -0.45 -1.00 Modern -0.76 -0.92 -1.62 Cash Crops Labor-Intensive 15.59 -1.31 15.37 (10.0) (10.0) Capital-intensive -0.63 4.59 4.59 (10.0) (10.0) Livestock Production -0.15 -1.58 -2.10 Nonagricultural Production -0.02 -0.05 -0.07 Wages and Migration Wages 0.96 0.21 0.76 Migration -2.75 -0.62 -2.19 Remittances -1.82 -0.41 -1.45 Household Incomes Landless 0.76 0.16 0.59 Subsistence 0.11 0.16 0.34 Commercial 0.55 0.37 0.82 Market Linkages Marketed Surplus of Staples Traditional 1.19* 0.32* 0.86* Modern -3.87 -2.92 -5.66 Rural-Urban Linkages 0.51* 0.86* 1.23* (nonagricultural trade volume) * Denotes that marketed surplus is negative in the base model. A large percentage change in marketed surplus reflects a small marketed surplus in the base. 30 Table 5. Simulated Impacts of Income Transfers on Rural Production, Incomes, Migration, and Trade, Under Perfect Price Transmission Percentage Effects of Direct Subsidy Equal to 10% of Base Income to... A B C Landless Subsistence Commercial Sector or Household Group Households Households Households Village Production (Prices) Staples Traditional -0.24 -1.38 -1.47 Modern -0.08 -0.46 -0.49 Cash Crops Labor-Intensive -0.42 -2.43 -2.59 Capital-intensive -0.07 -0.38 -0.41 Livestock Production -0.02 -0.09 -0.10 Nonagricultural Production 0.25 1.44 1.53 (0.90) (5.23) (5.57) Wages and Migration Wages 0.10 0.58 0.62 Migration -0.29 -1.68 -1.79 Remittances -0.19 -1.11 -1.18 Household Incomes Landless 10.08 0.46 0.49 Subsistence 0.37 12.20 2.34 Commercial 0.45 2.64 12.81 Market Linkages Marketed Surplus of Staples Traditional 1.63* 12.16* 6.50* Modern -3.58 -26.40 -14.56 Rural-Urban Linkages 2.16* 4.08* 4.41* (nonagricultural trade volume) * Denotes that marketed surplus is negative in the base model. A large percentage change in marketed surplus reflects a small marketed surplus in the base. 31 Table 6. Simulated Impacts of an Exchange Rate Devaluation on Rural Production, Incomes, Migration, and Trade Sector Percentage Effects of a 10% Currency Devaluation Village Production (Prices) Staples Traditional -4.13 Modern -3.28 Cash Crops Labor-Intensive -8.87 Capital-intensive -2.87 Livestock Production -0.70 Nonagricultural Production -1.54 Wages and Migration Wages 6.39 Migration International 8.27 Internal -13.71 Remittances International 4.72 Internal -8.20 Household Incomes Landless 6.09 Subsistence 3.78 Commercial 1.52 Market Linkages Marketed Surplus of Staples Traditional 6.42* Modern -18.09 Rural-Urban Linkages 1.88* (nonagricultural trade volume) * Denotes a negative value in the base model. A large percentage change in marketed surplus reflects a small marketed surplus in the base. 32 t 8 0 1 cyner ation 65.1- ) (4.23 28.0- ) (5.05 32.9- 99.2- 67.0- 06.1- 76.6 06.7 43.4 ) ) ) Imperfect 6.26(5.6 4.02(3.1 1.96(1.6 Cur Devalu 0 0 0 0 0 0 A/N Perfec 74.52 70.41 97.31 56.8 91.5 t Rate laicre 72.1 )39.1( 46.0 )52.1( 90.0- 30.0- 20.0- 20.0- 60.0 81.0- 21.0- Imperfec )41.0-(50.0 )42.0-(70.0 ) 10.12(9.99 Comm Exchange me*ig sr tcefreP 0 0 0 0 0 0 A/N 0 0 0 0 01 and Re sfe n anrT t ) ) )2 ec 77.3 98.1 91.0 Transfere smission come Tra In tensisbuS )87.5( )47.3( 82.0- 90.0- 50.0- 50.0- 55.0- 63.0- Imperfec 0.15(-0.42 10.21(9.21 0.37(0.0 t Incom Price holdes 0 0 0 0 0 0 A/N 0 0 0 01 0 Perfec nda ) ) Hou t Crop, Cash sse 94.0 )57.0( 52.0 )84.0( )0 40.0- 10.0- 10.0- 10.0- 20.0 70.0- 50.0- Imperfec 10.02(9.94 0.03(-0.09 0.05(0.0 for Experiment Landl t 0 0 0 0 0 0 A/N 0 0 01 0 0 Perfec Markets se les Increa Stap tcefrep 90.0- ) (0.30 50.0- ) 83.4 50.0- 81.0- 24.0 32.1- 18.0- in mI (0.34 46.41 Price )01.1-(33.0 33 )31.0(91.0 )84.0(15.0 t ission Crop 0 0 sh Perfec 73.51 95.4 0 70.0- A/N 0 0 39.0 55.0 67.0 ansmrT Ca Price perfect Im of (Prices) e Production rationg Incomes plications Mi Im ntse la tensive tensiv Production s Production and 7. Crops ercial esg gesa ittance Household Experim Table Village Staples Tradition Modern Cash Labor-In Capital-in Livestock Nonagricultural Wa W Migration Rem Landless Subsistence Comm A/N A/N 61.2 81.11 -16.89 01.6 A/N A/N 27.2 60.3 26.4- 46.2 A/N A/N 96.2 12.9 19.31- 54.2 A/N A/N 29.1 81.1 87.1- 98.1 34 A/N A/N 14.0 48.0 72.1- 17.0 e) wages and volum Staples prices of trade Linkages staples Surplus Linkages Traditional Modern Market Marketed Rural-Urban (nonagricultural *Endogenous Table 8. Exports of Agricultural Products to the United States by Members of Proposed U.S./Central America Free Trade Agreement, 2000-2001 Products Rank 2001 2001 % Change 2000-2001 Horticultural Products 1 $596,645 13% Fishery Products 2 $487,261 6% Tobacco & Products 3 $89,435 2% Forest Products 4 $66,791 16% Livestock and Meats 5 $57,247 16% All Agricultural Export $1,914,553 -10% Source: U.S. Department of Commerce, Bureau of the Census 35 jr 1 jr 2 jr 3 jr 4 jr M (C) Acto Acto Acto Acto Acto ts en Ag s formeR Affectedy )B( ir 1 iro 2 ir N licyoP rectlid Acto Act Acto In of 36 cesne flunIe y t en wid (A) rectl y- Di Affected Ag om Econ Local 1. ore ge re guiF adrT licy Po anhC