75070 v3 Unlocking Central America’s Export Potential 3. Unlocking potential in rural areas: geographic analysis Finance and Private Sector Development Department Central America Country Management Unit Latin America and the Caribbean Region The World Bank October 2012 1 Acknowledgements This report was prepared by Maximo Torero (International Food Policy Research Institute, IFPRI), with assistance from Maribel Elias, Camila Alva and Marines Grandes. The methodology is based on Elias, Maruyama, and Torero 2009. The report was directed by Thomas Haven (World Bank) and we gratefully acknowledge the financing from the Development Economics Group (DEC) research funds and the Poverty and Social Impact Analysis (PSIA) trust fund that made it possible. 2 Table of contents Summary ......................................................................................................................................... 4 Introduction ..................................................................................................................................... 6 1. The stochastic profit frontier approach ................................................................................ 8 1.1 Why use stochastic profit frontiers to construct a new typology? ................................ 8 1.2 Theoretical framework ................................................................................................. 9 2. Market Accessibility analysis ............................................................................................ 10 2.1 The Accessibility Model ............................................................................................. 10 2.2 Accessibility results .................................................................................................... 14 3. The model and estimation .................................................................................................. 15 3.1 The model ................................................................................................................... 15 3.2 Estimation ................................................................................................................... 16 3.3 The distinction between production inputs and environmental factors ...................... 17 3.4 Moving from household level estimations to spatial analysis .................................... 17 3.5 The Data used for the implementation ....................................................................... 19 3.6 Empirical estimation ................................................................................................... 19 4. Results for Honduras, Nicaragua and Panama ................................................................... 24 4.1 Profit frontier estimation ............................................................................................ 24 4.2 Constructing the typology .......................................................................................... 30 4.3 Typology results ......................................................................................................... 31 Conclusions ................................................................................................................................... 45 Bibliography ................................................................................................................................. 46 3 Summary Using a ground-breaking and innovative approach, this section uses a geographic lens to shed light on an area not covered by the sector analyses: specific geographies with potential to improve their productivity, and ultimately, exports. Given the size of the rural economy (over half of the population in some countries), concentration of poverty in rural areas, and importance of agriculture in the region’s export basket, the analysis is a powerful tool to allow policymakers to prioritize investments in agriculture and rural areas. The analysis characterizes micro-regions of Honduras, Nicaragua, and Panama1 along four dimensions: poverty levels, agricultural potential, average farm efficiency, and market access. Each micro-region2 is assigned to a category, e.g. areas with high poverty, high potential, and lower efficiency would be considered “high priority�. Figure A shows all of the categories for Nicaragua. Figure B breaks out the micro-regions identified as being “high priority�, while adding the market access dimension. Having a precise identification of such areas allows interventions, such as agricultural extension, to be targeted to optimize agricultural potential. This analysis utilizes an econometrically rigorous stochastic profit frontier estimation approach, and introduces a new and innovative tool to policymakers in Central America. 1 A similar analysis was undertaken for Guatemala and can be found in the 2010 World Bank report entitled SME Development in Guatemala: Let 10,000 Firms Bloom. 2 It should be noted that this methodology can be conducted with differing definitions of regions. While the current analysis is essentially of micro-regions, it can be calculated for different definitions in subsequent studies. 4 Figure A: Categories of Micro-Regions for Nicaragua3 Figure B: “High priority� areas in Nicaragua Source: Authors’ analysis. 3 There are several smaller protected regions in Nicaragua that are not reflected in the maps. These small protected regions may influence the categorization of certain micro-regions. 5 Introduction This section uses a geographic lens to shed light on an area not covered by the sector analyses: specific geographies with potential to improve their productivity, and ultimately, exports. Given the size of the rural economy (over half of the population in some countries), concentration of poverty in rural areas, and importance of agriculture in the region’s export basket, this is a powerful tool to allow policymakers to prioritize investments in agriculture and rural areas. The section uses mapping technology and a variety of data to divide rural Honduras, Nicaragua and Panama into a typology of micro-regions that differ according to their characteristics, problems, and potential for development. The typology is based on relevant criteria, including climate and topography, production, access to roads and markets, off-farm job opportunities, population density, gender distribution and the presence of various institutions (formal and informal), such as credit providers. The analysis takes advantage of the availability of rich biophysical data on the geography of each country and a highly detailed geo-referenced household survey of the region to construct our typology. These data sources are combined to estimate the efficiency and potential of local farmers. The identification of the productive potential and efficiency is achieved through the estimation of an econometric stochastic profit frontier model that takes into account indicators of socioeconomic and market conditions as well as biophysical and accessibility factors. These indicators explain a big portion of the heterogeneity among rural households. An accurate classification of the areas in terms of agricultural potential is crucial to guide the type of interventions, which could be oriented to productive development, to market creation (agricultural or not agricultural) or even to basic social assistance. Figure 1: Advantages of a typology of micro-regions Productive projects differentiated to Conditional Cash Transfers and meet local needs and problems Nutritional Programs The inclusion of socioeconomic characteristics and access in the analysis allows for the What are the principal differences identification of bottlenecks in between high and low efficiency areas of high potential but low households in the area? or medium efficiency Productive and Efficiency High potential and low Low potential and low potential based on market, Typology average efficiency socioeconomic, bio-physical average efficiency and access characteristics. Diagnostic from High poverty areas High poverty areas Poverty map As shown in Figure 1, the typology of micro-regions, once built, can be combined with other relevant information such as malnutrition and poverty maps in order to provide a more detailed 6 diagnosis of the needs and the potential solutions for the distinct rural areas of a country. Table 1 is an example of the classifications that can be obtained mixing potential and efficiency with malnutrition or poverty. For example, we could identify areas with high levels of malnutrition or high poverty (left part of table 1). In addition, if those areas were of low productive potential, independently of its level of efficiency (red part of table 1), conditional cash transfer and nutritional programs would be recommended, at least in the short term. However, if those areas were of high/medium potential (dark green part of table 1), production strategies with (if necessary) nutritional programs should be promoted, according to its level of efficiency4. Table 1: Example of a 3-dimensional classification Micro-Regions Poverty Potential Efficiency Critical, lacking agricultural potential High Low High-Medium-Low Medium priority, no agricultural opportunities Medium Low High-Medium-Low Low priority Low Low High-Medium-Low High priority High Medium-High High-Medium-Low Medium priority, with agricultural opportunities Medium Medium-High Medium-Low Low priority, with agricultural opportunities Low Medium-High Medium-low High performance Low Medium-High High Armed with this typology, policymakers can geographically target areas in which there is both potential and inefficiency; appropriate policies will reduce these inefficiencies in production. Similarly, the typology could identify areas in which the only alternative, given the existing low potential of the land, is to reduce poverty though rural labor programs or safety nets programs. Figure 2 tries to summarize these alternatives. Figure 2: Prioritizing interventions based on agricultural potential and malnutrition Cash Prioritize transfers in agricultural the short-run; Potential of small holders development Agricultural interventions development in the long-run Agricultural Prioritize Cash non-agricultural transfers in the short-run; interventions Non-agricultural (rural labor development development) in the long-run Poverty 4 It is possible to obtain a more detailed characterization of each area in order to recommend ad-hoc policies for each particular reality. 7 1. The stochastic profit frontier approach 1.1 Why use stochastic profit frontiers to construct a new typology? Rural households in developing countries are extremely diverse in their economic characteristics. This diversity occurs for several reasons: (i) the heterogeneity in the quantity and quality of their assets; (ii) the technologies available to the household; (iii) the transaction costs in markets for outputs and inputs; (iv) the credit and financial constraints; (v) the access to public goods and services; and (vi) the local agro-ecological and biophysical conditions. Rural development policies need to take this heterogeneity into account in order to be effective, and generate a demand for analytical tools that combine socioeconomic information with mapping technology. Among the variety of typologies used to categorize territories, poverty maps are arguably the most widely developed because they allow policymakers to design spatially targeted poverty alleviation programs (see Elbers, Fujii, Lanjouw, Ozler, and Yin (2004)). By imputing consumption and income values from survey data estimations and extrapolating them to census data (see Elbers, Lanjouw, and Lanjouw (2003)), poverty maps give reasonable static diagnostics of welfare. This is extremely useful in ranking areas from poorest to richest when designing a transfer program. It is much less useful, however, in deciding how to invest resources when designing poverty alleviation programs. Another common tool used to construct typologies is cluster analysis. Cluster analysis methods have become popular because they are data-driven and can be used without formulating rigorous models to define the factors that determine the chosen welfare measure to be analyzed. When used to construct a typology to characterize communities’ welfare or economic performance, however, the resulting index offers very little information on what policies should be implemented to improve the current conditions of these regions because the groups are not constructed ordering all variables monotonically (ascending or descending), generating confusion about the interpretation of the results. Hence, cluster analysis works well only when differences are determined over a small and relatively homogenous group of variables. In this paper, we attempt to construct a typology that takes into account the heterogeneity of small farmers while at the same time being strongly based on economic foundations. By recognizing the fact those farmers are productive units optimizing an objective function subject to a set of constraints, the stochastic profit frontier analysis deals with many of these issues. The efficiency indicator is a continuous measure, similar to a score, and its interpretation is direct and simple. The functional form used for its estimation is flexible and imposes a limited structure on the analysis. Moreover, it is possible (with more and better data than what we have available at this point) to calculate the stochastic profit frontier using non-parametric estimation, in which case no parametric form is imposed for the analysis. Finally, the theory behind stochastic profit frontier estimation methods is standard theory, involving a constrained optimization process and allowing for random shocks, a setup that is suitable to model a farmer’s decision-making process and analyze the opportunities and challenges he faces. Profit frontiers have been used to estimate farm efficiency levels in developing countries. Using data for Basmati rice producers in Pakistan, Ali and Flinn (1989) find a mean level of profit inefficiency of 28 percent associated with the household’s education, nonagricultural 8 employment, credit constraints, water constraints, and late application of fertilizer. Also in Pakistan, Ali, Parikh, and Shah (1994) find an average farm profit inefficiency of 24 percent; they also find that the size of holding, fragmentation of land, subsistence needs, and higher age of farmers contribute positively to inefficiency. Rahman (2003) finds a mean level of profit inefficiency of 23 percent among Bangladeshi rice farmers, explained largely by infrastructure, soil fertility, experience, extension services, tenancy, and share of non-agricultural income. Using data for Chinese farm households, Wang, Wailes, and Cramer (1996) find a 39 percent mean profit inefficiency level, influenced by farmers’ resource endowment, education, family size, per capita net income, and family ties with village leaders. All these studies, however, treat farms as single-output firms. For the purposes of our typology, it is essential to work with multiple output profit frontier as this is a more realistic depiction of the farmers’ decision - making process. In that respect, we are unaware of studies that use the stochastic profit frontier approach in a multiple output farm setting. 1.2 Theoretical framework A major caveat encountered when designing a comprehensive typology of the communities in Honduras, Nicaragua and Panama is dealing with the difference between the economic potential of an area and its current observed status. For instance, a detailed diagnosis describing the present situation of local economies will help to identify the most deprived areas (i.e. poverty maps), but some of these areas might have already reached their maximum economic potential given current conditions, and short run investments will have little or no impact on the welfare of the local population. Hence, for policy making, a useful typology of local communities needs to differentiate and combine the notions of current status and optimal potential. In other words, it has to take into account the idea that agents’ attempt to optimize, but do not always succeed. The stochastic frontier approach provides an ideal framework to build the typology. Conceptually, it is developed from a theory of producer behavior in which the motivation is the standard optimization criteria (minimize costs or maximize profits), but in which success is not guaranteed. The associated estimation procedures allow for failures in efforts to optimize and different degrees of success between producers. This opens up the possibility of analyzing the determinants of variation in the efficiency with which producers pursue their objectives. Taking the farm as an example, the stochastic frontier approach accounts for the fact that, conditional on geographic location, elements such as prices, biophysical conditions, and (in the short run) fixed inputs are exogenous to the farmer’s decision process. Hence, holding this factors fixed there exists an optimal production technology and production plan that generates the maximum profit the farmer can obtain. With this approach it is possible to identify where the profit frontier lies, and how much of the difference between it and observed profits (i.e. profit loss) can be explained by the farmer’s choices resulting in profit inefficiencies. Adding the stochastic component allows for a better fit to the farm production process, which is very sensitive to unpredictable changes in exogenous conditions, like weather or international prices. In this context, profit inefficiency is defined as the monetary loss which results from not operating at the frontier given the prices and levels of fixed production factors faced by the farm. 9 2. Market Accessibility analysis GIS data has recently made it possible to investigate the market access question in a more sophisticated way. With this data one can calculate the shortest time or distance from any village to a regional or local market using the distance traveled on different road surfaces combined with an impedance measure which reflects the speed one can travel on roads of different qualities and on the slope of the terrain through which the road passes. The resulting market access measure can be expressed as a weighted average of the distance traveled on each type of road, where the weights are proportional to the impedance factor (see attachment 1 on details on how this is measured). There are two problems with these measures of access. The first is that they do not incorporate transportation costs which may well vary with distance and type of road surface in a different way than does time. Where that is the case, the time based measure will be misleading because it could imply that for a particular village, one market is closer than another in that it takes less time to get there even though it may cost more to get there. By the same token it could imply that one village is “closer� to market than another as measured by time but not when measured by cost. But presumably what the farmer wants to know is not how far it is to his market, but rather how much he can sell his produce for in that market or equivalently, what his farm gate price is, net of transportation cost. In this paper we use a measure that incorporates both aspects and report our measure of the merge market distance data for each village in one of the three countries where we had all the necessary information, i.e. Honduras, with a matrix of transportation costs by truck on two different classes of roads, and on rivers or by animal on trails where there are no roads This gives us a measure of accessibility in terms of costs (see Annex 1 for details on the costs used for the case of Honduras). The second problem with the typical market access indicator is that it considers only the local market. But the level of prices in local markets may well vary according to how far they are from the country’s largest market. It could well be that a farmer would get a higher price for his products by shipping them to a market which, while further from his village, is closer to the capital city or equivalently, in which the price of his product is higher. Therefore we estimate the costs to simultaneously access to the local market and to the largest market, being the variable reported the one that minimize the cost to access to both markets simultaneously. 2.1 The Accessibility Model To “connect� every household with the closest market we constructed a series of accessibility indicators. The notion behind them is that accessibility is not a discrete variable (i.e. have or not have access), but a continuum that reflects the difficulties each household faces when trying to access different types of infrastructure. This accessibility analysis was applied to the entire land surface of Honduras, Nicaragua and Panama. Accessibility is defined as how feasible it is to reach a location from others, considering factors like distance, moving costs, type of transportation and time. In other words, it is any indicator of effort to reach or access a particular location. The base of this analysis is assuming that people are likely to move through highways, major roads, or paths in the case those exist, but otherwise would walk their way around to the nearest market. The final objective is calculating the time a person invests on reaching the nearest market through the fastest way 10 The moving time on the land surface depends on different factors, the most important one being the distance, but there are other important factors like the existing road network and its specific characteristics, the slope and the presence of obstacles like rivers (except for those cases where rivers are used as a means of transport). The accessibility analysis was developed on a raster format, which means that the entire area of analysis was converted into a grid of cells measuring 92.6 by 92.6 meters. Each cell was assigned a “friction� value based on characteristics of slope, roads, and barriers, which allowed each cell to be allotted a value for the time required to reach the nearest facility (Figures 3 and 4). Having created the friction grid, the cost weighted distance algorithm runs over the raster surface, calculating the accumulated time departing from each market available replacing overlapping values with the least time consuming route. Figure 3: Friction surface between points A and B Figure 4: Values indicating the difficulty of crossing a “cell� The first variable is the slope, which has been used to calculate a walking travel speed that depends indirectly from it. Tobler’s (1993) Walking velocity has three variations, one corresponding to a footpath, another to a horseback and finally one for off path. The horseback walking velocity has been assigned to the dirt road tracks, while the footpaths velocity to 11 walking trails, on the case there are not even paths available the off path walking velocity has been assigned. The following calculations resulted, where (see Figure 5): Figure 5: Calculation of slope Walking velocity on footpath = [6 × exp(−3.5 × abs(S +0.05))] Walking velocity on horseback = [6 × exp(−3.5 × abs(S +0.05))] × 1.25 Walking velocity offpath = [6 × exp(−3.5 × abs(S +0.05))] × 0.6 The following table presents the results for each of the road classifications: Figure 6: Times calculated only with the off path walking velocity 12 The second variable used in this analysis was transportation infrastructure, of which Honduras, Nicaragua and Panama has two major kinds: paved roads, and unpaved roads. In addition there are some few rivers though which navigation occurs. Each type of road was assigned an average travel speed, and the corresponding cell given a crossing time in seconds:     CellCros sin gTime( Seconds )  92.6   1    1000    Speed ( Km / hr )      3600   Average Speed cell Crossing travel KM/Hr Time in Seconds Paved Road 60 5 Unpaved Road 30 11 Navigation in River 10 33 The third and final variable used in this model corresponds to the presence of natural barriers as rivers, which prevent people from traveling a straight line if there is no bridge. Cells corresponding to areas with a river and no bridge are assigned a travel time 10 times their value, so that the crossing would only be considered where a bridge is available. Once the friction model is built and each cell has been allocated a travel time value, cost- weighted distance algorithms are run over the raster surface, calculating the accumulated time required to travel a particular route, choosing the one that is least time-consuming as in Figure 7. This information is then used to simulate the impacts of improvements of road segments. Specifically, if a road is improved from a walking trail to a dirt road track then the new average speed is assigned of the upgraded category and re-estimates all the accessibility measures. Figure 7: Friction Surface Map 13 Finally, and only for the case of Honduras, were able to combine the friction surface map of the time requires to travel with a matrix of transportation costs by truck on two different classes of roads, and on rivers or by animal on trails where there are no roads This gives us a measure of accessibility in terms of costs (see Annex 1 for details on the costs used for the case of Honduras). 2.2 Accessibility results Following the methodology explained in the previous sections we constructed the accessibility maps for each of the three countries. Figure 8 presents the results for the three countries. Figure 8: Accessibility (a) Honduras (b) Nicaragua Note: Accessibility to markets of more than 50,000 habitants 14 (c) Panama 3. The model and estimation 3.1 The model Let x denote a (1 ×m) vector of variable and quasi-fixed inputs and y denote a (1 × q) vector of multiple outputs involved in the farm production process. Let z denote a (1 × r) vector of environmental variables that, though not directly determining the farmer’s profits, could affect the farm’s performance. We will discuss later in this section our criteria to place specific variables as elements of x or z.5 Let be the set of feasible production plans of the farm. We define a measure of output technical inefficiency δ (Farrell 1957) for some production plan such that: δ0 = δ(x0,y0 |P) ≡ sup{δ | (x0,y0) P,δ > 0} (1) For (x0,y0) P,δ(x0,y0 |P) ≥ 1. We now define the restricted profit function π(p, z, δ) as the maximum profit attainable by a farm with characteristics z, facing output prices p P (z) and input prices w W (z): (2) Let πi be the observed profits for farmer i. The analyst is confronted with a set of observations (πi,pi,wi,zi) for i =1,...,n, which are realizations of identically, independently distributed random variables with probability density function f(π, p, w, z). This function has support over . 5 Deprins and Simar (1989) and Kumbhakar and Lovell (2000) discuss the rationale for placing certain variables as elements of x or z, admitting this issue is frequently a judgment call. In many cases, it is not obvious whether an exogenous variable is a characteristic of production technology or a determinant of productive efficiency. 15 We assume z is not independent from (π, p, w), i.e. f(π, p, w | z) f(π, p, w). This means that the constraints on farmers’ choices of prices p and w, and on observed profits π, due to the environmental variables z the farms face operate through the dependence of (π, p, w) on z in f(π, p, w, z). There exist several ways to formulate the model such that the production set is dependent on z (Coelli, Rao, and Battese 1998), however, we consider it is more appropriate given the empirical setup we are analyzing to assume the environmental variables z influence the mean and variance of the inefficiency process, but not the boundary of its support. Hence, in our formulation the conditioning in f(δi | zi) operates through the following mechanism: δi = exp(ziβ + εi) (3) where β is a vector of parameters, and εi is a continuous i.i.d. random variable, independent of zi.6 We assume the term εi is distributed N(0, ) with left truncation at −ziβ for each i. 3.2 Estimation Because the effect of covariates z operates through the dependence between π and z induced by equation 3, these assumptions provide a rationale for second-stage regressions. Kumbhakar and Lovell (2000) and Kumbhakar (1996) provide the typical setup in these cases, defining the stochastic profit frontier function as: πi = g(pi,wi) exp(νi − ξi) (4) where νi is the stochastic noise error and xii is a non-negative random variable associated with inefficiencies in production. Then the profit efficiency of farm i can be defined as: | | ∑ | (5) where Xdi are exogenous (to the production process) variables characterizing the environment in which production occurs and that can be associated with inefficiencies of the farm. As noted by Simar and Wilson (2007), regressing efficiency estimates obtained from maximum likelihood estimation of a parametric model for (p, w, δ) will very likely result in problems for statistical consistency because the covariates in the second-stage regression (z) are correlated with the one-sided error terms from the first stage (in order for there to be a motivation for a second stage).7 Consequently, the likelihood that is maximized is not the correct one, unless one takes account of the correlation structure. In order to do so we estimate (4) in the first stage modeling heteroskedasticity in the one-sided error term ξ as a linear function of a set of 6 See Simar and Wilson (2007) for estimation in a semi-parametric setup. 7 The errors and the covariates in the �rst stage will not be independent if the covariates in the second stage are correlated with the covariates in the �rst stage, which occurs in most empirical applications. 16 covariates. The variance of the technical inefficiency component is then modeled as (6) We use maximum likelihood estimation and a translogarithmic profit function correcting for heteroskedasticity as shown in (6), and then proceed to the second stage estimation of the technical efficiency term ξ on the environmental variables z. 3.3 The distinction between production inputs and en vironmental factors In this section we make explicit our criteria to distinguish quasi-fixed production inputs in x (which also includes variable inputs) from environmental variables z, as in some cases that distinction might seem arbitrary. An input is included in x when there its market is active and prices can be identified.8 In some cases, prices for certain inputs may not exist (or are not available to the analyst), particularly when studying rural poor populations in developing countries. Active markets and monetary transactions for land or weather-based insurance, for instance, are rare in these settings, so it is extremely difficult to find a reliable price for land (of varying qualities) and weather (and climate-risk) preferences. Under these conditions, we believe elements like land size, climatic and biophysical conditions should be included in x in order to capture their direct impact on production as fixed or quasi-fixed inputs, even though the argument can be made that these variables capture failures in the land and risk-coping markets to justify their inclusion in z.9 3.4 Moving from household level estimations to spatial analysis The procedure described in the previous section provides profit efficiency estimates at the farm level. A remaining task is to scale up these results to the regional level which can then be used for purposes of the typology. According to our model, differences in profits are given by differences in crop choices, local prices, biophysical conditions, and farm efficiency (and, there- fore, the exogenous factors affecting it). Hence, the econometric estimation of the model described in sections 3.1 and 3.2 makes it possible to recover technological parameters for the “representative� agricultural producer in rural Honduras, Nicaragua and Panama. As explained earlier, a primary objective towards the construction of the typology is to estimate the profit frontier and efficiency for a given region (e.g. community, district, province or department) in rural Honduras, Nicaragua and Panama. If the appropriate information (prices, 8 If there exists any evidence that these prices might not reflect actual market conditions for all the production units in the sample (due to accessibility problems or spatially incomplete markets) then the farm’s levels or stocks of these inputs can be included in z in order to capture these market failures through their impact on farm efficiency. The idea behind this is that the input price is among the determinants of the production frontier, and the market failures for that particular input influences the efficiency with which producers approach that frontier. 9 Forms of land ownership or non-market mechanisms to smooth consumption, however, should be included in z in order to capture their impact on productive efficiency if it is suspected that these markets do not work properly. 17 biophysical factors, farm characteristics, and factors influencing efficiency) at that area’s level is available, it can just be plugged in the estimated profit function to recover regional efficiency. Price data for key staple and high value commodities produced in each of the three countries comes from our household survey (see data section for specific dates of the household surveys used for each country). The information is extracted from the production balances of the household surveys of each country. Ideally, a multi-output frontier model would be estimated including every single output and input utilized by the farm in the production process. However, data limitations and computational feasibility makes this impossible. Therefore, it is necessary to group outputs and inputs into the categories mentioned in the previous section. Grouping, however, generates other problems. To assign a single price to broad groups as “Fruits� or “Vegetables�, the median price per kilogram of all products in that group for a given region is used. How precise this grouping procedure is will depend on how many products in a group is grown in the region, and on how different the prices of these products are. For example, if green and red apples are the only fruits grown in a region, and their prices are very similar, the median price of all the fruits produced in the region will be an adequate summary statistic. It follows that the choice of how large a region is matters as well. If a region is too large, the risk that the median price is a poorer summary statistic is higher because the probability that more products are included in each group and that there is higher price variability increases. However, if a region is too small the small amount of observations to calculate any reliable measure of central tendency will also be a problem. Given the data, household information is aggregated at the district level, and for those districts with too few observations the province level is used. Other farm or household specific variables used in the estimation procedures are calculated in a similar way, and subject to the same limitations. The biophysical data and the market accessibility data, on the other hand, have been specifically generated to map out in great detail the whole geography for each of the countries, so no aggregation or grouping issues occur. As mentioned before, with the price and farm data aggregated at the adequate level, and the availability of perfectly mapped biophysical and accessibility datasets, all that is left to do is to plug in all this information back into the model to predict profit frontiers and efficiency levels at the regional level. 18 3.5 The Data used for the implementation The following table describes the data sources used for the implementation of the typology of micro-regions. We have not identified any later source of data which have the variables that were needed to implement the typology. Data/Country Nicaragua Panama Honduras Household Survey LSMS EMNV LSMS ENV (2008) LSMS ENCOVI (2004) (2005) Agro ecological Data on land use FAO (2005-2006) Life zones, PRONADERS data collected from the (1998) Ministry of Agriculture Roads, rivers, lakes Solar and Wind DIVA – GIS Roads: Fuente: Energy Resource SOPTRAVI, elaborated Assessment (Swera) by: Sistema Nacional de Información Territorial (SINIT). 1999 Rivers: Instituto Geográfico Nacional. Sistema Nacional de Información Territorial (SINIT). Populated centers http://world- http://world- National Institute of gazetteer.com/ gazetteer.com/ Statistics Poverty data “Mapa de Pobreza “Pobreza y Desigualdad a “Estimación de Indicadores Extrema de Nivel de Distrito y de Pobreza y Desigualdad Nicaragua� (2001). Corregimiento� (2005), a Nivel Municipal en INIDE MEF. Honduras¨ (2001) – BID- INE 3.6 Empirical estimation The methodology employed to calculate the productive potential and the efficiency of the micro- regions, is similar to the one used by the World Bank to estimate poverty maps in which LSMS surveys and census data are combined to take advantage of the information richness of the first one and of the representativeness of the second one. The methodology is composed by two steps, described below and summarized on Figure 6. 19 Step 1: Estimation of the stochastic profit frontier function. Farmer level. The stochastic profit frontier function is defined as: ( ) ( ) (1) Where πij is the utility of the farmer i on the area j; Pij and Wij are the vector of median prices of products and inputs at the regional level faced by the farmer; vij is a two-tailed error or stochastic noise iid distributed with and independent of uij, which is a non-negative random variable associated with the production inefficiency distributed independently with a semi- normal distribution ( ); Zi is the socioeconomic characteristics and includes the farm fix factors (land and capital); Gj represents the biophysical conditions and Aj is the market access cost that the farmer faces. All the variables are at farmer level and obtained from the LSMS survey, with exception of Aj and Gj which comes from secondary data10. Step 2: Prediction of potential and efficiency. Regional level. After obtained the parameters in the step 1, the values of productive potential and efficiency representatives for each region are predicted. In order to obtain significant results, LSMS data is replaced by census data. The productive potential is estimated in a linear way, using a vector of median prices at region level, following the equation presented below: (2) Where y are the parameters obtained in the step 1. Given that the vector of prices is at regional level, there will be one prediction for each region. The efficiency is estimated in a non-linear way according to the following formula: { } ( ) (3) 10 Aj is based on an accessibility model that includes information on diesel prices, road infrastructure, distance, etc. Gj comes from a biophysical database. 20 Given that the efficiency depends on biophysical conditions and that each region can contain more than one biophysical condition, the prediction of the efficiency will be for each region j and for each biophysical condition g within each region, where � is a cumulative normal function and: The variance of the inefficiency, , depends on the biophysical and access conditions (is heteroskedastic). The variance of the random component, , is constant and was estimated in the step 1. is the prediction error of the productive potential: . To obtain unique values of efficiency per region, a weighted average of the efficiencies in each region is calculated. The weights will be assigned according to the total geographical extension that each biophysical condition occupies in each region. Figure 9 details the different steps we followed to estimate the typology. Figure 10 shows a graphic representation of the steps and Figure 10 and 11 provides an example for Honduras. Following we present the results for Honduras, Nicaragua and Panama. Figure 9: Steps of Methodology of Typology of micro regions 21 Figure 10: Implementing the typology – case of Honduras Targeting Criteria Data based on Efficiency Estimated Geo Layers cost of Market Access PPF: Input, Output, Profits Agricultur al Profit Frontier Variable X Variable Z .4 1 Group 1 Group 2 Available datasets: Efficiency .3 Land characteristics, in .9 Cumulative Density biophysical conditions, Density Group 1 .2 Group 2 socioeconomic Agricultural .8 characteristics, assets, Profits .1 market access, etc. .7 0 4 6 8 10 12 0 2 4 6 8 10 values X Values Z 22 Figure 11: Typology implemented for Honduras 23 4. Results for Honduras, Nicaragua and Panama 4.1 Profit frontier estimation Tables 2, 3 and 4 show the results of the frontier estimation for Honduras, Nicaragua and Panama respectively and Tables 5, 6 and 7 the results for the second stage regression of estimated technical inefficiency on environmental variables z for the three countries. The association of these covariates with inefficiency is also related to the level of frontier profits, so the model is fully interacted with the predicted level of potential profits and the coefficients for these interactions are shown in the second column of the respective tables. The correlation between frontier profits and technical inefficiency is negative and significant for Honduras and Nicaragua, indicating that farmers with higher potential are more likely to have higher efficiency levels in our sample. Farmers’ experience is associated with lower inefficiency, but this link weakens for those with higher profit potential. This is not the case in Panama where the correlation between frontier profits and technical efficiency is positive which could be results of the low of intensity of agricultural activities in this country. When analyzing the determinants of inefficiency on Tables 5 to 7 we found that in all cases household size is negative correlated with inefficiency for the case of Honduras and Panama, i.e. the more the number of members the lower the inefficiency. In the case of Nicaragua where we interact household size with plot size the effect with respect to inefficiency becomes positive, meaning that these larger farms are less labor constraint and at the same time they can also offer wages that are attractive enough to override some of the imperfections in local labor markets that small farms cannot. For farms with high potential, access to formal credit is significantly associated with lower levels of inefficiency. However in the case of Nicaragua this is related also to farms with large plot size. In the case of Honduras, land ownership is correlated with higher levels of efficiency (for all farmers but particularly for farmers with high potential), which goes in line with common notions of land ownership opening access to credit and encouraging productive investments. In the case of Nicaragua and Guatemala this is not the case and the opposite effect is obtained. Unfortunately, to arrive to any conclusion on this aspect we would need to have better controls for land quality, as there might be an endogenous component on which land is owned and which is rented, particularly for smallholders. Market Access costs (cost of transportation) and inefficiency is positively and significantly correlated. The association fades for farms with higher potential, which indicates that accessibility is a more important bottleneck for smallholders. Adequate policies that reduce transportation and transaction costs could be progressive if carried out jointly with assistance programs that increase the competitiveness of rural areas compared to urban areas. It is important to note that along this study we use the term “inefficiency� in ways that are consistent with the literature on stochastic frontiers. However, the model and data cannot capture all the complexities of the farm productive process, so the econometric calculation may identify as “inefficient� decisions that are perfectly rational but difficult to explain by the analyst who has incomplete information. For instance, a farmer facing extremely variable climate conditions, may opt for more resilient but less profitable crops in order to reduce the risk of losing all his harvest. 24 If the analyst cannot observe this high variability he will regard the farmer’s decision as sub - optimal. Another example is over-utilization of available resources such as land or water. In the short run, practices that over-exploit productive assets can result in high profits, but in the long run this can cause a premature exhaustion of the productive capacity of the farm. A farmer with better foresight could appear to be “inefficient� for missing a short run high profit opportunity when in fact he is maximizing his long run profits. Unfortunately, until household surveys or other auxiliary sources collect more (and better) information on risk preferences, climate variability over time, price variability of inputs and outputs, conservation practices of productive assets, etc., it will be impossible to differentiate the risk averse or non-depredatory producers from inefficient producers who are not optimizing long run profits. If the appropriate data is collected, then we could add covariates that capture preferences for risk, weather variability, etc. in z as environmental variables and test if they can capture a significant fraction of the variance of the technical inefficiency term. Table 2- Frontier Estimation for Hondurasf Frontier Estimation (dependent variable: lnprofit) lnprice1 2.787*** lnprice2_lnprice4 -0.195 lnprice1_lnfertilizer 0.299 lnwage_lnwage -0.366*** (0.872) (0.167) (0.341) (0.107) lnprice2 -0.571 lnprice2_lnprice5 0.122 lnprice2_lnwage 0.0713 lnwage_lnmanure 0.180 (0.612) (0.0806) (0.202) (0.123) lnprice4 1.699 lnprice2_lnprice6 0.247* lnprice2_lnmanure -0.117** lnwage_lnfertilizer 1.074* (1.675) (0.150) (0.0507) (0.633) lnprice5 -0.537 lnprice2_lnprice7 -0.0157 lnprice2_lnfertilizer 0.0975 lnmanure_lnmanure 0.00727 (0.660) (0.111) (0.219) (0.0320) lnprice6 -3.243*** lnprice4_lnprice4 1.346*** lnprice4_lnwage -0.113 lnmanure_lnfertilizer 0.377*** (1.255) (0.364) (0.512) (0.141) lnprice7 1.371 lnprice4_lnprice5 -0.333* lnprice4_lnmanure -0.0364 lnfertilizer_lnfertilizer -0.484*** (1.282) (0.182) (0.106) (0.177) lnwage 0.977 lnprice4_lnprice6 -0.746** lnprice4_lnfertilizer -0.0334 Profit Constant 8.763*** (0.763) (0.352) (0.326) (1.738) lnmanure -0.703 lnprice4_lnprice7 -0.843*** lnprice5_lnwage 0.0997 lnsig2v -0.315*** (0.439) (0.309) (0.194) (0.103) lnfertilizer -0.676 lnprice5_lnprice5 0.183*** lnprice5_lnmanure 0.103** (2.040) (0.0444) (0.0486) lnprice1_lnprice1 0.0883 lnprice5_lnprice6 -0.111 lnprice5_lnfertilizer -0.496** (0.112) (0.142) (0.212) lnprice1_lnprice2 -0.152 lnprice5_lnprice7 0.416*** lnprice6_lnwage 1.090*** (0.0935) (0.141) (0.383) lnprice1_lnprice4 0.695** lnprice6_lnprice6 0.0170 lnprice6_lnmanure 0.0250 (0.287) (0.0360) (0.0833) lnprice1_lnprice5 -0.275* lnprice6_lnprice7 0.546** lnprice6_lnfertilizer -1.797*** (0.141) (0.219) (0.396) lnprice1_lnprice6 0.281 lnprice7_lnprice7 0.104*** lnprice7_lnwage -0.680* (0.208) (0.0362) (0.378) lnprice1_lnprice7 0.143 lnprice1_lnwage -0.892*** lnprice7_lnmanure -0.220** (0.244) (0.273) (0.0897) lnprice2_lnprice2 0.0331 lnprice1_lnmanure -0.104 lnprice7_lnfertilizer 0.288 (0.0441) (0.0719) (0.359) Note1: Standard Errors in Parenthesis. ***p<0.01, **p<0.05, *p<0.1 Note2: Prices: (1) Fruits, (2) Industrial Crops, (3) Corn, (4) Cereals, (5) Vegetables, (6) Beans, (7) Tubers, (lnwage) Wages, (lnmanure) Manure, (lnfertilizer) Fertilizer 25 Table 3- Frontier Estimation for Nicaragua Frontier estimation (dependent variable: lnprofit) lnprice1 2.029*** lnprice2_lnprice4 -0.667*** lnprice1_lnw1_wage -1.037*** (0.508) (0.250) (0.229) lnprice2 -1.308 lnprice2_lnprice5 -0.572*** lnprice2_lnw1_wage -0.102 (1.232) (0.177) (0.662) lnprice4 -0.224 lnprice2_lnprice6 0.757 lnprice4_lnw1_wage 0.345 (1.015) (0.694) (0.379) lnprice5 -2.369*** lnprice2_lnprice7 -0.817*** lnprice5_lnw1_wage 1.477*** (0.778) (0.232) (0.321) lnprice6 0.243 lnprice4_lnprice4 0.0319 lnprice6_lnw1_wage -0.570 (2.716) (0.123) (0.753) lnprice7 -0.127 lnprice4_lnprice5 0.00620 lnprice7_lnw1_wage 0.398 (0.711) (0.108) (0.297) lnw1_wage 2.205 lnprice4_lnprice6 0.661 lnw1_wage_lnw1_wage -0.238 (2.892) (0.461) (0.549) lnprice1_lnprice1 -0.0918 lnprice4_lnprice7 0.163 lnsig2v -0.347*** (0.0616) (0.145) (0.0359) lnprice1_lnprice2 0.731*** lnprice5_lnprice5 -0.0972 Constant 6.132 (0.137) (0.104) (3.930) lnprice1_lnprice4 -0.0648 lnprice5_lnprice6 -1.373*** (0.0897) (0.316) lnprice1_lnprice5 -0.0382 lnprice5_lnprice7 -0.0582 (0.0757) (0.0969) lnprice1_lnprice6 0.288 lnprice6_lnprice6 0.428 (0.324) (0.460) lnprice1_lnprice7 0.0711 lnprice6_lnprice7 0.318 (0.0809) (0.380) lnprice2_lnprice2 0.273 lnprice7_lnprice7 0.206** (0.301) (0.0893) Note 1: Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 Note 2: Prices: (1) Fruits, (2) Industrial crops, (3) Corn, (4) Cereals, (5) Vegetables, (6) Beans, (7) Tubers 26 Table 4 – Frontier Estimation for Panama Frontier Estimation (dependent variable: lnprofit) lnprice2 -3.150 lnprice3_lnprice5 4.737*** (2.102) (1.773) lnprice3 1.474 lnprice4_lnprice4 2.682*** (2.429) (0.714) lnprice4 -2.516 lnprice4_lnprice5 -2.543** (2.340) (1.276) lnprice5 5.160*** lnprice5_lnprice5 0.141 (1.985) (0.761) lnw_wage -1.355 lnw_wage_lnprice2 2.472 (1.400) (1.595) lnprice2_lnprice2 -0.124 lnw_wage_lnprice3 -0.127 (0.606) (1.600) lnprice2_lnprice3 -4.310** lnw_wage_lnprice4 -1.152 (1.812) (1.040) lnprice2_lnprice4 4.156** lnw_wage_lnprice5 -0.577 (1.916) (0.842) lnprice2_lnprice5 -3.017 lnw_wage_lnw_wage -0.0738 (1.941) (0.358) lnprice3_lnprice3 3.693*** Constant 8.486*** (1.013) (1.767) lnprice3_lnprice4 -5.344** lnsig2v 0.549*** (2.338) (0.169) Note1: Standard Errors in Parenthesis. ***p<0.01, **p<0.05, *p<0.1 Note2: Prices: (1) Fruits, (2) Industrial Crops & legumes, (3) Cereals, (4) Vegetables, (5) Tubers 27 Table 5 – Inefficiency Estimation for Honduras Inneficiency Determinants (lnsig2u) Land -0.515*** (0.133) Land*Land 0.0183*** (0.00467) Capital -0.139*** (0.0261) Land Title -0.187 (0.164) Credit Access -1.206* (0.695) Market Access -0.252*** (0.0579) Education -0.0234 (0.0250) Household Size -0.105*** (0.0273) Technical Assistance -0.297 (0.432) Agroec1: Lower montane moist forest 0.00862 (0.337) Agroec2: Subtropical rainforest -0.500*** (0.173) Agroec3: Tropical rainforest -0.483 (0.303) Agroec4: Lower montane wet forest -0.440 (0.545) Agroec5: Subtropical wet forest -0.706** (0.284) Agroec6: Subtropical dry forest -0.289 (0.425) Efficiency Constant 1.065*** (0.394) Observations 1396 Note1: Standard Errors in Parenthesis. ***p<0.01, **p<0.05, *p<0.1 Note2: Some categories were not incorporated to avoid multicollinearity 28 Table 6 – Inefficiency Estimation for Nicaragua Inefficiency determinants x land extension Land -0.563*** -0.00131 (0.135) (0.00113) Log of Productive assets (self-valorization) -0.285*** 0.0629*** (0.0576) (0.0124) Household size 0.0365 -0.116*** (0.0524) (0.0244) Membership to a farmer organization 4.060 -5.681* (2.593) (3.414) Access to technical assistance -0.164 -0.160 (0.759) (0.254) Maximum years of schooling of the household 0.0185 0.00927 (0.0394) (0.0103) Land title 0.465** -0.0518* (0.224) (0.0311) Access to credit 0.830** -0.624** (0.420) (0.258) Access to credit * Land title -0.260 0.337 (0.747) (0.331) Log. of Cost to the nearest city 0.881** -0.533*** (0.407) (0.124) Land use1: Crops -1.131** (0.490) Land use2: Fruits -0.868* (0.464) Land use4: Pastures -0.700 (0.433) Land use5: Forest -0.400 (0.406) Land use6: Secondary Forest -0.788** (0.376) Constant 3.296*** (0.570) Observations 2265 Note1: Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 Note2: Some categories were not incorporated to avoid multicollinearity 29 Table 7 – Inefficiency Estimation for Panama Inneficiency Determinants (lnsig2u) x Land extension Log of Productive Assets (self-valorization) -0.294*** -0.0115 (0.0814) (0.0127) Land 0.0908 -0.00167 (0.304) (0.00483) Credit 0.264 -0.0432 (0.237) (0.0447) Hoursehold Size -0.00322 -0.0159 (0.0487) (0.0206) Land Title 0.580** -0.101* (0.265) (0.0537) Technical Assistance -11.84 0.555 (8.906) (0.459) Maximum years of schooling of the household 0.0344 0.00221 (0.0220) (0.00411) Market Access -0.233 0.0147 (0.153) (0.0563) Land Use 2 : Grasslands 0.303 (0.188) Land Use 3 : Secondary Forest 1.046*** (0.335) Land Use 4 : Cropland -0.0401 (0.219) Constant 1.830** (0.818) Observations 1589 Note1: Standard Errors in Parenthesis. ***p<0.01, **p<0.05, *p<0.1 Note2: Some categories were not incorporated to avoid multicollinearity 4.2 Constructing the typology The stochastic profit frontier estimation results in the previous section and the scaling-up steps described in Section 4.4, Figures 9-11, allow us to construct a typology of micro-regions using the notions of market access, profit potential (frontier), profit efficiency (technical efficiency), and priority (poverty rates). The first map of Figures 12, 15 and 18, present the results of the accessibility measurement constructed following the methodology explained in section 3 in dollars per equivalent kilograms of potatoes transported across the three different countries in Honduras, and in time to the closes market for the case of Nicaragua and Panama where we did not have the necessary data to develop the full cost of transportation analysis as in the case of Honduras. 30 We then assess what is the maximum profit farmers in a given area can generate given their average characteristics and assuming an efficient allocation of resources and skills through the scaling-up to the regional level of frontier profits. The results for Honduras, Nicaragua and Panama are shown in the second map of Figures 12, 15 and 18. Once the profit potential of an area is established, it is necessary to know how far those regions from that frontier are. This distance is given by the inefficiency component. The third map of Figures 12, 13 and 14 show the efficiency levels for rural households in rural Honduras, Nicaragua and Panama. More efficient areas or areas closer to their potential (frontier), are depicted in red, while less efficient areas are depicted in green. Some interesting patterns start to appear in the maps. Many areas classified as having a high potential, are nevertheless highly inefficient, which might explain some of the poverty rates in that area we will observe in the following map. To complete our typology, we need a measure of the priority in which investments should be made to have a greater impact on overall welfare levels of the rural population. For this a reasonable criteria is to use poverty rates, depicted in Figures 13, 16 and 19. Combining figures 12 and 13 we develop the typology for Honduras (Figure 14), and similarly for Nicaragua (combining Figures 15 and 16 and presenting the typology in Figure 17) and Panama (combining Figures 18 and 19 and presenting the typology in Figure 20). With only these 4 dimensions we can observe the extreme heterogeneity that exists in the rural highlands. 4.3 Typology results Because of the high heterogeneity between micro-regions (even when considering only 3 basic dimensions) it is difficult to interpret all the information presented in the maps of the previous section. For the purposes of exposition, we collapse the types in our typology to 7 groups that capture some key characteristics that matter for policy making as was shown in Table 1. These groups are described below and shown in Figures 14, 17 and 20: - Critical areas (high poverty and low potential [Dark Red] - High priority areas (high poverty and medium/high potential) [Dark green] - Medium priority areas without opportunities for agriculture (medium poverty and low potential) [Orange] - Medium priority areas with opportunities for agriculture (medium poverty, high/medium potential, and low/medium efficiency) [Green] - Low priority areas with opportunities for agriculture (low poverty, high/medium potential, and low/medium efficiency) [Light green] - High performance areas (low poverty, high/medium potential, and high efficiency) [Yellow] - Low priority areas (low poverty and low potential) [pink] Figures 21, 22 and 23 show 4 specific areas where different types of interventions can be implemented based on the typology developed for the three countries. In all figures we present four maps: (a) in the left top corner show critical areas, i.e. areas of extreme high poverty and 31 with very low potential for agricultural development given their low level of potential and efficiency; (b) in the right top corner we show high priority areas, i.e. areas with high poverty and low efficiency but with high agricultural potential for agricultural development; (c) in the bottom left corner we show the high performance areas, i.e. areas with low poverty, high efficiency and high potential, these are the areas from where the policy makers can learn and try to replicate what is happening there; and (d) finally, the bottom right corner we show medium priority areas with no opportunities for agriculture (medium poverty, low potential, and low efficiency) In the case of the critical areas without agricultural potential, policies in these areas should provide direct social assistance in the short term and major effort will be to properly target the benefits from such programs. For example in the case of Nicaragua it will be import to implement existing best practices on social protection programs as the conditional cash transfers, direct cash transfers, food for work, or school feeding. In the case of Panama the school feeding, cash transfers and food ration stamps are already in place although the targeting of these programs should be a priority. Similarly is the case of Honduras where cash transfers and school feeding programs already exist. The high priority areas are clearly areas where public investment can play a crucial role by identifying key bottlenecks that producers are facing and which explain why they show such low levels of efficiency despite they have land with significant potential for agriculture. The policies in these areas should therefore target the specific bottlenecks and market failures that don’t allow producers to reach and clearly take advantage of their high potential for production. Market failures not only limit the access of smallholders to other factors of production and modern technologies but also make them utilize their productive assets for non-productive purposes, for instance, as an insurance mechanism to smooth consumption through the sale of these assets. Therefore, the key idea is to put markets to work in rural areas such that productive resources are entirely devoted in an efficient way to productive purposes. This in turn will lead to higher labor productivity and labor incomes. Four markets deserve special attention: credit, insurance, land, and services. Specific policies must be identified to eliminate failures and in some cases create those markets, namely markets of agricultural services: extension, legal, accounting, marketing, management, etc. Finally, horizontal and vertical integration arrangements could also play a crucial role by allowing them to combine productive resources through non-market relationships. From this point of view, these institutional arrangements might be of critical importance in rural areas where market failures are widespread. In this way, labor resources can be combined with other productive resources, modern inputs capital, and services that otherwise wouldn’t be possible given market failures. In the case of vertical integration schemes among smallholders, such as contract farming, we foresee two potential benefits: (a) first, it allows the smallholder to resolve its market failures or bottlenecks through the access provided by the contracting party; (b) second, it allows the farmers to specialize in production activities, most probably where they have comparative advantage; and (c) allow for scales of operation. In the medium priority areas with no potential for agriculture the focus needs to be on non-farm 32 activities to move households out of poverty given the low level of agricultural potential. In addition public investment in hard infrastructure could help to create these opportunities. Finally, high performance areas are where lessons can be learned on how they produce at their maximum potential level, and from there extrapolate the lessons learned to other regions of the country. Figure 12: Honduras – Accessibility, Potential and Efficiency Accessibility Profit Potential Efficiency 33 Figure 13: Poverty Map for Honduras 34 Figure 14: Typology of Micro regions for Honduras 35 Figure 15: Nicaragua – Accessibility, Potential and Efficiency Accessibility Profit Potential Efficiency 36 Figure 16: Poverty Map for Nicaragua 37 Figure 17: Typology of Micro regions for Nicaragua 38 Figure 18: Panama - Accessibility, Potential and Efficiency Accessibility Profit Potential Efficiency 39 Figure 19: Poverty Map for Panama 40 Figure 20: Typology of Micro regions for Panama 41 Figure 21: Honduras- 4 key areas and types of public investment needed Areas of high poverty with low potential for agriculture and Areas of high poverty but with high potential for agriculture. Key significant number of microclimates. Mostly basic staples policies in these areas should focus in identifying and resolving subsistence agriculture. Key policies should focus on short term major bottlenecks to better use the quality of the land and targeted assistance as the use of conditional cash transfers, School optimize its potential. feeding, etc. Areas of low poverty rates with significant potential and high Areas with moderate poverty rates and low potential and medium efficiency. Given their high performance these regions could be and high efficiency. Despite the efficiency is high the agricultural used to learn and replicate from them. potential is low and as a result there are important levels of poverty. Policies in these areas should focus on non-farm activities as a way to increase the income of the households. 42 Figure 22: Nicaragua- 4 key areas and types of public investment needed Areas of high poverty with low potential for agriculture Areas of high poverty but with high potential for and significant number of microclimates. Mostly basic agriculture. Key policies in these areas should focus in staples subsistence agriculture. Key policies should identifying and resolving major bottlenecks to better use focus on short term targeted assistance as the use of the quality of the land and optimize its potential. conditional cash transfers, School feeding, etc. Areas of low poverty rates with significant potential and Areas with moderate poverty rates and low potential and high efficiency. Given their high performance these medium and high efficiency. Despite high efficiency, regions could be used to learn and replicate from them. agricultural potential is low, resulting in high poverty. Policies in these areas should focus on non-farm activities as a way to increase household income. 43 Figure 23: Panama- 4 key areas and types of public investment needed Areas of high poverty with low potential for agriculture and Areas of high poverty but with high potential for agriculture. Key significant number of microclimates. Mostly basic staples policies in these areas should focus in identifying and resolving major subsistence agriculture. Key policies should focus on short term bottlenecks to better use the quality of the land and optimize its targeted assistance as the use of conditional cash transfers, School potential. feeding, etc. Areas of low poverty rates with significant potential and high Areas with moderate poverty rates and low potential and medium and efficiency. Given their high performance these regions could be high efficiency. Despite the efficiency is high the agricultural used to learn and replicate from them. potential is low and as a result there are important levels of poverty. Policies in these areas should focus on non-farm activities as a way to increase the income of the households. 44 Conclusions This paper presents an alternative method to classify and analyze rural regions for Rural Honduras, Nicaragua and Panama using stochastic profit frontier estimation. Unlike other methods used to classify regions like poverty maps or cluster analysis, this procedure is based on a basic assumption of resource optimization under economic and physical constraints. We take advantage of extremely rich biophysical data on these countries geography and a highly detailed and fully geo-referenced socioeconomic household surveys to construct our typology. We combine these data sources using a profit frontier approach to estimate agricultural potential and efficiency at the regional level, and properly account for the impact of random shocks inherent to farm activities by adding a stochastic component. Adding information from poverty maps and a comprehensive calculation of access costs to local markets to the profit frontier results generates a multi-dimensional typology that is detailed enough to capture the high heterogeneity of rural populations, but simple enough to be practical for its results to be used as a tool for policy making and investment decisions. With this basic setup it is possible to start designing policies that ac cording to the regions’ characteristics. For instance, because of their low agricultural potential and high poverty rates, critical areas are the first candidates to receive funds from cash transfer programs and other immediate assistance. Areas that combine high potential with low efficiency need investments that can reduce transaction costs and increase productivity in order to take full advantages of the region’s opportunities. The agro ecological conditions, technologies, and economic environment of high performance areas should be analyzed carefully to see how these settings can be replicated elsewhere. Finally, the accessibility model results can identify areas that could develop quickly if properly connected, and others were access is not the main bottleneck for development. 45 Bibliography Ali, F., A. Parikh, and M. K. Shah (1994): “Measurement of pro�t efficiency using behavioural and stochastic frontier approaches,� Applied Economics, 26(2), 181–188. Ali, M., and J. C. Flinn (1989): “Pro�t efficiency among Basmati rice producers in Pakistan Punjab,� American Journal of Agricultural Economics, 71(2), 303–310. Coelli, T., D. S. P. Rao, and G. E. Battese (1998): An introduction to efficiency and productivity analysis. Kluwer Academic Publishers, Inc., Boston. Deprins, D., and L. Simar (1989): “Estimating technical inefficiencies with correction for environmental conditions with an application to railway companies,� Annals of Public and Cooperative Economics, 60(1), 81–102. Elbers, C., T. Fujii, P. Lanjouw, B. Ozler, and W. Yin (2004): “Poverty alleviation through geographic targetin: How much does disaggregation help?,� World Bank Policy Research Working Paper 3419, World Bank, Washington, DC. Elbers, C., J. O. Lanjouw, and P. F. Lanjouw (2003): “Micro-level estimation of poverty and inequality,� Econometrica, 71(1), 355–364. Fan, S. and P. Hazell (1999): “Are Returns To Public Investment Lower In Less -Favored Rural Areas? An Empirical Analysis Of India.� EPTD Discussion Paper 43, Environment and Production Technology Division, International Food Policy Research Institute. Washington, D.C. May. Fan, S., Hazell, P., and Haque, T. 2000a. “Targeting public investments by agro -ecological zone to achieve growth and poverty alleviation goals in rural India.� Food Policy 25(4): 411-428. Fan, S., Hazell, P., and Thorat, S. 2000b. “Government spending, growth and poverty in rural India.� American Journal of Agricultural Economics 82(4): 1038-51. Fan, S.; L. Zhang and X. Zhang. 2002. “Growth, Inequality, and Poverty in Rural China The Role of Public Investments� Research Report 125. International Food Policy Research Institute. Washington, D.C. Farrell, M. J. (1957): “The measurement of productive efficiency,� Journal of the Royal Statistical Society, Series A 120, 253–281. Kumbhakar, S. C. (1996): “Efficiency measurement with multiple outputs and multiple inputs,� The Journal of Productivity Analysis, 7, 225–255. Kumbhakar, S. C., and C. A. K. Lovell (2000): Stochastic frontier analysis. Cambridge University Press. Lopez, Ramón and Alberto Valdés; (1999). Rural Poverty in Latin America: Analytics, New Empirical Evidence and Policy. New York. St Martin’s Press, MacMillan. Maruyama, E and Torero, Maximo; (2009): Constructing a typology of rural communities of the Peruvian highlands using stochastic profit frontier estimation. IFPRI, Washington. Pandolfelli, Lauren, Dohrn, Stephan and Meinzen-Dick, Ruth; (2007). “Gender and Collective Action: Policy Implications from Recent Research�, Policy Brief 5, CAPRI. 46 Rahman, S. (2003): “Pro�t efficiency among Bangladeshi rice farmers,� Food Policy, 28(5–6), 487–503. Simar, L., and P. W. Wilson (2007): “Estimation and inference in two-stage, semi-parametric models of production processes,� Journal of Econometrics, 136, 31–64. Tobler, W. (1993): “Movement takes time: A note on the time of movement,� in Structure and change in the space economy: Festschrift in honor of Martin J. Beckmann, ed. by T. R. Lakshmanan, and P. Nijkamp. Springer-Verlag, Berlin. Wang, J., E. J. Wailes, and G. L. Cramer (1996): “A shadow-price frontier measurement of pro�t efficiency in Chinese agriculture,� American Journal of Agricultural Economics, 78(1), 146– 156. Zhang, X and S. Fan. 2000. “Public Investment and Regional Inequality in Rural China� Discussion Paper No. 71, Environment and Production Technology Division, International Food Policy Research Institute. Washington, D.C. December. 47 Annex 1: Calculation of transportation costs  Transportation costs in roads:  We assumed the cost of transportation of goods in a medium sized truck. Such transportation would involve: gas prices (Diesel 2), other indirect costs, and depreciation of the vehicle.  Regional prices of Diesel 2 are taken for Honduras. The regional prices were adjusted according to regional price differences. For this purposes, we used the values of regional baskets of basic goods and services. This basket is calculated by the respective National Statistic institute and reflects differences in prices and consumption patterns among different areas.  These baskets were used to construct deflators that are subsequently used to adjust the diesel prices. In this sense, our estimations consider diesel prices expressed in local currency for the capital of each country.  Our estimations also consider the following costs: tires, oil, filters, possible repairs, tuning and contingencies. We do not have available figures of these costs for each of the countries, but there is available information for Colombia. According to the Colombian Ministry of Transportation (2000), these costs represent around 1.97 times the expenditure in gas. So this ratio is used.  Considering a value of US$ 30,000 and a useful life of 200 thousand kilometers, we estimate a depreciation of US$/. 0.15 per kilometer.  Additionally, assume values for the following parameters: Unit Parameter Motor Performace Paved roads Km / gallon 12.5 Unpaved roads Km / gallon 10.6 Rivers Km / gallon 9.0 Speed Paved roads Km / hour 60.0 Unpaved roads Km / hour 30.0 Rivers Km / hour 10.0 Truck load Tons 22.5  Under these conditions, we calculate the cost of transporting one kg per second. Cost of transporting 1 1  Diesel Pr ice x Factor for Indir cos ts x x Speed x one kg per sec ond Performance 12.5 tons per load US $ gallon km  1 hour  1  1 ton   x1.97 x     gallon km hour  3600 sec onds  tons 1000 kg  US $  kg x sec ond 48  Transportation costs with no car:  When roads are not available, we consider that farmers transport their products walking or using mules.  The cost imputed to this means of transportation is the cost of opportunity of the farmer, which is assumed to be the hourly wage.  Using the National Household Survey of Honduras, we estimate the hourly wages for rural areas in each region (department).  Due to larger wages in urban areas, considering regional averages may significantly increase the hourly wages in rural areas. Thus, even when our estimations of hourly wages may not be representative, we decided to use the median of rural hourly wages in each region.  The hourly wages are adjusted to be comparable with the diesel prices. Firstly, they were geographically adjusted and are expressed in local currency of the capital. Secondly, they were temporarily adjusted by inflation.  We assume that a person by his or herself can transport one and that a mule can transport three bags of 60 kg per trip.  Thus, the transportation cost of one kilogram per second can be calculated as: Cost of transporting  hourly wage x kg per bag x number of bags one kg per sec ond US $ 1 hour 1 1 bag  hour 3600 sec onds # of bags 60 kg US $  kg x sec ond 49