www.uni-goettingen.de/globalfood RTG 1666 GlobalFood Transformation of Global Agri-Food Systems: Trends, Driving Forces, and Implications for Developing Countries Georg-August-University of Göttingen GlobalFood Discussion Papers No. 56 Distance to market and farm-gate prices of staple beans in rural Nicaragua Ayako Ebata Pamela Velasco Stephan von Cramon-Taubadel January 2015 RTG 1666 GlobalFood ⋅ Heinrich Düker Weg 12 ⋅ 37073 Göttingen ⋅ Germany www.uni-goettingen.de/globalfood ISSN (2192-3248) Distance to market and farm-gate prices of staple beans in rural Nicaragua Ayako Ebata*, Pamela Velasco, Stephan von Cramon-Taubadel Department of Agricultural Economics and Rural Development , Research Training Group “GlobalFood”, University of Göttingen, Heinrich-Dücker Weg 12, 37073 Göttingen, Germany *Corresponding author: [email protected] Abstract: While smallholder market participation is seen as a catalyst for poverty alleviation, farmers in rural areas face a number of challenges in doing so. One of the most important factors is considered transaction costs related to transportation. Our study quantifies the benefits associated with improvement of rural road infrastructure by scrutinizing farm-gate prices of beans in rural Nicaragua. We find that the longer the distance and traveling time are to major commercial centers from farming communities, the less farm-gate prices producers receive. We find that a decrease in distance and traveling time by one unit is associated with an increase in farm-gate prices by 2-2.5 cents/qq. If infrastructure development can reduce travel time by 25%, an average farm would increase its annual revenue from beans by between $27.69 and $125.96 (between 4% and 18% of annual revenue today). Given that such infrastructure development affects all farmers and all crops, our findings suggest a larger implication at the sectorial level. Keywords: Producer prices, Central America, Transactions costs, transportation infrastructure JEL codes: O13, O18, Q11 Acknowledgement: the authors thank the financial support from the German Research Foundation (DFG) and German Academic Exchange Services (DAAD). In addition, the data set was provided by the Catholic Relief Services (CRS) in Nicaragua. 1. Introduction In today’s changing agrifood system, smallholder participation in commercial markets has attracted attention as a potential catalyst for alleviation of poverty. Farmers who are included in the global procurement system are found to benefit from premium product prices (Gulati et al., 2007), reduced transactions costs in product marketing (Nagaraj et al., 2008; Vieira, 2008), and access to necessary assets (Minten et al., 2009; Nagaraj et al., 2008; Swinnen, 2007). As a result, participating farmers are able to improve productivity, household income and/or asset holdings (Minten et al., 2009; Miyata et al., 2009; Reardon, Barrett et al., 2009). However, participation in global supply chains requires good access to roads and other transportation infrastructure, production assets (e.g. irrigation system), and thorough knowledge of farming techniques among others (Barrett et al., 2012; Donovan & Poole, 2008; Hernandez et al., 2012; Michelson, 2013; Murray, 1991; Rao & Qaim, 2011). For lack of these factors, small farmers in rural areas are often excluded from the global retail markets and therefore unable to enjoy benefits that the global procurement system can provide. In response to the difficulties that small farmers face, empirical studies suggest mechanisms that assist small farmers’ participation in the global supply chain. For instance, Hellin et al. (2009) and Narrod et al. (2009) show the importance of collective actions by looking at cases in Central America, and Kenya and India, respectively. By forming farmer organizations, individual smallholders can conduct product marketing as a group, enabling access to improved market information as well as sales of larger quantities which can reduce transaction costs. Minten et al. (2009) argue that intensive farm technical assistance allows farmers to meet complex quality requirements imposed by buyers. They find that participating farmers in Madagascar are provided with necessary inputs by the buyer to ensure the quality of final products. Based on a negative experience in the pineapple industry in Ghana, Whitfield (2012) also highlights the importance of updating production technology as well as trade-friendly policy environments. In essence, such mechanisms aim to reduce the transactions costs that smallholders face when accessing markets. Transactions costs are seen as one of the key factors that influence market participation and the welfare of small farmers (Barrett, 2008; Pingali & Khwaja, 2005). Poor infrastructure in rural areas in particular can prevent smallholders in developing countries from participating in market-based economic activities (Mabaya, 2003; Moser et al., 2009). At the macro-level, geographically isolated areas demonstrate less market integration than those that are well-connected (Barrett, 1996; Baulch, 1997; Fackler & Goodwin, 2001; Ravallion, 1986). 1 Rapsomanikis et al. (2006) show that high transfer costs due to poor infrastructure and lack of communication methods can create large marketing margins. Renkow et al. (2004) estimate that fixed transaction costs are equivalent to a 15% ad valorem tax on maize farmers in Kenya, and Jacoby and Minten (2009) show that transportation cost can be up to 50% of final product price in the case of rice farmers in remote areas of Madagascar. As a result, high transportation costs encourage farmers in rural areas to stay in subsistence farming (Dillon & Barrett, 2013; Key et al., 2000). When markets are isolated, local players such as traders can acquire regional monopsony or oligopsony power (Barrett, 2008; Faminow & Benson, 1990; Graubner et al., 2011). As a result, commodity prices in geographically segregated areas often respond less quickly to changes in macro-level prices and are less integrated than in markets that are well linked to national and international markets (Getnet, Verbeke, & Viaene, 2005; Goletti, Ahmed, & Farid, 1995; Siqueira, Kilmer, & Campos, 2010). In dealing with market participants who have market power, smallholder will tend to pay more for inputs and receive less for their products, thus exacerbating the problem of low margins and poverty traps. All of these considerations underline the recognized importance of transportation infrastructure improvement (Jacoby, 2000). Given the potential for infrastructure development in rural areas to alleviate poverty, there is an increasing interest in developing rural infrastructure (World Bank, 2007). However, quantifying the optimal level of infrastructure investment is a difficult task. If policy makers ignore the impact of market segregation due to transportation cost on low farm prices, the optimal level of investment can be underestimated (Mérel et al., 2009). In order to take appropriate investment decisions, policy makers require quantitative information on the potential impact of rural road improvement. In this paper we generate such information by studying how farm-gate prices are affected by physical distance and traveling time from farms to markets. Building up on the hedonic price model, we identify product-, producer- and marketing- attributes, including physical distance and traveling time, which influence producer prices. As a case study, we select the bean sector in rural Nicaragua. Bean is one of the most important crops for food security in Nicaragua besides maize and rice (FAO, 2012; INIDE, 2011). In the recent years, Nicaraguan bean sector suffered from stagnation of productivity and restriction of agricultural land expansion (FAO, 2012). In addition, as a key staple crop, beans are subject to government policy interventions that have arbitrary impacts on bean producers. During 2 2010 and 2011, export restrictions were put in place by the government. This interrupted trade flows to important importers in neighboring Central American countries (FAO, 2012; La Prensa, 2011). Moreover, transportation costs within Nicaragua are high: on average, transportation costs within Nicaragua to local seaports account for 50% of total freight rates to the U.S. (World Bank, 2012). As a result, bean producers face difficulty in participating in commercial sales. The rest of the paper is organized as follows. The next section describes the bean sector in Nicaragua. In section 3 we then explain our conceptual framework, data set and econometric model. Descriptive statistics and regression results are presented in section 4, and we discuss the findings and conclude. 2. Background: beans in Nicaragua Beans are important for Nicaraguans not only as a staple food crop but also as a major income source for the poor (FAO, 2012; INIDE, 2011). Beans are produced throughout the country and especially in the Northeast (FAO, 2012). More specifically, production of beans is prominent in the departments of Jinotega, Matagalpa and Nueva Segovia (INIDE, 2011). Nicaragua’s agriculture is predominantly conducted by small producers. Approximately 50% of all producers in the country farm less than 3.5ha1 of land. These small producers account for only 19.2% of the land used for bean production. The bean sector has seen little improvement regarding production technology (FAO, 2012). As a result, yield growth has been stagnant over the last 20 years (FAO, n.d.). Table 1. Farm size and number of bean producers in Nicaragua Size Number of producers Bean cultivation area Ha All commodities % Ha % 0.4 or less 31,758 12.15% 1,114.43 1.15% 0.4 -- 0.7 16,660 6.38% 3,643.90 3.75% 0.7 -- 1.8 38,149 14.60% 1,3,903.30 14.32% 1.8 -- 3.5 35,580 13.62% 1,4,737.54 15.18% 3.5 -- 7.0 33,591 12.85% 1,4,768.51 15.21% 7.0 -- 14.8 29,775 11.39% 1,3,768.83 14.18% 14.8 -- 35.2 37,246 14.25% 1,7,642.24 18.18% 35.2 or more 38,562 14.76% 1,7,488.06 18.02% Total 261,321 9,7,066.82 Source: (INIDE, 2011) The majority of beans produced in Nicaragua are sold domestically but the export market 1 In Nicaragua, land area is measured using Manzanas (Mz). We convert the unit to hectares with a conversion rate 1 Mz=0.704ha. 3 has grown in the last decade (Figure 1). Between 2007 and 2010, on average 45% of total production was directed to the export markets (FAO, n.d.). Central American countries are the biggest importers of Nicaraguan beans (Table 2). Since 2007, Nicaraguan exports to El Salvador, Costa Rica and Honduras have increased. El Salvador is now the largest importer of beans produced in Nicaragua, while a relatively small share is directed to the U.S. The active exchange of the commodity in the Central American region may be due to the Dominican Republic-Central America Free Trade Agreement (DR-CAFTA) signed by the Dominican Republic, the U.S. and Nicaragua in 2004 (USTR, n.d.). Bean exports to Venezuela have also grown since 2008 (Table 2). Figure 1. Production, domestic supply and trade of beans in Nicaragua: 2000-2011 Quantity of beans (ton): 2000-2011 250,000 200,000 150,000 100,000 50,000 - Production Import Export Domestic supply Source: (FAO, n.d.) Two types of beans are produced in Nicaragua: red and black. Red beans are a staple commodity not only in Nicaragua but also in many other Central American countries. Therefore, production of red beans is significantly more than black beans. Although black beans may be exchanged domestically and regionally, they are mostly targeted for export almost exclusively to Venezuela (FAO, 2012). However, the sustainability as well as the potential of the Venezuelan market is questioned. Nicaragua and Venezuela do not have an official trade agreement such as DR-CAFTA, and exports to Venezuela are coordinated exclusively by the Nicaraguan government as a part of an alliance called ALBA (Bolivarian Alliance for the Peoples of Our 4 America, Spanish acronym) (FAO, 2012). As a result, the transactions lack transparency (COHA, 2010) and there are concerns that the recent surge in black bean export to Venezuela may be temporary and do not provide income-generating opportunity for all producers. Table 2. Destination of Nicaraguan bean export Destination 2006 2007 2008 2009 2010 2011 North America USA 3,744 3,789 5,523 5,732 4,886 2,540 Canada 80 20 Central America Guatemala 225 496 259 832 472 683 El Salvador 21,710 27,253 25,149 18,306 9,713 Costa Rica 17,981 14,264 14,525 12,675 3,766 Honduras 9,231 6,682 13,522 4,654 536 Panama 0 20 0 0 Others Venezuela 660 2,460 14,040 9,806 Source: (FAO, n.d.) As a key food security crop, beans are subject to policy interventions in Nicaragua. In 2010 and 2011, an informal restriction was put on red bean export in order to protect consumers in Nicaragua (The Economist, 2011). However, this policy was criticized for reducing Nicaragua’s share of the regional red bean market (FAO, 2012; La Prensa, 2011). As seen in Table 2, bean export to El Salvador, Costa Rica and Honduras decreased significantly in 2010 and 2011. The resulting shortage of red beans in these Central American markets has been replaced by competitors such as China (FAO, 2012), which could result in Nicaragua losing these markets permanently. Transportation costs are considered as one of the key factors that hinder both international and domestic product exchange in Nicaragua. According to World Bank (2012), Nicaraguan domestic transportation costs can make up more than 50% of the total freight costs to the U.S. For instance, transportation costs incurred within Nicaragua from Matagalpa, Jinotega and Nueva Segovia to the port of Corinto are 59%, 62% and 64%, respectively, of the total freight costs from these locations to Miami. 3. Empirical estimation strategy Conceptual framework Our model is based on the hedonic price model developed by Rosen (1974). The hedonic price model decomposes observed market prices based on implicit characteristics of the goods 5 exchanged. This model enables us to isolate product attributes of interest and assess how they influence market prices. In the context of agricultural commodities, the hedonic price model has been mainly used to analyze consumer preferences for product attributes. For instance, a number of hedonic analyses of coffee prices have been published(e.g. Donnet et al., 2007, 2008; Teuber & Herrmann, 2012). Faye et al. (2004) and Mishili et al. (2009) look at cowpea prices in Senegal and Nigeria, Ghana and Mali, respectively. These studies analyze consumer preferences for individual products attributes in order to understand the factors that influence consumer choices. Our study applies an analogous methodology to disentangle product characteristics that influence prices received at the farm level. Based on findings from the literature and the empirical context of Nicaraguan bean sector, we identify several variables that are potentially important determinants of farm-gate bean prices. Product quality is one of the most well-documented factors that influence prices (Donnet et al., 2007; Faye et al., 2004; Mishili et al., 2009). Quality characteristics can be implicit (e.g. reputation, brand, preferred production practices) or explicit (e.g. color, shape, size, taste). Marketing practices are often found to be important as well. When a large quantity is sold at once, per unit product prices tend to decrease (Donnet et al., 2007). This may be because sellers are willing to give discount for a larger quantity of sales. Gender might also play a role as female farmers may have less negotiation power than men and can face disadvantages when marketing (Dolan, 2001; Zhang et al., 2006). As a result, they may receive lower prices than their male counterparts. Distance and lack of access to markets can have negative effects on producer prices. For instance, Fafchamps and Hill (2005) show that coffee producers in Uganda are offered lower prices by traders in their villages than at commercial markets, because traveling to remote villages incurs transportation costs. In addition, remoteness can reduce competition and enable oligopsonistic traders to offer lower farm-gate prices (Graubner et al., 2011). Michelson et al. (2012) show that farm-gate prices are significantly lower than wholesale prices in the capital city in Nicaragua. This may result from the exploitation of market power by traders in farming communities when individual transportation to commercial markets is not easy due to poor transportation infrastructure. Based on these considerations, we employ various measures of product quality, quantity exchanged and transfer costs to major ports as explanatory variables in our analysis. We use total 6 distance and traveling time between farming communities and commercial markets as proxies for transfer costs. No matter who travels the distance, farm-gate prices are set lower if the overall transfer costs are high. Therefore, our analysis applies total distance and traveling time from communities to major commercial centers instead of markets where producers could sell their products. Data We analyze sales data recorded by an NGO that is active in Nicaragua, the Catholic Relief Services (CRS). The CRS implemented a development project in rural Nicaragua between September 2007 and October 2012. This project targeted small farmers in Nicaragua who own less than 10 hectares of land. Among the information that was collected are records of individual sales by farmers over the five-year project period. In total, there are 3,893 bean producers in the data. Each producer sold beans at least once during the five years and the average producer sold beans three times, which sums up to a total of 11,718 observations. We exploit the full unbalanced data set. The farmers included in the data set were not chosen randomly. Instead, CRS applied several criteria in selecting individuals to participate in its project. However, the project did not include any interventions that directly influence farm-gate prices. Moreover, the information provided by CRS is rich in the factors that may influence farm-gate prices. The credibility of the information is also considered high since the information on sales was collected every three months, which is approximately one cultivation cycle of beans. Price data are available for each individual sales transaction and include information on the buyers, destination countries, and product quality. The dependent variable, the farm-gate prices of beans, was originally recorded in the local currency, Nicaraguan Córdobas. We converted the values to USD, using the exchange rates recorded throughout the project period. For explanatory variables, we apply both non-binary and binary variables which are categorized as marketing-, product-, and farmer-related variables. For marketing-related variables, we use information about buyers and the intended destination of the beans exchanged. Buyers are divided into five categories: local markets, intermediaries, farmer organizations/cooperatives, private companies, and private export companies. In the analysis, we drop the dummy variable representing local markets as a point of comparison. We expect product prices to be higher when the buyer is a farmer organization/cooperative rather than the local market or a private company. This is because cooperatives’ main objective is not profit but rather 7 enhancing members’ welfare (Giannakas & Fulton, 2005). The information regarding destination countries was obtained through cooperatives. Approximately 90% of farmers in the sample belong to a cooperative and these cooperatives are aware of all the buyers outside local wholesale markets. Therefore, the cooperatives provided information regarding product destination countries corresponding to each buyer. All of the beans sold are destined for the domestic Nicaraguan market or for export to Costa Rica, El Salvador or Venezuela. In order to test whether prices differ by destination, we apply one dummy variable for each of the export destinations. Hence, the default destination is the domestic market in Nicaragua. While it is possible that beans destined for export markets fetch higher prices, in the case of Venezuela the prices may be lower due to the agreement between the governments. Therefore, the expected effect of these destination dummy variables is unclear a priori. For product-related variables, we apply product quality and variety. The quality variable is recorded as 1 if the bean sold is of a high quality. The variety variable equals 1 if the bean sold is red bean and 0 if it is black bean. We expect that the higher the quality of the product, the higher its price (Donnet et al., 2007; Faye et al., 2004; Mishili et al., 2009). Therefore, the quality variable is expected to have a positive coefficient. In terms of bean variety, red beans may receive higher and more volatile prices than black beans because black bean prices may be influenced by the Nicaraguan and Venezuelan governments while red bean prices are determined freely in the market. For farmer-related variables, we employ two farmer characteristics variables (gender and household head) as well as distance to major commercial centers, which is the variable of main interest. Gender of the producer is recorded as 1 if female and 0 if male. The household head variable equals 1 if the producer is the head of the household. The gender variable will have a negative coefficient if females face disadvantage when marketing compared with males (Dolan, 2001; Zhang et al., 2006). The effect of being a household head on producer prices is ambiguous. The exact location of each farm is not coded in the dataset, but for each farm we do know in which municipality it is located. For each farm we calculate distances and traveling time between three major commercial centers and the municipalities in which it is located using Google Maps. The three commercial centers are identified in terms of national and international product exchange: namely, Managua international airport, the Port of Corinto and the Port of Limón. The Port of Limón is the major seaport in Costa Rica while the Port of Corinto is in Nicaragua. In terms of Nicaragua’s total export values, 27.75%, 16.34% and 15.69% are 8