IFPRI Discussion Paper 00883 July 2009 Linking Urban Consumers and Rural Farmers in India A Comparison of Traditional and Modern Food Supply Chains Bart Minten Thomas Reardon Anneleen Vandeplas New Delhi Office INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE The International Food Policy Research Institute (IFPRI) was established in 1975. IFPRI is one of 15 agricultural research centers that receive principal funding from governments, private foundations, and international and regional organizations, most of which are members of the Consultative Group on International Agricultural Research (CGIAR). FINANCIAL CONTRIBUTORS AND PARTNERS IFPRI’s research, capacity-strengthening, and communications work is made possible by its financial contributors and partners. IFPRI gratefully acknowledges the generous unrestricted funding from Australia, Canada, China, Finland, France, Germany, India, Ireland, Italy, Japan, the Netherlands, Norway, the Philippines, Sweden, Switzerland, the United Kingdom, the United States, and the World Bank. AUTHORS Bart Minten Senior Research Fellow, International Food Policy Research Institute b.minten@cgiar.org Thomas Reardon Professor, Michigan State University reardon@msu.edu Anneleen Vandeplas, Ph.D. Fellow, Research Foundation–Flanders, and Researcher, LICOS, K.U.Leuven Anneleen.Vandeplas@econ.kuleuven.be Notices 1 Effective January 2007, the Discussion Paper series within each division and the Director General’s Office of IFPRI were merged into one IFPRI-wide Discussion Paper series. The new series begins with number 00689, reflecting the prior publication of 688 discussion papers within the dispersed series. The earlier series are available on IFPRI’s website at www.ifpri.org/pubs/otherpubs.htm#dp. 2 IFPRI discussion papers contain preliminary material and research results. They have not been subject to formal external reviews managed by IFPRI’s Publications Review Committee but have been reviewed by at least one internal and/or external reviewer. They are circulated in order to stimulate discussion and critical comment. Copyright 2009 International Food Policy Research Institute. All rights reserved. Sections of this material may be reproduced for personal and not-for-profit use without the express written permission of but with acknowledgment to IFPRI. To reproduce the material contained herein for profit or commercial use requires express written permission. To obtain permission, contact the Communications Division at ifpri-copyright@cgiar.org. mailto:b.minten@cgiar.org� mailto:reardon@msu.edu� mailto:Anneleen.Vandeplas@econ.kuleuven.be� http://www.ifpri.org/pubs/otherpubs.htm#dp� mailto:ifpri-copyright@cgiar.org� iii Contents Abstract vi 1. Introduction 1 2. Conceptual Framework 2 3. Rural-Urban Food Supply Chains in India 7 4. Traditional Food Supply Chains 13 5. Modernizing Markets in India 32 6. Conclusions and Way Forward 40 Appendix 44 References 45 iv List of Tables Table 1. Urbanization in India 9 Table 2. Average value of expenditures per person per 30 days in urban India 10 Table 3. Indian food consumption expenditures according to the level of processing (in %), 2004/05 12 Table 4. Descriptives of farmers and retailers 16 Table 5. Characteristics of the most recent transaction of farmers and retailers 18 Table 6. Determinants of vegetable prices (dependent variable = log [price per kg]) 19 Table 7. Role of wholesalers and brokers 20 Table 8. Information transmission on quality 21 Table 9. Information transmission on quantity 22 Table 9. Information transmission on quantity (continued) 23 Table10. Service delivery 24 Table 11. Choices in outlets and in brokers 25 Table 11. Choices in outlets and in brokers (continued) 26 Table 12. Marketing options and advantages of the broker channel 28 Table 13. Credit and insurance services provided by the broker 28 Table 11. Choices in outlets and in brokers (continued) 29 Table 14. Input advances provided by the broker 30 Table 15. Growth in modern food marketing channels versus food consumption in India 32 Table 16. Public procurement, offtake, and stocks of foodgrains, 1995/96 to 2005/06 33 Table 17. Rice and wheat consumption in 1999/2000 34 Table 18. Output and net value added (NVA) of food industries in India 36 Table 19. Characteristics of food processing industry 37 Table 20. Food service industry in India 38 Table A.1. Major agricultural marketing policy reforms, 1998/99 to 2006/07 44 List of Figures Figure 1. Conceptual framework for understanding changes in food supply chains between rural producers and urban consumers 4 Figure 2. Simplified diagram of the agricultural product flow from rural producers to urban consumers in India 8 Figure 3. Share of cereals in monthly per capita expenditure for total food items in India 11 v ACKNOWLEDGMENTS The authors would like to thank the nongovernmental organization (NGO) Rachna and Manoj Bhatt for the implementation of the survey; Yashodhan Ghorpade for help in setting up the survey and for his insights into the topic; Sunipa Das Gupta for research assistance; and Ashok Gulati, Shenggen Fan, an anonymous reviewer, and participants in a seminar at IFPRI–New Delhi for their useful comments on an earlier version of this paper. vi ABSTRACT Food supply chains are being transformed in a number of developing countries due to widespread changes in urban food demand. To better anticipate the impact of this transformation and thus assist in the design of appropriate policies, it is important to understand the changes that are occurring in these supply chains. In a case study of India, we find that overall urban consumption is increasing; the urban food basket is shifting away from staples toward high-value products; and modern market channels (modern retail, food processing, and the food service industry) are on the rise. We document differing practices in traditional and modern food supply chains and identify an agenda for future research. Keywords: agricultural marketing, market transformation, India, rural-urban linkages 1 1. INTRODUCTION Food supply chains from rural producers to urban consumers are being transformed in a number of developing countries due to widespread changes in urban food demand: overall urban consumption is increasing; the food basket is shifting away from staples toward high-value products such as dairy products, meat, fish, fruits, and vegetables (e.g., Pingali 2007; Gulati et al. 2007); consumers require more variety and choice; they consume more processed and ready-to-eat food; and they are increasingly concerned about food quality and safety (e.g., Jaffee and Henson 2004; Swinnen 2007). These changes are in turn being driven by income growth; urbanization; changes in lifestyle, with more women working outside the home; access to technology such as refrigerators and microwave ovens; the development of better packaging technologies; and the entrance of modern marketing channels, including modern retail, the processing sector, and the food service industry (e.g., Reardon and Timmer 2007; Reardon et al. 2003). To better anticipate the impact of this transformation and thus assist in the design of appropriate policies, it is important to better understand the changes that are occurring in these supply chains. Our objective in this paper is to conceptualize changes—and their impacts—in rural-urban food supply chains, to examine the evidence of changes in the rural-urban food supply chain in the case of India, and to indicate needed research in this area. We find that three important changes are taking place. First, there is significant growth in the rural-urban food supply chain. Based on a series of representative national household surveys—the National Sample Surveys (NSSs)—annual urban food expenditures were valued at about US$45 billion in total in 2006. Driven by increasing urbanization and population growth, this is a real threefold increase compared to 35 years earlier. Moreover, while urban food expenditures made up about one-quarter of total national food expenditures 35 years ago, this had increased to more than one-third by 2006. Given that this increasing food consumption has not been made up by imports—as India exports more than it imports (now and 35 years ago)—this implies that relatively and absolutely larger quantities of food are shipped from rural areas to urban areas. Second, the composition of the food shipped to urban areas has changed significantly, as the share of cereal consumption in the urban food basket has declined from 36percent in 1972 to 23percent in 2006. Third, all “modern” market channels—private-sector-led (modern retail, food processing, the food service industry) as well as public-sector-led (the parastatals)—show higher annual growth rates than do overall urban food expenditures.1 1 The processing sector and the food service industry consist of an informal and a formal, modern sector. With the data available, it is difficult to obtain separate numbers on their growth rates. However, for both of these sectors, the formal sector is significantly more important than the informal sector in terms of the share of output (for more details, see Section 5). The growth of modern retail in recent years has been estimated at 65 percent annually, the food service industry (restaurants, fast-food, takeaway, cafés/bars, food stalls / kiosks) overall at 9 percent, the processing sector at 7 percent, and parastatal marketing at 7 percent. As urban food expenditures grew more slowly (at 3.4 percent annually over the last 10 years, based on the NSSs), this indicates the increasing relative and absolute importance of modern channels in food supply chains. These growth and diversification trends in urban demand and the increasing importance of modern marketing channels are expected to continue in the future, and they might have important implications for all stakeholders in rural-urban food supply chains. The structure of the paper is as follows. In the following section, we present a simplified conceptual framework for understanding the changes in the food supply chain between rural producers and urban consumers. In Section 3, we describe the rural-urban food supply chain and the changes in urban food demand that are occurring in India. In Section 4, we discuss the benefits and drawbacks of the traditional marketing channel based on the results of a small primary survey that we conducted with farmers and traders who use this traditional channel for the marketing of vegetables. We examine the modern marketing channels in India in Section 5. We provide our conclusions and suggestions for further research in Section 6. 2 2. CONCEPTUAL FRAMEWORK Transformation of the Food Supply Chain A simplified conceptual framework for understanding the changes in rural-urban food supply chains is shown in Figure 1. Various drivers are changing urban food demand in most developing countries. These drivers include, most importantly, (1) urbanization (a larger share of the population in developing countries is living in urban centers; given that population growth in these countries is often high, a rapid increase in the urban population overall is usually seen); (2) income growth (an important increase in average incomes and a reduction in poverty levels has been seen in a number of developing countries in recent years); (3) changing lifestyle and female participation in the workplace (women have traditionally taken care of agricultural production and/or food preparation, but as they are increasingly entering into the urban labor force, they often have less time to spend on these activities); and (4) increasing access to better technologies (these include, at the household level, the spread of refrigerators, microwave ovens, and gas stoves, which allow for the use of different foods and food preparation methods, and at the industry level, access to better food packaging technology—with the rise of Tetra Pak, for example, which has made packaged milk and juices available in mass markets). These changes have led urban consumers in developing countries to demand a different food basket: (1) the quantity, per person and overall, that is demanded from urban food markets is increasing faster than in rural areas; (2) the composition of the food basket is different, as better-off consumers often shift away from grains and consume relatively more high-value products such as fruits and vegetables, dairy products, meat, and fish,2 The changing requirements of urban consumers lead to a restructuring of food supply chains. The final food supply chain arrangements are, however, shaped not only by these demand factors. Conditioning factors such as geography, as well as more processed food for convenience; (3) there is a demand for more choices per product and a greater variety of food products in general; and (4) urban customers in developing countries are also increasingly concerned about quality and safety issues with regard to their food, especially as safety issues tend to be more correlated with nonstaple foods. Demand by urban customers is transmitted through the marketing sector or supply chain to the rural producer. This chain ensures that a primary agricultural product is delivered to the customer in the right form, at the right time, and in the right location. Marketing margins, defined as the difference between the retail price and the producer price, are a reflection of the costs incurred in this process of delivering food from the rural producer to the urban consumer. Marketing costs reflect physical handling costs, transaction costs, and potentially rents. They differ per product, as physical handling costs (transport, handling, processing, storage, traceability, etc.) vary by product. The degree of perishability often affects the physical costs, as higher losses, as well as higher risks, are expected to be reflected in the final price of more perishable goods. On top of the physical marketing costs such as transport and handling, marketing margins also reflect costs incurred in the process of conducting transactions between different agents that have imperfect information. Williamson (1979) refers to these costs as transaction costs. Transaction costs are incurred in second-best contractual arrangements, as the first-best solutions are reached only in competitive markets with perfect information. These transaction costs include information costs, bargaining costs, enforcement costs, search costs, externality costs, and the costs of nonoptimal risk sharing. These costs might be substantial, especially in developing-country settings (Fafchamps 2004). A third potential component of marketing costs that are reflected in higher food prices (and partially borne by farmers) is oligopsonistic or monopsonistic profits extracted where middlemen have market power. 3 the population structure,4 2 This shift is more commonly known as Bennett’s law (Bennett 1941). 3 For example, Reardon, Stamoulis, and Pingali (2007) show how changes might be strongly related to geographical locations. the structure of the financial 3 sector, and the reliability of the justice system, among others, are important in shaping the final outcome of the chain. Policy factors also play an important role, be it regulation, hard infrastructure, institutions, international trade, or foreign direct investment (FDI) rules.5 Changes in the supply chain are ultimately transmitted to the rural producer. His or her production environment and livelihood might change due to the different crops that he or she is required to grow and due to changes in input as well as output prices. Moreover, other types of labor, land, inputs and technologies may be used, and new requirements of the market, including transaction requirements (such as postharvest handling) might translate into additional investments. The producer’s behavior is, however, influenced not only by market forces but by nonpolicy conditioning factors and policy factors as well. The rural nonfarm economy will often strongly condition the ability of the farmer to make the requisite investments to respond to the requirements of the transformed supply chain (Reardon, Stamoulis, and Pingali 2007). The differential pull and push factors lead to a difference in food supply chains across countries and products—as reflected in different types of institutional arrangements, which range from spot market exchanges to full vertical integration, in which the stages of marketing, transaction, and production are linked through ownership rather than through market exchanges (Swinnen 2007). These factors will also lead to a different growth path for the modern sector within the food supply chain. Better vertical coordination mechanisms might result in significant cost savings for the firms involved in the modern sector, and economies of scale could potentially lead to lower prices to the consumer, higher prices for the producer, and/or more quantity traded in the sector. However, growth might be constrained by policy factors and conditioning factors, as well as the specific structure of rural supply and urban demand (Swinnen et al. 2008). 4 Increasing urbanization leads to an increasing scarcity of labor in rural areas and might, through induced innovation, force the adoption of new, less labor-intensive technologies. 5 For a more detailed discussion, see Reardon and Timmer (2007). 4 Figure 1. Conceptual framework for understanding changes in food supply chains between rural producers and urban consumers Policy factors - Regulation - Infrastructure - Institutions - International trade - Foreign direct investment Drivers for change Changes in urban demand Changes in marketing sector Changes for rural producers - Urbanization - Income growth - Lifestyle change - Access to different technologies - Quantity - Quality - Variety - Composition - Food safety - Presentation - Transportation - Storage - Processing - Packing/grading - Certification - Traceability - Transacting - Crops planted - Quantity - Input/output price - Labor use - Land use - Input use - Technology use Conditioning factors - Geography - Land tenure - Population structure - Financial sector - Justice system 5 Impact on Poverty The effects of changes in food supply chains on poverty are strongly debated in the literature. Minot and Roy (2007) distinguish four pathways by which they might affect poverty: through a direct effect on farm income, through backward linkages to agricultural input suppliers, by changing wages and employment, and by affecting the food prices faced by consumers. Using this distinction, we briefly review below some of the relevant literature. First, significant research has examined the effects of these changes on farm income. Higher standards might lead to higher technical requirements, which in turn translate into physical investments, human capital investments, more coordination costs, and practice changes at the farm level (Reardon et al. 1999). Minot and Roy (2007) argue that the effects on smallholders in particular depend on five factors, including the relative costs of production, the relative costs of marketing, the agrarian structure, the nature of consumer demand, and the nature of marketing institutions. This theoretical differential effect is reflected in the empirical research (for a detailed overview, see Reardon et al., forthcoming). While some authors find that smallholders are left out of modern supply chains (e.g., Danielou and Ravry 2005; Key and Runsten 1999; Weatherspoon and Reardon 2003), others find that smallholders can also benefit significantly from the changes, often through better access to inputs, reduced production and marketing risks, improved technology use, and higher agricultural productivity (e.g., Natawidjaja et al. 2007; Hernández, Reardon, and Berdegué 2007; Maertens and Swinnen 2009; Swinnen 2007). Farmers involved in high-value agriculture are usually found to perform better than other farmers. However, Minot and Roy (2007) find that most of the reported research is plagued by a lack of controls for confounding factors and selection bias. Second, a few studies have investigated the impact of changes in backward linkages. For example, Kimenye (2002) finds that high-standard green bean production has significantly higher backward linkages with input markets through the increased demand for chemical inputs, irrigation services, and so on. Hernández, Reardon, and Berdegué (2007) and Natawidjaja et al. (2007) find similar results in the cases of Guatemala and Indonesia, respectively. Minot and Roy (2007) argue that changes in backward linkages should become more prevalent due to the growth of high-value agriculture. Due to the increased demand, the organization of supply chains for inputs might improve, as urban-produced equipment and variable inputs might become cheaper because of economies of scale (e.g., von Braun and Kennedy 1994). Third, some research has studied the effect of changing food supply chains on labor markets and employment in the food supply chains or in the agricultural sector itself. Researchers have hypothesized that poverty might be reduced because of the intensive use of unskilled labor in these new supply chains. Some recent studies have found that this impact pathway can be very important in developing countries (e.g., Barron and Rello 2000; Maertens and Swinnen 2009; McCulloh and Ota 2002; Neven et al., forthcoming; Jarvis and Vera-Toscano 2004). Fourth, food prices in rural as well as urban areas might be affected due to the transformation of the urban economy. This might be partly due to changes in demand for quality as prices for products of the same quality go up due to higher demand, or higher qualities are delivered that fetch a higher price overall. Food prices might be different in the modern sector compared to traditional ones. For example, in a cross-country comparison, Minten and Reardon (2008) find that the prices offered in modern retail are significantly lower for almost all types of products in developing countries where modern retail has achieved significant market share. However, when the share is small, as is the case in the poorest countries, prices are found to be equal or even higher, especially for fruits and vegetables (Minten 2008). The increase in high-value agriculture might further lead to substitution away from grain production, potentially leading to price increases for basic staples. Preliminary evidence suggests that rapid growth in high-value agriculture might not imply higher food prices for staples (Minot and Roy 2007). More in-depth analysis is called for, however. 6 Some researchers have further tried to understand how prices are transmitted from urban consumers to producers by examining in detail the restructuring of the value chains. For example, by relying on detailed data from actors within the value chain, Natawidjaja et al. (2007) and Dawe et al. (2008) document the different costs within value chains. Fafchamps and Hill (2008) show how prices are poorly transmitted in traditional markets in Uganda. In theoretical work, Swinnen and Vandeplas (2007), Stokke (2008), and Roe and Diao (2004) model the structure of value chains, analyze the emergence of modern marketing channels, and predict, among other things, the price effect for those connected and not connected to the modern channel. The spatial dimension has also been used explicitly to explore the effect of urban-rural linkages on agricultural marketing patterns and poverty. For example, a number of studies have analyzed how agricultural productivity and marketing patterns among rural households change with distance to urban areas (e.g., Fafchamps and Shilpi 2003; Stifel and Minten 2008; Minten and Kyle 1999). Other studies have examined the relationship between rural poverty and proximity to urban areas (Minot 2008; Minot, Baulch, and Epprecht 2006; Jacoby and Minten, forthcoming). 7 3. RURAL-URBAN FOOD SUPPLY CHAINS IN INDIA Structure of the Food Supply Chain Figure 2 presents a simplified diagram of the food supply chain from rural producers to urban consumers in India. While there might be large differences by product, producers are linked to urban consumers through a variety of channels. As exact numbers on the sizes of the different channels differ by product, and are often hard to obtain, we will limit ourselves to a generic discussion.6 6 Not all channels are relevant for all products—for example, only a limited number of products are procured and distributed by the parastatals, and some products are not being processed. The most important channel from rural producers to urban consumers is the traditional channel that passes through agricultural wholesale markets and traditional urban retail. Four alternative marketing channels are distinguished in our diagram: modern retail/cooperatives, the processing sector, the food service sector, and parastatal marketing. 8 Figure 2. Simplified diagram of the agricultural product flow from rural producers to urban consumers in India Rural Producer Processing Plant Consolidator/ Primary Market Wholesale Market Sub- Wholesaler Traditional Retail Urban Consumer Food Service Sector Modern Retail/ Cooperatives Distribution Center Collection Center Parastatals Public Distribution System (PDS) Shops 9 The alternative marketing channels might not or might partly rely on traditional marketing channels (Figure 1). It seems that the majority of the produce in the alternative market channels originates from the traditional wholesale markets (Reardon et al., forthcoming). However, the alternative channels also develop supply chains that are independent of the traditional sector. For example, some of the modern retailers and cooperatives have set up their own collection centers in rural areas, where they procure directly from the producers (Reardon et al., forthcoming). However, this direct procurement is still relatively less important for modern retail as well as in the rural-urban food supply chains overall, often due to prohibitions by law. The Indian government has also set up its own food procurement and distribution systems, often not relying on the traditional channels. These public intervention programs, however, focus on only a limited number of crops (most notably rice and wheat, but also some others). Even if the government is not directly involved in the food supply chain, it still heavily regulates private agricultural trade, for staples as well as non-staples. Regulation has been deemed important by successive governments as a means to protect vulnerable households, to allow smallholders to receive the best prices, and to prevent exploitation by unscrupulous traders (Shiva 2007; Acharya 2004). These regulations include the Essential Commodity Act (restricting movement and storage of agricultural products), the Agricultural Produce Marketing Act (under which agricultural marketing takes places through a licensed trader system), the Small Scale Industry Reservation (under which most food processing was reserved for small firms, until 1997), and more general policies that also affect other sectors, such as tax policy, border and commercial policies, food laws, and labor policies. However, a number of these laws have been relaxed over time, allowing for a potentially more important role and better investment opportunities for the private sector. Changes in Urban Food Demand Drivers for Change As discussed in Section 2, at least four drivers are contributing to a change in rural-urban food supply chains. They include urbanization, income growth, changes in lifestyle (with more women working outside the home), and access to technology such as refrigerators and microwave ovens. First, the number of people living in cities is increasing, requiring more food to be shipped from rural to urban areas.7 Table 1. Urbanization in India The share of the urban population in India increased from 18 percent in 1961 to 28 percent in 2001, based on the official definition of urban centers used by the Census of India (Table 1). Census Year Urban Population (million) % Urban Population 1961 78.93 17.97 1971 109.11 19.91 1981 159.46 23.34 1991 217.17 25.72 2001 286.20 28.54 2010 (P) 354.94 29.92 2020 (P) 429.61 32.20 Source: Population Census of India, various years; P = projections by Census of India 7 In the Census of India 2001, an urban area is defined as (1) any statutory place with a municipality, corporation, cantonment board, or notified town area committee, etc., or (2) a place satisfying the following three criteria simultaneously: (a) a minimum population of 5,000; (b) at least 75 percent of the male working population engaged in nonagricultural pursuits; and (c) a population density of at least 400 per km² (1,000 per sq. mile). 10 This share is projected to increase further, to 30 percent in 2010 and 32 percent in 2020 (Census of India). In absolute numbers, the urban population in India increased by 69 million between 1991 and 2001 and is expected to increase by another 143 million between 2001 and 2020—equivalent to about half the current population of the United States (Census of India).8 Second, India has been characterized by strong gross domestic product (GDP) and income growth over the past years. While GDP growth was only slightly higher than population growth from the ’60s into the ’80s, it has shown strong per capita growth since the liberalization of the economy in the ’90s. This high economic growth is reflected in household expenditure levels. Table 2 illustrates how per capita expenditure has evolved in the last decades in urban areas, based on the results of the NSSs, the official source for the calculation of national consumption expenditures in India. 9 Table 2. Average value of expenditures per person per 30 days in urban India Application of the commonly used consumer price index (CPI) for industrial workers as a deflator shows that real total consumption expenditures increased by almost 50 percent in 2006 compared to 1972. Current Expenditures (Rs) Real Expenditures (Rs)* Food Total Food Total 1972/73 40.83 63.33 525.97 879.80 1977/78 57.72 96.15 468.26 820.94 1983 96.92 164.0 478.20 865.50 1987/88 140.94 249.92 522.92 975.82 1993/94 250.53 458.00 528.24 1035.03 1997 320.26 645.44 470.92 1026.74 2000/01 402.04 932.80 486.24 1159.87 2006/07 517.00 1312.00 517.00 1312.00 Source: National Sample Survey (NSS) data, various surveys * deflated by the average consumer price index (CPI) for industrial workers, Indiastat (100 = 2006/07) Surprisingly, food expenditures changed little over that period and remained almost stable in real terms.10 Third, changes in lifestyle, with more women working outside the home, also lead to different consumption patterns, as women have less time to spend on food preparation. While attempts to gain a clear picture of the issue are plagued by definitional problems, it is estimated that the participation of women in the urban labor force in India increased from 13.4 percent in 1972 to 16.6 percent in 2004 (Indiastat), a very low number compared to other countries such as China and the United States, for example. Annual per capita urban food expenditures were equivalent to US$139, indicating that total urban food consumption amounted to about US$45 billion in 2006. Using the real expenditure levels from Table 2 and the urbanization rates for 1971 from Table 1, we can estimate that this represents a real threefold increase—driven by the increasing urban population—in the value of urban food expenditures in about 35 years. 11 Fourth, access to microwave ovens and refrigerators has increased over the years. Access to these technologies often leads to different food purchase behaviors, as household are able to buy food 8 In the development literature for countries such as India, urbanization is prone to serious measurement error. For a discussion on the topic, see Sivaramakrishnan, Kundu, and Singh (2005). 9 There is, however, an active debate in the literature on the differential results of the National Sample Surveys (NSSs) and the National Account Statistics (NAS). For an overview, see Deaton and Kozel (2005a; 2005b). 10 In comparing calorie consumption between 1934–84 and 2004–05, Deaton and Drèze (2009) find that average calorie consumption decreased by about 10 percent in rural areas and remained constant in urban areas over that period. Given the relative small price changes for commodities in that period, the numbers presented here are consistent with their findings, which are based on the same data set. 11 The changing age composition of the population—India is one of the youngest nations and could become still younger over time—might also be an important determinant of changes in food consumption. The impact of this change should provide fertile ground for future research. For a discussion on this topic, see Birdsall, Kelley, and Sinding (2001). 11 and store it for longer periods at home. Based on a nationwide National Council of Applied Economic Research (NCAER) survey conducted in 2007, it is estimated that approximately one-third of the Indian urban population had access to a refrigerator in 2007. Changes in Consumption Baskets The changes in these drivers are leading to a different urban—as well as rural—consumption basket. First, the importance of cereals in overall food expenditures is declining, and the importance of high- value products such as fruits and vegetables, meat, dairy products, and fish is on the rise. This trend is illustrated in Figure 3. The share of cereals in total food expenditures in urban areas declined from 36 percent in 1972/73 to 24 percent in 2005/06 (Indiastat). The same trend is seen in rural areas, where the share of cereals declined from 56 percent to 32 percent. Figure 3. Share of cereals in monthly per capita expenditure for total food items in India Source: Indiastat A second trend is the increasing consumption of processed food in richer households in urban areas. Morisset and Kumar (2008a) divide the food consumption basket into different levels of processed food: Primary products are products consumed without processing; these products include fruits and vegetables, eggs, and fluid milk at the farm. First-processing products with low value added are defined as products that undergo minimal processing such as dehusking, milling, drying, and grinding. Examples are rice, flour, pulses, spices, and dried fruits. Value addition is estimated at 0 percent–5 percent. First-processing products with high value added undergo more complicated processing and have a larger value added, of between 5 percent and 15 percent. There is no adding of ingredients, and products are not mixed. Examples are dairy products such as butter and curd, but also meat, fish, and sugar. Second-processing products are products that have as an input a first-processed product and to which another product (a flavor, a preservative, or another ingredient) is added. These include biscuits, bread, ghee, ice cream, and jam. Third-processing food is associated with ready-to-eat food, prepared and packaged meals, and takeout meals (Morisset and Kumar 2008a). 12 Morisset and Kumar (2008a) find that primary products that are not subject to any processing represent only about 17 percent of the value of the urban food consumption basket in India (Table 3). The majority of the consumed foods thus undergo some sort of processing. However, the value added is low (Table 3). Morisset and Kumar (2008a) further compare the level of consumption of processed food to income level in urban areas. When they divide the urban population into 12 income categories, they find that the poorest and richest groups spend about 30 percent and 58 percent, respectively, of their food budget on a relatively highly processed food category (high-value first-processing and second-processing products). Table 3. Indian food consumption expenditures according to the level of processing (in %), 2004/05 Urban Rural Primary products 16.8 15.3 First processing—low value added (0%–5%) 34.8 43.9 First processing— high value added (5%–15%) 38.2 35.1 Second processing 10.2 5.7 Total 100.0 100.0 Source: Morisset and Kumar (2008a) This result implies that when incomes rise, the consumption of processed food increases in importance. Unfortunately, no analysis has been performed on the significance of this trend over a longer period. 13 4. TRADITIONAL FOOD SUPPLY CHAINS Background At the time of India’s independence, agricultural marketing was perceived to be badly organized, leading to low prices for the producer, large physical losses, and high marketing costs. A large number of regulations were thus put in place—including controls on private storage, transport, processing, exports, imports, credit access, and market infrastructure development, as well as small-scale reservation of selected enterprises—to ensure a reasonable income for farmers as well as affordable prices for consumers (World Bank 2007). One of the government’s interventions was to regulate agricultural markets through Agricultural Produce Marketing Committees (APMCs) and to establish a large number of market yards (Acharya 2004). While some have argued that the regulated marketing system has served farmers well over time (Acharya 2004), this view is now increasingly questioned due to several problems with the regulated system. These include farmers being prohibited from buying outside the market yard, the large area served per market yard, the increased importance of bureaucrats in the management of the APMCs, the creation of barriers to entry for newcomers, and the use of market fees as a source of income for the government (Archarya 2004). Wholesale markets (mandis) numbered 268 at Independence. It is estimated that there were around 6,300 wholesale markets in India in 2007 (Chauhan 2008). Acharya (2004) estimates that 98 percent of these markets were in some way “regulated” in 2004. There are also a large number of rural primary/temporary markets (20,870), which might supply wholesale markets located in urban centers. While farmers might have the option—depending on regulation and on the enforcement of this regulation—to go through local village traders, who might be independent or who might work for specific commission agents in the wholesale markets, Fafchamps, Vargas-Hill, and Minten (2008) find that the majority of nonstaple foods is sold directly to brokers or traders on the wholesale markets by the rural producers themselves, even in cases where the law does not require them to do so. Sub- wholesalers, who buy on the wholesale markets but do not sell to consumers themselves, or retailers, who do sell directly to consumers, buy produce on these wholesale markets. The latter then distribute these products by pushcarts, in kirana stores, or at wet markets to urban consumers. Typically, farmers bring their produce to the wholesale market and to the shop of the broker with whom they would like to work. Buyers then pick the produce up from there. Transactions take place mostly by means of an open-outcry auction, managed by a broker who does not take possession but rather just takes commission (therefore called a “commission agent”).12 Agricultural marketing within a particular state is regulated by the Agricultural Produce Marketing (APM) Act, and variation exists between the states in terms of the extent to which the act is implemented. If it is implemented, an APMC is responsible for enforcing the act for each market area. Although more than half the members of this committee were representing the farmers of the market As lots are auctioned, new prices are set. The scant recent research on these traditional marketing systems indicates that these markets (1) are not efficient (Matoo, Mishra, and Narain 2007; Ramaswami and Balakrishnan 2002; Umali-Deininger and Deininger 2001; Thomas 2003), (2) lack integration (Palaskas and Harriss-White 1996), (3) are plagued by trader collusion (Banerji and Meenakshi 2004), and (4) are characterized by a high level of physical wastage (Matoo, Mishra, and Narain 2007). It has also been found that the wholesale market infrastructure for staple as well as nonstaple crops is not very developed (Fafchamps, Vargas-Hill, and Minten 2008). The majority of wholesale markets are not paved, and there are few grading or cold storage facilities. Sanitation facilities are largely deficient, with few public toilets, inadequate drainage, and little or no coordinated pest control. As can be expected, postharvest losses are rather large in this trading environment. 12 Unfortunately, no statistics exist on either their geographic coverage or the percentage of crops they handle compared to the wholesaler who takes possession. 14 area at the start of the regulated market system, elections have not been held regularly, and committees are now often administrated by bureaucrats (Acharya 2004). The committee is empowered to establish markets, control and regulate the admission of traders to the market, charge fees (market, license, and rental fees), issue and renew licenses, and suspend or cancel licenses. Over time, APMCs have emerged as a government-sponsored marketing-services monopoly that prohibits innovations such as contract farming and does not allow traders to buy outside the specified market yards (Acharya 2004). While the APMC also collects significant revenues from market fees, the infrastructure in most markets is largely deficient, as revenues are often directed toward other ends by the government (Umali-Deininger and Sur 2007; Fafchamps, Vargas-Hill, and Minten 2008). Under the APMC marketing system, it seems that the bulk of trade in agricultural commodities takes place at the wholesale market, run and operated by the APMC. The committee allots shops to desirous agents who meet basic eligibility criteria (based on nationality, solvency, and other not particularly restrictive criteria) upon payment of a (rather small) license fee. Typically, the number of license holders greatly exceeds the number of shops available in the yard of the typical wholesale market. Legally, such license holders have equal right to transact business, but effectively the lack of space for trading often severely curtails their business. Licenses once awarded can be easily renewed annually. Invariably, members of the same family transact business at the same shop, which is passed from one generation to the next. Since the number of physical shops is practically fixed and holders rarely return licenses, the advantage enjoyed by license holders who have secured a shop at a given rate is often only reinforced over time. The government has realized that regulation of agricultural markets is less needed in today’s environment and has thus made several agricultural marketing policy reforms in the last decade (see Appendix). Given the perceived problems with the existing regulated agricultural market system in particular, the central government, in consultation with state governments and the private sector, formulated an Amended Act, which was circulated to the states in 2003. The Amended Act proposes removal of the restriction of farmer direct marketing (under the regulated system, notified products can be sold only at markets, to licensed traders), the opening of market infrastructure development to other agencies (especially the private sector), and the establishment of a framework for contract farming. However, this act has not yet been amended by a large number of states; as of the beginning of 2007, of the 28 states in India, 11 states had amended the original act and 14 had not, while 2 had never had the original act in place and 1 (Bihar) repealed it (Chauhan 2008). Data and Methodology for Our Case Study As shown in Figure 2, a nodal point in the traditional food supply chain is the wholesale market, as the majority of marketed produce passes through it (Fafchamps, Vargas-Hill, and Minten 2008). To better understand the functioning of the wholesale market and how it fulfills its role in the chain between rural producers and urban consumers, we rely on a small but unique primary survey that was conducted in the state of Uttarakhand in northern India.13 The survey was conducted in December 2007 at the two main wholesale markets of Uttarakhand: the wholesale market of Dehradun and the wholesale market of Haldwani. Dehradun, the In this state, the APMC Act had not yet been amended at the time of the survey. The aim of the survey was to better understand the activities on these wholesale markets. After initial assessments, it seemed that wholesalers and brokers were not reliable sources of information, as their statements contrasted sharply with statements by the persons with whom they interacted. For this reason, buyers and sellers were interviewed who had just completed a transaction on these wholesale markets and piece together the functioning of brokers and wholesalers based on the interviewees’ declarations. 13 To understand the marketing activities of all agricultural households and to better understand the reasons for selling on wholesale markets (which presumably only a subgroup would do), we would have had to field a representative agricultural household survey. This survey focuses exclusively on marketing on the wholesale market. 15 state capital, is the most important market for the Garhwal hills and the surrounding plains areas.14 The Haldwani wholesale market is the main center of trade for fruit and vegetable produce from the Kumaon hills, as well as much of the Terai plains.15 The survey focused on cauliflower and green peas. At the time of the survey, these were two vegetables that were being marketed in large quantities in the areas around Haldwani and Dehradun. A total of 480 persons were interviewed, 240 at the Haldwani wholesale market and 240 at the Dehradun wholesale mandi. Half of the surveys were conducted with farmers and half with retailers. Half of the surveyed agents were involved in green peas and half in cauliflower. Farmers and traders were both randomly selected after an enumeration procedure. Farmers were interviewed at the wholesale market, while retailers were visited at the major retail markets in the city.16 We will first provide descriptive statistics on the farmers and retailers who participated in the survey (Table 4). There is little demographic difference between the farmers and retailers. While farmers are slightly older (47 years, versus 37 years for retailers), levels of education and sizes of households are similar. About 40 percent of each group are member of a scheduled caste, tribe, or other backward caste. Farmers seem slightly poorer: 29 percent of them carry a BPL (Below the Poverty Line) and 65 percent an APL (Above the Poverty Line) card. The survey contained detailed questions on the demographic background of the interviewee, the reasons for the choice of marketing channel and broker, and linkages with the wholesale market broker used in the last transaction involving cauliflower or green peas. Then, questions were asked on the last transaction, regarding prices and costs, the quality characteristics of the product, quality and quantity assessments by buyers, and the transaction costs that were faced during the transaction. The survey ended with questions on wholesale market practices in general. 17 14 The chief crops traded here include tomato, cauliflower, bottle gourd, radish, and green peas, as well as a large number of fruits. In 2007, the Dehradun market had 13 category A shops, 34 category B shops, 78 category C shops, 56 category D shops, 10 category E shops, and a number of traders and commission agents operating out of tin sheds. 15 The main crops traded here are potato, cauliflower, French beans, capsicum, green peas, and tomato, along with apple, peach, apricot, plum, and mango. The market has about 230 designated shops categorized as A, B, or C, given out to license-holding commission agents, and another 20-odd commission agents operate out of tin sheds (about 250 commission agent licenses have been given out). 16 The three or four largest retail markets were selected in every city. Twenty or fewer retailers of each crop were interviewed at each of these retail markets. They could not be interviewed at the wholesale market, as they were often under time pressure to leave the wholesale market to start their retail activities. 17 Both BPL and APL cards are ration cards distributed by the government to poorer households, to allow them cheaper access to basic necessities. This compares to 30 percent and 51 percent, respectively, for retailers. However, while 47 percent of the farmers own mobile phones, only 25 percent of the retailers do. 16 Table 4. Descriptives of farmers and retailers Unit Farmers Retailers Average or % Standard Deviation Average or % Standard Deviation Demographics Age years 47.2 10.4 37.3 9.8 Level of education years 5.3 4.1 4.8 3.4 Household size number 8.0 3.0 7.0 2.3 Member of scheduled or backward caste/tribe % 37 39 Wealth Has BPL (Below the Poverty Line) card % 29 30 Has APL (Above the Poverty Line) card % 65 51 Owns mobile phone % 47 25 Owns land begha (≈ 1/15 ha) 19.0 23.0 Owns tractor % 29 Owns cattle % 88 Product characteristics (cauliflower, green peas) Average sales Both products kg per day 36.2 31.7 Cauliflower kg per day 28.3 16.1 Green peas kg per day 44.2 40.3 Average production Both products tons per season 15.4 18.1 Cauliflower tons per season 25.4 18.9 Green peas tons per season 5.5 10.0 Sells other products Both products % yes 85 Importance in monetary income Both products avg % 75.2 23.9 41.3 30.3 Experience with: Both products years 12.6 10.9 11.6 8.3 Land cultivated with: Both products begha (≈ 1/15 ha) 21.5 23.5 Cauliflower begha (≈ 1/15 ha) 23.1 20.8 Green peas begha (≈ 1/15 ha) 20.0 25.9 Marketing behavior Distance to wholesale market km 42.1 34.8 3.5 2.1 Visits to this market visits last 2 weeks 10.0 3.4 visits this year 23.5 18.7 Time spent at market hours 3.5 2.2 2.3 1.0 Visits to other mandi % 16 2 The majority of farmers and retailers depend on other agricultural products in their business. Eighty-three percent of the farmers sold other products during the year, and 85 percent of the retailers had sold another product in the last two weeks. Both green peas and cauliflower do, however, have a 17 large importance for these agents: these products make up 75 percent of the annual monetary income of the interviewed farmers and represented 41 percent of the turnover of the retailers over the two weeks prior to the interviews. Both groups have similar experiences in dealing with the products under study. As could be expected, farmers and retailers have different visiting habits with regard to the wholesale markets. Retailers visit almost every day, while farmers come on average 23 times a year. Few farmers (16 percent) or retailers (2 percent) visit other markets. In our further analysis, we first turn to the question of what the major problems are in the current marketing system based on the results of this survey. We will discuss, consecutively, the effect of regulations, the role of brokers and wholesalers, information transmission, service delivery, and competition. In the next section, we will then discuss reasons for the resilience of the traditional system. Problems in Traditional Markets Ineffective Regulations An important rule of the APM Act in effect in Uttarakhand states that the broker rates are 3 percent and that a 2.5 percent tax on each transaction is to be paid to the market officials. The act states that neither of these charges is to be paid by the farmer. Using the data that were collected from interviews with farmers and retailers, we test to what extent these rules are respected. We do so by asking about the costs that farmers and retailers faced during their last transaction on the wholesale market as well as by looking at the effective net prices realized by farmers and retailers (Table 5).18 Table 5 first presents the different rates paid by farmers and retailers to brokers.19 We find that the average broker rate, presumably including the tax fee,20 paid by farmers and retailers is 6 percent and 4 percent, respectively. Fifty percent of the farmers pay 8 percent, while 15 percent state that they pay 0 percent (Table 5). Broker rates differ significantly from the prescribed ones, and regulations by the APMC are not respected in two ways. First, broker rates are charged to farmers, contrary to regulations.21 Second, the combined rates of farmers and retailers are higher than the prescribed 5.5 percent.22 18 It is noteworthy to first look at the sizes of these transactions. The average transaction size is small—for the retailer around 50 kilograms of produce, and representing between US$5 and US$15, depending on the crop. Transactions are significantly larger for farmers, indicating that brokers split up the farmers’ lots. 19 While taxes are in principle paid by the brokers, it is unclear whether they effectively do so for every transaction. In Haldwani, farmers receive slips when they enter the market, on which they indicate how much they are bringing in and which broker they will deal with. This gives the broker seemingly less room for maneuvering. 20 As we interviewed farmers and retailers only, and not the brokers, we could not test to what extent the charged rates were used to pay for the mandated tax. 21 We do not discuss the sense of this regulation. One would expect that in competitive markets, it should not matter for price formation at the farm level to whom the broker rates would be charged. 22 It is also interesting to note the different practices between the various wholesale markets. While at the Haldwani market most of the brokers charge only the farmers, both farmers and retailers are charged (but thus with higher rates in total) at the Dehradun market. It is unclear why rate setting evolved differently for these two markets, despite being governed by the same APMC laws. Table 5 further documents the other costs that farmers and retailers face on the wholesale market. The total reported costs faced by the farmer (7.4 percent) and by the retailer (4.6 percent) amount to 12 percent of the value of the lot. Labor costs (for loading and off-loading) represent 1.8 percent of the total value of the transaction, while phone and weighing costs are negligible. A number of farmers also face “sampling costs” representing 0.3 percent of the value of the transaction. This broker practice implies that brokers take a head of cauliflower or a quarter kilogram of peas and set it apart for themselves, in the name of sampling. 18 Table 5. Characteristics of the most recent transaction of farmers and retailers Unit Farmers Retailers Average or % Standard Deviation Average or % Standard Deviation Values and quantities Cauliflower Quantity traded, mean kg 915 579 56 55 Quantity traded, median kg 800 40 Value of transaction, mean Rs 4,241 2,793 282 250 Value of transaction, median Rs 1,400 200 Green peas Quantity traded, mean kg 220 190 52 40 Quantity traded, median kg 150 50 Value of transaction, mean Rs 1,933 1,549 514 461 Value of transaction, median Rs 3,600 355 Commission rates charged Average % 5.8 2.8 2.6 3.0 Median % 7.0 0.0 Percentage of brokers who charge: 0% % 15.4 56.7 4% % 7.9 0.0 5%–5.5% % 8.3 1.2 6%–6.5% % 18.3 41.2 8% % 49.6 0.4 10% % 0.4 0.4 Costs on wholesale market Commission rates* % of gross price 5.8 2.8 4.0 2.2 Labor costs % of gross price 1.3 1.6 0.5 0.7 Phone costs % of gross price 0.0 0.0 0.0 0.0 Weighing costs % of gross price 0.0 0.0 0.0 0.0 Sampling costs % of gross price 0.3 0.3 0.0 0.0 Total % of gross price 7.4 3.8 4.6 2.4 * includes mandi taxes While statements on costs are informative, the real benchmarks are the net payments that farmers receive and that retailers pay when they leave the wholesale market. To formally test for the size of the differences in these net payments, we run a regression in which we regress the log of the price per kilogram paid by the buyer and received by the seller on a dummy of the retailer and the location of the market. The results are shown in Table 6. They indicate that the price that the retailer pays is significantly higher than the price received by the farmer for the transaction on the wholesale market. The difference is as high as 13 percent in the case of green peas and 26 percent in the case of cauliflower. As price differences could be explained by other potential determinants, such as the quality of the product as well as the day of the transaction, we add these additional controls to the regression (bottom of Table 6).23 23 Quality is self-reported by farmers and retailers. Enumerators were trained in setting objective standards for these indicators. No correction has been performed for potential bias from systematic measurement error from self-reporting. The coefficients remain largely significant, and the size of the coefficient is robust. For both specifications we use a formal F-test to verify that the price difference between retailer and producer is higher than the prescribed 5.5 percent. The results indicate that the 19 margin is significantly higher than the prescribed rates and that marketing regulations on margins are thus not respected. Table 6. Determinants of vegetable prices (dependent variable = log [price per kg]) Cauliflower Green Peas Unit Coefficient t-value Coefficient t-value Parsimonious specification Retailer yes = 1 0.263 9.23 0.127 4.12 Dehradun market yes = 1 0.282 9.91 -0.405 -13.10 Intercept 1.296 52.47 2.345 87.54 Number of observations 240 240 F(2, 237) 91.69 94.25 Prob > F 0 0 R-squared 0.436 0.443 Adj R-square 0.432 0.438 F-test: Price difference between farmer and retailer is higher than prescribed 5.5% F(1, 239) 53.37 5.48 Prob > F 0.00 0.02 Including controls for quality and the day of transaction Retailer 0.244 8.37 0.128 4.53 Medium size yes = 1 -0.128 -3.68 Small size yes = 1 -0.272 -4.67 Mixed size yes = 1 -0.119 -3.13 Number of peas per shell number -0.023 -1.48 Presence of spots yes = 1 0.067 1.90 -0.015 -0.37 Rotten material yes = 1 0.054 1.37 -0.128 -3.05 Less bright color yes = 1 0.029 0.68 -0.098 -2.41 Dehradun market 0.308 7.99 -0.469 -14.11 day of transaction included but not reported Intercept 1.713 12.39 2.506 20.05 Number of observations 239 236 F(x, 220) 19.38 24.15 Prob > F 0 0 R-squared 0.613 0.622 Adj R-square 0.582 0.596 F-test: Price difference between farmer and retailer is higher than prescribed 5.5% F(1, 239) 42.06 6.66 Prob > F 0.00 0.01 The Confusing Role of Wholesalers versus Brokers Two different licenses are given out by APMC market officials for the wholesale market: a broker (a commission agent or a wholesaler who does not take possession) and a trader (a wholesaler who takes possession) license.24 24 The marketing system also features other licenses. Anybody involved in food and agricultural trade (except farmers) needs a license from the APMC. While a commission agent license allows the holder to organize an auction and to help sellers and buyers find each other, a trader license allows the wholesaler to take possession of 20 agricultural produce and to sell that produce, but not necessarily through an auction. We find in our surveys that the same persons often hold both licenses for the same products. While 95 percent of the farmers stated that they sold their produce through a formal auction, this might be less in practice (Table 7), as a quarter of the farmers stated that the buyer was the wholesaler. Two-thirds of the retailers stated that the wholesaler (and not the farmer) was the seller, indicating that the wholesaler took possession of the goods. Table 7. Role of wholesalers and brokers Unit Farmers Retailers Average or % Standard Deviation Average or % Standard Deviation Transactions through auctions % 95 88 Number of buyers in auction number 7.7 5.5 6.5 3.3 Type of buyer Broker % 26 Retailer in town % 29 Reseller at mandi % 18 Trader outside town % 19 Don’t know % 9 Type of seller Broker % 65 Farmer % 25 Trader % 9 Other % 1 Confusion with regard to the exact role of the broker might lead to a conflict of interest. As a broker is paid through a percentage of the final price, he or she has an incentive to obtain a high price for the seller. A wholesale trader, in contrast, profits from the difference between the buyer’s and seller’s prices, and thus has an incentive to lower the price that is paid to the farmer. If an agent acts as a wholesaler and a commission agent at the same time, this might lead to some perverse incentives.25 Imperfect Information Transmission First, it might lead to opportunistic behavior in which, if the farmer is not present at the auction, he or she is charged a commission rate for an auction that never took place. Second, traders would have an incentive to reduce the price information they provide to farmers to a minimum, in contrast with brokers which are supposed to organize transparent auctions. Fruit and vegetable lots are characterized by significant variation in quantity and quality, both observable and unobservable. For buyers to make correct valuation decisions, they need to obtain adequate information on the lot, and mechanisms are thus ideally in place to address the asymmetric information problem. As shown above, the majority of the transactions at wholesale markets take place through auctions. Given that often no personal relationships exist between farmers and the final retailers, and given that there are no repeated transactions between them, which might be an effective way of dealing with information asymmetries (e.g., Kranton 1996), information transmission is even more important. We discuss what practices are used in these markets to assess the quality, the quantity, and wastage, as well as unobservable characteristics related to production practices that might influence food quality. The large majority of retailers believe there are quality differences between the different lots (Table 8). 25 This practice is called “frontrunning” and is prohibited in modern exchange markets (e.g., Chicago Board of Trade). 21 Table 8. Information transmission on quality Unit Farmers Retailers Overall There are quality differences between lots A lot % 1 3 A bit % 93 92 None % 6 5 Buyer/seller receives/delivers lower quality than paid for Regularly % 0 0 Sometimes % 25 68 Never % 74 32 Buyer/seller receives/delivers higher quality than paid for Regularly % 5 3 Sometimes % 68 52 Never % 27 45 Last transaction Buyer had enough information before transaction % yes 83 Buyer checked quality himself % yes 85 If not: How was quality assured? No assurance on quality % 66 Assurance is based on trust with broker % 34 If yes: Time used for checking quality minutes 4.4 Means of quality checking By looks % 100 By touch % 62 By smell % 7 By taste % 34 Buyer was able to check whole lot % yes 34 If only part of the lot, was it representative? % yes 90 Buyer knew about production activities (pesticide use, etc.) % yes 22 Eighty-three percent of traditional retailers report having enough information on quality before placing their bid or completing the transaction. To assess product quality, traders rely exclusively on inspection themselves. In their most recent transaction, 85 percent of the traders checked the quality, spending an average of four minutes doing so. Quality was checked mostly by looking and touching the produce. One-third of the retailers reported even having tasted the produce. While only part of the produce could be checked in most transactions, almost all retailers believed that the checked sample was representative. While modern markets and especially international markets place a high premium on food safety, this is seemingly less the case in these traditional horticulture markets. The use of modern inputs is high in horticultural production in India, but in the current marketing system there is little or no transmission of information on the use of inputs (Fafchamps, Vargas-Hill, and Minten 2008). However, there might be important public health issues related to the lack of proper attention to and control of these issues (Umali-Deininger and Sur 2007; Athukorala and Jayasuriya 2003).26 26 For example, Marshall et al. (2003) tested fresh vegetables at various production sites and at the main wholesale market in Delhi. They found that 72 percent of the spinach samples exceeded the Indian maximum residue levels (MRL) and 100 percent exceeded the Codex MRL. Kumari et al. (2004) found that 26 percent of their samples of seasonal vegetables contained residues above the MRL. In our 22 sample, only one-fifth of the retailers stated that they were aware of the production activities relating to pesticide, fertilizer, and irrigation water use by the farmers. However, even if they are aware, this does not imply that they do not sell unsafe food, as recent research in India shows that there is no premium in traditional retail markets for these unobservable quality characteristics (Fafchamps, Vargas-Hill, and Minten 2008). To correctly assess quantities, lots should be weighed. Eighty percent of the farmers and 73 percent of the retailers say that they know the exact weight of the lot (Table 9). Table 9. Information transmission on quantity Unit Farmers Retailers Overall There are quantity differences between lots A lot % 1 3 A bit % 94 90 None % 5 6 Buyer/seller receives/delivers lower quantity than paid for Regularly % 0 0 Sometimes % 19 62 Never % 80 38 Buyer/seller receives/delivers higher quantity than paid for Regularly % 5 3 Sometimes % 64 29 Never % 31 58 Last transaction Buyer had enough information on quantity % yes 78 Buyer/seller knew exact weight of lot % 80 73 If weighed: Weighed in front of buyer/seller? % 80 93 Type of scale used was Mechanical % 67 73 Electronic % 33 27 Rounding off of weights % 88 86 Rounding off of weight to seller’s advantage % 16 11 Rounding off of weight to buyer’s advantage % 84 89 If not weighed: Differences between standard units? A lot of variation % 0 8 A bit of variation % 88 83 No differences % 12 8 Wastage Buyer knows quantity of wastage at purchase Exactly % 9 Approximately % 57 Not very well % 34 Is there less wastage than expected? Regularly % 46 Sometimes % 41 Never % 13 23 Table 9. Information transmission on quantity (continued) Unit Farmers Retailers Is there more wastage than expected? Regularly % 24 Sometimes % 28 Never % 48 Reported level of wastage* % 0.3 1.4 * Wastage level reported by farmer is during transport; wastage level reported by seller is after purchase. Weighing methods are still old-fashioned, as only about one-third of the weighing transactions were done on an electronic scale. When produce is weighed, farmers and retailers still complain about “rounding-off effects,” that is, where only whole kilograms are paid for. This practice was mentioned by 88 percent of the farmers and 86 percent of the retailers. In most cases, the rounding off was to the advantage of the broker or the buyer. Even when no weighing takes place, farmers and retailers might still feel comfortable with traditional units that are in use—that is, 78 percent of the retailers reported that they were satisfied with the assessment of the quantity. Wastage levels in horticultural marketing are often said to be high in India (Mattoo, Mishra, and Narain 2007), although few good empirical studies are available. Farmers estimated the percentage of the produce that could not be sold at 0.3 percent of all their transported produce.27 Limited Service Delivery by Brokers Retailers evaluated the percentage of wastage because of bad quality at 1.4 percent of the purchased lot. For buyers to correctly value lots, their expectations on wastage level must be well informed. If buyers are not able to ascertain wastage correctly, they may charge uncertainty premiums that are passed on to the farmers. About one-third of buyers state that they do not know very well the wastage of the lot that they will purchase. Nine percent say that they do know exactly, while the majority (57 percent) know the approximate level. Although a large majority of traders seem to have enough information to make a reliable assessment of the quality and quantity of the lot, there is some distrust of the existing system, as illustrated by the asymmetric responses by farmers and retailers to rewards and payments for quality and quantity. More than two-thirds of the farmers believe that they sometimes deliver higher quality and quantity than they are paid for, while one-quarter or less believe that they deliver lower quality and quantity than paid for. The complaints are similar, but in the opposite direction, for retailers. Two types of transaction and marketing costs, for retailers as well as farmers, can be distinguished: monetary costs and the opportunity costs of performing the transaction. The average farmer spends almost two hours to travel to the market. The large majority of farmers (94 percent) use motorized transport to do so (Table 10). 27 Wastage is often not very well defined. Here we refer to wastage as produce that disintegrates and is unsellable as any grade or for any use. 24 Table10. Service delivery Unit Farmers Retailers Average or % Standard Deviation Average or % Standard Deviation Transaction costs Auction Waiting time before auction minutes 68.5 96.3 18.8 15.0 Time that auction took minutes 11.2 9.5 5.6 3.8 Transportation Time to travel minutes 110.5 97.0 23.8 28.4 Time spent at mandi hours 3.5 2.2 2.3 1.0 Means of transport: Motorized transport % 94 53 Animal % 5 0 Bicycle % 4 On foot / with cart % 43 Payment for transport % yes 91 52 How much? Rs per kg 0.4 0.3 0.3 0.2 Payments for transaction When paid normally? Within 3 hours % 82 78 Between 4 hours and 1 day % 12 5 Between 2 and 5 days % 0 2 After 6 days or more % 6 15 Paid in cash % yes 99 100 Information In contact with broker before transaction % yes 5 2 If yes: Number of brokers in contact with number 1.1 0.3 1.4 0.6 Means of contact: By mobile phone % 77 100 By landline % 15 0 Through representatives in person % 8 0 Discussed price level with broker % yes 77 20 Main source of price information Personal observation % 26 40 Speaking with regular customers % 7 Speaking with other farmers % 36 Speaking with other retailers % 14 Speaking with commission agents/traders % 24 20 Screen with price information % 0 Observing prices at auctions % 14 19 Thirty-eight percent of farmers also bring the produce of other farmers when they come to the market. The transport must usually be paid for, and the cost amounts to almost 10 percent of the value of the produce on the wholesale market. Farmers then spend on average 3.5 hours at the wholesale market. 25 Farmers report that they must wait for almost an hour before the auction can start and that the auction takes on average only 10 minutes. However, farmers often still must wait for a while before buyers have taken possession and payment is settled. While a significant number of retailers and farmers own phones, they are still little used for their business. Market transactions are spot transactions in which farmers and retailers arrive without much prior contact (Table 10). Ninety-five percent and 98 percent of the farmers and retailers, respectively, reported that they had no contact with the broker before coming to the market. For those who had had contact, only a limited number had discussed prices with the broker. Most of the price information for retailers and farmers was obtained informally through personal observations or through contacts with fellow farmers or traders.28 Payments for the transactions are in most cases immediate for farmers as well as retailers. Eighty-two percent of farmers state that they are paid within three hours after the transaction. The majority of retailers also report paying immediately for the transaction. However, 17 percent of retailers do not pay immediately and are allowed late payments, often until after their produce has been sold (i.e., more than one day). Agricultural trading is largely a cash economy, as almost none of the transactions are settled by check or other more sophisticated means of payment. Similar results in studies on the importance of spot, unsophisticated, and cash transactions have also been found in other poor agricultural economies (Fafchamps 2004; Fafchamps and Minten 1999; McMillan 2002). Finally, while grading is said to be carried out on the lot, most of the grading is done by the farmer, and little by the broker. 29 Lack of Outside Competition with the Broker System Farmers and retailers were asked whether they had other potential outlets and procurement sources outside the wholesale market where they conducted their last transaction (Table 11).30 Table 11. Choices in outlets and in brokers Unit Farmers Retailers Average or % Standard Deviation Average or % Standard Deviation Outlet options Consolidator % 4 Village market % 26 Broker % 100 89 Reseller at wholesale market % 17 Reseller delivering at shop % 5 Direct marketing at mandi % 22 3 Visit to other market % 16 2 28 There are currently no functioning commodity exchange markets for horticulture that allow farmers to hedge risks, and farmers must thus live with the uncertainty of price formation during the days of their sales. 29 This practice seems to vary by state and product. Gabre-Madhin (2001) found that in Ethiopia most grading of grains was carried out by brokers themselves, while Berdegué et al. (2006) found that in Mexico, for guava, no grading was carried out by farmers in backward zones, in contrast with farmers in advanced zones. 30 It is important to remember that the only farmers who were interviewed were those who came to the wholesale market. They thus do not reflect the farming community as a whole. 26 Table 11. Choices in outlets and in brokers (continued) Unit Farmers Retailers Average or % Standard Deviation Averag e or % Standard Deviation Choice between brokers Perceived number of brokers who sell this product number 62.3 42.5 46.7 21.0 Number of brokers you know who sell this product number 5.3 4.8 6.1 5.0 Number of brokers you use number 2.0 1.7 3.8 2.9 Number of brokers considered today number 1.2 0.5 1.4 1.1 Number of auctions attended number 4.0 3.2 Number of auctions in which this retailer placed a bid number 2.1 1.3 Number of auctions in which this retailer won the bid number 1.2 0.6 Frequency of transactions Number of transactions through brokers This season number 9.6 9.7 Last year number 26.1 27.2 Last two weeks number 9.0 2.7 Number of brokers used for these transactions This season number 1.6 0.9 Last year number 1.8 1.2 Last two weeks number 3.5 2.6 Number of brokers used last season (farmers) / last 2 weeks (retailers) One % 57 17 Two % 33 19 More than two % 10 64 Time dealt with the broker of last transaction years 9.7 9.1 9.2 7.1 Twenty-six percent of the farmers reported that they could potentially sell their produce at a local village market, 4 percent to local consolidators, and 22 percent to retailers at retail markets in town. Farmers thus seem to have few marketing options outside the broker channel, a situation that is partly driven by the existing APMC legislation, as trade can in principle be conducted only by licensed traders. The same holds for retailers. Only 17 percent could go through a reseller (masahari) on the wholesale market, or sometimes they were able to have produce delivered to the shop itself by this type of reseller (5 percent). Few (3 percent) consider direct purchases from farmers to be an option. Although there are few options for farmers and retailers outside the broker/wholesale channel, there are a large number of choices within this channel. Farmers say they could potentially choose from among 60 brokers who deal in the produce that they are selling. Of these brokers, they know five personally, on average. However, they would use only a limited number of them for their transactions, that is, fewer than two, on average (Table 11). Fifty-seven percent of the farmers had considered only one broker to deal with for all their transactions in the last year, and a significant number of farmers thus self-select a specific broker relationship. Often they have a long-term relationship with this 27 broker, as they have dealt with the broker used in the last transaction for almost 10 years, on average. Retailers report a similar large number of brokers to choose from, but they usually work with a larger number than farmers do. Retailers placed a bid in two auctions, on average, and they won in one, on average. It thus seems that they place bids only when they are almost sure that they will win them. The research results indicate that there is little choice for farmers and retailers outside the broker system, but a lot within. Brokers as a group might thus have an incentive to keep the system as it is and try as much as possible to protect their interests from outside competition and potentially to collude inside the market (Olson 1965). They seem to effectively do this through the organization of broker unions that use their power—and money—to lobby with politicians as well as market authorities. This is evidenced by the noncompliance with the rules of the APMC. It is also reflected in the opposition that traders and brokers are able to successfully organize toward organizations that want to bypass them. For example, traders in the states of Uttarakhand and Uttar Pradesh succeeded in delaying modern retail investments in food markets by calling for large strikes. It is especially interesting that the modern retailer Reliance, which had a clear strategy of wanting to bypass brokers and buy directly from farmers, took most of the heat from these protesters.31 Reasons for the Resilience of the Traditional System However, large schemes such as the e-choupal model, which used brokers while bypassing the regular wholesale markets, were easily accepted (Goyal 2008). We thus find that margins for transactions at wholesale markets between the seller and buyer are high (between 13 percent and 26 percent), regulations are not respected, the wholesale trader and broker roles are being confused or potentially manipulated, information transmission is incomplete, transaction times are long, and outside competition with the broker system is limited. This seems to lead to lower sales prices for the producer and higher purchase prices for the retailer. However, farmers and retailers do come to these markets; they must by law, but they do so even when the marketing laws have been relaxed, as seen in other states (Fafchamps, Vargas-Hill, Minten 2008). In the new institutional economics literature, various authors (e.g., North [1990] and Fafchamps [2004]) have shown that institutions are not passive entities and do adjust to the constraints of particular economic environments. In the next section, we will look in greater detail at the services that brokers provide and how the traditional institutions have been able to persist in agricultural markets. Transaction Costs Retailers and farmers were asked why they had chosen the specific broker used in their last transaction. While habit formation is part of the explanation for using a specific broker (50 percent and 29 percent of the farmers and retailers, respectively, state that this is very important), most of the decision is based on an effort to reduce search costs, obtain the best price possible, and speed up transactions (Table 12). Farmers would argue, for example, that even if they have the freedom to sell their produce at fresh markets directly to consumers at higher prices, every transaction would require much more time that way, and they would likely suffer high levels of wastage. Interlinked market services provided by brokers are not terribly important for retailers. Only 15 percent and 4 percent of the retailers would choose a particular broker because they think it very important that the broker provide the option of late payments or access to credit, respectively. However, in the case of farmers, these numbers are 26 percent for the provision of input advances and 21 percent for the provision of credit (Table 12). Although the numbers are higher for farmers than for retailers, they are significantly lower than commonly believed. In the next sections, we turn to a more detailed discussion of these interlinked services. 31 Interestingly, farmers also staged protests to keep modern retail procurement there. They lost, however. 28 Table 12. Marketing options and advantages of the broker channel Unit Farmers Retailers Advantages of brokers considered to be “very important” Last transaction “He finds lots of potential buyers/sellers” % 46 34 “He offers better prices” % 55 64 “He offers higher quality” % 66 “He gives seasonal input advances” % 26 “He allows me to defer payment” % 16 “He offers loans in case of need” % 21 4 “I have the habit” % 50 29 “He completes transactions quickly” % 63 58 Regularity of supply/outlet Buyer/seller is unable to buy/sell this product Never % 89 54 Occasionally % 46 Often % 11 0 If it happens, buyer/seller: Takes it home / sells it next day % 91 Leaves it at wholesale market % 4 Sells to a wholesaler % 4 Trades another product % 75 Does no business that day % 7 Performs wage labor that day % 5 Access to Insurance and Credit A common problem for rural agricultural economies is the prevalence of various types of shocks for which poorer households in particular are ill prepared. Given the lack of formal insurance mechanisms, households must often rely on social capital and the selling of assets to deal with these shocks, and only those households that have these informal insurance mechanisms are able to successfully smooth their consumption (e.g., Dercon 2005; Rosenzweig and Wolpin 1993; Townsend 1994). Some farmers—and a limited number of retailers—use brokers as a source of insurance. Thirty-nine percent of the farmers report that in case of need, the broker would give loans to the farmer for sure (Table 13). Table 1. Credit and insurance services provided by the broker Unit Farmers Retailers Averag e or % Standard Deviation Average or % Standard Deviation Overall Number of brokers from whom farmer/retailer could obtain a loan number 1.1 1.1 0.6 1.2 Broker used in last transaction Broker gives loans in case of need Yes, for sure % 39 7 Probably % 18 22 No % 44 71 29 Table 11. Choices in outlets and in brokers (continued) Unit Farmers Retailers Averag e or % Standard Deviation Averag e or % Standard Deviation Buyer/seller ever received a loan from this broker % 22 2 Number of loans received in last five years number 1.9 1.1 2.4 1.7 Value of loan, mean Rs 8,263 9,429 21,500 38,464 Buyer/seller has other sources of loans % yes 96 95 If yes, from: Bank % 46 17 Friends/family % 78 81 Others % 2 14 Eighteen percent of the farmers think that the broker would “probably” do so. While more than half of the farmers think that they could rely on the broker in case of need, only 22 percent of the farmers have ever received a loan from the broker with whom they dealt in the last transaction. Twenty percent of the farmers have received a loan from this broker in the last five years. The average value of the loan was 8,263 Rs (more than US$200), or about two to four times the value of the last transaction. However, farmers seldom rely exclusively on brokers for access to credit, as 96 percent of the farmers report having alternative sources of credit. These include formal banks (46 percent of the farmers), but also, more importantly, friends and family (78 percent). Access to credit through brokers is less important for retailers. Only 7 percent of the retailers believe that the broker would give loans for sure in case of need, and only 2 percent of the retailers report ever having received a loan from a broker. Retailers also can rely on other options for access to credit, as stated by 95 percent of the retailers. Access to Input Advances A second advantage for farmers of working with specific brokers might be the access to input advances for their agricultural activities. This access proves problematic for some—and often the poorest—farmers all over the developing world, often due to seasonal liquidity constraints (Dercon and Christiaensen 2007). This year, 22 percent of the farmers had received an input advance from the broker they dealt with in the last transaction. For half the farmers, this advance was in kind—more specifically, seeds. No fertilizer or pesticides were given in kind to any farmer in our sample. The average value of the seed advances was calculated at 6,115 Rs, or US$155. The quality of the seeds was evaluated by most farmers to be good (68 percent), and almost half the farmers think they could not find the same quality themselves at the same price. However, they do not believe that the productivity of their vegetables was higher because of the use of these inputs. Almost none of the brokers (11 percent) monitored the use of these seeds. Only 4 percent of the farmers reported having to pay interest on this advance in kind. The other half of the farmers who received input advances (i.e., 11 percent of all the farmers) received cash. The amount was, on average, almost 7,000 Rs (almost US$180). Only one farmer who received this advance in cash was asked to pay interest on it (Table 14). 30 Table 14. Input advances provided by the broker Unit Farmers Average or percent Standard Deviation Overall Number of brokers from whom farmer received input advances this season None % 78 One % 22 Two % 1 Broker used in last transaction Farmers who received input advances % 21 Of those who received input advances: Input advances were partly in kind % yes 55 Details on input advances in kind Farmers who received seeds % 96 Value of seeds received, mean Rs 6,115 7,769 “Quality of seeds” % good 68 “Could have access to market access for same quality” % yes 56 “Productivity of these seeds is higher than if bought myself” % yes 22 “Farmer can find inputs himself at a cheaper price” % yes 15 “Broker checks if inputs were used on farmer’s fields” % yes 11 “Interest payments on this advance” % yes 33 Input advances were partly in cash % yes 55 Details on input advances in cash Amount of cash received, mean Rs 6,982 9,546 “Interest payments on this advance” % yes 4 Farmer receives input advances every year % yes 29 What would happen if these input advances were not paid back? “Broker would not work with me anymore” % yes 84 “Broker would complain to the market authorities” % yes 20 “Broker would complain to the other brokers” % yes 77 “Broker would use social pressure in the village” % yes 55 “Broker would bring me to the police or court” % yes 2 Surprisingly, only 6 percent of the farmers state that they receive input advances from this broker every year. It thus seems that they must make the case every year for their need for this money. We also asked farmers what the broker would do if the farmer did not pay back the input advance. As is usually the case in this type of market, formal enforcement mechanisms are little relied upon (e.g., Fafchamps and Minten 2001; Bigsten et al. 2000; McMillan and Woodruff 1999). Farmers report that it is very unlikely that the broker would go to the market authorities, to the police, or to the court. In the case of default, however, the broker would not work anymore with that farmer, other brokers would hear about it, and some brokers might use peer pressure in the village. This case study in Uttarakhand on horticultural wholesale markets thus illustrates how transactions are conducted in the traditional rural-urban food supply chain. The size of the transactions is small, and little aggregation takes place at the village. We further find that marketing regulations are ineffective, that there is confusion surrounding the function of brokers and wholesalers, that information transmission on quantity and quality between buyers and sellers is incomplete, and that the monetary costs of transactions on the wholesale market are significant. Despite these problems, the 31 broker system has remained prevalent in Indian agricultural markets, partly due to its ability to link credit and insurance markets for farmers to the agricultural output market. The interviewed farmers and retailers also state that they are still largely satisfied with the current system, seemingly because the wholesale market reduces search costs and allows for access to a marketplace with many sellers and buyers; however, in the case of Uttarakhand, there are no effective alternatives yet because of the APMC Act.32 32 An important policy question is whether the setup could be improved, and what the role of the brokers should be in these settings. Brokers in any market are rewarded for reducing the search costs of sellers and buyers (see, for example, Hsieh and Moretti [2003] and Gabre-Madhin [2001]). It is not evident to what extent agricultural brokers in India deliver these services, as buyers and sellers have few marketing options outside the wholesale markets, and both will be physically present (as required by law). As services for the reduction of search are thus limited, the broker system as currently practiced might then be just an expensive way to conduct auctions. Alternative, cheaper options could be explored to conduct such auctions. 32 5. MODERNIZING MARKETS IN INDIA Background Reardon et al. (forthcoming) distinguish two waves of modernization in food supply chains in developing countries. The first wave was led by the public sector, where governments intervened in agricultural markets and tried to streamline food supply chains. The second is more recent, is led by the private sector, and concerns the emergence in food supply chains of the private processing sector, the food service sector, and modern food retail. We follow the same distinction in our discussion below.33 Although modern marketing channels still make up only a minor part of rural-urban food supply chains, they are all growing in importance. Table 15 shows the different growth rates for the four “modern” marketing channels. While caution in the interpretation of the numbers is necessary (we do not separate formal from informal in the processing sector or in the food service sector; however, in both cases, it is estimated that the formal sector makes up the largest part of the sector), 34 Table 15. Growth in modern food marketing channels versus food consumption in India real growth rates were 7 percent for parastatal procurement, 7 percent for the processing sector, 9 percent for the food service sector, and 65 percent for modern food retail. For each channel, these growth rates are significantly higher than growth rates for food expenditures overall (2.5 percent) and for urban food expenditures in particular (3.4 percent). They thus indicate the growing relative and absolute importance of these modern channels. We discuss each of these sectors in greater detail below. Annual Growth Rate Unit Period Source Modern public-sector-led Parastatal procurement 6.59% tons 1996– 2006 Reserve Bank of India (Indiastat) Modern private-sector-led Modern food retail 65.5% real $* 2001– 2008 Reardon, Gulati, and Minten (2008) Processing sector overall 6.75% real Rs** 2002– 2006 Ministry of Food Processing (2008) Food service sector overall 9.20% real Rs** 2001– 2006 Euromonitor International (2007) Food consumption Total urban food consumption 3.41% real Rs# 1997– 2006 National Sample Survey Organisation (NSSO), several years Total food consumption (urban + rural) 2.47% real Rs# 1997– 2006 NSSO, several years * Deflated by urban consumer price index (CPI) (U.S. Department of Labor) ** Deflated by the authors in the referenced document # Deflated by the CPI for industrial workers, India (Indiastat) 33 Following Reardon et al. (forthcoming), we use the term modern to refer to “recent” evolutions in the marketplace. If modernization is equated with improved efficiency, then public-sector-led supply chains such as cooperatives and parastatals might most likely fail that test. 34 It is not always clear exactly which definitions were used to describe sectors. More details are provided in the Section Public-Sector-Led Food Supply Chains and the section, Modern Private-Sector-Led Food Supply Chains. 33 Public-Sector-Led Food Supply Chains Parastatals Parastatal marketing has in recent years become an increasingly important food distribution channel in India, in contrast with most other developing countries, which generally saw increasingly liberalized agricultural markets. Food procurement and distribution are organized by the parastatal Food Corporation of India (FCI) to ensure access to basic staples for the most vulnerable segments of the population. Rashid, Gulati, and Cummings (2008) argue that the Great Bengal Famine in 1943 was the event that provided the momentum for the Indian government to intervene in grain markets. At that time, it was believed that private markets had failed, as there was not a particular shortage of food in that year. The government has since intervened in grain markets with four objectives: (1) to maintain central reserves of food, (2) to import and produce foodgrains, (3) to regulate prices and control interstate movements of foodgrains, and (4) to construct and lease storage facilities (Rashid, Gulati, and Cummings 2008). To this end, the government has sold rice and wheat, as well as sugar and kerosene, at cheaper prices to ration-card holders through a public distribution system consisting of a network of fair-price shops. The degree of intervention in foodgrain markets by the FCI has increased significantly over the years. Its market share in wheat markets increased from 4 percent in 1967/68 to 25 percent in 2004. Its share in rice markets increased from 9 percent to 22 percent over the same period. In 2006, it procured 15 million tons of wheat and 27 million tons of rice, representing 21 percent and 29 percent, respectively, of total production in that year (Table 16). The total number of fair-price shops was estimated at 476,000 in 2004. Despite the program’s laudable objectives, Rashid, Gulati, and Cummings (2008) argue that the rationale for public interventions has mostly disappeared and that the public distribution system is considered to be an expensive and poorly targeted social safety net program. Table 16. Public procurement, offtake, and stocks of foodgrains, 1995/96 to 2005/06 Wheat (million tons) Rice (million tons) Procurement Stocks Targeted Public Distribution System (TPDS) Offtake Procurement Stocks TPDS Offtake 1995/96 12.33 7.76