Private standards and producer risk: a framework for analysis of development implications^**

Abstract

1. Introduction

While governments and development organisations struggle to reach the rural poor through their policies, in practice, agricultural standards are remaking the countryside. International agri-food firms have in recent decades required that supplying farmers follow private sector standards (Henson and Humphrey, 2010). These standards set out stringent rules for production and processing. When they are applied by small-scale farmers in Africa, Asia and Latin America, implications extend into the poorest households on the planet. These households tend to be risk averse (Moscardi and de Janvry, 1977; Hamal and Anderson, 1982; Rosenzweig, 1988; Gomez-Limon, Arriaza and Riesgo, 2003). The purpose of this study is to try to understand how their aversion to risk affects their decision to adopt standards.

Once the exclusive remit of the state, production standards have been taken up by private sector actors, particularly since the 1990s when consumers began demanding quality that went beyond minimum public standards, and agri-food corporations became concerned by quality risks and reputation (Henson and Humphrey, 2010). While public standards are mandatory for importing into a country and retailing a product anywhere within its borders, private standards are voluntary in the sense that they are used by specific firms, and suppliers decide whether or not to supply to those firms and adopt the standard. Private standards tend to be more stringent than public ones, providing a means to match supply to market requirements while helping to reduce the transaction costs of buyer-supplier coordination. They originate from several different sources. Plurilateral negotiating forums such as the International Standardization Organization (ISO) develop industry benchmarks with multistakeholder representation (Buthe and Mattli, 2011; Cao and Prakash, 2011). Consortia of private firms, such as the European retailers who created GLOBALGAP, or individual firm quality standards, such as Tesco’s Nature Choice, set out rules for product homogeneity, safety and other quality attributes. Civil society actors have created schemes such as the Forest Stewardship Council (FSC) to certify compliance with environmental and social norms (Auld, 2014).

An extensive literature shows that these private standards affect the quantity farmers sell, the prices they receive, the level of their income, poverty and how they allocate their labour (Bolwig, Gibbon and Jones, 2009; Maertens and Swinnen, 2009; Handschuch, Wollni and Villalobos, 2013; Subervie and Vagneron, 2013; Ayuya et al., 2015). The context shapes these impacts: the local institutional framework affects the capacity to take advantage of the new scheme (Mohan, 2016), while the original marketing arrangement – for sale to local markets or for export – defines a baseline against which the new standard-governed arrangement will be measured. There is also preliminary evidence that the adoption of standards affects the risks farmers face, including from volatility in prices and sale quantities (Bolwig, Gibbon and Jones, 2009; Handschuch, Wollni and Villalobos, 2013; Mohan, 2017). Given the riskiness of marketing cash crops and farmers’ risk aversion, it stands to reason that when a farmer decides whether to adopt a high-quality standard, the impact of adoption on risk could influence their decision.

Research on standards has focused on how they are used by consumers and their intermediaries. This research has, for example, shown how standards can be used to leverage intermediary power in input markets (Baake and von Schlippenbach, 2011; von Schlippenbach and Teichmann, 2012), signal quality to consumers (Dasgupta and Mondria, 2012; Auriol and Schilizzi, 2015), attain social objectives (Podhorsky, 2015) and incentivize quality revelation (Albano and Lizzeri, 2001; Weaver and Kim, 2002).

The theoretical literature on standards and development has generally assumed that producers are risk neutral. When such research has focused on producers and standards (Bazoche, Giraud-Heraud and Soler, 2005; Xiang et al., 2012; Swinnen et al., 2015; Yu & Bouamra-Mechemache, 2016), it has examined how farmers choose the stringency of standards they adopt, taking into account how the standard affects mean farmer income through prices and market access and in light of market power in the value chain. Yet this focus on the strategic choice of standards by developed-world farmers is less relevant to developing world conditions, where farmers tend to be standard takers rather than makers. Furthermore, this literature has not focused specifically on risk aversion although risk-related concerns are often considered (Bazoche, Giraud-Heraud and Soler, 2005; Swinnen et al. 2015).

To our knowledge, the literature on private standards and development has not yet explicitly analysed the link between risk and farmer welfare. However, several papers on standards have suggested that future research could relax the risk neutrality assumption (Albano and Lizzeri, 2001; Bazoche, Giraud-Heraud and Soler, 2005; Baake and von Schlippenbach, 2011). We intend to address this gap by investigating how developing world small-scale farmers’ aversion to risk affects their decision to adopt standards.

We consider farmer standard choice in an industrial organisation model characterised by quantity uncertainty. The market is set up as a differentiated exchange of high- and low-quality product between consumers, intermediaries and producers. There is a quality information asymmetry between the producers of an exported commodity and its consumers, which is resolved by an intermediary who signals the quality of the good. Although the standard model assumes that all that is produced is sold, this paper follows evidence from agricultural trade and assumes that some of what is produced goes to waste. Producers are risk averse and respond to quantity uncertainty by changing their production choices accordingly. The analysis proceeds by examining how producers optimally partition themselves between high and low-quality standards given quantity uncertainty and prices quoted by competing intermediaries. This market structure emerges through interactions between prices, risk aversion, quantity uncertainty and standard choice.

This framework enables us to study the relationship between standards and risk aversion as broadly as possible. This is important since the empirical setting in which farmers choose to get certified to a standard affects the relationship between the standard and risk. With a view to highlighting the pertinent aspects of the empirical setting that affect the nature of the relationship, this paper will focus on two widely occurring situations: the export enclave setting and a setting whereby farmers have the option to sell locally or into export markets.¹ In developing-country export enclaves, all production is exported, and as such is subject to rejection rates in importing countries. In a mixed export-domestic setting, producers face a choice between selling via a trader that supplies the local market and an exporter that supplies foreign markets. The paper analyses how the characteristics of these settings influence the nature of the risk–certification relationship.

Our main results include the following. We show that in an export enclave, risk-averse producers tend to adopt high-quality standards more than risk neutral ones. We also find that in a mixed export-domestic setting, relatively risk-averse producers find the low-quality standard most appealing. The relationship between risk and standard adoption is affected by market conditions including rejection rates of produce, profit margins and the extent of demand differentiation between the two varieties. Finally, the model suggests that the welfare impact of certification is sensitive to risk aversion and the degree of risk.

This research contributes to the literature on private sector standards by investigating the implications of farmer risk aversion for standard-governed value chains. The use of assumptions derived directly from the stylised facts of fieldwork and case studies among small-scale farmers in the developing world provides real-world decision sequences and modelling that adds institutional richness to the literature. The paper also adds to the theoretical literature on the production implications of risk aversion. Classic scholarship shows that producer risk aversion can lead to suboptimal output levels (Sandmo, 1971) and quality (Leland, 1972). More recent work has analysed how suppliers’ risk aversion affects whether moral hazard in their quality effort can be addressed (Yoo, 2014). We show that farmers’ risk aversion also affects their choice of quality standards.

The remainder of the article is organised as follows. The next section discusses the stylised facts of farmers’ production choices that motivate the assumptions in the paper. Section 3 presents the model. We go on to analyse the equilibrium and present theoretical propositions on the relationship between risk aversion and farmer certification choice. The welfare implications for producers are then studied. The last section summarises and concludes.

2. The nature of trade with private standards

In order to model the role of standards in farmer optimisation in developing countries, it is necessary to first understand how standards operate in agri-food systems there. A full survey would go well beyond the scope of this paper. Instead, this section presents a collection of selected cases and discusses the stylised facts they highlight. Notwithstanding the great diversity of agricultural institutions present in rural areas of the developing world, the cases presented here suggest common features that undergird the proposed model, suggesting it will be of reasonably general relevance.

The case of Nepali tea provides a good starting point to understand rural marketing arrangements in export enclaves (Mohan, 2014, 2016). Small-scale orthodox tea farmers in the South Asian country face a choice between supplying into wholesale bulk tea markets or high-quality value chains. They can produce conventional quality tea and sell it to the local factory, which processes and exports it to neighbouring India. Alternatively, farmers can adopt the organic certification scheme, in which case the factory pays them a higher price for their high-quality standard produce, processes the tea and sells the certified produce to top-price niche specialty markets. Farmers who adopt the high-quality standard tend to receive higher prices but also have higher costs than those who continued to produce using the low-quality standard (Mohan, 2016). Additional costs came from the certification process itself, the purchase of additional livestock for making organic fertiliser and higher levels of labour needed for organic methods. Local processing factories quote a price at the beginning of each growing season, and farmers produce as much as their land will yield, given weather and pest conditions. With an average tea farm of three-quarters of a hectare and a standard deviation of just 0.55, each farmer produced roughly the same quantity. That is, in the short run, with a limited amount of land, the quantity each farmer supplied was fixed. Although the factory accepted all the tea supplied by each farmer, the factory determined after delivery what proportion was deemed to be A-grade (which earned top price) and what proportion was B-grade (which garnered a lower return or none at all). The factory assessment was based in particular on the proportion of the supply that was sold to high-quality markets, which in turn was affected by how much of the farmer’s supply consisted of two leaf and a bud of the tea bush since this small plucking morsel yields the best quality made tea. Several months later, once overseas buyers had chosen which tea they wished to purchase, the factory informed the farmer what proportion was deemed top quality and paid him or her accordingly. Farmers who adopted the organic standard faced lower rates of product rejection and a lower standard deviation in that rate than their conventional counterparts (Mohan, 2017).

The case study illustrates that all produce in developing country agricultural export enclaves is subject to rejection rates in importing countries. Low quality production is for wholesale export through a loosely governed spot market and is often rejected at the border of the importing country or by consumers there because of quality flaws in the product. On the other hand, high-quality production is for lucrative niche export markets and occurs within a tightly governed value chain wherein intermediaries use certification schemes to closely support and monitor suppliers’ quality improvement processes, thereby assuring lower and more stable rejection rates². The two supply chains are often embedded within a single exporting company that sells into both types of markets in the export destination. In both institutional arrangements, the support received for high-quality production – say training sessions to abide by a quality certification scheme such as GLOBALGAP – tends reduce the average rejection rate and lower the variation in that rate experienced by high-quality producers. As such, farmers growing wholesale low-quality produce for export experience high and variable rejection rates, whereas those that adopt a high-quality standard can improve their production processes such that a lower and more stable rejection rate is attained.

The case of Tanzanian pigeonpeas, on the other hand, illustrates the features of a situation where farmers can sell either into domestic or international markets (Jones et al., 2006). Small-scale farmers in the East African country grow pigeonpeas for subsistence and as a cash crop. They can produce pigeonpeas for domestic markets, selling it to middlemen who bring it to urban supermarkets, open-air markets or process it into dhal before it is retailed in Tanzania. Alternatively, they can produce for export including into niche markets for high-quality grain in Great Britain and the EU. These overseas buyers require colour homogeneity, demand compliance with seed cleanliness and breakage standards, and prefer a pre-cleaned product. In order to meet these standards, middlemen in Tanzania buy pigeonpeas from smallholders and sort compliant from non-compliant peas by hand, rejecting any produce that doesn’t meet the grade. In an effort to capture the higher prices associated with these markets and with assistance from a regional NGO, some village groups have tried to increase the proportion of produce that is compliant. This has entailed training sessions in village-level grain cleaning and handling practices, farmer-level sorting and selected adoption of new improved varieties of pigeonpeas that resist disease and have homogeneously cream-coloured seeds.

The case study illustrates that in a mixed export-domestic sales setting in a developing country, domestic markets tend to be relatively lenient concerning quality standards compared to overseas ones. Whilst the local trader has lower quality expectations and thus rejects a very small and stable proportion of the produce, the exporter has to meet the quality requirements of foreign markets and thus rejects a higher and more variable part of the supply. The costs of production are often similar in the two value chains. Low-quality production is sold within the country, either through wholesale or supermarket value chains, with low rates of rejection of produce and little variation in those rejection rates. Produce sold for export has to meet the high-quality standards of governments, importing firms and consumers in the importing countries, but it is sold on wholesale international spot markets with little guidance or investment for producers trying to meet the standard. The adoption of high-quality standards for export can thus go hand-in-hand with a high rate of rejection of produce that varies widely based on weather, infrastructure and the capacity of the farmer. As such, farmers growing low-quality produce for domestic markets experience low and stable rejection rates, whereas those that adopt a high-quality standard are involved in a more exacting value chain that leads to a higher and more variable rejection rate.

Private quality standards are adopted in other empirical contexts. Although the potato farmers in India’s Punjab region all sell into domestic value chains, the lead firms in the marketing channels differ – from large retailers and international multinational corporations to spot traders – and demand different standards and have different rates of product rejection (Singh, 2008). Singh reports, for example, that Frito-Lay rejected 46 per cent of produce supplied by farmers, with some farmers experiencing a rejection rate as high as 62 per cent (Singh, 2008: 301). In other cases, all production may be for bulk wholesale export, but to different countries with differing standards requirements: export to an emerging economy, for example, could require just a basic lower-quality standard, while export to the EU would require compliance with a high-quality standard.

Regardless of the empirical setting, the welfare of small-scale farmers is affected by the uncertainty they face. Uncertainty is thus a key factor in their decision-making calculus. In rural areas of developing countries, it appears that buyers often quote a given price for a whole season and price uncertainty is minimised (Bingen, 2006). As such, the rest of this paper focuses on quantity uncertainty, abstracting from price uncertainty, which has been studied elsewhere (c.f. Bourguignon, Lambert and Suwa-Wisenmann, 2004). Quantity uncertainty can be particularly problematic and arises, inter alia, from production and marketing risks.

Research suggests that farmers’ choices are sensitive to production risks, including from pests, rainfall, flooding and droughts. Rosenweig and Binswanger (1993), for example, find that farmers reduce the responsiveness of their asset portfolio (including livestock, machines and land) to rainfall variation when there is greater expected rainfall variability, while Ogada, Nyangena and Yesuf (2010) show that farmers minimise investment in technologies (such as fertiliser) when there is a higher possibility of crop failure (downside risk). Unfortunately, there is very little data on how standard adoption affects such production risks. The study by Handschuch, Wollni and Villalobos (2013) among Chilean raspberry producers provides the best evidence on production risk and standards, finding that certified and uncertified farmers had virtually identical variation in yields (standard deviations of 4,061 kg per ha for the former and 4,040 kg per ha for the latter). This suggests that it is reasonable to assume that production risks are independent of standard adoption, and as such this paper follows this approach and abstracts away from production level risks³.

Marketing risk in quantity arises when produce is rejected by a buyer owing to poor quality. Some of what farmers grow is categorised as A-grade produce and sells at top price, but the rest is categorised as B-grade and either goes to waste or sells on a separate subsidiary market. Rejection from the target market occurs either on the basis of product-specific quality characteristics such as colour, cleanliness and size of produce, or because of sanitary criteria such as freshness (e.g. the produce goes rotten in transit), defects (e.g. the produce has bruising and damage) or infestation (e.g. high levels of insects in the produce). The fact that a significant and uncertain proportion of product can be rejected because of quality issues has been echoed in other papers. De Janvry, McIntosh and Sadoulet (2011) find that only one-half to one-seventh of Fair Trade certified output actually sells on the Fair Trade market.

The scope of quantity uncertainty differs across marketing channels and risk-averse producers looking for more secure, stable income streams take these uncertainty profiles into account. The risk management tools used by their developed country peers are often unavailable to farmers in the developing world, including because of limited assets and credit market failures (Basu, 1989; Udry, 1994; Ghosh, Mookherjee and Ray, 2001). In an economic environment where farmers have little market power yet live close to subsistence, risk mitigation objectives are met through diversification of income streams (Kurosaki, 1997) and strategic crop (Fafchamps, 1992) and marketing choices (Zusman, 1989). In some settings, farmers have no marketing choices for their cash crop and sell an undifferentiated product to a middleman. As suggested by the cases above, however, farmers sometimes have the option of voluntarily adopting a private standard and selling directly to a high-quality intermediary, who will in turn sell to a downstream wholesaler or retailer who requires compliance with that standard (Jang and Olson, 2010). Each contractual option has attendant income implications as well as risks. High-quality standard-governed contracts usually pay a higher price (Jones et al. 2006). However, there can be high costs of compliance with quality standards, including for the certification process, associated technologies as well as more labour-intensive production methods (Ransom, 2006: 170).

The stylised facts arising from these case studies indicate that farmers decide which crop to grow, taking into account inter alia prices offered and agro-ecological conditions. Taking the crop as a given, they then choose what marketing channel to supply into. Different marketing avenues generate different production, profit and utility functions for the farm. Farmers exercise choice over crop and contract, but can’t influence quantity or price. Their influence over quantity sold is foreclosed by the finite amount of land they have, their helplessness in the face of weather and assuming that they already use best practices in input use and effort against pests and weeds. They have little control over price because of market competition. Finally, farmers are embedded in a complex institutional matrix that includes interlinked markets, informal norms and organisational rigidities (Harriss-White, 1996; Bardhan, 1989, 2003). The next section incorporates these stylised facts as assumptions to build a realistic institutional setting for an industrial organisation model of the market structure that emerges from producer choice of standards.

3. Model

Producers, intermediaries and consumers interact in this model of quality differentiated exchange. Consumers want to buy an agri-food commodity from producers. Some of the consumers would prefer to buy a high-quality good and others would prefer low quality. Producers are small-scale farmers in a developing country who grow a cash crop and cannot successfully signal the quality of their goods to consumers on their own. Intermediaries solve this problem by purchasing the good from producers, signalling its quality and reselling it to consumers. Intermediaries are wholesalers who buy products from farmers and sell on to consumers. There are two types of intermediaries: those who sell a product that meets a high quality standard, and those who sell low-quality produce. The standards themselves are exogenous and their levels taken as a given. Each intermediary invites producers to produce for them and follow the corresponding standard in return for its price.

Every farmer produces according to a standard, choosing whether to adopt the high- or low-quality standard.⁴ I assume that producers do not coordinate with one another or act strategically: instead, each farmer chooses a standard independently. If the other standard yields higher expected utility, s/he will switch to the other standard, which increases its supply and lowers its relative price. The population will continue to sort itself in this manner until the expected utility of the two types of production are equated and no farmer has the incentive to switch, defining the equilibrium organisation of the market. Decisions are made only with respect to short-term outcomes in the single period framework of the model.

If the high standard is adopted, then the farmer faces a contract that corresponds to the high standard. Specifically, farmers that choose the high standard grow a high-quality product, incur high standard costs, face high standard prices and encounter high standard sale quantities. Farmers that instead decide to adopt the low-quality standard face low-standard quality, costs, prices and quantities. In an export enclave, products made according to the high-quality standard garners higher prices, but also incur greater costs than low-quality goods. In a mixed setting, products made according to the high-quality standard garners higher prices and have the same cost as low-quality goods. Getting certified to a standard is the only source that can create these different qualities, costs, prices and quantities.

Farmers face uncertainty which affects which standard they adopt. Specifically, there is uncertainty about what quantity of goods they will sell, and the uncertainty profile differs between the two standards.⁵ Although I assume that every farmer has the same amount of land and produces the same quantity of crop, as per the stylised facts presented in the previous section,⁶ a fraction of what they produce gets rejected by the A-grade target market.⁷ The proportion of the produce that is rejected by A-grade markets is, as per the stylised facts presented in the last section, classified as B-grade; I assume that B-grade produce goes to waste and producers are not paid for it. The proportion of their output that is lost in this fashion, which we call the rejection rate, is a random variable that has a different realisation for each farmer.

In an extension, which is not included here, I explore the possibility that rejected produce is sold on a secondary, lower-value market. For example, rejected high-quality produce could sell on low-quality markets and rejected low-quality produce could sell on animal feed markets. The findings of this revised model are similar to those found in this chapter, but highlight that the income generated through subsidiary markets provides insurance to farmers that reduces the sensitivity of standard choice to risk aversion.

The model is used to analyse two of the most common empirical settings in which agricultural commodities are exported from developing countries: the export enclave and the mixed export-domestic marketing environment. Following the discussion in the previous section, farmers in export enclaves who have adopted the low-quality standard have, on average, more of their produce rejected than those who adopt the high-quality standard. Furthermore, there is more variation among the rejection rates of low-quality farmers. On the other hand, for crops that are sold into both export and domestic markets, the high-quality exporting farmers face higher and more variable rejection rates than those that grow for domestic low-quality markets. Farmers do not know what proportion of their produce will be rejected at the beginning of the season when they decide which standard to adopt.

Consumers have preferences for differentiated quality wherein their demand is partitioned between high- and low-quality varieties based on the quality-to-price ratio. When the two varieties are grown in an export enclave, they may both be sold to the same importing country. In this case, they may share shelf space in stores and individual consumers substitute between them. Alternatively, the two product types may be destined for different countries or regions, as when they are exported to two different importing countries, or in the mixed environment when they retail in an importing country and the producing country. When they are sold across country lines, increases and decreases in produce prices lead to smaller and larger market shares, respectively, in the consuming country market for the variety that meets local quality requirements (assuming that the exporter is a small country). In this case, the demand substitution occurs between markets. In both the single and dual destination market cases, total demand for the product can be conceptualised as a continuum that is divided between high- and low-quality consumers in response to price.

The game timing is as follows. Each intermediary offers a contract to producers that consists of a price that corresponds to its standard. Farmers choose which standard to adopt, thereby partitioning themselves into farmers that produce according to high standards and those that produce according to low-quality standards. The product is then made and the agreed-on quantities are transferred from producers to intermediaries. Intermediaries sell the product to consumers, who reject a proportion of each farmer’s output. Prices are realised as markets clear and intermediaries pay the farmers for the proportion of their produce that has actually been purchased by consumers. The game timing is outlined in Figure 1.

Fig. 1.

Game timing.

Open in new tabDownload slide

An analogy is useful to clarify the nature of quality-related information asymmetries. Suppose we are in a store buying groceries and we wish to buy some grapes. Approaching the fruit section, we see there are no-name, bagged grapes and there are some higher-priced top-grade organic grapes. Given the difference in price between the two goods, a certain proportion of consumers would prefer to buy the higher-quality organic grapes. Their preferences are based on product attributes they cannot see at the produce counter: the richness of flavour, say, or the pesticide content. The intermediating wholesaler, however, has imposed a set of organic and production standards and selectively tested the product for compliance, and so knows whether the product meets these requirements; he signals as such through his brand and certification label, consumers trust the signal and are able to buy the quality level they wish. This corresponds to high-/low-quality information asymmetry.

Yet there is always a chance that some grapes are simply defective: say they are bruised, or have fungus growing on them (Roy and Thorat, 2008). Since neither the producer nor the intermediary can inspect every grape, they do not have full information on this (although the intermediary can predict, based on previous knowledge of his entire shipment, the rough proportion of overall product that is flawed). Consumers, on the other hand, can see the bruised or rotten character directly, and respond by rejecting the product: they refuse to buy it, instead buying a different one. This leads to a certain rate of rejection of produce. There are thus two dimensions to quality: a high/low dimension that is not observable by consumers, and a defective/acceptable dimension that is observable by neither farmers nor intermediaries. These two dimensions of quality information asymmetry roughly reflect two types of quality regulated by standards, with the high-/low-quality reflecting product and process method (PPM) issues and the quality failure reflecting sanitary and phytosanitary (SPS) issues.⁸ Since producers are homogenous, have the same quality information as intermediaries and always produce according to the standard they have chosen, there is no adverse selection or moral hazard in the model.⁹

3.1. Demand

3.2. Production

There is an infinite number of farmers, indexed from zero to 1. A proportion $N$ of them produce according to low-quality standards and $1−N$ produce using high-quality standards, where $N∈[0,1]$⁠. $N$ is determined endogenously. To simplify, we assume that there is no entry of new producers and quantity per farmer is fixed and identical: each farmer is endowed with one unit of land and can produce one unit of output.

Although all actors can perfectly observe whether the product is of high- or low-quality once the intermediaries have signalled as such, there is a positive probability $δ$ that goods have a quality failure (i.e. rotten fruit) and will be rejected by consumers. We assume that in the event of quality failure, the rejected product goes to waste (Weaver and Kim, 2002). There is, thus, a third, implicit quality that is so low the product goes to disposal. The high-quality intermediary knows the total proportion $δ¯$ of product certified to high-quality standards that will be rejected. Likewise, the low-quality intermediary knows that $δ̲$ of the total low-quality product will be rejected.

These industry-level rejection rates are also known by producers, but an individual producer may lose more or less than this proportion. Individual producers do not know the rejection rate $δi$ they will experience. The rejection rate faced by an individual producer is a function of weather, chance, labour availability, overseas purchases and other influences on quality and demand. Indeed, to a given farmer it must seem rather random.¹¹

Following on these stylised facts – particularly that the rejection rate is a proportion ranging from 0 to 1 and that it is randomly distributed – we assume that the proportion of an individual producers’ output that is rejected, $δi$⁠, is a uniformly distributed random variable.¹² If a producer chooses to produce high-quality goods, $δi∈Φ¯$⁠, $Φ¯∼U(0,2δ¯)$ and $E[Φ¯]=δ¯$⁠. If s/he chooses to produce low-quality goods, $δi∈Φ̲$⁠, $Φ̲∼U(0,2δ̲)$ and $E[Φ̲]=δ̲$⁠.

In export enclaves, the industry-level average probability of high-quality product rejection is lower than that of low quality such that $0 <δ¯e<δ̲e< 0.5$⁠. In a mixed export-domestic setting, the industry-level average probability of high-quality product rejection is higher than that of low quality such that $0<δ̲m<δ¯m<0.5$⁠. Average rejection rates are below 50 per cent given that a random, uniform distribution with equal probabilities below and above the mean, ensures that all individual rejection rates are less than 100 per cent: that is, each and every draw of the random variable is a proportion between 0 and 1. The survival rate of high-quality produce can be defined as $(1−δ¯)$ while that of low quality produce is $(1−δ̲)$⁠.

3.3. Intermediaries

There are two types of intermediaries, high and low. We assume here that there are no costs to intermediation and that there are many of each type of firm,¹³ which generates perfect competition in the sector and drives profits to zero. As such, the prices $p¯,p̲$⁠, which intermediaries pay to producers who are certified to high and low standards, respectively, are the same prices that they charge consumers.

Notwithstanding their passive role in transmitting the product, intermediaries serve a crucial quality signalling function that enables trade in both high and low-quality products (Albano and Lizzeri, 2001; Dasgupta and Mondria, 2012; Auriol and Schilizzi, 2015), as we now show.¹⁴

4. Equilibrium market structure

We begin with the most simple scenario, that is, where there are no intermediaries and no risk aversion. We then introduce, in turn, intermediaries and risk aversion.

4.1. Scenario 1: no intermediaries

4.2. Scenario 2: intermediaries and risk neutral producers

By signalling quality to consumers, intermediaries allow producers to sell differentiated goods to consumers. As described above, we assume that there are many perfectly competitive intermediaries of two types: high and low. Recall that the total quantity supplied by producers is always one, regardless of the proportion of them that supply high quality, but this proportion will affect the total that is actually consumed because the rejection rate differs between the two varieties.

So long as we have a separating equilibrium where some producers make high-quality goods $(N< 1)$⁠, total quantity consumed changes under intermediated trade. In an export enclave, the quantity consumed under intermediation will be higher than without intermediation $(Qe>Q˜e)$ since some producers access high-quality tightly governed value chains with lower rejection rates. In regions selling to the domestic and export market, the opposite is the case: the quantity consumed decreases with intermediation $(Qm<Q˜m)$ since a fraction of producers are now exporting into demanding markets that reject a higher proportion of their produce.

More farmers will produce high-quality in equilibrium when consumer preference for high-quality $(α¯)$ is higher, consumer preference for low-quality $(α̲)$ is lower and the cost differential between high and low $(c¯−c̲)$ is smaller.

Let us now introduce producer risk aversion into the model.

4.3. Scenario 3: risk-averse producers

We wish to solve this condition for the equilibrium partitioning of risk-averse producers, $Na⁎$⁠, which is a function of the market-clearing prices specified in equations (10) and (11) and risk aversion.

A specific assumption on the functional form is necessary to solve this framework. I follow the classic paper by Townsend (1994) and recent scholarship in expected utility analysis (Torkamani and Haji-Rahimi, 2001; Kirkwood, 2004; von Gaudecker, von Soest and Wengstrom, 2012; Yoo, 2014) and use the negative exponential utility function $U=a−be−λπ$⁠. The theoretical literature notes that it is relatively well-behaved in expected utility analysis (Hassett, Sears and Trennepohl, 1982; Torkamani and Haji-Rahimi, 2001; Loistl, 1976: 909), where its negative second derivative makes it particularly useful in the analysis of risk aversion: indeed, its central parameter, $λ$⁠, corresponds to the Arrow–Pratt absolute risk aversion (ARA) measure.¹⁶

This equation as well as the market-clearing price equations (10) and (11) make up a system of three equations that define equilibrium partitioning with risk-averse producers $Na⁎$⁠. As the closed-form solution for $Na⁎$ is rather long and involved, we withhold it here and analyse it in more detail analytically.

Recall that $Na⁎$ measures how farmers divide themselves between high-quality standards and low-quality standards. We wish to know how that market structure is affected by changes in risk aversion in the population. If a population becomes more averse to risk, such that $λ$ increases, do more farmers choose to adopt high-quality standards such that $Na⁎$ decreases? The rest of this subsection takes up this question, considering first the export enclave and then the mixed export-domestic setting. 

Analysis shows that three conditions are relevant to the sign of the relationship between producer standard choice and risk aversion. The first condition concerns the relative sizes of the pre-rejection profit margin, $p−c$⁠. Recall that in an export enclave, produce certified to high-quality standards has very high costs and so the low contract has a relatively large pre-rejection profit margin, $p̲−c̲>p¯−c¯$⁠.¹⁹ Indeed, evidence suggests that the high costs of getting certified and high marginal costs from certified production can outweigh the price benefits, such that the per unit profits from certified production are inferior to low-quality profits. This condition says that if the profit margin condition holds, then in export enclaves, more risk-averse populations will be more interested in certification to high-quality standards.

The second condition requires that high- and low-quality product varieties are well differentiated on consumer markets (small $γβ$⁠): this is supported by the great consumer differentiation between certified high-quality goods (e.g. fair trade organic products) and uncertified low-quality goods (e.g. unbranded bulk goods). Finally, there is a large difference in their survival rates (large $1−δ̲1−δ¯$⁠), which is consistent with empirical evidence as well (Mohan, 2017). The  Appendix shows that given this, and if the size of the effects are reasonable, then $dNa⁎dλ<0$⁠: as risk aversion $λ$ increases, $N$ decreases and more producers choose to adopt high-quality certification as their risk aversion increases.

This model shows a situation in which consumer preferences for high-quality goods provide an incentive for intermediaries to insure producers who undertake the process of meeting high-quality standards. Specifically, by adopting such standards, producers reduce the risk of their produce being rejected in end-markets. Although such standards can be demanding in terms of cost and field-level practice, they thus provide an opportunity for improved risk management. The findings here reflect this phenomenon: a relatively more risk-averse population will be attracted to the lower average $δ$ offered by certification to high-quality produce, as well as the lower uncertainty about the value of the rejection rate in any given draw. A relatively more risk-averse population will thus tend to have a higher proportion of farmers who get certified: equilibrium $N$ will be lower.

Figure 2 illustrates Proposition 1. It represents the equilibrium partitioning of producers as a function of the level of risk aversion in the population.²⁰ The figure shows that the equilibrium partitioning of farmers $Na⁎$ is decreasing and convex in risk aversion $λ$⁠. That is, as the level of risk aversion in the population increases, the proportion of producers who adopt the high-quality standard increases.

Fig. 2.

Equilibrium partitioning of farmers for a standard choice given risk aversion.

Open in new tabDownload slide

When some producers export and some produce for domestic markets, these conditions affect the relationship between risk aversion and standard adoption. 

In a mixed context, export prices are likely to be much higher than those earned on local markets. Furthermore, the evidence presented in Section 2 highlighted that in such regions, exports are generally into global wholesale markets where costs are unlikely to be dramatically higher than for local production. As such, the high-quality contract is likely to have higher per-unit profits: $p¯−c¯>p̲−c̲$⁠. Recall that in this setting the rate of rejection is relatively low for producers who adopt the low-quality standard such that $δ̲<δ¯$⁠. These conditions imply $dNa⁎dλ>0$⁠: as risk aversion $λ$ increases, $N$ increases and fewer producers choose to adopt high-quality certification as their risk aversion increases. Intuitively, the low-quality contract has lower uncertainty, but lower profits and will be more attractive to a more risk-averse population.

More generally, it can be shown that the sign of the relationships between $Na⁎$ and $λ$ depend on market conditions, as the following proposition notes. 

There are several scenarios under which the first condition can be met. The  Appendix shows that if, for example, consumer preferences in importing destinations for high-quality goods $α¯$ increases or $α̲$ decreases, then the producer premium for high-quality production increases and the $p̲e−c̲e<p¯e−c¯e$ condition will be more likely to hold in export enclaves. In mixed settings, if the costs of exporting are very high, then the premium from exporting decreases and $p̲m−c̲m>p¯m−c¯m$ is more likely to hold. In these cases, ceteris parebis, the relationship between risk aversion and adoption is inverted. In an export enclave, as risk aversion increases, the proportion of high-quality producers will decrease: $dNa⁎dλ> 0$⁠, while in a mixed environment $dNa⁎dλ< 0$ will hold.

While the variance of profits is a function of the variance of the random variable $δ$⁠, the latter is ‘amplified’ by the square of the price corresponding to that standard. So, for example, farmers who are certified to the high-quality standard in the export enclave model have a RV with a relatively low variance: $Var(δ¯)<Var(δ̲)$²². Yet they face prices that are relatively high: $p¯>p̲$⁠. If the price differential is not great, then it can be assumed that the random variable effect will dominate and $Var(π¯)<Var(π̲)$⁠. If, however, consumer preferences for high-quality are so strong or the availability of high-quality goods so limited that high-quality goods fetch very high prices,²³ this price effect amplifies the random variable variance so as to make the high-quality contract the relatively high-variance option. The low-quality contract may have a relatively higher rate of quantity uncertainty, but the margins at stake are relatively low, so the impact of uncertainty on total income risk is muted. As such, the low-quality contract becomes the lower risk option. This amplification effect makes certification the ‘high stakes’ game and risk-averse producers will turn in greater numbers to the low-quality option. The stylised facts from empirical studies of certification do at times reflect this phenomenon: that while certification can lead to high prices and lowered quantity volatility, it can be a high-stakes game that could scare away producers. In a mixed setting, if the per-unit profits of the low-quality contract are higher than those of the high-quality – for example, if the costs of production for export are dramatically higher than for local production – then the high profits of producing for local low-quality markets amplify the risk of the contract, as in the analysis above. In such a case, the high-quality export contract becomes the relatively low-risk one and will be relatively more attractive to a relatively more risk-averse population.

One can also imagine a situation where the two varieties become more similar on consumer or producer markets. If, say, the degree of consumer differentiation between the two decreases, then the high-quality product may become a small niche market with a high price. This would once again trigger the amplification effect and higher risk aversion would induce a move away from the high-quality standard. If the survival rates of the two varieties become more similar, then the ratio of them decreases and $dNa⁎dλ$ will once again be negative. The effect of a combination of these changes – for example, a decrease in consumer preference for high-quality $α¯$ alongside a increase in the survival rate of high-quality goods – depends on the sizes of the changes and the parameters. Changes in the size of the parameters can shift the relative sizes of terms in the derivative and also lead to changes in the equilibrium relationship between risk aversion and certification.²⁴

Figure 3 illustrates Proposition 3. It represents the equilibrium partitioning of producers in an export enclave as a function of risk aversion given that $p¯−c¯>p̲−c̲$⁠.²⁵ The figure shows that the equilibrium relationship between risk and standard choice has effectively flipped: the equilibrium partitioning of farmers $Na⁎$ is increasing and concave in risk aversion $λ$⁠. That is, as the level of risk aversion in the population increases in this market context, the proportion of producers who adopt the high-quality standard decreases.

Fig. 3.

Equilibrium partitioning with changed market conditions.

Open in new tabDownload slide

More generally, Proposition 3 suggests that the popularity of certification among risk-averse producers depends on market parameters, such as the ex ante consumer demand for certified goods, $α¯$⁠, the cost of producing certified goods $c¯$⁠, the differentiation of goods on end-markets and relative rejection rates. It implies that the underlying demand affects whether a relatively more risk-averse population will tend to be more, or less, interested in certification.

5. Producer welfare

Do farmers benefit from certification? This section answers this question by studying the welfare impact of certification in the model. We examine how certification affects small-scale farmers, starting with a comparison to the pre-certification state. This is followed by a discussion of how the welfare impact of certification interacts with risk aversion and market conditions.

The impact of certification on producer welfare can be found by comparing the expected utility of producers before certification to the expected utility of producers after certification. Before certification, all production and consumption is of low-quality goods, so the utility of agents is a function of low-quality profits, $N= 1$⁠, and the pre-certification price is $p̲=α̲−β(1−δ̲)$⁠. Expected utility of producers pre-certification is thus $E[U(π̲|p̲,N= 1)]$⁠. Once the intermediaries are in place, some farmers can get certified to the high-quality standard and $N< 1$ with $q¯> 0$⁠. In this case, the price attained by low-quality producers is the market-clearing one noted above, namely $p′̲=α̲−γ(1−δ¯)+N[γ(1−δ¯)−β(1−δ̲)]$⁠. Since in equilibrium the expected utility of high- and low-quality producers will be identical, we can focus on the expected utility of low-quality producers to understand the welfare impact of certification; that of high-quality producers follows by implication. The utility obtained by low-quality producers given certification is then $E[U(π̲|p̲',N<1)]$⁠. The welfare gain from certification $ΔW$ is then the difference between the post- and pre-certification expected utility. 

The analysis in the  Appendix yields three conditions under which producer welfare improves with the introduction of high-quality certified production. The first condition notes that the existence of welfare benefits for producers from certification is contingent on a high rejection rate and/or low price for low-quality produce. In export enclaves, this holds since low-quality production tends to have a high rejection rate; in mixed settings, it holds since low-quality production has relatively low prices. Since one of the main benefits from certification for producers in export enclaves is access to a lower average rejection rate and the higher prices from export are seriously appealing in locations that previously just produced for local markets, it makes intuitive sense that these conditions foster welfare benefits from certification. The second and third conditions echo those described in the previous section: the two varieties cannot be close substitutes in consumer markets and there must be a large difference in rejection rates. This suggests that for producers to benefit from the introduction of the high-quality standard, the latter must be quite different in its demand conditions. If these conditions are true, then it is shown in the  Appendix that producers benefit from the introduction of certification.

This analysis illuminates the boundary condition under which certification is welfare-improving for producers (⁠$ΔW> 0$⁠). If, for example, there is little consumer substitution between the two varieties, then certification will increase prices for producers substantially and certification will be welfare-improving. Yet the space provided for the condition to hold is generated by the balance between the two ratios. For example, if the high-quality rejection rate increases, such that $δ¯$ increases, then the cross-price effect $γ$ can increase and the welfare condition will continue to hold. If, on the other hand, there is a high cross-price effect or the difference in rejection rates is low, then $γβ>1−δ̲1−δ¯$ and there will be a negative impact of certification on the welfare of producers. If the low-quality rejection rate is low, there may also be a negative welfare impact from certification. As such, the welfare impact of certification can be negative in our model.

How does the welfare impact of certification change with risk aversion? Shifts in producers’ attitudes to risk affect the risk aversion parameter in their utility function, which in turn affects both the decisions they make and how they evaluate how things have changed with certification. This relationship, which is captured in the present model by how $ΔW$ changes with $λ$⁠, is not simple since risk aversion (⁠$λ$⁠) affects welfare directly through expected utility as well as indirectly through the partitioning rate $N$⁠. The direction of the relationship between the level of risk aversion in the population of farmers and the welfare impact of certification is ambiguous.²⁶ It is not just certification itself, but how certification affects prices, costs and quantities sold that influences how the welfare impact of standards changes in producers’ risk profile. As noted in the discussion of Proposition 3, certification’s new rejection rates interact with price changes to alter the variance in profits, and this affects the welfare of risk-averse agents.

The finding of a positive welfare impact in Proposition 4 and the analysis of the welfare–risk relationship are illustrated through simulation in Figure 4. This figure is generated using the simulation parameters noted previously, which are consistent with the three conditions on Proposition 4. It shows that the difference between post- and pre-certification welfare is positive (⁠$ΔW> 0$⁠), that is, that producers gain from certification, as per Proposition 4. It also indicates that with this parameterisation, an increase in the level of risk aversion $λ$ leads to a larger welfare gain from certification $ΔW$⁠.

Fig. 4.

Welfare effect of certification across risk aversion levels.

Open in new tabDownload slide

The welfare impact of certification varies in the other parameters of the model. It is increasing in the degree of quantity uncertainty, the degree of consumer differentiation among standards and the differential in rejection rates. The welfare impact is also affected by costs. If the cost of production of low quality increases, ceteris paribus, then the welfare gain from certification increases. If the cost of high-quality production increases, then the welfare gain from certification decreases.²⁷ There are lower welfare benefits from certification if ex ante consumer preferences for the low-quality good $α̲$ are higher. If consumers become more willing to pay more for the high-quality good, such that $α¯$ increases, the welfare impact of certification increases.

6. Conclusion

How does the adoption of high-quality standards affect the market structure and welfare of risk-averse farmers in the developing world? This article has taken up this question through an industrial organisation model of trade between consumers, intermediaries and producers. In particular, it has examined how the introduction of risk aversion affects the market structure of agricultural economies in the developing world. When farmers exercise choice over how their crop is produced and marketed – but not over quantity or price – they choose marketing options that maximise their welfare, which includes risk mitigation objectives. In a setting where producers choose whether to get certified to a high-quality standard or not, this model has shown that this choice is acutely sensitive to the degree of producers’ risk aversion.

In particular, the model predicts that given uncertainty, risk-averse farmers in export enclaves will tend to get certified to high-quality standards more than risk neutral farmers, given reasonable empirical assumptions. This highlights a relatively neglected aspect of the development implication of certification schemes such as organic, Fair Trade or Rainforest Alliance: they may also help risk-averse, vulnerable small-scale farmers in the developing world get secure, stable access to lucrative markets. Yet the relationship between the standard’s market structure and risk aversion is sensitive to market conditions in exporting regions. If some producers are exporting into global wholesale markets and some selling into low-risk, low-price domestic markets, the relatively risk averse will tend to avoid the adoption of high-quality standards for export. This analysis sheds light on how key conditions in export settings affect the risk–standard relationship. Important factors include differences in the rejection rates of the contracts associated with high- and low-quality standards; the relative profit margins of high- and low-quality standards and the degree of their differentiation on consumer markets. The configuration of stylised facts in a particular setting affects the risk–standard relationship.

The findings of the model are necessarily contingent on the assumptions – including that price uncertainty is insignificant and that there is perfect information and rationality. Furthermore, on several fronts, the model highlights how welfare implications are contingent on the specific market conditions that obtain in a given case. Nonetheless, several general implications remain. One is that risk aversion has a monotonic impact on farmers’ choice of standards in a closed economy. Another is that producer welfare is increasing in the degree of differentiation of certified and uncertified goods. The nature of the status quo marketing arrangement will affect a farmer’s perception of the relative riskiness of a high-quality standard. In particular, standard adoption is perceived differently in export enclaves and in mixed domestic-exporting regions. Lastly, the relative profitability of certified and uncertified markets is shown to be a crucial ingredient affecting risk-averse farmers’ choice calculus. Indeed, high certified profits can actually amplify the riskiness of getting certified and deter the risk-averse from adopting.

The empirical facts provide preliminary evidence in support of the findings of this paper. Although there is relatively little quantitative evidence on risk aversion and standard adoption, a regression of standard adoption status on an experimental measure of respondent’s risk aversion with Nepali data revealed a positive relationship between risk aversion and certification status, consistent with the prediction of the theoretical framework in this paper (Mohan, 2017). Unfortunately, the absence of data on consumer substitution between high quality and low quality in the same market prevents us from using the data from this study to fully calibrate the model in this paper. More generally, the very sparse data on rejection rates experienced by developing country farmers, and mixed evidence on risk aversion among certified farmers, makes it difficult to test the model here. Further empirical study of quantity uncertainty and risk aversion is thus in order. However, the facts do suggest that it is a simplification to assume that farmers choose a standard solely to access a lower final rejection rate. In reality, one is more likely to find a tiered system of acceptance and rejection, with produce rejected by A-grade markets bought on B-grade markets. In this context, adopting the standard may be a way for farmers to sell a higher and more stable proportion of their produce on lucrative A-grade markets.

These findings, while illustrative, yield an addition series of questions. Given that the real world features both price and quantity uncertainty, how is producer welfare affected by risk aversion given price and quantity uncertainty in a partial equilibrium setting that abstracts from consumer markets? Do certified markets really offer less risky marketing options? Are there selection effects, such that risk lovers choose to get certified first and tend to benefit more from certification? How do farmers choose standards given that in practice they sell into multiple end-markets? These questions illustrate how much remains to be learned about the impact of private sector standards on risk-averse farmers’ welfare.

Footnotes

Review coordinated by Giannis Karagiannis

References

Appendix

 Appendix for private standards and producer risk: a theoretical analysis

Note: The material contained herein is supplementary to the article named in the title.

Producer choice of standards and risk aversion