Abstract
We presented six novel farming vignettes to U.K. farmers that describe trade-offs between pecuniary and nonpecuniary benefits. What farmers would choose corresponds with what they think would make them happier, which supports the use of subjective well-being as a proxy for decision utility in agricultural research. Where a disparity between choice and well-being exists, farmers seem willing to trade happiness for pecuniary benefits. Our results also suggest that farmers often trade pecuniary gains for nonpecuniary benefits. The utility derived from nonpecuniary benefits may help explain farmer behaviors, such as unsubsidized environmental improvements and reluctance to adopt efficiency-enhancing technologies.
1. Introduction
Economists typically infer utility by observing actual behavior under the assumption that individuals make choices based on utility maximization (revealed preference). In recent years, there has been increasing interest in the use of subjective indicators of well-being as a proxy for individual utility. One justification for this approach is that individuals often make choices that do not appear to improve overall subjective well-being, such as taking on stressful high-income jobs. In such situations, it has been argued that self-reported well-being (e.g., measures that ask how happy or how satisfied with one’s life one is) is a more accurate representation than observed choice of the utility one will eventually experience (Stutzer and Frey 2010; Hirschauer, Lehberger, and Musshoff 2015). A further argument for measuring subjective well-being is that it enhances the economist’s toolkit by allowing for welfare evaluations to be conducted in situations where they are difficult to do credibly with revealed preference methods. Central to the appropriateness of using subjective well-being as a proxy for utility instead of revealed preference is the assumption that people make choices that they think would maximize their happiness (Benjamin et al. 2012).
The main question we ask in this study is: do farmers make choices that exclusively maximize their happiness? Although farmers are likely to behave in a similar way to people in general, there are a couple of reasons we believe it is sensible to apply Benjamin et al.’s (2012) approach to a sample of farmers. First, farmers have a set of quite context-specific nonpecuniary goals that feed into their well-being. The literature has made significant progress in generalizing the well-being objectives and motivations of farmers (e.g., Howley 2015). However, we do not yet know which of these motivations farmers would prioritize when faced with a trade-off or whether there are specific contextual tensions associated with the farming lifestyle that would lead to a discordance between choice and well-being maximization. Second, the behavior of farmers is an important driver of the agricultural sector of the economy and thus is crucial to debates surrounding both food security and environmental sustainability. In contrast to a general population sample, farmers can be characterized as self-employed business owners with a different and perhaps richer set of competing objectives. Replicating Benjamin et al.’s (2012) findings in an agricultural context would also help strengthen the case for the use of subjective well-being data as a suitable proxy for utility (and hence a predictor of actual choice behavior) in different decision-making domains.
To answer these questions, we studied six hypothetical trade-off scenarios designed to explore what choices farmers would make when faced with competing benefits (e.g., more environmental conservation vs. higher farm profits) and which they believe would make them happier if they were given the option. Our approach is close in spirit to previous studies comparing well-being and choice in the general population by Benjamin et al. (2012) and Adler, Dolan, and Kavetsos (2017). The main novelty of this study is that we combine the study design from this prior literature with insights from the agricultural literature to study trade-offs that are pertinent to farmers. By looking at response patterns across our scenarios, we ascertain the extent to which stated choice and predicted well-being align. In other words, do farmers seek to maximize their happiness when making farm decisions?
A supplementary aim of this study is to use farmers’ stated preferences from these trade-offs to explore whether nonpecuniary benefits form an important part of a farmer’s utility function as well as result in any associated consequences for farmer decision-making. Given that utility is difficult to measure, models of farm behavior have often treated farms as enterprises and made the simplifying assumption that farms, like firms, are profit maximizers. This is evident in many mathematical economic models of farm behavior, where money is treated as an adequate substitute for utility (Edwards-Jones 2006). We test to what degree nonpecuniary benefits are determinants of farmer utility (and therefore choice) by tailoring our scenarios to represent trade-offs that farmers might commonly face when it comes to competing pecuniary and nonpecuniary benefits. In essence, we are interested in exploring whether farmers are willing to sacrifice some pecuniary gain (e.g., farm profits) in return for some nonpecuniary benefit (e.g., preserving the environment).
In keeping with studies that use more general samples of the population, our results suggest that choice and well-being correspond closely to one another in farmers. In short, farmers appear to make choices that they believe will make them the happiest. Where there are disparities, they appear to arise from a tension between farm income maximization and the preservation of specific nonpecuniary benefits associated with farming. This suggests that farmers do care about things other than their individual happiness. We also find across many scenarios (irrespective of whether we look at stated choice or predicted well-being) that many farmers are willing to sacrifice income or profitability for nonpecuniary benefits. The utility-enhancing properties of nonpecuniary benefits, we suggest, can help explain some outcomes that appear puzzling if looked at from a purely profit-maximizing perspective. For example, farmers often engage in unsubsidized (and even loss-making) environmental practices (Mills et al. 2018; Marr and Howley 2019). Also, the uptake of efficiency-enhancing technologies and new farm practices is often much lower than a simple cost-benefit analysis based on financial returns would predict (Pannell et al. 2006). Such behaviors may not be profit-maximizing but are still consistent with utility maximization once one takes into account the utility-enhancing properties of nonpecuniary benefits associated with many farm activities. Although our findings suggest that nonpecuniary benefits are important, we also offer some preliminary evidence to suggest that the degree to which nonpecuniary benefits influence farmer decision-making may vary across farmers, depending on characteristics such as age, education, and farm type.
Background and Literature
Before proceeding, we present a brief overview of the background literature in choice versus well-being as well as the literature on farmer well-being more generally. First, it is instructive to understand the distinction between choice and well-being. The distinction made by Kahneman, Wakker, and Sarin (1997) between experienced utility and decision utility is a useful starting point. Decision utility is closer to the notion of utility commonly used in an economic context, where the chosen option is the one with the greatest utility attached to it. This form of utility is abstract, difficult to measure, and usually inferred by means of revealed preference; that is, for two real-valued goods x and y: if ∃ x ∈ X where x > 0 such that x > y for all y ∈ X where y ≠ x and y > 0, then u(x) ≥ u(y). On the other hand, experienced utility is more indicative of one’s subjective quality of experience. Experienced utility is commonly measured using self-response questions regarding an individual’s level of subjective well-being. In recent years, principally using primary survey data on stated preferences, a number of studies have explored the degree to which choice (decision utility) and well-being (experienced utility) align for people in general (Benjamin et al. 2012; Adler 2013; Benjamin, Heffetz, Kimball, and Rees-Jones, 2014; Benjamin, Heffetz, Kimball, and Szembrot 2014; Fleurbaey and Schwandt 2015; Adler, Dolan, and Kavetsos 2017). Although there is a separate philosophical debate on the precise interpretation of subjective well-being measures, the literature following Benjamin et al. (2012) has empirically investigated whether measures of experienced utility can be used as an adequate proxy for decision utility. The general conclusion from these studies is that stated choices and predicted well-being broadly align for general population samples (i.e., people make choices that would maximize their happiness). More generally, this literature suggests that the concepts of happiness and well-being constitute the most important component of stated preference, adding weight to arguments that support the use of measures of subjective well-being as a proxy for utility.
In the agricultural literature, a wide array of work has examined decision utility using revealed preference to explain behaviors related to technology adoption and conservation (e.g., Foltz 2003; Lichtenberg 2004). However, even Foltz (2003) highlighted the limitations of this indirect approach for explaining the reasons behind farmer behavior compared with directly asking farmers about their preferences. Increasingly, stated preference methods, such as contingent valuation and choice experiments, have also been used to understand which choices farmers would make in hypothetical scenarios, such as which agricultural policy is preferred between two alternatives with different features (e.g., Ruto and Garrod 2009; Kuhfuss et al. 2016). The main advantage of studying stated/hypothetical preferences is that this approach allows the researcher to consider behavior in a wider array of scenarios than would otherwise be possible.
Aside from measuring stated or actual preferences, there is a nascent literature that uses measures of subjective well-being (i.e., experienced utility) as a means for measuring farmer utility and conducting welfare evaluations. For example, the degree of independence and autonomy associated with different farming practices has been shown to be a strong contributing factor to increased farmer self-reported well-being (Markussen et al. 2018). A further study has shown that organic farmers appear to be happier than nonorganic farmers (Mzoughi 2014), which may signify that conservation motivations feed directly into farmer well-being. O’Brien, Berry, and Hogan (2012) point out the importance of community connectedness for self-reported farmer life satisfaction. Howley et al. (2017) reported that farm income was only weakly related to life satisfaction, suggesting that farmer life satisfaction can be distinct from business success. To the best of our knowledge, existing studies in the farmer well-being literature have not assessed whether the two forms of utility (decision and experienced) correspond with each other (i.e., whether farmers make choices that maximize their subjective well-being).
The findings from more general samples (e.g., Benjamin et al. 2012) suggest that it is quite likely that the choices farmers make also maximize their subjective well-being. To what extent is this true? If farmers care about things other than their happiness, then the level of subjective well-being obtained from an action may not provide a very good approximation of the choices that farmers will make. Taking advantage of the specific trade-offs that farmers face, another question we pose is: do nonpecuniary benefits form an important argument in a farmer’s utility function? Specifically, we aim to understand whether farmers would trade pecuniary gains in return for nonpecuniary benefits and whether we can better understand farmer decision-making as a result.
2. Method
Conceptual Framework
In standard economic theory, it is typically assumed that work is a source of disutility with individuals having to choose the amount of labor to supply to maximize utility (see Rätzel 2012; Spencer 2014 for a review). Increasing time allocated to labor generates utility from consumption via greater income but also reduces utility by reducing leisure time. There is an emerging literature that suggests assuming a disutility of labor may be unrealistic for farm operators or other self-employed individuals (see Howley 2015). Therefore, we refine the traditional labor versus leisure decision problem by positing that farmers have to make choices between competing pecuniary and nonpecuniary activities. Pecuniary benefits generate utility from increasing one’s consumption opportunities, though they may also feed directly into well-being in other ways (e.g., from relative comparison). Nonpecuniary activities directly contribute to utility. Although they may be mutually compatible (e.g., one can derive utility from farm work and that can increase income), there are often situations where they are at odds. As an illustration, we suggest that certain efficiency-enhancing technologies or farm practices may increase pecuniary returns but lead to a reduction in nonpecuniary benefits by reducing the utility experienced from more “traditional” ways of working.
Following Benjamin et al. (2012), we assume that subjective well-being is a uniquely important argument of the utility function. This means that self-reported happiness data could serve as a proxy for decision utility. We test whether this assumption holds in the case of farm operators by developing a series of trade-offs, where we ask farmers which option they would choose and which option would make them happiest. We can then test for any discordance between responses to these questions. Following this, we look at the emergent preferences to see if they can reveal anything about the motivations underpinning farmer behavior. We are particularly interested in the degree to which farmers are willing to trade pecuniary returns for other nonpecuniary benefits.
Well-Being and Choice Trade-Off Scenarios
To examine how farmers approach trade-offs between pecuniary and nonpecuniary benefits and whether there is a discordance between stated choice and anticipated well-being, we followed a similar approach to Benjamin et al. (2012) in developing a series of vignettes. For each vignette, we described two possible courses of action: option A and option B. Immediately below this, participants were shown two questions. First, they were asked to consider which option they believed would provide them with a happier life as a whole. Second, they were asked which option they would actually choose if they were limited to these two options. Respondents were asked the happiness question followed by the choice question immediately afterward, allowing for the comparison of responses within-subjects (Figure 1). Responses were marked on a six-point ipsative scale. Our aim with this design was to answer two main questions for each scenario: (1) Do well-being and choice preferences align when faced with trade-offs involving competing benefits? (2) To what extent are farmers willing to trade pecuniary benefits for nonpecuniary benefits? We were also interested in how farmers would behave when faced with trade-offs between different types of nonpecuniary benefits.
In constructing the scenarios, we aimed to present options that were representative of important decisions that farmers make (e.g., maintain land for conservation or convert to growing crops for additional income). Based on the existing literature, we identified four key and often competing benefits that would underpin these scenarios. These are (1) pecuniary benefits, (2) environmental conservation, (3) social and lifestyle benefits, and (4) farm labor. Our selection of farmer motivators does not preclude the existence of many more. However, we present an argument in the following subsection to explain the rationale behind this selection and why we believe these are applicable and key to most farmers.
Happiness and Choice Trade-Off Question
Note: The descriptive text and options shown in this screenshot correspond to scenario 1. The two questions in italics and the response scales are identical across all six scenarios.
Farming Motivations
Broadly, motivation is a multidimensional construct that aims to provide an explanation for why people are driven to take action at all, both in terms of what people choose to do and in terms of how much energy and effort they devote to an activity (Ryan and Deci 2000). Economic explanations for behavior often focus on extrinsic motivations, such as income maximization, which drive behavior through externally provided rewards or punishments. Indeed, pecuniary benefits are likely to be central to farmer behavior. Farmers (to varying degrees) rely on income from their farm business to support themselves and their dependents. However, there is a rich literature also supporting the importance of nonpecuniary benefits to farmers, many of which are intrinsic motivations in the sense that behavior is driven by personal interests and internal rewards (e.g., Ryan and Deci 2019). Early work by Gasson (1973, 1974), for instance, reported that in addition to the desire to make money, farmers reported that social (e.g., farming tradition), expressive (creativity), and intrinsic (enjoyment of work tasks) aspects were important to them. Following on from this literature, several studies across the social sciences have pointed out that farming is a vocation that is often valued in and of itself, and that farmers often seek to balance the need to maximize incomes with familial and lifestyle objectives (Beedell and Rehman 1999, 2000; Willock et al. 1999; Vanclay 2004; Maybery, Crase, and Gullifer 2005; Howley 2015).
In further support of the idea that nonpecuniary benefits can influence farmer decision-making, previous research has highlighted a number of cases where farmers exhibit behavior that would be against their financial self-interests. Examples include engaging in loss-making production activities (O’Donoghue and Howley 2012), disinvestment reluctance even when land prices are significantly higher than the annualized returns (Musshoff et al. 2013), and working more on the farm even if the off-farm labor market provided greater income gains (Key and Roberts 2009). These behaviors appear to contradict a solely profit-maximization focus but can be explained by the nonpecuniary returns associated with various farm practices. In further support of the importance of nonpecuniary factors, a number of studies have shown that noneconomic motivations can be significant predictors of farmer behavior across a variety of domains (e.g., Darnhofer, Schneeberger, and Freyer 2005; Läpple and Rensburg 2011; Mills et al. 2018; Cullen et al. 2020).
When it comes to the types of nonpecuniary benefits that may be relevant for farmers, environmental conservation has been highlighted as being valuable not just because the deterioration of environmental capital threatens long-term production but also simply due to a general pro-environmental attitude or concern with environmental issues (Marr and Howley 2019). In support of this idea, Mills et al. (2018) found that 25% of environmental activity on arable farms in England are unsubsidized. However, we note that pro-environmental behavior may provide benefits that are not just nonpecuniary. Certain environmental behaviors may enhance the long-run sustainability and financial viability of the farm (e.g., by preserving the quality of soil and fresh waterways, which may boost future yields).
A variety of other nonpecuniary benefits associated with the farming lifestyle have been highlighted as potentially important for farmers. Howley (2015) classified these motivators into two distinct categories: farm labor and social and lifestyle. We adopt this classification in the design of our scenarios. Farm labor is reflective of the intrinsic rewards that farmers derive from farm work independent of any material benefits that may arise from it. Social and lifestyle benefits are derived from farming as a choice of lifestyle more generally (e.g., the “rural idyll,” social interaction, benefits for raising children).
Scenario Development
We suggest that farmers may commonly be faced with trade-offs between the four key benefits that we identified from the existing literature: (1) pecuniary benefits, (2) environmental conservation, (3) social and lifestyle benefits, and (4) farm labor. We designed six hypothetical vignettes that each describe a particular trade-off scenario. For each scenario, we were interested in which option farmers predict would maximize their happiness and which option they would actually choose. Each vignette traded two of the four benefits with each other, resulting in six scenarios.1 The first three scenarios offer a trade-off between pecuniary benefits and one of the nonpecuniary benefits. The remaining three scenarios offer trade-offs between the different categories of nonpecuniary benefits to identify which (if any) is valued most highly on aggregate. Each vignette was also kept intentionally generic to apply to as wide a range of farmers as possible.
Scenario 1 offers respondents a trade-off between (1) and (2): either using land for environmental conservation or growing crops on it for additional pecuniary benefits. Farmers are commonly faced with decisions to undertake activities that would boost farm output but have adverse environmental consequences (Maybery, Crase, and Gullifer 2005). While agri-environment schemes (AESs) such as Countryside Stewardship have continued to operate at least in the short term in the United Kingdom after Brexit, anecdotal comments from an environmental farmer group meeting that one of the authors attended in 2019 suggested that there may be a host of smaller environmental practices that farms incorporate into their operation that are not acknowledged or compensated. They are undertaken for reasons such as personal pride or a sense of responsibility toward the natural environment. This is supported by existing research suggesting that, although certain cohorts of farmers appear unwilling to participate in an AES even when they can set the price themselves (Vanslembrouck, Huylenbroeck, and Verbeke 2002), many others appear willing to engage in unsubsidized pro-environmental behaviors (Lokhorst et al. 2011).
Scenario 2 offers a trade-off between (1) pecuniary benefits and (3) social and lifestyle benefits by offering a choice between a machine that replaces farmhands but generates more profits or to retain human workers (reflecting traditional farming culture) but with less profit. Scenario 3 trades (1) pecuniary benefits with (4) farm labor by offering a choice of whether to adopt a hypothetical technology that would require indoor computer control and yield extra profit versus continuing to work outdoors (farm labor) without extra profit. Scenarios 2 and 3 are designed to represent the types of trade-offs or pressures that farmers face when it comes to “sustainable intensification,” wherein some aspects of farming may have to evolve to incorporate new technologies and farm practices (Wezel et al. 2014). Though changes such as the adoption of new technologies and farm practices may enhance profitability or long-run sustainability, it may also affect (positively or negatively) other nonpecuniary aspects of farming life with associated consequences for overall utility.
Scenario 4 trades (3) social and lifestyle benefits with (4) farm labor. This vignette offers a choice between adopting a hypothetical technology that would save one hour of farm labor to use for leisure (providing social and lifestyle benefits) or continuing to work on the farm without adopting this technology (providing farm labor benefits that stem from an intrinsic motivation to undertake farm work). The final two scenarios deal with the trade-off between wider environmental benefits (which can be seen as a public good) and the more private nonpecuniary benefits that apply only to the farmer in question. Scenario 5 trades (2) environmental conservation with (3) social and lifestyle benefits by offering a choice between spending time to maintain agri-environment measures that are no longer being compensated for or using this time instead to spend with family or in the local community. Scenario 6 trades (2) environmental conservation with (4) farm labor by offering a choice of whether to adopt a new technology that has the potential to benefit the environment at the expense of removing a portion of time required to work on-farm. The full text for all six vignettes used in the survey can be found in Appendix A.
Data
We collected data from two surveys (in summer 2019 and summer 2020) that were sent to a random sample of U.K. farmers selected from a 1999–2013 database of Common Agricultural Policy (CAP) subsidy recipients from farmsubsidy.org as part of a wider project on agricultural policy and the environment (Howley and Ocean 2021, 2022; Ocean and Howley 2021). Invitation letters containing a link to an online survey were mailed to each selected farm address. Some 12,000 farmers were invited in the first survey, and 30,000 were invited in the second survey. The overall response rate of survey 1 was approximately 7.6%, and the overall response rate of survey 2 was approximately 8.5%. Because the surveys contained a large number of questions overall, we minimized survey burden by displaying only two trade-off scenarios to each participant. We obtained 344 responses for scenario 1, 335 responses each for scenarios 2 and 3, and 338 responses for scenario 4. These responses were collected from the first survey. Scenarios 5 and 6 were presented only in the second survey and received 807 and 810 responses, respectively, owing to the larger overall sample size.
For those farmers that completed the demographic questions at the end of the survey, their demographic and farm characteristics appear to correspond broadly to U.K. data (though we do not claim that they are representative of all U.K. farmers and farms). According to the 2016 Farm Structure Survey, 29% (26% in our first survey, 21% in our second survey) of U.K. farms were cropping farms, while 64% (44%, 49%) were livestock (including dairy).2 There was a higher proportion of mixed farms in our sample than in the United Kingdom (24% in each survey vs. 6% nationally). In terms of income (standard output), 61% of U.K. farms in 2016 were below €50,000 and 14% were €250,000 or more. This corresponds reasonably well to our data, where 49% were below ₤45,000 in both samples; 16% were above ₤190,000 in the first survey, while 19.2% were above ₤180,000 in the second survey. However, our sample does represent a higher proportion of large farms. U.K. government farm size data from June 2017 show that 19% of holdings were greater than or equal to 100 ha compared with 55% and 57% in each of our survey samples.3 On the other hand, the demographic characteristics of the farmers were virtually identical to national-level data. We had an identical median age band (50–59) and proportion of females to males (15% in the first survey and 17% in the second survey) in our samples compared with U.K. data.
3. Results
Aggregate Happiness and Choice Preferences for Each Scenario
For each of our six scenarios, we first look at the proportion of farmers preferring option A to option B, both in terms of what they feel would make them happier and in terms of what they would choose. Here, the goal is to establish whether farmers as a whole show a clear preference between option A and B in each scenario as well as to show whether mean preferences are the same between happiness and choice. Table 1 shows that predicted happiness seems to be a good predictor of choice at the aggregate level because the mean responses for scenarios 3–6 generate the same aggregate preference relationship in terms of well-being and choice. That is, for these four scenarios, the option that on aggregate farmers thought would make them happier is also the option that the majority of farmers would choose. In addition, the percentage of farmers who feel they would be happier with option A and the percentage who state that they would choose option A are very similar for scenarios 4–6, which trade nonpecuniary factors against each other. It may also be worth noting that a clear majority prefers the social and lifestyle option to the farm labor option in scenario 4 and the environmental conservation option in scenario 5. This suggests that social and lifestyle benefits seem to be particularly valuable for farmers on aggregate.
In contrast, we observe an aggregate disparity between choice and well-being preferences for scenarios 1 and 2. Looking at scenario 1 in Table 1, we can see that the majority of farmers (52%) believe that maintaining additional environmental features at the cost of additional farm income would make them happier, but a minority (44%) report that they would choose this option. For scenario 2, less than half (45%) report that they would feel happier purchasing a new technology that would save labor effort and boost farm income over continuing to work with other farmers on the farm, but 51% would choose the pecuniary option. In scenario 3, which also trades a pecuniary option with a nonpecuniary one, we do not observe an aggregate happiness-choice discordance as we do with scenarios 1 and 2. Forty-four percent of farmers would be happier to adopt a profit-enhancing technology that would require indoor computer control as opposed to continuing to work outdoors (farm labor), and 48% would choose this option. As in scenarios 1 and 2, however, a higher percentage of farmers indicate that they would choose the pecuniary option compared with the percentage that said that this is the option that would make them happiest.
Considering the scenarios as a whole, we can see that choice and predicted happiness broadly align. In other words, what farmers choose on aggregate broadly corresponds with what they feel would make them happiest. However, a t-test for the difference between the proportion of farmers that would be happier with option A and the proportion of farmers that would choose option A finds significant differences for scenarios 1–3 (Table 1). Specifically, there appears to be a systematic tendency among respondents to favor the pecuniary option more in the choice question than in the happiness question. This suggests that income or profit maximization may be acting at odds with well-being maximization in farmers.
Do Choice and Well-Being Preferences Align within Farmers?
Table 1 outlines that choice and well-being broadly align at the aggregate level, but there may be systematic happiness-choice discordances in trade-offs involving pecuniary benefits. To further scrutinize the relationship between choice and well-being, we examine whether preferences align within each farmer by examining the congruence of response pairs per individual. First, we plot heat maps that provide a visual representation of the pairwise response distribution in terms of responses to the happiness and choice questions for each scenario. Figure 2 illustrates that happiness and choice preferences align within farmers in the majority of cases. This is evident given the concentration of responses along the diagonal. That choice and well-being coincide closely within farmers is supported by high correlations between what farmers said would make them happier and what they said they would choose. These correlations range from a minimum of 0.85 for scenario 1 to a maximum of 0.95 for scenario 4. Therefore, in keeping with analyses on more general population samples (Benjamin et al., 2012), well-being is likely to form a very large portion of the choice utility function for farmers.
Notwithstanding the close correlation between choice and well-being, we still observe a number of cases where there is an incongruence between choice and happiness. This is mostly evident in scenarios 1–3, which involve a trade-off between a pecuniary and a nonpecuniary option. For example, looking at scenario 1 in Figure 2, we see that 11.9% of responses are in the top-left or bottom-right quadrants. This means that 11.9% of respondents preferred a different option depending on whether they were predicting their choice or whether they were predicting their happiness. That 10.1% of this total of 11.9% are concentrated in the top-left quadrant suggests that when there is a preference disparity, it is heavily weighted in favor of farmers who felt that the environmental option would make them happiest but would actually choose the pecuniary option. Table 2 summarizes this information by specifically outlining the proportion of respondents that demonstrated a disparity between their chosen option and their happiness-maximizing option. The first row presents the same information as discussed for scenario 1 in the first heat map. Ten percent of farmers in scenario 1 were happier with option A (environment) but would choose option B (farm income), whereas only 2% of respondents were happier with option B but would choose option A.
Table 2 echoes Table 1 in that it suggests little disparity between predicted well-being and choice for scenarios 4–6 and slightly more for scenarios 1–3. We can formally test whether the happiness question is statistically equivalent to the choice question by comparing the direction of disparity between the cases where choice and happiness diverge using the Liddell test (Liddell 1983). The Liddell test is an exact version of McNemar’s test, which measures the degree of disagreement between matched pairs in terms of two binary outcome variables. The null hypothesis is P (choose A ∩ happier with B) = P (choose B ∩ happier with A), which is a test of whether the top-left and bottom-right quadrants in Figure 2 contain the same proportions of respondents. Table 2 shows that the null hypothesis is rejected for scenarios 1–3, which trade pecuniary benefits for nonpecuniary benefits. This means that when preferences diverge, there is an asymmetry in how they diverge. For farmers who would choose differently to what makes them happier, they appear to be willing to trade reduced happiness for higher income.
Heat Maps of Within-Person Responses to the Six Trade-Off Scenarios
Note: Each cell contains the proportion that responded with the corresponding response pairing. For example, the bottom left cell in scenario 1 indicates that 16.6% of respondents replied “Definitely option A” to both the happiness and the choice questions. Observations on the main diagonal show responses where choice and happiness correspond exactly. Observations in the top-left and bottom-right quadrants represent farmers for whom choice and well-being were maximized with different options.
Heterogeneity in Choice-Happiness Concordance
We estimated separate probit regressions (Appendix Table B1) to determine whether individual and farm characteristics could predict the likelihood of a disparity between choice and well-being preferences.4 The included covariates are gender, age band, income, farm type, education, marital status, children, health, and life satisfaction. The dependent variable is a dummy that is equal to one if the option selected by an individual for a particular scenario differed between the happiness and choice questions and zero otherwise. Specifically, an individual expressing different degrees of preference for the same option would be given the value zero (e.g., possibly choosing A but probably being happier with A is coded as zero). Because relatively few people expressed such a preference disparity overall, to maximize explanatory power, we pooled scenarios 1–3 into a single dependent variable that is one if there was a preference discordance shown for at least one of these three scenarios. This allows us to estimate a single regression to model the characteristics that make it more likely that there will be a conflict between pecuniary and nonpecuniary benefits.
Appendix Table B1 shows that individual and farm characteristics do not seem to predict the likelihood of a farmer having a happiness-choice discordance very well. Although the first three scenarios involve a pecuniary benefit as part of the trade-off, farm income is not a strong predictor of happiness-choice discordance. This suggests that although scenarios involving income are most likely to generate a difference between happiness and choice, this appears to be down to an individual farmer’s utility function rather than the objective level of farm income. The strongest predictor of a happiness-choice discordance in the pecuniary versus nonpecuniary scenarios appears to be farm type. In particular, dairy farmers appear significantly less likely to exhibit a happiness-choice discordance than other kinds of farmers. For example, the predicted probability of having a discordance in responses is approximately 14 percentage points higher for cereal or cropping farms relative to dairy farms. Of the individual characteristics, age appears to have the largest influence on discordance. In the pooled regression for scenarios 1–3, farmers in higher age bands were less likely to express a happiness-choice discordance, which likely indicates that there is less of a dilemma for older farmers when it comes to a choice between pecuniary and other quality-of-life benefits. Farmers younger than 50 were approximately 11 percentage points more likely than farmers 70 or older to express a happiness-choice discordance when they face a trade-off between pecuniary and nonpecuniary benefits. This age pattern becomes stronger if we change the dependent variable in the regression to only capture cases where pecuniary benefits were chosen but where participants felt that nonpecuniary benefits would make them happier (these results are not shown but are available on request). Age is also a predictor of discordance in scenarios 5 and 6, which trade the environment against different types of nonpecuniary benefits. Again, the youngest farmers (those younger than 50) are more likely to exhibit a happiness-choice discordance in these scenarios, although the probability of discordance begins to increase again in older age.
The Importance of Nonpecuniary Benefits
Having looked at the consistency of preferences in terms of predicted choice and happiness, we now scrutinize the actual preferences expressed in the scenarios involving pecuniary versus nonpecuniary trade-offs. First, let us consider scenario 1. Here, farmers were faced with a scenario involving a trade-off between maintaining environmental features on the farm once an AES ends (option A) versus converting the land back to growing crops for an additional 10% increase in farm income (option B).
The preferences expressed here are pertinent in a U.K. context, where many AES contracts are ending and may not be renewed in their original form now that the United Kingdom no longer falls under the EU CAP. At present, the main financial incentive for environmental practices in the U.K. agricultural sector are AESs, which form part of the CAP. They were originally introduced in the EU in the 1980s and have been mandatory for member states since 1992, though voluntary for farmers (Hodge and Reader 2010). By paying farmers for providing environmental services, AESs incentivize them to engage in more pro-environment farmpractices than would otherwise be the case (McGurk, Hynes, and Thorne 2020). Some examples of the types of behavior AESs encourage include crop rotation, reducing fertilizer use, enhancing wildlife habitats, and maintaining buffer strips. Farmers commit to these practices for a set contract length (typically around five years) but are free to return their land to previous uses once the scheme ends (European Commission Directorate-General for Environment 2017). To illustrate the scale of these schemes, in 2020, the total area of land in AESs in England came to 3.6 million ha, which represents approximately 40% of the usable agricultural area (Joint Nature Conservation Committee 2021). Despite their wide coverage, AESs (in combination with other rural development initiatives that fall under pillar II of the CAP) represent less than 20% of the total CAP budget.
While the role of AESs in the EU has grown since their inception, quantifying the environmental benefits has proven challenging, and significant doubts about their environmental effectiveness have been raised. This is because requirements and payments are generally uniform across all farmers, which means the potential for adverse selection remain high (Chabé-Ferret and Subervie 2013). The difficulty is that farmers whose usual practices already satisfy a portion of the scheme’s requirements are those who are most likely to participate. Notwithstanding this problem, some studies have employed quasi-experimental methodologies and found that AESs have provided significant environmental benefits (Pufahl and Weiss 2009; Kuhfuss and Subervie 2018).
The environmental cost of removing AESs will depend on the degree to which farmers would voluntarily maintain environmental features once the scheme contracts end. Scenario 1 suggests that a significant proportion of farmers seem to be willing to sacrifice a degree of pecuniary gain in favor of environmental preservation through intrinsic motivation alone. Just under half of all farmers reported that, when faced with this specific trade-off, they would choose the environmental option, and just over half indicate that this option is the one that would make them happiest. Of course, if option B were to provide a 20% increase in farm income, we would expect to see a much larger proportion of farmers in favor of the pecuniary option. Even so, it is still informative to observe that a significant proportion of farmers appear willing to sacrifice some profitability for environmental conservation. This is consistent with other research, which suggests that many farmers plan to maintain pro-environmental farm practices even after the end of an AES contract (Kuhfuss et al. 2016; Howley and Ocean 2022).
Moving to the other nonpecuniary factors, 49% of farmers in scenario 2 indicated that they would choose to continue to work with other workers on the farm (social and lifestyle benefits) in favor of a machine that replaces the workers with a subsequent 10% increase in profitability (pecuniary benefits). In scenario 3, we can see that, despite the profitability gains, a small majority of farmers (52%) would choose to continue working outdoors (option B) as opposed to adopting a labor-saving and profitability-enhancing technology that requires indoor control. This response pattern illustrates the difficulty in encouraging uptake of new technologies and farm practices more generally. Policy makers often lament that farmers often fail to adopt profit-enhancing technologies or new farm practices despite it being financially optimal to do so (Pannell et al. 2006; Howley 2015). Our results highlight a possible explanation for this: adopting certain technologies may result in the loss of highly valued nonpecuniary benefits.
Overall, we suggest that farmers are commonly faced with difficult trade-offs where they have to weigh up utility gains from increased farm income associated with new farm practices against the disutility stemming from losses in other nonpecuniary benefits. That many farmers would still prefer nonpecuniary options to pecuniary ones highlights the fact that they cannot be treated as solely profit-maximizing entities. This is supported by observed behavior (Pannell et al. 2006; Mills et al. 2018; Marr and Howley 2019), where there has been a tendency for some farmers to engage in unsubsidized or even loss-making environmental practices, reluctance to adopt new efficiency-enhancing technologies or farm practices, and allocating more time to farm labor even when faced with higher returns in the off-farm labor market. A further contraindication for the use of neoclassical models in agriculture is the apparent intrinsic motivation toward farm work in itself. In scenario 4, although a clear majority of farmers report that they would be happier and would choose extra leisure time instead of extra farm labor, over one-third of farmers still preferred extra outdoor farm work to extra leisure time. This contradicts standard economic models of labor supply, which treat labor as only a means to generate income.
Heterogeneity in Preferences across Farmer Subgroups
The analysis illustrates that nonpecuniary benefits form an important part of the utility function for many farmers. To add further richness to our understanding of which types of farmers might be more likely to prefer one of the two options in general, we split each scenario into subgroups based on the following covariates: farm type, education, gender, age, and whether they had children. Our aim with this analysis is to offer a preliminary descriptive examination of whether the behavior of farmers with specific characteristics are more likely to be affected by the presence of nonpecuniary benefits.
Appendix Table B2 shows the proportion that preferred option A for both happiness and choice across the subgroup categories for each scenario. Looking first at scenarios 1–3, we can see that across the major farm classifications, dairy and crop farmers appear to value pecuniary benefits relatively more than do livestock farmers. One possible explanation for this is that dairy and crop farms in the United Kingdom tend to be much more intensive, less reliant on subsidies, and more business oriented than livestock farms. In addition, scenarios 4 and 6 suggest that farm labor may be relatively less important to crop and dairy farmers than to livestock farmers. This could be a result of the differences in what constitutes “farm work” across different types of farms. It may suggest that the cases in which farmers might paradoxically prefer to spend more time laboring on the farm are likely only when the type of farm work involved has sufficient intrinsic rewards and these rewarding types of work may only be available on certain types of farms.
In terms of individual characteristics, Appendix Table B2 suggests that less educated farmers may have more of an attachment to the nonpecuniary benefits associated with farm labor than more educated farmers. For example, the proportion of farmers preferring the pecuniary option over the farm labor option (scenario 3) is 19 percentage points higher in those with a degree relative to those without a degree. The other two scenarios involving farm labor (4 and 6) also see a significantly higher proportion of farmers with a degree preferring the alternative nonpecuniary option over farm labor. In scenario 4, the proportion of farmers preferring the social and lifestyle option over the farm labor option is 13 percentage points higher for those with a degree, whereas in scenario 6 the proportion preferring the environmental benefits option to the farm labor option is 18 percentage points higher in farmers with a degree.
Second, it appears as though female farmers may place a lower value on pecuniary benefits than do male farmers. Scenarios 1–3 show that a relatively higher proportion of male farmers would be happier and would choose the pecuniary option over the nonpecuniary option, though the relatively low proportion of female farmers in our samples reduces the reliability of any gender comparison. In relation to age, we observe in scenarios 1–3 that older farmers are more likely to choose the nonpecuniary as opposed to pecuniary option than comparatively younger farmers. For example, in scenarios 2 and 3, 62% of farmers younger than 50 years choose the pecuniary in favor of the social and lifestyle and farm labor option, but this compares to 45% and 44%, respectively, for farmers older than 60. A similar picture is evident in scenario 1 where 49% of farmers older than 60 would choose the environmental as opposed to pecuniary option, but this falls to 32% for those younger than 50. Although we cannot distinguish between life-cycle effects and generational effects from these data, it is perhaps intuitive that older farmers are less concerned with income. For example, they may already have previously accumulated wealth or have reduced family expenditure requirements, which would lead to a comparatively lower marginal utility of income.
Finally, one might expect farmers with children to be more interested in income maximization due to a bequest motive. Our results suggest that this is unlikely. The differences between preferences in farmers with children and farmers without children top out at a maximum of 4 percentage points across scenarios 1–3. In scenario 5, there is a 12 percentage point difference: relatively more farmers without children would be happier with environmental benefits to social and lifestyle benefits than those with children. However, this pattern is not repeated in scenario 6, where a similar proportion of farmers with and without children prefer environmental benefits to farm labor.
4. Discussion
In summary, this study has made two main contributions to the literature. First, we have confirmed that choice and well-being align strongly in farmers, as it appears to do so in a wider population sample more generally (Benjamin et al. 2012). In what we believe is the first study to analyze farmer-specific trade-off scenarios in a farming population, we find that farmers would generally choose what they also believe would make them happier (if they were given the option). This supports the theoretical idea that subjective well-being (i.e., experienced utility) corresponds closely to the more traditional economic concept of decision utility. It also lends some credibility to the use of subjective well-being measures as a proxy for farmer decision utility.
Although our results support the finding that happiness is a uniquely important argument of the utility function (Benjamin et al. 2012), we identify situations where farmer choices do not appear to maximize their happiness. This is apparent in the scenarios that trade pecuniary gains against nonpecuniary gains. Such a disparity between happiness and choice appears to be more likely to occur in younger farmers. One possible explanation may be due to responsibility utility (Comerford and Lades 2022). Farmers may feel a responsibility to seek an income-maximizing choice but may not desire this option overall from the perspective of their individual happiness (e.g., because of family). Instead, it may only be possible to maximize their well-being when an exogenous third-party is able to choose on their behalf (Comerford and Lades 2022). Another explanation for the disparity could be that farmers feel as though they will become happier in the future from choosing to maximize income today (i.e., maximizing lifetime happiness rather than short-run happiness). However, they may be overestimating the future well-being impact of more income today because of the “impact bias” (Wilson and Gilbert 2005). Specifically, people are poor at predicting the intensity and duration of feelings caused by future events. Therefore, it is possible that any perceived future increases to happiness from extra income today may be less effective than anticipated.
Second, our findings also suggest that nonpecuniary attributes, such as environmental conservation and intrinsic rewards from farm work, are important arguments in a farmer’s utility function. Therefore, they are likely to influence decision-making in an agricultural context. Our scenarios were designed to illustrate how farmers are commonly faced with difficult trade-offs, where they have to weigh up increases in farm income associated with new farm practices against losses in other nonpecuniary benefits. That many farmers appear willing to sacrifice a pecuniary gain in return for nonpecuniary benefits demonstrates that farm owners cannot be treated solely as profit-maximizing entities. In turn, the omission of nonpecuniary benefits for farmer behavior from economic models of agriculture may lead to an inaccurate assessment of the impact of a change in agricultural or environmental policy. Our subgroup analysis also points to differences across farmers regarding the extent to which nonpecuniary benefits matter. For example, relatively older farmers, those with comparatively less education, and livestock farmers all appear to be relatively more likely to be impacted by the presence of nonpecuniary benefits in their decision-making.
The presence of nonpecuniary benefits makes the challenge of predicting how farmers will react to new interventions such as new policies (e.g., payments for ecosystem services) or the development of new technologies challenging, especially because traditional models may not accurately capture farmer motivations. To illustrate this point, consider the recently passed U.K. Agriculture Act 2020, which promises to change how agricultural subsidies are distributed following the United Kingdom’s withdrawal from the EU and subsequently the CAP (Coe and Finlay 2020). The CAP is designed to provide financial support to farmers in EU member states. Payments under CAP account for over half of all farm incomes in the United Kingdom and so have been central to the financial sustainability of farm businesses. AESs cofunded by EU member states by the CAP have been the main mechanism used to deliver environmental benefits on agricultural land in Europe. Notwithstanding the support for AESs, the CAP has been criticized as contributing to widespread biodiversity loss (Pe’er et al. 2020). This is because the majority of CAP support consists of direct payments to farmers with limited requirements to meet environmental objectives (so-called cross-compliance). The Agriculture Act 2020 plans to move toward paying farmers to produce public goods and services as opposed to payments based on how much land is farmed or historical production levels (the basis of current CAP payments). One implication of such a policy change is that there will be more scope for farmers to obtain payments for environmental practices and possibly fewer situations where they have to make a choice between environmental conservation and pecuniary benefits. However, this would require a significant change in farming practice, and the consequences of such changes on farmer behavior and welfare are unclear.
It is an open question, for instance, as to whether farmers will derive the same utility from producing environmental public goods and services as they do from traditional farm work, such as producing food. This is not to say that farmers do not care greatly about environmental issues. Indeed, scenario 1 suggests quite the opposite. Instead, we suggest that the success of new policy initiatives will not only depend on farm income but will also depend on the degree to which participation engenders losses in other utility-enhancing nonpecuniary benefits. For example, participating in a new subsidy scheme (such as payments for ecosystems services) might maximize farm profits but not necessarily be the preferred option for each individual farmer when it comes to maximizing their utility. A similar argument follows for the adoption of a new agricultural technology. There is a rich literature exploring the determinants of technology adoption, which has pointed out the importance of economic considerations such as usability and profitability (see Pannell et al. 2006). The responses we collected suggest that even if technologies are profit enhancing, this may not maximize the utility of some farmers if it results in the loss of other key nonpecuniary benefits.
We note that, as with other vignette-based studies, the present study has the limitation of not being able to measure actual choices in practice as well as presenting scenarios that are quite scant in terms of the information relative to the details that would be accessible in a real decision setting. However, insofar as the broad conclusions of this article match similar previous work that was replicated with high-stakes choices (Benjamin, Heffetz, Kimball, and Rees-Jones 2014), this is unlikely to be of particular concern. Using vignettes has the advantage of allowing us to address a much wider variety of relevant real-world choice scenarios than would otherwise be the case. We also acknowledge that, because of our recruitment method, we cannot generalize the results to all farmers, particularly to those who may not have been able to respond because of limited internet connectivity. To supplement the approach followed here, future studies may wish to adopt a reflective approach that asks farmers about their satisfaction with previous choices, though this is also problematic in that there are well-documented biases in terms of remembered utility, where experiences are evaluated predominantly by the peak and end of the memory (e.g., Redelmeier and Kahneman 1996).
5. Conclusion
Using tailored scenarios that address important trade-offs for farmers, our study shows that farmers choose in a way that broadly but not exclusively maximizes their well-being. This suggests that using subjective indicators of well-being as a proxy for decision utility may be a useful tool for studying farmer decision-making as well as conducting welfare evaluations more broadly. In cases where a disparity between choice and well-being preference was observed, choices favored greater pecuniary benefits (i.e., income or profits) at the expense of happiness. This means that using subjective well-being measures alone may understate the importance of income in farmer choice. The finding highlights the ongoing struggle between seeking out the lifestyle benefits that farmers seem to value highly in well-being terms and generating enough profit to thrive. Finally, our scenarios highlight the importance of nonpecuniary benefits for farmer utility, particularly those associated with the farming lifestyle. Farmers may decline an opportunity to increase farm income such as adopting a new technology or practice if it diminishes other sources of nonpecuniary utility. Failing to account for the presence of these nonpecuniary benefits can lead policy makers to draw incorrect conclusions when it comes to predicting farmer responses to policy changes.
Acknowledgments
This study was funded through the Global Food Security’s “Resilience of the UK Food System Programme” with support from BBSRC, ESRC, NERC, and the Scottish Government.
Footnotes
Supplementary materials are available through online subscription at: https://le.uwpress.org/.
↵1 This is because (42) = 6.
↵2 See “National Statistics: Farm Labour Profiles from the England and UK Farm Structure Survey,” Gov.uk, August 27, 2015. Available at https://www.gov.uk/government/sta tistics/farm-labour-profiles-from-the-england-and-uk-farm-structure-survey.
↵3 See “Statistical Data Set: Agriculture in the United Kingdom Data Sets,” Gov.uk, July 9, 2012. Available at https://www.gov.uk/government/statistical-data-sets/agriculture-in-the-united-kingdom.
↵4 We note that this analysis is exploratory in the sense that we had no prior expectations for the direction of specific subgroup differences.
References
This article requires a subscription to view the full text. If you have a subscription you may use the login form below to view the article. Access to this article can also be purchased.
Log in using your username and password
Purchase access
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.