The permanence of land management practices adopted under agri-environmental schemes (AESs) is often questioned. This paper investigates the drivers of farmers’ decisions as to whether to maintain “proenvironment” practices beyond the duration of a contract, and in particular the effect of social norms. Our results, based on the stated intentions of 395 French farmers, show that both pecuniary and nonpecuniary motivations drive farmers’ decisions, which are also significantly influenced by information about a social norm. Therefore “nudging” farmers, by conveying information to them on other farmers’ proenvironmental practices, appears to be a means of maintaining the long-run benefits of AESs. (JEL Q18, Q28)


Agri-environmental schemes (AESs) have been used in the European Union, United States, and Australia to address a wide range of environmental issues, from the conservation of biodiversity to water quality enhancement and landscape protection. These schemes are based on individual contracts signed with farmers who volunteer to implement proenvironmental management practices in return for an annual payment. This payment is calculated so as to compensate average compliance costs and foregone farming revenue due to the adoption of new management practices. Budgets dedicated to AESs are significant and are therefore under public scrutiny. Over the 2007–2013 financial period, total payments made by the European Union for AESs1 amounted to €22.7 billion, with an approximately equivalent amount of additional spending by member states.

All AES contracts have an end point, with contracts lasting from 5 years in French “territorialized agri-environmental measures,” to 10 years in the U.K. Higher Level Stewardship scheme, 15 years for some of the contracts of the U.S. Conservation Reserve Program, and 20 years in the now-defunct Environmentally Sensitive Areas scheme in the United Kingdom. At the end of the contract, farmers are free of any contracted commitment concerning their land management choices and can therefore revert to environmentally damaging practices even if this destroys the accumulated natural capital result-ing from participation (Hanley, Whitby, and Simpson 1999). This issue has been referred to as the “end of the contract problem” (Whitby 2000), and is an important criticism to be made of AESs and more generally of payments for environmental services schemes (Swart 2003), especially when budget constraints are tight and under public scrutiny. Policy-makers’ interest in investing in AESs would increase if the land management practices induced by the contract were permanently adopted. This end-of-contract problem is particularly problematic when new practices are less profitable than less environmentally beneficial alternatives.

However, motivations other than profit can also be expected to influence farmers’ choice to contribute to the provision of environmental services, even without monetary compensation. Indeed a growing literature demonstrates that information about one’s own behavior relative to that of others (an indicator of a “social norm”) can influence individual behaviors (Croson and Treich 2014). Thaler and Sunstein (2008) show that individual choices are shaped not only by information about what others in the same social group do, but also by the way this information is formulated and provided, the so-called framings of information. They introduce the concept of “nudge” as the use of a specific policy design, type of information, and framing of information that influences people’s decisions without changing the structure of economic incentives or restricting their available options. We use information on a social norm—specifically, the behavior of others in a reference group— as a behavioral nudge and investigate its effects on stated intentions. We also investigate the impacts of changing the framing of this information on the social norm.

A first objective of this work is therefore to investigate the drivers of farmers’ land management intentions at the end of AES contracts. Will farmers keep providing enhanced environmental services even in the absence of any payment, or does a shortterm contract necessarily lead to a short-term provision of conservation benefits? This paper reviews existing studies on this question and focuses more specifically on behavioral drivers that may induce a continuation of proenvironmental actions after the end of the contract, even when the new practices are less profitable to the farmer. The main focus of this paper is to investigate the effect of: (1) providing information about what other farmers do or intend to do, that is, giving them an indication of what the prevailing “social norm” might be, in terms of farmers’ willingness to maintain the land management practices they adopted under the AES after the contract ends; and (2) whether the framing of this information about the behavior of others matters to individual’s stated intention.

The behavioral motives underlying the decision to maintain proenvironmental practices beyond the duration of the contract and the effect of these nudges are tested through a national survey conducted in France in 2013. We sampled 395 French farmers engaged in agri-environmental contracts. Our results show that information about what other farmers intend to do can greatly influence a farmer’s stated decision whether to maintain the practices adopted during the AES after the contract ends and payments cease. However, changes in the framing of this information have no significant effect on a farmer’s stated intention.


Farmers engaging in AESs can provide environmental services in two ways: through land retirement or by modifying their resource use or farming practices, that is, by “land sparing” or “land sharing” (Lipper et al. 2009; Balmford, Green, and Phalan 2012). Land sparing options, such as wetland or grassland creation on farmland, require setting the farm plot aside from production. Such options usually create significant and long-lasting opportunity costs for participants in terms of the net value of production foregone. Other options, pertaining more to the land sharing approach, offer payments to farmers who agree to reduce the intensity of agricultural production, such as a limitation in stocking rates or a reduction of chemical pesticide or fertilizer use. Typically, these changes also come at a cost in terms of profits foregone (Armsworth et al. 2012) since they usually induce lower and more variable yields or may require higher management costs. Assuming that farmers make their decisions based on relative profits from alternative land management options, it is logical to expect that farmers will not maintain more costly practices without compensatory payments. This can be reinforced by a tendency to refuse to provide for free an environmental service for which they were paid during the contract period. As Engel, Pagiola, and Wunder (2008) argue, “there cannot be any expectation of permanence in the absence of payments,” as the logic of AESs (as well as payments for environmental services) turns public good supply into a marketable service.

However, some studies show that land management changes induced by AESs become permanent. Roberts and Lubowski (2007) show that more than 40% of farmlands engaged in the U.S. Conservation Reserve Program would not have been returned to crops if the program had ended in 1997. Although this striking conclusion might be due to the context of their analysis (especially in terms of crop prices), it points to the fact that some newly adopted practices might be maintained by farmers even without monetary compensation. Other evaluations (European Court of Auditors 2011) also find that there is only a partial reversal to previous management practices at the end of contracts.

There are several possible explanations for this observation. The first one is that farmers enrolling into AESs would have changed their practices even without any financial incentives for enrollment. In such case, AES contracts have provided windfall gains to farmers without environmental additionality (Chabé-Ferret and Subervie 2013). Therefore, farmers have no reason to change their practices after the end of the contract. But new practices can also be adopted permanently under AESs that have induced a true change. Since landowners base their choices on their beliefs about the relative pay-offs of alternative land management options they face, enrolling into an agri-environmental contract offers them the opportunity to test the true costs and constraints associated with the adoption of new practices. For example, in the case of land sharing options, the transition toward proenvironment practices may require short-term additional costs, such as investments in mechanical weeding equipment to replace chemical weeding, but may reveal itself less costly in the longer run than conventional farming methods. The AES payment supports farmers during this investment period, along with opportunities for them to acquire new skills and better knowledge of the risks. Assuming that these risks can be reduced with time and experience, and that the new practices are privately profitable after the fixed starting costs are overcome, the switch to low-input practices can become permanent with no additional incentive. The payments provided by agri-environmental schemes thus represent an opportunity to learn more about such pay-offs and to change initial beliefs, to break away from existing production “habits” and form new habits, potentially motivating the supply of environmental services even in the absence of AES payments (Hiedanpää and Bromley 2014).

Beyond financial motives, there are other drivers of changes. Motivations can also be nonpecuniary but selfish (this is the case when a participant is motivated by gaining a better reputation, a warm-glow feeling, or enjoying social acknowledgment that she contributes to the public good) or purely altruistic, when the participant genuinely seeks to contribute to the improvement of the public good (Glaeser 2014). As we discuss in the next section, social norms can “supercharge” these nonpecuniary motivations and thus increase the likelihood that farmers maintain proenvironmental practices despite the end of the financial incentives.


Farmers’ decisions whether to maintain proenvironmental practices after the end of an AES contract can be considered as a public good supply problem. Farmers who decide to maintain proenvironmental practices may bear private costs, whereas environmental improvements will benefit all members of the community. A large amount of research effort has been focused on understanding why people contribute to public goods when the main game theoretic prediction would be not to contribute. One interpretation is that a large proportion of people are conditional cooperators: people tend to contribute more when other people contribute too. In a seminal article, Fischbacher, Gä chter, and Fehr (2001) demonstrate, using the strategy method2 in public good games where players choose their contribution depending on others’ contributions, that about 50% of people are conditional cooperators. In other experiments, subjects are even willing to pay to get information about others’ contribution in a public good game in order to decide on their voluntary contribution (Kurzban and DeScioli 2008).

These experimental results have been confirmed in the field. Frey and Meier (2004) carried out an experiment at the University of Zurich where students were all asked to contribute to a charity fund but were given different information on other students’ contribution rates. This information had a significant effect: more students contributed when they had the information that 64% of the other students contributed than when they had the information that only 46% contributed. The choice to contribute or not was also significantly correlated with students’ expectations of others’ behavior. This approach has also been used to analyze the phenomenon of tax evasion. Paying taxes can be considered irrational if the probability of detection and the penalty if caught are analyzed. Tax evasion should therefore normally be much higher than what it is in most countries. Tax payers seem to be largely influenced in their tax morale by the perception that they have of the behavior of others and can therefore also be considered as conditional cooperators (Frey and Torgler 2007). There are a number of interpretations to explain conditional cooperation: people may value conforming to a social norm, have some preference for fairness such as reciprocity, or could consider that contributions of others are an indicator of the quality or importance of the public good (Frey and Meier 2004).

Social norms are traditionally considered to be divided into two categories: descriptive norms and injunctive norms (Cialdini, Reno, and Kallgren 1990). A descriptive norm describes behavior that is in some sense typical within a group. People tend to comply with descriptive norms because they reveal useful information about appropriate behavior in particular situations: “if others do that, it must be a good thing to do.” An injunctive norm refers to what constitutes morally approved and disapproved conduct, that is to say, what ought to be done. Adherence to injunctive norms is linked to other people’s ability to administer social punishment or reward (Thøgersen 2006). Bicchieri (2006) considers that people are influenced by their subjective beliefs about what others do and think, rather than by the actual behavior and opinions of others. These beliefs may change when new information is received. Providing social information about others’ behavior may therefore modify subjective estimation of norms and thus have a positive impact on the adoption of prosocial behavior.

In the context of a payment for environmental services scheme subsidizing farmers for reforestation, in China, Chen et al. (2009) show through a choice experiment survey that individual intentions to reenroll can be positively influenced by the information that neighbors also intend to reenroll. Farmers also stated that they would require lower subsidies to carry out environment protection activities if a large proportion of farmers reenrolled than if few farmers would do so (Chen et al. 2009). In a rather different context, Czajkowski, Hanley, and Nyborg (2015) find that adherence to a social norm codetermines the desire to engage in higher levels of home recycling for a large group of their sample of Polish households. The positive effect of social information on prosocial behavior has also been demonstrated in other contexts mainly in the social psychology literature: dictator games in the laboratory (Bicchieri and Xiao 2009), charity giving (Croson, Handy, and Shang 2009), littering (Cialdini, Reno, and Kallgren 1990), energy consumption (Schultz et al. 2007), and student alcohol consumption (Neighbors, Larimer, and Lewis 2004).

However, many examples from the literature also show that the framing of information can significantly influence individual choices. Framing effects have been studied in psychology, in medical and clinical decision-making, consumers’ choices, and bargaining behaviors (Levin, Schneider, and Gaeth 1998). Framing can be defined as “presenting individuals with logically equivalent options in semantically different ways” (Krishnamurthy, Carter, and Blair 2001, 383). One particular type of framing is of interest when a social norm is being presented to respondents, namely, attribute framing (Levin, Schnittjer, and Thee 1998). Attribute framing is a case of valence framing where one of the attributes of the choice is presented either positively or negatively. It is usually found that a positive attribute framing triggers a positive reaction. For example, experiments (Levin et al. 1988) show that respondents are more likely to wish for surgery if they are told that the technique used has a 50% success rate than if they are told that it has a 50% failure rate. The authors explain this effect by the way information is processed: positive framing creates positive associations in memory, which lead to a more favorable judgment of the event or object. In order to test this framing effect on farmers’ intentions, but also to avoid weakening the social norm effect of information, we tested the effect of a negative and a positive framing of information.


The survey was targeted at farmers eligible for the main French AES scheme called Mesures Agro-Environnementales territorialisées, or MAEt. The MAEt scheme was introduced in France under the second pillar of the Common Agricultural Policy for the 2007–2013 period, to target agri-environmental efforts on environmentally vulnerable areas, that is, the most sensitive areas for biodiversity conservation and water quality issues. Concerning water quality, the scheme is open to farmers located in the most contaminated drinking water catchment areas and/or in priority watersheds, where the risk of failing to achieve good ecological status for water bodies set by the European Water Framework Directive is the highest. Concerning biodiversity, the scheme is intended to attain the conservation objectives of the Natura 2000 network sites, defined by the European Union’s Habitat and Birds Directives. The MAEt scheme provides payments for a change in farmers’ practices or to maintain farming practices or activities that benefit the environment but are at risk of disappearing. In this scheme, farmers can adopt a wide range of land management options such as the reduction of input use (pesticides or fertilizers), the conversion of croplands to grasslands, or the restoration of hedgerows. They get a compensation payment that is calculated so as to cover the average additional costs and income foregone associated with the chosen land management options.

Survey and Treatments

We used an online survey3 to question farmers participating in the MAEt scheme about their land management intentions after the end of their contract. This survey was initially set up to evaluate the MAEt scheme over the 2007–2013 Common Agricultural Policy programming period. One section of the questionnaire focuses on land use and land management changes that farmers made when joining the MAEt scheme and on their intention to maintain these changes after the end of the contract, in the event that it is not renewed by the government. In order to test the effect of the social norm and framing effects, we constructed three treatment groups within which the question on whether farmers intended to maintain their land management practices was put differently (Table 1).



The software randomly selected respondents and assigned them to one of the three treatments. Respondents from groups 1 and 2 were both given the same information, which states the results obtained from a pilot survey4 that was implemented in the Languedoc-Roussillon region before the implementation of the national survey. However the framing of the information differed: it was positively framed for respondents from group 1 and negatively framed for respondents from group 2.

Considering the literature on conditional cooperation and on social norms, we expect that the information on rates of continuation of proenvironmental practices provided to groups 1 and 2 will have a positive impact on farmers’ intentions to also continue with their newly adopted practices after their contract ends. However, considering the framing literature, we expect that the magnitude of this positive impact will differ depending on whether this information is put in a positive or a negative framing. In our case, the impact of this information should be higher when highlighting the rate of farmers willing to continue, as in Treatment 1, than when highlighting the rate of farmers not willing to continue, as in Treatment 2.

Econometric Specification

As the respondents were randomly assigned to the groups, the treatment effects of information on the social norm and the framing of this information are causal and can directly be identified. In order to distinguish the two effects, we proceed in two steps. First, we introduce the dummy variable T, which takes the value 1 if the respondent received information on the social norm (groups 1 and 2), and 0 otherwise (control group). The effect of information on the probability that farmers decide to continue proenvironmental land management after the end of the contract (y = 1) is obtained through a maximum like-lihood estimation of the α parameter: Embedded Image [1] where F(⋅) is the cumulative distribution function of the logistic distribution.

Next, we distinguish two framing effects: T1 and T2. T1 is a dummy variable that takes the value 1 if the respondent received positively framed information on others’ behavior (group 1), and 0 otherwise (control group or group 2). T2 is a second dummy variable that takes the value 1 if the respondent received negatively framed information (group 2), and 0 otherwise (control group or group 1). We run the following econometric specification in order to identify the effect of framing: Embedded Image [2] where F(⋅) is again the cumulative distribution function of the logistic distribution.

Finally, so as to control for the effects of individual characteristics X on farmers’ decisions to maintain their newly adopted practices, we also introduce these characteristics as covariates in the regression: Embedded Image [3]

Vector X includes variables describing general farm characteristics: utilizable agricultural area (UAA) in hectares, the type of AES currently subscribed to, and type of farming activities. We assume that a higher UAA can increase the probability to maintain land sparing options and that the burden of continuing with better environmental practices without payment may differ across AES options and farming activities. Also included in vector X are variables aimed at signaling potential low additionality of farmer’s participation, that is, whether the respondent states that he already (almost) complied with the scheme’s requirement before joining, and to what extent he had to change his farming practices to comply with these requirements (low changes, medium changes, or major changes). Indeed, as discussed in Section II, alternative hypothesis can explain why farmers may continue to use their proenvironmental land management practices at the end of the program. First, we hypothesize that minor changes are easier and less costly to maintain than more major ones. However, if important investments have been made to comply with the AES option, it might be more difficult to revert to old practices. Second, we introduce proxy variables to capture different types of motivations for continuing MAEt practices after the end of the contract (Glaeser 2014): pecuniary, nonpecuniary selfish, and nonpecuniary altruistic motivations. Farmers who could earn a higher gross margin, who could sell their products at a higher price, and who faced no technical difficulties with the AES requirements might have pecuniary motives to maintain the adopted practices. Farmers who, during the AES, state that they experienced a better life quality (in terms of health, labor constraints, etc.) and/or explain that they gained an acknowledgment that their farming activity contributed to the protection of the environment and to high quality land management might have nonpecuniary selfish motives to do so. Finally, farmers who state that protecting the environment through their participation in the AES is a source of satisfaction by itself are likely to have purely altruistic motivations.


A total of 525 farmers participating in the MAEt scheme answered the national online survey, of which 83 stated that their joining the MAEt scheme had not changed their practices and 442 who, on the contrary, have adopted new practices. These 442 farmers were asked whether they intended to continue with these newly adopted practices when the payments ceased, and 395 answered the question. Hence, the answer rate for the question concerning the permanence of changes is almost 90%, with only 47 farmers choosing not to address this question.5 The sample used for analysis is therefore constituted of these 395 farmers randomly distributed among the three groups, with 128 respondents in the control group, 126 in group 1, and 141 in group 2.

As described in Table 2, the sample includes farmers engaged in AES options aiming at a reduction in fertilizer use (variable name AES fertilizers); a reduction in phytosanitary products use (AES phytosanitary); management of land cover, pastures, and moors (AES land cover), introducing for example constraints on mowing periods to favor biodiversity conservation; the creation or up-keep of grassland (AES grassland); the management of specific structural landscape features like hedgerows or ditches (AES linear); or finally AES options for conversion to organic farming (AES organic). Other minor options, concerning the management of specific environments (for example reed beds or salt marshes) or landscape are also represented in the sample and have been grouped together under the AES other variable. Farmers included in the sample have adopted 1.8 options on their farm, on average (standard deviation 1.02), 80% being engaged in one or two options. The most common farming activity in the sample is field crops (41.3% of the sampled farmers), followed by mixed farming (31.7%), and livestock farming (20.3%).


Descriptive Statistics of the Sample

There was a range of feedback from respondents on their experience with the MAEt scheme. Twenty percent of them declared that joining the AES enabled them to sell their agricultural products at a greater price, and 42% increased their total gross margin. But, almost half of them stated that they had experienced difficulties with the technical constraints imposed by the AES contract.6 On the other hand, a large majority of respondents (89%) stated that their participation in the scheme provided them with greater social acknowledgement of their contribution to the protection of natural resources and to local land management, and they almost unanimously (96%) stated that their participation provided them with the individual satisfaction of participating in the protection of the environment.7 Nearly 50% have experienced an improvement in their quality of life due to their participation in AES. Some 68% of the farmers of our sample acknowledged that they joined the AES partly because their practices were already in line with AES requirements. Nevertheless all of them stated that they have changed their land management practices after their enrollment in the AES. Forty-six percent of the interviewed farmers stated that they had to implement “low levels” of change in their practices to conform to the AES requirements, 39% have implemented “medium” changes and only 15% have implemented “major” changes. Remember that the 83 respondents who chose the fourth option (no changes) are excluded from our sample.

Some of this feedback varies depending on the type of option chosen by farmers. For example, farmers who had to reduce their phytosanitary products use state significantly more often they have implemented major changes and had technical difficulties than farmers not concerned with this type of option. Farmers implementing land cover options seem to perceive more acknowledgement for their environmental effort from society, which can be explained by the better visibility of these practices, for example, the management of pastures to prevent forest fires. We also find, not surprisingly, that those who chose to enter into an AES for the reduction of pesticide use experience better life quality than farmers who did not adopt this option type. In the following we will thus control for the effect of these differences.

Table 3 shows that, overall, random assignment among treatment groups has created three groups with similar characteristics for most of the variables we control for. However, we observe a few differences that we have to account for during the analysis. Farmers who adopted options for structural landscape features management (AES linear) are overrepresented in the group who received information (T = 1), and especially in the group who received positively framed information (T1 = 1), while those who adopted organic options are underrepresented in the group with information. Farming activities as well as farmers under phytosanitary constraints (AES phytosanitary) are also unevenly distributed between the two framing groups (T1 = 1 and T2 = 1). Finally, fewer farmers have altruistic motivations in the group that received negatively framed information.


Balancing Tests and Mean Value of the Dependent Variable in Each Treatment Group


To the question “Would you continue your newly adopted practices after the contract ends?” (see Table 1), farmers could choose one of the four responses: “absolutely,” “probably yes,” “probably no,” or “not at all.” Figure 1 shows the percentage of answers in the three informational treatments. We also observe, as shown in Figure 1, an increase in the percentage of respondents stating “probably yes” or “absolutely” between control group (no information), group 2 (negatively framed information), and group 1 (positively framed information). The second part of this section will therefore focus on measuring the effect of the treatments, in particular, in testing the significance of the difference observed in Figure 1. In the following analysis, we pool responses to work with a binary variable: y =1 if the answer is “absolutely” and “probably yes,” y = 0 if the answer is “not at all” and “probably no.”


Percentage of Farmers Intending to Maintain the Proenvironment Practices after the Contract End, According to the Three Treatments


On average, 55% of farmers (219 of the 395 who answered this question) were willing to maintain the practices adopted during the AES after the end of the contract. This percentage remains high, 43%, when we consider the control group only, excluding the influence of the treatments. Table 4 presents the results of the logit models. Since the marginal effects of each variable cannot be directly observed from the coefficients of the logit models, we present the odds ratios. The odds ratio indicates the effect of an increase of one unit of the considered independent variable on the odds that farmers intend to continue the AES land management practices rather than abandon them. Therefore an odds ratio lower than 1 indicates a negative effect of the variable on the dependent variable. Logit 1 and logit 2 present the results on the effects of the information on the social norm and on the framing of this information. Results will be discussed in the next subsection. To analyze how farmers’ characteristics (X) impact their intention to continue the AES land management practices, we now consider the Logit 3 model in Table 4.


Model Results (Odds Ratios)

As expected, the likelihood of continued implementation of agri-environmental practices postcontract decreases if farmers have experienced technical difficulties during implementation. The odds of continuing the new practices are more than 50% lower in that case. Conversely, if the new agri-environmental practices have generated a better sale value for production, the odds of continuing these are more than doubled (but this effect is significant at only 10%).

Farmers who experienced acknowledgment for their contribution to the protection of the environment or a better life quality are more likely to maintain the adopted practices even in the absence of payment, which indicates that they might have nonpecuniary selfish motivations to do so. Farmers who experienced acknowledgment may value external positive judgments and might fear social disapproval if they return to their less environmentally friendly practices. On the other hand, farmers who did not experience acknowledgment may feel fewer qualms about reverting to their old practices.

No significant effect of altruistic motivation was detected (captured by “Contribute to the environment”). Indeed, over 384 responses, 369 respondents (96%) state that they enrolled in the AES to participate in the preservation of the environment. Altruism is one of the drivers of participation for almost all respondents. This lack of variation across farmers in our data prevents us from giving any conclusion about the effect of this motivation on farmers’ choice to maintain the adopted practices at the contract end (Table 2). Not surprisingly, farmers are more likely to continue agri-environmental practices if they implemented small rather than major changes to conform to the AES requirements or if they already conformed before joining the AES (Table 4, Logit 3). This result confirms the intuition that a long-term upkeep of the practices is linked to a low additionality of the scheme.

Finally, and more surprisingly, farmers who participate in an AES phytosanitary option (aiming at a lower use of pesticides) display a greater propensity to maintain the adopted practices, while options of grassland management or reduction of fertilizer use decreases it. This is rather counterintuitive since the reduced use of pesticide may result in greater yield variability. However, it can be explained by the fact that farmers have to invest in greater knowledge of pest and weed management techniques in order to comply with the AES requirements. Once such investment has been made, it might be less profitable to revert to previously used techniques.

Effect of Social Norm and Framing

The results also show that being provided with the information that a majority of farmers would not revert to their old (detrimental) practices is sufficient to trigger a higher proportion of positive responses concerning future commitment to maintain agri-environmental practices. Indeed, α is positive and significant (Table 4, Logit 1), and the odds ratios show that the odds that farmers maintain the adopted practices is more than two times higher (2.1 in Logit 1) when information about the social norm (T = 1) is given than without such information. This effect is even stronger, with an odds ratio of 2.8, when controlling for the observable characteristics of the respondents in Logit 3 (Table 4), which were slightly unbalanced between treatment groups (Table 3). This effect is also directly observed in the proportion of farmers who state that they would maintain the AES practices after the contract ends: 61% of farmers who received information, compared with only 43% in the control group (Table 3). The treatment variable T also stays highly significant when we run logit regressions by type of option (AES fertilizers, AES phytosanitary, AES land cover, and AES grassland).

However, a test of equality of parameters for variables T1 and T2 in Logit 2 (Table 4) reveals that there is no significant difference between the two estimates of the parameters β1 and β2 as defined in equation [2], which means that the way information is framed, positively or negatively, has no significant effect here. This is contradictory with the literature, where an attribute framing effect is considered “a reliable phenomenon” (Levin et al. 2002, 413). Note that in our survey the information about the social norm is quite strong, since the rate of farmers stating that they would maintain their practices at the end of the contract was 80% in our pilot survey. This may lessen the impact of the negatively framed information (that only 20% do not continue with the newly adopted practices).

In an attempt to identify if some of the characteristics included in X might influence positively or negatively the susceptibility of farmers to social norms, interaction variables T×X have been included in Logit 4. We can in particular expect farmers with nonpecuniary selfish motivations, in our case those who experienced social acknowledgement, or farmers with purely altruistic motivation to be influenced by the information that other farmers intend to maintain the adopted practices. Indeed, if farmers state that social judgement of their contribution to the environment is important for them, they might equally be sensitive to their peers’ judgement and have a strong preference for conformity to social norms. Purely altruistic farmers might also be more likely to maintain the adopted practices if they know that others do so, as it increases the chances of their own actions having an impact on the environment. However, we could not detect any significant effect of these two interaction variables, Acknowledgment×T and Contribute to the environment× T, suggesting that farmers’ sensitivity to the social norm is not dictated by these motivations.


The first result of this paper is that the end-of-the-contract problem in AESs might not be as problematic as previously thought. Indeed, 43% of the surveyed farmers intend to maintain the practices they adopted under the AES requirements, even in the absence of financial incentives or knowledge of others’ intentions. This result conforms to those obtained by Roberts and Lubowski (2007) and the study by the European Court of Auditors (2011). We show that pecuniary and nonpecuniary selfish motivations, like social acknowledgement or a better life quality, can partly explain this intention. However, we also show that low levels of land management change are more likely to be permanent than major changes. Therefore, the long-lasting transition toward more environmentally friendly practices in agriculture can be expected to be slow and incremental. This means that the decision to renew contracts should be partly based on the environmental additionality of schemes: schemes that produce bigger changes in farm practices are more likely to suffer a reversion to precontract management than those that produce smaller changes. They are more susceptible to suffer from a relatively higher rate of loss of environmental capital once contracts expire.

More interestingly, we find that farmers participating in the French MAEt scheme are conditional cooperators. Hence, providing information on what others intend to do, as an indicator of a social norm, can be a powerful nudge to increase the permanence of proenvironment practices. As such, this paper adds to a series of results that are increasingly inspiring public economists for more ambitious policies targeted at farmers (DEFRA 2008; World Bank 2015). Much attention has been granted to the design of incentive policies such as taxes or subsidies to reduce polluting activities from agricultural activities. The recent economic crisis in Europe, which makes green taxes more politically sensitive and reduces the margin of maneuvers for public spending, has given momentum to new kinds of policies relying more on suasion and psychology than on monetary incentives.

Of course, this can raise ethical issues, extensively discussed by Thaler and Sunstein (2008), since there is a risk for public authorities to “manipulate” citizens’ choices in a “paternalistic” manner that does not coincide with what their choices under free will would be. While much of the social nudge literature is focused on changing behavior toward a mode that is deemed better for the subject’s own well-being by those implementing the policy (e.g., encouraging children to eat more healthily), our use of the term involves a different context. We use information on a social norm to nudge farmers into continuing with proenvironment practices at the end of an agri-environment contract. Note that such behavior is entirely voluntary for the farmer, as indeed was their uptake of the contract in the first place. We also assume that the continuation of proenvironmental benefits yields a social benefit in terms of enhanced water quality and human health. However, farmers are not being coerced into stating that they would continue with these practices: it is an entirely free choice made by them, which they presumably arrive at by comparing the utility to themselves of alternative actions. Moreover, it can be underlined that this utility can include a pay-off from bringing one’s own behavior closer to the social norm, as Czajkowski, Hanley, and Nyborg (2015) show.

One potential limit of this paper is that it relies on stated intentions rather than actual behavior to study the social norms effect. For strategic reasons, farmers might over- or understate their intention to maintain the adopted practices and more (or fewer) farmers than were found through the survey will actually maintain them. However, there is no reason to think that the treatment effects of giving information on others’ intentions influence this strategic behavior, nor that strategic behavior will vary systematically across treatments. As the treatment is randomly assigned across participants, we can then expect that the impact we capture by comparing the relative levels of permanence between the treatment and control groups reflects its likely actual impact on farmers’ decisions to maintain proenvironment land management practices after AES contracts end.

To conclude, this paper contributes to the literature showing that in general, people have preferences for following social rules and may suffer disutility when violating social norms. Farmers are no different: their individual behavior is likely to be influenced by the behavior of others. This should be kept in mind when designing an AES. As shown in this paper, informing a farmer on the choices made by her peers can induce her to conform. Communicating the average adoption rate of an agri-environmental contract—through articles in agriculture magazines or information via farmers’ organizations—could thus help to persuade more farmers to enroll, if this average adoption rate were high enough. Proposing contracts that include a specific reward for a collective success can help to signal the social norm to farmers. For example, Kuhfuss et al. (2015) show with a choice experiment survey that a monetary bonus paid to all contractors if the adoption rate is above a given target can improve farmers’ participation and increase land enrollment for lower overall budgetary costs. Indeed, this study shows that wine growers in the south of France do value this conditional bonus much more than its expected monetary value. In addition, they suggest that the introduction of this conditional bonus contributes to increased expectations of farmers on others’ participation, therefore shifting a proenvironmental social norm and favoring the adoption of less pesticide intensive farming practices. In such case, combining a financial incentive with a behavioral nudge can increase the efficiency of public policy with no added costs.


The authors would like to thank the French Ministry of Agriculture (Bureau des Actions Territoriales et Agro-environnementales) for its collaboration on the implementation of the survey. This survey was funded by the ONEMA in the framework of the 2011 call for research projects “Changer les pratiques agricoles pour préserver les services écosystémiques,” supporting the implementation of the French National Action plan Ecophyto 2018.


  • The authors are, respectively, postdoctoral research fellow, Department of Geography and Sustainable Development, University of St. Andrews, St. Andrews, Scotland, United Kingdom; researcher, INRA, UMR, Montpellier, France; professor, Montpellier Supagro, UMR, Montpellier, France; professor, Department of Geography and Sustainable Development, University of St. Andrews, St. Andrews, Scot-land, United Kingdom; Ph.D. student, University of Montpellier, UMR, Montpellier, France; lecturer, Montpellier Supagro, UMR, Montpellier, France.

  • 1 Financial plan of the European Agricultural Fund for Rural Development (EARDF) axis 2 measure 214 (agri-environment).

  • 2 The strategy method, in which a responder makes conditional decisions for each possible information set, is usually opposed to the more standard direct-response method.

  • 3 Using the software Limesurvey®.

  • 4 Based on the responses of 91 farmers participating to the MAEt scheme.

  • 5 There are no significant differences in the answer rates of the three groups.

  • 6 Some examples reported by farmers in an open-ended question of the survey include the following: difficulties in managing weeds without herbicides during rainy years, difficulties in respecting the objective of input reduction each year of the 5-year commitment, timing issues for AES options, which include constraints on the period of interventions (mowing date for example) that can be incompatible with weather conditions or workforce availability. Previous surveys with farmers participating in the MAEt scheme in France have shown that field slopes or narrow rows in vineyards and orchards can hinder mechanical interventions, which are usually substitutes for the use of phytosanitary products. All these “technical difficulties,” which vary from one farmer to another according to his farming skills, equipment, and local constraints, do have an impact on his decision to maintain proenvironment practices without contract payments.

  • 7 In open-ended questions of our survey, many farmers highlight their concern for the environment (altruistic motivations). They are often upset that their efforts for preserving landscape, biodiversity, and the environment in general are not sufficiently acknowledged by policy makers and society in general. Therefore, many respondents claim that they enrolled in an AES to demonstrate and make more visible their environmental contributions (and to gain greater social acknowledgment of their efforts).