Environmental policy integration in the EU’s common agricultural policy: greening or greenwashing?

ABSTRACT The increasing multidimensionality of agriculture, linking the domain with environmental, trade and food safety concerns, has mobilized new policy actors bringing new preferences and ideas into the Common Agricultural Policy (CAP) debate. This article investigates the extent to which this has contributed to Environmental Policy Integration (EPI) in the CAP. It puts forward the claim that an incomplete transformation in European agricultural policy from exceptionalism to post-exceptionalism explains the limited extent of EPI in the CAP. This claim is substantiated by a longitudinal comparative analysis of the CAP reforms over the last two decades, applying a multidimensional concept of EPI as process (how the formal and informal procedures and institutions in place allow for the integration of environmental concerns in policy deliberation), output (the translation of such concerns in changes in policies) and outcome (the performance of the new policies in terms of environmental benefits).


Introduction
The European Union's (EU) Common Agricultural Policy (CAP) has undergone significant changes during the nearly 60 years of its existence. Although the sector is characterized by a relatively insulated policy networkbringing together agricultural policy experts, ministers and interest groupsthe increasing multidimensionality of agriculture has resulted in new concerns entering the debate, ranging from development and trade to environment and food safety. This contribution will focus on the role of environmental concerns in processes of CAP reform, addressing the question of in what form and to what extent Environmental Policy Integration (EPI) has taken place in the CAP since the 1990s with respect to (a) the decision-making process; (b) the policy output in terms of both substance and underlying policy ideas; and (c) the policy outcome in terms of effectiveness.
Existing research on EPI in the agricultural sector often concludes that environmental discourse or rhetoric has certainly increased, but that it seems to be difficult to move from political commitment to genuine EPI (Buller 2002), due to low priority of environmental issues and a closed agricultural policy network (Lowe and Baldock 2000: 31-33). The dominance of economic and producer interests resulting from these institutional power relations, together with budgetary and trade concerns, are emphasized in rationalist International Political Economy accounts of CAP reform, explaining the watering down of environmental concerns on the basis of material factors (Ackrill 2000;Swinnen 2015). This contribution, instead, applies a more discursive take on the CAP  emphasizing the role of policy ideas in constraining and enabling EPI. The argument I will develop in this contribution is that the incomplete transformation from exceptionalism to post-exceptionalism in European agriculture has had a limiting effect on EPI in the CAP (Daugbjerg and Feindt 2017). While the increasing multidimensionality of agriculture and food policy have instigated a shift in the four dimensions of exceptionalism and post-exceptionalism as applied to the CAPpolicy ideas, institutions, interests and policy instrumentsthis shift has not resulted in a full or stable 'reframing' (Persson 2007) of agricultural policy ideas in more environmentally oriented terms. The integration of environmental goals in agricultural policy and the policies' effectiveness remain limited as a result. This paper will build on and contribute to existing research on EPI in general and EPI in the CAP in two ways. First, it will introduce a multidimensional conceptualization and measurement of EPI that includes the analysis of policy ideas and discourse. This conceptualization enables a more nuanced analysis of EPI. The importance of ideas cannot be overstated, for changes in policy process and instruments are unlikely to endure without changes in the underlying policy ideas (Nilsson and Persson 2003). Existing analyses of EPI in the CAPwith the notable exceptions of Feindt (2010) and Lynggaard (2007) tend to underestimate how policy changes are shaped by transformations in ideas. Secondly, this contribution will present a longitudinal and upto-date multi-dimensional analysis covering the EU-level CAP reforms since the 1990s. In doing so it seeks to remedy the shortcoming in the literature observed by Nilsson and Persson (2003) and Feindt (2010) that studies in EPI are often constrained to single-country studies focusing on one or two dimensions of EPI only.
In the remainder of this contribution I will first engage in the debate on the concept and dimensions of EPI and introduce the conceptualization applied here. In the second section, I will explain how exceptionalism and post-exceptionalism are expected to affect the different dimensions of EPI. After explaining the methods that will be applied in the empirical research in section three, the fourth section will present the outcome of the empirical analysis. Finally, in the conclusion I will reflect on the relation between the ongoing transition to post-exceptionalism in European agriculture and the shape and degree of EPI in the CAP, and discuss the limits of this study as well as avenues for further inquiry.

Conceptual distinctions
In the environmental policy studies literature there is an ongoing debate on how EPI should be conceptualized and measured. In essence, EPI is about 'the incorporation of environmental concerns in sectoral policies outside the traditional environmental policy domain' (Runhaar et al. 2014: 233). The underlying rationale is that sustainable development can only be achieved if environmental perspectives 'become a natural part of the goals, strategies and decision-making procedures of all major parts of public policy' (Nilsson et al. 2009: 228). A number of conceptual distinctions are made with respect to EPI which are relevant in this research: (a) horizontal versus vertical; (b) weak versus strong; and (c) process versus substance.
Horizontal EPI focuses on environmental co-ordination across policy sectors, while vertical EPI is concerned with environmental co-ordination in one particular sector without cross-sectoral modes of interaction (Lafferty and Hovden 2003: 12-14). Although vertical integration is in part instigated by horizontal requirements (Feindt 2010), this paper will limit itself to vertical EPI within the agricultural sector.
The distinction between weak and strong EPI revolves around the degree to which EPI requires a revision in the traditional hierarchy of policy objectives. Weak EPI merely requires that environmental considerations are taken into account in sectoral policy-making. Strong EPI additionally requires that environmental considerations are given principled priority over other considerations (Lafferty and Hovden 2003: 9). This paper takes a pragmatic point of departure: that for EPI to have occurred some degree of revision in priorities needs to have taken place, increasing the role of environmental considerations, but not necessarily giving them principled priority.
Finally, EPI can simultaneously be considered a process 'leading to changes in policy-making and policy outputs' and a 'substantive result of changes in policy decision-making, behaviour or ideas' (Kivimaa and Mickwitz 2006: 731, emphasis added). Those interested in the policy process focus on how environmental considerations are institutionalized into the policy process, while those focusing on the substantive result of that process investigate whether it led to more environmentally friendly policies (output) and environmental performance (outcome) (Persson 2007: 30-1). A sole focus on the process dimension would uncover the institutional strategies applied to enhance EPI but neglect the connection with the ends this process aims to achieve. This paper will therefore take both process and substance into account, differentiating between a process, output and outcome dimension of EPI.

The CAP and dimensions of EPI
This section will present indicators for the three dimensions of EPI, inspired by the existing literature (Lafferty and Hovden 2003;Nilsson and Persson 2003;Nilsson et al. 2009), and adapted to the case where appropriate. Secondly, it will introduce ideational and discursive elements and connect these analytically to the process and output dimensions of EPI.
The measurement of process EPI focuses both on actors' structural power position (based in institutions) and their effect on policy formulation. 1 For the former, it analyses the procedures that are put in place to (a) co-ordinate decision-making; (b) enhance the comprehensiveness and inclusiveness of actor representation; and (c) institutionalize evaluation of and reporting on environmental impact. For the latter, it investigates whether these procedures contributed to the inclusion and relative weight attached to environmental concerns in the policy-formulation process. Although the first group of indicators may be considered institutional in nature (see Nilsson et al. 2009: 340), they are studied in and affect the policy-formulation process and are therefore included in the process dimension here.
The output dimension focuses on the policies introduced or adapted through the different CAP reforms. I distinguish between changes in policy settings, instruments and objectives, evaluating whether these have resulted in (potentially) greener policies. Whether these policy changes signify successful EPI, however, still depends on their environmental effects. This is the focus of the analysis of EPI as outcome where first of all, the implementation of the policies will be analyzed, checking whether the policies are applied with the actual environmental aim of the policy in mind or if they are watered down during implementation. Secondly, the environmental performance of the policies is analyzed: do the policies have positive environmental effects?
The significance of a change in underlying policy ideas to achieve EPI is highlighted by the less commonly used 'reframing approach', emphasizing that EPI requires the reframing of 'fundamental problem perceptions, causal narratives and overall policy goals into more environmental terms' (Persson 2007: 43). 2 This resonates with the discursive approach taken in this contribution, allowing for the incorporation of ideational and discursive dimensions in the analysis . In the field of agricultural policy studies, the role of ideational factors is usually the subject of the debate on rival policy paradigms and their supporting discourses. Policy paradigms contain ideas with respect to the understanding of the policy problem, appropriate policy goals, and the proper instruments to achieve them (Hall 1993). Policy actors apply a certain justificatory discourse in the policy-making process legitimating the policy objectives and instruments associated with the policy paradigm they prefer to be 'institutionalized' in the policy output Skogstad 1998; see also Weber and Driessen 2010). In this paper the ideational aspects of EPI are therefore conceptualized in the dimensions of EPI as output and process (see Table 1).
First, the state-assisted agriculture paradigm has a productivist focus, emphasizing that the farm sector contributes to a secure and safe food supply. Due to unstable natural conditions and the sub-optimality of the price mechanism to secure income stability and productivity, adherents to this paradigm argue that the farm sector is exceptional and warrants special treatment (i.e., public intervention) (Daugbjerg 2003;Skogstad 1998). Secondly, the multifunctionality paradigm emphasizes the environmental and social services of farming. Farmers are not sufficiently rewarded by the market for these public goods and should therefore legitimately receive public money (Coleman 1998, Daugbjerg 2003. Finally, the competitive paradigm argues that agriculture is an economic sector like any other in which the farmer should be treated as an entrepreneur and market forces should take precedence over state intervention (Coleman 1998;Skogstad 1998). The ideas contained in the multifunctionality paradigm resonate with the ideas underlying EPI, attaching importance to environmental concerns, whereas the other paradigms tend to prioritize economic concerns. A shift towards the multifunctionality paradigm should therefore be conducive to EPI (see also Feindt 2010: 296).

EPI and (post-)exceptionalism
Agricultural exceptionalism is about treating the agricultural sector differently from other economic sectors and includes a 'belief system that provides cognitive justification and political legitimation' for this special treatment (Daugbjerg and Feindt 2017). The idea of agricultural exceptionalism was part of the stateassisted policy paradigm both in European and US agriculture after the Second World War (Skogstad 1998) and became the dominant ideational framework . It became challenged since the 1980s, however, due to a mix of budgetary, trade-related and environmental considerations, and a transformation to a post-exceptionalist agriculture set in. 3 In this paper I take exceptionalism and post-exceptionalism as ideal-types, each characterized by a specific combination of ideas, institutions, interests and policies (see Daugbjerg and Feindt 2017). I hypothesize that each of these two specific combinations of variables has different effects on the likely forms and degrees of EPI in agriculture, affecting EPI as process, output and outcome (see Table 1).
During the process of policy formation the institutional and ideational aspects of exceptionalism and post-exceptionalism are likely to affect EPI. Institutionally, post-exceptionalism is likely to be more conducive to EPI than exceptionalism, because it is characterized by a more open policy network. Ideationally, agricultural exceptionalism fits with the state-assisted policy paradigm, while post-exceptionalism resonates with the multifunctionality paradigm (Daugbjerg and Feindt 2017). In the case of exceptionalist agriculture, I therefore expect a productivist discourse and a representation of interests in which farmers' economic interests are pitted against environmental interests while the former carry the day. In the case of post-exceptionalist agriculture, I expect the legitimating discourse also to be based on environmental considerations and a positive-sum (Persson 2007: 27) representation of economic and environmental interests.
The policy aspect of exceptionalism and post-exceptionalismconnected to the state-assisted and multifunctionality paradigms respectivelylinks up with the output dimension of EPI. Exceptionalism will result in policies aimed at production and farm-income related objectives. If environmental side effects occur, these are unintended (Lenschow 1999). To the extent that policies are considered (partly) 'environmental' in naturethey are likely to be voluntary. Post-exceptionalism is likely to result in the introduction of relatively more demanding and mandatory environmental policy objectives and instruments.
With respect to the outcome dimension of EPI, more positive environmental results are likely in a post-exceptionalist agriculture than in an exceptionalist agriculture, as the former will have more (partially) environmentally instigated policies. Environmental policies may be watered down during the implementation process, however, particularly when an environmental discourse was applied strategically during the policy-making process. I consider the use of environmental ideas strategic when actors apply them instrumentally in a discourse 'manipulating public images and ideas in order to build support for their policies' (Legro 2000: 423), not reflecting genuine motivation for action based on the internalization of environmental ideas. When environmental arguments are strategic justifications rather than genuine motivations for action, then environmental aspects of the CAP reform are more likely to be sacrificed in the negotiating compromise allowing for flexibility, and hence further watering-down, in the implementation process. Neither exceptionalism nor post-exceptionalism are thus likely to result in a high degree of outcome EPI as long as underlying policy ideas have not genuinely become more environmentally oriented.

Methods and sources
This contribution applies a qualitative research design. The sources used to conduct the empirical analysis include: (a) all official Commission proposals and communications to the Council and European Parliament (EP) with respect to the different CAP reforms as well as the relevant impact assessments; (b) 10 speeches for each Commissioner of Agriculture and Rural Development; (c) opinion papers of agricultural and environmental organizations; (d) four interviews; and (e) secondary literature. The policy documents enable measuring the policy output of the different reforms and the legitimating discourse applied to justify these reforms in the policy process. The speeches are selected on the basis of the variation in the audiences they address (EP, Council, interest groups) as well as their timing (public consultation/preparatory phase, decision-making, communicating the policy outcome). This enables verifying whether environmental discourse is strategically applied to certain audiences or over time, during policy preparation and communication of the results rather than during the actual decision-making. Together with the official EU documents, opinion papers and secondary literature analyzing previous CAP-reform processes, interviews provide additional insights in the policy-making process and the degree to which the mobilization of environmental actors was effective. The interviews were conducted with officials of COPA-COGECA (as representative of farm interests), Birdlife (as representative of environmental organizations) and representatives of two different DGs. Both of the latter had been involved in CAP reforms and related environmental measures since the 1990s, covering the whole of the period analyzed in this paper. With respect to EPI as outcome, it should be noted that the scope of this contribution and limited access to primary data do not allow for an independent assessment. The analysis will be based on existing evaluations and secondary sources. Many of these sources contain 'discourse on' or 'interpretation of' policy effectiveness rather than presenting verifiable primary data. Although this impedes drawing strong and objective conclusions, source triangulation was appliedusing not only Commission evaluations, but also European Court of Auditors (ECOA) reports and publications of Environmental Research institutesin an attempt to arrive at as reliable an assessment as possible.
A qualitative content analysis program (NVIVO) was used to code the policy documents and speeches on problem definitions and appropriate policy objectives and instruments (indicators of underlying policy paradigms). This enabled a structured analysis of the discourse in the process dimension of EPI.
5. The CAP 1993-2013: environment as genuine objective or justification strategy?

EPI as output
Over the last decades, the CAP has witnessed a shift from emphasis on price and market intervention to direct income paymentspartly conditional on environmental requirementsand rural development. A number of the policy changes introduced by the five CAP reforms in this periodthe 1992 MacSharry, 1999Agenda 2000Fischler, 2008 Health Check and 2013 CAP reformsmay be considered 'environmental' in nature in the sense that they either have environmental objectives, or could have positive environmental side effects (for more elaborate analyses see Buller 2002;Feindt 2010;Lowe and Baldock 2000;Mathews 2013). The first policy is the set-aside introduced in the 1992 MacSharry reform. In order to be eligible for the direct income payments, (larger) farms had to take land out of production. Combined with the 29 per cent reduction in intervention prices, this measure was expected to make farming more extensive. Secondly, the 1992 reform also included agri-environment measures as an accompanying measure. 4 These obliged member-states to develop schemes providing payments to farmers for voluntarily implementing specific environmentally friendly farming practices. Thirdly, the 1999 reform introduced voluntary cross compliance, which, if introduced by member-states, made the direct income payments conditional on meeting a number of basic environmental and animal husbandry conditions. The same reform also merged agri-environment measures together with other (non-environmental) measures in the socalled 'pillar II', introducing a Rural Development policy next to the existing market and price policy of 'pillar I'. The 1999 reform also introduced the instrument of modulation, which allowed member-states to reduce the direct payments in pillar I by a small percentage and shift these funds to pillar II. The subsequent 2003 reform made cross compliance and modulation compulsory, while the CAP Health Check in 2008 raised the modulation percentage, expanding the amount of money shifting towards rural development. The latter only increases potential spending on agri-environment measures, however, as pillar II also contains many non-environmental policies. Finally, the 2013 reform introduced the greening criteria, which made the payment of 30 per cent of the direct income payments dependent on compliance with three environmental requirements (maintaining permanent grassland, 5 5 per cent environmental focus area and crop diversification). The same reform also introduced a change affecting the settings of the policy instruments in pillars I and II by enabling (but not requiring) member-states to decrease direct income payments (pillar I) above a certain level (degressivity) and inject the 'capped' funds into pillar II.
These measures could potentially have positive environmental effects, but these were not necessarily their main objective (Feindt 2010). Commission documents show that in 1992, set-aside was mainly instigated by economic considerations, such as over-production, farm income and budgetary constraints (Lenschow 1999: 101;MacSharry 1991aMacSharry , 1991band Buller 2002 reach a similar conclusion). That environmental objectives were never a key objective (Lowe and Baldock 2000: 42) is evidenced by the fact thatwhen set-aside was no longer necessary or even undesirable in the light of changing market conditions and increased anxieties for food securitycompulsory setaside was first set at zero in the 2008 reform (with Fischer-Boel [2007] arguing that 'set aside is a support management tool, not an environmental tool') and phased out in 2013. Contrary to set-aside, the agri-environment measures in the 1992 reform were genuinely focused on alleviating environmental pressures, marking a start of more integrated policies (Buller 2002;Mathews 2013). They were, however, only an 'accompanying' measure, signifying merely a small part of total CAP spending. The cross compliance and greening criteria served environmental objectivesranging from preventing soil erosion and water pollution to securing a minimum level of maintenance of the land and biodiversity. They introduced only limited additional environmental objectives, however, as part of the cross compliance criteria linked existing legislation to direct payments. The slow development in policies serving environmental purposes in the CAP (but their increasing compulsoriness) is in accordance with the fact that in 1992 the transformation from exceptionalism to post-exceptionalism was only in its early stages, while it had progressed but was still not complete by 2013. Although an increasing number of policy objectives and instruments in accordance with both the multifunctionality (e.g., agri-environment measures and cross compliance) and competitive paradigm (e.g., guarantee price reductions) became institutionalized in the CAP over time, this never fully eclipsed the existing institutionalization of the dependent agriculture paradigm (e.g., income support measures) that had resulted from the longstanding discursive hegemony of a productivist discourse rooted in agricultural exceptionalism (cf. Feindt 2017).

Procedures and institutions
Environmental considerations are brought into the CAP debate by environmentally oriented services and staff within DG AGRI, interlocutors of DG ENVI, environmental groups and by the EP's COMENVI. The formal institutional rules on co-ordination are that with respect to Commission proposals all DGs are consulted. Apart from that many informal contacts between the DGs exist. It is particularly through these informal contacts and consultations that DG ENVI influences the CAP (interviews with Commission officials, 1 April 2016; 28 April 2016).
Turning to access and influence, both farm interest groups and environmental groups have formal access through the Advisory Groups (Civil Dialogue Groups since 2014) of DG AGRI. The number and composition of these groups have changed over the years, improving the representation of environmental interests (Feindt 2010; interview with Birdlife official 19 May 2016). 6 Nevertheless, agricultural producers and traders still make up the vast majority of representatives in Civil Dialogue Groups and farm interest representatives have extensive informal access to decision-makers (interview with COPA-COGECA official 10 May 2016).
In the decision-making process on CAP reform, the Council of Agriculture Ministers plays a dominant role, and agriculture ministers tend to be more open to lobbying efforts of farm interest groups than those of environmental groups (Greer and Hind 2012). This is therefore a likely venue for the wateringdown of environmental elements in reform proposals. Originally, the EP could only give its opinion on CAP reforms, but since the Lisbon Treaty, it acquired co-decision. The post-2013 reform is therefore the only reform until now in which we can see how this enhanced role of the EP played out (see Greer 2017).
Finally, when it comes to evaluation and reporting on environmental objectives a procedure of ex ante impact assessments and ex post evaluations is institutionalized. The degree to which environmental objectives are part of the impact assessments and evaluations compared to other criteria evolved over time. In the impact assessment for the 1999 and 2003 CAP reforms, the emphasis was on the expected consequences of the proposed policy for farm income and market stability mainly, while only a few pages of the extensive reports were devoted to the environment and rural development (Commission 1998(Commission , 2003. This conclusion is supported by existing research on EU ex ante impact assessments, indicating the narrow focus of such assessments and bias towards economic impacts (Hertin et al. 2009). From the 2008 Health Check onwards, estimated environmental consequences of different policy options are more consistently and substantially discussed in the impact assessments (Commission 2008, 2011).

The CAP reform negotiating process and legitimating discourse
With respect to the 1992 reform, the Commission champions a productivist discourse, but new discursive elements are introduced, including the 'dual role of the farmer as a producer of our food supplies and as a guardian of the countryside' (MacSharry 1991a). The latter 'multifunctional' environmentrelated consideration, however, surfaces in the policy preparation and result communication phases rather than in the decision-making arena, pointing at potentially strategic usage of the argument. Daugbjerg (2003) and  reach similar conclusions, claiming that the Commission's recourse to a multifunctionality discourse appears to be its response to the neoliberal challenge in the early 1990s, enabling a reframing of the CAP in order to preserve agricultural exceptionalism, albeit based on a different rationale.
The role of DG ENVI and environmental groups was limited in the 1992 reform. Even the agri-environment measures were fully instigated and developed by DG AGRI (interviews with Commission officials, 1 April 2016; 28 April 2016). The demands of environmental groups had instead focused on environmental conditionality. When such conditionality through cross compliance rules was introduced on a voluntary basis in 1999 and made mandatory in the 2003 reform, it was genuinely co-developed by DG AGRI and DG ENVI (interviews with Commission officials, 1 April 2016; 28 April 2016). This was possible partly because within DG AGRI only a small number of reformminded people, with close contacts to environmentalists, were instructed by Commissioner Fischler to prepare the reforms (interview with Commission official 28 April 2016). Member-state opposition to cross compliance and modulation, however, caused the Commission to leave the usage of these instruments up to member-states' discretion in the 1999 reform (Schwaag Serger 2001: 103-4).
Analysis of the discourse applied during the decision-making process on the two reforms allows for a good illustration of what can be expected (Table 1) under the circumstances of an incomplete shift towards postexceptionalism. On the one hand, the Commission presents a consistent discourse (no significant variation in terms of audience or timing): (a) assuming potential harmony between economic and environmental objectives, combining producer interests and environmental concerns (Fischler 2003); and (b) justifying direct income payments on the basis of a multifunctionality discourse as 'payment for the delivery of public goods' (Commission 1998: 103;Fischler 2002). On the other hand, the Commission's arguments in favor of cross compliance emphasize political considerationsthat consumers and society at large demanded more environmentally friendly productionrather than independent environmental goals (Fischler 1997). Environmental discourse appears strategically applied instead of indicating a 'reframing' of agricultural policy ideas in more environmental terms. Despite the application of productivist and multifunctionality discourses during the policy-making process, the CAP's policy heritage emphasizing income policy as the overarching goal remained intact (Daugbjerg and Swinbank 2016;. Together with DG ENVI, environmental groups such as Birdlife and the European Environment Bureau instigated the greening component in the 2013 CAP reform debate (Roederer-Rynning 2015: 346; interview Birdlife official 19 May 2016), which became a justification for continued direct payments. A stronger focus on pillar II policies could have been an alternative, but the Commission preferred measures that would be applied universally in the different member-states (Mathews 2013). Environmental organizations judged the eventual Commission proposals on greening insufficiently ambitious and feared the policy's fate once the Council and EP had their say (EEB andothers 2010, 2012).
Both the Council of Ministers and the EP contributed to watering down the environmental aspects in the 2013 CAP reform. 7 The European Parliament's COMAGRIwhose members tend to be biased in favor of farm interestshad acquired the formal lead position in the deliberation, and demanded relaxation of the greening requirements (Roederer-Rynning 2015; interview with Commission official 1 April 2016; interview with Birdlife official 19 May 2016). The final agreement reduced the greening requirements in terms of crop rotation and ecological focus area and allowed them to be fulfilled by 'equivalent' (among others, agri-environment) measures.
The justificatory discourse applied in the 2013 reform process shows important commonalities with Fischler's discourse a decade earlier. Commissioner Cioloș assumed that the CAP could reconcile economic competitiveness and environmental sustainability (2010b), argued that the greening criteria would establish a closer link between the direct payments and public environmental services (2011), but also repeatedly emphasized that the underlying need of greening was to make the payments more acceptable to society (2010a, 2011). Environmental considerations therefore seem particularly important as a political legitimation of existing policies, rather than as an objective in itself (for a similar conclusion see Daugbjerg and Swinbank 2016: 275). What is different and striking with respect to the latest reform, however, is thatnotwithstanding the application of a mix of discourses (Alons and Zwaan 2016) the post-2013 CAP reform debate shows the most 'productivist' focus compared to the other four reform debates, supporting the assisted agriculture paradigm (see also Erjavec and Erjavec 2015). This renewed emphasis can be explained by the experience of food shortages after several price-hikes during the previous decade. It had a decisive impact on the Council and EP during the 2013 reform (Swinnen 2015), supporting  claim that in the decisive policy arenas, agricultural exceptionalism remains dominant.

EPI as outcome
In the 1992 reform, the compulsory set-aside and agri-environment measures could be expected to have positive environmental (side-)effects. The way setaside was implemented in most member-states did not contribute to realizing this potential, however, both because non-productive land was taken out of production instead of environmentally vulnerable land, and because the land often was not fallowed, but original crops were replaced with allowed non-food crops (ECOA 2000;IEEP 2008: 16).
Existing evaluations of agri-environment measures tend to emphasize their effect on agricultural practices, rather than their actual effects on the environment, assuming that the agricultural practices supported by agri-environment measures will consecutively have positive environmental effects (Primdahl et al. 2003). Such analyses find that the uptake in terms of the percentage of utilizable agricultural area (UAA) enrolled in the programme has increased from 16 per cent in 1997 to 21 per cent in 2009, while the spending on agrienvironment measures as a percentage of total expenditure has increased from 0.6 per cent in 1993 to 4.5 per cent in 19974.5 per cent in (Buller 2000, which many still consider limited (Lowe and Baldock 2000). This upward trend was not continuous either, and the 2013 reform even brought a decrease in agri-environment spending in real terms.
More importantly, though, the design and monitoring of agri-environment measures often undermine their assumed environmental effects as well as their assessability. First of allnotwithstanding good exceptionsthe member-states' implementation of the measures overall seems to be geared towards maintaining existing practices and supporting farm income rather than ameliorating environmentally damaging practices (ECOA 2000(ECOA , 2011Hart 2015), impeding the policy's potential positive environmental effects. Next to this targeting problem, the objectives of the measures tend to be imprecise, lacking baseline levels and time frames for measurable achievements, which impedes monitoring and makes it difficult to judge whether or not objectives in terms of environmental effects are achieved (ECOA 2011;EEA 2006). The Commission, nevertheless, claims that the measures have prevented further intensification and resulted in a decreased usage of chemical fertilizers (Commission 1997). Whether this claim can be substantiated is questionable, as 'still very little information [is] available on the environmental benefits of agri-environment payments' (ECOA 2011).
For the 1999 reform, the ex-post evaluation shows that positive environmental externalities were limited. The implementation of voluntary cross compliance was lacking in many member-states and 'national envelopes' were used for income support rather than agri-environment measures (IEEP 2002). The fact that the more innovative and radical elements of the reform were optional, combined with the limited budget for pillar II, constrained the reform's impact (Buller 2002). After cross compliance became compulsory, research commissioned by DG AGRI found that, despite a wide variation in implementation between member-states, the policy had improved compliance with environmental obligations (Alliance Environment 2007). Reports of the ECOA countered these claims, lamenting the merely partial implementation of the requirements by member-states, their unreliable reporting on compliance and the weak sanctions (ECOA 2008), noting an increase in infringements from 21 per cent in 2011 to 27 per cent in 2014 (ECOA 2016). The Court furthermore argued that the indicators the Commission collected data on (percentage of CAP payments and acres covered by cross compliance) did not measure the effectiveness of cross compliance. Although, at the overall level of CAP, data is also collected on farmland bird indexes, water quality and soil erosion, the specific impact of cross compliance on these indicators remains unknown (ECOA 2016). As was the case with agri-environment measures, the (additional) environmental effects of cross compliance appear limited.
The positive estimates of the Commission's impact assessment of the 2013 reform were called into question by research institutes like the Institute for European Environmental Policy (IEEP) (2013). It is argued that farmers would hardly have to change their policies to be eligible for the greening payments and that, therefore, the overall environmental benefits would be limited (Bureau and Mahé 2015: 107-8). Moreover, the greening measures were watered down further in the implementation phase (Hart 2015). Commission officials also doubt that greening will have more positive environmental effects than the previously existing policies (interview with Commission officials 1 April 2016; 28 April 2016). While a recent Commission working document draws positive conclusions on the amount of ecological focus area, it emphasizes that farmers have 'optimized their EFA choices on economic grounds' rather than environmental ones (Commission 2016: 19). Although it is too early to draw robust conclusions, the greening requirements are unlikely to have significant environmental effects.

Conclusion
This paper developed the argument that the incomplete transformation from an exceptionalist agriculture to a post-exceptionalist agriculture policy in Europeboth in terms of the associated institutional power structures and dominant ideashas brought only limited EPI in the CAP. While changes in policy-making procedures and institutions have provided environmental actors with enhanced access to the policy-making process, their impact on this process did not significantly increase. Although the legitimating discourse has increasingly emphasized environmental concerns, this discourse seems to be particularly applied strategically as a vehicle to legitimate existing practices (as is also claimed by Daugbjerg and Swinbank 2016). Environmental objectives have become a variable in the agricultural policy-making equation, but its coefficient remains small. This explains why dissatisfying effectiveness of environmental measures has not instigated more far-reaching policy adaptations.
The conclusions of this study are limited to agriculture policy in the European Union. Further research in the agricultural sector could investigate how EPI may have been shaped differently in countries where exceptionalism has given way to more market liberal perspectives on agriculture, such as in the United States. This study is furthermore limited in depth, as longitudinal comparison and development was prioritized over presenting the full intricacies of the reform process. Finally, the exclusion of horizontal EPI from the analysis may have resulted in insufficient appreciation of the indirect effects on the CAP of environmental obligations based on horizontal European regulations and agreements.
What does the analysis of EPI in the specific case of the CAP add to the broader EPI literature? First of all, the three-dimensional conceptualization developed in this paper can also be applied to EPI in other fields. Though exact measurements may need to be adapted, it is a rather general conceptualization that allows for a relatively comprehensive analysis of EPI and nuanced conclusions based on clear distinctions between the different dimensions of EPI. The importance of such a multi-dimensional approach is underlined by the findings in this paper that increased access of environmental actors to the decision-making process does not necessarily give them much influence, and that more mandatory policy instruments (output) do not guarantee more effective outcomes. This brings me to the second contribution of this paper, which is the focus on the crucial role of ideas in shaping EPI. The case study has shown that as long as institutional changes are not accompanied by a change in policy ideas, environmentally oriented policies will remain unstable and their effectiveness limited. While a shift from exceptionalism to post-exceptionalism may not be applicable to other domains (or may differ in similar domains in other countries), these domains will have their own policy paradigms affecting EPI trajectories. In the case of market-oriented agriculture, for example, environmental concerns would have to be connected to considerations of competitiveness and a level playing field, while a public goods justification would only be acceptable to the extent that these 'goods' are quantifiable to some degree in order for the idea of supply and demand to apply. This has important implications for both policy-makers and researchers. The former would have to apply domain-specific justificatory discourses to accomplish a durable 'reframing' of different sectors in more environmental terms, while the latter have to adapt their analytical lenses to a domain's ideational peculiarities.