Éric Montpetit (Université de Montréal)

Séminaire du Staff / Printemps 2008 Éric Montpetit (Université de Montréal) Policy Learning in the European Union: A Comparison with North America Discussant: Karin Ingold Lundi 21 Avril 12h15 14h00 M4276

Policy Learning in the European Union: A Comparison with North America Éric Montpetit Université de Montréal Département de science politique C.P. 6128, succursale Centre-ville Montréal (Québec) Canada H3C 3J7 Phone : (514) 343-7299 Fax : (514) 343-2360 Email : e.montpetit@umontreal.ca The author acknowledges the financial support of the Social Science Research Council of Canada and thanks Rukmini Canape-Brunet and Catherine Pelletier for excellent research assistance as well as Peter Loewen, Nathalie Schiffino and David Aubin for their comments and suggestions.

Abstract This article tests the hypothesis suggesting that policy processes in the European Union are particularly conducive to policy learning. The statistical analysis provided in the article rests on a comparative survey of actors involved in biotechnology policy development in Europe and North America. The article focuses on policy learning intensity, on capacity for consensus formation and on policy transfers. In contrast to some of the scholarly work on policy-making in the European Union, the survey fails to provide evidence that policy actors in Europe learn more than policy actors in North America. Introduction Do policy actors in the European Union learn more than policy actors in North America? An important share of the scholarly work on the European Union suggests an affirmative answer to this question. It argues that continental integration has given shape to a unique governance structure, in which authority is no longer the purview of nationstates. Delegated to various non-governmental actors or shared between public and private actors, authority in this governance structure does not depend on hierarchy so much as it rests on persuasion. Unlike conventional nation-states in which actors at the top of hierarchies can make decisions alone, the horizontal governance structure of the European Union requires diverse actors to discuss with each other, learn and reach a consensus before any decision can be made. Several Europeans take pride in this governance structure, which places dialogue, reasoned argumentation and learning above 1

the coercive means and the discretionary authority of top officials. These Europeans will even look down on American politics, which they view as too adversarial (Kelemen 2006). To my knowledge, however, no empirical study offers a comparison of the propensity of European and North American policy actors to learn. The purpose of this article is to provide such a comparison. Specifically, the article proposes three tests of the propensity of European policy actors to learn. These tests rest on a comparative survey of actors involved in the development of biotechnology policy in Europe and North America. The biotechnology sector appears well-suited as it involves natural systems, in which learning capacity is higher according to Sabatier and Zafonte (2001). Moreover, biotechnology involves subsystems in which the European Union plays an important role (e.g. GMOs) and subsystems in which it does not (human genetics). This characteristic of the biotechnology sector adds comparative possibilities, although the article s main comparison is between Europe and North America. The survey provides information on learning intensity, on capacity for consensus and on policy transfers. All three tests failed to find any significant differences in inclinations to learn between Europe and North America. These results cast a shadow on theories suggesting that European integration has produced a unique governance structure, featuring learning as its main mechanism for decision making. The article is divided in five sections. First, I review the literature suggesting that European policy processes promote learning. Second, I introduce the biotechnology actor survey. Third, I present the three tests. Fourth, I summarize the results. Lastly, I revisit the literature in light of the results. 2

Policy Learning in the European Union Studies of policy learning can variously center on organizations or individuals and refer to technical objects or relatively abstract ideas (Bennett and Howlett 1992; Zarkin 2008). In this paper, I am interested in learning about various policy options, some technical and others related to general ideas, by individuals involved in policy-making. I adopt Sabatier and Zafonte s (2001: 11566) definition of policy learning as relatively enduring alterations of thought or behavioral intentions. Sabatier and Zafonte (2001: 11566) add that learning normally result[s] from experience and/or the assessment of new information involving the precepts of belief systems. The change of thought involved in learning, therefore, does not come lightly, without some form of confrontation between external information or experience challenging the ideas and the beliefs of the individual who learn has held for some time. In short, policy learning should occur in processes involving face-to-face deliberations among actors who have diverse ideas and beliefs (Mutz 2006: 63). Sabatier s advocacy coalition framework provides one of the most influential understandings of policy learning (Jenkins-Smith and Sabatier 1993). The framework suggests that actors choose their coalition allies based on the proximity of core policy beliefs (Leach and Sabatier 2005). In any issue area a subsystem in the framework s terminology actors should divide themselves in at least two adversarial advocacy coalitions competing for the realization of conflicting core beliefs. The framework hypothesizes that members of competing advocacy coalitions rarely interact with each other and carefully filter incoming information and evidence: The result is a dialogue of the deaf in which members of different coalitions talk past each other (Sabatier and 3

Zafonte 2001: 11566). As underlined by critics, Sabatier s advocacy coalition framework is closer to a non-learning model (Fisher 2003). In fact, the framework s prediction regarding policy learning is consistent with recent research suggesting that people dislike political debates (Hibbing and Theiss-Morse 2002). According to Mutz (2006), citizens prefer avoiding politics rather than advocating ideas fellow citizens might dislike. Therefore, face-to-face deliberations among people who hold different views would occur rarely. The advocacy coalition framework, however, acknowledges that in some subsystems members of competing coalitions can be attracted by the same reputed professional forums. When face-to-face meetings between members of competing coalitions occur in these forums, learning is encouraged, engendering alterations in the thoughts of participants. These forums are more likely to exist in sectors dealing with natural systems, such as biotechnology, rather than social systems (Sabatier and Zafonte 2001). However, empirical tests of the Advocacy Coalition framework have provided more evidence of adversarial relationships than evidence of learning between coalitions (Sabatier and Jenkins-Smith 1999). Sabatier (1998) acknowledges that his framework was developed with the adversarial American political system in mind. Thanks the separation of powers, the fragmentation of governance and ensuing venue shopping strategies, the American political system might not encourage policy learning to the same extent as more consensual systems. However, Sabatier (1998) argues, European integration increases the complexity of policy relevant relationships and opens up policy-making to new actors, bringing Europe closer to the United States. Likewise, Kelemen (2006) argues that the 4

vertical rivalry between the European Union and member-states on the one hand and the horizontal rivalry between the Commission, the Council of Ministers and the European Parliament on the other hand have encouraged the production of detailed laws and regulations, which are key features of adversarial systems. In fact, these detailed laws and regulations have strengthened the European Court of Justice, solicited regularly to resolve litigations over their enforcement. A number of years ago, Streeck (1992) had already noted that continental integration has moved Europe away from corporatism and closer to American pluralism. In the European Union of the Single European Act, he argues, state and key civil society actors no longer cooperate in stable relationships based on trust. National peak associations, once partners of nation-states in these relationships, face the challenge of coordinating their efforts at the European level without eroding their internal cohesion. The challenge is rarely surmounted and the result often is increased fragmentation in interest systems. In contrast with corporatist environments, interest groups and other private actors in this new environment become adversaries in competitions for access to policy-makers and influence, both at the member state and European Union levels. Stable relationships based on trust are slowly replaced by formal rules governing actor competition and dispute resolution. Thanks to changes engendered by European integration, policy-making in Europe would become increasingly indistinguishable from policy-making in the United States, where policy learning is limited by adversarial attitudes. Not all analysts, however, would agree that the European Union is evolving in the direction of the American political system (Howlett 2007). Surely, European integration has transformed the corporatist arrangements of several member-states, notably by 5

dispersing power within large networks of actors in several sectors (Héritier and Lehmkuhl 2008; Skogstad 2003). These networks, however, would not be characterized by an adversarial style. In fact, some analysts claim that European Union policies are frequently made in networks characterized by a consensual policy style (Risse-Kappen 1996; Peterson 1997). In other words, the consensual policy style associated with several member-states proved resilient despite the extraordinary institutional and network changes engendered by continental integration (on member-states styles see Richardson 1982; Katzenstein 1985; Lijphart 1999). Logically, large horizontal networks, in which actors seek a consensus, should encourage face-to-face discussions between actors who have different policy views. Because these encounters are consensus-oriented, actors cannot filter out inconvenient views and associate themselves only with actors who share similar beliefs. Consensus building requires actors willing to hear what the other side has to say. In short, European consensus-oriented networks encourage policy learning (Börzel 1998: 264; Kohler-Koch 1996; Mutz 2006). Surely enough, several analysts agree that the European Union, far from undermining policy learning, encourages it to a larger extent than American institutions. Perhaps, however, that the institutional difference between Europe and North America does not so much create differences in the quantity or intensity of policy learning between the two continents, as it simply encourages different propensities on the part of actors to reach consensus. Scholars working on legitimacy deficits would indeed suggest so much. Consensus in the European Union should be more important than in the North America, as actors taking part in policy development are also influenced by different concerns over legitimacy deficits. With the direct election of the chief executive 6

every four years and regular congressional, state and local elections, the American political system, as well as the Canadian political system, compare favorably with the European Union in terms of democratic legitimacy. Analysts of the European Union have complained that the limited power of the European Parliament, the exercise of quasilegislative functions by member states executive officials through the Council of Ministers and the key role played by the bureaucrats of the Commission in policy development leave the European Union with a large democratic deficit (Risse-Kappen 1996). As illustrated by the failure of the European constitution, any project to change the unusual institutional architecture of the European Union in view of improving democratic legitimacy would face serious obstacles. European officials, therefore, look elsewhere for sources of legitimacy (Skogstad 2003). Students of the European Union have argued that European officials frequently resort to expertise to legitimize their policy decisions (Verdun 1999; Haas 1998). Moreover, the expertise would not be centrally located within the European Commission whose resources are limited (Kohler-Koch 1996). Rather, the Commission, sometimes in concert with the Council of Ministers, would resort to committees of experts, encouraging the formation of expert networks (Peterson 1997; Risse-Kappen 1996; Marks, Hooghe and Blank 1996). Experts, invariably drawn from various member states, would be brought together on these committees to provide advice to European officials. Made of people speaking different languages, belonging to different cultures and with different educational backgrounds, these expert committees and networks are the sites of dense interactions among diverse actors, creating policy-learning opportunities. More importantly, the mandate of the committees to provide advice would encourage experts to 7

come to a consensus (Wessels 1998). Skogstad (2003) speaks of the integrative function of expert networks, as an essential source of democratic legitimacy in the European Union. Only when deliberations and dialogues among a diversity of actors produce a consensus around policy options can European officials be reassured that knowledge will comfort their decisions with some legitimacy. In contrast, American officials do not seek expertise with the same legitimacy concerns in mind. Expertise in the American system will often be used in a partisan fashion, simply as a way for an official to give credibility to her or his position vis-à-vis that of her or his opponents (Jasanoff 1990; Weiss 1989). If actors can learn from this adversarial process, they are likely to draw all kinds of lessons. The same expertise can convince one Republican-leaning policy-maker to support a position advocated by Democrats and convince another similarly Republican-leaning policy-maker of the correctness of a Republican position. American policy actors, in contrast to their European counterparts, do not need to make any effort to reach a consensus since final policy decisions belong to elected representatives, upon whom rests the legitimacy of decisions. In short, learning would be more consensual in the European Union than in North America, thanks to different conceptions of legitimacy. Radaelli (2000) pays attention to the particular form of consensus the legitimacy concerns of European officials engender. He argues that the committee system of the European Union encourages the diffusion of knowledge about the policy experiences of member states. Moreover, committees will often make a judgment about which experiences work and which do not. These judgments would be particularly useful to European officials who would feel comfortable drawing lessons from the successful 8

policy experiences of member states, which cannot be suspected of lacking democratic legitimacy. European consensus, therefore, would be about the experiences of member states, from which transfers should be contemplated. In North America, actors who participate in policy development are unlikely to possess direct knowledge of the policy experiences of countries from which a transfer could proceed. In fact, North American actors should not value policy transfers to the same extent as their European counterparts in the first place, as they have distinctive legitimacy concerns. Following Radaelli (2000), therefore, policy development in North America should not be as favorable to policy transfers as it is in the European Union. In sum, learning refers to a relatively enduring change of views about a policy. Sabatier, who has provided the most influential understanding of policy learning, argues that policy processes fail frequently at encouraging learning with any intensity. Actors hold to their core beliefs, changing only their thoughts at the periphery of their belief systems. Sabatier (1998) adds that policy processes in Europe are not sufficiently different from those in North America to warrant a different prediction about learning. Several analysts disagree, arguing that policy-making in Europe occurs in large networks characterized by a consensual style more conducive to learning than the adversarial American style. The difference between the two continents would be particularly salient over the capacity of learning to generate consensus. In conjunction with policy styles, concerns in Europe about a democratic deficit would encourage the formation of consensus around successful member state experiences, from which policy transfers should proceed. 9

This article proposes statistical tests of three hypotheses, which emerge from the above literature review. Specifically, the article tests hypotheses suggesting that policymaking in Europe is more conducive to learning, in terms of intensity (1), consensus building (2) and policy transfers (3), than policy-making in North America. The data used to test these proposals come from a survey of policy actors, presented in the next section. The Biotechnology Actor Survey The Biotechnology Actor Survey (BAS) is a comparative survey of actors involved in human genetics and agri-food biotechnology policy development in France, the United Kingdom, Canada and the United States. Respondents are civil servants, interest group leaders and independent experts. The survey was designed to test hypotheses about policy learning with questions on policy opinions, their evolution over a 12 month period and the reason for change. The sector of biotechnology appeared particularly appropriate as it involves a natural rather than a social system (Sabatier and Zafonte 2001). Moreover, biotechnology knowledge evolves at a fast pace with several applications being popularized in the western world in the last 20 years. The survey relied on an internet instrument. Potential respondents were first invited to participate via email in June 2006. To obtain their names and contact information, newspaper articles, parliamentary hearings and scholarly work on biotechnology was examined. This examination was followed by a web search for these individuals and their organizations. After the first round of invitations was sent, the snow ball method was used to multiply invitations, until saturation was reached in December 10

2006 when 1967 invitations had been sent. The survey, hosted on a web site accessible only through a link on email invitations, took at least 30 minutes to fill out. 1273 out of the 1967 email invitations ended up in the inbox of potential respondents, the remaining ones were either blocked by anti-spam software or were sent to addresses which were no longer valid. Among the 1273 who received the email invitation, I do not know how many individuals refused to follow the link for fear of virus infection, fraud or who simply did not read the email out of negligence. Of the 1273 potential respondents, 270 actually filled out the survey in part or entirely. Taking into account only the 229 respondents who answered more than 70% of the survey, and using the 1273 figure as denominator, the response rate is 18%. Respondents are fairly distributed between the two subsystems, along actor types and continents (see Table 1). The only comparable survey of which I am aware, a mail survey, has fewer respondents and a similar distribution along actor types (Aerni and Bernauer 2005). [Table 1 about here] The survey provides information on policy learning through a question asking respondents how their opinion had changed in the last year about 17 policy options, 10 related to the agri-food sector and 7 to human genetics. The exact phrasing of the question, the 17 options and the answer possibilities are presented in Appendix 1. Using information on these 17 options, a measure of learning intensity was constructed. Specifically, learning intensity refers to the percentage of options about which a respondent indicates having altered her or his opinion. The closer a respondent is to 11

100%, the more intensively she or he claims to have learned in the past year. The capacity for consensus is indicated by the importance of movements of opinion on each of the 17 options. Lastly, information on policy transfers was collected through a question on the factors which influenced the opinion of actors most. The policy experience of foreign countries was one of the answer options. Respondents were also asked to list the countries whose policy experience was worth drawing from. The descriptive statistics for these measures of learning are presented in Appendix 2. 1 The Appendix reveals that, on average, North American actors learn more intensively than European actors. Surprisingly, the capacity for consensus also appears slightly higher in North America. The last column of the Appendix s second table presents the results of a subtraction of the percentage of actors having become less favorable to the percentage of actors having become more favorable. The difference provides an indication of the magnitude of opinion movements in one direction or another. The farther the figure is from 0, the more important is the movement. Assuming that a percentage above 10 suggests a movement of opinion large enough to indicate consensus formation, North American actors display such a capacity on more options than European actors. In fact, movements of opinion above 10 percentage points occur on 10 options in North America and only on 8 options in Europe. Discounting for the options on which an important movement occurs simultaneously on both continents (because learning about options can occur at the international level), North Americans display a capacity for consensus on 4 options and Europeans on only 2 options. Lastly, North American actors claim to be influenced by the policy experiences of foreign 1 The list of lesson-inspiring countries is presented in Figure 1, discussed bellow. 12

countries slightly more frequently than Europeans. These results provide evidence against the three hypotheses presented above. These statistics, however, are only descriptive statistics which do not allow comparisons with other factors likely to influence policy learning. To enable such comparisons, I now move to more sophisticated statistical tests. The Tests The data collected through the survey provide several possibilities to test differences between Europe and North America on policy learning. First, besides the continent of origin of the respondent, the survey provides information on the biotechnology subsystem and the group to which a respondent belongs. This information can be used in a regression analysis to compare the extent to which geography and institutions prevail over the policy function of groups as far as learning is concerned. 2 Given the first hypothesis, I expect a significant coefficient for continent of origin, suggesting that European actors learn more intensively than North American actors. Moreover, a significant coefficient for an interaction between continent and subsystem, suggesting that agri-food biotechnology actors in Europe learn more than the other actors, would be evidence that the institutions of the European Union matter. In fact, the European Union plays an important policy role in the area of agri-food biotechnology, creating interaction opportunities among actors from various member states. 3 As 2 Learning intensity is not distributed normally. Forty percent of the observations are left censored at 0. Therefore, I use Tobit estimations. 3 Unfortunately, however, the survey does not provide any information on the extent to which respondents participates in committees of the European Union. Therefore, I have to assume that actors in the agri-food sectors participate more in European committees, because the European Union is more involved in this subsystem than in the human genetics subsystem. 13

explained above, such interactions could be conducive to learning. In contrast, human genetics has largely remained the domain of member states (Abels 2007). The European Union has fewer committees preparing policy proposals in this latter area, limiting the opportunities for interactions among actors from different member states. In the human genetics subsystem, European actors should therefore learn less than in the agri-food subsystem. To measure adequately the possibility of an institutional or geographical influence on policy learning, it would be useful to compare it with an alternative source of learning. The information derived from the survey allows a comparison with the potential influence of the policy functions of respondents on policy learning. It could indeed be argued that governmental actors have policy functions sufficiently distinctive from interest groups or independent experts to learn in a distinctive manner, independently of the country in which they work (Aberbach, Putnam and Rockman 1981). Strong evidence behind our hypothesis on learning intensity in the European Union would be provided by significant coefficients for the variables related to continent and subsystem, in conjunction with insignificant coefficients for group related variables. Such results would indicate that the distinctive policy function of groups have less influence on policy learning than the geographical and institutional environment in which all groups perform these policy functions. On each of the 17 policy options presented in Appendix 1, an important movement of opinion in either direction would indicate a capacity to form a consensus. Using the categorical variables created from the question on the 17 options (see Appendix 2), multinominal logistic regressions enable comparisons of the capacity to form 14

consensus between two continents, the two subsystems, and within the three groups of actors. To support the hypothesis about consensus formation, the variable for continent should indicate a capacity for consensus in Europe on a larger number of agri-food options than in North America. Continental differences should not be as pronounced over the human genetics options, although member-state policy styles might have an effect. Moreover, the group variables should have significant coefficients less frequently than the continent variable. The third and last test relates to policy transfers. According to the third hypothesis, European actors should claim to be influenced by the experience of foreign countries more frequently than North American actors. I suggest a logistic regression to test this last hypothesis. In a manner similar to the regression on learning intensity, I expect the interactive variable between continent and subsystem to indicate that transfers are more prevalent in Europe in the agri-food subsystem. The variables for groups should be insignificant. The Results The results of the first and third tests are presented in Table 2. The findings for learning intensity do not match the expectations related to the first hypothesis. European actors do not learn more intensively than North American actors, even in the agri-food subsystem. In contrast to what the descriptive statistics suggest, North American actors do not learn significantly more than European actors either. The results of the first regression in fact suggest that group related policy functions have a greater influence on learning that geography and institutions. Non-governmental actors, independent of their 15

continent of origin, learn more intensively than independent experts, the reference category. However, governmental actors do not learn any more intensively than independent experts. In short, the first test provides evidence against the hypothesis suggesting that European Integration encourages more learning than North American adversarial policy style. [Table 2 about here] The results for our second test are not presented in a table format because they did not lend themselves easily to a brief presentation. As a reminder, the second test involves 17 multinominal logistic regressions, each accounting for movements of opinion in the direction of more and less favorable about a policy option. In any case, the regressions do not yield any support to the hypothesis suggesting that European actors have a greater capacity to form a consensual opinion than North American actors. In fact, the variable continent is significant (at p 0.1) for one policy option in human genetics rather agrifood biotechnology: data collection about patients and donors. Moreover, the sign of the coefficient indicates that North American actors are more likely than European actors to reach a consensus around this policy option. In comparison, non-governmental actors, independent of their continent of origin, are more likely than independent experts to experience a movement of opinion, in either of the two directions, on five policy options. 4 These results, just like the results of the regression on learning intensity, suggest that the distinctive policy functions of groups are more important for learning than geography and institutions. 4 Policy options 1, 6, 14, 15 and 16 in Appendix 1. 16

The third test also provides evidence against the idea that policy learning follows distinctive patterns in Europe. In the last column of Table 2, the coefficients for the two variables related to continents are not statistically significant. This result suggests that European actors are no more likely to be influenced by policy transfers than North American actors. Figure 1 provides additional evidence. The figure indicates the number of respondents who checked listed countries as providing inspiring policy lessons in the area of biotechnology. 5 The European Commission might attempt to legitimize itself by persuading policy-makers that the experience of a member-state is worth borrowing, the results nevertheless indicate that European policy actors are not any more inspired by the policy experiences of members-states than North American policy actors. Moreover, North American actors are just as likely as European actors to be inspired by the policy experience of European member states. [Figure 1 about here] In contrast to learning intensity and consensus capacity, however, policy transfers are not influenced by the policy functions of groups of actors. Government actors and non-government actors are no more or no less likely to be influenced by the policy experiences of foreign countries. The results also show that actors in the human genetics subsystem are more likely to draw lessons from foreign countries than actors in the agrifood biotechnology subsystem. 5 A similar question asked respondents to list countries whose policy experience suggests avoiding some options. The list of countries was significantly different from that of Figure 1. However, the pattern between the two continents was identical, 17

In short, none of the three tests yielded results supporting the three hypotheses about policy learning in Europe. European actors in the area of biotechnology policy do not learn any more intensively than North American actors. European actors do not have any more consensual learning capacity than North American actors. And European actors do not claim to learn from the policy experiences of foreign countries, especially European member-states, any more than North American actors. Discussion Comparative policy research has emphasized the role of institutions, or other geography-specific variables such as policy styles or culture, in explanations of policy outputs. This trend was encouraged by students of the European Union who have treated this system of governance as unique; not quite a state, nor an international organization. The structure of governance gradually constructed with European integration would have given form to distinctive ways of making public policy. Policy-making in the European Union would occur in networks, in which government actors often take a back seat (Peterson 1997; Marks, Hooghe and Blank 1996; Héritier and Lehmkuhl 2008). Power in these networks would be exercised softly, that is through persuasion, rather than through coercion (Kohler-Koch 1996; Risse-Kappen 1996; Skogstad 2003). Expert knowledge would play a key role in decision-making (Verdun 1999; Haas 1998; Wessels 1998). Authors who have argued that European Union institutions create distinctive patterns of policy-making have not always referred directly to the concept of policy learning; however, most of them suggest clearly that policy learning is a by-product of several of the Union s policy processes. Risse-Kappen (1996: 73) speaks about the deliberative 18

and communicative processes of the European Union, capable of opening space for new and innovative solutions. Kohler-Koch (1996: 372-3) insists on the role of the European Community in processes of social learning to propagate the idea of cooperative problem-solving, which itself is an idea concerning governance structure conducive to joint-learning. In other words, several analysts have insisted, since the 1990s, that the policy-making structures developed in the European Union promote policy learning, that is relatively enduring alterations of thoughts, often in a consensual direction. To my knowledge, this distinctive policy-making feature of the European Union has not been empirically examined through comparisons with countries from other continents. This article fills this gap by providing such a comparison. Specifically, the article examines the propensity of European actors to learn in comparison with North American actors. In addition, it compares two distinctive policy subsystems, one in which the European Union is highly involved and one in which it is not. These comparisons fail to confirm that policy learning is more prevalent in Europe or that European Union institutions promote it. Interestingly enough, more support was found for differences in learning capacity between groups of actors having distinctive policy functions than for continental differences. Interest groups leaders were indeed found to learn more intensively and more frequently in a consensual manner than independent experts or governmental actors, independent of their continent of origin. This finding suggests that studying actor-related variables is at least as important as studying the distinctiveness of institutions and related policy styles. 19

Moreover, differences were found between the human genetics and agri-food biotechnology subsystems as far as drawing policy lessons from foreign countries is concerned. Actors claim to learn more from foreign countries in the human genetics subsystem, independent of their continent of origin. This finding makes sense for anyone who has knowledge of these two subsystems. The regulation of risks related to agri-food biotechnology is an internationally contested issue, with every country claiming to have the best system. 6 In contrast, the regulation of human genetics, outside of the United States, was inspired by the early model provided by the United Kingdom. In 1990, British policy-makers adopted a law to enable the creation of an autonomous governmental agency to regulate this sector at arm s length. The British experience with this agency was seen as sufficiently positive by French and Canadian policy-makers to convince them to draw legislations enabling the creation of similar agencies in 2004 (Montpetit, Rothmayr and Varone 2007). Interestingly enough, Appendix 2 reveals that the creation of such an agency was the object of large movements of opinion on both sides of the Atlantic. Moreover, when asked which country has a lesson inspiring policy experience, 61 out of the 117 respondents in the human genetics subsystem mentioned the United Kingdom. In contrast to Radaelli s (2000) argument, I found no evidence to suggest that the European Union needs or obtains policy transfers to legitimize its policy decisions to a larger extent than the United States or Canada. Does the failure to find evidence of a difference in policy learning between Europe and North America confirms the prediction of Sabatier s advocacy coalition framework? As a reminder, the framework suggests that in systems where policy making responsibilities are dispersed, learning occurs slowly, if at all. Moreover, Sabatier (1998) 6 Even Canada and the United States disagree on which country has the best system. See Montpetit (2005). 20

argues that European integration has moved policy-making on this continent closer to the United States. Similarly, Kelemen (2006) argues that European integration encourages the development of an adversarial policy style, similar to the American policy style. Therefore, Kelemen and Sabatier would not be surprised by this article s finding about the absence of a difference between Europe and North America. The absence of a continental difference, however, should not be interpreted as evidence that the European Union encourages an adversarial policy style or that it limits actors capacity to learn. In fact, it might simply be that consensus and learning are more prevalent in the United States than mainstream theories suggest. Several scholars disagree with Sabatier s argument that policy actors in adversarial systems, the United States in particular, do not learn much (Innes and Booher 2003; Fisher 2003; Schön and Rein 1994). Washington has a wealth of think tanks producing a large amount of knowledge (Rich 2004). In this fragmented system, policy-makers from all sides of any issue demand sophisticated knowledge to substantiate their policy positions and convince their fellow policy-makers to give their support (Weiss 1989). As a result, policy actors in the United States, if they do not directly deliberate against adversaries, they are exposed to a wealth of diverse knowledge, resting on serious research, on any policy issue. In this context, American policy actors are exposed to different views, including views coming from the other side. In addition, the quality of the research behind the views of the other side might convince actors to accept some arguments coming from the other side. Actors might even come to the view that deliberation and cooperation with actors from the other side is better than confrontation. Again, deliberations among actors with different views engender policy learning. Surely enough, some American political scientists argue or 21

suggest that Sabatier s advocacy coalition framework underestimates the amount of learning resulting from American policy processes (Schneider and Ingram 2007; Mansbridge 1993). To prove Kelemen and Sabatier right or wrong, information on the attitudes of European actors toward actors who hold ideas and policy positions different from their own would be necessary. Unfortunately, the biotechnology actor survey does not provide such information. To engage in a discussion over the validity of the advocacy coalition framework, a comparison with the attitudes of policy actors involved in sectors dealing with social systems would have been useful. Again, the framework predicts less learning in these latter sectors than in sectors dealing with natural systems, such as biotechnology. Therefore, social versus natural subsystem differences should be more important than continental differences, according to the framework. Unfortunately, I do not have the data to make such a comparison. In short, the results of the three tests presented in this article call for a significant revisions of the theories suggesting that governance in the European Union is particularly conducive to policy learning. However, these results should not be straightforwardly interpreted as supporting theories suggesting that European integration nourishes adversity and undermines policy learning. Conclusion The objective of this article was to assess the extent to which policy-making in Europe involves learning to a larger extent than in North America. While Sabatier and others claim that European integration moves policy processes closer to what they are in 22

North America, the United States in particular, several students of the European Union have insisted alternatively on a distinctive policy style, on unique governance structures and on policy transfer practices making up for the particular legitimacy deficit facing the Union institutions. If students of the European Union are right, we should observe distinctive patterns of policy learning between Europe and North America. Using a survey of biotechnology policy actors, this article proposes three tests assessing differences between the two continents on learning intensity, capacity for consensus and propensity to resort to policy transfers. All three tests failed at finding differences between Europe and North America. European actors learn, but no more than North American actors. This been said, from the three tests presented in this article, I cannot conclude that policy-making processes in the European Union have no distinctive feature or effect. My conclusion is only relevant to policy learning, measured in terms of intensity, consensus formation and transfers from other countries. The European Union might transform policy-making in a manner that has an effect on power distribution among actors or levels of government for instance and eventually on substantive policy decisions. Further comparative studies with North America would be necessary to test such hypotheses. These additional tests could resort to methods other than survey methods, perhaps to capture policy change better. The survey presented in this article provides a convincing measure of the extent to which actors change their views on policy over a short period of time. However, policy change might not only occur through alteration of thoughts on the part of actors contributing to policy development. As Héritier and Lehmkuhl (2008) show, the structural change caused by European integration brought new actors in policy 23

processes and enabled strategies likely to engender policy change, independently of policy learning. References Abels, G. (2007) Trade and Human Rights: Inter- and Supranational Regulation of ART and GM Food, in É. Montpetit, C. Rothmayr and F. Varone (eds) The Politics of Biotechnology in North America and Europe: Policy Networks, Institutions, and Internationalization, Lanham: Lexington Books. Aberbach, J.D., Putnam R.D. and Rockman B.A. (1981) Bureaucrats and Politicians in Western Democracies, Cambridge: Harvard University Press. Aerni, P. and Bernauer, T. (2005) Stakeholder Attitudes Toward GMOs in the Philippines, Mexico, and South Africa: The Issue of Public Trust, World Development, 34(3): 557-575. Bennett, C.J. and Howlett, M. (1992) The Lessons of Learning: Reconciling Theories of Policy Learning and Policy Change, Policy Sciences, 25: 275-294. Börzel, T.A. (1998) Organizing Babylon: On Different Conceptions of Policy Networks, Public Administration 76: 253-273. 24

Fisher F. (2003) Reframing Public Policy: Discursive Politics and Deliberative Practices, Oxford: Oxford University Press. Haas, P.M. (1998) Compliance with EU Directives: Insights from International Relations and Comparative Politics, Journal of European Public Policy, 5(1): 17-37. Héritier, A. and Lehmkuhl D. Introdution: The Shaddow of Hierarchy and New Modes of Governance, Journal of Public Policy, 28: 1-17. Hibbing, J.R. and Theiss-Morse E. (2002) Stealth Democracy: Americans Beliefs about How Government Should Work, Cambridge: Cambridge University Press. Howlett, M. (2007) Analyzing Multi-Actor, Multi-Round Public Policy Decision- Making Processes in Government: Findings from Five Canadian Cases, Canadian Journal of Political Science, 40(3): 659-684. Innes J.E. and Booher D.E. (2003) Collaborative Policymaking: Governance Through Dialogue, in M.A. Hajer and H. Wagenaar (eds), Deliberative Policy Analysis: Understanding Governance in the Network Society, Cambridge: Cambridge University Press. Jasanoff, S. (1990) The Fifth Branch: Science Advisers as Policymakers, Cambridge: Harvard University Press. 25

Jenkins-Smith, H.C. and Sabatier P.A. (1993) The Dynamics of Policy-Oriented Learning, in P.A. Sabatier and H. Jenkins-Smith (eds), Policy Change and Learning: An Advocacy Coalition Approach, Boulder: Westview Press. Katzenstein, P.J. (1985) Small States in World Markets: Industrial Policy in Europe, Ithaca: Cornell University Press. Kelemen, D. R. (2006) Suing for Europe: Adversarial Legalism and European Governance, Comparative Political Studies, 39: 101-127. Kohler-Koch, B. (1996) Catching Up with Change: The Transformation of Governance in the European Union, Journal of European Public Policy, 3(3): 359-380. Leach, W.D. and Sabatier P.A. (2005) To Trust an Adversary: Integrating Rational and Psychological Models in Collaborative Policymaking, American Political Science Review, 99(4): 491-504. Lijphart, A. (1999) Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries, New Haven: Yale University Press. 26

Mansbridge, J.J. (1992) A Deliberative Theory of Interest Representation, in M. Petracca (ed), The Politics of Interest: Interest Groups Transformed, Boulder: Westview Press. Marks, G., Hooghe, L. and Blank K. (1996) European Integration from the 1980s: State- Centric v. Multi-level Governance, Journal of Common Market Studies 34(3): 341-378. Montpetit, É. (2005) A Policy Network Explanation of Biotechnology Policy Differences between the United States and Canada, Journal of Public Policy 25(3): 339-366. Montpetit, É, Rothmayr, C. and Varone F. (eds) (2007) The Politics of Biotechnology in North America and Europe: Policy Networks, Institutions, and Internationalization, Lanham: Lexington Books. Mutz, D.C. (2006) Hearing the Other Side: Deliberative versus Participatory Democracy, Cambridge: Cambridge University Press. Peterson, J. (1997) States, Societies and the European Union, West European Politics, 20(4): 1-23. Radaelli, C. (2000) Policy Transfer in the European Union: Institutional Isomorphism as a Source of Legitimacy, Governance, 13(1): 25-43. 27

Rich, A. (2004) Think Tanks, Public Policy, and the Politics of Expertise, Cambridge: Cambridge University Press. Richardson, J. (ed) (1982) Policy Styles in Western Europe, London: Allen & Unwin. Risse-Kappen, T. (1996) Exploring the Nature of the Beast: International Relations Theory and Comparative Policy Analysis Meet the European Union, Journal of Common Market Studies 34(1): 53-80. Sabatier, P.A. (1998) The Advocacy Coalition Framework: Revisions and Relevance for Europe, Journal of European Public Policy, 5(1): 98-130. Sabatier, P.A. and Jenkins-Smith, H.C. (1999) The Advocacy Coalition Framework: An Assessment in P.A. Sabatier (ed), Theories of the Policy Process, Boulder: Westview Press. Sabatier, P.A. and Zafonte M.A. (2001) Policy Knowledge: Advocacy Organizations in N.J. Smelser and P.B. Baltes (eds), International Encyclopedia of the Social & Behavioral Sciences, Volume 17, Amsterdam: Elsevier. Scharpf, F.W. (1994) Games Real Actors Could Play: Positive and Negative Coordination in Embedded Negotiations, Journal of Theoretical Politics, 6: 27-53. 28

Schneider, A. and Ingram, H. (2007) Ways of Knowing: Implications for Public Policy, Paper prepared for the annual meeting of the American Political Science Association, Chicago. Schön, D.A. and Rein M. (1994) Frame Reflection: Toward the Resolution of Intractable Policy Controversy, New York: Basic Books. Skogstad, G. (2003) Legitimacy and/or policy effectiveness?: network governance and GMO regulation in the European Union, Journal of European Public Policy, 10: 321-338. Streeck, W. (1992) From National Corporatism to Transnational Pluralism: European Interest Politics and the Single Market in T. Treu (ed), Participation in Public Policy- Making: The Role of Trade Unions and Employers Associations, Berlin: Walter de Gruyter. Verdun, A. (1999) The Role of the Delors Committee in the Creation of EMU: An Epistemic Community?, Journal of European Public Policy, 6(2): 308-328. Weiss, C.H. (1989) Congressional Committees as Users of Analysis Journal of Policy Analysis and Management, 8(3): 411-431. 29

Wessels, W. (1998) Comitology: Fusion in Action. Politico-Administrative Trends in the EU System, Journal of European Public Policy, 5(2): 209-234. 30

Table 1: Distribution of Respondents Subsystems Agri-Food 153 Human Genetics 117 Governmental 77 Actor Types Non-Governmental 77 Independent Experts 107 Other 9 Continents North America 197 Europe 73 Total 270 31

Table 2: Results for Learning intensity and Policy Transfers Variables Learning Intensity (Tobit Regression) Policy Transfer (Logistic Regression) Subsystem -11.02 (9.03) -2.30*** (0.44) Continent -10.28 (11.53) -0.24 (0.48) Continent*Subsystem -2.14 (16.80) -1.10 (1.18) Non-governmental actors 15.34* (8.71) -0.29 (0.43) Governmental actors 3.27 (9.57) -0.25 (0.49) Constant -2.94 (10.69) -3.29** (1.02) N 233 232 Pseudo R2 0.00 0.09 *** p 0.01, ** p 0.05, * p 0.1 Standard errors are indicated into parentheses. 32