Origins of the EU 51 cross-cutting groups of people in similar material positions.This does not mean that ideas have stronger causal effects where they cross-cut groups;ideas shared within powerful groups may be equally (or more)significant for important outcomes.It simply means that ideas'autonomous effects should be particularly demonstrable in these cases. To see this assertion in greater detail,consider how it responds to the"how much" problem.We could isolate ideas precisely if we found an extremely close compar- ison,contrasting actors in near-identical places in the objective world to highlight the purely subjective variations in their behavior.Such comparisons are available at the individual level,within groups.Close organizational peers share positions in the objective world;comparing their views of their groups'interests can separate variation in their ideas from variation in objective pressures.Take two French diplomats,with similar social backgrounds and party sympathies,in the same office of the foreign ministry in 1950.One insists on French interests in a new "suprana- tional"Franco-German federation;the other sees French interests in policies based on an informal partnership with Britain.These similarly placed individuals face all the same objective pressures but seem to interpret them differently.If other similarly placed individuals across France also disagree (displaying a pattern that cross-cuts parties,bureaucracies,economic groups),and if these debates persist through important decisions (showing they were not just an initial vetting of options),we can conclude that objective pressures at these levels leave French interests indeterminate across these alternatives.If,in addition,we have the kind of"interpretive"evidence typically offered by ideational accounts-actors consistently say and write that they believe certain things,and that their peers think differently-we have strong evidence that ideas alone are causing individual variation across that range.7 This conclusion does not rule out that historical or psychological factors other than simple"ideas"(like past socialization,psychological dispositions,or individual experiences)may have led to this pattern of debate.But as long as we cannot trace current patterns of mobilization to current objective conditions (and can trace related patterns of rhetoric),we have evidence that subjective factors are currently influencing action.Psychological or historical factors may help explain why actors came to hold certain ideas,but neither factor invalidates the claim that subjective ideas are now causing variation in behavior.In other words,explaining the distribution of ideas as dependent variables is legitimately separate from showing their presence and effects as independent variables.s Far from exaggerating the range of purely subjective variation in individuals' views,this approach should minimize interpretive biases.Unlike in most ideational arguments,it is the actors,not the observer,who define the range across which ideas matter (see Figure 1).One actor wants to pursue strategy X;one of her close 17.Provided there are no clear individual-level incentives to disagree,as in career competition.I see no pattern of such incentives in my case.Disincentives to interal disagreement seem more common in most organizations. 18.See Kingdon 1984.77:Kier 1997.148:and Kowert and Legro 1996.469
cross-cutting groups of people in similar material positions. This does not mean that ideas have stronger causal effects where they cross-cut groups; ideas shared within powerful groups may be equally (or more) significant for important outcomes. It simply means that ideas’ autonomous effects should be particularly demonstrable in these cases. To see this assertion in greater detail, consider how it responds to the “how much” problem. We could isolate ideas precisely if we found an extremely close comparison, contrasting actors in near-identical places in the objective world to highlight the purely subjective variations in their behavior. Such comparisons are available at the individual level, within groups. Close organizational peers share positions in the objective world; comparing their views of their groups’ interests can separate variation in their ideas from variation in objective pressures. Take two French diplomats, with similar social backgrounds and party sympathies, in the same office of the foreign ministry in 1950. One insists on French interests in a new “supranational” Franco-German federation; the other sees French interests in policies based on an informal partnership with Britain. These similarly placed individuals face all the same objective pressures but seem to interpret them differently. If other similarly placed individuals across France also disagree (displaying a pattern that cross-cuts parties, bureaucracies, economic groups), and if these debates persist through important decisions (showing they were not just an initial vetting of options), we can conclude that objective pressures at these levels leave French interests indeterminate across these alternatives. If, in addition, we have the kind of “interpretive” evidence typically offered by ideational accounts—actors consistently say and write that they believe certain things, and that their peers think differently—we have strong evidence that ideas alone are causing individual variation across that range.17 This conclusion does not rule out that historical or psychological factors other than simple “ideas” (like past socialization, psychological dispositions, or individual experiences) may have led to this pattern of debate. But as long as we cannot trace current patterns of mobilization to current objective conditions (and can trace related patterns of rhetoric), we have evidence that subjective factors are currently influencing action. Psychological or historical factors may help explain why actors came to hold certain ideas, but neither factor invalidates the claim that subjective ideas are now causing variation in behavior. In other words, explaining the distribution of ideas as dependent variables is legitimately separate from showing their presence and effects as independent variables.18 Far from exaggerating the range of purely subjective variation in individuals’ views, this approach should minimize interpretive biases. Unlike in most ideational arguments, it is the actors, not the observer, who define the range across which ideas matter (see Figure 1). One actor wants to pursue strategy X; one of her close 17. Provided there are no clear individual-level incentives to disagree, as in career competition. I see no pattern of such incentives in my case. Disincentives to internal disagreement seem more common in most organizations. 18. See Kingdon 1984, 77; Kier 1997, 148; and Kowert and Legro 1996, 469. Origins of the EU 51
52 International Organization Ideas as sole cause Leads to outcome A Leads to outcome Strategy X Strategy Y Actor I Actor2 (in same objective position) FIGURE 1.How much cross-cutting ideas matter peers,in the same objective position,wants to pursue strategy Y.Given solid evidence of differing preferences,and careful verification that they face identical objective constraints,this control for objective causes is as free from bias as qualitative observations can be. Furthermore,this method may tend to underestimate the overall impact of ideas on individuals'views.Open disagreements may well be narrow debates within a consensus built on other ideas.19 Nor may individuals voice the full range and strength of their views in internal debates,especially in hierarchical organizations like goverments.Even when they do,obtaining information about internal dissent is often difficult.This approach highlights the effects of ideas only where ideas separate most demonstrably from objective factors,without making claims about their less visible impact. Leaders'Ideas and Entrepreneurial Coalition-Building Variation in individuals'ideas does not mean ideas matter in government strategies. Whatever individuals debate,objective pressures in coalition building may ulti- mately impose one view on policy choices.Under certain conditions,however,they may not.A massive game-theoretic literature notes that when democratic groups face three or more alternatives on a given issue,or a decision involving two or more issues,multiple outcomes may be able to attract majority support.In the extreme situation of a pure"Condorcet paradox,"where even thirds of the group rank three alternatives in incompatible ways,no option spontaneously garners a majority.The result is that actors with agenda-setting authority can pass any option.20 Even with less perfectly divided preferences,bargaining over multiple issues can introduce similar flexibility.21 Leaders may use payoffs across issues to make winners out of 19.For example,French elites in the 1950s may share ideas about the state,democracy,capitalism, and so on that limit their debate over European policies. 20.Given options x,y.z,even thirds have rank-ordered preferences .ryz,yzx,and zry.See Arrow 1951; and Riker 1980.For a recent overview,see Nurmi 1999. 21.See Kramer 1973:and McKelvey and Wendell 1976
peers, in the same objective position, wants to pursue strategy Y. Given solid evidence of differing preferences, and careful verification that they face identical objective constraints, this control for objective causes is as free from bias as qualitative observations can be. Furthermore, this method may tend to underestimate the overall impact of ideas on individuals’ views. Open disagreements may well be narrow debates within a consensus built on other ideas.19 Nor may individuals voice the full range and strength of their views in internal debates, especially in hierarchical organizations like governments. Even when they do, obtaining information about internal dissent is often difficult. This approach highlights the effects of ideas only where ideas separate most demonstrably from objective factors, without making claims about their less visible impact. Leaders’ Ideas and Entrepreneurial Coalition-Building Variation in individuals’ ideas does not mean ideas matter in government strategies. Whatever individuals debate, objective pressures in coalition building may ultimately impose one view on policy choices. Under certain conditions, however, they may not. A massive game-theoretic literature notes that when democratic groups face three or more alternatives on a given issue, or a decision involving two or more issues, multiple outcomes may be able to attract majority support. In the extreme situation of a pure “Condorcet paradox,” where even thirds of the group rank three alternatives in incompatible ways, no option spontaneously garners a majority. The result is that actors with agenda-setting authority can pass any option.20 Even with less perfectly divided preferences, bargaining over multiple issues can introduce similar flexibility.21 Leaders may use payoffs across issues to make winners out of 19. For example, French elites in the 1950s may share ideas about the state, democracy, capitalism, and so on that limit their debate over European policies. 20. Given options x, y, z, even thirds have rank-ordered preferences xyz, yzx, and zxy. See Arrow 1951; and Riker 1980. For a recent overview, see Nurmi 1999. 21. See Kramer 1973; and McKelvey and Wendell 1976. FIGURE 1. How much cross-cutting ideas matter 52 International Organization
Origins of the EU 53 positions that would lose in a single-issue vote.Especially in some combination of these conditions-three alternatives on one issue,and in multidimensional issue space-individual leaders may enjoy broad autonomy to build the majorities they prefer. There is nothing inherently ideational about such leadership.Game-theoretic analyses tend to picture leaders using this leverage for personal material gain.When connected to ideational variation at the individual level,however,these dynamics can tie variation in ideas alone to variation in strategies and outcomes.The cross-cutting debates described earlier not only highlight ideational variation at the individual level;they also create broad patterns of divided preferences and rich possibilities for issue linkages.Leaders may then enjoy opportunities to choose among"multiple equilibria"on the basis of their personal ideas.Showing that this happened requires three kinds of evidence:that individual ideas varied over distinct alternatives (as described earlier),that divided preferences or issue linkages made those alternatives viable options,and that leaders indeed asserted their personal views and used agenda-setting power and issue linkages to assemble coalitional support behind them.Given such evidence,we can say that leaders'ideas,as autonomous factors,caused one strategy across the range of active alternatives.By contrasting the outcome to the counterfactual consequences of alternatives,we can specify a range of variation in outcomes caused by variation in ideas alone. Less abstractly.how would this process look?As a new cross-cutting debate emerged,top leaders'views on that issue would disconnect from their previous coalitional support.Support for various strategies would be scattered across parties and other organizations.As long as a leader retained office and agenda-setting power due to support on other (non-cross-cutting)issues,the leader could build majorities for any of several choices on the cross-cutting issue.Note that this does not require that the cross-cutting issue be obviously less important than the non-cross-cutting issue.In democracies,voters may be less informed or concerned (or both)about certain debates than policymakers.This seems particularly common in the domain of foreign policy.22 Electoral coalition building and office holding can continue to operate along the "normal"issues resonant with voters,even though policymakers perceive a new debate as crucial.This disconnect can allow a leader to retain control of the agenda during a major cross-cutting debate.We would see the leader's choices contested by elites who shared the leader's main coalitional. party,bureaucratic,and economic positions,differing only in holding other ideas in the cross-cutting debate.But the leader's agenda-setting power,or use of pressure or payoffs on other issues,would decide the outcome. The next section argues that French policymaking in early European institution building illustrates this process.Across three strategies that were viable domesti- cally and internationally,leaders built majority support for their own ideas. 22.See Rosenau 1961:Margolis and Mauser 1989;and the annual Index to International Public Opinion
positions that would lose in a single-issue vote. Especially in some combination of these conditions—three alternatives on one issue, and in multidimensional issue space—individual leaders may enjoy broad autonomy to build the majorities they prefer. There is nothing inherently ideational about such leadership. Game-theoretic analyses tend to picture leaders using this leverage for personal material gain. When connected to ideational variation at the individual level, however, these dynamics can tie variation in ideas alone to variation in strategies and outcomes. The cross-cutting debates described earlier not only highlight ideational variation at the individual level; they also create broad patterns of divided preferences and rich possibilities for issue linkages. Leaders may then enjoy opportunities to choose among “multiple equilibria” on the basis of their personal ideas. Showing that this happened requires three kinds of evidence: that individual ideas varied over distinct alternatives (as described earlier), that divided preferences or issue linkages made those alternatives viable options, and that leaders indeed asserted their personal views and used agenda-setting power and issue linkages to assemble coalitional support behind them. Given such evidence, we can say that leaders’ ideas, as autonomous factors, caused one strategy across the range of active alternatives. By contrasting the outcome to the counterfactual consequences of alternatives, we can specify a range of variation in outcomes caused by variation in ideas alone. Less abstractly, how would this process look? As a new cross-cutting debate emerged, top leaders’ views on that issue would disconnect from their previous coalitional support. Support for various strategies would be scattered across parties and other organizations. As long as a leader retained office and agenda-setting power due to support on other (non-cross-cutting) issues, the leader could build majorities for any of several choices on the cross-cutting issue. Note that this does not require that the cross-cutting issue be obviously less important than the non–cross-cutting issue. In democracies, voters may be less informed or concerned (or both) about certain debates than policymakers. This seems particularly common in the domain of foreign policy.22 Electoral coalition building and office holding can continue to operate along the “normal” issues resonant with voters, even though policymakers perceive a new debate as crucial. This disconnect can allow a leader to retain control of the agenda during a major cross-cutting debate. We would see the leader’s choices contested by elites who shared the leader’s main coalitional, party, bureaucratic, and economic positions, differing only in holding other ideas in the cross-cutting debate. But the leader’s agenda-setting power, or use of pressure or payoffs on other issues, would decide the outcome. The next section argues that French policymaking in early European institution building illustrates this process. Across three strategies that were viable domestically and internationally, leaders built majority support for their own ideas. 22. See Rosenau 1961; Margolis and Mauser 1989; and the annual Index to International Public Opinion. Origins of the EU 53
54 International Organization French Ideas and the Origins of the EU Western Europe first developed uniquely strong international institutions in the 1950s.At the end of that decade,six governments committed to deep and broad cooperation in the EEC,including the delegation of considerable powers to"supra- national"agents in the EEC's executive commission and court. Experts agree that French choices were particularly important to the EEC's birth. All three of the major supranational projects of the 1950s-the ECSC,the failed European Defense Community (EDC),and the EEC-took shape because govern- ments that preferred other European projects conceded to the French.When the French proposed the ECSC in 1950,the British refused to join any supranational plan.German and Benelux leaders,though skeptical of supranationality and reluc- tant to proceed without Britain,eventually followed the French lead.23 When the French proposed the EDC as a similar framework for West German rearmament,all the other Europeans argued for a simpler plan within NATO.France's ECSC partners(Germany,Benelux,Italy)eventually agreed to negotiate EDC but quickly dropped it when the French tured against it in 1954.24 While the EEC itself was not a French proposal-originating with ultra-Europeanist Dutch foreign minister Jan Willem Beyen-it was partly crafted to appeal to the French.Beyen notwithstand- ing,most Benelux and German leaders preferred trade liberalization in the broader and institutionally weak format championed by the British.They hesitantly sup- ported Beyen's EEC plan because they feared the French would only accept liberalization within a"little Europe"with elaborate economic safeguards.25 In sum, no one contests that without French insistence on institutionally strong,geograph- ically limited institution building in the 1950s no such framework would have resulted. Two alternatives to my account,however,see this French position as strongly determined by nonideational factors.For structuralists,objective economic and geopolitical constraints selected French strategies.Historians like Milward and William Hitchcock argue that France needed a new basis for controlling and cooperating with a resurgent West Germany in 1950;the ECSC's supranational oversight of basic industries was the straightforward solution.26 Similar concerns about overseeing German military power pushed the French to the EDC.The incentives of this larger delegation of sovereignty were less clear,however,and the EDC died.But soon economic interests,transmitted through interest groups, refocused the French on an institutionally strong "little Europe."Weak French business feared British plans for broad trade liberalization,preferring a smaller and more managed trade pact.Growing farm surpluses led French farmers to insist that 23.See Kersten 1988:Milward 1988;Ludlow 1997,17:and Buhrer 1995 24.See Fursdon 1980:and de la Gorce 1979.91-159. 25.See Asbeek-Brusse 1997,157:Milward 1992.192:Griffiths 1990:and Bouwman 1995,141. 26.Milward emphasizes French economic needs for the ECSC:Hitchcock emphasizes geopolitics. See Milward 1984:and Hitchcock 1998
French Ideas and the Origins of the EU Western Europe first developed uniquely strong international institutions in the 1950s. At the end of that decade, six governments committed to deep and broad cooperation in the EEC, including the delegation of considerable powers to “supranational” agents in the EEC’s executive commission and court. Experts agree that French choices were particularly important to the EEC’s birth. All three of the major supranational projects of the 1950s—the ECSC, the failed European Defense Community (EDC), and the EEC—took shape because governments that preferred other European projects conceded to the French. When the French proposed the ECSC in 1950, the British refused to join any supranational plan. German and Benelux leaders, though skeptical of supranationality and reluctant to proceed without Britain, eventually followed the French lead.23 When the French proposed the EDC as a similar framework for West German rearmament, all the other Europeans argued for a simpler plan within NATO. France’s ECSC partners (Germany, Benelux, Italy) eventually agreed to negotiate EDC but quickly dropped it when the French turned against it in 1954.24 While the EEC itself was not a French proposal—originating with ultra-Europeanist Dutch foreign minister Jan Willem Beyen—it was partly crafted to appeal to the French. Beyen notwithstanding, most Benelux and German leaders preferred trade liberalization in the broader and institutionally weak format championed by the British. They hesitantly supported Beyen’s EEC plan because they feared the French would only accept liberalization within a “little Europe” with elaborate economic safeguards.25 In sum, no one contests that without French insistence on institutionally strong, geographically limited institution building in the 1950s no such framework would have resulted. Two alternatives to my account, however, see this French position as strongly determined by nonideational factors. For structuralists, objective economic and geopolitical constraints selected French strategies. Historians like Milward and William Hitchcock argue that France needed a new basis for controlling and cooperating with a resurgent West Germany in 1950; the ECSC’s supranational oversight of basic industries was the straightforward solution.26 Similar concerns about overseeing German military power pushed the French to the EDC. The incentives of this larger delegation of sovereignty were less clear, however, and the EDC died. But soon economic interests, transmitted through interest groups, refocused the French on an institutionally strong “little Europe.” Weak French business feared British plans for broad trade liberalization, preferring a smaller and more managed trade pact. Growing farm surpluses led French farmers to insist that 23. See Kersten 1988; Milward 1988; Ludlow 1997, 17; and Bu¨hrer 1995. 24. See Fursdon 1980; and de la Gorce 1979, 91–159. 25. See Asbeek-Brusse 1997, 157; Milward 1992, 192; Griffiths 1990; and Bouwman 1995, 141. 26. Milward emphasizes French economic needs for the ECSC; Hitchcock emphasizes geopolitics. See Milward 1984; and Hitchcock 1998. 54 International Organization
Origins of the EU 55 any deal include a sheltered agricultural regime.This further alienated the British (with their Commonwealth agricultural preferences).The EEC was the result:The Germans and Benelux agreed to liberalization largely on French terms.Thus,claims Moravcsik,we can trace French choices to clear majorities based on objective societal interests.27 For institutionalist pioneer Ernst Haas,structural pressures led the French to the ECSC,but institutional feedback led on to the EEC.He suggested that French interest groups and parties did not spontaneously perceive interests in extending the ECSC to the EDC or EEC.Only the ECSC's supranational agents led domestic groups to perceive such interests.Chief among them was Jean Monnet,the French bureaucrat who presided over the ECSC's"High Authority"from 1952 to 1955,and thereafter led a lobby called the "Action Committee for the United States of Europe."Without the initial creation of supranational agents in the ECSC,implied this "neofunctionalist"account,leaders in France (and elsewhere)were unlikely to pursue the EEC.French choices can be traced to a majority of sectoral associations and their party representatives,but only supranational(not national)leadership built that majority.28 I accept elements of both approaches but strongly revise their causal claims.The institutionalist story requires more blatant revision.Between the initial,narrow ECSC deal and the much more extensive EEC,literally no major French actors shifted their policy preferences in response to supranational lobbying.Nonetheless, once the broader EEC bargain was struck,we need an institutionalist logic of path-dependence to understand why it was consolidated into Europe's fundamental architecture in the 1960s.My qualification of the structuralist case is less direct but no less profound.Any explanation of European institution building must begin with the economic and geopolitical context.Yet structuralists overlook the immense political battle that took place over different interpretations of that context.As of 1950,otherwise-similar French elites debated three ideational "models"of their interests in Europe.The divide between“community,”“confederal,”and“tradi-- tional"views did not quite display the impasse of a Condorcet paradox,but the availability of many issue linkages made all three options domestically viable.2 Corresponding deals were viable in international bargaining."Pro-community" leaders who obtained power on other,disconnected issues used their agenda-setting authority to assemble majorities behind their personal ideas and to strike particular 27.Moravcsik 1998.86. 28.For similar arguments about the EEC's later development,see Fligstein and Mara-Drita 1996;and Sandholtz and Stone Sweet 1998. 29.Labeling“community"views S,for“supranational,.”“confederal'"views C,and“traditional" views T.French elites divided into preference orderings SCT.CTS.and TCS.In single-issue voting.this would produce a confederal majority.Only issue linkages made"community"policies a viable outcome
any deal include a sheltered agricultural regime. This further alienated the British (with their Commonwealth agricultural preferences). The EEC was the result: The Germans and Benelux agreed to liberalization largely on French terms. Thus, claims Moravcsik, we can trace French choices to clear majorities based on objective societal interests.27 For institutionalist pioneer Ernst Haas, structural pressures led the French to the ECSC, but institutional feedback led on to the EEC. He suggested that French interest groups and parties did not spontaneously perceive interests in extending the ECSC to the EDC or EEC. Only the ECSC’s supranational agents led domestic groups to perceive such interests. Chief among them was Jean Monnet, the French bureaucrat who presided over the ECSC’s “High Authority” from 1952 to 1955, and thereafter led a lobby called the “Action Committee for the United States of Europe.” Without the initial creation of supranational agents in the ECSC, implied this “neofunctionalist” account, leaders in France (and elsewhere) were unlikely to pursue the EEC. French choices can be traced to a majority of sectoral associations and their party representatives, but only supranational (not national) leadership built that majority.28 I accept elements of both approaches but strongly revise their causal claims. The institutionalist story requires more blatant revision. Between the initial, narrow ECSC deal and the much more extensive EEC, literally no major French actors shifted their policy preferences in response to supranational lobbying. Nonetheless, once the broader EEC bargain was struck, we need an institutionalist logic of path-dependence to understand why it was consolidated into Europe’s fundamental architecture in the 1960s. My qualification of the structuralist case is less direct but no less profound. Any explanation of European institution building must begin with the economic and geopolitical context. Yet structuralists overlook the immense political battle that took place over different interpretations of that context. As of 1950, otherwise-similar French elites debated three ideational “models” of their interests in Europe. The divide between “community,” “confederal,” and “traditional” views did not quite display the impasse of a Condorcet paradox, but the availability of many issue linkages made all three options domestically viable.29 Corresponding deals were viable in international bargaining. “Pro-community” leaders who obtained power on other, disconnected issues used their agenda-setting authority to assemble majorities behind their personal ideas and to strike particular 27. Moravcsik 1998, 86. 28. For similar arguments about the EEC’s later development, see Fligstein and Mara-Drita 1996; and Sandholtz and Stone Sweet 1998. 29. Labeling “community” views S, for “supranational,” “confederal” views C, and “traditional” views T, French elites divided into preference orderings SCT, CTS, and TCS. In single-issue voting, this would produce a confederal majority. Only issue linkages made “community” policies a viable outcome. Origins of the EU 55