It is the cache of It is a snapshot of the page. The current page could have changed in the meantime.
Tip: To quickly find your search term on this page, press Ctrl+F or ⌘-F (Mac) and use the find bar.

EIoP: Text 1998-006: Full text
Institutionalism and Commission’s Executive Discretion: an Empirical Analysis(*)
Fabio Franchino
European Integration online Papers (EIoP) Vol. 2 (1998) N° 6;
Date of Publication in EIoP: 24.7.1998
| Abstract | Back to homepage | PDF | PS |
| This paper's comments page | Send your comment! to this paper |



I. Introduction

In their Pathologies of Rational Choice Theory (1994), Green and Shapiro see the empirical under-performance of rational choice as a ubiquitous weakness of this method of inquiry. They, sarcastically but correctly, observe that ‘the empirical contributions of rational choice theory … are few, far between, and consistently more modest than the combination of mystique and methodological fanfare surrounding the rational choice movement would lead one to suspect’ (Green and Shapiro, 1994: 179). This article takes up their challenge, testing the propositions I have generated in a rational choice model developed in Franchino (1998) and comparing their explanatory power with predictions based on more sociological premises.

In Franchino (1998), I have used Bayesian game theory to dissect the determinants of Commission’s executive discretion and, especially, the conditions under which these factors affect its autonomy in implementing Community legislation. Three key variables, information asymmetry, legislative procedure and distribution of preferences, seem to be at the core of the delegation of policymaking authority, in secondary legislation, from Member States to the Commission. For reasons explained below, I will add a fourth independent variable, that is legislative instrument.

This article compares the explanatory power of three models that operationalize, with different degree of rationalist assumptions, those propositions. The rationalist model is a straightforward application of the Bayesian game. It assumes that only formal procedures and instruments, information asymmetry and preferences distribution explain the degree of executive discretion of the Commission. A second model, named ‘diluted rationalist’, introduces informal rules in one policy area where there is ample evidence that informal procedures, or norms, affect policy outcomes. Finally, the ‘socio-rationalist’ model includes variables such as opinions of the European Parliament and the Economic and Social Committee as explanatory factors. The article also assesses the role of the Parliament and tests the correlation between the degree of Commission’s executive discretion and Member States’ ex-post control as it has been suggested in Franchino (1998) and in works on American institutions.

This article will show that the Commission enjoys more autonomy when the policy issue is highly technical and implementation has to be carried out mainly at Community level. However, if the act delegating policy making functions needs (formal or informal) unanimous approval in the Council of Ministers, the Commission will enjoy less autonomy than in the case of qualified majority. Moreover, more delegation of powers is accompanied by more stringent ex-post procedural control.

The article is organized in five parts. First, I discuss the issue of delegation in the European Community and the method that I have used to operationalize executive discretion. Second, I describe the hypotheses I have put forward in Franchino (1998), that is the four core independent variables and the conditions under which they have explanatory power. This section discusses also their operationalisation. Third, after a brief comparison of the theoretical premises of rational choice and sociological institutionalism, I test the explanatory power of the rationalist and diluted rationalist model. The statistical analysis continues in the last two parts of the article where I evaluate the role of the Parliament, the socio-rationalist model and the correlation between discretion and ex-post control. The article concludes with comments on the explanatory power of each variable and on the strengths and limits of rational choice theorizing.


II. The dependent variable: executive discretion

A. Delegation in the European Community

The delegation of policy making powers to the Community and the subsequent control of such powers are dynamics at the core of the European integration process. In Regulating Europe (1996) Giandomenico Majone dedicates an entire chapter to the issue of regulatory delegation and Pierson’s (1996) argument of unintended consequences leading to integrationist outcomes is a function of the type of and the control over delegated policies.

Borrowing from Tsebelis’s (1990) idea, there are two nested games of delegation in the Community: Treaty and secondary law delegation. In the first game, policies are delegated to the Community institutions, in the second one, policy making authority might (or might not) be delegated to the Commission itself. The contributions of Majone and Pierson, and others including Moravcsik’s (1993) liberal intergovernmentalism and Pollack’s (1997) functionalist analysis of delegation, are primarily concerned with the first game. I will focus on the activities delegated by EC secondary legislation.

These games differ in at least three important aspects. First, the institutions conferred of delegated powers vary. In Treaty delegation, the Council of Ministers, the Commission and the European Court of Justice are empowered of execution, proposal and judicial review respectively, with the added option of excluding an institution from performing these functions such as in the second and third pillar of the Maastricht Treaty(1). In secondary legislation, the addressees of delegation are primarily the Commission and Member States with the few exceptions of European agencies and standardization bodies (e.g. the European Environmental Agency, CEN and CENELEC). Second, the politics of delegation differs. In Intergovernmental Conferences and other similar arenas the debate is more on whether to have a common policy rather than on the substantive details of the policy (Edwards and Pijpers, 1997). Non cooperative bargaining theory provides the underlining logic (Moravcsik, 1993; Schneider and Cederman, 1994; and on bargaining theory Raiffa, 1982) and procedural rules of Treaty reform affect outcomes. Side payments such as the EU regional policy is a classical mechanism of Community task expansion in this arena (Pollack, 1994). In secondary legislation, Community institutions coordinate their strategies through the binding procedural and substantial boundaries of the Treaty. Enforcement of such agreement is not an issue, although enforcement of specific legislation is. From this point of view, cooperative games(2) underline the policy making process and legislative procedures have independent explanatory power over policy outcomes (Crombez, 1996; Tsebelis, 1994). Further, issue specific regulatory spillovers are the most common mechanisms of task expansion. Finally, the technical complexity increases in secondary legislation. This is in fact one of the reasons of having institutions with proposal and legislative powers in the first place (Pollack, 1997).

My interest is on the activities delegated in EC secondary legislation, at the so-called meso-level, not on those delegated in the Treaty. Given the nested nature of this game, Treaty delegation is certainly a precondition of secondary law delegation because the Community cannot legislate without a Treaty base. There are however important differences that allow us to treat the two games, both theoretically and methodologically, in a separate way. Moreover, the functions delegated to the Commission by the Treaty itself are not a good predictor of those delegated in secondary legislation (see note 26).

B. A classification of activities delegated in secondary legislation

The first hurdle to jump is to classify the type of activities that Member States and, eventually, the Parliament delegate to the Commission. They fall into eight categories that are based on Keohane’s (1984) functional regime theory and on the literature on the socio-economic activities of government. According to regime theory, states establish international institutions to lower transaction costs of international cooperation, thereby overcoming collective action problems. As a bare minimum, international institutions act as a secretariat that circulates information and ensures coordination amongst states (categories 1-3). If this is not sufficient to monitor compliance and detect transgressors because of the complexity or incompleteness of treaty obligations, supranational agents might be asked to actively oversee principals’ behavior and give unbiased recommendations (categories 4-5). However, regime theory can only go that far to explain the activities delegated to the Commission. The Community shows many similarities to a domestic political system. The supremacy and direct effect of Community legislation, the abundance of Brussels-based interest groups and an institutional structure and a policy process that are gradually developing domestic-like features invite us to use the analytical lenses of comparative politics.(3) In this system of governance, the Commission, as the Community bureaucracy and the institution with ‘executive vocation’ (Lenaerts, 1991: 30), carries out the traditional regulatory, administrative (categories 6-7) and redistributive (category 8) functions.(4)


To sum up, the categories are 1) receiving information from Member States and other Institutions, 2) providing information to Member States and other Institutions, 3) ensuring coordination and consultation amongst Member States, 4) giving opinions and recommendations, 5) monitoring, examining, reviewing and investigating 6) taking decisions, 7) administering, implementing and regulating, 8) managing and financing programmes.

Below I describe in more detail these activities and provide examples from actual legislation.

1-2) Receiving/providing information from/to Member States and other Institutions.

The first two cases concern the exchange of information between the Commission and other institutions or bodies. The Commission can be simply the depositary of information when Member States and other institutions must notify it of, for example, the adoption of specific national laws or administrative acts. Conversely, the Commission is the provider of information when it must give notice of its activities to the other Community institutions or, in some cases, to the public at large. The large majority of Community legislation provides for this exchange of information with two most notable exceptions, that is agriculture legislation fixing guide-prices and intervention prices for specific products and commercial policy acts suspending import levies or defining details of tariff quotas. General publication requirements established by the Treaty suffice in these cases.

3) Ensuring coordination and consultation amongst Member States

A third activity concerns consultation and coordination. This is when there is a general call for the Commission to be consulted about certain administrative acts taken by national administrations or for the Commission to consult interest groups and other bodies. A typical example is when a regulation opens and provides for the administration of tariff quotas. Here, a general call for coordination between the Commission and national administrations and for the Commission to favor cooperation amongst Member States is usual.

4) Giving opinions and recommendations

A fourth type of delegated activity is when the Commission is formally asked to give an opinion or recommendation on a certain matter. For example, some directives on the approximation of laws ask for the Commission’s opinion when a Member State temporarily suspends or restricts their application for health and safety reasons. In this case the Commission produces a formal, but not legally binding, act and national courts must refer to the European Court of Justice questions concerning its interpretation or validity (Weatherill and Beaumont, 1995: 139).

5) Monitoring, examining, reviewing and investigating

This set of functions can be classifiable under the general label of monitoring. The Commission is asked to examine, control, investigate, or review certain Community policies. This is common in environmental legislation but also in transport, agriculture, commercial policy and approximation of laws. An example is Regulation 1602/92 temporarily derogating from implementation of some anti-dumping measures. The Commission is requested to monitor and periodically review the import into the Canary Islands of products that are exempted from anti-dumping duties. A second example is Directive 271/91 on urban waste-water treatment where the Commission has to monitor Member States compliance and examine whether the technical difficulties encountered in the implementation warrant an extension of the compliance period.

6) Taking decisions

When the legislation delegates power to the Commission to take legally binding decisions on certain matters, this type of activity falls into the sixth category. It includes decisions on safeguard measures in commercial policy and decisions to reduce or suspend payments of financial support in environmental or transport programmes. It also includes authorizations to Member States to perform certain activities such as the use of statistical units for the observation and analysis of the production system in the Community (Regulation 696/93).

7) Administering, implementing and regulating

The penultimate type of activity is when the Commission is clearly invited to adopt measures not only for the administrative management and implementation of a certain policy but also for its regulation and definition of principles and criteria. Legislative acts that delegate this type of activity include Council regulations opening and providing for the administration of tariff quotas under Articles 28 or 113 (EC) and for the establishment of support systems for farmers. Power to make regulations are conferred on the Commission especially in Council directives on the freedom of movement and approximation of laws, but also in regulations on the import of, for instance, wild species.


8) Managing and financing programmes

Finally, the last type of delegated function is when the Commission is granted the power to manage the allocation of financial resources of action programmes. The Commission is first asked to select projects pursuing the objectives of the programme. These can be, for instance, the protection of the environment (e.g. in coastal areas and coastal waters, Regulation 3908/91) or the development of transport infrastructure (Regulation 3359/90). Then, it has to decide the form of financing (e.g. capital grants, financial contribution, interest rebates or repayable advances). There is also an important set of legislation establishing support systems for farmers that delegates this type of activity.

An index of executive discretion is then created from this classification. A sample of directives and regulations (see Appendix) has been checked against this set of activities and a value of 1 has been assigned to the act for each function that is clearly conferred on the Commission(5). The sum of these values is a measure of the degree of discretion the Commission enjoys in implementing the legislation. This index ranges from minimum of zero (no discretion) to maximum of eight (extensive discretion). In other words, I have used the observable variable, number of delegated activities, to measure the latent variable, degree of executive discretion.(6)

III. The delegation trade-off, the independent variables and the initial model

Whatever the rationale for the conferral of powers to an institution, delegation entails risks in the form of agent’s shirking or slippage (Kiewiet and McCubbins, 1991; Milgrom and Roberts, 1992). The agent can exploit information asymmetries and institutional constraints to pursue its interest at the expense of principals. In Franchino (1998), I have designed an extensive form game to analyze this trade off between efficiency gains and agency losses in the European Community(7). Here principals (i.e. Member States and the Parliament) delegate a certain amount of executive discretion to the agent (i.e. the Commission) and all actors are uninformed about future contigencies. The solution to the game consists in finding out the optimal amount of discretion that minimizes losses to the principals and that gives enough latitude to the Commission to deal with unexpected events. Comparative statics have then been carried out to analyze the impact on executive discretion of 1) different procedures for adoption of secondary legislation, 2) changes in the preferences of the agent and the principals and 3) changes in the degree of asymmetry of information.

There are then three variables that independently affect Commission’s executive discretion. These are information asymmetry, legislative procedure and distribution of preferences. Below I will discuss the relation they bear upon executive discretion and their operationalisation. I will also compare these propositions with those of similar formal works carried out on American institutions. I will then add a fourth independent variable, legislative instrument, that I consider relevant but has not been formally discussed.


A. Uncertainty and information asymmetry

All things equal, in an environment where uncertainty about the choice of the best policy alternative is high because of the complexity of the issue or lack of reliable information about future events, the scope of delegated authority is broader. As McCubbins and Page (1987: 417) point out, ‘[w]ith little or no information with which to evaluate the possible alternatives, and with conceivably large political risks associated with uncertain choices, legislators would prefer to delegate an increasingly large domain of alternative regulatory targets to the agency’. Epstein and O’Halloran (1994) and, in a similar way but applied to the EC, I (1998: 16-17) make this proposition dependent on the distribution of principals’ and agent’s preferences and on the symmetric or asymmetric change of uncertainty across the policy spectrum.

The main message of these works is that uncertainty affects the distribution of information at the expense of legislators who are not only unable to discern an optimum policy but probably are also unaware beforehand of what their interests will ultimately be (McCubbins and Page, 1987: 417). The higher the uncertainty, the larger the information asymmetry the broader the discretion of the delegated authority is(8).

Uncertainty is more than policy specific; it is actually issue specific. In commercial policy, a law that alters an import duty in the common customs tariff regulates a relatively straightforward matter compared to an act that designs the anti-dumping regime of the Community. Similarly, in agriculture, the fixing of intervention prices for some products is certainly less complex that establishing guidance funds and support systems for farmers. I would argue that an acceptable way to quantitatively operationalize information asymmetry is using the word count of the specific legislation excluding annexes, tables and the introductory statements(9). It seems to me that, generally, the complexity of the issue affects the length of the legislative act and that the word count assures objective cross-policy and cross-issue comparability. For instance, the word count of relatively simple legislation fixing duties, prices and import quotas amounts to less than one hundred words. Establishing a system of quota administration requires about five hundred words. Highly technical directives on environmental policies and approximation of laws easily extend over a thousand words.

B. Legislative procedures

Since Gordon Tullock’s (1981) question on the apparent inconsistency between the theoretical results on the instability of equilibrium outcomes (McKelvey, 1976) and the empirical existence of many stable policies, political scientists have gone a long way to explain the role of rules and procedures in determining outcomes(10). But the rationalist camp of EC students has only recently approached this topic (Crombez, 1996; Garrett and Tsebelis, 1996; Scharpf, 1988; Steunenberg, 1994; Tsebelis, 1994). Following this theoretical stream, the formal model in Franchino (1998) has outlined the conditions under which legislative procedures affect executive discretion. In particular, Proposition 1i posits that, when the status quo ante discretion is small, ‘the equilibrium discretion in the Assent procedure is always smaller than the equilibrium in the other two procedures [i.e. Consultation and Cooperation procedure]’(Franchino, 1998: 10). Put simply, the legislative procedure has an independent causal effect on the extent of executive discretion delegated to the Commission only in non amending legislation. In case of amending directives or regulations, the legislative procedure loses explanatory power.

In this section, I delineate the independent role that decision rules have on non amending secondary legislation. I first look at the Council of Ministers alone, I will then move to the European Parliament.

Ceteris paribus, from Proposition 1i (Franchino, 1998: 10), the executive discretion of the Commission is likely to be larger under qualified majority vote rather than unanimity. The Commission can make full use of its agenda setting powers by targeting its proposal to gain support from the pivotal Government in the qualified majority vote that prefers the largest amount of executive discretion.(11) In unanimity, the Commission is constrained by the ideal discretion of the Government whose preference is farthest away from its own. In non amending legislation where there is no status quo ante discretion, the equilibrium discretion will be smaller than in qualified majority.(12) I have used a dichotomous variable taking the value of 0 for unanimity and 1 for qualified majority vote.

The Parliament affects outcomes under the Assent and Co-operation procedures where a majority of its members must approve the legislative measure. The EEC Treaty, as amended by the Single European Act, provided for only two cases where Assent was necessary, that is accession and uniform procedure for elections to the Parliament (Artt.138.3 and 237 respectively). Since there are no legislative acts approved under Assent that satisfy the selection criteria of the data set, I will disregard this procedure.


On the contrary, Co-operation applied to a wide set of policies ranging from freedom of workers’ movement and of establishment to law approximation. Briefly speaking, a vote of the Parliament is needed after the Council adopts a common position. Rejection obliges the Council to adopt the act by unanimity while amendments have to be reconsidered by the Commission and re-voted by the Council. As Tsebelis (1994) puts it, the Parliament gains a conditional agenda setting power.(13) For our purposes, if Parliament’s ideal discretion is greater than the discretion preferred by the pivotal Government in the qualified majority vote, the outcome in non amending legislation remains unchanged. If the opposite applies, the equilibrium discretion is the Parliament’s ideal discretion(14). Thus, a variable on the role of this institution effectively measures whether Commission’s and Parliament’s preferences are similar. If they are, the variable should not have explanatory power; if they are not, it should have a negative and statistically relevant role. I will explain below in more detail the way I have operationalized and tested the role of the Parliament.

C. Distribution of preferences

If a principal delegates authority to an agent with similar preferences, discretion will be complete because the agent’s shirking is unlikely (or better, irrational). On the contrary, conflicting interests lead the principal to reduce the agent’s room of maneuver using control mechanisms such as administrative and oversight procedures (McCubbins and Page, 1987; Pollack, 1997). Formal models (Epstein and O'Halloran, 1994; Gilligan and Krehbiel, 1989; O'Halloran, 1994) generally confirm this finding and, in the American literature, the debate revolves around the efficacy of such control mechanisms (Kiewiet and McCubbins, 1991; McCubbins and Schwartz, 1984; Moe, 1987; Weingast and Moran, 1983).

In the EC framework, the agency relationship is complicated by two factors. First, the Commission enjoys a monopoly proposal power; second, there are multiple principals (Member States and, occasionally, the Parliament). As I have formally demonstrated in Franchino (1998: 15, 17-18), the proposal power of the Commission does not affect the central finding. All else equal, a convergence of principals’ preferences toward the Commission’s leads to more discretion being delegated to the agent. To be more precise, it is the distance between the preferences of the pivotal actor in a legislative procedure and those of the Commission that is positively related to discretion. The fact that there are many principals slightly complicates the picture instead. An increase in conflict between them decreases discretion because the distance between the pivotal actor and the Commission is likely to be greater.(15)

Measuring this distance in all procedures, years and policy areas is a task that is seriously jeopardized by the lack of objective and comparable data across these three dimensions. Eurobarometer surveys of the Commission, for instance, do sometimes focus on public opinion attitudes toward certain policies such as agriculture, the single market or EMU. However, because of the different questions being asked, they are not comparable and are discontinued throughout the period under analysis. Opinions and statements issued by various institutions provide valuable information about the interinstitutional dynamics of legislative policymaking, but Member States’ officials seldom make their concern public and opinions are issued by less relevant institutions in the decision-making process such as the Economic and Social Committee (see more on opinions below).

Taken into account these difficulties, there is however a survey published annually in the Eurobarometer on the general public attitude toward the EC that can be used for our purposes. Citizens of all Member States are asked, generally twice a year, the following question: “Generally speaking, do you think that (your country’s) membership of the European Community (Common Market) is a good thing, a bad thing, or neither good nor bad?” (EC Commission, 1994). I have derived from this survey a measure of both convergence of governmental preferences toward Commission’s and interstate conflict. These are

formula 1

where supporti is the percentage of those who answered that EC membership is good for them in country i and year y, while wi is the voting weight as in article 148 (EEC). In the first variable, I imply that this support can be interpreted as a convergence toward Commission’s preferences. The higher the mean, the higher the support for EC level activities, the more policymaking functions are delegated to the Commission. The second variable captures the conflict between Member States. An increase in conflict tends to diminish the extent of executive discretion enjoyed by the Commission (but see below). The weights measure the relative importance of the countries in the decision making process. In other words, they measure the relative probability of each country to be the pivotal actor.(16)


Chart 1 provides descriptive statistics for these two variables in the time period under study.

Chart 1

Notice that conflict and convergence are uncorrelated. A t-test cannot rejects the null hypothesis of independence between the two variables.(17) In the years between 1987 and 1993, an increase in convergence could actually be associated to both a decrease and an increase in conflict. Popular support for the Single Market initiative rose considerably, but not homogeneously, across Member States up until 1992. Hence we have increased convergence of preferences but the same degree of interstate conflict. The permanence of such conflict was mainly due to the still rather limited response to the Single Market initiative in traditionally laggard states such as the United Kingdom and Denmark (EC Commission, 1994). On the contrary, the fall in support in 1992-3, probably due to the collapse of the European Monetary System, was relatively more homogeneous across Member States. In those two years we have a decrease of both convergence and conflict.

From Chart 1 we also see that the way conflict and convergence operate on executive discretion is contextual. Low intestate conflict has a positive impact of discretion only if there is a convergence toward Commission’s preferences. Member States can agree not to delegate policymaking functions to the Commission (this would mean low conflict and convergence). Similarly, high convergence increases executive discretion only if interstate conflict is low. This is because there is a smaller probability that the pivotal actor is very distant from the Commission.

I would argue that an acceptable way to measure these effects and to minimize problems of multicollinearity is to use the following index

formula 2

Here convergence is normalized by the degree of interstate conflict. This index captures the idea that if there is a general support for EC level activities and limited conflict, Member States are more likely to delegate policymaking powers to the Commission. While if convergence is high but not across all states, less will be delegated. Similarly, there could be harmonious relations (low conflict) but lack of support for common policies and hence limited delegation.

I have weighted this index to have it approximately distributed across a value of one thousand. The index had its lowest point at 906 in 1987 where there was high conflict and low convergence of preferences, it reached a value of approximately one thousand in 1990-1 where high conflict remained but preferences of Member States converged toward Commission’s. The peak (i.e. 1403) was in 1993 where, although both convergence and conflict declined, the former decreased by less that 10 percent of its 1990 value and the latter by more than 30 percent. This is because the fall in support in traditionally pro-European countries such as Spain and Belgium was not paralleled by a similar fall in more skeptical ones. Support in Denmark actually rose by almost 6 percent.

I acknowledge that this is a less than perfect measure of preferences and conflict. For instance, it disregards policy areas and assumes an efficient mechanism of transmission of preferences from citizens to state officials.(18) I can only urge researchers, especially in the rationalist camp, to develop statistics about Member States’ revealed preferences in a similar way as, for instance, the ADA liberal support scores of the US Senate.

D. Policy types and legislative instruments

The EC institution on which powers are conferred to implement Community legislation is traditionally the Commission with a limited facility for the Council, in specific cases, to reserve to itself implementing powers as from the third indent of article 145 (EEC). The extent of Commission involvement is however a function of the policy type. Responsibility in many areas is shared between the Commission and national administrations. Generally speaking, the two most important legislative instruments of the Community, regulations and directives, mirror the distribution of implementing competence between European and national authorities.


A regulation is directly applicable in its entirety in all Member States. It is normally used in policies where EC institutions are extensively involved while the functions of national administrators are minimal. Regulations establishing quotas, suspending customs duties, setting intervention prices in agricultural produces fall under this category. Regulations are also used when there is a clear-cut criterion for the division of powers between European and national levels such as in competition policy. The Commission intervenes when certain conditions are fulfilled. For instance Regulation 4064/89 sets the aggregate turnover above which the Commission can investigate the compatibility of mergers with articles 85-90 of the Treaty.

A directive instead is not directly applicable and allows Member States the choice of implementing form and methods in their national laws. Directives are binding on Member States as to the result to be achieved (article 189 EEC). They are common in areas of free movement of persons and services, approximation of laws and environmental policy. Examples include directives on pollutants emissions from combustion plants (88/609/EEC) and on the approximation of the laws relating to active implantable medical devices (90/385/EEC ).

Although there are some notable exceptions, regulations are used in types of policies where the Commission directly administers and sets rules, while directives are used in policies where national administrations perform these functions and the Commission supervises implementation. The legislative instrument, as a variable measuring this division of responsibility, has an independent effect on executive discretion. Regulations are for policies where discretionary powers are rather extensive while directives imply policies with functions delegated to national administrations at the expense of the Commission. A dummy variable taking the value of 0 for directives and 1 for regulations should hence have a relevant positive effect on executive discretion.

To sum up, the initial model with the expected signs is:

formula 3

The extent of executive discretion (DISCR) delegated to the Commission in non amending secondary legislation is a positive function of preference distribution (PREF), legislative procedure (PROC), legislative instrument (INST) and information asymmetry (INFO).

This model and the contending hypotheses have been tested on a stratified sample of non amending secondary legislation approved between the first of July 1987 and the first of November 1993, the dates on which the Single European Act and the Maastricht Treaty respectively came into effect. A search in the CELEX database has generated approximately a thousand directives and regulations that fit such criteria and, after a cross check in the EC Official Journal, the exact figure amounts to 1033 (see Appendix on population and sampling strategy). Chart 2 shows the legislative production in different policy areas and according to the three main legislative procedures: unanimity, qualified majority and Cooperation.

Chart 2

Unsurprisingly, the large majority of non amending legislation, slightly more than 80 percent, has been in the areas of customs union, agriculture and commercial policy. However, important legislation has also been produced in the areas of environment, transport and, especially, approximation of laws. Similarly, qualified majority (QM) has been the predominant procedural rule, more than 83 percent of the cases, followed by Cooperation, almost 11 percent, and unanimity, around 6 percent.


IV. The rationalist and diluted rationalist hypotheses: norms and behavior

Rational choice and sociological institutionalisms sharply differ in their interpretation of the way institutions affect behavior.(19) Rational choice sees actors with given preferences and beliefs pursuing objectives within a set of constraining factors (i.e. distribution of capabilities and information) and strategically interacting with other actors within a set of commonly known rules (i.e. institutions). Hence we have Member States that, for functional reasons and using various legislative procedures, delegate policy making authority to the Commission and create control mechanisms to limit agency losses arising from information asymmetries or institutional design (Kiewiet and McCubbins, 1991). What is problematic in rational choice is the “givens”. The formation of preferences and beliefs are rarely investigated. In this arena sociological institutionalism has more explanatory power because they focus on ideational, normative, cultural and cognitive factors that facilitate intersubjective communication and “constitute” actors’ preferences. This literature is gradually emerging in European studies with a special emphasis on constructivism and on the role of norms, ideas and culture.(20)

In fact , an important subset of the rational-sociological dialectic is on the role of norms. There are three standard rationalist treatments of norms (i.e. collective understanding of proper behavior)(Yee, 1997). First, norms are considered as given decision rules. Second, they affect behavior because of the sanctioning costs of their violation. Third, they are incorporated as a type of interest in the actor’s utility function. However, ‘[n]orms whose autonomous effects on behavior do not stem from (dis)utility-generating interests and sanctions cannot be simply “given”’ (Yee, 1997: 1021). Rationalists cannot explain the properties of those norms that have an independent effect on behavior and this is where the sociological tradition has made powerful contributions. Legro (1997: 34), for instance, has suggested that clarity, durability and acceptance amongst actors are the key criteria to gauge the robustness of norms in autonomously affecting behavior.

It is against this background that I have generated my hypotheses. Although the epistemological and ontological basis of this article is evidently rationalistic, my attempt is to test different degrees of rationality with special emphasis on decision rules and legislative procedures. This section compares the explanatory power of two hypotheses. The first one – rationalist - takes the model as it is laid out in the previous section with formal decision rules as one of the independent variables. The second hypothesis – diluted rationalist - acknowledges comments of analysts of the Common Agricultural Policy (CAP) that the perceived appropriate behavior (i.e. norm) of Member States in this policy area has been to adopt legislation using unanimity even when qualified majority was provided by the Treaty (see Fennel, 1987: 73; Keeler, 1996: 136; Patterson, 1997: 144; Peterson, 1989: 468; Runge and von Witzke, 1987; for an opposing view see Wallace, 1989: 200). CAP has also been seen as a joint decision system entailing unanimity by Scharpf (1988 : 251, 257) in his classic study on joint decision traps.

Unanimity in agriculture seems also to withstand Legro’s criteria of robustness more that in other EU policies. It is a clear rule, it is long-standing (since, at least, the Luxembourg Compromise of 1966) and there is a general acceptance of its use by Member States. Swinbank (1989: 309), for instance, observes that Governments have been generally supportive of unanimity in agriculture and objections to this rule have only been raised by Belgium and Italy. Further, the probability of simply threatening to invoke unanimity to protect a ‘vital’ national interest is a function of the extent of disagreement over the content of policy reform and of the magnitude of sunk costs (Pierson, 1996). Agriculture was a clear policy failure in the 80s and object to strong international pressures (Keeler, 1996; Woolcock and Hodges, 1996). Reforming it entailed substantial transaction costs that were asymmetrically distributed across the Community.(21) The adoption of qualified majority seemed even less likely. Eventually, the commissioner MacSharry managed to reform the CAP not because governments adopted qualified majority but because their preferences and strategies changed in a complex three level game of national, European and international negotiations (Patterson, 1997).


Analysis of the results

The standard procedure would be to run an ordinary least squared (OLS) regression and compare the statistical results of the two hypotheses. However, this type of parametric inference requires a set of assumptions to ensure that regression coefficients are BLUE (best linear and unbiased estimators)(22). More specifically for our case, OLS inferential statements assume that the random error in the model is normally distributed. If that was not the case, ‘our confidence intervals and hypothesis tests could have a greater than nominal probability of error. Bootstrapping may be a way of overcoming this problem’(Mooney and Duval, 1993: 55).

In our context, while the traditional assumptions of linearity, homoscedasticity and low collinearity are generally respected(23), the error structure of the model is not normal. A Jarque-Bera omnibus test for normality has rejected the null hypothesis that the distribution of residuals is normal. This is probably because the dependent variable is bounded by zero and has a bimodal distribution.(24) As suggested by Mooney and Duval, I have used bootstrapping to solve this problem (see Appendix).

The results are shown in Table 1. In the first column there are the mean bootstrapped values of the regression coefficients. The other columns show the endpoints of the confidence intervals of the null hypothesis calculated using different techniques (see Appendix). A coefficient outside these endpoint values is significantly different from zero at 95 level of confidence.

Table 1

The results show that the diluted rationalist hypothesis is the most accurate. It explains, on average, about 55 percent of the variation in executive discretion compared to a value of more that 48 percent of rationalist hypothesis. However, is this difference statistically relevant? What is the probability that the bootstrapped rationalist model has a R2 of 55 percent? As shown in Table 1, I can reject at 95 percent confidence level the null hypothesis that a R2 of 55 percent is generated by the rationalist model. The higher goodness of fit of the diluted rationalist model is statistically relevant.(25) Only the normal approximation interval does not allow me to reject the null hypothesis but this is probably more a sign of the inappropriateness of this method (see Appendix). It is also worth pointing out that 45 percent of the variance still remains unexplained thus inviting researchers to put forward better specified models.(26)

Results also indicate that information asymmetry (INFO) is consistently the most important variable exerting a strong influence on the extent of executive discretion. Ceteris paribus, an increase in length of five hundred words from an act suspending import levies to one setting up the administration of a tariff quotas leads to an increase of more than one point of the discretion index. Moving from tariff quota legislation (approximately 700 words) to acts on environmental policy or on the approximation of technical standards (approximately 3000 words) increases the discretion index by almost five points. The more technical the policy issue the more authority will be delegated to the Commission.

Legislative instrument (INST) is another important determinant of delegated authority. All else equal, the use of regulations rather than directives increases by one point the discretion index. This means that when the implementation of the policy has to be carried out at EU level, the Commission becomes the obvious candidate to which functions are conferred. When national bureaucracies administer the policy, the Commission is probably relegated to a supervisory role with less policy making activities.

The substantive and statistical significance of the index of preference distribution (PREF) tells us different stories. The regressor of this index is significantly different from zero at 95 percent confidence level according to almost all methods, the only exception being the normal approximation in the diluted rationalist model probably because of its too stringent assumptions (see Appendix). Its substantive significance is however rather limited. Recall that this variable should measure the distance between the pivotal player and the Commission in the different procedures, years and policy areas, and it has been operationalized with an index derived from an Eurobarometer survey. The higher the index, the shorter the distance and the more discretion is delegated to the Commission. The effect of this index on executive discretion is very small. An increase from its lowest value of 906 in 1987 to its highest of 1403 in 1993 barely increases the discretion index of one point (ceteris paribus).


Finally, Commission’s executive discretion is likely to be larger under qualified majority rather than under unanimity only if we take into account the informal rule of using unanimity in agriculture. The statistical significance of the legislative procedure (PROC) in the rationalist model is not conclusive. The coefficient is significant, but with the wrong sign, according to the percentile and bias corrected methods. It loses significance for the normal approximation and percentile-t methods. Since statisticians tend to prefer the bias corrected and percentile-t methods (see Appendix), we cannot reject the null hypothesis. Substantively, the use of unanimity or qualified majority has no effect on the discretion index. In the diluted rationalist model the legislative procedure is substantially and statistically significant. Ceteris paribus, the formal and informal use of unanimity decreases the index of discretion by one point. When unanimity is (formally or informally) needed to approve a Community act, less policy making functions are delegated to the Commission. The pivotal Government is the one with preferences farthest away from the Commission’s, less will be conferred upon the agent than in the case of qualified majority.

It is important to comment on this result. On one side, rationalist accounts on the effect of decision rules on policy outcomes are confirmed. Legally or normatively enforced rules affect the content of legislative acts. On the other side, it shows an important weakness of rational choice institutionalism. Such approach is in effect mute about which norms matter. It cannot tell us under which circumstances actors perceive as appropriate a certain behavior even if its violation can be interest based and is not sanctioned by enforcement mechanisms.(27)

These results also confirm that clear, long-standing and agreed upon norms have explanatory power. The critical reader may point out that the re-coding of legislative procedure (PROC) to account for informal behavior could have merely increased its variance and, consequently, its significance. I have re-coded in a similar way all the other policy areas in the sample, even though there is less evidence of long-standing and concordant use of unanimity. Legislative procedure is not substantively significant in these tests.

V. The role of the Parliament and of opinions

A. The Parliament in the Co-operation procedure

Does the Parliament affect the discretionary powers delegated to the Commission? In Franchino (1998: 11-12), I have set out the conditions under which the Parliament can have an effect on discretion. In non amending legislation approved under Co-operation procedure, the Parliament will have a substantial influence on discretion only if its preferences are more divergent from the Commission’s than the pivotal Government’s in the qualified majority vote. Put simply, it is when the Parliament has pivotal role in the procedure. This seems to be a rare occurrence (Fitzmaurice, 1988: 394), and in the sampled legislation approved under Co-operation I have found no clear evidence of Parliamentary opposition to delegation.(28) A variable on the role of this institution should be excluded from the model because it has no explanatory power.

It is however possible to argue from a sociological perspective that the simple use of this procedure has an effect on Member States’ strategies and that the Parliament can exercise a broadly defined agenda-setting influence (Earnshaw and Judge, 1996: 97-109), beyond the strict conditions set by Tsebelis (1994). In order to account for this effect, the role of the Parliament has been operationalized in two ways, both of which however fail to relevantly improve the explanatory power of the model.

A first operationalisation is to convert legislative procedure (PROC) into a multichotomous variable taking the value of 2 for legislation approved under Co-operation.(29) The results of this test are shown in the upper part of Table 2; they take as benchmark for comparison the diluted rationalist model because of its higher explanatory power. Unfortunately, the model explains a lower percentage of the variance of the dependent variable and we can only just reject the null hypothesis that it can be randomly generated from the diluted rationalist one.


The improvement in explanatory power of legislative instrument (INST) and information asymmetry (INFO) is generally at the expense of the importance of distribution of preferences (PREF) and legislative procedure (PROC) as collinearity diagnostics has already told us. Ceteris paribus, the use of regulations increases the discretion index by one and an half points and an increase of one thousand words leads to a rise of more than two points of the discretion index. Distribution of preferences is still substantively insignificant. Most importantly, a unit increase of the new variant of legislative procedure causes a rise of less than 0.8 of the discretion index (ceteris paribus). This value is lower than in the model without the Parliament. Before stating that this institution has a substantive effect on executive discretion, we need to run a second test because the conversion of legislative procedure into a multichotomous variable could have increased the measurement error of its regressor.

Table 2

The second operationalisation consists in adding a dummy Co-operation variable (PARL) to the diluted rationalist model. This new variable takes the value of 1 when the Parliament’s vote is needed to adopt legislation under the Co-operation procedure. The new model would become:

formula 4

Given the substantial similarity of Commission’s and Parliament’s preferences and the hypothesis about a general legislative influence of the Parliament, we should expect a positive influence of this variable on discretion. However, this addition is not statistically relevant. With an F(1,94) statistics of 1.28 we cannot reject the null hypothesis of significantly improved explanatory power. Moreover, the Co-operation variable has practically no substantive meaning. All else equal, it determines less than a fifth of a point of executive discretion (its mean bootstrapped regression coefficient is 0.19).

To sum up, the role of procedures and distribution of preferences on discretion as hypothesized in Franchino (1998) remains valid. Being no clear evidence of parliamentary opposition to delegation, variables on parliamentary role do not have explanatory power and models incorporating parliamentary preferences have lower coefficients of determination.(30) More specifically, these tests disconfirm the existence of a sociologically induced influence on Member States’ preferences on discretion generated by the more important role assigned to the Parliament in Co-operation. This does not however mean that the hypothesis on parliamentary effect on discretion is confirmed because we cannot state that a pivotal Parliament has an impact on discretion.

B. The impact of opinions

The Treaty extensively provides for the making of opinions by various institutions. Article 155 (EEC), for instance, confers a general power on the Commission to make opinions and recommendations whenever it deems necessary. The Parliament has to give its opinion in the Consultation procedure and in the first reading of the Co-operation and Co-decision procedures. The areas requiring Parliament’s consultation range from custom duties, to agricultural legislation, transport policy and approximation of laws. The Single European Act has extended consultation to some aspects of research and technology and environmental policy. There is also an emerging practice of optional consultation and the jurisprudence has set some criteria for reconsultation (Weatherill and Beaumont, 1995: 108-115). Another institution whose opinion is compulsorily required is the Economic and Social Committee (ESC).(31) Its origin can be traced back to the European Steel and Coal Community but its establishment came with the Rome Treaties. It is purely an advisory body representing socio-economic interests with members appointed by Member States. The policy spheres on which the ESC must be consulted include agriculture, social policy, freedom of workers movement and some internal market issues.

The question this section tries to answer is whether opinions of the Parliament and of the ESC have an impact on delegated functions. Students applying game theory to EC legislative procedures normally disregard opinions because they are not legally binding. Actors do not condition their strategies on the signals sent with the opinions especially when the sender has no role in determining the final payoffs of the game.(32) Opinions are however important soft laws (Wellens and Borchardt, 1980) that cast light on the interpretation of legal provisions and on institutional preferences and can have a substantial impact on policy implementation (Bulmer, 1994: 367-368). Taking into account the role of opinions means shifting the theoretical framework toward a more sociological understanding of actors’ behavior because it includes variables that have ideational or cognitive relevance and are more likely to affect preferences and beliefs rather than strategies.


I term this last model socio-rationalist because it includes both rationalist and sociological variables. I have added to the diluted rationalist model two dummies for when the Parliament and the ESC have issued opinions. The new model variant, then, is:

formula 5

where Parliament opinion (EPO) and opinion of the Economic and Social Committee (ESCO) take the value of 1 when an opinion is issued by the Parliament and the Committee respectively. Since there is no evidence of opposition to delegation in the opinions of the sampled legislation, my hypothesis is that the issuance of such an opinion favors the conferral of powers upon the Commission. The lower part of Table 2 shows the results; they also take as benchmark for comparison the diluted rationalist model.

The inclusion of opinions does not relevantly improve the explanatory power of the model. We cannot reject the null hypothesis that the new model can be randomly generated from the diluted rationalist one. This is because the adjusted R2 (54.93) falls within the 95 percent confidence intervals of the R2 of the diluted rationalist model. This is according to all four methods of computation of the intervals. Also the F-test on variable addition (F(2,93) = 0.07) fails to reject the null hypothesis.

The substantive and statistical significance of the original four variables (PREF, PROC, INST and INFO) is not affected by the introduction of the opinion of the Parliament and the Committee. The new variables (EPO and ESCO), while significantly different from zero, have a very marginal effect on executive discretion. Ceteris paribus, they determine around a fifth of a point of executive discretion each. Notice also that the sign of Parliament opinion (EPO) is unexpected.

From these results I think we can safely confirm many qualitatively based statements on the ineffectiveness of the ESC and the Parliament in its consultation role (e.g. Nugent, 1994: 239-242, 309; Weatherill and Beaumont, 1995: 108-113, 151). More specifically, results disconfirm the notion that opinions can have a certain effect on delegated powers and, at least to this extent, rationalist students are correct in disregarding them.

VI. Delegation and ex-post control

The final section of this article analyses the causal link between delegation and ex-post control. In Franchino (1998: 19-20) I have shown how Member States strongly prefer to be informed about implementation and how they set up a structure of committees of national experts (i.e. Comitology) to oversee Commission’s execution of policy decisions. As Hayes-Renshaw and Wallace put it (1997: 182) “’Comitology’ is […] a rather normal tool of the policy maker and policy implementer, namely the convening of groups through which the Commission discusses in advance policy ideas with relevant ‘clients’ or, subsequently, the progress of policy implementation’.(33) In agency theory parlance, Comitology is an oversight mechanism of the ‘police patrol’ type (McCubbins and Schwartz, 1984; Pollack, 1997) that comprises active, and costly, monitoring of agent’s decisions by the principals with the aim of limiting agency losses.

The link between discretion and control has been first suggested by McCubbins and Page (1987: 416-418) in two hypotheses where they propone a positive relation between the scope of regulatory authority delegated to the agent and strictness of implementing procedures. The empirical works of McCubbins, Noll and Weingast (1987, 1989) confirm this proposition and formal models (e.g. Epstein and O'Halloran, 1994) show the circumstances under which principal and agent prefer ex-post control. Although with exceptions, this relation is also confirmed in case of multiple principals as I have formally shown in Franchino (1998: 21-24).(34)


Council Decision 87/373/EEC has introduced seven distinct procedures for the operation of implementation committees in the Community. They can be arrayed along two dimensions that measure the degree of executive autonomy of the Commission. First, the extent to which Council control takes place after or prior to Commission’s measure. In the former case, the Commission is free to adopt implementation measures that can be subsequently reviewed by the Council. In the latter case, the Commission cannot adopt measures without Council approval. The second dimension is the decision rule to modify or annul Commission’s measures used in the relevant committee. Table 3 briefly summarizes committees’ characteristics and the indexes that I have assign to measure the increasing level of ex-post control.(35)

Table 3

Results in Table 4 confirm the positive relation between the extent of executive discretion and the severity of ex-post control. I have used both Spearman and Kendall rank correlation tests. Both statistics reject the null hypothesis of no correlation between executive discretion and ex-post control, the only exception being the normal approximation endpoints of the null hypothesis of the Kendall S.(36) There is a rather strong increasing monotonic relation between executive discretion and ex-post control according to the Spearman test. The mean bootstrapped value of rs is 0.99, in case of perfect relation rs equals one. The weaker statistical significance of the Kendall S is probably due to the high number of cases in the same ordinal scale (i.e. draws) in the sample.

Table 4

To conclude, when Member States decide to delegate extensive executive autonomy to the Commission, they also tend to establish rather confining procedures of implementation in order to oversee Commission’s activity and limit agency losses.

VII. Conclusion and discussion

‘Much of the problem of poor empirical performance alluded to by Green and Shapiro, however, derives from the very virtue of rational choice theory. This is that it does tend to be rigorous theory, generating well-specified propositions that with the appropriate evidence can be determined as being either true or false, in a discipline in which much of the other theorizing is imprecise, casual, ad hoc and therefore much easier to find some sort of generalized empirical support for’(Laver, 1997: 14-15).

Michael Laver strikes the right cords in his reply to Green and Shapiro. Rational choice allows the researcher to develop rigorous, well-specified models and to sharpen the business of empirical evaluation, sometimes also for the discomfort of the student when his or her model do not stand up to these tests.

For the relevance of this article, the first clear benefit of rational choice modeling is the clarification on the conditions under which variables have (and have not) explanatory power. Given the monopolistic power of proposal the Commission for instance, legislative procedures affect the degree of executive discretion only in non amending legislation while this factor loses relevance in other cases. This has allowed me to rigorously justify the selection criteria and the controlling variables of the data set. It has also defined the limits of my findings because no analysis has been carried out to test the validity of the controlling variable on amending legislation.


Rational choice scores fairly well in the statistical part of the article. On the positive side, information asymmetry is, as predicted, the most robust and important variable determining Commission’s autonomy. An increase of one thousand words in the length of the act induces a two point increase in the discretion index in all the above models (ceteris paribus). Students of EC affairs such as Majone, Moravcsik, Pierson and Pollack are correct in stressing the relevance of the distribution of information as an important factor of the integration dynamics. Technical complexity leads to more delegation and autonomy or even to the establishment of new institutions such as in case of judicial review, environmental and monetary policies. For instance, students of EC judicial politics have stressed the role played by the European Court of Justice in the integration process in such an highly technical subject of judicial review (Alter and Meunier-Aitsahalia, 1994; Burley and Mattli, 1993; Moravcsik, 1995: 623). For the same reason and for matters of credibility, the management of a single currency requires an independent institution, the European Central Bank, solely dedicated to price stability.

On a similar positive note, but based more on disconfirming rather than confirming evidence, the inclusion of other procedural variables (Parliament and opinions) does not improve the explanatory power of the basic model. Their substantive effect on executive discretion is rather marginal (ceteris paribus, a variation of one fifth of a point of the discretion index). We can say that the Parliament has no impact probably because its preferences are consistently similar to the Commission’s. However we cannot say that, if pivotal, the Parliament can make a difference because of the rarity of such occurrence. Similarly, we can say that opinions do not make a difference. This is not surprising, and it is consistent with the views of many EC students.

Another aspect where the model has scored relatively well is on the correlation between discretion and ex-post control. When Member States confer extensive policymaking functions on the Commission, they also set rather confining implementation procedures to monitor Commission’s activity as predicted by the hypothesis and confirmed by empirical works on American institutions.

On a less positive note, although the parsimony of rational choice is a much celebrated strength of this method, sometimes the researcher might find himself or herself in the situation of acknowledging the importance of variables that have been excluded from the formal model. Legislative instrument is a variable that has been consistently relevant throughout the different tests. When the Community approves a regulation rather than a directive, it also delegates more functions to the Commission (ceteris paribus, the discretion index increases by one point). Legislative instrument has not been formally analyzed mainly for reasons of tractability. Parsimony has certainly its cost.

Less satisfactory is also the performance of other two independent variables, preference distribution and legislative procedures. They pose different challenges to rational choice. Measuring actors’ preferences is a far more tricky business in political science than in economics (where rational choice originated). In the latter discipline, consumers, companies and governments reveal their preferences in their purchasing behavior and their pricing, expenditure and investment decisions. These are easy to find and quantify. In political science, preferences can be deliberately concealed behind the doors of smoke-filled rooms of diplomatic negotiations or of the EC Council of Ministers for that matter. And even if revealed, preferences can be difficult to quantify for statistically intensive studies. My operationalisation has not been successful. The substantive effect on discretion has been marginal (ceteris paribus, less than one point change in executive discretion when the distribution of preferences varies across the two extremes of 1987 and 1993). For statistical intensive and rational choice studies, much more work needs to be done because of the analytical importance of state preferences. In his exhortation for detailed research into domestic preference formation, Moravcsik points out that ‘[p]references determine the nature and intensity of the game that states are playing and thus are a primary determinant of which systemic theory [i.e. realist and regime theory – and EC politics for us] is appropriate and how it should be specified’ (1997 : 542).

Legislative procedure, the second variable, poses a more serious challenge to rational choice as we traditionally see it because it questions the ontological premises of thin rationality. Only informal procedures have a substantive effect on discretion (ceteris paribus, the index decreases by one point when unanimity is formally or informally used). Formal rules are substantively insignificant. I do not wish to enter the debate on the compatibility of different institutionalist schools (see e.g. Dowding, 1994; Hall and Taylor, 1996; Kato, 1996; Ostrom, 1991) but the results from this article stressed the importance of normatively enforced rules in complementing or substituting formal rules. We need to sharpen our concept of rationality by incorporating heuristics, norms and rules as powerful explanatory variables in ‘second-generation models‘ of rationality (Ostrom, 1998; Yee, 1997).



Population characteristics

The population includes all 1987-93 non amending secondary legislation (regulations and directives). I have selected this period because of the intensive legislative activity of the Community and because the analysis needs to control for EC Treaty reform (Single European Act in 1987 and Maastricht Treaty in 1993). An interesting replication of the findings could include post-1993 acts and the Co-decision procedure.

The CELEX database and the Official Journal have been the main sources used. Unfortunately, both are slightly deficient. CELEX has some regulations whose reference cannot be found in the Official Journal. Given the legal requirements of publication, this seems to be a flaw of the database. Conversely, there is not a requirement of publication of Directives in the Official Journal, which is then incomplete in this respect.

I have disregarded decisions because of their administrative and addressee-related nature, and opinions and recommendations because they are not legally binding. Directive and regulations amending decisions, protocols or conventions have been included.

In the few cases where there were more than one Treaty base, the legislative act has been assigned first to the more stringent procedure then to the policy so that to ensure the widest distribution across policies and procedures. This is in order to maximize the efficiency of the sampling strategy (see below). Apart from references to articles 235 and 227.2 which require unanimity but do not affect the allocation to a policy area, there are only 25 (i.e. less that 2.5 percent) directives and regulations which are based on more that one Treaty provision and to which I have apply this allocation criteria.

Sampling strategy

A sampling procedure needs to trade off feasibility and representation. A feasible sample minimizes sampling costs and analytical complexity. A representative sample mirrors the key characteristics of the population and minimizes the sampling error.

In our case, the population shows highly skewed frequency distributions across the two key variables of policy area and legislative procedure (see Chart 2). A simple random sample could easily under-represent a policy area or a legislative procedure. For instance, in a simple random sample with repetition of 100 cases drawn by the author, only one regulation has been approved under unanimity and five directives were approved using Cooperation. Both procedures were heavily underrepresented. Conversely, customs union, agricultural and commercial legislation amounted to slightly more than 85 percent.

In order to decrease such sampling error without increasing the sample size, I have instead drawn a stratified random sample of 100 cases. Each stratum is characterized by a different Treaty base and legislative procedure to ensure internal homogeneity and external heterogeneity. The sample size of each stratum is proportional to the stratum population. This procedure is termed stratified random sampling with proportional allocation or constant sampling fraction. In this way, first and second order probabilities of inclusion of a case in a stratum equal simple random sampling probabilities and variance and total formulae are similar. There is no need to modify values of observations (Frosini et al., 1994: 87-8). Further, bootstrapping (see below) obviates eventual problems of probability distribution.

Although only simple random sampling generates samples with independently and identically distributed cases, this proportionate stratified sampling improves representation without complicating too much the analysis (Frosini et al., 1994 : 41-45).



Bootstrapping is a computationally intensive non-parametric approach to statistical inference. It is of particular utility in our case when traditional distributional assumptions of parametric inference are violated (Mooney and Duval, 1993; Mooney and Krause, 1997). Moreover, bootstrapping has also allowed me to make clear inferential statements about the goodness-of-fit of the models under study because it constructs an estimate of the sampling distribution of statistics with weak statistical theory such as the adjusted R2 and co-graduation indexes.

In regression models, its basic procedure is to take 1000 re-samples with replacement of the residuals of the original regression model, to calculate the bootstrapped dependent variables, regression coefficients and R2 and to develop bootstrap confidence intervals. Statisticians point out that this is the most appropriate procedure to bootstrap a regression model. An alternative could be to resample cases of data but this ignores the error structure of the model (Mooney and Duval, 1993: 17).

Since bootstrapping is primarily a tool for inferential statistics, I have used as point estimator of the population variables the mean bootstrapped values. This is based on principles similar to those of the estimates of the jackknife technique (Mooney and Duval, 1993: 22-27).

To test whether these observed values were in the critical region, I have computed the a/2 and 1-a/2 double-tailed endpoints of the null hypothesis (e.g. b=0 for regressors). The techniques used to compute the confidence intervals were the normal approximation, percentile, bias corrected and percentile-t. I have also computed the ‘bootstrapped t(or z)–statistics’ using the standard error estimated with the bootstrapped sampling distribution (Mooney and Duval, 1993: 35).

I have developed my own GAUSS code to carry out the computations, for alternatives see Mooney (1994a, 1994b).

Although I have reported all four bootstrap confidence intervals that are generally used, statisticians tend to agree that the BC and percentile-t methods are the most accurate. Normal approximation fails to use the entire bootstrapped estimate of the sampling distribution and requires parametric assumptions about the empirical probability distribution. Further, it is probably less appropriate in case of regression coefficients because parametric statistical theory tells us that they are likely to have a t-distribution. Percentile assumes an unbiased bootstrapped estimate of the sampling distribution (see Mooney and Duval, 1993: 33-42).


Alter, Karen J. and Sophie Meunier-Aitsahalia (1994) 'Judicial Politics in the European Community: European Integration and the Pathbreaking Cassis de Dijon Decision.', Comparative Politics Studies 26:535-561.

Aspinwall, Mark D. and Gerald Schneider (1998) 'Same Menu, Separate Tables. the Institutionalist Turn in Political Science and the Study of European Integration'. Paper read at ECPR Joint Sessions of Workshops, 23 - 28 March 1998, at Warwick.

Berry, William D. (1993) Understanding Regression Assumptions. London: SAGE.

Bradley, Kieran St Clair (1992) 'Comitology and the Law: Through a Glass, Darkly', Common Market Law Review 29:693-721.

Bulmer, Simon J. (1994) 'The Governance of the European Union: A New Institutionalist Approach', Journal of Public Policy 13:351-380.

Burley, Anne-Marie and Walter Mattli (1993) 'Europe Before the Court: A Political Theory of Legal Integration', International Organization 47:41-76.

Checkel, Jeffrey T. (1998) 'Social Construction, Institutional Analysis and the Study of European Integration'. Paper read at ECPR Joint Sessions of Workshops, 23 - 28 March, at Warwick.

Cram, Laura (1998) 'The EU Institutions and Collective Action: Constructing a European Interest?'. In Collective Action in the Europan Union. Interest and the New Politics of Associability, edited by J. Greenwood and M. Aspinwall. London: Routledge.

Crombez, Christophe (1996) 'Legislative Procedures in the European Community', British Journal of Political Science 26:199-228.

Dowding, Keith M. (1994) 'The Compatibility of Behaviouralism, Rational Choice and 'New Institutionalism'', Journal of Theoretical Politics 6:105-117.

Earnshaw, David and David Judge (1996) 'From Co-operation to Co-decision'. In European Union. Power and Policy-Making, edited by J. Richardson. London: Routledge.

EC Commission (1994) Eurobarometer. Trends 1974 - 1994. Brussels: EC Commission.

Edwards, G. and A. Pijpers eds. (1997) The Politics of European Treaty Reform. London: Pinter.

Epstein, David and Sharyn O'Halloran (1994) 'Administrative Procedures, Information, and Agency Discretion', American Journal of Political Science 38:697-722.

Fennel, Rosemary (1987) 'Reform of the CAP: Shadow or Substance?', Journal of Common Market Studies 26:61-77.

Finnemore, Martha (1996) 'Norms, Culture, and World Politics: Insights from Sociology's Institutionalism', International Organization 50:325-347.

Fitzmaurice, J. (1988) 'An Analysis of the European Community's Co-operation Procedure', Journal of Common Market Studies 26:389 - 400.

Franchino, Fabio (1998) 'Commission's Executive Discretion, Information and Policy Preferences'. Paper read at ECPR Joint Sessions of Workshops, 23 - 28 March 1998, at Warwick University, Warwick.

Frosini, Benito V., Mario Montinaro and Giovanna Nicolini (1994) Il Campionamento da Popolazioni Finite. Metodi e Applicazioni. Turin (Italy): UTET.

Garrett, Geoffrey and George Tsebelis (1996) 'An Institutional Critique of Intergovernmentalism', International Organization 50:269 - 299.

Garrett, Geoffrey and Barry R. Weingast (1993) 'Ideas, Interests, and Institutions: Constructing the European Community's Internal Market'. In Ideas and Foreign Policy. Beliefs, Institutions, and Political Change, edited by J. Goldstein and R. O. Keohane. Ithaca: Cornell University Press.

Gilligan, Thomas and Keith Krehbiel (1989) 'Collective Choice without Procedural Commitment'. In Models of Strategic Choice in Politics, edited by P. C. Ordeshook. Ann Arbor: University of Michigan Press.

Green, Donald P. and Ian Shapiro (1994) Pathologies of Rational Choice Theory. New Haven, CT: Yale University Press.

Greenwood, Justin and Mark Aspinwall eds. (1998) Collective Action in the European Union. Interest and the New Politics of Associability. London: Routledge.

Greenwood, Justin and Karsten Ronit (1994) 'Interest Groups in the European Community: Newly Emerging Dynamics and Forms', West European Politics 17:31-52.

Hall, Peter A. and Rosemary C.R. Taylor (1996) 'Political Science and Three New Institutionalisms', Political Studies 44:936-957.

Hayes-Renshaw, Fiona and Helen Wallace (1997) The Council of Ministers. London: MacMillan.

Hix, Simon (1994) 'The Study of the European Community: The Challenge to Comparative Politics.', West European Politics 17:1-30.

Hosli, Madeleine O. (1996) 'Coalitions and Power: Effects of Qualified Majority Voting in the Council of the European Union', Journal of Common Market Studies 34:255-273.

Jorgensen, Knud Erik ed. (1997) Reflective Approaches to European Governance: MacMillan.

Josling, Timothy E. and Angela C. Mariani (1993) 'The Distributional and Efficiency Implications of the Mac Sharry Proposals for Reform of the CAP', Journal of Regional Policy 13:27-49.

Kato, Junko (1996) 'Review Article: Institutions and Rationality in Politics - Three Varieties of Neo-Institutionalists', British Journal of Political Science 26:553-582.

Keeler, John T.S. (1996) 'Agricultural Power in the European Community. Explaining the Fate of CAP and GATT Negotiations', Comparative Politics 28:127-149.

Keohane, Robert O. (1984) After Hegemony: Cooperation and Discord in the World Political Economy. Princeton, N.J.: Princeton University Press.

Kerremans, Bart (1996) 'Do Institutions Make a Difference? Non-Institutionalism, Neo-Institutionalism, and the Logic of Common Decision-Making in the European Union', Governance: An International Journal of Policy and Administration 9:217-240.

Kiewiet, D. Roderick and Mathew D. McCubbins (1991) The Logic of Delegation. Congressional Parties and the Appropriations Process. Chicago: The University of Chicago Press.

Kreps, David M. (1990) 'Corporate Culture and Economic Theory'. In Perspectives on Positive Political Economy, edited by Alt and Shepsle. Cambridge: Cambrige University Press.

Laver, Michael (1997) Private Desires, Political Action. An Invitation to the Politics of Rational Choice. London: SAGE.

Legro, Jeffrey W. (1997) 'Which Norms Matter? Revisiting the "Failure" of Internationalism', International Organization 51:31-63.

Lenaerts, Koen (1991) 'Some Reflections on the Separation of Powers in the European Community', Common Market Law Review 28:11 - 35.

Lowi, Theodore J. (1964) 'American Business, Public Policy, Case-Studies, and Political Theory', World Politics 16:677-715.

Majone, Giandomenico (1996) Regulating Europe. London: Routledge.

March, James G. and Johan P. Olsen (1989) Rediscovering Institutions: The Organizational Basis of Politics. New York: Free Press.

Marsh, John (1997) 'The Common Agricultural Policy'. In New Challenges to the European Union: Policies and Policy-Making, edited by S. Stavridis, E. Mossialos, R. Morgan and H. Machin. Aldershot: Darthmouth.

McCubbins, Mathew D., Roger G. Noll and Barry R. Weingast (1987) 'Administrative Procedures as Instruments of Political Control', Journal of Law, Economics, and Organization 3:243 - 277.

McCubbins, Mathew D., Roger G. Noll and Barry R. Weingast (1989) 'Structure and Process, Politics and Policy: Administrative Arrangements and the Political Control of Agencies', Viginia Law Review 75:431 - 482.

McCubbins, Mathew D. and Talbot Page (1987) 'A Theory of Congressional Delegation'. In Congress: Structure and Policy, edited by M. D. McCubbins and T. Sullivan. Cambridge, MA: Cambridge University Press.

McCubbins, Mathew D. and Thomas Schwartz (1984) 'Congressional Oversight Overlooked: Police Patrols versus Fire Alarms', American Journal of Political Science 28:165-179.

McKelvey, Richard D. (1976) 'Intransitivities in Multidimentional Voting Models and Some Implications for Agenda Control', Journal of Economic Theory 12:472-482.

Milgrom, Paul and John Roberts (1992) Economics, Organization and Management. Englewood Cliffs, NJ: Prentice Hall.

Moe, Terry M. (1987) 'An Assessment of the Positive Theory of Congressional Dominance', Legislative Studies Quarterly 12:475-520.

Mooney, Christopher Z. (1994a) 'Constructing Bootstrapped Confidence Intervals Using GAUSS', Political Methodologist 6:27-29.

Mooney, Christopher Z. (1994b) 'GAUSS Code for Re-sampling Cases and Residuals in Boostrapping a Regression Model', Political Methodologist 6:26-27.

Mooney, Christopher Z. (1997) Monte Carlo Simulation. London: SAGE.

Mooney, Christopher Z. and Robert D. Duval (1993) Boostrapping. A Nonparametric Approach to Statistical Inference. London: SAGE.

Mooney, Christopher Z. and George A. Krause (1997) 'Review Article : Of Silicon and Political Science - Computationally Intensive Techniques of Statistical Estimation and Inference', British Journal of Political Science 27:83-110.

Moravcsik, Andrew (1993) 'Preferences and Power in the European Community: A Liberal Intergovernmentalist Approach', Journal of Common Market Studies 31:473-524.

Moravcsik, Andrew (1995) 'Liberal Intergovernmentalism and Integration: a Rejoinder', Journal of Common Market Studies 33:611-628.

Moravcsik, Andrew (1997) 'Taking Preferences Seriously: A Liberal Theory of International Politics', International Organization 51:513-553.

Morrow, James D. (1994) Game Theory for Political Scientists. Princeton, New Jersey: Princeton University Press.

Moser, Peter (1996) 'The European Parliament as a Conditional Agenda Setter: What are the Conditions? A Critique of Tsebelis (1994)', American Political Science Review 90:834 - 838.

Moser, Peter (1997) 'A Theory of the Conditional Influence of the European Parliament in the Cooperation Procedure', Public Choice 91:333 - 350.

Nugent, Neill (1994) The Government and Politics of the European Union. Third ed. London: MacMillan.

O'Halloran, Sharyn (1994) Politics, Process, and American Trade Policy: The University of Michigan Press.

Ostrom, Elinor (1991) 'Rational Choice Theory and Institutional Analysis: Toward Complementarity', American Political Science Review 85:237-243.

Ostrom, Elinor (1998) 'A Behavioral Approach to the Rational Choice Theory of Collective Action. Presidential Address, American Political Science Association, 1997', American Political Science Review 92:1-22.

Patterson, Lee Ann (1997) 'Agricultural Policy Reform in the European Community: A Three-Level Game Analysis', International Organization 51:135-165.

Peterson, John (1989) 'Hormones, Heifers and High Politics - Biotechnology and the Common Agricultural Policy', Public Administration 67:455-471.

Pierson, Paul (1996) 'The Path to European Integration. An Historical Institutionalist Analysis', Comparative Political Studies 29:123 - 163.

Pollack, Mark A. (1994) 'Creeping Competence: The Expanding Agenda of the European Community', Journal of Public Policy 14:95 - 145.

Pollack, Mark A. (1997) 'Delegation, Agency, and Agenda Setting in the European Community', International Organization 51:99-134.

Radaelli, Claudio M. (1995) 'Corporate Direct Taxation in the EU: Explaining the Policy Process', Journal of Public Policy 15:43-71.

Radaelli, Claudio M. (1997) 'How Does Europeanization Produce Domestic Policy Change? Corporate Tax Policy in Italy and United Kingdom', Comparative Political Studies 30:553-575.

Raiffa, H. (1982) The Art and Science of Negotiations. Cambridge, Mass.: Harvard University Press.

Richardson, Jeremy (1996) 'Policy-Making in the EU. Interests, Ideas and Garbage Cans of Primeval Soup'. In European Union. Power and Policy-Making, edited by J. Richardson. London: Routledge.

Runge, C.F. and H. von Witzke (1987) 'Institutional Change in the Common Agricultural Policy of the European Community', American Journal of Agricultural Economics 69:213-222.

Scharpf, Fritz W. (1988) 'The Joint-Decision Trap: Lessons from German Federalism and European Integration', Public Administration 66:239-278.

Schneider, Gerald and Lars-Erik Cederman (1994) 'The Change of Tide in Political Cooperation: a Limited Information Model of European Integration', International Organization 48:633-662.

Shepsle, Kenneth A. (1979) 'Institutional Arragment and Equilibrium in Multidimentional Voting Models', American Journal of Political Science 23:27-59.

Shepsle, Kenneth A. and Barry R. Weingast (1981) 'Structure-induced Equilibrium and Legislative Choice', Public Choice 37:503-520.

Shepsle, Kenneth A. and Barry R. Weingast (1984) 'Uncovered Sets and Sophisticated Voting Outcomes with Implications for Agenda Institutions', American Journal of Political Science 28:49-74.

Stein, Eric (1981) 'Lawyers, Judges, and the Making of a Transnational Constitution', American Journal of International Law 75:1-27.

Steunenberg, Bernard (1994) 'Decision Making under Different Institutional Arrangements: Legislation by the European Community', Journal of Institutional and Theoretical Economics 150:642-669.

Steunenberg, Bernard, Christian Koboldt and Dieter Schmidtchen (1996a) 'Beyond Comitology: a Comparative Analysis of Implementation Procedures with Parliamentary Involvement', Aussenwirtshaft 52:87 - 112.

Steunenberg, Bernard, Christian Koboldt and Dieter Schmidtchen (1996b) 'Policymaking, Comitology, and the Balance of Power in the European Union', International Review of Law and Economics 16:329 - 344.

Steward, Jon (1991) Econometrics. London: Philip Allan.

Swinbank, Alan (1989) 'The Common Agricultural Policy and the Politics of European Decision Making', Journal of Common Market Studies 27:303-322.

Tsebelis, George (1990) Nested Games: Rational Choice in Comparative Politics. Berkeley: University of California Press.

Tsebelis, George (1994) 'The Power of the European Parliament as a Conditional Agenda Setter', American Political Science Review 88:128-142.

Tullock, Gordon (1981) 'Why so much Stability?', Public Choice:189-202.

Wallace, Helen (1989) 'The Best is the Enemy of the "Could": Bargaining in the European Community'. In Agricultural Trade Liberalisation and the European Community, edited by S. Tarditi, K. Thomson, P. Pierani and E. Croci-Angelini. Oxford: Claredon Press.

Weatherill, Stephen and Paul Beaumont (1995) EC Law. The Essential Guide to the Legal Workings of the European Community. London: Penguin Books.

Weiler, Joseph H. H. (1991) 'The Transformation of Europe', The Yale Law Journal 100:2403-2483.

Weingast, Barry R. and Mark J. Moran (1983) 'Bureaucratic Discretion or Congressional Control? Regulatory Policymaking by the Federal Trade Commission', Journal of Political Economy 91:756-800.

Wellens, K. and G. Borchardt (1980) 'Soft Law in the European Community Law', European Law Review 14:267-321.

Widgrén, Mika (1995) 'Probabilistic Voting Power in the EU Council: The Cases of Trade Policy and Social Regulation', Scandinavian Journal of Economics 97:345-356.

Williamson, O.E. (1985) The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. New York: Free Press.

Woolcock, Stephen and Michael Hodges (1996) 'EU Policy in the Uruguay Round'. In Policy-Making in the European Union, edited by H. Wallace and W. Wallace. Oxford: Oxford University Press.

Yee, Albert S. (1997) 'Thick Rationality and the Missing "Brute Fact": The Limits of Rationalist Incorporations of Norms and Ideas', The Journal of Politics 59:1001-1039.

Zetterholm, S. ed. (1994) National Cultures and European Integration: Exploratory Essays on Cultural Diversity and Common Policies. Oxford: Berg.


(*) Comments on previous versions of the paper from Fulvio Attiná, Jens Bastian, Claudio Radaelli, Cheryl Schonhardt-Bailey, two anonymous EIoP reviewers and participants at the workshop on ‘Lo Spazio Politico Europeo’, 1998 SISP Conference, University of Milan, are very much appreciated.

(1) Judicial review and Commission’s proposal power are practically excluded from Justice and Home Affairs and Foreign and Security Policy. There are also few EC Treaty articles, such as 48.3d, that accord powers of implementation directly to the Commission.

(2) My use of game theory takes the EC Treaty as the key binding agreement (see Morrow, 1994: 75). Cooperative games are played at the Intergovernmental Conferences in terms of international public law and non cooperative games are played in secondary legislation in terms of implementation.

(3) On the ‘constitutionalization’ of Community law see Stein (1981) and Weiler (1991), on interest groups see Greenwood and Ronit (1994) and Greenwood and Aspinwall (1998), on institutions, policy process and comparative politics Lenaerts (1991), Hix (1994) and Richardson (1996).

(4) Categories 6, 7 and 8 are based on the works of Majone (1996) and (more loosely) of Pollack (1994). Majone (p. 54) uses a functional classification of policies (stabilization, regulation and redistribution). Categories 6-7 and category 8 overlap with the latter two. Pollack uses Lowi’s (1964) classification of regulatory, distributive and redistributive policy types to explain task expansion in the Community. Categories 6-7 are similar to the first one, category 8 to the last two.

(5) Only the part that includes the actual article of the legislation has been considered for this purpose. Introductory statements giving reasons for adoption, tables and annexes have been disregarded for that matter.

(6) There are two problems with this procedure. The first concerns whether the number of delegated activities actually measures the degree of executive discretion. One could object that the delegation of such activities is frequently accompanied by a list of implementation criteria that limits the room of manoeuvre of the Commission. This is certainly correct, but a measure of the stringency of these criteria as an intervening or independent variable fails the test of cross-policy comparability mainly because of the technical complexity of secondary legislation. Is, for instance, a criterion that sets the quantities of head of bovine quotas stricter than a purity criterion of foodstuffs flavourings? It is my opinion that researcher’s measurement bias plays a too great a role here to assure objectivity. Moreover, the effect of these criteria on discretion loses at least some of its significance if we see EC laws from the perspective of Williamson’s (1985) economic theory of contracts. From this viewpoint, any legislative act is an incomplete contract because it cannot specify precisely what each institution is to do in all possible circumstances. To solve this contractual incompleteness, legislators rely on so called ‘relational contracts’ (Milgrom and Roberts, 1992; see also Majone, 1996) where they specify only general goals and establish ex-post control procedures while specific criteria tend to play a lessen role. In fact, a long list of criteria can easily present inherent contradictions (see e.g. article 39 EC) and then impose no effective control on the empowered institution, especially in highly complex policy environment. Here, it is more likely to have general objectives (e.g. price stability in monetary policy) and either control mechanisms such as implementation committees or an institutional framework where reputational factors are effective constraints on behavior (on reputation see Kreps, 1990; on control mechanisms see e.g. Moe, 1987).

A second possible objection to the construction of this index is that the different activities should be weighted for the degree of discretion they bestow upon the Commission. If in one regulation the Commission is asked to provide information on a certain matter while in a second regulation it is asked to regulate the matter, surely more discretionary authority has been delegated in the latter case. Although it raises the issue of appropriate weighting, this is another correct point but it turned out to be of less relevance empirically. To test this, I have assigned an increasing value from one to eight starting from the top activity listed above and computed a weighted index of discretion in a similar way. I then applied the statistical analysis described below and found no appreciable difference in the results of the study. This is because a law that delegates only regulatory powers (and for that matter, activities at the bottom of the list) is a very rare occurrence. Normally the act asks the Commission also to collect and provide information from and to Member States and, probably, to give opinions (that is activities at the top of the list). Hence, there is not substantial difference between the two indexes. For instance, Council Directive 92/80/EEC on the approximation of taxes on manufactured tobacco requires the Commission to simply receive and provide information. This makes a discretion index of two and a weighted index of three. A regulation providing for the administration of tariff quotas (e.g. Regulation 786/88) delegates administrative functions but asks also for the exchange of information. In this case, the discretion index would be four and the weighted one nine. Both indexes gauge the actual difference in discretion. Since this is the way the index is constructed and laws are drafted, I have decided not to use weights rather than assigning arbitrary ones. Weights, without a specific justification, may introduce a bias by making inappropriate assumptions.

(7) Franchino (1998) is currently under review. In this paper I will frequently refer to that article. This is because some statements made here have been mathematically, but not empirically, proved in Franchino (1998). For reasons of space I cannot recall the formal argument, if interested please contact the author. Unfortunately there are no formal works (that I know of) on agency discretion in the EC that I can relate to, so I will refer in some cases to similar formal works on US institutions, such as Epstein and O'Halloran (1994).

(8) McCubbins and Page (1987) add that, as a result, legislators will also design more stringent control mechanisms. This proposition will be tested at the end of the article.

(9) The part of the legislative act which is counted for the number of words covers the text from the first article to the name of the President of the Council of Ministers included.

(10) It is impossible, in this context, to do justice to an impressive set of contributors from different intitutionalist streams. Remaining in the rationalist camp, Shepsle and Weingast’s (1979, 1981, 1984) seminal work on structure induced equilibrium is one of the prominent theories.

(11) The Commission always prefers complete discretion as I have proved in Appendix 1b of Franchino (1998: 27-28).

(12) See Proposition 1 and Figure 4 in Franchino (1998: 10). In case of Assent procedure, it can be the parliamentary preferences that set the equilibrium discretion but only in the very unlikely case that Parliament’s ideal policy is the farthest away from Commission’s. This occurrence is not relevant for us because there is no legislation approved under Assent in the data set.

(13) For a detail description of the Co-operation procedure see, for instance, Nugent (1994). On the ‘conditional agenda setting’ debate see also Moser (1996, 1997).

(14) The equilibrium discretion in the Co-operation procedure is [min(2dGb, 2dP)] where dGb and dP are the ideal discretion for the pivotal Government and the Parliament respectively. See proof in Franchino (1998: 11-12).

(15) See Propositions 2i and 3 and the section on conflict between principals in Franchino (1998: 14-15) for proof and a more detailed analysis.

(16) On voting weights and power indexes see Hosli (1996) and Widgrén (1995).

(17) The correlation coefficient is 0.56, the observed t value is 1.51 which is lower than the critical value t5 (0.05) of 2.02.

(18) Technically, it assumes the absence of agency losses. I thank an anonymous referee from EIoP, Francesca Longo and Claudio Radaelli for pointing this out.

(19) I limit my comments to rational choice and sociological institutionalism given their clearer ontological premises. The different shades of historical institutionalism can be traced back to share common features with either one of the two or both. It is not my aim to deal with all aspects of the institutionalist debate. Good reviews are by Finnemore (1996), Hall and Taylor (1996), Yee (1997) and, focused on the EU, Aspinwall and Schneider (1998).

(20) In between these extremes, many studies have a historic-institutionalist approach that merges the logics of appropriateness and consequentiality (Hall and Taylor, 1996; March and Olsen, 1989). Formal EU rationalist studies include Crombez (1996), Garrett and Tsebelis (1996) and Hosli (1996). The emerging sociological literature is on constructivism (Cram, 1998; Jorgensen, 1997), on the role of norms (Checkel, 1998) and of culture (Zetterholm, 1994). Apart from the extensive literature on epistemic communities, EU works on ideas include Garrett and Weingast (1993) and Radaelli (1995, 1997). Works that specifically deal with norms across EU policies are, to my knowledge, lacking, although there are more general socio-historical works on institutions ( e.g. Bulmer, 1994; Kerremans, 1996).

(21) See, for instance, Marsh (1997) on the distributional effects (partially considered as sunk costs in Pierson’s words) of the CAP. On the distributional implications of the MacSharry proposals see Josling and Mariani (1993).

(22) See, for instance, Steward (1991: 25) or Berry (1993) on the assumptions of the classical regression model. There might also be the risk of a measurement error with OLS because executive discretion (DELEG) is an ordinal index. However, this is likely to affect only the residuals of the equation and bootstrapping (see below) has been specifically adopted in this case in order to deal with the distribution of the error structure.

(23) There is no evidence of hetescedasticity, while there is a certain degree of collinearity between legislative procedure (PROC) and instrument (INST) and between preference distribution (PREF) and information asymmetry (INFO). The condition index is however well below 30 in both cases, so it is not a serious problem.

(24) I have carried out one-sided Jarque-Bera tests using the GAUSS code suggested by Mooney (1997). The null hypothesis of normality has been rejected at 5 per cent significance level for the rationalist model and at 10 per cent for diluted rationalist model. The dependent variable distribution approximates a highly right skewed Pareto distribution [Par(0,0.1)] mixed with a Chi squared distribution [c2 (4)] at a factor of 0.7.

(25) Notice that this test is different from the standard F-tests on model specification because the operationalisation of variables differs in this case.

(26) Below I test the inclusion of procedural variables. Apart from these tests, I have also added other variables such as a discretion index created on the basis of the Treaty text and a policy area variable based on the number of treaty articles providing for unanimity or qualified majority. The variables are either statistically insignificant or have no effect (if not collinear) on the signs and significance levels of the other variables. To be included, they also need a strong theoretical justification that I cannot find in the literature.

(27) The introduction of social sanctions or interests to explain such behavior simply shifts the level of debate to which socio-psychological mechanisms are in force (Yee, 1997: 1015-8).

(28) There are no amendments in the sampled legislation that cut back functions conferred on the Commission. However I must acknowledge that even in case of Parliamentary opposition to delegation, it would not have been possible, for lack of relevant data, to gauge whether the opposition was stronger that the pivotal Government’s.

(29) In this way we test whether the Parliament has a positive effect on discretion. Alternatively, I have tested a model with legislative procedure (PROC) taking the value of 0 for Co-operation, 1 for unanimity and 2 for qualified majority to see whether the Parliament has a negative effect on discretion. Its explanatory power is even lower than the one of the rationalist model.

(30) This also applies if we eliminate legislative instrument (INST) whose role was not formalized in Franchino (1998).

(31) I disregard the Committee of the Regions because it was established by the Treaty on European Union in 1993.

(32) This is the so-called babbling equilibrium in signaling games. Opinions can, on the contrary, enhance coordination if actors’ moves are interdependent (e.g. in Co-operation) and preferences are similar (Morrow, 1994: 250-256).

(33) For a less benevolent analysis, see Bradley (1992) on how Comitology is incompatible with the prerogatives conferred upon the Commission and the Parliament by the Treaty.

(34) The trade-off between discretion and control is not always present in case of multiple principals because the implementation game besets the danger of collusion between the agent and some principals. When Commission’s preferences are very distant from Government’s, the Member State prefers both less discretion and more ex-post control. While when they are very close, it prefers more discretion and less control (see Franchino, 1998: 22-23). However, these types of Governments are unlikely to be the pivotal actors because their preferences tend to lie at the extremes of the policy spectrum. As such, they will no affect the equilibrium outcome.

(35) Management IIb and Regulatory IIIa have the same value because they allow a similar degree of autonomy to the Commission. They have also a similar equilibrium (see Steunenberg et al. 1996a, 1996b). Safeguard procedures have higher values because they do not require the establishment of a committee of national experts acting as gatekeeper for measures that can be referred to the Council. In this case, the Council is immediately involved and control takes place at a higher level (Bradley, 1992).

If there are no implementation procedures the ex-post control value is zero. While I have disregarded the case when there are procedures that, although termed implementation, are clearly legislative ones. Finally, if there are two procedures in an act, the mean value has been recorded.

(36) In this case Kendall S is significant at 10 percent level. Kendall S is also significant at 5 percent level in the original sample.

©1998 by Franchino
formated and tagged by MN, 17.7.1998