Finding Patterns - The Supranational Scenario

Contents

List of Tables

List of Figures

Introduction

1. Theoretical Background and Relevance of the Research Question

2. Research Design: Dataset, Missing Values and the Construction of the Independent Variable

3. Analysis
a. New or Amended Legislation
b. The procedure and the Council voting rule
i. Co-decision vs. consultation
ii. Co-decision vs. Consultation Depending on the Council Voting Rule
iii. Summary
c. The Supranational Scenario and the Outcome
d. The Proposal Level

Conclusion

Literature

Appendix

List of Tables

Table 1. Mean distance from the European Parliament and the Commission to the Status Quo for new and amended legislation, and for both of them

Table 2. The mean position of the Commission and the European Parliament, and the Status Quo, depending of the procedure: co-decision or consultation and both of them. The mean position of the difference between the European Parliament and the Status Quo and between the Commission and the Status Quo

Table 3. The mean position of the Status Quo, the European Parliament and the Commission, their mean distance to the Status Quo and the difference between their position depending on the voting rule in the Council: Qualified Majority Voting and Unanimity and on the procedure: co-decision or consultation

Table 4. The correlation coefficient between the issue- and the proposal-level for: the distance between the Commission and the European Parliament, the distance between the Commission and the European Parliament to the Reference Point

Table 5. Absolute and relative frequency of Council voting rule: Qualified Majority Voting or Unanimity depending on the procedure: consultation or co-decision. (Appendix F.)

Table 6. Value of the adjusted R, taking into account the number of cases, for the linear regression of the mean position between the European Parliament and the Commission, controlling for the Council voting rule, the procedure and the interaction between the Council voting rule and the procedure. (Appendix G.

List of Figures

Figure 1. The supranational scenario

Figure 2. Whisker box for the difference between the position of European Parliament (EP) and the Commission’s (EC). (Appendix A.)

Figure 3. Histogram of the repartition of the Euclidian distance between the European Parliament (EP) and the Commission (EC) in steps of 10. (Appendix B.)

Figure 4. Whisker Boxes for the position of the European Parliament (EP) and the Commission (EC) depending on the procedure: consultation (CNS) or co-decision (COD). (Appendix C.)

Figure 5. Whisker boxes of the difference in the positions of the European Parliament (EP) and the Commission (EC) depending on the procedure: consultation (CNS) or co-decision (COD). (Appendix E.)

Introduction

When studying the European Union, one will find references to the “supranational scenario” quite soon. In current literature, there is a trend to think about an integration dimension in the European Union. According to this idea, the Status Quo is located at one end of the integration continua, the supranational actors, that is, the Commission and the European Parliament, at the other one. The Status Quo is assumed to be formed under the Luxemburg Compromise and therefore to be at the point of the least integrationist member state. Contrary to this, the Commission and the European Parliament are more pro-integrationist than all member states of the Council of Ministers. One of the most well-know theorists in the analysis on European Union in an institutionalist way is George Tsebelis.

My research question will be: Is there empirical evidence that one can easily assume that the European Parliament and the Commission are (a) located closely together in their preferences for concrete policies? Are (b) these positions generally opposite from the Status Quo? And are there (c) some factors which influence the likelihood that EP and EC will be positioned more closely together and at the same time far away from the Status Quo? Does (d) the “supranational scenario” influence the outcome?

The quantitative check of Tsebelis’ assumption will be made with the Decision-Making in the European Union Dataset. This database includes the position, the salience and the power of all (for the moment) 15 members of the Council, the outcome after the bargaining process, the reference point, and the two relevant actors in my study: the European Parliament and the Commission. To proceed, I will use the statistic program Stata 9.1, and run statistical tests on the data.

The goal of this paper is to provide further insights into the positioning of the Parliament and the Commission and to see if the supranational scenario survives a quantitative analysis. As my goal is to find patterns, I will emphasize on an explorative study, without, however, undermining the principles of a social inquiry.

1. Theoretical Background and Relevance of the Research Question

Since the very first steps in the European Union, so since the Treaty of Rome in 1957, a proposal by the Commission has required a qualified majority in the Council to be accepted, but unanimity to be overturned (Tsebelis / Kreppel 1998: 43). This mechanism existed in the consultation procedure in the Treaty of Rome. For the cooperation procedure, which was introduced with the Single European Act (1986), the higher threshold for overturning a proposal remained the same. But with the co-decision procedure, introduced in Maastricht (1993) and reformed in Amsterdam (1997), the European Parliament became a co-legislator of its own, because the last step of decision-making is now in the Conciliation Committee. Therefore, the Parliament and the Council have the agenda-setter-power. In the following, when talking about co-decision, I will refer to the reformed one of post-Amsterdam, because the data I have refers to this co-decision procedure (see chapter 2.: Research Design).

For his rational-choice institutionalist analysis, Tsebelis distinguishes several different “scenarios” how one could imagine the European Union: The “national delegates” scenario stipulates that Members of the European Parliament are “subject to significant domestic constraints that reduce their latitude to act as pro-Europe policy entrepreneurs” (Tsebelis / Garret 2000: 10). The “strong parties” scenario assumes that coalitions are formed among party, rather than national, lines. The “regulation” scenario sees conflict about the level of regulation, which may array on a right-left dimension (op. cit.: 11).

The “supranational scenario”, on which my work will focus, assumes that there is a underlying dimension of “less or more integration”, on which each actor has a unique preference (their preferences are single-peaked). These actors have Euclidian preferences, so that they prefer options closer to their ideal points. They are indifferent toward two alternatives which are equally distant to their ideal point. The Status Quo and both (the median member of) the European Parliament and the (median member of the College of Commissioner from the) Commission are on opposite sides of the integration-dimension, where the supranational actors favour integration the most^[1]. The Status Quo is assumed to be formed under the Luxemburg Compromise. Therefore, it’s at the point of the least integrationist Council member. Also, the actors have perfect information about each others’ preferences, at least at the final step of the EU-decision making (Tsebelis / Garret 2000: 17). To represent the supranational scenario in a simple way, Tsebelis reduces the 15 members of the Council with weighted votes to seven virtual players. The threshold is therefore approximately the same as saying that the third player on the integration dimension is the pivotal player. A pivotal player can turn a losing coalition into a winning coalition. Figure 1. gives an impression of the supranational scenario.

Abbildung in dieser Leseprobe nicht enthalten

Figure 1. The supranational scenario

1, 2, 3, 4, 5, 6 and 7 are members of a virtual seven-member Council. Q is the Status Quo under the Luxemburg Compromise. The Commission (Comm) and the European Parliament (EP) are on the most pro-integrationist side. Source: Tsebelis / Garret 2000: 16.

Even if Tsebelis’ model wasn’t accepted all over the political science field, the biggest debate arose when he declared that the “new” co-decision (I) procedure actually reduced the power of the European Parliament instead of increasing it^[2] (Kreppel 2002: 788). This is because, from a theoretical point of view, the Council could make a “take-it-or-leave-it” proposal to the European Parliament in the final stage of decision making. In fact, if there wasn’t agreement in the Conciliation Committee, the Council could return to the “common position”. The Parliament would not reject this proposal, because, again, it would prefer any agreement to no agreement at all. With the co-decision procedure after the reforms in Amsterdam, the Parliament became an “unconditional” veto-player and therefore a “co-equal legislator with the Council” (Tsebelis / Garret: 2000: 15).

The goal of my research question is simply to check how often Tsebelis’ assumption of the “supranational scenario” does occur in reality. Furthermore, I will attempt to distinguish if there is a difference in the frequency of the European Parliament and the Commission building the supranational scenario based on whether they decide on a new or an amended law, which procedure applies (consultation or co-decision) and what Council working rule applies (Qualified Majority Voting or Unanimity). Then I will look for a difference in the outcome taking into consideration whether there is a supranational scenario or none. Last, I will transpose the positions of the actors on the proposal-level to see if one can generalize my findings from the issue level on the proposal level.

2. Research Design: Dataset, Missing Values and the Construction of the Independent Variable

For my analysis, I will use the Decision-Making in the European Union (DEU) Dataset, which contains data for 66 proposals of the Commission. These proposals had to fulfil three criteria to be kept in the dataset: The procedure had to be either co-decision or consultation. The proposal had to be introduced before December 2000, and on the agenda in 1999 and/or 2000. In addition, the proposals had to have a minimum level of controversy. This political importance criterion was fulfilled when the proposal was mentioned in Agence Europe and the experts agreed on the controversy. The dataset contains the positions and the saliences of all 15 member states, the Commission and the European Parliament. For my analysis, I will just take into account the positions of the actors, as Tsebelis’ institutionalist theory does. The actors are defined as unitary actors with single peaked preferences. The positions were coded between 0 and 100, and for the analysis, it’s assumed to be a measurement scale. In fact, there are issues which are dichotomous, so which present two alternatives. Others have a ranked ordering, that means “closer to” or “further from”. The last positions can be placed on a measurement scale, with a natural point zero (Thomson / Stokman 2006).

As the independent variable I will use the Euclidian distance between the position of the European Parliament (EP) and the position of the European Commission (EC). This seems to be a simple and useful tool to aid my research. Additionally, this fits with the assumption that actors have Euclidian preferences and are indifferent between two alternatives which are equally distant.

First of all, one can note that for 53 out of the 162 issues the EP and the EC have the exact same position (the difference between the position of the EP and the position of the EC equals 0).

In other words, only one of three times (including missings) or in only 40 percent of the cases, the EP and the EC are exactly on the same position (excluding missings). As we see, there is a slight difference in the result, depending on whether I include or exclude missings. This causes – as so often in social sciences – a problem concerning how I will handle missing values.

One possibility would be to replace the missing values with the position of the Commission. There are two problems with respect to my research question. First, this method can be used for missing positions of member states, arguing that a missing one is equal to no position at all and that the state therefore would agree with the Commission’s proposal. The most famous example is Austria’s position on fishery^[3]. I don’t think this argumentation applies to the EP, because Members of the European Parliament (MEPs) come from all countries of the European Union (EU), with at least a majority – when not even an overwhelming majority – having a position on each specific topic. Second, if I’d simply replaced missing values for the EP’s position through the position of the Commission, it would have biased the results in favour of the supranational scenario, with more cases in which the position of the Commission is the same as the EP’s.

But, as discussed in König et al (2001), there could be a rational reason for actors to hide their positions on a specific topic. Contrary to the “fishery in Austria”-assumption, actors would not be without position, but have a position far away from the Status Quo^[4] (SQ). In fact, when near to the SQ, players can gain concessions from other’s which are positioned at the extremes, while such actors have to make offers in order to change the actual policy (König et al 2001: 13). With regard to this, simply dropping cases with missing values would also bias my results.

For my research question, the only cases I can drop without problems at the moment are those with missing values for both the EP and the EC. This is the case for the issues numbered n00358i5 (labelling of olive oil), n96115i2 (use of absorbents in fruit juice), n99072i2 (application of medicinal products for bovine and swine on horses), and n99072i3 (application of medicinal products for bovine and swine on minor species).

For the rest of the data, when the values are missing, I calculate the (virtual) position of the EP which is indifferent between the Commission’s position and the Status Quo, as proposed in Thomson / Stokman: “a good solution is the assumption that the indifferent actors are located half way between the reference point and the most favoured position of the […] agenda-setter” (2006: 52).

Therefore, the position of the EP represents the sum of the EC’s position and the position of the SQ for each issue, divided by two. There are four dichotomous cases which are influenced by this calculation. For d00067i2, n00127i3, n96114i3 and n98189i2, the results of the indifferent calculations would put the EP on the virtual position 50. As this is not possible, I’ll drop these four cases, because the results I’ll get should be interpretable in reality. There wouldn’t be a real compromise between two alternatives for dichotomous cases.

For my research on the supranational scenario, I should define what I mean by numbers when I say “the European Parliament and the Commission are located close together”. A simple and useful statistical tool is the median. As shown in the Whisker Box in Appendix A., the median lies on the value 22. From now on, when I say that the EP and the EC are “close together”, I will mean that the Euclidian difference between their positions is smaller than, or equals, 22 points.

How often can this first definitional requirement be found in the dataset? As defined by the use of the median itself, there are about 50 percent of the cases above and 50 percent below 22 points. The mean is 34,37, with a standard deviation of 36,57. This shows that there is a rather large scope of values which form this mean. In 38,36 percent of the time, the EP and the EC are on the exact same position, as the difference between their position is zero. In 15,72 percent of the time, the EP and the EC are on complete opposite sides, having a difference of a full 100^[5]. Definitely, there is a bias to the extremes because of the dichotomous cases that have been coded by either 0 or 100. But, excluding dichotomous cases gives a similar picture of the repartition: In 45,9 percent of the time, the difference between the EP and the EC is less than 22 points. In only 32,03 percent of the time, the two are on the exact same position, in 11,72 percent of the cases the difference is a full 100^[6]. Knowing that, I can conclude that even if the dichotomous variables bias the results to a minor extent, there is no need to exclude them from my analysis. Furthermore, I can, after this very first step, question the supranational scenario as an absolute axiom for the position of the EP and the EC for every single issue in European daily policy-making: The EC and the EP do not necessarily act together all the time!

But just to say that the EP and the EC are close together isn’t enough to construct my independent variable: It could be that these two actors actually have the same position, but just at the same position as the Status Quo. In order to exclude this, I have to take the distance of both the EP and the EC from the Status Quo into account. Only such cases, in which the EC and the EP act together and are far away from the SQ, can be considered as the supranational scenario.

For further analysis, I now have to assume that there is an underlying dimension of integration. This means, that 0 is much less integrative, 100 is highly pro-integrative. As so often in social sciences, this isn’t an absolute rule, but, as Thomson / Stokman noted, there is a “tendency for the Commission to be located at one end of the issue continua and the reference point at the other” (2006: 40). I subtract the position of the SQ from both the EP and the EC. This procedure will guarantee positive results when EP and EC are more integrationist than the SQ and I will get negative ones in cases when they want less than the actual SQ. There are seven cases in which the position of either the EP or EC is minor to the SQ, four in which they both are. None of these cases are dichotomous. Let’s have a look at theses issues to see how to deal with these cases. For d00032i3, d96161i2, d98325i1, d98325i2, d98325i3, n99235i2 and n99246i1, the SQ is higher than the position of at least one of the supranational institutions. For most of these proposals, I can’t find an underlying integration dimension as assumed above. In fact, there are some examples in which a lower position would be more integrative than 100^[7].

In order to define the criterion for both the EP and the EC being far away from the SQ, I take the median as the statistical frontier. My criterion “far from the SQ” is therefore: farther than, or exactly, 80 points for the Commission, and farther than, or exactly, 65 points for the European Parliament.

To summarize, the supranational scenario will be defined from now on as the following: (a) The difference between the positions of the European Parliament and the Commission is less than, or equals, 22 points. (b)

For the same issue, the distance from the Commission to the Status Quo is more than, or equals, 80 points. (c) Again for the same issue, the distance from the EP to the Status Quo is more than, or equals, 65 points.

In a second step, I will see if I can find some patterns in the data depending on different variables: First of all, are the EP and the EC more often on a similar position for new or amended laws?

3. Analysis

a. New or Amended Legislation

The first part of my analysis will deal with the following question: Is there a difference in the positioning of the EC and the EP depending on whether they decide on a new or an amended law? The underlying assumption I have for this chapter is that the dimension of “less or more integration” is substituted by another ideological (or Right-Left) one when there is already a legislation on the EU-level. One could e.g. imagine a priority-ranking of the European Parliament. As a unitary actor (defined as such in the database), the EP would acts in favour of integration first (that is, a position close to the Commission when the legislation is new, far away from the SQ), so that the policy-field is dealt with on the supranational level. By contrast, when there already is legislation which the EU-level deals with, the EP takes a substantive look at the specific issue and doesn’t necessarily follow the Commission’s proposal. In the actual literature, there is a tendency to think of a “Right-Left” or an ideological dimension when the Status Quo already represents EU-wide rules (Rittberger 2000: 558; Tsebelis / Garret 2000: 10).

Hypothesis 1.1. The European Parliament takes a closer position to the Commission for new laws than for amendments.

Hypothesis 1.2. If it follows the position of the Commission, the European Parliament takes a more pro-integrative position for new laws.

I do a cross tabulation of the Euclidian distance between the EP and the EC over the “New or Amended”-variable. I will first describe the results: The Euclidian distance between the EP and the EC is a maximum of 22 points (the definitional frontier of the EP and the EC being close together) in 52.63 percent of the cases for new legislation, in 42.22 percent of time for amended legislations. They are on exact the same position in 40,37 percent of the time when deciding on new laws, but only in 33,33 percent of time for amendments. As a fact, EP and EC act more closely together for new legislations. The EP and the EC are on opposite sides (more than 75 points) nearly one out of four times for new legislation, only 14,29 percent of the cases for amended ones.

Therefore, one can note that the EC and the EP act more often together when deciding on a new law than for amendments. The hypothesis 1.1. is therefore confirmed^[8].

Unfortunately, a t-test of the cross tabulation “difference of the EP and the EC” with the variable “New or Amended” gives a probability value (probability of error) of Pr = 0.117, so that I can’t propose a statement in a statistically significant way. In other words, there is about a 12 percent chance that the repartition between the position of EP and EC and the variables new or amended is a pure coincidence.

Now, I will check if there is an integration bias for these supranational actors. I will take a look at the distance of both the EP and the EC to the SQ, depending on the variable “New or Amended”. The control group is the mean distance of the EP and the EC to the SQ for all issues. The results are summarized in table 1.^[9].

Abbildung in dieser Leseprobe nicht enthalten

Table 1. Mean distance from the European Parliament (EP) and the Commission (EC) to the Status Quo (SQ) for new (New) and amended (Amend) legislation, and for both of them (All). Standard deviation in brackets, n is number of cases. Source: DEU, own calculations.

The Commission’s position doesn’t seem to be influenced by the fact whether the legislation is new or amended, as there is only a slight difference. In contrast, the EP’s position seems to be less integrationist when deciding on amended laws. The European Parliament is actually more pro-integrationist when deciding on new legislation. Therefore, my hypothesis on the positioning of the EP does fit with the quantitative findings. It seems that the MEPs choose the European way first – when there is a new law – and only look afterwards at the more substantial level, when there is an amended law. My hypothesis 1.2. is therefore confirmed.

For these findings, it may have been interesting to have data on the coalition which formed the majority. This would not only go beyond the scope of this research paper, it would also go against the assumption of the DEU Research Group of a European Parliament acting as a unitary actor. Further research could verify if there is a correlation between the coalition voting for the law and the more or less integrative character of the position of the EP. I’d suppose that the position of the Parliament is more integrative when the Grand Coalition (Party of European Socialists and European People’s Party – European Democrats) vote for the law. This would be a evidence for the MEPs not voting along a Right-Left continua, but along an Integration dimension.

Now, I will check how likely the supranational scenario is, depending on the procedure. Does another procedure matter for the positioning of the European Parliament?

b. The procedure and the Council voting rule

i. Co-decision vs. consultation

As mentioned above in the theoretical discussion, the question whether the legislation has to pass under the consultation or the co-decision procedure plays an immense role for the power of the EP. In fact, most scholars think of the EP being a co-legislator under co-decision, but a rather unimportant player under consultation. My assumption for this chapter would be that the EP takes a closer position to the position of the EC under consultation than under co-decision. My argument would be that the EP, which is just allowed to formulate an “opinion”, acts closer with the EC, because the Commission could accept little changes in its proposal even if, institutionally, it’s not obliged to. In that way, I would see the EP having a “power to delay” legislation (Hix 2005: 78; Tsebelis / Garret 2001: 372), because the Commission must wait for an “opinion” from the EP. This especially after the “isoglucose” ruling of the European Court of Justice, when it annulled a legislation because the Council didn’t wait for an EP’s “opinion” (Hix 2005: 77).

Amendments which clarify the legislation without changing the overall frame could be accepted from the EC (cf. Kreppel 2002). In contrast, I suppose that under co-decision, the EP’s position is farther away from the EC’s one as it can act as a co-legislator. The Commission, as only legislator with the Council under the consultation procedure, should be more integrative under this procedure. This would be a result from the first hypothesis. The Commission would foresee that the Parliament takes a position close to its own. Therefore, it would have a greater weight against the Council and would propose a more integrative regulation.

Hypothesis 2.1. The European Parliament takes a closer position to the Commission when acting under the consultation procedure.

Hypothesis 2.2. The Commission, and, if Hypothesis 2.1. is verified, the European Parliament, take a more integrative position under consultation.

I do a cross tabulation of both the difference between the position of the EC and the EP, depending on the procedure. Furthermore, I have a closer look at the frequency of the EC and the EP acting in a supranational scenario, as defined above. As one can easily see in Appendix C., there is a difference in the density of the frequencies of the EP and the EC depending on the procedure. For co-decision, the EP and the EC have a higher position as for consultation. The mean position of the EP is 70.73 points with a standard deviation of 37.47 for co-decision, and 60.02 (38.73, standard deviation, in brackets after the mean in the following) for consultation. Quite the same can be observed for the EC: It’s on average on position 71.01 with approximately the same standard deviation of 39.33 for co-decision, and 61.26 (40.40) for consultation. After this very first step, there is no noticeable difference between the institutions for the two procedures, the Commission being a little more integrative. But, for both EP and EC, the position tends more toward 100 when deciding under co-decision. These results falsify the hypothesis 2.2.

Mean position of

Abbildung in dieser Leseprobe nicht enthalten

Table 2. The mean position of the Commission (EC) and the European Parliament (EP), and the Status Quo (SQ), depending of the procedure: co-decision (COD) and consultation (CNS) and both of them (all). The mean position of the difference between the European Parliament and the Status Quo (|EP - SQ|) and between the Commission and the Status Quo (|EC - SQ|). Standard deviation in brackets, n is the number of cases. Source: DEU, own calculations.

As table 2. shows, the average difference between the positions of the EC and the EP is 38.15 for co-decision, 31.16 for consultation.

It therefore seems that the two bodies act closer together for consultation. To put it in other words, the EP tends to agree with the EC more when it has no power, and imposes its will when being a co-equal legislator. The hypothesis 2.1. is therefore confirmed.

Additionally, the average distance from the EP to the SQ is 67.80 when deciding under co-decision. In contrast, its mean is 50.91 under consultation. The mean distance to the reference point of the EC is 73,82 under co-decision, and 56,43 under consultation. Hence, there is a trend for both EP and EC to be more integrationist under co-decision than under consultation. This confirms the results I got above by just looking at their average position on the continua, without taking into account their distance to the reference point, like I did now. Again, this findings falsify my hypothesis 2.2.

In order to explain that, one could think of a Commission which doesn’t think that the EP always follows it. In other words, the Commission doesn’t see legislation as a one-shot game and doesn’t want to pull the legislation too much and lose the approval of the Parliament. In fact, that would considerably decrease its power against the Council.

ii. Co-decision vs. Consultation Depending on the Council Voting Rule

The fact that there is such a difference between the procedures for both the EP and the EC is quite interesting. But I think it would be a wrong interpretation just to say that when there is a co-decision procedure, the EP and the EC should be more integrative than under consultation. This can be shown in the data I based my research on: The cases chosen for the test of co-decision are mostly decided in the Council by Qualified Majority Voting (QMV). As a matter of fact, in 83.56 percent of the time, the voting rule is QMV under co-decision. This means that 62 out of 73 issues don’t need the approval of all member states in the Council. By comparison, in only in 53.93 percent of the cases dealt under consultation, the voting rule in the Council is QMV. For 43.82 percent of the time, consultation means unanimity^[10]. As a result: I must check if there is the same integration bias in co-decision procedures, but taking into account the Council voting rule this time. My hypothesis for this sub-chapter is: When deciding on a legislation which the Council has to accept by unanimity, both the EP and the EC are less integrationist than when the Council votes by QMV. With this, I follow Steunenberg / Selck (2006: 64) who estimate the Council unanimity win set as being smaller than the Council’s qualified majority win set. As the preferences are Euclidian and single peaked, the Commission can make a proposal which the Commission (in fact, the median voter of the College of Commissioners) could accept. This is the well-known power of agenda setter, which the Commission is.

Hypothesis 2.3. Under all procedures, the Commission’s position is less integrative when the Council voting rule is unanimity.

To see if the actual procedure matters, I must see the positions of EP and EC for each of the procedures of consultation and co-decision, this for both Council vote rules QMV and Unanimity. Table 3. summarizes my results.

Abbildung in dieser Leseprobe nicht enthalten

Table 3. The mean position of the Status Quo (Mean SQ), the European Parliament (Mean EP) and the Commission (Mean EC), their mean distance to the Status Quo and the difference between their position (Mean |EC-EP|) depending on the voting rule in the Council: Qualified Majority Voting (QMV) and Unanimity (Unan) and on the procedure: co-decision (COD) or consultation (CNS). Standard deviation in brackets, n is the number of cases. Source: DEU, own calculation.

As one can easily see, the average distance between the Commission and the Status Quo is not really affected by the Council voting rule for both procedures. Nevertheless, the Commission’s distance to the SQ is a little farther under unanimity for both procedures. This falsifies my hypothesis 2.3. As I now checked the possible correlation with the Council voting rule, I can conclude that the EC act more in favour of integration under co-decision than under consultation.

This conclusion can be enforced when looking at the standard deviations, too: For the co-decision procedure under both voting rules in the Council, the standard deviation is smaller than it is for consultation. This means that the distances from the Commission to the Status Quo are located closer around the mean for co-decision as for consultation. But, a look at the means of the distance from the Parliament to the Status Quo shows us another picture of the repartition. The average distance between the EP and the SQ is quite high for co-decision under QMV. It even fulfils the “Far away”-criterion from my operationalization of the supranational scenario. Compared to the consultation procedure, one can note a big difference in the average distances to the SQ. But when looking at the differences within the procedures when the Council decides by unanimity, the contrary occurs. The average distance from the EP to the SQ is lower under co-decision than under consultation. It therefore seems that, when the Council decides by unanimity, the EP is more integrationist under consultation, which is exactly the contrary as under QMV. But these results shouldn’t be interpret to early as there are only 12 cases for which the Council voting rule is unanimity and the procedure is co-decision. The results aren’t therefore reliable at all.

Next, it is noticeable that under consultation, the EP and the EC act much closer together when there is unanimity rule in the Council than QMV. The average difference nearly fulfils my “close together” criterion of 22 points. This could be interpreted as following: Having no institutional power, the European Parliament enforces the position of the Commission more for unanimity, as it puts pressure on the least integrationist member of the Council.

In contrast to this, when there is the co-decision procedure, the supranational actors act closer together under QMV than under unanimity^[11]. Additionally, the EP is much farther away from the Status Quo for that constellation. An explanation would be quite intuitive: The win set for QMV voting is a lot bigger under QMV. Therefore, the EP can take a position which is much more integrative, because, one shouldn’t forget, it’s an “unconditional” veto-player under co-decision (Tsebelis / Garret 2001: 373-74).

Anyway, even if one doesn’t agree with the explanation, it is noticeable and it should be remembered that the Council voting rule has an influence on the positioning of the European Parliament. When doing a linear regression for the difference between EP and EC controlling for the Council voting rule, the procedure, and the interaction between the theses two variables, one will find that there is a weak explanation force of the results^[12]. Neither the Council voting rule nor the procedure as such explain the differences in the positioning of EP and EC. The interaction between these two variables explain at least a little. The adjusted R equals 0.0372 for this last case. This means that only 3.7 percent of the variance of the average difference between the EP and the EC is explained by the Council voting rule and the procedure all together.

Anyway, only this constellation gives results which are statistically significant at a p=0.01 level. But, I should note here that the results of this linear regression must be taken into account with a lot of caution. The difference between the EC and the EP isn’t normally repartitioned, its curve isn’t gaussian^[13], which is a condition for doing a linear regression. Also the variables “Council voting rule” and “ Procedure” have dichotomous parameter values.

iii. Summary

One should remember that (a) the Commission is more integrative than the EP. (b) Under co-decision, the EC is more integrative than under consultation. (c) The EP is more integrative under co-decision, but only when there is QMV (one shouldn’t forget the fact that there are only 12 cases, which turn this into a non-reliable statement). (d)

The Council voting rule plays an important role, as the positioning of the EP differs heavily on whether it is unanimity or QMV. (e) Even if the conditions for a linear regression weren’t fulfilled, one should note that only the interaction between the Council voting rule and the procedure gives statistically significant results when looking at the difference in the positioning of the EP and the EC.

c. The Supranational Scenario and the Outcome

The next step in my analysis will be to see whether there is or is not a difference in the outcome when the EP and the EC act together in a supranational scenario. My assumption will be quite intuitive, namely that when acting together, the EC and the EP get a result which is closer to their positions than when not acting in the supranational scenario.

Hypothesis 3.1. The outcome is closer to the positions of the Commission and the European Parliament when they form a supranational scenario.

I defined the supranational scenario in a quantitative manner as: (a) The difference between the positions of the EC and EP equals or is less than 22 points and (b) the distance to the SQ is greater than the median (|SQ-EP|>=65; |SQ-EC|>=80). To see differences, the control group will be all possible constellations.

The control group of all outcomes has a mean position at the point 55.11 (33.78). The average of the positions of the outcomes in cases where there is a supranational scenario is 58.81 (36.66). As a result, this is a slight proxy for outcomes being more supranational when the EC and the EP act together.

Here, I should fall back upon the skewness of the repartition in my analysis. The skew of the outcome is -0.4029 when the EP and the EC act in the supranational scenario. This means that, when drawing the repartition of the position of the outcomes, the left side of the graph is more even than the right side. In comparison to the skew of -0.2328 for my control group, this simply implies that, when having a supranational scenario, there will be more outcomes closer to 100 than when the EC and the EP do not act together.

In order to see how the Commission and the European Parliament influence the outcome, I take the absolute value of the distance between both the EP and the EC to the outcome. The reason why I take the absolute value is that, as defined above in the theoretical part of this paper, the actors are assumed to be indifferent towards two alternatives which are equally distant. The histograms in Appendix H. give an impression of the repartition of the distance of the EP to the SQ and of the EC to the SQ.

As one can see, there is a trend for both the EP and the EC to be winners. Note that at the two ends of the continua, there is a statistical artefact due to the dichotomous cases. The average difference between both the EP and the EC and the outcome is respectively 34.22 (35.20) and 34.05 (33.03) in all cases. These are my control groups. When acting in the supranational scenario, the average difference in the distance from the EP to the outcome is 31.55 (37.42) and 31.75 (37.31) for the EC. The difference with the control group isn’t immense, but there is a trend to see my hypothesis being confirmed. In fact, the distance to the outcome is lower for the supranational bodies when they act in the supranational scenario. My hypothesis 3.1. is therefore confirmed.

But I also have to take the size of change into consideration for each issue. The “size of change” is the difference between the old and the new Status Quo. To put it into other words, I have to see if the distance between the reference point and the outcome is bigger when there is the supranational scenario. I will take the absolute values and drop the missings. The mean size of change is 51.81 (35.14). This is my control group. For the issues where there is a supranational scenario, the average size of change is 61.91 (37.60). Thus, one can note that there is a significant difference in the change of legislation when the two bodies act together. In that way, the outcome is closer to the position of the EC and the EP when they act in the supranational scenario, and therefore, it’s more pro-integrative.

d. The Proposal Level

All my findings were based on the underlying assumption that the issues were independent from each other. But this is in fact quite a strong statement, as day-to-day politics have a lot to do with logrolling and issue-linkage. And even more because the issues are dealt with within a specific proposal. To avoid criticism on this assumption, I will now see if my findings on the issue-level correspond to the findings on the proposal level.

Hypothesis 4.1. One can consider an independence of the issues.

In order to reach the proposal-level, I take the square root of the sum of the squared distances from both the EP and the EC to the SQ over the number of issues. With the same procedure, I get the distance from the EP to the Commission in up to a six-dimensional space. I drop the cases in which there are missings for one or more issues. After doing this, I still have data for 50 proposals on the proposal-level. But the results aren’t interpretable yet because of the different dimensionalities. In one dimension, the maximum distance between two points is 100. In two dimensions, the maximum is already about 141.42, or √2 times 100, as the two axes of the spaces form a square of 100 times 100. In the third dimension, the max is 173.21, or √3 times 100, because, once again, the cubic is 100 times 100 times 100. This continues the higher the dimensionality gets. Therefore, to get comparable results, I have to make the findings uniform with √n, in which n is the number of dimensions (that is, issues)^[14].

After having done that, I calculate the correlation coefficient between the issue- and the proposal-level: For (a) the distance between the EP and the EC, for (b) the distance from the EP to the SQ and for (c) the distance from the EC to the SQ. The results are summarized in table 4.

Abbildung in dieser Leseprobe nicht enthalten

Table 4. The correlation coefficient between the issue- and the proposal-level for: the distance between the Commission and the European Parliament (|EC-EP|), the distance between the Commission and the European Parliament to the Reference Point (respectively, |EC-SQ| and |EP-SQ|). Souce: DEU, own calculations.

As one can see, all correlation coefficients are higher than 0.7, which is the usual value in the social sciences above which one can speak about a high correlation between two variables. I can conclude that my underlying assumption of an independence of the issues wasn’t that wrong. Also, the reader will note that the correlation coefficients imply that I can generalize my findings from the issue dimension on the proposal level as well – even if there is not a perfect correlation to one. Anyway, all the calculations to reach the proposal-level were important to countering possible criticism on the independence of issues assumption.

Conclusion

The aim of my analysis in this paper was to find patterns in the DEU dataset regarding the supranational scenario. Therefore, most of this paper was in an explorative shape. For each chapter, there was a hypothesis derived directly from the theoretical part. The overall research questions were: Is there empirical evidence supporting the assumption that the European Parliament and the Commission are located closely together in their preferences for concrete policies? Are these positions generally on the opposite side of the Status Quo? And are there some factors which influence the likelihood that the two supranational actors will be positioned more closely together and far away from the Status Quo? Does the supranational scenario influence the outcome?

In the theoretical part, I discussed what the supranational scenario is. I took George Tsebelis’ definition into account. To summarize, the supranational scenario assumes that the European Parliament and the Commission are located on the pro-integrationist side of an integration dimension. The Status Quo is located at the other end, at the point of the least integrationist member state of the Council. Directly from this definition, I derived the operationalization of the independent. The analysis started with an overall view: I could question the supranational scenario as an absolute axiom of everyday politics in the EU.

Then, I had a look at whether the supranational scenario occurs for new and amended legislation. I stipulated that MEPs chose the European way first, when deciding on a new legislation. I supposed that an ideological dimension overrides the integrational one when deciding on an amended law. These theses were confirmed by my quantitative analysis. In addition to that, I could observe that the Commission’s position isn’t influenced by either new or amended laws.

Then, I looked at the frequency of the supranational scenario depending on the procedure. My hypothesis for this chapter was that the European Parliament is located closer to the Commission under consultation. This would be because is has no power under this procedure. I confirmed this hypothesis and also found that the Commission and the Parliament are more integrationist under co-decision. This went against my second hypothesis for this chapter.

After that, I had to look for these procedures, but depending on the Council voting rule. The Commission acts more in favour of integration under co-decision. I also found that under consultation, the EP and the EC act much closer together when there is unanimity rule in the Council. In contrast, under co-decision they act closer together under QMV than under unanimity.

Afterwards, I had a look at the outcome depending on whether there is a supranational scenario. My hypothesis was that the outcome is closer to the positions of the Parliament and the Commission when they act in a supranational scenario. This thesis was confirmed statistically. The legislation changed more under the supranational scenario.

In the last step, I had to see if I could transpose my results from the issue- on the proposal level. After having calculated the positions for the different dimensions, I calculated the correlation coefficient between values from the issue and the proposal level. My results were that the two units of analysis are highly correlating. Therefore, I concluded that one can generalize the findings from the issue level to the proposal level.

To conclude, one should remember from this paper that one can’t simply assume the supranational scenario for each piece of legislation. But, in fact, the supranational scenario is more likely to occur for new legislation. It also occurs more often under consultation when there is a unanimity rule in the Council and under co-decision when there is QMV. Also, when acting in a supranational scenario, the legislation changes more toward more integration.

Literature

Crombez, Christophe / Steunenberg, Bernard / Corbett, Richard. 2000. “Understanding the EU Legislative Process. Political Scientists’ and Practitioners’ Perspectives.” European Union Politics 1 (3): 363-381.

Hix, Simon. 2005. The Political System of the European Union. 2nd edition. London: Macmillan.

König, Thomas / Finke, Daniel / Daimler, Stephanie. 2001. “Ignoring the Non-ignorables?: Missingness and Missing Positions.” European Union Politics 6 (3): 266-89.

Kreppel, Amie. 2001. “Moving Beyond Procedure: An Empirical Analysis of European Parliament Legislative Influence.” Comparative Political Studies 35: 784-813.

Steunenberg, B. / Selck, T. J. 2006. “Testing procedural models of EU legislative decision-making.” In: Robert Thomson et al (ed.). The European Union Decides. Cambridge: Cambridge University Press: 54-85.

Rittberger, Berthold. 2000. “Impatient legislators and new issue-dimensions: a critique of the Garret-Tsebelis ‘standard version’ of legislative politics.” Journal of European Public Policy 7(4): 554-575.

Thomson, Robert / Stokman, Frans N. 2006. “Research Design: Measuring Actors’ Positions, Saliences and Capabilities.” In: Robert Thomson et al (ed.). The European Union Decides. Cambridge: Cambridge University Press: 25-53.

Tsebelis, George / Garrett, Geoffrey. 2000. “Legislative Politics in the European Union.” European Union Politics 1 (1): 9 –36.

Tsebelis, George / Garrett, Geoffrey. 2001. “The Institutional Foundations of Intergovernmentalism and Supranationalism in the European Union.” International Organization 55 (2): 357-89.

Tsebelis, George / Kreppel, Amie. 1998. “The history of conditional agenda-setting.” European Journal of Political Research 33: 41-71.

Appendix

Appendix A.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2. Whisker box for the difference between the position of European Parliament (EP) and the Commission’s (EC). Source: DEU, own representation

Appendix B.

Abbildung in dieser Leseprobe nicht enthalten

Figure 3. Histogram of the repartition of the Euclidian distance between the European Parliament (EP) and the Commission (EC) in steps of 10. Source: DEU, own representation.

Appendix C.

Abbildung in dieser Leseprobe nicht enthalten

Figure 4. Whisker Boxes for the position of the European Parliament (EP) and the Commission (EC) depending on the procedure: consultation (CNS) or co-decision (COD). Source: DEU, own representation.

Appendix D. How to deal with negative distances to the Status Quo?

For d00032i3, the issue was: Can the public have direct access to documents on terrorism and defence? The EP had the position 0, what meant in favour of full access. SQ=100 meant that only a small group should decide which documents are secret and which are not. Therefore, I can take the absolute value of the difference between the Status Quo and the position of the EP.

For d96161i2, the issue was: Should it be a notification time for consumers to inform producers about a fault in a product? EP=0 meant: There should be no restrains. SQ=50 meant, that the state could decide. Here, we can see that the position 0 is in favour of harmonisation. Therefore, I will take the absolute Euclidian distance.

For d98325i1, the issue was: Which law should be applied for e-commerce? EC=0 meant that there should be the “Country of Origin”-rule, which would harmonize the Internal Market. The SQ=100 meant, that International Law should be applied, which is more difficult for the construction of the Internal Market. Here, the absolute value of the distance between EC/EP and the SQ would give a better picture, as 0 is more integrative.

For d98325i2, the issue was if there should be exceptions to the regulation from d98325i1 (the first issue of the same proposal). Again, more integration (that means, no exceptions) was coded by 0. Again, I take the absolute value of the distance.

For d98325i3, the issue was on the liability for the contents and dealings on the internet services which provider of e-commerce provide. The Commission was in favour of no liability for neither hosting, mere nor catch illegal content. This because the control isn’t technically possible at this moment. As there is no underlying integration dimension here, I can take absolute values.

For n99235i2, the issue was the timing of the liberalization of banana imports, after the transitory quota-rule ran out. The SQ=100 meant: No agreement can be found, so the transition quota-rule keeps running. But a look inside the EP shows that the working group on that question was lead by a French and two Spanish MEPs. France and Spain would retard the new timing as long as possible, as banana imports are important to them. So they chose the EP=75 2010 year for liberalization, which was the latest date they could get. They would have preferred the intermediate solution of quotas as these protected their industries. Therefore, I can take the absolute value for the distance from EP to SQ as it’s a coding problem and not a theoretical one.

For n99246i1, the issue was on the part of financing of member states in the milk for schools program. EC=50 wanted 50% paid by EU-funds, against the SQ=100, 100%. Here, there is an Integration Dimension and the argumentation for me taking the absolute distance is a little more tricky. But: The EC tries to achieve multi-level governance. In that way, the co-financing of the milk program at schools would oblige the member states to coordinate and in that way to get integrated in EU-politics a little more. In addition, one should note that, as the budget is fixed, the gain of 43 million Euro could be spent on a new program, which would be worked out as usual in cooperation with the member states and their regional administration. Therefore, I will take the absolute value of the Euclidian distance between EC and SQ for n99246i1, too.

Appendix E.

Abbildung in dieser Leseprobe nicht enthalten

Figure 5. Whisker boxes of the difference in the positions of the European Parliament (EP) and the Commission (EC) depending on the procedure: consultation (CNS) or co-decision (COD). Source: DEU, own representation.

Appendix F.

Abbildung in dieser Leseprobe nicht enthalten

Table 5. Absolute and relative frequency of Council voting rule: Qualified Majority Voting (QMV) or Unanimity depending on the procedure: consultation (CNS) or co-decision (COD). Source: DEU, own calculations.

Appendix G.

Value of the adjusted R for the Mean |EC-EP|

Abbildung in dieser Leseprobe nicht enthalten

Table . Value of the adjusted R, taking into account the number of cases, for the linear regression of the mean position between the European Parliament and the Commission (Mean |EC-EP|), controlling for the Council voting rule, the procedure and the interaction between the Council voting rule and the procedure (Council voting rule x Procedure). ***: p<0.001; **: p<0.01; *: p<0.05; ~: p>0.5. Source: DEU, own calculations.

Appendix H.

Abbildung in dieser Leseprobe nicht enthaltenAbbildung in dieser Leseprobe nicht enthalten

Figure 6. Histogram of the frequency of the distance from the Commission to the Outcome (left) and the European Parliament to the Outcome. Source: DEU, own representation.

[...]

---

^[1] By the way, in Tsebelis’ representations of this one dimension, he places the European Parliament more integrationist then the Commission (1998: 44). As the reader will see below, my analysis will show that for the cases studied, the Commission tends to be the more integrationist one of the two bodies.

^[2] See for example Crombez et al. 2000

^[3] Austria doesn’t have any position in fishery, as it has no coast with an ocean or a sea.

^[4] I will use the words Status Quo and reference point interchangeably in the following, as there is only a little conceptual difference.

^[5] The histogram in Appendix B. gives an impression of the frequency of the difference in position of the EP and the EC.

^[6] The median of the repartition of the difference between the positions of the EC and the EP is 25. But I see no need to change my first definition, fixing the virtual frontier on 22.

^[7] See discussion in Appendix D.

^[8] An other possibility for verifying this difference is to use the skewness for both the EP and the EC, depending on whether the law is new or amended. The skewness of the Euclidian distance between EP and EC is 0.7178 for new legislation, 0.6268 for amended ones. The skew I take as a reference is the one for all legislations, both new and amended. This skewness is 0.7027. As a result, for new laws, the EC and the EP tend to be a little closer together (because the modus is left from the mean), while they’re less for amended ones. But the difference between these skews are too small to make any statistical significant statement.

^[9] I should note here that the mean position of the SQ is 13.27 (29.87) for new, 6.07 (22.07) for amended laws. Therefore, there is an average difference of about seven points between the SQ of a new and the one of an amended legislation. This finding is counterintuitive, but it’s resulting from the higher dimensionality of the new proposals.

^[10] See Appendix F.

^[11] One should note here that the findings concerning the QMV Councill voting rule are even more important nowadays. Ín fact, the scope of QMV was extended with the reform of Amsterdam. But the DEU dataset took into account only proposals for which the procedure wasn’t affected by the treaty reform.

^[12] See Appendix G.

^[13] See Appendix B.

^[14] The same would have be to divide the sum of the squared distances with the number of dimensions before taking the square root. For reasons of simplicity in programming, I chose the method described below. This is no problem because √x / √y = √(x / y).