In recent years, the decision-making process in the Council of the European Union became the focus of interest for an growing number of scholars. However, it appears to be difficult to achieve the degree of detailed information that is needed to understand how this institution takes its decisions, and it seems to be difficult to obtain specialized knowledge about what really goes on behind the Council's closed doors. Yet such information on actors' policy positions would represent crucial input for studies that attempt to test explanatory models of how the Council takes its decisions.Footnote ¹ The focus of interest for research on EU politics should not be limited to further analyses of existing data. New research in the field should rather attempt to increase our knowledge of the formal and informal processes of Council decision-making by improving data quality.Footnote ² Such information could then be used both to assess performance of existing models and to create better theoretical frameworks for understanding how the European Union works. New data-generation techniques could therefore open up a completely new dimension for research in this field.Footnote ³

Jonathan Sullivan and Tim Veen,Footnote ⁴ in a review article recently published in this journal, address this challenge. They argue that data-generation processes for studies of decision-making in the EU Council might be significantly improved by linking EU documents with text analytical tools. The authors aim at ‘shedding light’ on the Council, which they label an ‘opaque institution’.Footnote ⁵ They refer to the weaknesses of the two main approaches in the field, applied theoretical models and voting studies. According to Sullivan and Veen, existing weaknesses could be overcome through improved data quality resulting from the application of new methods of data-gathering. They first provide a brief review of the state of affairs in the literature by comparing formal bargaining models and voting models of Council decision-making. Both methods, they assert, have problems regarding their explanatory power that are ‘the result of difficulties researchers have faced in generating appropriate data’.Footnote ⁶ Sullivan and Veen then suggest that researchers might benefit from two recent developments: an increasing availability of primary sources (e.g. Council minutes, Eur-Lex, etc.) and the possibility of applying new data-generation techniques based on document analysis.

Although Sullivan and Veen's article attempts to introduce new ways of studying decision-making in the Council, it falls short of spelling out what type of data to use to enable researchers to employ these techniques. Despite their increasing availability, the sole use of EU primary sources is still too limited to conduct quantitative studies oriented towards testing bargaining models, because they do not report on the actors' policy positions. Our article demonstrates which other textual resources might be used. To fill the gap in the literature, we suggest the use of newspaper analysis. EU-related studies hitherto only use newspapers to investigate the bias of media on political behaviour of the European public, for example citizens' attitudes towards the EU and European integration, or the development of a European public sphere.Footnote ⁷ In contrast to these approaches, we regard newspapers as a proxy of what really happened during the bargaining process. Newspapers are a readily available body of non-political text and are a sustainable source of information to derive the policy positions of political actors.Footnote ⁸ We suggest linking procedural information issued by the EU Council and the other institutional actors with data gained from newspapers to enhance data quality. Improved data on changes in actor positions will help us to understand better the bargaining and voting processes in the Council and EU politics in general.

Various formal models of EU decision-making have been developed, yielding a remarkable degree of sophistication in the theoretical literature. Empirical testing, however, as suggested by Hörl et al., has largely been unsuccessful.Footnote ⁹ For models that were empirically evaluated, a substantial mismatch between theoretical predictions and empirical outcomes has been observed, possibly because of a lack of accurate data. Overall, formal studies on EU decision-making suffer both from a scarcity of empirical testing designs and from poor model fit.Footnote ¹⁰

Most of the decision-making models that have been put forward to study the EU are based on rational choice theory. The models assume that all the actors involved in a bargaining situation seek to further their own interests. These in turn determine their negotiating behaviour. Linking the two stages of first negotiating a bill and then voting on it in the Council would require information on the actors' policy preferences. Voting studies only take into account the final stage of the decision-making process. They neglect both actors' preferences and their initial strategic positions and disregard the extent to which these might have changed during the course of the bargaining process. Hence these studies do not explain whether and to what extent voting outcomes are affected by bargaining dynamics. In order to link these two, information is needed about actors' ideal preferences, their revealed positions, the strategies involved (e.g. issue linkage) and their eventual voting behaviour. Only if this type of information is available will it be possible to test rational choice models of Council decision-making and to put forward theories informed by both the bargaining and the final voting stages.

Sullivan and Veen refer to the increasing availability of primary documents issued by the EU itself as a promising way of generating new data. It is true that the European Parliament and the Council are subject to increased transparency.Footnote ¹¹ Two important sources of legislative process data issued by the institutions are the Council's Public Register and the database Eur-Lex.Footnote ¹² The Council's Public Register comprises the Council minutes, which record how the individual member states voted on legal acts put before the Council. These minutes, as pointed out by Hagemann,Footnote ¹³ provide information on a large number of procedural detail: the type of legal procedure, the introduction date of the bill, the adoption date, the policy area, the involvement of preparatory bodies and the Commission, the title of the proposal, details about the policy content, the inter-institutional reference number, the Sectoral Council, the stage of the legislative process when the vote was taken, the stage of the legislative process when the proposal was adopted, the identity of the member holding the presidency, each member state's decision to support, abstain, or oppose a bill, and even formal statements added by individual states. The Public Register also provides monthly summaries of the Council acts. These include details about the legislative acts adopted, the applied voting rule and the results of the voting. Apart from the Council's Public Register, the Eur-Lex database is a legal database that reports on inter-institutional procedures.Footnote ¹⁴

These online resources undoubtedly provide valuable information on the EU legislative process. However, the Council's Public Register only provides summaries of the decisions taken. In most cases these are no longer than two sentences at most, and do not represent transcripts of the discussions. The information in Eur-Lex refers to the application of particular procedures, voting rules in specific sectors and general trends about the number or type of legislative decisions over time.Footnote ¹⁵ Both sources lack information regarding member states' policy preferences and how these have shifted over time, and are therefore inadequate to test bargaining models.

In addition to the described sources, we suggest the systematic use of newspaper articles to identify and track the policy positions of the EU members represented in the Council. Newspapers offer a vast body of non-political text on the EU legislative processes.Footnote ¹⁶ They can be used to reveal details of the day-to-day decision-making processes in the Union, thereby filling the gap left by relying too heavily on the textual sources that have previously been employed in the literature. While national newspapers have undoubtedly the potential to present different versions of national bargaining positions, most of them report rather sporadically on the EU decision-making process. We therefore suggest focusing on those newspapers that specifically concentrate on the daily workings of the Union. Some of the most prominent of these are Agence Europe, the ‘Brussels’ section of the Financial Times, Europolitics and the Economist Group's European Voice. If newspaper articles – coupled with procedural information taken from primary documents issued by the EU institutions – were to be analysed by means of electronic content analysis, a hitherto unknown depth of understanding of the internal dynamics of the EU decision-making processes could be achieved.

The application of analysis of newspapers faces several challenges. So far, political studies that apply newspaper analysis to EU affairs have mainly done so by manually collecting relevant articles and by more often than not failing to spell out what precisely the sampling strategy was. This selection is prone to bias since it is based on an often-subjective distinction between important and less important cases. The literature on lawmaking in the US Congress has been facing similar challenges and can provide useful guidelines for identifying and selecting cases. David R. Mayhew has defined legal acts as important when they are both ‘innovative’ and ‘consequential’.Footnote ¹⁷ Judgements of observers and participants in the process can be used to indicate the extent to which an act is innovative and consequential, and hence important.Footnote ¹⁸ Another indicator refers to whether – and if so, how often – the negotiations surrounding a particular legal act are covered by major media outlets.Footnote ¹⁹

A further issue regarding newspaper coverage of political affairs concerns objectivity. Recent work in this field, focusing on the US media market, discusses bias and how it can be overcome.Footnote ²⁰ Mullainathan and Shleifer discuss possibilities for gaining unbiased information from newspapers. They suggest that a hypothetical reader who reads many different newspapers' coverage of the same event should get a fairly objective idea of what ‘actually happened’. The different media sources would offset each other, and a relatively unbiased aggregate outcome should occur as a result.Footnote ²¹ Following this reasoning, newspaper-based text analysis should include as many different newspapers as possible in order to approach objectivity. In the context of the EU decision-making process, this means that studies should not solely be based on single EU bulletins; in order to circumvent bias, it would be crucial to include other newspaper sources too. If applied properly, newspaper analysis is able to overcome problems that are sometimes associated with it. It has the potential to provide the type of valid and reliable data that is needed to improve model-testing and our understanding of EU affairs.

The example of the EU Services Directive helps to illustrate the benefits of employing newspaper analysis for gaining information on the EU bargaining process. We chose this directive because many observers, participants and journalists alike, have declared it to be one of the most crucial EU legal acts ever negotiated. Arlene McCarthy, an MEP, referred to it as ‘the single most important and most disputed piece of legislation in the EU’.Footnote ²² During the Finnish EU presidency, Trade and Industry Minister Mauri Pekkarinen praised the compromise on this directive as ‘historic’. Internal Market Commissioner Charlie McCreevy stressed the importance of the directive for all EU citizens and businesses.Footnote ²³ The salience of the Services Directive also becomes apparent through the number of articles published in the Economist newspaper, which ran no fewer than 48 articles on this single measure. The Services Directive aims at further liberalizing the internal market for the European service sector. It allows service providers to offer their services in any country of the EU without facing any legal or administrative barriers. At the heart of the directive are two fundamental freedoms enshrined in the EC Treaty: the freedom to establishment and the freedom to provide services.Footnote ²⁴ The negotiation and adoption of the Services Directive was a long and arduous process. However, it was not a secretive one. Newspapers reported extensively on the actors' preferences, their expectations, the choices they made, the obstacles that needed to be overcome and the final voting outcome. An article published in Europolitics illustrates and nicely sums up the kind of information provided by newspapers in the course of negotiating the directive:

This short passage reveals substantial pieces of information on the bargaining situation for this particular case. It highlights the coalitions that have developed in the course of the process, and it identifies the different blocks' positions in relation to the European Parliament. This brief example shows that newspaper articles can be a vital source of information on the bargaining process of EU decision-making. To date, this source is dramatically underexploited.

Researchers studying the EU legislative processes need to move on by opening up new data sources. The systematic analysis of newspaper articles, linked to the use of primary documents issued by the Council and by the other institutions, presents an important method of significantly improving the data quality and furthering our understanding of how the EU really works. The Council of the European Union is not such an opaque institution after all. The data is out there; all we have to do is use it.