German ‘LexIconSpace’: Policy Positions and their Legislative Context

Abstract

This article presents a new approach for estimating the policy positions of political actors in the German multi-party policy space. The approach consists of two steps, ‘smart tagging’ in the data generation process and Bayesian factor analysis in the estimation process. ‘Smart tagging’ relates the statements of political parties and governments to the keywords of German federal legislation, which we use to estimate the policy positions in portfolio-specific n-dimensional policy spaces. Our G-LIS approach (German ‘LexIconSpace’) provides several advantages for scholars evaluating policy-seeking theories, in particular by providing context-related variation of policy positions across portfolios and over time. Our findings for the portfolio of ‘labour and social policy’ reveal a two-factor solution which unfolds a latent ‘resource’ and ‘value’ dimension in Germany during the period from 1961 to 2009. We find changes in the policy positions of German political parties and governments, which existing approaches can hardly identify in n-dimensional spaces under the specification of the error term for each dimension and actor.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the support for this research provided by the German Research Foundation (DFG) and the Mannheim Centre for European Social Research (DFG-project ‘Policy Change and Reform: The Determinants of Success and Duration of German Legislation between 1961 and 2005’, KO 1415/9-1).

 The authors also owe a considerable debt of gratitude to Daniel Finke for delivering the Gauss code for the Bayesian factor analysis.

Notes

See Reimut Zohlnhöfer, Die Wirtschaftspolitik der Ära Kohl. Eine Analyse der Schlüsselentscheidungen in den Politikfeldern Finanzen, Arbeit und Entstaatlichung 1982–1998 (Opladen: Leske und Budrich, 2001); Christoph Egle, Tobias Ostheim and Reimut Zohlnhöfer (eds.), Das rot–grüne Projekt: Eine Bilanz der Regierung Schröder 1998–2002 (Wiesbaden: Westdeutscher Verlag, 2003); Manfred G. Schmidt and Reimut Zohlnhöfer (eds.), Regieren in der Bundesrepublik Deutschland. Innen- und Außenpolitik seit 1949 (Wiesbaden: Verlag für Sozialwissenchaften, 2006); Christoph Egle and Reimut Zohlnhöfer (eds.), Ende des rot–grünen Projektes: Eine Bilanz der Regierung Schröder 2002–2005 (Wiesbaden: Verlag für Sozialwissenschaften, 2007).

See Hans-Dieter Klingemann, ‘Election Programs in West Germany: Explorations in the Nature of Political Controversy’, in Ian Budge et al. (eds.), Ideology, Strategy and Party Change (Cambridge: Cambridge University Press, 1987), pp.294–323; Hans-Dieter Klingemann and Andrea Volkens, ‘Coalition Government in the Federal Republic of Germany: Does Policy Matter?’, in Michael Laver and Ian Budge (eds.), Party Policy and Government Coalitions (Basingstoke: Macmillan, 1992), pp.189–222; Thomas König, Andrea Volkens and Thomas Bräuninger, ‘Regierungserklärungen von 1949 bis 1998. Eine vergleichende Untersuchung ihrer regierungsinternen und -externen Bestimmungsfaktoren’, Zeitschrift für Parlamentsfragen 30/3 (1999), pp.641–59; Thomas König, Till Blume and Bernd Luig, ‘Policy Change without Government Change? German Gridlock after the 2002 Election’, German Politics 12/2 (2003), pp.86–146; Susumu Shikano and Franz U. Pappi, The Positions of Parties in Ideological and Policy Space: The Perception of German Voters of their Party System (Mannheim: Mannheimer Zentrum für Europäische Sozialforschung, Working Paper No. 73, 2004); Franz U. Pappi and Susumu Shikano, Ideologische Signale in den Wahlprogrammen der deutschen Bundestagsparteien 1980 bis 2002 (Mannheim: Mannheimer Zentrum für Europäische Sozialforschung, Working Paper No. 76, 2004); Simon Franzmann and Andrè Kaiser, ‘Locating Political Parties in Policy Space. A Reanalysis of Party Manifesto Data’, Party Politics 12/2 (2006), pp.163–88; Eric Linhart and Susumu Shikano, Die Generierung von Parteipositionen aus vorverschlüsselten Wahlprogrammen für die Bundesrepublik Deutschland 1949–2002 (Mannheim: Mannheimer Zentrum für Europäische Sozialforschung, Working Paper No. 98, 2007); Thomas Bräuninger and Marc Debus, ‘Der Einfluss von Koalitionsaussagen, programmatischen Standpunkten und der Bundespolitik auf die Regierungsbildung in den deutschen Ländern’, Politische Vierteljahresschrift 49/2 (2008), pp.309–38; Jonathan Slapin and Sven-Oliver Proksch, ‘A Scaling Model for Estimating Time-Series Party Positions from Texts’, American Journal of Political Science 52/3 (2008), pp.705–22; Ralf Schmitt, Die politikfeldspezifische Auswertung von Wahlprogrammen am Beispiel der deutschen Bundesländer (Mannheim: Mannheimer Zentrum für Europäische Sozialforschung, Working Paper No. 114, 2008).

See Frank Castles and Peter Mair, ‘Left–Right Political Scales: Some Expert Judgements’, European Journal of Political Research 12/1 (1984), pp.73–88; Michael Laver and Ben W. Hunt, Policy and Party Competition (New York: Routledge, 1992); Kenneth Benoit and Michael Laver, Party Policy in Modern Democracies (London: Routledge, 2006).

Ian Budge, Hans-Dieter Klingemann, Andrea Volkens, Judith Bara and Eric Tanenbaum (eds.), Mapping Policy Preferences. Estimates for Parties, Electors, and Governments 1945–1998 (Oxford: Oxford University Press, 2001); Hans-Dieter Klingemann, Andrea Volkens, Judith Bara, Ian Budge and Michael McDonald, Mapping Policy Preferences II. Estimates for Parties, Electors, and Governments in Eastern Europe, European Union, and OECD 1990–2003 (Oxford: Oxford University Press, 2006); compare Kenneth Benoit and Michael Laver, ‘Benchmarks to Text Analysis: A Response to Budge and Pennings’, Electoral Studies 26/1 (2007), pp.130–35; Kenneth Benoit, Michael Laver and Slava Mikhaylov, ‘Treating Words as Data with Error: Uncertainty in Text Statements of Policy Positions’, American Journal of Political Science 53/2 (2009), pp. 495–513.

Michael Laver, Kenneth Benoit and John Garry, ‘Extracting Policy Positions from Political Texts Using Words as Data’, American Political Science Review 97/2 (2003), pp.311–31; compare Ian Budge and Paul Pennings, ‘Do they Work? Validating Computerised Word Frequency Estimates against Policy Series’, Electoral Studies 26/1 (2007), pp.121–9.

Slapin and Proksch, ‘A Scaling Model for Estimating Time-Series Party Positions from Texts’; Sven-Oliver Proksch and Jonathan Slapin, ‘Position-Taking in European Parliament Speeches’, British Journal of Political Science (2009, forthcoming).

See Laver et al., ‘Extracting Policy Positions from Political Texts Using Words as Data’; Slapin and Proksch, ‘A Scaling Model for Estimating Time-Series Party Positions from Texts’; compare Lanny Martin and Georg Vanberg, ‘A Robust Transformation Procedure for Interpreting Political Text’, Political Analysis 16/1 (2008), pp.93–100; Andrea Volkens, ‘Strengths and Weaknesses of Approaches to Measuring Policy Positions of Parties’, Electoral Studies 26/1 (2007), pp.108–20.

Laver and Budge, Party Policy and Government Coalitions.

Michael Laver and Kenneth A. Shepsle, Making and Breaking Governments. Cabinets and Legislatures in Parliamentary Democracies (Cambridge: Cambridge University Press, 1996).

Keith Krehbiel, ‘Institutional and Partisan Sources of Gridlock: A Theory of Divided and Unified Government’, Journal of Theoretical Politics 8/1 (1996), pp.7–40; Thomas König and Thomas Bräuninger, ‘Wie wichtig sind die Länder für die Politik der Bundesregierung bei Einspruchs- und Zustimmungsgesetzen?’, Zeitschrift für Parlamentsfragen 28/4 (1997), pp.605–28; König et al., ‘Policy Change without Government Change?’.

An online-service for the initiatives from the 8th to the 15th legislative term is available by the ‘State of Federal Legislation’ (GESTA) information system at http://dip.bundestag.de. Additional GESTA data about proposals from the 14th to the 16th legislative term of the German Bundestag (until 22 August 2007) can be found at http://www.bundestag.de/bic/standgesetzgebung. The old GESTA database was replaced at the beginning of the parliamentary summer break in 2007. The new (modified) information system DIP21 is available at http://dip21.bundestag.de. Unfortunately, we had to manually reconstruct comparable data about the proposals from the 4th to the 7th legislative term by means of the ‘proceedings’ of the German Bundestag/German Bundesrat (available electronically since 1976). We cannot provide the relevant information about all proposals, because a standardised documentation of legislative activities is missing before 1972. Further differences compared to other statistics stem from mergers of proposals or different ways of counting. Regarding the ‘proceedings’ see Deutscher Bundestag, Verhandlungen des Deutschen Bundestages und des Bundesrates 1949–2009 ff (München: Saur, 2009); regarding the German federal legislation in general see Peter Schindler, Datenhandbuch zur Geschichte des Deutschen Bundestages 1949 bis 1999. Band 2 (Baden-Baden: Nomos, 1999), pp.2318–635; Michael F. Feldkamp, Datenhandbuch zur Geschichte des Deutschen Bundestages 1994 bis 2003 (Baden-Baden: Nomos, 2005), pp.571–612; compare Thomas König and Thomas Bräuninger, Gesetzgebung im Föderalismus (Speyer: Forschungsinstitut für öffentliche Verwaltung, Speyerer Forschungsberichte No. 237, 2005); Bernhard Miller and Christian Stecker, ‘Consensus by Default? Interaction of Government and Opposition Parties in the Committees of the German Bundestag’, German Politics 17/3 (2008), pp.305–22.

Latest downloading of legislative proposals as at 31 December 2008.

‘SIMSTAT/WordStat’ is a suite of software packages produced by Provalis Research. The text mining tool ‘WordStat’ runs at the base of the data analysis tool ‘SIMSTAT’. See Tyler Johnson, ‘Review of WordStat 5.1, SIMSTAT 2.5, and QDA Miner 2.0’, The Political Methodologist 15/1 (2007), pp.11–14; Angelique Davi, Dominique Haughton, Nada Nasr, Gaurav Shah, Maria Skaletsky and Ruth Spack, ‘A Review of Two Text-Mining Packages: SAS TextMining and WordStat’, The American Statistician 59/1 (2005), pp.89–103.

The thematic classification (Sachgebiet) refers to the official directory of federal legislation (Fundstellennachweis). It is comparable to the ‘Directory of Community Legislation’ at the EU level. The foremost fee-based database of German Law ‘Juris’ provides information about the thematic classification of laws at http://www.juris.de.

‘WordNet’ is a lexical database of the English language, which is available for download from ‘Princeton University Cognitive Science Laboratory’ (http://wordnet.princeton.edu); Christiane Fellbaum (ed.), WordNet: An Electronic Lexical Database (Cambridge, MA: MIT Press, 1998).

‘GermaNet’ is the German version of ‘WordNet’, developed at the Division of Computational Linguistics of the Linguistics Department at the University of Tübingen (http://www.sfs.uni-tuebingen.de/GermaNet); see Claudia Kunze and Lothar Lemnitzer, Computerlexikographie: Eine Einführung (Tübingen: Narr, 2007), pp.136–65.

The CDU and its Bavarian sister party CSU presented election programmes on their own in the first decades of the Federal Republic of Germany and again in 1990 (German reunification). In these cases we refer to the CDU programme.

In 1990 the West German Greens presented a programme that was different from the common programme of West German and East German Greens (‘Wahlplattform Die Grünen/Bündnis90’). We refer to the common programme.

See Hans U. Behn, Die Regierungserklärungen der Bundesrepublik Deutschland (München: Olzog, 1971); Manuel Fröhlich, ‘Regierungserklärungen mit Geschichte’, Das Parlament, 20 Nov. 1998, p.13; Karl-Rudolf Korte (ed.), Das Wort hat der Herr Bundeskanzler. Eine Analyse der Großen Regierungserklärungen von Adenauer bis Schröder (Wiesbaden: Westdeutscher Verlag, 2002); Klaus Stüwe, Die Rede des Kanzlers. Regierungserklärungen von Adenauer bis Schröder (Wiesbaden: VS Verlag, 2005). In 1961 Vice-Chancellor Erhard had to read the declaration on behalf of Chancellor Adenauer.

The four exceptions are caused by Chancellor Adenauer's resignation in 1963, Chancellor Erhard's resignation in 1966, Chancellor Brandt's resignation in 1974 and the constructive vote of no-confidence against Chancellor Schmidt in 1982.

We are currently not in a position to precisely assess the correctness of all manifestos. Until now, we calculated on the basis of nine SPD election programmes an error rate less than 4 per cent (GESIS files in comparison with the scans of the original programmes). Nevertheless, this error rate does not reflect some fundamental problems, e.g. that speeches are considered as election programmes.

The length of the government declarations is artificially inflated in the WZB files; compare Stüwe, Die Rede des Kanzlers, p.121.

See Eneko Agirre and Philip Edmonds (eds.), Word Sense Disambiguation. Algorithms and Applications (Dordricht: Springer, 2007).

The German refinement of the CMP approach also differentiates between ideology (CMP category, variable ‘Position’) and policy area (variable ‘Politikfeld’). But because no individual data are available, researchers cannot combine the analyses. See Andrea Volkens, Handbuch zur Inhaltsanalyse programmatischer Dokumente von Parteien und Regierungen in der Bundesrepublik Deutschland (Berlin: Wissenschaftszentrum Berlin für Sozialforschung, Discussion Paper FS III 02/203, 2002); compare Shikano and Pappi, The Positions of Parties in Ideological and Policy Space; Pappi and Shikano, Ideologische Signale in den Wahlprogrammen der deutschen Bundestagsparteien 1980 bis 2002.

This portfolio standardisation is complicated by the fusion of competencies or the separation of competencies (see e.g. the ‘Federal Ministry for Economics and Labour’ under Wolfgang Clement in the 15th legislative term) and the creation of new ministries (see e.g. the ‘Federal Ministry of the Environment, Nature Conservation and Nuclear Safety’ which was created a few weeks after the Chernobyl disaster in 1986. Until then, the Federal Ministry of the Interior the Federal Ministry of Agriculture and the former Federal Ministry of Youth, Family and Health were responsible for environment policy. As a consequence, instead of the original portfolio classification the thematic classification (Sachgebiet) serves as a linkage to the 14 standard portfolios. The thematic classification allows a refined analysis of broad policy areas (see e.g. the Federal Ministry of the Interior that has competencies in issues related to internal security, managing immigration, public administration, sports, etc.).

Portfolios (which consist of themes) and ideological dimensions are independently conceptualised. We do not assume a transformation. In contrast to this, see James M. Enelow and Melvin J. Hinich, The Spatial Theory of Voting. An Introduction (Cambridge: Cambridge University Press, 1984), pp.36–79.

See John D. Huber and Ronald Inglehart, ‘Expert Interpretations of Party Space and Party Locations in 42 Societies’, Party Politics 1/1 (1995), pp.73–111; Benoit and Laver, Party Policy in Modern Democracies.

Seymour M. Lipset and Stein Rokkan, ‘Cleavage Structures, Party Systems, and Voter Alignments. An Introduction’, in Seymour M. Lipset and Stein Rokkan (eds.), Party Systems and Voter Alignments: Cross-National Perspectives (New York: Free Press, 1967), pp.1–64; Seymour M. Lipset, ‘Cleavages, Parties and Democracy’, in Lauri Karvonen and Stein Kuhnle (eds.), Party Systems and Voter Alignments Revisited (London: Routledge, 2001), pp.3–9.

Regarding ideological (bi-)polarity compare Michael Laver and John Garry, ‘Estimating Policy Positions from Political Texts’, American Journal of Political Science 44/3 (2000), pp.619–34; Benoit and Laver, Party Policy in Modern Democracies.

See Teije J. Euverman and Adrianus A. Vermulst, Bayesian Factor Analysis (Groningen: University of Groningen, 1983); Simon Jackman, ‘Estimation and Inference via Bayesian Simulation: An Introduction to Markov Chain Monte Carlo’, American Journal of Political Science 44/2 (2000), pp.375–404; Jeff Gill, Bayesian Methods: A Social and Behavioral Sciences Approach (Boca Raton: Chapman and Hall, 2002); Andrew Gelman, John B. Carlin, Hal S. Stern and Donald B. Rubin, Bayesian Data Analysis (London: Chapman and Hall, 2003); Kevin M. Quinn, ‘Bayesian Factor Analysis for Mixed Ordinal and Continuous Responses’, Political Analysis 12/4 (2004), pp.338–53.

Considering duplications, portfolio words and exclusion words we can realise the compilation of the ideology wordbook with reasonable expenditure. The ideological wordbook can be transferred from one legislative term to another in order to determine the adequacy. Definitely, the more words/terms the wordbooks already contain the better is the relationship between cost and benefit.

Daniel Finke wrote the Gauss code. He will publish the code at http://www.rzuser.uni-heidelberg.de/~dfinke.

Regarding the salience and (non-)separability of voters' preferences see Susumu Shikano, ‘Die Eigendynamik zur Eindimensionalität des Parteienwettbewerbs. Eine Simulationsstudie’, Politische Vierteljahresschrift 49/2 (2008), pp.229–50; Enelow and Hinich, The Spatial Theory of Voting.

The 1973 and 1991 grand government declarations are coded as R72 and R90, because they are related to the beginning of the legislative terms after the federal elections in 1972 and 1990, respectively.