Jump to ContentJump to Main Navigation
Parties and DemocracyCoalition Formation and Government Functioning in Twenty States$

Ian Budge and Hans Keman

Print publication date: 1993

Print ISBN-13: 9780198279259

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/0198279256.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 26 February 2017

(p.213) Appendix B: Data: Sources and Preparation

(p.213) Appendix B: Data: Sources and Preparation

Parties and Democracy
Oxford University Press

The data used for analysis throughout the book are all aggregated and organized by country. The period studied covers at most the post‐war era (1945 to 1984), but quite often lack of data or cross‐national limitations on availability restricted it to 1950 onwards or even 1965 onwards (particularly in regard to economic data).

The data—apart from those related to policy—are organized on different levels of aggregation:

  1. 1. On the level of the political system (in addition, some specific variables were added to the files or derived from system‐level information)

  2. 2. On the level of governments as a whole

  3. 3. On the level of separate ministries

  4. 4. For sectors of ministries

  5. 5. For modes of termination

The following democracies are the ones for which at least some data were available: Australia (1), Austria (2), Belgium (3), Canada (4), Denmark (5), Finland (6), France (Fourth Republic) (7), France (Fifth Republic) (8), the Federal Republic of Germany (9), Ireland (10), Italy (11), Japan (12), The Netherlands (13), New Zealand (14), Norway (15), Sweden (16), Switzerland (17), the United Kingdom (18), Iceland* (30), India (40), Israel* (50), Luxemburg* (60), Sri Lanka (70). Asterisked countries are those omitted at some points through lack of policy‐relevant information. Because of difficulties of comparability India and Sri Lanka are only referred to in the course of particular analyses.

The following variables have been used to describe the various parliamentary systems of the countries under review:

  • Number of parliamentary seats (total)

  • Code‐number of parties in Parliament according to Mackie and Rose

  • Number of parliamentary seats held by each party (only those, naturally,

  • which have ever entered Parliament)

  • Percentage of votes for each party

  • Role of party in government (2 = participating; 1 = supporting; 0 = other)

  • Number of portfolios held by each party

  • Percentage of portfolios held by each party

  • Party code of party of Prime Minister

(p.214) Most of these variables are straightforwardly taken from Mackie and Rose1 and updated by means of either the European Journal of Political Research or Keesing's Contemporary Archives.2 The party‐in‐government variable as well as the portfolio variables are primarily based on Keesing's Contemporary Archives, but at the same time checked with the work of von Beyme and Paloheimo.3

In operationalizing the parliamentary variables one major problem arose: whether to score the actual ministers, or ministries. We opted for the latter option, that is to say we counted the number of portfolios (i.e. competences) held by each party. For in our analysis we seek to discover the impact of parties within governments, in terms of both the office‐seeking capacity of a party and its capacity to transform its ideology into public policy formation and performance. Hence the minister is not the carrier of the supposed influence; it is the competence of the ministry in a policy area that is important to the party to which the office‐holder is affiliated.

Both von Beyme and Paloheimo inform us on the role of parties in government, and in general the differences between them are not great. Their definition of a government, however, is different from ours: we return to this later. If there was a difference of opinion we resorted to Keesing's Contemporary Archives, unless it concerned the Federal Republic of Germany or Austria, or Finland where we considered von Beyme and Paloheimo respectively country‐specialists. There were differences between Keesing's Contemporary Archives and Paloheimo with regard to the number of portfolios held by each party. The reason for this is twofold: on the one hand, Paloheimo counts persons rather than competences; on the other, it is genuinely difficult in some cases (in particular, Parliaments on the ‘Westminster model’) to know what is a governmental organization and what is not. Hence there are problems in determining the exact number. Again, when in doubt, we resorted to Keesing's Contemporary Archives.

The variables listed below have been used to depict the format and working of the various party systems:4

  • Anti‐democratic feelings (1 = yes; 0 = no)

  • Left–Right feelings (1 = yes; 0 = no)

  • Absolute majority of government (1 = yes; 0 = no)

  • Anti‐system party (1 = yes; 0 = no)

  • Normal party of Government (1 = yes; 0 = no)

  • Tendance of party (1 = Bourgeois; 2 = Socialist)

  • Party Family (1 = Conservative; 2 = Liberal; 3 = Christian Democratic;

  • 4 = Social Democratic; 5 = Special Interest)

Although most of these variables appear to be quite static, we have nevertheless assumed them to be dynamic. Therefore the most problematic issue with respect to them has been less the distinction as a dummy and more if and when change occurred. Invariably the cut‐off points used will show (p.215) some arbitrariness. However, we are convinced that the decisions made would be agreed to by most specialists.

Another problem has been to determine whether a party belongs to the Conservative or the Christian Democratic party family. The same applies to the distinction between Liberal and Conservative. In this we have followed the categorization of parties made by Keman5 and applied it to this sample.

The third cluster of variables consists of those representing specific features of governments. Before discussing the separate elements of each variable we shall elaborate (again) the definition of ‘government’ as it has been used throughout this study: it encompasses any administration that is formed after an election and continues in the absence of

  1. (a) change of Prime Minister;

  2. (b) change in the party composition of the Cabinet; or

  3. (c) resignation in an inter‐election period followed by re‐formation of the government with the same Prime Minister and party composition.

This definition is obviously more strict than most others, particularly those used in much of the literature on coalitions. This is particularly true in regard to formal resignation. Yet it makes sense to include this as a cut‐off point, since a resignation generally changes the political situation in some significant aspects. The effect of our decision has been to increase the total number of governments, hence the total number of cases. However, the differences with both von Beyme and Paloheimo, although they exist, are not enormously great, and are explicable by our inclusion of caretaker governments and the operation of the resignation rule.

Another respect in which we differ from some of the other specialist treatments concerns the date of the initial constitution of a government. Whereas there is great unanimity as to dates of termination there is less in regard to formation. The main reason lies in the different ways in which governments are brought into being in some countries and in their constitutional processes: in a number of the continental European countries a considerable lapse of time occurs between the ending of a government and the installation of a new government by the Head of State. Most conspicuous in this respect are Belgium, Denmark, and The Netherlands (the maximum lapse of time being almost nine months, in which there is no new government and the old one is not allowed to introduce new bills or to make politically controversial decisions). To a lesser extent these lapses of time can be found in Finland, Italy, Norway, and Sweden. Von Beyme only gives the beginning not the ending of a government, whereas Paloheimo follows the same practices as we have. Any difference has been checked by means of Keesing's Contemporary Archives.

In addition to these government variables (dates of initiation and termination) we have created the following: (p.216)

  • Government number (1 . . . n) (numerically coded)

  • Identity of Prime Minister (alphanumerically coded)

  • Duration of government (in years, up to the nearest half‐month)

  • Percentage of ministries held by party

  • Type of government (1–5)

‘Ideological complexion of government’ is an indicator which introduces a more qualitative aspect to government formation. It attempts to account for the relative strength of parties in government with reference to the Left–Right dimension, through a five‐point scale in which the proportional shares of the Left, Centre and Right are transformed into scores (1 to 5) representing the degree of dominance of either party:

  1. 1 = right‐wing dominance (share of Cabinet seats of these parties greater than 66.6 per cent)

  2. 2 = Right–Centre complexion (share of Cabinet seats of Right and Centre parties between 33.3 and 66.6 per cent each)

  3. 3 = balanced situation (share of Centre greater than 50 per cent; or if Left and Right form a government together not dominated by one or the other)

  4. 4 = Left–Centre complexion (share of Cabinet seats of Left and Centre parties between 33.3 and 66.6 per cent each)

  5. 5 = left‐wing dominance (share of Cabinet seats of these parties greater than 66.6 per cent)

These scores are collected and calculated for the period 1950–83 on an annual basis and then compiled for each government.6

The following are the five types of government in our classification:

  1. 1 = One‐party government: one party takes all government seats.

  2. 2 = Minimal winning coalitions: all participating parties are necessary to form the government.

  3. 3 = Surplus coalition: this comprises those coalition governments which exceed the minimal‐winning criterion.

  4. 4 = Minority government: the party or parties in government do not possess a majority in Parliament.

  5. 5 = Caretaker government: the government formed is not intended to undertake any kind of serious policy‐making, but is only temporarily minding the shop.

As mentioned in Chapter 4, Cabinet Ministries are lightly standardized into common types—which in fact vary only in minor ways across countries. Coding is as follows: 01 = Prime Minister, 02 = Deputy Prime Minister, 03 = Foreign Affairs, 04 = Defence, 05 = Interior, 06 = Justice, 07 = Finance, 08 = Economy, 09 = Labour, 10 = Social Affairs, 11 = Education, 12 = Health, 13 = Housing, 14 = Infrastructure, 15 = Agriculture, 16 = Industry/Trade/Commerce, 17 = Religion, 18 = Other. These data have not yet been (p.217) collected cross‐nationally in any other study known to us. On the one hand, this implies that there are no differences with other data sources. On the other, it has meant that there were some problems concerning comparability and reliability.

In many countries different labels are used for the same competence (for instance, the Treasury in the UK is similar to Finance in most continental countries; in other cases Justice is differently organized, not only in a ministry, but also in a separate non‐governmental institution). These problems have been solved by consulting country studies or, if available and understandable, government and/or parliamentary year‐books.

In most countries some of the competences listed above are combined in one ministry; in others they are separated. For example, in some countries Defence is organized in different ministries (Navy, Air Force, and Army in The Netherlands, for instance, up to 1959); in Scandinavia Agriculture is subdivided into Forestry, Fisheries, and Agriculture itself. Conversely, Labour and Social Affairs (often labelled ‘Employment’) may be headed by one minister, as may the Economy, Trade, and Industry. In each case of doubt we consulted Keesing's Contemporary Archives. If more than one of the coded competences was mentioned we scored them both. Where there was a subdivision, we either attributed it to the party holding all relevant ministries or scored the party holding the main competence—in the Scandinavian cases, Agriculture; in the Dutch case, Navy, Air Force, and Army together as Defence (since after 1959 Navy and Air Force became sub‐ministries within Defence).

To increase comparability and reliability we have developed the concept of sectors of ministries. We contend that the main areas of policy concern are generally not covered by just one ministry, but by various ministries. For our analysis of the relation between policy formation, policy performance, and party control of certain offices, we have designated as ‘relevant ministries’ the following:

  • Economic Policy sector: Finance, Labour, Economy

  • Social Welfare sector: Social Affairs, Education, Health,

  • External Security sector: Prime Minister, Foreign Affairs, Defence

Not all ministries existed in a country for the whole period of the study, so there were variations in the absolute number of relevant ministries. We therefore employed proportional representation of parties within a sector rather than absolute values. The dominance of a sector was decided by looking at the party (family) with the greatest representation. If two parties had equal numbers of specific ministries the Prime Minister's party—if represented—was considered to be the dominant one.

A separate file has been generated to deal with modes of termination of governments. Apart from von Beyme's work,7 we have not come across such (p.218) data in the existing literature. Following von Beyme,8 we developed a classification for modes of termination:

  1. 1.1 Fixed elections. These include any election stipulated by law or constitution—for the Anglo‐Saxon countries, in the fifth year of the electoral calendar.

  2. 1.2 Anticipated elections. These include all elections not required as under 1.1, and those, in Anglo‐Saxon countries, occurring before four years have gone by.

  3. 2.1 Voluntary resignation of the Prime Minister.

  4. 2.2 Resignation of the Prime Minister due to health reasons. Both these last two reasons can be considered as non‐political ones, but mode 2.1 may well be a cover‐up for factional dispute within a party or government (as for instance in Japan). Yet, as we cannot distinguish ‘real’ from ‘fake’ reasons, we have accepted them on face value.

  5. 3.1 Dissension within government. This covers those instances when either a coalition breaks up without external pressure or there are publicized quarrels and/or movement of personnel. Often these incidents are not discussed in the literature since in many cases they have no visible consequence for a government defined in a more relaxed way than we have defined it.

  6. 3.2 Lack of parliamentary support. This reason of termination, of course, lies at the heart of any parliamentary democracy. We have counted here every instance when either parties withdrew support from government, or there occurred a (successful) vote of no confidence (or similar parliamentary action).

  7. 3.3 Intervention by the Head of State. Apart from semi‐presidential regimes (such as Finland and France), however, the role of the Head of State is much less essential for government termination than with respect to government formation.

We collected these data from Keesing's Contemporary Archives. In addition we compared our data with those of von Beyme. There was little disagreement except for the fact that we have identified more governments and thus have found more reasons of termination. Another difference has been that electoral reasons are not subdivided by von Beyme as by us. Finally, more than one reason is often mentioned by von Beyme: in these cases we have followed Keesing's Contemporary Archives. In a number of cases (about 10 per cent) we were not able to locate a reason of termination.

Finally we have developed a number of variables relating to policy formation (or outputs) and policy performance (or outcomes). These include indexes of economic policy, social welfare, and one measure of the trade‐off between military expenditure and welfare. In addition we used indexes representing the extent to which socio‐economic welfare has been achieved.

Policy formation indexes. Measurement of economic policy output is based (p.219) on the interpretation of certain policy measures which supposedly form, in combination, a nation's macro‐economic policy strategy. To this end we have examined the degree to which a country allowed a Budget deficit and a high increase in money supply; and the size of the public economy (total outlays on all government services). Unlike social welfare output, we have defined the index qualitatively in the following manner:

  1. 1 = High levels of spending, regardless of the Budget deficit and the velocity of money circulation.

  2. 2 = Moderately high levels of public spending, with a relatively low Budget deficit.

  3. 3 = Modest levels of public spending, with avoidance of Budget deficits and of an increase in the money supply.

  4. 4 = Low levels of public spending, hardly any Budget deficits, and no increase of money supply.

The index is based on rank‐ordering scores on each dimension, for each country, on a yearly basis (1963–84). The data source is OECD (various volumes).9

External security is simply based on total expenditure on defence expressed as a percentage of GDP (source: SIPRI, various volumes). The ‘trade‐off’ measure is a constructed variable. It is the combination of the annual growth‐rates of social welfare and military expenditures. If they both increase or decrease simultaneously we have labelled this a positive and negative pay‐off respectively. If one increases and the other decreases we called this a positive trade‐off (if social welfare grows) or a negative trade‐off (if defence grows). Thus there are four distinctive outcomes:10

  1. 1 = Positive trade‐off

  2. 2 = Positive pay‐off

  3. 3 = Negative pay‐off

  4. 4 = Negative trade‐off

Policy performance index. This is identical to the ‘misery index’ as used by OECD: (rate of inflation + rate of unemployment) divided by 2. This measure shows the relative success of a country in coping with socio‐economic circumstances. The index is based on annual OECD data.

All policy‐related variables are based on yearly figures and are transformed into rank‐ordered values. This procedure has been undertaken in order to increase comparability, sometimes at the expense of precision. However, we feel that, since the data are collected and calculated on an annual basis, this certainly improves cross‐national comparisons and analysis over time.

In conclusion, we should note that this appendix represents only a summary description of the data we have employed. With the aid of a grant from the British Economic and Social Science Research Council (R000231598) work will be going on to refine and complete our data collection, and to publish all (p.220) variables in full in a handbook, in association with other collections of data on parties in government. In addition we intend to lodge the fully computerized data set or sets at the major European Social Science Archives. Thus all the information we have drawn up should be available for public use and secondary analysis from the end of 1990 onwards. Obviously it may undergo some changes of detail in the mean time, but this appendix should remain valid as a broad description of the leading features of our collection. We should like to end with a tribute to the work of our predecessors in the field, whose works have been cited frequently in the notes to this appendix, and hope our data may prove as useful to future investigators as theirs has proved to be to us.


(1.) T. Mackie and R. Rose, The International Almanac of Electoral History (2nd edn.; London: Macmillan, 1982).

(2.) Bristol: Longman, 1945–.

(3.) K. von Beyme, Political Parties in Western Democracies (Aldershot: Gower, 1985), appendix pp. 377–406; H. Paloheimo, Governments in (p.229) Democratic Capitalist States 1950–1983: A Data Handbook (Tampere: Finnish Political Science Association, 1984).

(4.) Most of these variables reflect the literature on parties and party systems (see G. Sartori, Parties and Party Systems (Cambridge: Cambridge UP, 1976)) as well as certain dimensions of what might be called commonsense observation in most Western parliamentary democracies. In addition to Sartori, these data are derived from the following works. With respect to party families: von Beyme, Political Parties; Paloheimo, Governments in Democratic Capitalist States; J. Raschke (ed.), Die politischen Parteien in Westeuropa (Reinbek‐Bei‐Hamburg: RoRoRo Verlag, 1978). With respect to tendance: I. Budge and D. J. Farlie, Explaining and Predicting Elections (London: Allen & Unwin, 1983). With respect to anti‐democratic and Left–Right feelings, and tendance of party: I. Budge, ‘Parties and Democratic Government: A Framework for Comparative Explanation’, Western European Politics, 7 (1984), 95–118. Finally, various country studies were used where necessary.

(5.) H. E. Keman, The Development toward Surplus Welfare (Amsterdam: CT Press, 1988).

(6.) For this, see also M. G. Schmidt, ‘The Welfare State and the Economy in Periods of Economic Crisis: A Comparative Analysis of 23 O.E.C.D. Nations’, European Journal of Political Research, 11 (1983), 1–26; Keman, The Development toward Surplus Welfare.

(7.) Von Beyme, Political Parties, appendix.

(8.) Ibid., pp. 329 ff.

(9.) For an extensive discussion of this variable, see H. E. Keman and T. van Dijk, ‘Policy Formation as a Strategy to Overcome the Economic Crisis’, in F. Castles and R. Wildermann (eds.), The Future of Party Government (Berlin: De Gruyter, 1987), 127–62.

(10.) For a full discussion of this method of calculation, see H. E. Keman, ‘Welfare and Warfare: Critical Opinions and Conscious Choice in Public Policy’, in F. Castles, F. Lehner, and M. G. Schmidt (eds.), Managing Mixed Economies (Berlin: De Gruyter, 1988).