Do Parties make a Difference? A Comparison of Party and Coalition Policy in Ireland using Expert Coding and Computerised Content Analysis

Transcription

1 Do Parties make a Difference? A Comparison of Party and Coalition Policy in Ireland using Expert Coding and Computerised Content Analysis Lucy Mansergh Department of Political Science Trinity College Dublin Paper prepared for presentation at the ECPR Meeting Workshop Estimating the Policy Positions of Political Actors Mannheim, March 1999 Introduction

2 The assumption that parties make a difference is fundamental to the traditional conception of representative democracy. What political scientists tend to disagree upon is the extent and nature of the difference that parties can and do exert on public policy. Two extreme positions may be identified; that parties matter little to the formulation and implementation of most public policy, in that economic and socio-economic imperatives tend to dictate, and the opposing view, which is that the scope for partisan influence is great. The first paradigm, which claims the prevalence of socio-economic factors in the determination of public policy, has supporters from many different intellectual bents. These include Convergence Theorists, Rational Choice Theorists, End of Ideology proponents, and neo-marxists 1. The Convergence Theorists claim that technological imperatives in advanced societies override political factors making for diversity, and progressively shape social structures and public policies in a similar mould 2. Schofield (1993), a rational choice theorist, posits that government policy is based on the policies of all the parties in a system rather than just those in government, on the basis that policy outputs go to all parties in the system whether they are in or out of government 3. The End of Ideology theory propounds the view that with the development of post-modern or post-industrial societies 4, fewer issues exist around which parties may take extreme positions. Lastly, the Neo-Marxists argue that the ability of the state or constituent parties to determine policy outcomes is itself determined by the nature of the dominant class-struggle in society. The opposing paradigm attributes observed variance in public policy outcomes to political factors 5. According to Schmidt 6, supporters of this view rely on a number of assumptions regarding the relations between social constituencies, parties and public policy. A political party, it is suggested, pursues preferences which broadly correlate with those of its voters, either as an aside to or in conjunction with its pursuit of office. Once in government, that incumbent party has (or those incumbent parties have) the resources to implement those policies advocated in the campaign period. Policy differences between governments, therefore, are a function of the party composition of government controlling for the distribution of power in parliament and in extra-parliamentary arenas, institutional arrangements, adaptation to environments, socio-economic circumstances and international independence 7. Not merely however is government policy a function of its party composition, it may further be a function of the political position of the minister holding the particular portfolio 8. Lastly, it is contended that it is not enough that parties are capable of implementing policies; for voters preferences to be given true representation, it is necessary for them actually to do so 9. If voters hold different policy preferences, then parties must stand for different policies for representative democracy to be representative. Furthermore, not only must they stand for these policies before the election but they must also do so afterwards. The scope of this paper is too narrow to test the final proposition that parties do all or much of what they said they would prior to the election, once in office. Rather, the aim is to gauge, in the Irish context, whether the political parties do hold substantively different policy positions from each other, and whether the programmes of the governments that these parties form reflect the substantive positions of the incumbent parties. In order to evaluate this, it is necessary to estimate the policy positions of both the political parties and the governments of which they are a part. This paper draws on two different approaches to the estimation of party and government positions from political texts. These are the two Laver and Garry content analyses schemes; their expert coding scheme and Castles, p.6, Budge and Laver, p.426, 1992 Bell, 1972 Castles, p.5, 1982 Schmidt, p.156, Laver and Shepsle, p.14, 1995 Ginsberg, p.41, 1976

3 computerised content analysis using their English language dictionary. By comparing the estimated policy positions of the Irish political parties both in and out of Government as derived by the different methods, this paper seeks to evaluate an hypothesis concerning the policy input of political parties into Programmes for Government. Part 2 of the paper sets out a short characterisation of the Irish party system and the perceived role of policy within that system. Part 3 describes the circumstances of each change of Government that will be dealt with, and sets up an hypothesis concerning the relationship between party policy and coalition government policy. Part 4 discusses different approaches to the estimation of policy positions of political actors that this paper will utilise. Part 5 summarises the results of the different approaches to the measurement of the actors policy positions and part 6 concludes by looking at the implications of these results for our hypothesis. Policy and the Party system in Ireland The Rational Choice tradition claims three models of competitive party behaviour. These are based on three assumptions about the behaviour of political parties; vote-seeking, officeseeking or policy-seeking 10. In an attempt to identify which categories best fit the Irish political parties, Laver and Hunt (1992) conducted a survey of about forty specialists in Irish politics and asked them to rate the extent to which politicians valued cabinet portfolios as rewards of office or as a means to affect policy. A similar exercise was performed for 17 other European systems. Of these, only in Greece and Italy were politicians rated as being more interested in office rather than policy 11. In the Irish case, the party system has not only been characterised by low policy-motivation, but has also been charged with having little ideological policy competition. Laver and Budge 12 declare that the lack of a strong left-right dimension and the role of traditional loyalties in determining patterns of party support have meant that intense ideological competition has rarely been presented as the basic motivation of Irish politicians. Traditional loyalties so-called, date back to the early years after Independence. The split in Sinn Fein in the 1920s over whether or not to accept the Treaty of Independence, culminated in the formation of two parties, Fianna Fáil, which had been the anti-treaty faction and Fine Gael, the pro-treaty faction. The Labour Party, whose origins predate both of these other parties, failed to take up a party position on the issue of the acceptance or rejection of the treaty. The nationalist cleavage, it is claimed, has permeated Irish politics up to the present day, and has led to the dimunution of the importance of the more common European division between left and right-wing economics. This picture, however, is misleading. The issue of Northern Ireland is rarely rated highly by voters as an election issue. Economic issues have invariably been the issues that voters see as most important, such as unemployment and taxation 13. In the 1980s, a break-away group from Fianna Fáil formed a new party, the Progressive Democrats and its defining stance was free-market economics. Though the Progressive Democrats electoral performance has waned considerably since its best result winning 14 seats in 1987, it has survived to the present day. In the aftermath of the 1992 General Election, a near all-time high figure of 33 seats for the Labour Party was perceived as presaging a resurrection of the Left. While this may have been premature, what is clear is that parties not identified with the nationalist cleavage can Strom, p.565, May 1990 Laver and Hunt 1992; in Budge and Laver, p.43, 1992 Laver and Budge, p.43, 1992 Though it must be admitted that policy as a whole has been rated less highly than choice of candidate by all voters, as a determinant of their vote for at least the past four elections. The declared importance of policy, however, does vary according to voting intention for a party see Laver, Mair and Sinnott, p.135, 1987

4 and do survive. What is also clear is that neither Fine Gael nor Fianna Fáil has been afforded the chance to campaign purely on the nationalist platform. Further corroboration that perhaps other policy areas are a more important influence on Irish politicians than previously thought is to be found in the results of the Laver and Hunt expert survey of They discovered that a left-right dimension, operationalised in terms of party policy on the public ownership of business, was seen as very important by most of the Irish parties 14. Garry and Mansergh 15, studying the emphases placed on different policy concerns by the different parties in the 1997 General Election, found that for Fianna Fáil, Fine Gael and the Progressive Democrats, the policy category that was most emphasised was the economy, each party devoting over a quarter of its manifestos to this domain. For the Labour Party and Democratic Left, concern for welfare (ie, the welfare state) and the quality of life predominated their manifestos, but for both, the economy came a close second. Budge and Farlie 16 cite Saliency Theory in their own previous study findings from 1977, based on British manifestos and American election platforms. When [the documents ] sentences were systematically examined, a major finding was the very limited extent to which the rival parties mentioned each other at all, let alone mentioned the other s policies 17. This finding would indicate that party competition is not run along the confrontational lines in the countries studied and that what parties compete on are bundles of salient issues. The bundle that strikes a chord with the largest number of voters wins the election. In a study by Mair 18, looking at the extent to which saliency theory may hold true for the Irish case, he found that the parties differed in the extent to which they shared issues with the others Fianna Fáil was found to be the most consensual, ie, shared its most emphasised policy categories with at least one other party. Fine Gael was found to have the most well-emphasised policy areas that were party-specific, whilst Labour came somewhere in between the Fianna Fáil and Fine Gael extremes 19. One of the implications of this finding is that Fine Gael and Labour should have fewer difficulties in reaching agreement on the shape of a government programme than Fianna Fáil and Labour, as the number of policy areas to which both attach importance are fewer in number. Testing this hypothesis however is not within the scope of this paper. What Mair s 20 finding doesn t tell us, however, is whether the parties held similar or diverging policy positions within those categories that were equally emphasised. It is to this question that we shall later return, when we look at the party and government policy positions in the consensual areas of the economy and social policy 21. All of the Irish political parties, in or out of office, produce policy documents from time to time. Their number is generally greater when out of office, and the frequency of their publication tends to accelerate as the election draws closer 22. A definitive statement of the party s policy positions, the manifesto, is published during the campaign period and is often the compilation of previously published policy documents and those as-yet undisclosed. Interviews conducted by Garry and Mansergh 23 with the 1997manifesto writers of the five main parties, record that party policy formulation in Ireland in the 1990 s is an ongoing process and manifestos are not produced on an ad hoc basis but rather in a carefully organised and pre-meditated manner. Laver and Budge, p.48, 1992 Garry and Mansergh, p.24, forthcoming Budge and Farlie, p.23, 1983 Mair in Budge, Robertson and Hearl, p.144, 1987 Though admittedly we are not quite comparing like with like here. Later when comparing party policy positions to those of the government programmes, I shall do so on the basis of the Laver and Garry coding schemes, and to a lower degree of specificity than Mair, who used the detail of the MRG coding categories. The lesser detail of my approach brings together policy categories under the economic and social domains in which the MRG approach would have found both consensual and party-specific policies. Garry and Mansergh, p.5, forthcoming

5 Manifestos do play an important role. Debate at election time centres on these documents. Many of the shots fired at rival parties rely on inadequacies or inconsistencies in their party manifesto. While few voters ever read the document, it may act as a sop to the party faithful to get them out onto the doorsteps, and is certain to be disseminated by the media 24. More importantly though, in the Irish context, the manifestos are the basic ingredients of the coalition policy document, the Programme for Government. The manifestos act as bargaining chips in negotiations and as such, could fairly be seen as simply detailed shopping lists for government 25. Coalition Politics and Policy in Ireland Coalition politics in Ireland before 1989 was a question of finding some alternative to a single-party Fianna Fáil minority government 26. In the period , [Fianna Fáil] won an average of 46% of the vote and 49% of all seats in the legislature 27 and up until 1977, had managed on a number of occasions to secure an overall majority. The Party was vehemently opposed to the notion that it could ever be part of anything short of a single-party Government. This attitude was to change in 1989, when Fianna Fáil, in pursuit of another overall majority in the Dáil, prematurely called an election. Not only did the Fianna Fáil party fail to win that majority, they actually suffered a loss of seats. Rather than abandoning office in a period of economic growth, Fianna Fáil made an historic pact with the Progressive Democrats after the election, and entered into coalition with them 28. Three years later and the Progressive Democrats pulled the plug. In this next election of 1992, Fianna Fáil garnered 68 seats, Fine Gael, 45, Labour 33 and the Progressive Democrats 10. The Labour Party, which had been riding high in the opinion polls prior to the election and had refused to announce their preferred coalition partner in the hope of maximising transfers, were wooed into negotiations by both Fine Gael and Fianna Fáil. The Labour officials on entering negotiations with Fianna Fáil were apparently dumbstruck by their opposites opening presentation of a proposed Government Agreement for the two parties, and further by the extent to which Fianna Fáil had gone in accepting Labour policies. One of the negotiators on the Labour side later wrote of the document it was all there, in the document we received. A third banking force, reform of confessional legislation, significant investment in social services. Anything that we were likely to find contentious in their policy position was simply dropped 29. An agreement was struck. Yet despite Fianna Fáil s elaborate preparations for government, the coalition government had fallen within 18 months. A series of revelations exhibiting the lack of openness and exchange of information between Fianna Fáil and its partner in government caused Labour to pull out of Government. The then leader of Fianna Fail, Albert Reynolds, took the brunt of the blame and resigned. The parties in the Dáil, reluctant to endure another election, set about negotiating a new government. Under the new Fianna Fáil leader, Bertie Ahern, approaches were made again to Labour to return into coalition, but further revelations about the conduct of Fianna Fáil in the 1992 Government, put paid to these. Negotiations between Fine Gael, Labour and the Democratic Left were to prove successful. Again Labour was seen to win out in policy terms - Fine Gael should have been in the driving seat but in the end, conceded key ministries to Labour, including Finance, Foreign Affairs, Education and Environment 30. The circumstances of the 1997 election were more salubrious than those surrounding government formation in The three parties, Fine Gael, Labour and Democratic Left, felt that they had worked well together and set about campaigning as the next coalition. Their Garry and Mansergh, p.15, forthcoming Laver and Budge, p.41, 1992 O Reilly in Nealon, p.170, 1997 Finlay, p.136, 1998 O Reilly in Nealon, p.170, 1997

6 competition was Fianna Fáil and the Progressive Democrats who also announced an electoral pact. The 1997 election therefore produced a choice of government for the electorate prior to the election. Both groups published Joint Programmes for Government. Fianna Fáil and the Progressive Democrats combined won 81 seats, the other block, 75 seats 31. Labour, which had been pivotal in the previous two sets of negotiations, denied itself the same possibility at this election. It performed disastrously, winning only 17 seats, down 16 on the previous election. The Progressive Democrats, also, who were to go into Government with Fianna Fáil after the 1997 election, lost 60% of their seats as compared with the previous election, and went into government with Fianna Fáil with only 4 seats. Policy blunders in the Progressive Democrat s manifesto, in both the economic and social policy spheres, undoubtedly contributed to this result 32. Previous studies have shown differences in parties policy emphases, but also some similarities in the policy areas of the economy and society 33. This paper is concerned with ascertaining whether the political parties in question do indeed stand for differing policy positions within those areas. If we were to discover that this were not the case, then the supposition would be either that the parties do not have differing policy stances, and that our votes are indeed wasted as far as expressing a policy preference is concerned, or that they do stand for different policies, but not policy areas that this paper has concerned itself with. In other words, each party would own its own issue, none of the other parties declaring policy intentions in the same area. Once the policy positions are gauged, I shall compare the results of the estimated party positions with those of the government programme. The circumstances surrounding the government formations that are being studied are all distinct; in 1992, Fianna Fáil and Labour went into government together after a process of post-electoral bargaining. Both parties knew how many seats each had in the Dáil. Assuming that parties care enough about policy not to be bought off with dubious promises of cabinet seats, then we might suppose that coalition policy is located at the weighted mean of the policy positions of government members; each member s policy position being weighted by the share that it controls of the total number of legislative seats controlled by all government members taken together 34. The justifications for such a supposition rest on two premises; the first is that the larger party is the more powerful in negotiations, the second, that representative democracy, in its purest form, warrants this to be the case. In 1994 bargaining again took place in a situation in which the exact number of seats each party held in the Dáil was known by all of the parties to the Rainbow Agreement 35. However, no election preceded this coalition formation process. Therefore the policy positions of the parties to that agreement are those of Fine Gael, Labour and Democratic Left as measured by their 1992 party manifestos. Despite the lapse of time, the hypothesis stands that coalition policy in this case is at the weighted mean of the members of government. The fact that no new manifestos were issued for the bargaining occasion means that we must still take their 1992 documents as the best official indicator of party policy. In the case of 1997, we might hypothesise the same again. All of the five major parties grouped into two possible coalition alternatives prior to the election. Fianna Fáil and the Progressive Democrats entered into government. Their programme for government was written after the election results were known. It is possible that the policy positions of both parties moved closer together in anticipation of post-electoral government formation, but this should not detract from our original hypothesis. O Reilly in Nealon, p.172, 1997 see Mair, ch.7 in Budge, Robertson and Hearl, 1987 Laver and Budge, p.426, 1992, based on Schofield 1993 Four by-elections were held in the period up to December 1994 when the new government was formed. Changes in the number of seats held by each party has been taken into account.

7 Unfortunately due to the limited number of cases being looked at, testing these hypotheses in a statistical sense is not possible. All this paper can claim are observed patterns and possible hypotheses for future testing when more data become available. Two Different Approaches to the Measurement of the Policy Positions of Irish Political Actors There are two basic processes involved in both the computer and expert coding of party manifestos; the first is data reduction, the process of making the text easier to analyse, the second is the data manipulation process, from which scales may be produced 36. The data reduction process involves a number of stages. For the expert coder, the first is to identify a coding scheme. The scheme adopted for the purposes of this paper is that suggested by Laver and Garry 37, which incorporates five hierarchically-structured domains: these are the economy, the political system, the social system, external relations and a general category. Within each of these domains, lie a number of more detailed branches. For example, within the economic domain, there are four alternative coding branches: more state involvement, less state involvement, neutral on state involvement and a stance on economic growth, pro, con and neutral. For each of these branches, further branches exist, and for each there exists a pro, con and neutral category. The second step for the expert coder is to define the unit of text to be coded, and again, this paper adopts Laver and Garry s 38 convention of an average of ten words per line. The coder then proceeds to place each text unit into one category of the coding scheme. That each unit of text is assigned to either a pro/con/neutral coding category allows us not merely to measure the policy emphases of the particular political actors being studied, as the MRG scheme allowed, but further to assign them to policy positions. That each branch is structured hierarchically allows for discretion on the part of the coder as to how detailed an analysis is warranted. Computer coding, by contrast, relies on the definition of a content analysis dictionary of words systematically associated with particular coding categories in relevant text 39. Words (as the units of text to be coded) are mechanically allocated to particular coding categories 40. There are pros and cons to each process of data reduction. Computer coding is a fast, reliable method of allocating text to a particular coding scheme. The dictionary is devised to parallel particular coding categories of a scheme, and whilst particular words may be more associated with one coding category than other, it need not be the case that a word s meanings are exclusive to one category. Computer coding can in this way fall foul of criticisms concerning the validity of its results 41. Expert coding on the other hand, while far less reliable than computer coding, is likely to be less queried as to the validity of the codings. This may, however, be to overestimate the possible objectivity of the coder. Once the document has been apportioned, text unit by text unit, to the coding scheme (either by expert coder or computer), the question of how to interpret the data arises. For a particular policy area to which some text units have been assigned, the assignations may fall into any combination of pro, con or neutral positions. Being concerned exclusively with the substantive policy positions of the political actors, we ignore those numbers of text units that have been coded neutral. To derive a policy position for the political actor in question, as approximated by the manifesto codings, we look at the relative balance of pro and con text Laver and Garry, p.6, 1997 Laver and Garry, p.13, 1999 Laver and Garry, p.14, 1999

8 units, taken as a proportion of all text units conveying information on this matter 42. Taking the example of units coded on the economy where +State+ indicates more state involvement in the economy, -State-, less involvement: Econ L = ECONOMY/+State+ Econ R= ECONOMY/-State- The following is the scale that estimates the policy positions; ECON LR = (Econ R-Econ L) / (Econ R + Econ L) The higher the number resulting on the scale, the more right-wing the policy position, the lower, the more left-wing. We now turn to look at the estimated policy positions for each of the Irish parties and governments, as estimated by applying the two coding approaches to the manifestos and Government Programmes of the period The Estimated Policy Positions of the Political Actors Table 1 shows the numbers of raw codings for computer and expert coding for each of the parties manifestos and each of the government programmes for the economic left-right and liberal-conservative values scales (each a particular coding category within the Garry and Laver schemes). Table 1: Counts of texts units and coded words in 1992, 1994 and 1997 Irish party manifestos/government Programmes Expert coding Economic Left Economic Right Values Liberal Values Conservative Irl DL Irl Lab Irl FF Irl FG Irl PD Irl Gov Irl Gov Irl DL Irl Lab Irl FF Irl FG Irl PD Irl Gov Laver and Garry, p.15, 1999

9 Computer coding Economic Left Economic Right Values Liberal Values Conservative Irl DL Irl Lab Irl FF Irl FG Irl PD Irl Gov Irl Gov Irl DL Irl Lab Irl FF Irl FG Irl PD Irl Gov Table 2 shows the raw scales derived from each of the documents. Each number in the Economic Policy Columns, is the raw coding Economic Left subtracted from the raw coding Economic Right divided by their sum. Each number in the Social Policy Columns is the raw coding values liberal for the actor in a particular year subtracted from the number of raw codings conservative, again divided by their sum Table 2: Raw Economic left-right and social values liberal-conservative scores for 1992, 1994 and 1997 party manifestos/government Programmes Economic Policy Social Policy Computer Revised expert Computer Revised Expert Irl DL Irl Lab Irl FF Irl FG Irl PD Irl Gov Irl Gov Irl DL Irl Lab Irl FF Irl FG Irl PD Irl Gov

10 Table 3 compares the standardised scores for both computer coding and expert codings on all thirteen observations for each scale. Table 3: Standardised economic left-right and social values liberal-conservative scores for 1992, 1994 and 1997 party manifestos/government Programmes Economic Policy Social Policy Computer Revised Expert Computer Revised Expert Irl DL Irl Lab Irl FF Irl FG Irl PD Irl Gov Irl Gov Irl DL Irl Lab Irl FF Irl FG Irl PD Irl Gov What we find is that the relationship between the party and government scores on the leftright dimension and on the values liberal-conservative is varied. On the economic scale, the Programme for Government 1992 and Fine Gael 1997 each have radically different policy positions depending on whether you take the computer coding results or the expert coding as valid. The experts systematically rate Fine Gael as being more right-wing than does the computer. The Labour position in 1992, also varies according to method of measurement used: left-wing by expert coding and mildly centrist by computer. All of the other party positions recorded in the economic category by one method are closely mirrored by the other. More substantively however, Fianna Fáil shows up as centrist for both elections when coded by either method. Labour in 1992, as we have seen, as centrist or left by computer and expert coding respectively, and scores left-wing in 1997 by both methods. Fine Gael is centrepositioned by computer in 1992, and on the left in The Progressive Democrats score as very right-wing by either method, Democratic Left as most extreme left. The economic policy position of the government programme for 1992 as estimated by expert coding seems to lie very close to the Labour Party economic policy position, but by computer, bears no relation to either party s position. The 1994 government position is centrist by both types of estimation. The 1997 results on the economic scale show a very high association between each party s policy positions, as estimated by the two alternative methods, with the exception, as mentioned earlier, of Fine Gael. The government position in 1997 is mildly left-wing, and certainly to the left of both of its constituent parties, Fianna Fáil and the Progressive Democrats. As regards the liberal-conservative scale, a larger number of cases produced different results according to the coding method used. By whatever, method, however we find Fianna Fáil 1992 to be conservative, Labour and Democratic Left to be liberal. Both of the programmes for government, 1992 and 1994, score liberal by the expert coding approach, yet the 1992 programme scores conservative by computer coding. Via expert coding, the 1992 programme scores closest to the position of the Labour Party, the 1994 to the position of Fine Gael. The 1997 scores exhibit quite a good correlation between the two coding

11 methods, excepting the estimations of the Fine Gael values position. The 1997 government manifesto scores closer to the Fianna Fáil position by computer coding, yet closer to the Progressive Democrats position by expert coding. Both approaches, however, systematically place the government programme as more liberal than either of the party s positions. Before looking more substantively at the relationship between party positions and their governments, I should point out the high face validity of the expert coding method, especially for Democratic Left and Labour score quite left on the economic scale, Fianna Fáil centre and the others right. In terms of social policy as well, the results ring true - Fianna Fáil comes out as conservative, the rest liberal. The 1997 scores show the same party positions on the economic scale, though some movement has occurred since 1992 (Fine Gael, and the Progressive Democrats moving markedly more towards the right). In 1997 the Progressive Democrats and Fine Gael show a move to the conservative side of the values scale. What this analysis does illustrate is that the parties by and large are distinguishable by policy position in these two domains. Testing the hypothesis We now want to test the hypothesis that coalition policy is located at the weighted mean of the policy positions of government members; each member s policy position being weighted by the share that it controls of the total number of seats in the legislature 43. Table 4 gives the policy positions for each government on the basis of the weighted means of their member parties and compares this with their measured government positions as estimated using both coding methods. Table 4: Coalition policy the weighted means of the policy positions of the government members compared with their coded programme positions Economic Policy Expert Coding Weighted Mean Actual Gov. Pos Computer Coding Weighted Mean Actual Gov. Pos Social Policy Expert Coding Weighted Mean Actual Gov. Pos Computer Coding Laver and Budge, p.426, 1992

12 Weighted Mean Actual Gov. Pos What we find, is that in more than half of all the cases, the weighted mean of the government member s policy positions approximates that position recorded by a coding of the government document. In 1992 the weighted policy positions of Labour and Fianna Fáil on the economic scale, as determined by expert coding, corresponds quite well to its coded programme, but not at all when determined by computer coding. In 1994 the economic policy positions of Fine Gael, Labour and Democratic Left when expert-coded and computer-coded lie close to that of their government programme. But in 1997, there is no correspondence between the 1997 economic positions as determined by either method of estimation and the weighted means of the government parties positions. It is on the social policy dimension however that our hypothesis seems to work best. The expert coding score for the 1994 coalition programme of Fine Gael, Labour and Democratic Left closely approximates the coalition position as determined by the weighted mean of their manifestos. Further the social policy positions of all three governments, as estimated by computer coding, mirror closely their weighted mean positions. Overall the results are varied. There are more cases of the coalition policy as determined by the weighted mean positions of the incumbent parties wildly deviating from their coded programme positions than there are with any seeming relationship in the economic policy sphere. On the other hand, our hypothesis seems to work well with regards to social policy. We might therefore suppose that economic policy is more hotly fought over. Ultimately, the hypothesis that coalition policy is determined by the weighted mean of its parties policies seems to explain the determination of the coalition policy positions of the Rainbow Government in What seems to have happened in 1992, according to the expert coding policy position estimations (see Table 3) is that Labour managed to secure a disproportionate number of its economic policy positions in the government programme. Anecdotally, as we have seen, this is what the Labour Party claimed 44. The situation in 1997 (see Table 3) is that the economic policy position of the government is far closer to that of Fianna Fáil than the Progressive Democrats. What we found is that in 1992, the smaller party, Labour, achieved greater endorsement of its economic policies in the programme. Yet in 1997, the smaller party, the Progressive Democrats, seems to have little of its agenda endorsed in the coalition programme. The answer to this puzzle lies in the fact that for the government formation of 1992, Labour was not merely the pivotal party, but had won a phenomenal number of seats as compared with its earlier electoral performances. While Labour had gained 16 seats, Fianna Fáil had lost 9. In 1997, the Progressive Democrats were promised partnership in government by Fianna Fail prior to the election. Fianna Fail kept this promise and did not, by all accounts, broach the subject of forming a government with Labour. However the Progressive Democrats weak performance at the election, at least partially driven by some well-publicised policy blunders, must have seriously weakened their bargaining position. Fianna Fáil had performed well returning 77 deputies, a gain of nine seats. Compounding the Progressive Democrat s difficulties is that much discussion about them since the election has centred on the question of their viability as a distinct party, and has gone as far as mooting the possibility of their merging with Fianna Fáil. Clearly, they had a poor bargaining position once it came to coalition negotiations. Conclusion So what might we conclude? If we accept the computer coding scores for the coalitions social policy positions, then our hypothesis works well. It does seem to be the case that the coalition policy position is derived from the weighted mean of the incumbent parties positions, for each of the three governments studied. In the case of the Rainbow Coalition of 1994, we would similarly fail to reject our hypothesis in the economic sphere, where Finlay, p.136, 1998

13 estimations by both computer and expert coding place coalition policy at the position of the weighted means of the government parties positions. However, it does not work as well with the 1992 and 1997 economic positions of the governments. What we found in the expert-coded cases of the 1992 and 1997 governments, was that the economic policy position of the coalition as a whole was biased towards the position of that coalition partner whose seat share had increased as compared with the its previous share of Dáil seats. Perhaps on reflection such an outcome is not that surprising given the emphasis placed on economic policy by all the parties 45 in their manifestos, and the saliency of economic issues amongst the electorate. We might suppose that this is the policy area in which the parties are most concerned to leave their distinctive mark. Of course, in 1994 the absence of an election prior to coalition bargaining meant that the immediacy of a good electoral performance was not carried into the negotiations. This perhaps explains why coalition policy was located at the weighted mean of the positions of the government parties. A further exercise, at a later date, might therefore be to see whether or not in the case of economic policy, coalition policy is weighted in favour of that party on the relative ascendancy going into the post-electoral coalition negotiations. Simply put however, what a government stands for may not be placed in its government document. If the parties vehemently object to some of each other s policies, these policies may just be left out of the programme. Whilst these policy desires may be subdued to get into government once in, they may re-emerge. The defining variable, according to Laver and Budge 46 and Laver and Shepsle 47, is who gets to control which government portfolios. Thus their argument would be that our estimation of policy payoffs has been premature and that a decision on whether or not the incumbent parties got their policies adopted by their partners, should be deferred until the allocation of portfolios has been completed. That, however, is another paper. see Garry and Mansergh, p.24, forthcoming p.44,

14 Bibliography Budge I. and Farlie D., Explaining and Predicting Elections (George, Allen and Unwin, 1983) Budge I., Robertson D. and Hearl D. (eds.), Ideology, Strategy and Party Change: Spatial Analyses of Post-War Election Programmes in 19 Democracies (Cambridge University Press, 1987) Castles F., The Impact of Parties (Sage, London, 1982) Garry J. and Mansergh L., Irish Party Manifestos in the 1997 General Election, forthcoming chapter in How Ireland Voted Finlay F., Snakes and Ladders (Dublin: New Island Books, 1998) Ginsberg, Elections and Public Policy, APSR Vol. 84, No.1, March 1990 Laver M. and Budge I., Party Policy and Government Coalitions (London, 1992) Laver M. and Garry J., Estimating Policy Positions from Party Manifestos, UMEA paper, June 1997 Laver M. and Garry J., Estimating Policy Positions from Political Texts (Mannheim, March 1999) Laver M., Mair P. and Sinnott R., How Ireland Voted: The Irish General Election, 1987 (PSAI Press, 1987) Nealon T., Nealon s Guide to the 28 th Dail & Seanad (Gill and MacMillan, 1997) Schmidt M., When Parties Matter; A review of the possibilities and limits of partisan influence on public policy in the EJPR Vol. 30, pp , Sept Sinnott R., Irish Voters Decide (Manchester University Press, 1995) Strom K., A Behavioural Theory of Political Parties, in the AJPS Vol. 34, No.2, May 1990

15 Manifestos Analysed The following are the manifestos and programmes for government used in the analysis: Progressive Democrats Get the Nation Working 1992 Democratic Left A Programme for Change and Progress 1992 Labour Making Ireland Work 1992 Fine Gael Let s Bring Out the Best in the Country 1992 Fianna Fail We Can Make It Happen 1992 Fianna Fail and Labour Programme for Government A Government of Partnership 1992 Fine Gael, Labour and Democratic Left A Government of Renewal 1994 Progressive Democrats A New Deal 1997 Labour Making the Vital Difference 1997 Democratic Left Manifesto 1997 Fine Gael Securing and Sharing our Prosperity 1997 Fianna Fail People Before Politics 1997 Fianna Fail and Progressive Democrats Programme for Government An Action Programme for the Millenium 1997

16