Ministerial Autonomy or Ministerial Accommodation? Contested Bases of Government Survival in Parliamentary Democracies

British Journal of Political Science

Volumn 29, Issue 2, 1999, Pages 369-394

  Warwick, Paul V ^a  

^a SIMON FRASER UNIVERSITY   (Canada)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0033439443     PISSN: 00071234     EISSN: None     Source Type: Journal    
DOI: 10.1017/S0007123499000162     Document Type: Article

References (46)

1. Richard McKelvey, 'General Conditions for Global Intransitivities in Formal Voting Models', Econometrica, 47 (1976), 1085-111.
2. See, for example, the analyses of party manifestos reported in Ian Budge, David Robertson and Derek Hearl, eds., Ideology, Strategy, and Party Change: A Spatial Analysis of Post-War Election Programmes in Nineteen Democracies (Cambridge: Cambridge University Press, 1987), which showed most West European party spaces to be multidimensional.
3. The approach was introduced in Michael Laver and Kenneth Shepsle, 'Coalitions and Cabinet Governments', American Political Science Review, 84 (1990), 873-90, and elaborated much more fully in Michael Laver and Kenneth Shepsle, Making and Breaking Governments: Cabinets and Legislatures in Parliamentary Democracies (Cambridge: Cambridge University Press, 1996).
4. The approach was introduced in Michael Laver and Kenneth Shepsle, 'Coalitions and Cabinet Governments', American Political Science Review, 84 (1990), 873-90, and elaborated much more fully in Michael Laver and Kenneth Shepsle, Making and Breaking Governments: Cabinets and Legislatures in Parliamentary Democracies (Cambridge: Cambridge University Press, 1996).
5. The first assumption can be relaxed to allow a single ministry to have jurisdiction over more than one policy dimension; as far as is known at present, the freedom of ministers to impose their parties' policies in their ministries cannot (Laver and Shepsle, Making and Breaking Governments, pp. 235-45).
6. Laver and Shepsle, Making and Breaking Governments, pp. 66-89.
7. Joseph Kadane, 'On the Division of the Question', Public Choice, 13 (1972), 47-54. Kadane's result assumes a continuous policy space, single-peaked voter preferences and separable dimensions. Laver and Shepsle (Making and Breaking Governments, p. 81) show that the theorem applies when alternatives are limited to those satisfying the portfolio allocation assumptions. That the dimensions are separable is embodied in their assumption that policy dimensions correspond with the jurisdictions of key ministries and that ministers are autonomous from one another.
8. Joseph Kadane, 'On the Division of the Question', Public Choice, 13 (1972), 47-54. Kadane's result assumes a continuous policy space, single-peaked voter preferences and separable dimensions. Laver and Shepsle (Making and Breaking Governments, p. 81) show that the theorem applies when alternatives are limited to those satisfying the portfolio allocation assumptions. That the dimensions are separable is embodied in their assumption that policy dimensions correspond with the jurisdictions of key ministries and that ministers are autonomous from one another.
9. Michael Laver and Kenneth Shepsle, eds., Cabinet Ministers and Parliamentary Government (Cambridge: Cambridge University Press, 1994).
10. Laver and Shepsle, Making and Breaking Governments, pp. 158-92.
11. The two examples they considered were Germany in 1987 and Ireland in 1992-93 (Laver and Shepsle, Making and Breaking Governments, pp. 208-14).
12. An 'out-of-equilibrium' or 'non-equilibrium' government refers to a government that does not embody one of the equilibria of the portfolio allocation model; these governments may occupy equilibria that emanate from other theoretical perspectives.
13. Laver and Shepsle, Making and Breaking Governments, p. 78.
14. Whether an MSP forms a government by itself or enters a coalition in its winset depends on whether it can credibly hold out for a single-party government. Either outcome, however, is in equilibrium. The argument that government stability may suffer because this credibility is likely to be highly variable would undercut the usefulness of the equilibrium concept and is not asserted by Laver and Shepsle.
15. Laver and Shepsle, Making and Breaking Governments, pp. 102-6, 184-5.
16. This is true both for equilibrium governments and between them and out-of-equilibrium governments. The latter implication derives from the fragility of out-of-equilibrium governments and the fact that there is no particular reason for them to be compact. The former implication is suggested by Laver and Shepsle's finding (Making and Breaking Governments, pp. 205-6) that equilibria based on strong parties are less robust to shocks than those based on empty-winset DDMs. Since VSPs occupy the dimension-by-dimension median, the consequence could be a spurious association between ideological compactness and robustness or longevity. I owe this latter observation to Laver and Shepsle. Note, however, that this reasoning cannot explain the link between majoritarianism and survival. Since Laver and Shepsle use VSP status to explain the formation of minority governments (pp. 262-6), we should expect minority governments to survive longer than majority governments, which is not the case.
17. The most recent study advancing the importance of majority status and ideological compactness for government survival is Paul V. Warwick, Government Survival in Parliamentary Democracies (Cambridge and New York: Cambridge University Press, 1994).
18. The systems are: Austria, Belgium, Denmark, Finland, Iceland, Ireland, Italy, Luxembourg, the Netherlands, Norway, Sweden and (West) Germany. Legislatures with majority parties were excluded because, although majority parties are also VSPs, these situations conform to PA theory in only a trivial sense - all theories expect such parties to form stable governments. The only other exclusions were legislatures containing sizeable parties that had disappeared by 1989 and consequently were not considered in the Laver and Hunt survey (see below); the Netherlands before the amalgamation of the three confessional parties into the Christian Democratic Appeal is a prominent example. The WINSET program is available for downloading at FTP.TCD.IE; the files are located in the directory /PUB/POLITICS.
19. Michael Laver and W. Ben Hunt, Policy and Party Competition (New York: Routledge, 1992).
20. Laver and Shepsle's (Making and Breaking Governments, pp. 152-4) procedure was to pick the two highest rated portfolios for each system unless there was no statistically significant difference between the ratings allocated to the second and third portfolios, in which case all three were included.
21. Laver and Shepsle, Making and Breaking Governments, pp. 149-57.
22. Jaap Woldendorp, Hans Keman and Ian Budge, 'Political Data 1945-1990: Party Government in 20 Democracies', European Journal of Political Research, 24 (1993), 1-120; Keesing's Contemporary Archives (London: Keesing's Publications, 1945-90). In some instances, Woldendorp et al. list more than one minister for a given portfolio whenever two or more ministries partake of the same generic type. Thus, if a system has both a finance and a budget minister, Woldendorp et al. would list both under the 'Finance' category. The ranking of ministries in Laver and Hunt provided the means by which these issues could be resolved in most cases. One exception is Finland after 1982, where 'Finance' and 'Finance (Second)' ministers are listed. Maximum latitude has been granted the theory here: if either portfolio holder fits the prediction, it has been counted as a success for the theory.
23. Jaap Woldendorp, Hans Keman and Ian Budge, 'Political Data 1945-1990: Party Government in 20 Democracies', European Journal of Political Research, 24 (1993), 1-120; Keesing's Contemporary Archives (London: Keesing's Publications, 1945-90). In some instances, Woldendorp et al. list more than one minister for a given portfolio whenever two or more ministries partake of the same generic type. Thus, if a system has both a finance and a budget minister, Woldendorp et al. would list both under the 'Finance' category. The ranking of ministries in Laver and Hunt provided the means by which these issues could be resolved in most cases. One exception is Finland after 1982, where 'Finance' and 'Finance (Second)' ministers are listed. Maximum latitude has been granted the theory here: if either portfolio holder fits the prediction, it has been counted as a success for the theory.
24. Few systems range over an entire twenty-point scale; in fact, the average range of party systems is 12.24 points. This means that the changes being simulated are somewhat larger as a proportion of the extant or 'realistic' range of party systems, as opposed to the maximum imaginable.
25. Laver and Shepsle, Making and Breaking Governments, pp. 209-14.
26. Laver and Shepsle, Making and Breaking Governments, p. 196. Note that this implies that out-of-equilibrium governments formed because no equilibrium government is possible in a given legislature should be considered as explanatory failures.
27. This definition was first proposed by Eric C. Browne, Dennis W. Gleiber and Carolyn Mashoba, 'Evaluating Conflict of Interest Theory: Western European Cabinet Coalitions, 1945-80', British Journal of Political Science, 14 (1984), 1-32. We shall follow Warwick's precedent in counting all resignations, not just those that are accepted by the head of state, as marking the breakdown of a government. This nuance affects relatively few governments and does not substantively alter any of the results to be reported. The decision to count all elections, even those that a government survives, as terminal events is discussed below.
28. David Cox, 'Regression Models and Life Tables (with Discussion)', Journal of the Royal Statistical Society, B34 (1972), 187-220; and David Cox, 'Partial Likelihood', Biometrika, 62 (1975), 269-76. Two useful introductions to event history analysis are: Paul Allison, Event History Analysis: Regression for Longitudinal Event Data (Beverly Hills, Calif.: Sage, 1984); and Hans-Peter Blossfeld, Alfred Hamerle and Karl Mayer, Event History Analysis (Hillsdale, NJ: Erlbaum, 1989).
29. David Cox, 'Regression Models and Life Tables (with Discussion)', Journal of the Royal Statistical Society, B34 (1972), 187-220; and David Cox, 'Partial Likelihood', Biometrika, 62 (1975), 269-76. Two useful introductions to event history analysis are: Paul Allison, Event History Analysis: Regression for Longitudinal Event Data (Beverly Hills, Calif.: Sage, 1984); and Hans-Peter Blossfeld, Alfred Hamerle and Karl Mayer, Event History Analysis (Hillsdale, NJ: Erlbaum, 1989).
30. David Cox, 'Regression Models and Life Tables (with Discussion)', Journal of the Royal Statistical Society, B34 (1972), 187-220; and David Cox, 'Partial Likelihood', Biometrika, 62 (1975), 269-76. Two useful introductions to event history analysis are: Paul Allison, Event History Analysis: Regression for Longitudinal Event Data (Beverly Hills, Calif.: Sage, 1984); and Hans-Peter Blossfeld, Alfred Hamerle and Karl Mayer, Event History Analysis (Hillsdale, NJ: Erlbaum, 1989).
31. David Cox, 'Regression Models and Life Tables (with Discussion)', Journal of the Royal Statistical Society, B34 (1972), 187-220; and David Cox, 'Partial Likelihood', Biometrika, 62 (1975), 269-76. Two useful introductions to event history analysis are: Paul Allison, Event History Analysis: Regression for Longitudinal Event Data (Beverly Hills, Calif.: Sage, 1984); and Hans-Peter Blossfeld, Alfred Hamerle and Karl Mayer, Event History Analysis (Hillsdale, NJ: Erlbaum, 1989).
32. Gary King, James Alt, Nancy Burns and Michael Laver, 'A Unified Model of Cabinet Dissolutions in Parliamentary Democracies', American Journal of Political Science, 34 (1990), 846-75, and Warwick, Government Survival in Parliamentary Democracies.
33. Gary King, James Alt, Nancy Burns and Michael Laver, 'A Unified Model of Cabinet Dissolutions in Parliamentary Democracies', American Journal of Political Science, 34 (1990), 846-75, and Warwick, Government Survival in Parliamentary Democracies.
34. It turns out that censoring these cases does not affect the patterns of results to be reported in any significant way.
35. These rates do not depend on accurate measurement of party positions or dimensional saliences. In the simulations involving 1,000 sets of shocks to these parameters for each government formation situation, PA equilibrium governments could still be formed 91.8 per cent of the time, on average.
36. An analysis of variance yields an F-ratio of 0.062 (p = 0.804).
37. All dichotomous covariates used in this analysis are coded 1 for the possession of the trait in question and 0 otherwise.
38. This is also true when the covariates are entered separately.
39. Since both Equilibrium Status and Equilibrium Robustness range from 0 to 1, the magnitudes of the coefficients can be directly compared. The robustness coefficient is approximately four times as large.
40. A government that is highly diverse on one dimension but non-diverse on another, for example, could receive a total or average diversity score that is very like that received by a government with moderate diversity on both dimensions, even though it is arguably much more vulnerable. An obstacle to using range or diversity on each policy dimension separately is that the second and third dimensions do not always measure the same thing in each country. These dimensions can be unpacked so that the second dimension only measures attitudes towards the Soviet Union and the third dimension (where it exists) only measures interior affairs, but the consequence is that diversity on the dimensions turns out to be fairly highly correlated. For instance, the correlation between range as measured on the first and second (Soviet Union) dimensions is 0.747. This means that it would be inappropriate to try to assess independent effects for three separate measures of policy range.
41. Warwick, Government Survival in Parliamentary Democracies, p. 32.
42. The source for this information is Keesing's Contemporary Archives (1945-90).
43. For any one shock specification, a stream of 1,000 simulations has to be performed on each of the 173 legislatures in the dataset. This undertaking would have to be repeated for every shock specification one wanted to try.
44. Laver and Shepsle, 'Events, Equilibria and Government Survival', American Journal of Political Science, 42 (1998), 28-54, pp. 44-5.
45. The coding of party positions on these dimensions is given in Lawrence C. Dodd, Coalitions in Parliamentary Government (Princeton, NJ: Princeton University Press, 1976), pp. 246-55. The connection of government survival with the degree of clerical-secular and regime-support diversity was explored in Warwick, Government Survival in Parliamentary Democracies.
46. As noted earlier, Laver and Shepsle (Making and Breaking Governments, p. 157) found that using differential dimension saliences did not alter their empirical results significantly; casual observation suggests the same conclusion here.