Resumé

"Termination of government coalitions" is an ambiguous notion. While the concepts of "termination" and "coalition" do not present insurmountable problems, the concept of "governments" is indeed very tricky. Empirical findings on termination of governments are highly dependent upon the definition of government, as shown by a Danish case study of 1945-93. A government is defined on the basis of party composition. It is found that Danish majority coalitions terminate because they lose the first upcoming election, whereas minority coalitions terminate for party strategic reasons, notably decisions made by the pivotal Radical Liberal Party.

Introduction

This article* analyses some empirical and theoretical aspects of government termination in Denmark since 1945. As it turns out that the idea of "government coalition termination" is more ambiguous than one would initially asssume, it is necessary to deal with a number of conceptual problems and to resolve some important definitional matters from the very beginning. This is done in the following two sections. Then some general and comparative studies of government termination causes are briefly reviewed in a fourth section. Having thus cleared the ground, attention is turned to actual experiences with the termination of various types of governments in Denmark. In view of the findings obtained, the concluding section offers a few suggestions as to how the present understanding of government termination could be improved.

What is "Termination of Government Coalitions"?

Any analysis of coalition government termination, whether it deals with the phenomenon in general or whether it focuses on specific countries and cases, should begin with a clarification and definition of the core terms and concepts used. In particular, the object under study (the "dependent

variable") should be precisely defined from the outset. To do so in the present context turns out to be a lot more problematic than one would initially assume, not because of Danish peculiarities but because of what seems to be a general theoretical confusion about the proper answer to the crucial question of what constitutes a "government".

The concept of "termination" does not appear to present insurmountable problems. A government may be said to terminate when it, for some reason, is replaced by another government of some sort. A government (or cabinet, the two terms are used interchangeably in this paper) thus exists until another government takes over. The real problem is not with the concept of termination, nor with the concept of coalition for that matter, but, as will be discussed below, with the conceptualization of a "government".

A coalition government may simply be defined as a government in which at least two parties occupy ministerial posts. Such a definition excludes from the group of coalition members parties supporting and cooperating with a government without formally participating in it by having a share of the cabinet posts. This leaves us with the crucial question of how to define a "government".

Recent surveys of the relevant literature on government formation and duration (Budge & Keman 1990; Laver & Schofield 1991) show that there is no general agreement on how best to decide when a government has ended and by implication when a new government has been formed. The first coalition theory analysis of Danish governments, which also included supporting parties (Damgaard 1969), took it for granted that a change of government was solely determined by a change in the parliamentary party basis of the cabinet. Everybody agrees that a change in the party composition of a cabinet amounts to a change of government,¹ but most authors apply additional criteria according to which a government may be said to have ended. To mention only a few recent examples (we shall return to the matter below), Laver & Schofield (1991,147) consider a new government as forming whenever there is an election; Strøm (1984) adds a change of prime minister for whatever reason as a third relevant criterion, and Budge & Keman (1990) argue that even a formal resignation of a government in the inter-election period followed by the re-formation of the government with the same prime minister and party composition counts as a new government ²

Whatever the merits of the various criteria in relation to specific research purposes, it is obvious that the definition selected affects the number of governments to be counted as such. This can be seen in Table 1 for Danish coalition governments in the period after World War 11. A total of 15 coalition governments can be identified using the criteria mentioned. Ten of the coalitions are identified on the basis of the first criterion (I: party composition). Inclusion of a change of prime minister (II) adds two coalition

governments and general elections (III) a further three coalition governments. However, no additional coalitions are added by the fourth criterion of resignation/re-formation of the "same" government (IV). In the table it is shown that two majority governments had a new prime minister appointed during the inter-election period, but the table does not reveal that the reason was serious illness in both cases. The fact that general elections produced three minority coalitions actually means that a minority coalition survived an election on three occasions. Clearly, then, it matters how a government is conceptualized. This raises the question of how a government can most sensibly be defined. A closer look into the various possibilities is obviously needed.

As mentioned, scholars generally agree that a change in party composition is a change of government. The question therefore is whether additional criteria should be accepted as indicating a change of government, and, if so, which ones. The answer given by Lijphart (1984), Dodd (1976) and Damgaard (1969, 1974) is "no". To quote Lijphart (1984, 80-81): "We shall consider a cabinet to remain the 'same' cabinet if its party composition does not change". This definition could presumably provide a solid and unequivocal basis for investigations of cabinet terminations. But, as already mentioned, there are at least three other possible answers.

Blondel (1968) and Taylor & Herman (1971) also took a change of prime minister to constitute a new government even if the party composition was unchanged. The same procedure was adopted by Browne et al. (1986), Strøm (1986) and Budge & Keman (1990). The rationale is basically that a change of prime minister is normally thought to indicate that a government crisis has occurred. But that may not actually be true. Thus Taylor & Laver (1973), Warwick (1979) and Powell (1982) only accepted the criterion in cases where the replacement of a prime minister was clearly politically motivated and not just due to non-political factors such as death or ill health. With such a qualification the criterion looks sensible. It is certainly

reasonable to claim, for example, that the United Kingdom got a new Conservative government when John Major replaced Margaret Thatcher in 1990. Such involuntary replacement of a prime minister in a government of the same party or parties, by the way, has not occurred in Denmark since the first decade of the 20th century.

Most authors further consider any government to end when general elections are held even if the "same" government continues after the elections (e.g. Taylor & Laver 1973; Powell 1982; Strøm 1984; Budge & Keman 1990; Laver & Schofield 1991; Katz 1993). A priori it seems very odd that elections should always produce a "new" government even if the old one appears to continue in office, and in fact does so. One may be permitted to assume that the election criterion has also been devised to enlarge the sometimes very small numbers of observation (N) in quantitative analyses, and therefore to suspect that the resulting statistical figures are not always valid and sometimes perhaps even misleading. Lijphart (1971) once suggested that increasing the number of cases as much as possible could sometimes minimize the "many variables, small N" problem of the comparative method. In this context, however, it seems that the number of cases is increased by partitioning real cases into two or more artificial "sub-cases".

Explicit reasons for the "election means the end of government" criterion are not always stated. But Strøm (1984, 201) argues that his analysis of minority governments requires the election criterion to ensure that majority and minority governments are mutually exclusive categories, which may be correct. In addition, Strøm claims that the criterion helps to prevent a government from varying significantly over time in parliamentary support, which is correct but not relevant for all purposes.

Budge & Keman (1990, 14-15) note that we normally think of elections as one of the events defining a government's period in office, which is not universally true. Some governments may be thought of as being good at winning or surviving elections, in particular if they possess the prerogative to call elections before the end of a constitutionally fixed maximum election period. Such governments are even expected to use that opportunity more or less skilfully.

Laver & Schofield (1991, 147), focusing on the duration of coalitions, are somewhat more outspoken on the issue. They argue that even if a cabinet does not resign, does not change its party composition, and does not change its prime minister, it will almost invariably after an election experience a different "constellation of party weights and policy positions" and in that sense be a different cabinet. While this may be true, the cabinet is still the same in all the other respects mentioned. Furthermore, it may be argued that a government can find itself in even more different positions during the inter-election period because of, for example, changes in the

political agenda resulting from international or domestic events, major shifts in public opinion ratings, party strategic moves or intra-party disagreements. Hence, there are many other senses in which a government may become a "different" government. If this is so, then why should these not also be taken into account?

Finally, there is the criterion of formal resignation. The issue is whether a new government should be considered formed if it consists of the same parties and is headed by the same prime minister as the one having formally resigned, and whether so even without an intervening election. Some authors think so. Thus, Warwick, seeking "as accurate a measure of cabinet instability as possible", frankly states that: "A cabinet terminates when it terminates, after all, regardless of the composition of the cabinet that succeeds it" (1979, 468). Well, does it? That is the question. Warwick's real argument is that when studying cabinet instability it would not be proper to gloss over cases where a cabinet fell apart but was subsequently reformed by the same parties. He may be right, but one could also argue that cabinet instability can occur without resignation, and that resignation/ reformation may indicate a certain government stability because "no viable alternative can be found" (to use Warwick's own words).

Budge & Keman argue that there are good reasons why governments resign even if they return unchanged in terms of party composition and prime minister: "It is likely that after such a re-formation power relationships and/or policy priorities will have changed, so it makes sense to distinguish separate administrations before and after such events" (1990, 15; ef. Woldendorp et al. 1993, 5). Although the argument does make sense, it is very similar to the one used by Laver and Schofield to justify the election criterion and therefore open to similar counter-arguments. It is also possible that a resignation is used as a purely tactical weapon to make clear that the cabinet can stay in office. Budge & Keman (1990, 215) further note that their decision to include formal resignation increases the total number of government termination cases. That is true, but of course not a valid argument for the decision made.

Laver & Schofield (1991,145-146) reject the resignation criterion on the grounds that national rules, written or unwritten, differ too much with respect to situations in which a government is supposed to resign. In some systems, governments resign although everybody knows that they will almost immediately be reconstituted while this is not the case in other systems. In other words, different national procedures and norms may prevent genuine cross-national comparability if the formal resignation criterion is adopted.

In what follows the primary criterion of "government" is party composition
because after all it is the most basic one. Involuntary change of a
prime minister also appears to be a reasonable criterion. In substantive

terms it is much more difficult, if not impossible, to justify elections and formal resignations as criteria for government change. But even if the two criteria appear dubious it might still be instructive to have a look at them, if only to see what differences they might make for the understanding of causes and consequences of coalition government termination.

Coalition Termination, Single-Party Cabinets and Support

Before the problem of coalition termination causes is addressed, a couple of preliminary questions must be considered. The first question relates to the overall theme of termination of government coalitions which is narrower than simply the termination of governments. It obviously excludes the dissolution of single-party governments. This makes sense in view of the rather strong efforts made by scholars in the past two decades to understand coalition formation processes and outcomes: if the formation and duration of government coalitions are relevant and interesting to political scientists, why should attention not also be paid to the disappearance of the same coalitions? The challenge contained in this question stems from the fact that theories of coalition formation do not necessarily imply explanatory statements about the dissolution of government coalitions.

The study of government coalition formation was originally heavily influenced by William Riker's pioneer work. In The Theory of Political Coalitions (1962) Riker proposed a theory of coalitions applicable in various social situations and not a theory of the formation and termination of government coalitions in multiparty systems. But, interestingly, his theory of minimum winning coalitions ("the size principle") was also capable of explaining why overwhelming majorities, or coalitions of the whole, formed by reason of some accidental circumstance would break down (Riker 1962, ch. 3). According to Riker's size principle, such coalitions simply did not pay off to the participants involved. In this limited sense Riker also contributed, perhaps unwittingly, to a theory of coalition termination, but most of the work remained to be done.

Subsequent research on government coalitions did not pay much attention to the dissolution of coalitions although it was concerned with their "durability". However, as Budge & Keman argue (1990, ch. 6), breakdown and termination should be studied in their own right. To this must be added that some possible causes of termination may be relevant to all kinds of governments while others may only affect coalition governments, and that - as illustrated above - some causes of termination may simply follow from a selected definition of government.

The second preliminary point relates to the "support party" phenomenon.

In some multiparty systems, and not least in the Scandinavian ones, minority governments often rely on support from parties outside of the cabinet. Such more or less developed or formalized support arrangements can obviously also break down for various reasons that can be empirically investigated. But since it is sometimes very difficult, or perhaps even impossible as Lijphart (1984, 60) claims, to determine which parties are support parties, it may be sensible to focus only on parties actually and formally participating in cabinets, at least in comparative projects involving large-scale collections of data that cannot be sufficiently inspected in detail.

The upshot of these remarks is that to focus exclusively on the termination of formal government coalitions is reasonable, at least as an initial effort. But it is also sensible to be aware of the existence of informal coalitions, and of the possibility that some termination causes are common to governments in general and not just to coalition governments.

In Table 2 all Danish single-party and coalition governments since 1945 are listed according to the four criteria discussed. The figures for the two types of governments are surprisingly similar in all respects. Table 2 buttresses the argument that it matters a great deal how a government is defined, be it a one-party or coalition government. If the purpose is to study the duration of governments, for example, it makes a difference whether 18 or 28 governments are counted since 1945. The table also suggests that some of the reasons for government termination may be relevant for both types of government. Formally at least, an equal number of the two types of government ends because of a non-political change of prime minister and general elections. Further, the inter-election resignation/re-formation criterion is inapplicable in both cases.

These observations once again suggest that the most interesting question to pursue is why a government ends in terms of its replacement by a cabinet with a different party composition. In this respect single-party and coalition governments might differ to a considerable extent.

Causes of Government Termination

In somewhat different ways at least three comparative studies have dealt with the causes of government termination. They all used an inclusive definition of a government, that is, for practical purposes they focused on a maximum number of cases to be explained. Klaus von Beyme (1982, 388) listed several possible causes for the termination of governments in 17 Western democracies in the period of 1947-80. Although he found that it could sometimes be difficult methodologically to determine the "real reason" for government dissolutions, the evidence he provided showed that elections were by far the most frequent cause of termination, but a substantial number of government dissolutions were caused by coalition breakdowns. Other causes, such as defeats in parliament, death or ill health of the prime minister, voluntary resignations and the widening of coalitions were less frequent. However, von Beyme's figures also showed that the distribution of causes varied quite a lot across countries.

Budge & Keman (1990, ch. 6), in their analysis of almost the same countries in almost the same time period (1950-83), found elections to be the most frequent cause of termination.³ Although their operationalizations are not strictly similar to those of von Beyme, their overall results are pretty much the same.

Browne et al. (1986) took a rather different approach in their "critical events perspective" on the dissolution of governments in the Nordic countries 1945-80. But with the partial exception of Finland, elections none the less came out as the most frequent type of government terminating events compared to intra-party, international and ill health events. Browne and associates also tried to map the policy area and the source of events in an attempt to get hold of the complexity involved in many government termination events or processes. For example, according to the authors, the Danish J. O. Krag's Social Democratic minority cabinet (supported by the Socialist People's Party 1966-68) was dissolved in January 1968 because of a series of terminal events including an unscheduled election, intra-party disputes (in the support party, presumably) in an economic matter, and having parliament, opposition parties and the prime minister as important actors.

All the studies mentioned are quite useful. They also tend to support one of the concluding remarks of Budge & Keman: "Breakdown and termination are perhaps the most difficult aspects of government behaviour to analyse, in part because of the lack of previous research and in part because of difficulties in pinning down and putting into focus such diffuse phenomena" (1990, 187).

What seems to be needed in this research situation is studies in more
detail of concrete cases of government termination that go beyond the mere
coding of information collected from various data archives and reference

books. We shall therefore look at what is known about the termination of Danish governments and government coalitions since 1945 and, if possible, to make theoretical sense out of this knowledge in view of what has been suggested in the relevant literature.

Termination of Danish Governments

There are three relevant types of government to study, viz. majority coalitions, minority coalitions and single-party minority governments. Table 3 lists the Danish post-war cabinets as defined by the party composition criterion.

Majority Coalitions

One of the four majority coalitions in Table 3 is atypical in most respects. A National liberalization government was formed in May 1945 according to an agreement among the political parties and the resistance movement to ensure cooperation during a transitional period until new elections could be held after the occupation of the country by Nazi Germany (Kaarsted 1977, 265-313; Thomas 1982, 119-120). The National coalition only lasted for six months. With the return to normal parliamentary conditions party political disagreements again flourished. In Riker's terms it was a coalition of the whole that became almost useless to its members (Damgaard 1969, 36). The remaining majority coalitions terminated so far are:

• 1957-60: Social Democrats, Radical Liberals, Justice Party
• 1960-64: Social Democrats, Radical Liberals
• 1968-71: Radical Liberals, Liberals, Conservatives.

The three coalitions were all of minimal winning size with respect to the number of participating parties, and they all included the Radical Liberals. Their stories (Kaarsted 1964,1969,1992; Thomas 1982) can be summarized in the following observations on government termination, durability and succession.

First, all three governments terminated after having lost their majority, although by a narrow margin, in the first upcoming election.⁴ Thus, no Danish majority coalition in the post-war period has survived an election.⁵ But it is equally worth noting that although the internal life of coalition governments always involves inter-party strains and difficulties, none of the three majority governments terminated because of inter-party disagreement. In fact, no Danish majority coalition government has ever broken down because of internal disagreement.

Second, all three governments were pretty durable in the sense that they ruled for almost a full election period. The 1957-60 and the 1968-71 coalition leaders decided to call elections six months and four months before the end of the constitutional four-year term, respectively, hoping (in vain as it turned out) that the timing would be favourable to their parties. The 1960-64 coalition actually ruled in a full election period.⁶ The average lifetime of the three majority coalitions was three years and eight months.⁷

Third, there is no uniform pattern as to the government replacing a losing majority coalition. The national coalition of 1945, the 1960-64 coalition and the 1968-71 coalition had single-party minority governments as successors, Liberal in 1945, Social Democratic in 1964 and 1971. However, in 1964 the Social Democrats and Radical Liberals were able to continue government cooperation in a majority position by including an

MP from Greenland as a cabinet minister, even though their former
coalition partner (The Justice Party) failed completely at the polis.8

To this should be added that the coalitions of 1957-60 and 1960-64 experienced a change of prime minister, in 1960 and 1962 respectively, due to ill health and without a formal resignation. But, and as argued above, there is no reason why this should count as termination and formation of two new governments.

In sum, and disregarding the special National coalition of 1945, the postwar history shows that the few Danish majority coalitions were very durable and internally cohesive. They did not break down because of disagreement among the parties involved, but they clearly tended to lose when they eventually faced the electorate. In 1960, however, the Social Democrats and the Radical Liberals were able to stay in office, and in 1964 the Social Democrats continued as a single-party government.

Minority Coalitions

The five minority coalitions of the post-war period (again defined by party
composition) are the following:

• 1950-53: Liberals, Conservatives
• 1978-79: Social Democrats, Liberals

• 1982-88: Conservatives, Liberals, Christian People's Party, Centre
Democrats

• 1988-90: Conservatives, Liberals, Radical Liberals
• 1990-93: Conservatives, Liberals

The experiences with minority coalition governments (Kaarsted 1964,1977;
Thomas 1982; Fonsmark 1992; Damgaard 1992) may also be summarized
in a number of observations on termination, durability and succession.

First, the causes of minority coalition termination are different from those of majority coalitions. Without going into details it can be said that party strategic considerations loom large in decisions terminating minority coalitions, and that the Radical Liberal Party has played a crucial and pivotal role in most cases. The Liberal-Conservative coalition of 1950-53 had to leave office after having successfully completed a process of constitutional amendments because a majority in the Folketing, including the Radical Liberals, no longer wanted it in power. In 1988 the Radical Liberals, who had supported the four-party coalition of 1982-88, wanted and obtained cabinet positions at the expense of the Centre Democrats and the Christian People's Party. The three-party coalition of 1988-90 ended because the Radical Liberals decided to leave the government after an electoral defeat. However, the normally pivotal Radical Liberals had nothing to do with the termination of the unusual Social Democratic-Liberal

coalition in 1979. That government broke down because of disagreements between the two parties on how to fight the very difficult economic problems (Damgaard 1989). The Conservative-Liberal coalition of 1990-93 terminated because a report of an investigation concluded that government ministers had not provided the Folketing with true and complete information in a matter concerning Tamil refugees with a legal right to family reunion. The prime minister resigned in January 1993 without calling elections in a pre-emptive move to avoid a vote of no-confidence in parliament. He hoped that the coalition could continue with another Conservative prime minister and that the Radical Liberals would support such a solution, but he was disappointed.

Second, the durability of minority coalitions varies quite a lot, from only 14 months to almost six years. The Poul Schliiter "four-leaf clover" government (1982-88) even outperformed the post-war majority coalitions in terms of longevity, although it had to pay a high price in terms of an unprecedented large number of legislative voting defeats (Damgaard & Svensson 1989; Damgaard 1992; Madsen 1992). The average duration of the minority coalitions was two years and 11 months.

Third, and as in the case of majority coalitions, there is no uniform pattern as to the type of government succeeding a minority coalition. The first two were replaced by Social Democratic minority cabinets, while the minority coalitions since 1982 have been succeeded by other "bourgeois" (or "non-socialist") minority coalitions and most recently by a majority coalition led by the new Social Democratic leader Poul Nyrup Rasmussen. Poul Schliiter, leader of the Conservative party, headed three different minority coalitions after 1982, but his four-, three- and two-party coalitions always included the Liberals. As far as it goes, this does show a clear pattern: the Conservatives and the Liberals managed to stay in office for more than ten years and in three different cabinets, and they were only able to do so because of Radical Liberal support, which was withdrawn in 1993. Instead, the Radical Liberals decided to take part in a new coalition.

The five terminated minority coalitions are defined on the party composition basis. If other criteria of government were also used, three additional minority coalitions could be counted. The first would be a Liberal-Conservative cabinet (April-September 1953) "formed" because of elections required in a process of constitutional amendments. The 1950-53 Liberal-Conservative coalition even proposed formally to resign in April 1953 for purely tactical reasons (Kaarsted 1977, 511). The two remaining additional minority coalitions would be "four-leaf clover governments" after the elections of January 1984 and September 1987. The governing coalition actually survived on both occasions and even formally resigned on 9 September 1987, only to be reappointed on 10 September 1987 (Bille 1988). Again, however, it does not make sense to inflate the number of governments with artificial cases.

The post-war history thus tells us that Danish minority coalitions vary considerably in terms of durability. Further, and with the exception of the unusual and shortlived alliance of Social Democrats and Liberals 1978-79, they have been internally cohesive. They have not terminated because of defeats in elections but for party strategic reasons, notably decisions made by the Radical Liberals on how that party could best serve its interests in varying circumstances.

Single-Party Governments

The most numerous category of Danish post-war governments is the singleparty (minority) cabinet (Kaarsted 1964, 1969, 1977, 1988, 1992; Thomas 1982; Damgaard 1992). It contains the following governments (still defined according to the party basis only):

• 1945-47: Liberals

• 1947-50: Social Democrats
• 1953-57: Social Democrats
• 1964-68: Social Democrats
• 1971-73: Social Democrats
• 1973-75: Liberals

• 1975-78: Social Democrats
• 1979-82: Social Democrats

The Social Democratic preponderance in single-party governance is obvious from the above list. The party, which has been the largest group in the Danish parliament since 1924, relied on support from various combinations of left and centre parties. But there were also two Liberal minority governments. A short account of the single-party cabinets includes at least the following points.

First, the causes of government termination are quite varied. The two Liberal governments were defeated by no-confidence votes in parliament. The first of these (1945-47) called elections upon the defeat and then had to realize that the election results did not create a basis for its continuation in office. The second (1973-75) had called elections shortly before the defeat, and although it had increased its number of seats considerably, it was no longer acceptable to a parliamentary majority. Two of the Social Democratic cabinets (1947-50, 1979-82) resigned voluntarily without calling elections when they were no longer able to obtain support for their desired policies. The destiny of the four remaining Social Democratic cabinets was different. The 1975-78 government resigned voluntarily but only to form a, numerically speaking, strong minority coalition government with the Liberals. The 1953-57 cabinet, based on support from the Radical

Liberals, resigned after a regular election when the two parties could no longer form a parliamentary majority. However, they were able to build a majority coalition with the Justice Party (1957-60). The 1964-68 government, which after the election of 1966 was based on support from the Socialist People's Party, called elections because of a split within the supporting party (Mader 1979) and lost its majority basis. The 1971-73 cabinet, also based on cooperation with the Socialist People's Party, called elections because of a tie in legislative voting reportedly caused by a defecting Social Democratic MP who ran out of petrol on his way to parliament (Nannestad 1989) and lost its majority basis.

Second, the durability of single-party governments varies from 14 months to almost four years. On the average, their life-time (two years and eight months) is about three months shorter than that of minority coalitions, which again is about nine months shorter than the duration of majority coalitions (cf. Table 4). Third, the governments replacing single-party cabinets tend to be very different with respect to party composition, which is not all that surprising given that the party composition criterion is selected to define a government. None the less, six of the eight cabinets were succeeded by governments formed by parties considered to belong to the "opposing camp", while the party of two cabinets (1953-57, 1975-78) managed to stay in power by building a new coalition (cf. Table 3 above).

As previously mentioned, the numbers of majority and minority coalition governments are artificially expanded if additional criteria of government are used. This also applies in the case of single-party governments. Instead of eight cabinets, one could count 13. One of the possible additional cabinets would be a new Social Democratic government in 1955 as a result of the death of the prime minister, and another in 1972 when J. O. Krag voluntarily decided to leave politics after the successful referendum in October on Danish entry into the European Community. A further four cabinets would result from the elections in 1950, 1966, 1977 and 1981, which the respective Social Democratic governments survived for at least some time. In 1981 the government even formally resigned after the election only to be reappointed immediately afterwards.

Single-party governments terminate for various reasons. They may be

defeated in parliament and/or not able to obtain support after an appeal to the electorate; they may resign voluntarily; they may expand the party composition of the cabinet. Just like minority coalitions they are always dependent upon outside support, which means that party strategic considerations ultimately determine their destiny. Again, it could be added, the usually pivotal Radical Liberals play a crucial role.

Conclusion and Discussion

This paper has argued, and illustrated with a survey of Danish postwar government experience, that it makes sense to define a coalition government on the basis of its party composition. The alternatives (PMchange, elections, resignation/re-formation) do not appear persuasive except perhaps for the involuntary change of prime minister in a cabinet of the same party(ies). The definition selected has obvious consequences for the number of governments to be counted, for government durability, for the causes of government termination, and thus for the conclusions arrived at in analyses of such phenomena.

Comparative studies of government termination, using a more inclusive definition of a new government, find elections to be the most frequent cause of termination. The same conclusion seems to hold for Danish majority coalitions. Although the relatively few post-1945 majority coalitions were internally cohesive and durable, they all lost the first upcoming election. With one exception (1978-79), the minority coalitions were also cohesive, but their durability and termination, in addition, depended upon relationships with other parties, exactly as in the several cases of single-party government.

Patricia Hogwood has proposed a possibly fruitful distinction between strategic and contextual factors in coalition termination. Strategic considerations refer to frustrations with respect to the utility anticipated at the outset by coalition partners: "A termination brought about exclusively by such factors would constitute a strategic termination, one subject to the purposive choice of one or more of the participants" (Hogwood 1993, 12). As an example, a party might discover that the costs of sustaining a coalition exceed the benefits of cooperation. Contextual considerations, on the contrary, are not subject to actor choice, but can nevertheless erode the viability of a coalition. If they do so, the result will be an involuntary coalition termination. This could happen, for example, if elections change the partisan distribution of seats. Strategic and contextual termination may be seen as pure types; in practice, termination may involve interaction between the two types of factors.

A cursory application of this reasoning seems to make sense in most of

the Danish cases examined above. Thus, it appears that majority coalitions formed in normal times terminate for purely contextual reasons (election results), while the special National coalition of 1945 ended for strategic reasons (it soon became almost useless to the participants). On the other hand, the destiny of minority coalitions seems to be determined by strategic considerations among governmental as well as non-governmental parties. Thus, the terminations of the 1978-79 and 1988-90 coalitions were decided solely by "frustrated" government parties, while the remaining minority coalition terminations involved actions by non-governmental parties.

Single-party minority governments often end because of various forms of strategic considerations. First, they may become "frustrated" and leave office voluntarily if they can no longer get their policies adopted to a satisfying extent. Second, they may be defeated in parliament by a combination of opposition parties and be unable to redress the defeat through an appeal to the electorate. Third, they may resign in order to form an agreed upon coalition government. However, minority governments may also end because of contextual factors, as when an informal support arrangement breaks down through elections changing the partisan distribution of seats in their disfavour.

In Table 5 these various main reasons for the termination of all cabinets within the different types of governments are summarized. The table of course simplifies the sometimes very complex events involved, but as a general overview it is perhaps useful.

A further understanding of Danish coalition termination requires studies of the goals and strategies of relevant parties. As this paper (it is hoped) has shown, it would be a good choice to start with the Radical Liberal Party. This party has participated in all majority coalitions, and has usually strongly influenced the formation and termination of minority governments, be they of the coalition or single-party type. With the exception of a few periods (1966-68 and 1971-73 when the Social Democrats could and did form a majority with the Socialist People's Party, and 1978-79 when they cooperated with the Liberals) the Radical Liberals have been active in the formation and termination of governments. The strong position of the Radical Liberal Party derives from its central political location and a usually beneficial distribution of seats across the political spectrum (cf. Damgaard 1969; Pedersen 1987). But the bargaining power of the party was drastically reduced in the three periods just mentioned. In the first two, a centreleft majority could be established without the Radical Liberals. In the negotiations between the Social Democrats and Liberals on government formation in 1978 the Radical Liberals were excluded because the Liberals rejected a Social Democratic proposal to include the Radical Liberals in the prospective coalition by demanding the inclusion of the Conservative Party as well, which was unacceptable to the Social Democrats. But since 1979 the Radical Liberal Party has again played a crucial role.

Party strategy is of paramount importance in multi-party systems of government (Sjöblom 1968), and the strategies of some parties are usually more important than those of others. But, in as far as Danish majority coalitions are concerned, voter choice has been decisive in terminating the lives of governments. In the cases of minority coalitions and single-party governments, the causes of termination are basically strategic, even if support relationships may break down for other reasons.

---

NOTES

* The paper was originally presented at the workshop on "Termination of Coalitions: Theories and Cases", ECPR Joint Sessions of Workshops, Leiden, April 1993.

1. However, Finnish scholars (Törnudd 1969; Jansson 1992) deviate somewhat from the general consensus among specialists within the area of study. In their studies of cabinet coalitions and parliamentary government in Finland after the introduction of parliamentarism in 1917 they seem to follow a special Finnish convention. This convention maintains that even considerable changes of governments (of which there used to be many in Finland), including the withdrawal of a party from the cabinet which might even change the government from majority to minority status, do not constitute a change of government if the prime minister remains in office. The experience of stable cabinet coalitions in recent decades (Anckar 1992; Jansson 1992) suggests that such a special Finnish definition of government may no longer serve a useful purpose.

2. To mention a further variation of the theme, Katz (1993) counts "major reshuffles" as indicating a change of government.

---

3. Woldendorp et al. concluded similarly for 20 countries in the period 1945-90.

4. The 1957-60 coalition lost the election and its majority primarily because one of the three government parties (The Justice Party) did not manage to obtain any seats at all in the 1960 election.

5. Whereas the Social Democratic and Radical Liberal majority coalition 1929-40 survived the elections of 1932, 1935 and 1939.

6. The previous election was held on 15 November 1960 and the election called on 22 September 1964. However, the election law at the time required that an election be held before the month of October if the previous Folketing had been elected during the period of October-December (cf. Rasmussen 1972, 283-285).

7. According to the Constitution, the four-party majority coalition which came to power in January 1993 when the previous government, formed after the election in December 1990, left office without calling an election, will have to face the electorate in December 1994 at the latest.

8. Previously, the two MPs from Greenland had not participated in normal Danish party politics (Kaarsted 1969, 28-31).

REFERENCES

Anckar, D. 1992. "Finland: Dualism and Consensual Rule", in Damgaard, E., ed., Parliamentary
Change in the Nordic Countries. Oslo: Scandinavian University Press, pp. 151—
190.

Bille, L. 1988. "Politisk kronik: 2. halvår 1987", Økonomi og Politik 61, 52-59.

Blondel, J. 1968. "Party Systems and Patterns of Government in Western Democracies",
Canadian Journal of Political Science 1, 180-203.

Browne, E., Frendreis, J. P. & Gleiber, D. W. 1986. "Dissolution of Governments in
Scandinavia: A Critical Events Perspective", Scandinavian Political Studies 9, 93-110.

Budge, I. & Keman, H. 1990. Parties and Democracy. Coalition Formation and Government
Functioning in Twenty States. Oxford: Oxford University Press.

Damgaard, É. 1969. "The Parliamentary Basis of Danish Governments: The Patterns of
Coalition Formation", Scandinavian Political Studies, Yearbook 4, 30-57.

Damgaard, E. 1974. "Stability and Change of the Danish Party System over Half a Century",
Scandinavian Political Studies, Yearbook 9, 103-125.

Damgaard, E. 1989. "Crisis Politics in Denmark 1974-1987", in Damgaard, E., Gerlich, P.
& Richardson, J. J., eds., The Politics of Economic Crisis. Aldershot: Avebury, pp. 70-88.

Damgaard, E. 1992. "Denmark: Experiments in Parliamentary Government", in Damgaard,
E., ed.. Parliamentary Change in the Nordic Countries. Oslo: Scandinavian University
Press, pp. 19-49.

Damgaard, E. & Svensson, P. 1989. "Who Governs? Parties and Policies in Denmark",
European Journal of Political Research 17, 731-745.

Dodd, L. C. 1976. Coalitions in Parliamentary Government. Princeton, N.J.: Princeton
University Press.

Folketinget efter valget 1990. 1991. Copenhagen: J. H. Schultz.
Folketingstidende. Various years. Copenhagen: J. H. Schultz.

Folketingsårbog. Various years. Copenhagen: J. H. Schultz.

Fonsmark, H. 1992. Schluters Danmark. Historien om en politisk karriere. Copenhagen:
Børsen Bøger.

Hogwood, P. 1993. "Strategic and Contextual Factors in Coalition Termination", Paper,
ECPR Workshop on Termination of Coalitions, Leiden.

Jansson, J. M. 1992. Från splittring till samverkan. Parlamentarismen i Finland. Helsinki:
Söderström.

Kaarsted, T. 1964. Regeringskrisen 1957. En studie i regeringsdannelsens proces. Aarhus:
Universitetsforlaget.

Kaarsted, T. 1969. Dansk politik i 1960'erne. Taktik og strategi. Copenhagen: Gyldendal.

Kaarsted, T. 1977. De danske ministerier 1929-1953. Copenhagen: Pensionsforsikringsanstalten.

Kaarsted, T. 1988. Regeringen, vi aldrig fik. Regeringsdannelsen 1975 og dens baggrund.
Odense: Odense Universitetsforlag.

Kaarsted, T. 1992. De danske ministerier 1953-1972. Copenhagen: PFA Pension.

Katz, R. S. 1993. "Electoral Systems, Party Governments and Democracy", in Bryder, T.,
ed., Party Systems, Party Behaviour and Democracy. Copenhagen: Copenhagen Political
Studies Press.

Laver, M. & Schofield, N. 1991. Multiparty Government. The Politics of Coalition in Europe.
Oxford: Oxford University Press.

Lijphart, A. 1971. "Comparative Politics and the Comparative Method", The American
Political Science Review LXV, 682-693.

Lijphart, A. 1984. Democracies. Patterns of Majoritarian and Consensus Government in
Twenty-One Countries. New Haven and London: Yale University Press.

Mader, E. 1979. SF under det røde kabinet. Odense: Odense Universitetsforlag.

Madsen, J. M. 1992. Firkløverregeringen og det parlamentariske princip. Aarhus: Institute
of Political Science. Unpublished thesis.

Nannestad, P. 1989. Reactive Voting in Danish General Elections 1971-1979. Aarhus: Aarhus
University Press.

Pedersen, M. N. 1987. "The Danish 'Working Multiparty System': Breakdown or
Adaptation?", in Daalder, H., ed., Party Systems in Denmark, Austria, Switzerland, The
Netherlands, and Belgium. London: Frances Pinter, pp. 1-60.

Powell, G. B. 1982. Contemporary Democracies. Participation, Stability, and Violence. Cambridge,
Mass.: Harvard University Press.

Rasmussen, E. 1972. Komparativ Politik 11. Copenhagen: Gyldendal.

Riker, W. H. 1962. The Theory of Political Coalitions. New Haven and London: Yale
University Press.

Sjöblom, G. 1968. Party Strategies in a Multi-Party System. Lund: Studentlitteratur.

Strøm, K. 1984. "Minority Governments in Parliamentary Democracies. The Rationality of
Nonwinning Cabinet Solutions", Comparative Political Studies 17, 199-227.

Strøm, K. 1986. "Deferred Gratification and Minority Governments in Scandinavia", Legislative
Studies Quarterly, XI, 583-605.

Taylor, M. & Herman, V. M. 1971. "Party Systems and Government Stability", The American
Political Science Review LXV, 28-37.

Taylor, M. & Laver, M. 1973. "Government Coalitions in Western Europe", European
Journal of Political Research 1, 205-248.

Thomas, A. H. 1982. "Denmark: Coalitions and Minority Governments", in Browne, E.
C. & Dreijmanis, J., eds., Government Coalitions in Western Democracies. New York:
Longman, pp. 109-141.

Törnudd, K. 1969. "Composition of Cabinets in Finland 1917-1968", Scandinavian Political
Studies, Yearbook 4, 58-70.

von Beyme, K. 1982. Parteien in westlichen Demokratien. Miinchen: R. Piper & Co. Verlag.

Warwick, P. 1979. "The Durability of Coalition Governments in Parliamentary Democracies",
Comparative Political Studies 11, 465-498.

Woldendorp, J., Keman, H. & Budge, I. 1993. "Special Issue: Political Data 1945-1990.
Party Government in 20 Democracies", European Journal of Political Research 24, 1.