When a frustrated Premier Borten left office after the 1971 cabinet crisis
in Norway, he stated:

You can carry a heavy burden if it is well packed and if it is yours - you have a single party behind you. I do not know if it sounds pretentious, but perhaps my task best could be compared to that of carrying a number of fence poles. Many who have tried that know that difficulties may arise on the way, if the poles start going in different directions.²

Borten was referring to his experience as the leader of a coalition marked by great internal disagreement, and his statement is frequently quoted at times when problems arise within Norwegian coalitions. A striking feature of the problems that have led to the demise of Norwegian coalitions is that they appear to be caused by events of a non-economic nature (Browne, Frendreis & Gleiber 1986a, 102). This is puzzling, since coalition-building takes place along the predominant socio-economic left-right dimension, and, clearly, economic issues are the most important in this regard. Lægreid & Olsen (1986, 187) for instance point out: "Most cabinet crises in Norway in this century have not been struggles over large substantive stakes. They have rather been linked to issues that in a specific situation were redefined and given great political significance." This is compatible with the "critical events perspective" applied by Browne et al. (1986a) in their study of government stability in Scandinavia. They highlighted some important

Resumé

This article focuses on coalition termination in Norway and explores the questions: Which issues are severe enough to bring down a coalition government? To what extent are coalition crises caused by a complex bargaining environment? On the basis of a unified model for government solution, three cases are examined: the termination of Borten II in 1971, Willoch 111 in 1986, and Syse in 1990. The analysis demonstrates that the complexity of the bargaining environment is conditioned by the dimensionality of the party system. The stability of coalitions is restricted by the cleavage structure and the ideological diversity of the system, as parties are polarized along several conflict dimensions. Terminal issues are fundamentally related to the parties' position in the policy space. In order to preserve party identity and unity, political parties change from a cooperative to a competitive strategy when issues belonging to the "heartland" of the parties concerned become salient.

trends like the striking absence of economic issues from terminal events in Norway. The intriguing questions are thus: What is the nature of terminal events in Norway? Which issues are severe enough to cause the death of a coalition?

This article explores these questions by examining the stability (duration) and events surrounding coalition cabinets in the Norwegian political system. A coalition government is defined as a government consisting of two or more parties. The analysis consists of five parts, the first of which provides a theoretical overview of coalition termination. Note that some of the literature referred to is not limited to coalition governments, but also includes governments in general. The next section describes Norwegian coalition formations and relates cabinet stability to changes in the party system. In order to identify the types of conditions which lead to the actual termination of a coalition, the third part provides a detailed analysis of three cases. In section four these case studies are discussed in the light of the ideological polarization between Norwegian parties. In doing so, this part of the analysis incorporates data from two different levels of the party: the elite and the voter level. The conclusion relates these empirical findings to the initial theoretical framework.

Coalition Stability Revisited

The study of government stability (duration) can be divided into two main approaches. One approach relates the duration of governments to a set of attributes with which the cabinets are associated (Warwick 1979; Strøm 1985). The other main approach accounts for government duration in terms of the actual process that leads to the death of a government coalition; in other words the events behind cabinet dissolutions (Browne et al. 1986a, b, 1988). The first set of models are causal ones; they seek to explain cabinet duration as a result of systemic and governmental attributes. The events approach, on the other hand, models cabinet duration as a product of randomness. Acknowledging the explanatory value of both static and stochastic elements, some recent publications have tried to synthesize the two seemingly incompatible approaches to government dissolution (King et al. 1990; Laver & Schofield 1990; Warwick 1992; Gleiber, Frendreis & Granger 1992). Set out below is a short explanation of these two main approaches, followed by a discussion of the most recent studies that attempt to synthesize the two.

Within the attributes approach, there are three major types. The "regime attributes" approach focuses on the relationship between cabinet stability and a number of features of the system in general. The most important systemic features are the fragmentation of the party system and the polarizationof

izationofthe opposition. The greater the partisan fragmentation and the more polarized the opposition, the more short-lived the cabinet. In addition, the "coalition attributes" approach takes into account properties of the coalition itself, rather than attributes of the regime as such. Majority status is considered to be the most important coalition attribute, and empirical evidence clearly shows that majority governments tend to be more stable than minority governments (see, e.g., Strøm 1990a; Laver & Schofield 1990). The third approach examines the level of policy disagreementbetween or the bargaining environment within which coalitions must exist. The most important explanatory variables in the "bargaining approach" as suggested by Dodd (1976) are the degree of "cleavage conflict" or ideological diversity in the system, together with the level of fragmentation in the party system. The works of Dodd represent the first attempt at examining the relationship between cabinet stability and general features of the bargaining system, and his work is acknowledged by Laver and Schofield, as they attempt to explore the impact of the environment on cabinet stability (Laver & Schofield 1990, 156-158). In doing so, they make some qualitative distinctions between different bargainingsystems: systems, unipolar systems and multipolar systems. The key distinction that emerges from this classification contrasts multipolar systems on the one hand with bipolar and unipolar (centre) systems on the other. In bargaining systems that are not multipolar, election results might change the allocation of seats, but not necessarily the balance of power. Changes in the policy positions of the parties have the same limited effect. In multipolar systems, on the other hand, small changes in the pattern of seats or policies, of the type that occur in any election, are likely to reallocate any bargaining power.

A major common feature of the theories outlined is their "deterministic" nature; the information used to predict how long a government will last is known when the government is formed. They are attributes of the "regime", of the "coalition" or of the "bargaining" system, and nothing that happens after a cabinet takes office is used to explain its duration. Yet, when examining actual cases, we also find unexpected events that bring governments Thus, as an alternative approach to the "attribute theories", Browne et al. (1986a, b) have proposed a model of cabinet stability that rests fundamentally upon the assumption that the actual fall of governments is caused by random events. The vital point for Browne et al. is that the timing of these events is unrelated to the durability of a given government.³ The timing of its dissolution is largely determined by the occurrence of a critical event and not by its inherent durability. The possibility of such events occurring is, in other words, constant throughout the lifetime of a government. On this point, however, the arguments put forth by Browne et al. raise an important question: what accounts for a given government's inherent durability?

To resolve this problem, a combined approach has been elaborated, which links stochastic and static components in a single, unified model of government solution. This argument was first developed by King, Alt, Burns & Laver (1990) in an article that considers both the occurrence of systemic and governmental properties, such as ideological diversity and majority status, and at the same time allows for developments occurring during a government's lifetime, such as the occurrence of random critical events, to influence government duration. The analysis of King et al. suggests that the complexity of the bargaining environment, mainly conditioned the fragmentation and polarization of the party system, has an important impact on government stability. Thus, the more complex bargaining environment of the larger party systems are more sensitive to random critical events. In less fractionalized and less polarized party systems, on the other hånd, larger - and less frequent - random shocks may be needed to produce the same disturbing effects (King et al. 1990, 869). More recently, the combined approach has been defined by Warwick (1992) and Gleiber, Frendreis & Granger (1992), both studies using economic when discussing the ideological diversity of governments (see also Laver & Schofield 1990).

This point leads back to the initial part of this article, where termination of coalitions in the Norwegian system was considered. Even though the electoral history and the cooperative patterns of Norwegian parties clearly indicate that economic conflicts related to the left-right cleavage play an important role, the absence of such issues acting as terminal events in Norway is striking (see Browne et al. 1986a, 102). If issues related to this dimension are the most important in terms of coalition-building, why are they so rarely connected to cabinet crisis?

The answer to these questions may be found in the "nature" of party competition, which is conditioned by the cleavage structure of the party system. The Norwegian system has been politically defined around six dimensions of cleavage, determined by economic, geographical and historicalcircumstances & Valen 1964; Rokkan 1970; Valen & Rokkan 1974). The class cleavage and sectoral cleavage were determined by economic conflicts in the labour market and the commodity market respectively. A territorial cleavage between centre and periphery was partly overlapped by three cultural cleavages: a socio-cultural conflict between two different versions of the Norwegian language, a moral conflict articulated by the teetotalist movement, and a religious conflict over control of the Lutheran State Church. The last two cleavages are highly correlated and must thus be characterized as a moral-religious cleavage.⁴ While most voter studies acknowledge that policy differences in the Norwegian system can be best understood in terms of a multidimensional conflict structure (Rokkan1970; & Rokkan 1974; Valen 1981; Aardal & Valen 1989),

it has been less common to extend leadership studies beyond that of unidimensional models (see, however, Strøm & Leipart 1989; Valen 1990a, b). In this article I assert that the bargaining environment is more complex than is predicted by the unidimensional models. Polarization between Norwegian parties, not only at the electoral level, but also at the parliamentarylevel, be understood in terms of a multidimensional conflict structure. In this respect, the recent models of coalition termination may contribute a great deal to the understanding of the Norwegian system, because they allow seemingly random, critical events to be related to attributes of the party system, the most important attribute being ideologicalpolarization.

Norwegian Coalitions

Set out below is the general pattern of cabinet formation in Norway. Table 1 provides an overview of Norwegian governments, 1945-94.⁵ The table demonstrates the predominance of "bloc-thinking" in Norwegian politics: for a number of elections, dating back to 1965, the Norwegian electorate has been confronted with two clear-cut government alternatives, a single Labour Party government based on a socialist majority or a coalition consisting of bourgeois parties (the Centre Party, the Christian People's Party, the Conservatives and in earlier periods the Liberals).⁶These alternatives, fit the division of the left-right axis, have remained the same; despite the fragmentation of the system after 1973, the old parties have made no attempt to include any of the new parties in a formal coalition.⁷ The Labour Party has consequently avoided a formal coalition with anyone, either on the left (the Socialist Left Party) or the right. As a consequence, no efforts have been made to form an alliance between the socialist and non-socialist blocs. In addition, the traditional coalition partners on the bourgeois side have refused a formal alliance with the party on the far right (the Progress Party). Yet, in numerous situations the two "wing" parties have formed an important part of the parliamentary basis for the governing parties. Thus, in spite of the dynamic character of party competition over the last two or three decades, Norwegian coalition alternatives have been rather static in their outlook.⁸ In total, there have been seven coalition governments in the post-war period.

Another interesting aspect of coalition formation in Norway is related to government status. As the table indicates, all governments between 1945 and 1961 were majority Labour governments. In the first half of the 19605, minority solutions prevailed, while the latter part of this decade was dominated by majority governments. Since 1971 all governments but one (Willoch 11, from June 1983 to September 1985) have been minority solutions, with a share of seats ranging from 26 to 49 percent.⁹

Given the frequency of minority governments (single party or coalition), one might find the Norwegian system to be an unstable one. This is the conventional view of minority governments that Strøm (1990a) sets out to test in a cross-national study of 15 different nations.¹⁰ Contrary to the traditional view, Strøm's analysis gives a surprisingly favourable impression of such governments. Even though they are somewhat less durable than majority coalitions, they enjoy substantial advantages in regard to other dimensions, for instance in terms of electoral success (see Strøm 1990a, 128-29). According to Strøm, Norwegian minority governments in particularshow much more enduring pattern than conventional theory admits. Table 2 shows the average duration figures for Norwegian governmentssince From the table we can see that minority governments have been less durable than majority governments in the Norwegian system,

but post-war minority governments have enjoyed approximately 17 percent greater longevity than the cross-national mean for all governments in the Strøm analysis, which is 17.5 months. Consequently, regardless of the status of the governments, Strøm finds that the duration of Norwegian governments in the post-war period has been 40 percent higher on average than for all 15 countries analysed (Strøm 1990a, 227-228). If we compare Norwegian single-party governments with coalitions, Table 2 indicates that the latter have been considerably less stable. Minority coalitions, in particular, reveal a bad record in terms of stability with an average duration of only 6.7 months.¹¹ How can we explain these tendencies? To what extent can these results be related to the ideological polarization of the party system? This question is analysed by examining the "record" of coalition termination in Norway. Since all government coalitions have been nonsocialist,the will be limited to these. This study examines three cases: the termination of Borten II in 1971, Willoch 111 in 1986, and Syse in 1990.ⁿ

Coalition Termination: Three Cases

The story of Norwegian coalitions, which is the story of non-socialist coalitions, is a rather traumatic one. It started more or less as an "accident", or was at least triggered by one, when the Labour government was forced out of power in 1963 as a result of an accident in a government-operated coal mine in Svalbard.¹³ The first non-socialist government in 1963 did not last for long. After a month in office, the new coalition government was brought down by the socialist majority in parliament, and a new Labour administration took office. However, the symbolic importance of this shortlivedcoalition not be underestimated. Labour Party "hegemony" had then lasted for 28 years. It finally subsided when a credible non-socialist

alternative appealed to the voters.¹⁴ In 1965, the four non-socialist parties joined in a pre-electoral coalition, and the electoral result gave them a majority in the new Storting (parliament). Thus, the first ordinary coalition, Borten I, was formed in 1965, and lasted for four years. It was reformed after the election of 1969. Set out below is a discussion of the events which led to the termination of the Borten II coalition, followed by an examination of the Willoch 111 and Syse coalitions.

Borten II (1969-71)

With a total of 76 seats after the 1969 election, the four non-socialist parties held a majority in the new Storting. The Labour Party had 74 seats. The single strongest non-socialist party was the Conservative Party with 29 seats. However, the parties of the centre, the Liberals, the Christian People's Party and the Centre Party collectively had almost twice as many seats, and together they had a dominant position in the coalition. Thus, the largest of them, in this case the Centre Party, was rewarded with the appointment of a prime minister from its own ranks. Even if cooperation in the previous four years had gone quite smoothly, internal tensions started to emerge during the spring of 1970. The European Common Market (EC), after having been closed to new members since the early 19605, finally opened up for expansion after de Gaulle's resignation as President of France. In April 1970, the Storting decided to apply for EC membership. However, shortly after this decision, disagreement concerning the question of membership was evident in the coalition, with the Centre Party and the Conservatives as the main antagonists. The Conservatives were unanimously favour of membership, the Centre Party was opposed to it, while the Liberal and the Christian People's parties were split on the issue.

The EC issue created a number of conflicts within the non-socialist camp, and as the time for a decision on the matter came closer, the tensions increased. What finally triggered off the cabinet crisis was a trivial indiscretion the part of premier Borten in regard to a confidential state paper concerning the projected EC membership. Premier Borten handed in the cabinet's resignation in the early spring of 1971, and on 17 March a new Labour administration took office.

The traumatic experience due to the EC issue left considerable bitterness to be resolved between the non-socialist parties. Not until 1976 did the former coalition partners (this time without the Liberals) declare that they were ready to assume government responsibility if they obtained a majority in the 1977 election. The voters, however, did not give them the mandate to do so, and another four years were to pass before a non-socialist government (Willoch I) was formed after the election of 1981. This time it was a single-party Conservative government, based on the support of the

parties of the centre. Because of the Conservatives' stand on the abortion issue, the Christian People's Party declined to enter into a formal coalition with the party, but changed its mind in 1983, when it joined the government together with the Centre Party. The new coalition (Willoch II) was a majority coalition, and it was reformed after the election of 1985 (Willoch III).

Willoch III (1985-86)

The general election of 1985 resulted in an impasse for government alternatives the Norwegian system. Together, Labour and the Socialist Left obtained 77 seats in the new Storting, whereas the three coalition parties, the Conservatives, the Centre Party and the Christian People's Party held 78 seats. Thus, neither of the traditional groupings obtained a majority status, and with its two representatives, the Progress Party held the balance between the two blocs. Being a part of the non-socialist bloc, the party clearly preferred a non-socialist coalition to a single-party Labour administration. Willoch 111 was formed as a minority government dependent on the support of the tiny right-wing Progress Party. The possibility of including the Progress Party in the coalition, however, was never considered. The two parties of the centre, in particular, were reluctant to entertain such cooperation, and they refused even to have informal talks with the Progress Party to clarify the government's parliamentary basis.

Nevertheless, by holding the balance of two seats the Progress Party was in a pivotal position, and thus had the capacity to demand policy payoffs from the coalition. The coalition was severely constrained by this situation, but it was forced to enter into negotiations with the Progress Party on its budget proposals in the Storting. The climate of negotiation was very chilly indeed, and, at first, the coalition government was unable to obtain a majority. Only after harsh bargaining with the Progress Party did the budget proposal go through. However, during the spring session of 1986, it was evident that the country was facing severe economic problems. A drastic decline in government income due to the fall in oil prices forced the coalition government to propose a new budget. One of the initiatives was a proposal to increase taxes on petrol, which the Progress Party refused to back. The government turned to the socialist parties for support, but they in turn rejected the budget. In the spring of 1986 the Progress Party voted with the socialist opposition against increased petrol taxes, and Willoch 111 resigned after having been outvoted in parliament.

Labour again took over (Brundtland II). In the intermediate period, however, attempts were made by the previous coalition partners to topple the Labour government. When they finally found a suitable issue, they again ran into trouble with the Progress Party. In opposition to the government's

proposal for agricultural subsidies they suggested an increase in the budget on this point. The Progress Party, however, was unable to support a policy of increased funds for agriculture, and the Brundtland administration was able to continue in office for the rest of the electoral period.¹⁵ Thus, this administration lasted till 1989 based on support partly from the Socialist Left, partly from one of the non-socialist parties. Brundtland II was replaced by a new non-socialist coalition (Syse) after the election of 1989, consisting of the same parties as the previous Willoch 111.

Syse (1989-90)

Rommetvedt (1991, 315) has interpreted the Syse coalition as a result of "limited rationality" on behalf of the parties involved. The non-socialist parties obtained a slight majority in the new Storting, but the outcome of the election had drastically changed the distribution of party strength. It resulted in a strong increase in the support for the two wing parties on the left-right dimension, the Progress Party and the Socialist Left Party, at the expense of the two dominant parties competing along this dimension, Labour and the Conservatives. The parties of the centre were fairly unaffected the electoral results.

As in 1985, the three coalition parties were faced with the possibility of inviting the Progress Party to join them in a formal coalition. The party was still in a pivotal position, and it enjoyed much stronger public support than in 1985. This possibility, however, was never seriously considered, and the coalition that was actually formed had a parliamentary basis of 38 percent. Thus, not only were the coalition parties dependent on the support of the Progress Party, which indeed had proved to be rather unreliable, but long before the election the polls had also indicated that popular support had decreased sharply for the non-socialist alternative (Valen 1990c, 280). Again, a growing debate on Norway's future relationship with the EC created tensions within the bloc. It was therefore surprising that the parties entered into a new alliance.

Nevertheless, even if the prospects of the new government seemed rather poor, the Conservative, Christian People's and Centre parties formed a united front behind the 22-point programme they had agreed on prior to the election. Interestingly, the wording of the actual agreement provided for a dissolution of the coalition if severe disagreements were to arise on the EC issue (Rommetvedt 1991, 39; Thomassen 1991). The plan, therefore,was keep this issue on ice for as long as possible. It was, however, evident during the spring session of 1990 that disputes on the membership issue created great internal tensions. And after little more than a year in office, the Conservative premier Syse announced that the cabinet would resign because of disagreement on the EC issue. Only hours later, the

parliamentary leader of the Centre Party proclaimed that the party would
welcome a new Labour government.

In November 1990 Brundtland 111 took office, to be replaced by Brundtland after the general election of 1993. Since then, Labour has sought parliamentary backing partly from the left, partly from the centre of the system. This situation has not improved the climate between the traditional coalition parties. It leaves the Conservatives with the dilemma of where to seek governmental backing. Should the party lean on the support of the Progress Party in future government negotiations, or struggle to rebuild a centre-right coalition? The answer may be presented to us before the general election of 1997. For the present, however, let us concentrate on some important points related to the case studies described here.

Party Competition and Party Ideology

Both differences and similarities may be observed between our three cases. If we compare government parties and opposition parties as the main source of termination, this overview shows that two coalitions, Borten II and Syse, were terminated from within. Willoch 111, however, was brought down from the outside, as a result of being outvoted on a motion in parliament. Further, the parliamentary basis of the three cases differed: one was a majority coalition, whereas two of them were minority solutions. However, there is no systematic relationship between government status and source of resignation as far as these three cases are concerned. We can easily picture a situation where termination of a coalition is due to its minority status since minority governments constantly run the risk of being outvoted by the opposition. In our two cases, however, one minority coalition was outvoted in parliament (Willoch III) and the other (Syse) was terminated from within.

The EC issue was the actual cause that brought two of the coalitions down: Borten II and Syse. Both were severely constrained by this issue, and even though the termination of Borten II externally had the appearance of a scandal, it was in reality the result of great internal division between the alliance partners on this particular issue (Björklund 1982; Sejersted 1984). The Willoch 111 coalition, on the other hand, was outvoted on the petrol tax issue. The important point, however, is why this happened on the basis of a non-socialist vote. Why did the Progress Party pursue such a competitive strategy, and why was the coalition unable to secure its parliamentary basis? By negotiating with the Progress Party before the election, the three coalition partners could have clarified their political base. Put differently: by refusing to have proper discussions with Progress Party leaders, the coalition parties deprived themselves of important information,which

mation,whichmight have been used to avoid a cabinet crisis. In addition, the Progress Party might possibly have pursued a less competitive strategy, had the parties of the centre been willing at least to have informal talks with the party. Thus, two puzzling questions arise: first, why does the EC issue stir up so many conflicts in the political system? And second, why is there such a strained relationship between the parties of the centre and the Progress Party? Answers to these questions are inherent in politics itself; in the saliency of some specific issues, and in the diversity which these issues create among the parties. Let us therefore have a look at how political leaders themselves perceive politics. Which issues are important to the parties? And, on which issues are they divided? These questions are examined by using data about party representatives in the Storting between 1985 and 1989. The data were collected and administered by members of the Norwegian Program of Electoral Studies.¹⁶ In the survey, each representative was asked to assess 18 statements about certain policy areas.¹⁷ Naturally, such data only give information about political priorities from a particular period of time (in this case between 1985 and 1989), but most of the statements presented to the representatives concern issues that have preoccupied voters as well as parties with different intensity over a number of years. We may therefore assume that they reflect a fairly stable policy orientation on behalf of the parties. Table 3 gives an overview of which questions the party representatives assessed as most important.¹⁸

The table reveals interesting variations between the parties. The majority of the Socialist Left Party regards the question of nuclear arms as being the most important, whereas a significant number of the Labour representatives value the question of social equality as most vital. However, 16 per cent of them also find the question of nuclear arms important. Generally, the Labour representatives are spread over a number of issues. The percentages are affected by the size of the parties, and the priorities of the largest parties are of course more scattered than those of the smallest parties. Thus, the Progress Party, with its two representatives, will sum up to 100 percent only on two issues. Nevertheless, their priorities are consistent with the basic ideological platform of the party. Halting the growth of public expenditure forms an important part of the economic liberalism advocated by the party, and was one of the prime reasons why the party emerged in the first place. The emphasis on strong military defence reflects a typical "rightist" position. The two parties of the centre, the Christian People's and Centre parties give priority to areas that are not connected to the socio-economic left-right dimension. Moral-religious issues are given high priority by Christian leaders; an impressive 93 percent of them mention this area as the most important, whereas a plurality of Centre Party representatives mention more economic support to peripheral districts as vital. Forty-five percent of the Conservatives report "strong military

defence" as the most important issue, but quite a lot of them also give priority to the question of halting the growth of public expenditure. The priorities given by party representatives are thus consistent with those of the Progress Party representatives, but are spread over a variety of issues, since the Conservatives are far more numerous.

The information we have so far, however, provides us with little or no knowledge about the degree of polarization between the parties. How far apart are they on the various policy issues? Let us have a closer look at their positions on the issues they evaluated as most important. For each of the statements, the representatives were asked to indicate to what extent he or she agreed or did not agree on a scale running from 1 to 5, with 1 indicating total agreement and 5 total disagreement with the statement. Table 4 shows party representatives' position on seven issue-statements.

There are several interesting points to be noted from Table 4. First, the set of statements connected with economic issues and defence issues shows a clear left-right distribution. Consequently, we find the Progress Party and the Conservative Party most in favour of strong military defence and least in favour of prohibiting nuclear arms. With quite opposite views, the

Socialist Left and the Labour Party constitute the other extreme on these issues. These four parties are also the main opponents on the issues that concern more equality between social groups and classes, growth in public expenditure and a removal of state regulations. On all these issues, the Progress Party is at one extreme and the Socialist Left at the other, with the Conservatives and Labour respectively as their closest neighbours. The two centre parties, the Christian People's Party and the Centre Party, are where they "should" be: in the centre.

The picture gets a little more complicated if we look at the two remaining questions. On the question of more economic support to peripheral districts, we find the Centre Party and the Progress Party as the main opponents, the Centre Party being most in favour of economic support and the Progress Party most opposed to it. Although the question of economic support to peripheral districts is economic in character, it is not related to the socioeconomic dimension. Instead, it reflects divisions along the old sectoral cleavage in the Norwegian system. It includes considerations such as subsidies agriculture and government support of rural areas. These are important issues to the agrarian Centre Party, which relies for its support on people employed in rural areas. Equally interestingly, the table shows that the Christian People's Party and the Socialist Left Party are the main opponents concerning the question of strengthening Christian morality and beliefs. To "strengthen Christian morality and beliefs", as a generally formulated statement, is connected with issues like abortion, temperance and other non-economic issues.

The priorities indicated by representatives of the individual party reflect the dimensionality of the Norwegian system quite well. Representatives from the Labour Party typically give priority to issues related to social equality and other traditional "leftist" issues. The negative valuation of nuclear arms by the party's neighbour to the left, is consistent with leftwingscepticism defence policy within NATO. The parties of the right

give priority to typical "rightist" issues, such as strong military defence, lowering public expenses and taxes. The parties of the centre give strong considerations to issues related to the commodity market and the moralreligiousdimension, are interests associated with the cleavages to which these parties owe their existence. Norwegian parties are thus polarizedalong conflict dimensions. An interesting question is: to what extent is this conflict pattern reflected on the mass level?

Let us use voters' evaluation of party position as a supplement to the leadership data. For this purpose data from the 1989 Norwegian national election studies will be applied. In addition to the traditional left-right scale, these data sets contain respondents' position and party position on several other issues. Voters are then asked to place each of the seven parties on a 1-10 point scale.¹⁹ In this article I use three sets of questions in addition to the left-right scale: agricultural support, regulation of alcohol, and privatization of health care. Again, in a strict empirical sense, the results are only valid for the actual year in question (1989). On the basis of previous findings, however, I assume that these issues reflect an underlying conflict potential between the parties (see, e.g., Valen 1981; Aardal & Valen 1989; Macdonald et al. 1991).

Figure 1 provides voters' perceptions of party position on the left-right scale. The question was worded as follows: "There is so much talk about the conflict between left and right in politics. Here is a scale that goes from 1 on the left, that is, those who are placed politically furthest to the left, to 10 on the right, that is, those who are politically furthest to the right. Where would you place the various parties on such a scale?" The figure

confirms that voters place ti parties from left to right consi₁ "ntly with the traditional government alternatives. The closest neighbours on this scale are the Conservatives and the Progress Party on the right-hand side, and the Socialist Left Party and Labour on the left-hand side. Further, voters place the parties of the centre fairly close to each other. Note, that on this scale both the Centre Party and the Liberals are located closer to Labour than to the Conservatives.

Let us examine an issue that is assumed to structure the vote along the left-right continuum, the issue of whether or not to allow private health services on a commercial basis. The question was worded as follows: "Value 1 expresses the wish for a much greater role for private health care on a commercial basis, while value 10 expresses a wish that health care should be a public responsibility only. Where would you place the various parties on such a scale?" Observe that, in order to be consistent with the direction of left-right preferences (as they appear in Fig. 1), the value numbers on this scale are reversed in relation to the wording of the question, so that "leftist" views get the lowest value numbers (close to 1), whereas "rightist" views get the highest value numbers (close to 10). On this particular issue, the parties are distributed in "groups" from left to right. In a middle position, very close to each other, we see the Liberal Party, the Christian People's Party and the Centre Party. The Labour Party and the Socialist Left Party are located very close to each other on the left hand side, whereas we find the Conservatives and the Progress Party at the other extreme. The perception of parties as "leftist", "rightist" and "centrist" is in accordance with previous findings from the Norwegian system (Valen 1981; Aardal & Valen 1989; Macdonald et al. 1991).

However, as indicated by the leadership data, issues related to conflicts in the commodity market and along the moral-religious dimension are given great priority by the parties of the centre. The question we now turn to is therefore how voters perceive distances between parties on issues that relate to these dimensions. Let us start with agricultural policies, a question typically related to the sectoral cleavage. The voters were asked the followingquestion: say that Norwegian agriculture should manage without government support and tariff protection against foreign competition.Let assume that those who hold that position are placed at 1 on this scale. Others hold the opinion that the present system of government support for agriculture should be maintained. Let us assume that those who hold this opinion are given the value 10 on the scale. Of course, there are some who are located between these extremes. Where would you place the various parties on such a scale?" Figure 3 shows how parties were located on this particular issue. Consistent with the results from the leadership data, polarization on this dimension does not follow the left-right continuum. The "extreme" parties are now the agrarian Centre Party on the one hand,

giving whole-hearted support to agricultural subsidies, and the Progress
Party on the other, opposing such subsidies. In a middle position, right in
the centre of this dimension, we find all other parties.

The last question we turn to concerns alcohol policy, one of the traditional moral-religious issues in Norwegian politics. The voters were given the following options: "Value 1 denotes the position that alcohol should be sold without restrictions and at greatly reduced prices. Value 10 means that the sale and production of alcohol should be more strongly regulated than it is today. Where would you place the various parties on such a scale?" Figure 4 shows how the parties were placed on this dimension. The voters perceive the Christian People's Party and the Progress Party as the main opponents on the alcohol issue; both parties are placed in extreme positions on the 10-point scale. The Conservatives are placed fairly close to the Progress Party, whereas all other parties have a "centrist" position on this issue. There is almost no difference between the two socialist parties, the Socialist Left and Labour Party on the one hand, and the Liberals and the Centre Party on the other.

The empirical results produce significant findings. First, concerning the Willoch 111 government, the evidence clearly explains the "allergic" relationship between the Christian People's Party and the Centre Party on the one hand, and the Progress Party on the other. Being forced to base their government on parliamentary support from a party that advocates such moves as abolition of the state monopoly on the sale of alcohol and removal of farm subsidies must have been pure horror to these two coalition

parties. Once it became a part of their parliamentary basis, the antagonistic relationship between the Progress Party and the parties of the centre was like an undetonated bomb under the coalition. Thus, an important message resulting from the analysis of Willoch 111, is that the termination of coalitions cannot be regarded as independent of their actual formation: the conflicts that contaminated the bargaining climate during the period of formation continued to haunt the coalition throughout its lifetime.

But, why did the Progress Party join the socialist opposition in defeating the coalition on the petrol tax issue? Again, we must consider the nature of this issue. One of the Progress Party's main objectives has always been to lower taxes and reduce government regulation; it has been the Progress Party's raison d'etre, we might say. The proposition to increase taxes on petroleum was fundamentally opposed to the party's basic values, and supporting such a motion would draw into question its credibility in the eyes of the voters. Thus, the collapse of Willoch 111 can best be understood as a result of a complex bargaining system, due to polarization of the parties along several conflict dimensions.

Second, the multidimensional conflict structure is critical for understanding conflicts that arose in the Borten II and Syse governments owing to the EC issue. Virtually all underlying conflict dimensions are aroused by the EC issue (Rokkan & Valen 1964; Valen 1973, 1976, 1981). Of most relevance to the non-socialist coalitions is that the Centre Party represents rural economic interests and the peripheral parts of the country, while the Conservatives advocate the interests of urban areas, particularly industrial production and export. For the Centre Party, it is inconceivable to be a part of a coalition that is in favour of EC membership, simply because membership threatens the very existence of farming in Norway. Entrance into the EC would probably mean reduction of efforts to protect Norwegian agriculture from the more profitable farming of community countries. Further, Norwegian fishermen feel that their fish resources are greatly endangered. Thus, the Centre Party fiercely opposes any form of formal attachment to the EC. By contrast, in the Conservatives' view, industry (and thus urban interests) will profit considerably from EC membership. addition, values advocated by the Christian People's Party are also threatened by the Community, since membership involves adopting a more liberal legislation for the sale of alcohol. The EC issue, therefore, revives conflicts that are rooted in the old cleavage structure (Converse & Valen 1971).

Finally, with regard to the Syse coalition, all elements of the previous discussion are relevant. Internal division existed because of the EC issue, and indeed this issue proved to be fatal for the coalition. However, because of its minority status, the Syse cabinet faced the same problem as the Willoch 111 government. Since the Progress Party was still in a pivotal

position, the coalition was dependent on the support of this party. Moreover, the Progress Party was in a much stronger position after the election of 1989 than it had been after the election of 1985. The party had every incentive, therefore, to display its strength in the coming budgetary debate. Given these circumstances, for how long could the Syse coalition expect to last? Like Willoch 111, the Syse coalition could easily have collapsed because of a negative vote on the budget proposal in parliament. In this respect, the EC issue just won the race as the critical event that actually brought down the government.

Conclusions

This article began by considering the nature of events for coalition termination. claim that economic issues do not play an important part in the events that bring governments down in Norway is only partly supported. The EC issue has been terminal for two of the coalition cases, whereas one coalition was brought down by the petrol tax issue. Clearly, the membership question concerns important economic matters that correlate with the commodity dimension in particular and left-right controversies in general. However, the saliency of the EC issue stirs up virtually all cleavages in the system: centre vs. periphery as well as cultural and religious disputes. Taking into account that different dimensions are important for different parties, the stability of Norwegian coalitions is conditioned by the saliency of multiple dimensions. Parties seek to politicize those dimensions that are most favourable to them and attempt to depoliticize those that are not. Thus, shifts in lines of salient cleavages produce a complex bargaining system. The "nature" of the events depends on the party's position along some issue dimension. If the party holds an extreme position on some issue, it is likely to change from a cooperative to competitive strategy when the issue becomes salient. The saliency of issues, therefore, seems to be vital for the actual timing of the termination. The important point to note is that terminal issues are fundamentally related to the identity of the parties. They belong to the parties' "heartland" (see Heath et al. 1987) - their natural domain on which they build their public support. To preserve party identity (and possibly party unity), political parties will take strong stands on such issues, even if this means breaking the alliance.

In relation to the termination models outlined in part two of this article, these results are of considerable theoretical interest. First, they demonstrate the complexity of the "centre" concept in the analysis of multiparty systems. The former leader of the Labour Party, Reiulf Steen, once said during a round table discussion (Oslo, 4 December 1990): "There are two extreme parties in the Norwegian system. Those are the parties at the centre". This

seemingly paradoxical remark becomes more meaningful in view of this analysis. The idea of a centre depends largely on the possibility of voters to locate parties on a left-right scale. Extending the model to a multidimensionalsetting, complicates the matter rather dramatically (see also Daalder 1984).²⁰ Both of the Norwegian centre parties have "extreme" positions on the moral-religious dimension and the sectoral dimension, along which they face another non-socialist party as their main opponent, the Progress Party. This is consistent with the findings of Macdonald, Listhaug & Rabinowitz (1991). Interestingly, the Conservativesare close to the latter party on these dimensions. Thus, the polarization of the parties along these issue dimensions constitutes a considerable conflict potential between the parties of the bourgeois bloc.

Second, the analysis demonstrates the destabilizing effect that seemingly small changes in the distribution of party strength may have on the system. Sartori (1976) once argued that, despite the number of parties represented in the Storting, the Norwegian party system should not be regarded as one of polarized pluralism. It is doubtful, however, whether this argument holds good in a situation where the system of alternation between governmental has entered a phase of unpredictability. Owing to the strategic position of a party that in no sense is treated as "koalitionfähig" by the other parties, the level of conflict in the 1980s increased remarkably in the Norwegian system.²¹ Thus, focusing on factors like ideological polarization and party system format clearly contributes to the understanding of coalition termination in multiparty systems.

While this analysis has focused on the context in which termination has occurred, perhaps it is worthwhile then to examine also the behavioural aspects of the participants. Parties are constantly faced with the necessity to make strategic choices when a crisis situation arises. In this respect, they are neither passive agents nor unconstrained actors. Strøm's (1990a, b) trade-off model from the coalition formation literature illustrates this point. In "A Behavioral Theory of Political Parties" Strøm discusses three models of competitive party behaviour: the vote-seeking party, the office-seeking party, and the policy-seeking party. On the assumption that political leaders are rational actors, Strøm argues that in relation to these three goals they will make cost-benefit calculations before entering into a coalition. Pursuing one objective, office (and thus policy), might be done at the expense of the other objective, future votes (for similar reasoning, see Sjöblom 1968).²²

We can assume that parties do similar calculations when a critical event occurs. As a part of a coalition, the party has a choice between two alternative strategies: it may continue to cooperate or it may defect from the coalition. In the case of a minority situation, parties outside the coalition have the same options: they may choose to bring down the government, or they may choose to cooperate with it. To keep it simple, let us just

concentrate on the dilemma facing coalition partners. For each party it is a question of cost and benefit: what do I lose if I break the alliance, and what do I gain? Alternatively, what is there to win by staying, and what is there to lose? On the one hand, policy benefits can be obtained by participationin On the other hand, by using a competitive strategy, the party clarifies its ideological position and may thereby gain votes. Interestingly, the present analysis indicates that considerations of this type may have played an important part in strategic choices made by the relevant actors.²³

Just as Strøm's model assists in explaining coalition formation, it may be useful also to explain coalition termination. Particularly in systems where the question of being in government or being in opposition is decided by a marginal number of votes, we must assume that parties act in a highly strategic manner. The question of which factors motivate party decisions is thus of interest.

---

NOTES

1. A first draft of this article was presented to the workshop on The Termination of Coalitions: Theories and Cases, ECPR Joint Sessions, University of Leiden, 2-8 April 1993. The discussion there has been useful for the revision of the article. In addition. I express my thanks to Henry Valen, Kathy Dunn-Tenpas and Peter Mair for critical comments on an earlier draft of the article.

2. The quotation is taken from Rommetvedt (1991, 31), and originates from a radio recording of Borten's speech as he left office in 1971. However, I am fully responsible for the translation from the Norwegian.

3. Note the distinction between durability (as a theoretical expectation) and actual duration (as an observed value). See Browne et al. (1984) and Strøm (1990a).

4. Membership in the teetotalist movement and in the religious movements have traditionally substantially intercorrelated, while there is a low correlation between language position and the other two counter-cultures (Valen 1981, 152).

5. Note that Norwegian parliamentary elections are held every fourth year. The constitution not allow for a dissolution of parliament between elections. However, cabinets terminate and reform thereby allowing the number of governments to exceed the number of elections.

6. The term "bourgeois" is here used as synonymous with '"non-socialist", as is the custom in Norwegian politics.

7. Ranging from left to right on the socio-economic class dimension, the "old" parties that developed around 1920 were: the Communist Party, Labour, the Liberals, the Agrarian Party (later the Centre Party) and the Conservative Party. The first deviation from this model occurred in 1933, when a fundamentalist religious faction broke away from the Liberals and founded the Christian People's Party. In addition, a left-wing faction broke off in 1961, and formed the Socialist People's Party, due to disagreement within the Labour Party on foreign policy issues (NATO). The "new" parties referred to in the text are the two wing parties that emerged after the EC referendum in 1972. On the left wing a new socialist party emerged, as a result of a union between the Socialist People's Party, the Communist Party and an anti-EC faction from the Labour Party (AIK). A new party emerged also on the right wing: Anders Lange's Party, named after its founder. It was a right wing populist party which called for strong reduction of taxes and public expenditure. It later changed its name to the Progress Party. For a thorough description of the development of the party system, see Valen (1981, 51-71).

---

8. On the electoral level, party competition is clearly reflected in the increasing number of floating voters. After the early 1970s about one out of three electors shifted position from one election to the next, as compared to the 1960s when the corresponding figure was about one out of four (Valen & Urwin 1985, 61). At the Storting election of 1993. individual volatility reached almost 44 percent (Ukens Statistikk, December 1993, pp. 4-5). Thus, since the beginningof the 19705, the Norwegian electoral scene hasbecome increasingly unpredictable.

9. The radically undersized government of Korvald (with a parliamentary base of 26 percent) from 1972 to 1973 is a rather exceptional case, since its main purpose was to reach a tråde agreement with the EC.

10. "Stability", or. more specifically, duration of a government in terms of months in office. is defined as changes of government each time there is an election, whenever there is a change of prime minister or change in the party composition of the cabinet.

11. Note that Table 3 only goes up to 1987, leaving out the Syse coalition of 1989-90. Thus, the number of minority coalitions only amounts to three. In addition, the figures are greatly influenced by the short-lived Lyng coalition of 1963. However, if we add the Syse coalition to the analysis, the average duration of minority coalitions is still only eight months.

12. Actually, these cases are all there is, apart from termination of coalitions due to electoral defeat. The only exceptions are the first short-lived Lyng coalition in 1963 and the Korvald coalition of 1972-73. Both these cabinets were formed and terminated under very special circumstances, and they are not given much attention in the analysis.

13. The government was accused of withholding information from the Storting about safety precautions in the mines.

14. For a thorough discussion of this issue, see Rokkan (1966).

15. See Strøm (1994) for a detailed analysis of this particular case in "The Presthus Debacle".

16. The study was directed jointly by Henry Valen and Gudmund Hernes.

17. After having been presented with 18 issue statements, the representatives were asked: "Which one of these issues do you consider as most important?"

18. For the period in question, the Liberal Party was not represented in the Storting. Consequently, it is not included in the table.

19. The scales were designed to test out the so-called "directional theory of party competition", have been analysed previously by Macdonald, Listhaug and Rabinowitz (1991).

20. See Daalder (1984): "In Search for the Centre of European Party Systems" for an excellent discussion about the relevance of the notion of centre parties for different party systems. One of Daalder's conclusions, based on analysis of the Dutch system, is that a perceived centre is contingent upon the view you take: it all depends on which issues the parties are related to. Thus, referring to an analysis of the 1972 Dutch election study, Daalder states: "The left-right dimension was sufficiently robust to allow the singling out of some parties as having a centre location. But on certain issues (abortion, defense and law-and-order issues), neither the scale nor the location held. Clearly, presumed centre parties are not centre parties on all issues" (p. 96).

21. Heidar (1989) and Madeley (1990) make similar observations.

22. In addition to the three goals mentioned, Gunnar Sjoblom discusses a fourth party objective: party cohesion. This is an interesting point to be considered, since the question of party unity clearly constrains the parties' efforts to achieve vote-maximization the parties' possibilities of obtaining maximum parliamentary influence. See Sjoblom (1968, 183-184).

23. Narud and Irwin have analysed Dutch cabinet crises within the framework of the trade-off model (see Narud & Irwin 1994).

REFERENCES

Aardal, B. & Valen, H. 1989. Velgere, Partier og Politisk Avstånd. Oslo: Statistisk Sentralbyrå.

Browne, E., Frendreis, J. & Gleiber, D. 1984. "An Events Approach to the Problem ot
Cabinet Stability". Comparative Political Studies 17, 167-197.

Browne, E., Frendreis, J. & Gleiber, D. 1986a. "Dissolution of Governments in Scandinavia:
A Critical Events Perspective", Scandinavian Political Studies 9, 93-109.

Browne, E., Frendreis, J. & Gleiber, D. 1986b. "The Process of Cabinet Dissolution: An
Exponential Model of Duration and Stability in Western Democracies", American Journal
of Political Science 30, 625-650.

Browne, E., Frendreis J. & Gleiber, D. 1988. "Contending Models of Cabinet Stability",
American Political Science Review 82, 930-941.

Converse, P. & Valen, H. 1971. "Dimensions of Cleavage and Perceived Party Distances in
Norwegian Voting", Scandinavian Political Studies 6, 105-152.

Daalder, H. 1971. "Cabinets and Party Systems in Ten Smaller European Democracies", Acta
Politica 6, 282-303.

Daalder, H. 1984. "In Search of the Centre of European Party Systems". The American
Political Science Review 78, 92-109.

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

Gleiber, D., Frendreis, J. & Granger. G. 1992. "Governmental Lifecycles in Post-War
Parliamentary Democracies: Continuity and Change". Paper, the Annual Meeting of the
Western Political Science Association, San Francisco.

Heath, A., Jewell, R. & Curtice, J. 1987. "Attitudes, Values and Identities in the British
Electorate", paper, the Annual Meeting of the American Political Science Association,
Chicago.

Heidar, K. 1989. "Norway: Levels of Party Competition and System Change", West European
Politics 12, 143-156.

King, G., Alt, J., Burns, N. & Laver, M. 1990. "A Unified Model of Cabinet Dissolution in
Parliamentary Democracies", American Journal of Political Science 34, 846-871.

Lægreid, P. & Olsen, J. P. 1986. "The Storting - a Last Stronghold of the Political Amateur,"
in Suleiman, E. N., ed., Parliaments and Parliamentarians in Democratic Politics. New
York: Holmes & Meyer.

Laver, M. & Schofield, N. 1990. Multiparty Government. The Politics of Coalition in Europe.
Oxford: Oxford University Press.

Lipset, S. M. & Rokkan, S., eds. 1967. Party Systems and Voter Alignments: Cross National
Perspectives. N.Y.: Free Press.

Macdonald, S. E., Listhaug, O. & Rabinowitz, G. 1991. "Issues and Party Support in
Multiparty Systems", American Political Science Review 85, 1108-1131.

Madeley, J. 1990. "Norway's 1989 Election. The Path to Polarised Pluralism?" West European
Politics 13, 287-292.

Narud, H. M & Irwin, G. 1994. "Must the Breaker Pay? Cabinet Crises and Electoral Tradeoffs".
Politica 29, 265-284.

Rokkan, S. 1966. "Norway: Numerical Democracy and Corporate Pluralism", in Dahl, R.,
ed., Political Opposition in Western Democracies. New Haven: Yale University Press.

Rokkan, S. 1970. Citizens, Elections, Parties. Bergen: Universitetsforlaget.

Rokkan, S. & Valen, H. 1964. '"Regional Contrasts in Norwegian Politics", in Allardt &
Littunen, eds., Cleavages, Ideologies and Party Systems. Helsinki: Westermarck Society.

Rommetvedt, H. 1991. Partiavstand og Partikoalisjoner. Ph.D. dissertation. University of
Bergen.

Sartori,G. 1976. Parties and Party Systems: A Framework for Analysis. Cambridge: Cambridge
University Press.

Sejersted, F. 1984. Høyres Historie 3: Opposisjon og Posisjon. Oslo: J.W. Cappelens Forlag
a/s.

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

Strøm, K. 1985. "Party Goals and Government Performance in Parliamentary Democracies",
American Political Science Review 79, 738-754.

Strøm, K. 1988. "Contending Models of Cabinet Stability", American Political Science Review
82, 930-941.

Strøm, K. 1990a. Minority Government and Majority Rule. Cambridge: Cambridge University
Press.

Strøm, K. 1990b. "A Behavioral Theory of Competitive Political Parties", American Journal
of Political Science 34, 565-598.

Strøm, K. 1994. "The Presthus Debacle: Intraparty Politics and Bargaining Failure in
Norway", American Political Science Review 88, 112-127.

Strøm, K. & Leipart, J. 1989. "Ideology, Strategy and Party Competition in Postwar Norway",
European Journal of Political Research 17, 263-288.

Strøm, K. & Leipart, J. 1992. "Norway: Policy Pursuit and Coalition Avoidance", in Laver
& Budge, eds.. Party Policy and Government Coalitions. N.Y.: St. Martin's Press.

Thomassen, P. 1991. En Regjerings Fall. Oslo: Cappelen.

Valen, H. 1973. "'No" to EEC", Scandinavian Political Studies 8, 214-226.

Valen. H. 1976. "National Conflict Structure and Foreign Politics: The Impact of the EC
Issue on Perceived Cleavages in Norwegian Politics", European Journal of Political Research
4, 47-82.

Valen, H. 1981. Valg og Politikk. Oslo: NKS-forlaget.

Valen, H. 1990a. "Coalitions and Political Distances", in Sänkiaho, R., ed.. People and their
Polities. Helsinki: The Finnish Political Science Association.

Valen, H. 1990b. "Velgere, Politisk Avstånd og Koalisjoner", Norsk Statsvitenskapelig
Tidsskrift 6, 17-32.

Valen, H. 1990c. "The Storting Election of 1989: Polarization and Protest", Scandinavian
Political Studies 13, 277-290.

Valen, H. 1992. "Norway: A Changing Party System", in Franklin, Mackie & Valen, eds..
Electoral Change. Cambridge: Cambridge University Press.

Valen, H. 1994. "List Alliances: An Experiment in Political Representation", in Jennings &
Mann, eds.. Elections at Home and Abroad. Ann Arbor: The University of Michigan Press.

Valen, H. & Rokkan, S. 1974. "Norway: Conflict Structure and Mass Politics in a European
Periphery", in Rose. L., ed., Comparative Electoral Behavior. New York: Free Press.

Valen, H. & Urwin, D. 1985. "De Politiske Partiene", in Nordby. T., ed., Storting og
Regjering 1945-1985. Oslo: Kunnskapsforlaget.

Valen, H., Aardal, B. & Vogt, G. 1990. Endring og Kontinuitet: Stortingsvalget 1989. Oslo:
Statistisk Sentralbyrå.

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

Warwick, P. 1992. "Economic Trends and Government Survival in West European Parliamentary
American Political Science Review 86, 875-887.