Between cabinet membership and opposition: Commitment and responsibility of support parties

Existing research suggests that government participation is crucial for how voters evaluate party performance and how they cast their votes. However, in real life the distinction between government and opposition is not as straight-forward as one would think. Minority governments often enjoy the legislative support of external support parties, which play an ambiguous role in politics: while they are formally part of the opposition, they are simultaneously committed to keeping the government in office and passing its bills. How do voters evaluate parties that support a minority government? Will they respond to different frames about the significance of the commitment that support parties have made to government policy and survival? In a survey experiment, I test whether framing a written agreement as effective will cause voters to attribute more or less responsibility to the support parties and find that they respond to different frames by altering their perceptions of the importance of the agreement, but not their responsibility attribution.

In a non-majority situation, government formation and policy making requires coalition building of some sort. However, commitment to cooperation, which is the essential feature of a coalition, does not necessarily imply a majority governing coalition (Strøm, 1990, 24). Minority governments often rely on the stable support of external parties, which have considerable influence on government policy because they are needed to pass any legislation and stay in office. Yet, when studying accountability and retrospective voting, researchers customarily focus on actual governing parties while counting support parties as members of the opposition. The idea that voters will be able to punish support parties for poor economic performance apparently seems farfetched. The assumption that external support parties can enjoy influence without suffering blame is the foundational assumption with which Strøm (1990) was able to explain the very existence of minority governments.
We still know too little about how voters perceive the role of external support parties and to what extent they hold external support parties responsible for policy outcomes. A recent study by Tromborg et al. (2017) has greatly contributed to our understanding of how voters classify support parties. Using observational survey data, they found that Danish voters can identify the parties, which serve as parliamentary support for the government, but they do not recognize their influence on policy making to the same extent as Dutch voters. On average, Dutch voters do not attribute significantly less influence to support parties than to junior coalition members, while Danish voters perceive support parties as having less influence that cabinet members and in some cases even less influence than other opposition parties (Tromborg et al., 2017).
In this current paper, I argue that we ought to think of external support parties as existing somewhere on the spectrum between cabinet membership and opposition. Where exactly a given support party is located depends on its degree of commitment to government policy and survival. One important difference between the Dutch and Danish cases, which Tromborg et al. (2017) correctly point out, is the importance and formality of the support agreement. In the Netherlands, the PVV entered a support agreement in 2010, while the Danish minority governments usually rely on less formal support. This changed in 2019.
In this research note, I find that the otherwise wellinformed electorate has problems identifying the cabinet parties, which confirms the ambiguous role of support parties. Furthermore, I explore how much responsibility for minority government policy voters attribute to external support parties and when. This project elaborates on the existing findings in two ways: First, I explore the theoretical notion that external support parties can be committed to government policy and survival to varying degrees. Second, I experimentally manipulate voter's perceptions of this degree of commitment. After the Danish 2019 general election, the social-democratic single-party minority government signed a written agreement with three other leftwing parties. By framing this agreement as either an explicit, comprehensive, and long-term commitment or a worthless piece of paper, I can manipulate how important respondents' find the agreement, but, paradoxically, this does not have an effect on responsibility attribution.

Pseudo-opposition and quasi-majorities
In his seminal explanation of why minority governments form, Strøm (1990) suggested that when the parliament offers opposition parties sufficient opportunities to influence policies without being in government, some parties choose to serve as external support parties to avoid losing distinctiveness and suffer an electoral cost. Support parties can claim credit for successful policies, but avoid blame when a policy fails. Recent studies have explored this argument in more detail and found that external support parties tend to successfully attain their policy goals (Anghel & Thürk, 2021), that they tend to downplay their support of the government during election campaigns (Müller and König, 2021) and on issues that are salient to them (Müller, 2022). Generally, they are less constrained in what issues they emphasize, which might explain why their electoral performance is similar to that of "true" opposition parties (Thesen, 2016). While these studies teach us a lot about the strategies of support parties, they do not provide much insight into the perceptions of voters. The direct evidence that voters exempt support parties from responsibility is still lacking.
Just like coalition members can be placed on a continuum of "tight" to "lose" commitments and coalition loyalty (Müller and Strøm, 2000, 20-21;578-581), external support parties differ in the extent of their discipline and the nature of commitments to the government. Specifically, Strøm (1990, 108) suggested that minority governments come in two forms: Substantial minority governments, which are not supported by a parliamentary majority, but instead find support for their policy proposals on an ad hoc basis, from day to day and from issue to issue, or alternatively, formal minority governments, which strike a deal with one or more pseudo-opposition parties receiving concessions in return for permanent support.
According to Strøm (1990, 61-62), there are three defining features of a formal minority government: (1) Parliamentary support is negotiated prior to government formation, (2) makes the difference between minority and majority status, and (3) takes the form of an explicit, comprehensive, and long-term commitment to the policies and survival of the government. Because of their secure basis in parliament, formal minority governments might also be described as majority governments in disguise. In particular, when they involve a written and public support agreement, something Bale and Bergman (2006) refer to as contract parliamentarism, can the minority government appear and function as if it controlled a parliamentary majority. For instance, unlike substantive minority governments, formal minority governments are not different from majority governments in terms of their stability and duration (Krauss and Thürk, 2021).
Having established that participation in coalition government can represent varying degrees of commitment to government policy and survival, it is theoretically sound to conceptualize support parties as representing two centrist points on the spectrum between opposition parties and coalition members. Furthermore, I conceptualize support parties with a high commitment towards the government as more like coalition members than support parties with low commitment. Pseudo-opposition parties offer commitments, which will sustain the government in office, ease cooperation, and make the legislative process more efficient. Additionally, it sends a strong signal to voters that government policy is sanctioned by support parties. Ad hoc support parties, on the other hand, only provide vague, limited, and short-term commitments. Thus, I hypothesize that voters will attribute more responsibility for government policy to support parties with high degrees of commitment than to support parties with low degrees of commitment to cooperation.

Danish coalition politics 2020: Continuity and change
Denmark holds the world record in minority governing: The country has not seen a majority cabinet since the early 1990s. Instead, the political system has developed informal institutions for interparty cooperation, such as broad legislative agreements, to build parliamentary majorities and secure effective policy-making. Denmark is a prime example of a consensual style democracy with fragmented responsibility. Because voters are accustomed to minority governments, Denmark represent a most likely case for finding responsibility attribution-of all voters, Danes ought to be the most aware of support party responsibility.
Danish minority governments are remarkably stable and efficient due to strongly entrenched ideological blocs, which represent clear government alternatives. The bloc affiliation is usually clear to voters, because of historical alignments and because parties announce before the election which bloc they support (Green-Pedersen and Skjaeveland, 2019). The Social Liberals have historically belonged to both blocs, but the party's increasing emphasis of socio-cultural issues has made cooperation with the right-wing bloc highly unlikely (Green-Pedersen and Kosiara-Pedersen, 2019). The bloc structure implies that parties belonging to the same bloc have no incentive to topple the government as they risk that the next election will result in a majority for the competing bloc. Thus, their occasional threats to bring down the minority government are often dismissed as cheap talk. This severely constrains the bargaining position of the support parties (Green-Pedersen and Skjaeveland, 2019).
At some points leading up to the 2019 election, it seemed that a break with the tradition of uni-dimensional bloc politics was on the way. One year before the election, the Social Democrats announced that they intended to form a single-party government, thereby effectively distancing themselves from their long-term coalition partner, the Social Liberals. This maneuver was widely interpreted as a signal to voters that the Social Liberals would not have any influence on immigration policy after the election. Another unexpected development was the Liberals suggesting to form a grand coalition spanning the political center, which provided the Social Democrats with an alternative they could use as leverage when negotiating with the other leftwing parties. Left-right placements are illustrated in Figure  1; two-dimensional placements are in the appendix.
After 3 weeks of negotiations, the Social Democrats signed a written agreement with the Social Liberals, the Socialist People's Party and the Red-Green Unity List. This solution was novel because both single-party governments and formal recognition of support parties are rare. Danish minority governments have usually relied on less comprehensive and less committal legislative accommodations (Bale and Bergman, 2006). The 18 page long document is rather detailed on policy areas where the four parties agree, such as welfare services and environmental initiatives, while other areas are entirely absent, for example, immigration policy and European integration. This reflects that the left-wing bloc is strongly united on environmental issues, but disagrees on immigration after the Social Democrats have moved to the right on these issues (Green-Pedersen and Kosiara-Pedersen, 2019).
I cannot convincingly manipulate the two first elements of the formal minority government definition (the support was negotiated in advance and does secure the government a majority), but the commitment element clearly leaves more up to interpretation. The aftermath of the 2019 election conveniently represents a situation with substantial ambiguity over how to interpret the cooperation between the government and the support parties. While governments in Sweden and New Zealand have relied on support agreements since the mid-90s (Bale & Bergman, 2006), a written agreement is a new invention in Danish politics. Since there is no clear precedence for how to interpret it, it provides an ideal context for experimental manipulation of commitment to cooperation. While it might seem that the mere existence of an agreement creates a "high commitment" scenario, this particular coalition formation was unusually long and the subsequent party relations relatively conflictual. Thus, voters are unlikely to have a firm idea about how explicit, comprehensive, and long-term the external support parties' commitment to the government is.

Experimental research design
To explore how voters react to different frames about the commitment implied by the support agreements, I conducted an original survey in collaboration with Yougov. A total of 2164 computer-assisted web interviews of Danish citizens 18 years and older were completed during April 2020. Invitations were sent via email to participants in the Yougov panel. The sample matches the general population on key demographic variables.
I included a control group, which received no treatment, to explore which interpretation of the current political situation is more common. Since we do not know how much voters know about the written agreement, how they interpret it, and how they think about support parties generally, the observational data will be of interest in itself. The preexisting knowledge or attention to the government composition and the written agreement were tested using two simple knowledge questions asking participants to list first the parties currently in government, and secondly the parties that took part in the agreement.
The remaining participants were randomly assigned to either a "low commitment" or a "high commitment" condition. Despite having conceptualized commitment to government policy and survival as continuous, I operationalized it with two distinct points. The government was either presented as a formal minority government and the support parties as pseudo-opposition parties, or the government was presented as a substantial minority government and the support parties as ad hoc support parties.
In both conditions, participants were requested to read three excerpts of a mock news article. After the first two excerpts, participants were asked which of two statements best summarized the content. This served as an attention check, but also to motivate participants to engage with the material. The answers are reported in the appendix. Unfortunately, only 826 out of 1437 participants (57%) in the two treatment groups managed to summarize the article they read correctly. Afterward, participants were asked to evaluate the impact of the support agreement.
The experimental treatments were completely void of any issue content and did not mention any directions or locations such as left, right, "closer too," or "further away." I used budget negotiations because they touch on virtually all policy areas and thus reveal whether the commitments are truly comprehensive as required (Strøm, 1990). Additionally, since these negotiations were finalized months before the survey was fielded, there was a smaller risk that voters will have a clear recollection of how easy/hard they were. The survey was fielded during the first COVID-19 lockdown, which was both unfortunate and unintended. It is possible that the general tendency to "rally around the flag" made respondents perceive party cooperation as tighter or that the prominent leadership of the prime minister made them attribute more responsibility to the Social Democrats. However, this should affect respondents similarly and thus not impact the experimental analysis.
After the treatment, I asked all participants to attribute responsibility to the four parties in the support agreement on the general legislative process, on immigration policy, and on economic policy. Both policy issues were highly salient, but admittedly more so before COVID-19, and citizens were most likely to recognize cohesion or conflict on those issues.

Observational results: Governing versus support parties
The 727 participants in the control group were asked to name the parties in government. The prompt did not mention the existence of external support parties. Only 59% correctly mentioned the Social Democrats exclusively. One might pessimistically interpret this as evidence of widespread political ignorance, but that is far from the case. Table 1 shows how often participants who could or could not identify the governing party correctly answered much more refined questions about politics. Predictably, the participants who exclusively mentioned the Social Democrats also do better on all other knowledge questions, but there is still a majority of participants in the other group who does very well.
Rather than outright ignorance, the failure to name the governing party could indicate a fundamental confusion about what is meant by "government". 17% of participants mention at least one of the support parties but none of the "true" opposition parties as belonging to the government and 7% mention all the support parties. The electorate does not have clearly separated concepts of the government and support parties. These theoretical categories are blurred for voters.
Afterward, participants were asked to name the signatories of the written agreement. Here participants had even more trouble. 285 out of 727 (39%) correctly name all the parties, but almost as many (36%) refuse to even make an attempt. Of the remaining 179 participants, 47 failed to mention one of the support parties, 26 could only mention one support party, and 45 mistakenly included the Alternative. There are many ways to be wrong, but they are not equally detrimental to a voter's ability to understand and navigate Danish politics.
Experimental results: Does high commitment imply responsibility? How will voters respond to different frames about the significance of the written agreement? I test whether framing the agreement as either insignificant and nonbinding or as a serious commitment will cause voters to perceive the four participating parties as sharing the responsibility of government policy. Immediately after the treatment participants are asked how large impact they think the support agreement had. In other words, how constrained are parties by the commitments they have made in the agreement? Answers range from "no impact" to "decisive impact" and have been rescaled to 0-1. Mean and standard deviation is shown in Table 2, cross-tabulations can be found in the appendix.
Participants in the "high commitment" condition on average evaluated the support agreement to be 0.13 units more important than participants in the "low commitment" condition. This difference is significant at the 0.001 level, which strongly suggests that the stimulus material worked as intended. Furthermore, a larger share of the respondents in the treatment groups felt confident answering the question, while almost a third of respondents in the control group did not. Table 3 details the effect of the treatments on participants' attribution of responsibility for the general legislative process. There is no significant change. One might suspect that the null result is due to low engagement with the treatment, but a robustness check including only respondents who passed both attention checks shows that this is not the reason. Furthermore, for the control group, there is little to suggest that voters engage in wishful thinking and significantly overestimate the responsibility of their own preferred party. Results are reported in the appendix.
Similarly, one might worry that the null results are due to the general legislative process being synonymous with COVID policy in this period. However, there is no significant effect on perceived responsibility for immigration nor economic policy. Responsibility attribution on these specific policy areas are unlikely to be affected by the pandemic. The lack of effect on immigration policy is expected and in line with the support agreement content and the rhetoric of the Social Democrats who have made a strong point of not conceding on the issue. However, the null effect on economic policy is surprising as this was one of the most important points in the written agreement. Please refer to the appendix for these results.
In all cases, the control group attributes less responsibility to the support parties than either of the treatment groups does. The results are not directly comparable, since there could be question-order effects at play, but it at least suggests that the mere mentioning of the parties and the negotiations over the national budget, make them seem marginally more responsible for policy in the minds of voters. Overall, across issues and treatments, the amount of responsibility attributed to support parties is rather high with a plurality of respondents choosing the middle category. This result contradicts recent elite-level findings suggesting that support parties are less willing to cooperate with the government on issues they emphasize (Müller, 2022).

Conclusion
This paper explores how voters perceive the role of external support parties and to what extent they attribute responsibility to these parties. I rely on the distinction between substantial and formal minority governments, originally formulated by Strøm (1990), to argue that the degree to which external support parties make an explicit, comprehensive, and long-term commitment to government policy and survival will vary. Does that determine the degree of responsibility attributed to the party? The answers are mixed.
The observational results suggest that there is widespread confusion about whether the external support parties are part of the government or not. Only 59% of survey participants correctly identified the Social Democrats as the only governing party. Furthermore, when asked to name the signatories of the written agreement only 39% could correctly identify all support parties. This should not be interpreted as general political ignorance. Instead, it suggests that voters are unsure about the details of the written agreement and how to interpret it. The experimental results are puzzling. The manipulation check clearly shows that the experimental stimuli worked: respondents who received the "high commitment" condition perceived the agreement to represent a substantial commitment for the involved parties, while respondents in the "low commitment" condition thought it had a smaller impact. However, there are no effects of commitment on the perceived responsibility of the external support parties. Regardless of framing, support parties are held responsible for the policy to a high degree.
Overall, the results suggest that voters are aware of the ambiguous role of support parties but are not equating commitment to government policy and survival with policy responsibility and influence. The paradox could not be resolved within the current study, but opens up interesting new questions for others to elaborate on. Written agreements are certainly not a central element of how Danish voters evaluate party performance. Written agreements are new in Denmark, but they are a growing phenomenon across parliamentary systems (Krauss and Thürk, 2021) and even in places with an established tradition do the support agreements become longer, more specific and more comprehensive (Bale and Bergman, 2006). I encourage a comparative study to examine the responsibility attribution and electoral performance of support parties. Whether external support parties remain insulated from the cost of governing or if they will be perceived of selling themselves short is still unclear and potentially has important implications for the cost-benefit analysis of parties deciding whether to join a coalition government. That will be a worthwhile topic for future research.