Distributing Chairs and Seats in Committees: A Parliamentary Perspective

This article presents an argument for committee assignments based not on the traditional congressional theories, but on elements central to parliamentary systems: government formation. The argument of the article is that it is necessary to include the link between committees and cabinet governance for understanding parliamentary committees. This is tested on 40 years of committee assignments from the Danish parliament. The ﬁndings suggest that an approach inspired by a classic portfolio allocation model works best in explaining the distribution of seats and chairs between parties. Shadowing of coalition partners appears to matter little, if at all.


Introduction
In recent years, there has been an increased focus on how parties assign members to committees in parliamentary systems and how we can understand these assignments. Much of the work naturally takes it point of departure from the congressional theories of committee assignments. However, the evidence as to how well these theories travel are mixed. Mickler (2013Mickler ( , 2017 argues that they can be used on parliamentary systems, while Hansen (2011Hansen ( , 2013 believes that they are less useful and the focus should instead be on developing specific parliamentary theories. Taking a step back and defining what a parliament is can help us understand this position. One view, which is indeed the starting point of the argument made in this article, is that when voters elect members of parliament in a parliamentary regime, they do nothing more than elect a government (Laver, 2006, p. 121). This has significant implications for how we view the organisation of parliament, for instance, the formation of committees and what role parties play in parliaments. When the votes have been cast and counted, thus determining how many seats each party will get, the formation of a government can begin, which is of key importance also for committee assignments.
The formation of the government determines the approach of the particular polity for the time until a new election is called or a new government is formed. If a single party has won more than half of the parliamentary seats, it is clear that it will form the government. However, within the proportional electoral systems in place across Europe this is rarely the case and coalitions are formed. Minority governments in which the government must negotiate its majority from case to case are a normal occurrence in many parliamentary systems. The importance of the relationship between minority governments and committees is well established (see Strøm, 1984), although there is generally a lack of focus on government formation and status when considering committee design in general and assignments in particular. Legislation needs to be shepherded through parliament, and by not controlling a majority of the members of the committees this may complicate the legislative process. Oftentimes these minority governments are coalitions, and coalition formation and coalition governance has been a focus of much scholarly research. These approaches have moved into analysing legislative organisation as an explanation for choices made in coalition formation. Yet, when studying committee assignments in parliamentary systems, focus on government formation and use of the related literature are mostly eschewed in favour of theoretical explanations formed on the US Congress. The argument of this article is that we should not attempt to understand parliamentary committees without actually understanding the link between the committees and the cabinet governance.
The interesting element is that committees are a common feature of modern parliaments. Indeed, all work within highly specialised committee structures to ensure a highly effective legislative system. Specialised committees with limited membership and a significant scope of power are seen as necessary conditions for national parliaments being significant actors (LaPalombara, 1974, p. 123). While it is rare that one party has a majority of seats in the parliament and hence also controls the majority in the committees, the minority governments are supported by other parties which allows for policy coalitions-of a changing nature-to be formed, thus leading to a de facto control of many, if not all, of the committees.
In this article, it is argued that the impact of what seats a party holds in government will play a role in how the parties act when seeking committee memberships. Two hypotheses are presented to understand whether parties seek to maximise membership of committees where they control the corresponding ministerial portfolio or whether parties are using committees as control mechanisms in coalitions.
This argument is tested on a data set covering 40 years of committee assignments in the Danish parliament, 1973-2015, a period with a stable institutional design and with only minority governments. Several measures for committee membership at party level are created and it is examined what role the control of government portfolios play. The choice of Denmark as a case is based upon the stability of the committee system since 1973, the possibility to trace the development over time, and not least due to the existing research on Danish committees. For instance, Jensen (1995, p. 157) contends that the Danish committees are political institutions in which the parties help shape policy. Martin and Vanberg (2011) rate them as strong committees that are seen as powerful tools for policing coalition bargaining, though Carroll and Cox (2012) view Danish committees as relatively weak, especially their chairs, which is also supported by Sieberer and Höhmann (2017). There is an extensive literature on the functions of the Danish committees (e.g. Damgaard, 1977;Jensen, 1995). Quantitative examinations of the committee assignments in the Danish parliament are a recent phenomenon. Hansen (2010) analysed assignment to committees at the individual MP level from 1994 to 2007 andHansen (2013) analysed change in assignments at the individual level during the same time period. This article moves away from the level of individual MPs and instead directs its focus towards the party groups represented in the Danish parliament. It is through the party groups the MPs are assigned and it is the party groups who are the significant actors in the Danish parliament.

Parliamentary committees
A common feature of modern parliaments is that they all work with highly specialised committee structures, therefore ensuring a highly effective legislative system. In most parliamentary systems, the most important role of the committee is to debate and report on bills before and/or while they are being debated in the parliament (Mattson and Strøm, 1995, pp. 269-270). Most often there is a close link between the portfolio of the committees and the ministerial portfolios. This means that the bills reported on by the committees are initiated by ministers who are responsible for the portfolio corresponding to that of the committee. Though, it should be noted that a ministerial portfolio may cover several committees. In some cases, the link between minister and committee also means that the minister can be compelled to appear before the committee to answer questions (Mattson and Strøm, 1995, pp. 293-294). Recently, Zubek (2015) has proposed that when committees have extensive powers, for example, having influence early in legislative process, control of the timetable for legislative deliberations and having the power to seek information about ministerial proposals, we should expect coalition parties to use these powers to counteract ministerial drift. Thus, the committees can be seen as the agents of the parliament as a whole in the day-to-day dealing with ministers. When this is the case, it becomes important how parties distribute the committee seats among themselves, especially in connection with which party controls the corresponding ministerial portfolio.
The vast majority of theories on committees and committee assignments are based on the US Congress (e.g. Shepsle, 1978;Weingast and Marshall, 1988;Krehbiel, 1991;Shepsle and Weingast, 1994). It is only recently that, in the last decade or so, the non-US focus has become more pronounced (e.g. McElroy, 2006;Ciftci et al., 2008;Yordanova, 2009;Hansen, 2010Hansen, , 2011Hansen, , 2013Mickler, 2013Mickler, , 2017Fernandes, 2016). When analysing non-US legislative bodies and drawing extensively on the literature from the US Congress, it is very important not to translate the assumptions and findings directly, as they stem from a very different, and unique, institutional setup (Gamm and Huber, 2002). Take for example the relationship between committee assignment and district (e.g. Adler and Lapinski, 1997), which in majoritarian electoral systems is a clear relationship, but in proportional electoral systems, for example, open, closed or semiclosed lists, and different tiers of seat allocation, it is nearly impossible to argue for the presence of a meaningful relationship. This is but one of the challenges using theories based on the US experience, with another being that the parties, not the individual legislators, are typically the most important actor in parliamentary systems. For instance, this is clear when it comes to controlling the agenda and the time allotted to debate (Döring, 1995), and also when it comes to controlling the members of parties serving on committees (Damgaard, 1995).
Another issue separating the parliamentary systems from the Congress is the connection between the portfolios covered by committees and those found in the government (Mattson and Strøm, 1995, p. 275). Committee members therefore also play a key role in dealing with the proposed legislation in that area, and in keeping the responsible minister in check. However, if we look at the committees of the parliaments of Western Europe, for example, the variations in the power of committees are large (see Mattson and Strøm, 1995, pp. 285-295). In some committee systems, there are no bill re-writing powers vested in the committees, whereas in other systems the committees responsible for the bill reporting also have a second function in controlling the executive. Some features are, nevertheless, more universal than others. First, the party groups are the important actors in nearly all parliaments. Secondly, the link to the ministerial portfolio is also nearly universal. This means that if the aim is to look elsewhere than the congressional theories for inspiration in explaining committee assignments, it is necessary to focus on aspects that are important for parliaments. One of these significant aspects is coalition formation and how coalition bargains are policed.

Government formation models
The workings of coalition governments have seen much scrutiny both empirically and theoretically (e.g. Strøm, 1984Strøm, , 1986Saalfeld, 2000;Strøm et al., 2010). For this article, the two most relevant theoretical models developed are the portfolio allocation model of Laver and Shepsle (1996), and the legislative institutions model of Martin and Vanberg (2004, 2011. The rationale for examining coalition formation and coalition governance is that the allocation of government portfolios is central to the process of government formation, and it is possible to view the control of relevant ministries as a crucial link between party and government policy (Bäck et al., 2011).
The first coherent model of coalition delegation was presented by Laver and Shepsle (1996), who in their portfolio allocation model argue that parties accept the fact that once a minister is appointed, it is not possible for other coalition partners to successfully influence that minister. In other words, once appointed, ministers have a near-monopoly on policy for their portfolio. Indeed, their model specifies that as long as the policy enacted by the relevant ministers is preferable to the status quo for all coalition partners, it is a viable coalition and coalition partners will not interfere in policy decisions of other parties. Laver and Shepsle (1996) assign no particular role to the legislative institutions in their model as it is policy positions of the coalition partners that will ensure a functioning coalition. This is counter to the argument put forward by Martin and Vanberg (2004, 2011, who place specific emphasis on legislative institutions in upholding a coalition bargain. Martin and Vanberg (2011, pp. 34-35) argue that whether government parties can use legislative institutions to police a coalition bargain depends on whether the parliamentary institutions allow (i) information about policy details of bills and feasible alternatives and (ii) the opportunity for change. They argue this is most importantly done in the legislative committee system where parties can acquire information and propose amendments. The main argument of their model is that parties engaging in coalition governments will, in strong legislatures, use the legislative institution to keep tabs on coalition partners and ensure that ministerial discretion on policy is not misused. Their study is based on legislative proposals for five countries that are tracked through the legislative process and examined with regards to amendments, policy area, government issues divisiveness and other factors. 1 1 When considering the strength of the committee system, the evidence of monitoring is less clear: Sieberer and Höhmann (2017) find that monitoring does not increase when the position of chair becomes more powerful. See also Zubek (2015) and André et al. (2016) for analyses on the needs of parties in coalition government and the level of power given to committees.
These two perspectives on cabinet governance in parliamentary system have implications for how we should understand the role played by the parliamentary committees. The argument of this article is that we should not attempt to understand parliamentary committees without actually understanding the link between the committees and the cabinet governance.

Linking coalition governance and parliamentary committees
First off, it is important to note that while it is not possible to equate the theories of government formation and coalition governance to parliamentary committees directly, there are lessons to be learned. There are very clear elements from these theories that can be used to inspire a parliamentary explanation for the assignment of members to parliamentary committees. This section will attempt to link the two sides and explain why the point of departure should be taken in the government formation process. Martin and Vanberg (2004, 2011 argue that coalition partners use committees to police the coalition bargain through amendments posed in the committees. This implicitly assumes that coalition partners are represented in the committees in order for them to police the bargain and hence the allocation of seats matters, not least because the power that is given to members of parliamentary committees in systems means that they are viewed as strong committees (Martin and Vanberg, 2011). Other work on delegation and coalition governments has focused on what takes place at the cabinet level. For instance, Thies (2001) argues that smaller parties are disadvantaged in cabinet committees as seats are usually divided proportionally which provides them with a disadvantage in keeping tabs on all coalition partners equally. 2 Others focus on the use of committees and chairs as instruments for shadowing coalition partners (Kim and Loewenberg, 2005;Carroll and Cox, 2012) and most recently Sieberer and Höhmann (2017) questioning the use of chairs as a monitoring device. This means that it is necessary to explore further how it is possible to elaborate upon the argument of coalition governance and committee assignment more in detail.
The most important prize a party can win is to gain seats in the cabinet. This provides the party with the potential for stronger pursuit of its policy goals, as well as generally influencing policy to benefit the party's short-term and longer term goals to benefit the party electorally. However, ministers are also seen as having very strong policy influence within their own portfolio (Laver and 2 Cabinet committees exist in Danish cabinets and have played an increasingly important role with the main committees changing composition to facilitate the coalition in place. Since 1994 written coalition agreements are also found which highlights the policy priorities of the coalition governments. These two elements are currently not examined in this article. Shepsle, 1996). This creates the necessity for some form of bargaining mechanism where the coalition partners can seek to pull the ministers in the direction of their preferred policy. Depending on the institutional setup of the particular country this can happen through several mechanisms, for example, cabinet committees, junior ministers from other parties and the focus of this article: parliamentary committees. This means that when a party is allocated a particular portfolio in the government, there is little reason for the party to seek to maximise the committee chairs or size of membership of the particular committee. Though, it is important to keep in mind that committee chairs are valuable commodities in legislative institutions, as some scholars term such positions 'mega-seats' (e.g. Carroll et al., 2006). Additionally it has been shown that parties at times distribute committee seats to keep tabs on their coalition partners (Kim and Loewenberg, 2005). Thus, it is necessary to look at both the committee chairs and the ordinary committee members in order to gauge the importance of controlling the government portfolio for the assignment of these two elements.
H1: Parties controlling the government portfolio will seek to gain the chair of the corresponding committee and/or maximize its representation in the committee.
If this hypothesis is confirmed and we see that the chair is from the party holding the ministerial portfolio and/or an over-representation of committee members from the party holding the government portfolio, this would be an indication that the logic of portfolio allocation as bargaining at the government level is also present at the parliamentary level, and that parties aim to bolster the control of the portfolio by seeking parliamentary control to augment the executive control. If the hypothesis is rejected, this could point towards the committees being used as shadowing or bargain policing entities. However, this will depend on the next hypothesis being tested and confirmed, as not having strong representation by the party holding the portfolio is not necessarily equal to coalition partners seeking to maximise their respective positons. This leads us to the second hypothesis: H2: Parties in government will seek to maximise representation on committees (seats and chairs) where the portfolio is held by another party in government.
These two hypotheses are overlapping, but it is necessary to have both included. This is due to the fact that coalition governments might not be a majority government, which is the case in the analysis presented in this article where no majority coalitions exist. Under minority governments, the government will need to rely on either support parties, that is, parties with a natural affinity for the policies of the government, or parties getting paid for their support through legislative accommodations or other spoils. Alternatively, they may also allow the main opposition parties to influence policy in return for their support. Hence, it is possible that no systematic allocation of the parliamentary committee seats between coalition partners can be found as they do not control the majority outright. This might be down to the view that under minority government, legislative committees are the most important legislative institution for opposition parties (Strøm, 1990, p. 43). It might also be evident that the policing of the coalition bargain takes place outside of the legislative committees and it is instead necessary to look at the committees at the executive level, such as the cabinet committees.
The two hypotheses are not the only ones that could be formulated and tested, but are those that will provide an answer to the extent to which government portfolio distribution influences the distribution of parliamentary committee posts, as is argued in this article. The questions raised in this article are tested on the case of Denmark which for these purposes is an excellent case. The institutional stability in terms of legislative organisation, the constant occurrence of minority coalition governments from 1982, and an average of eight parties continuously represented, albeit with change in which parties, allow for the best possible case for examining the questions presented in this article. In Section 5, these elements are explored more in detail.

The Danish parliament and its committee system: a descriptive analysis
Since 1953, the Danish Folketing has been a unicameral legislature with 179 members; 4 members elected from Greenland and the Faroe Islands (2 each) and 175 from Denmark itself. The main actors in the Danish Folketing are the party groups. The institutional setup has been stable for nearly 40 years. Stability in numerical terms can also be ascribed to the party system. Since its last upheaval in 1973 there has constantly been between 8 and 11 different parties represented in the Folketing, and also constant minority governments with no single-party government from September 1982 until June 2015.
The last substantial change to the committee system was the introduction of standing committees in 1972 (Damgaard, 1990, p. 35). 3 Following the change, standing committees were established to correspond with the ministerial portfolios. 4 The committee members are allocated by the parties who are assigned 3 They were introduced as a replacement for the ad hoc committees formed to deal with each separate piece of legislation. Even though some standing committees can be traced back to the 1870s, the system in its entirety did not change until 1972. It should also be noted that it is still possible for the Folketing to form ad hoc committees. 4 It is the prerogative of the Prime Minister to name and distribute the ministerial portfolios. It is tradition that the Parliament follows with change to the committee name and structure, although there are seats at the beginning of each parliamentary year. The allocation of seats between the parties is governed by the Constitution ( §52) where it is stated that all elections to commissions and committees must happen by proportionality. How many committees and how many seats in each is governed by the Standing Orders of the Folketing (Folketinget, 2006), making it possible to change the Standing Orders with a simple majority. Despite the possibility to change the setup of the committees, this rarely happens.
At the beginning of each parliamentary year two blocs of parties form, one consisting of the parties in government and those who support the government, and another made up, usually, by the rest of the parties. There is no guarantee for a party to be admitted to either bloc, and examples exist of parties standing outside of the blocs, thus getting no seats in any committee, though this has not happened since the late 1980s. The committee places are then assigned through the D'Hondt allocation procedure. It should be noted that while the allocation of seats is proportional, the allocation of chairs is not proportional. If the largest bloc wish to fill all the committee chairs they are allowed to do so. However, in practice the largest bloc nearly always allow some committee chairs to be filled by the smaller bloc.The formal powers vested in the Danish committee chairs is not particularly strong (see Sieberer and Höhmann, 2017) although Christiansen and Hansen (2017) found that committee chairs receive significantly more attention in the media on issues relating to the policy brief of the committee and generally reports intra-party competition for the chair positions.
The intra-bloc allocation happens through negotiation. Smaller parties might get more seats or higher ranking committees than they would otherwise be entitled to, or even in some cases be given a chair (see Hansen, 2010). Given that there are many parties represented in the Danish parliament, a complete proportional representation of all parties in all committees is not possible. This means that there will be an over-representation of a party in some committees and under-representation in others, and this is not simply a function of the way the D'Hondt allocation works. Before 2011, smaller parties that is, those with four or five members, had no proportional right to a seat on all committees and needed to prioritise and negotiate for their assignments with the other parties in their bloc. The parties have the possibility to change the assignments of their own MPs during the year by informing the parliamentary offices that they wish to do so (Hansen, 2013). Changing the number of seats allocated to a party in a committee can only take place at the start of a new parliamentary year, which begins two weeks after an election and at every year on the first Tuesday in October.
examples of the Parliament wishing to keep the existing committee structures in place which could leave a committee linked to two or even three ministers.

Distributing Chairs and Seats
The functions of the Danish committees are that they can call the minister covering the portfolio to appear before the committee at any time and to pose questions regarding the legislation the committee is working on. This makes the Danish committees very influential in the Danish parliamentary system, especially in connection with legislation where it is the report of the committees which forms the basis of the decision of the Folketing (see also Damgaard, 1990). 5 The Danish committees cannot initiate or re-write legislation directly, although they have the right to propose amendments and can refuse to report a bill out of committee, until ordered to do so by a majority of the parliament (Döring, 1995, p. 238;Mattson and Strøm, 1995, p. 300). For a multi-party system like the Danish parliament, where minority governments are the norm, strong committees are important. This is the case for both the opposition parties to monitor the government policy and for the parties supporting the government, while not directly participating. The day-to-day work of the committee is governed by the chair, who sets the agenda regarding the order of business and when a bill or a ministerial questioning should take place. The chair is also responsible for the latter.

Data
This article draws on a complete dataset of all committee assignments in the Danish parliament from 1973 to 2015. Previous studies either relied on data from 1973 to 1976 (Damgaard, 1977), from a single parliamentary year (Jensen, 2002), or from 1994 to 2007 (Hansen 2010(Hansen , 2013. Elections are held every four years unless called earlier, yet each parliamentary year starts on the first Tuesday in October and ends on the first Monday the following October. In the case where an election is called between these dates, as most are, the parliamentary year is split into two sessions, one before the election and one after the election. This is not trivial for the analysis presented in this article. Parties do change their committee members between general elections, as they also do so between parliamentary years (see Hansen, 2013) where the committees are formed anew. The unit of analysis is each party in each committee in each legislative term, which in the full data set leaves us with 11,916 observations. 6 5 Another powerful tool for a committee is the possibility of calling a minister to appear in front of the committee to answer questions on current and developing political issues. Thus, the committees have become an important part of the parliamentary control with the government (Damgaard, 1990, p. 37) 6 The committees included in the analysis are all legislative committees, plus the Rules Committee, Citizenship, Rural Affairs and the European Committee. The committees on Greenland, Faroe Islands, Small islands, Intelligence Services and other similar committees are not included as they serve a different purpose than dealing with both legislation and government scrutiny as the other committees.
The first dependent variable examined measures whether the party in the committee in question holds the committee chair in the particular legislative term. Independent variables included are whether the party in the committee in question controls the corresponding ministerial portfolio and whether the party is a member of the government or not. A variable measuring the size of the party, in terms of number of parliamentary seats is also included. The second dependent variable used in the analysis is one measuring the level of over-or underrepresentation per party per committee per legislative term. First, the average number of seats a party is entitled to on each committee is calculated by multiplying the proportion of parliamentary seats held by a party by the total number of statutory committee seats available. This creates a measure of seats available to a party which is subtracted from the observed number of seats, thus producing a measure of over-and under-representation not unlike the one used by Hedlund et al. (2009) in their study of stacking in the US state legislative committees. A positive number is equal to over-representation and a negative number is equal to under-representation. However, as this is fractional measure, all the fractions have been rounded to the nearest whole number following traditional rules of rounding. This is a crude measure, but one that is realistic as either a party has the number of seats on a committee as it proportionally should have, or it is over-or under-represented by one or two seats, and if the latter case it has involved some form of trade-off in which committees to prioritise.
As committees and ministerial portfolios are linked, it is a relatively small step to use the ministerial portfolio ranking as a proxy for committee ranking. Here the method used is the one proposed by Hansen (2010) who in his study of individual MPs and their committee assignments used the portfolio saliency rankings by Druckman and Warwick (2005). Druckman and Warwick (2005) provide not only the latest portfolio salience measure available, it is also the most comprehensive covering all portfolios, even those unique to the country in question. As Hansen (2010) argues, there are challenges in using portfolio salience measures for measuring committee salience. Committees exist which have no direct link to a portfolio, such as, in this context, the committees for Europe, Rules and Elections. Thus these committees need to be ranked according to their salience. Europe is without a doubt among the most important committee providing Danish members of the Council of Ministers with a mandate for negotiating, and the latter Elections committee is created by virtue of the Constitution. This committee is ranked by assigning it the average of all committees with a score of more than 1 in the Druckman and Warwick (2005) survey.
Another challenge is whether the parties actually rank the committees equally and whether this is constant across time. Groseclose and Stewart (1998) found that for the US House of Representatives committee, ranking does change over time. However, for parliamentary systems we lack similar evidence. What we do know from the literature of portfolio allocation in parliamentary systems is that certain parties value certain government portfolios higher than others. Social Democratic parties prefer the portfolios linked with the welfare state, Conservative parties prefer the economic portfolios as well as the justice and defence portfolios. This suggests that it is necessary to at least control for the potential effect of variation in saliency at the party level. For this data, it has been decided to rely on the saliency measure proposed by Bäck et al. (2011) based on the Comparative Manifesto Project. The problem with this measure is that it does not cover all policy areas and as will be seen in the analysis it effectively halves the size of N. These variables are those that will be used in the analysis and in Table 1 below are the descriptive details of the variables presented.

Distributing the chairs and seats
Once the committee seats are allocated, the committee chair is formally elected among its committee members, although the distribution of chairs is negotiated between the electoral alliances in the beginning of each parliamentary year. This entails that a committee chair need not have a majority behind him or her in the committee. For this part of the analysis, the dependent variable used is one that measures whether the party holds the chair of the particular committee. It is a binary variable, that is, either the party holds the chair or not, and for that reason a logistic regression has been chosen. The independent variables included measure the number of parliamentary seats, the importance of the portfolio measured through the portfolio saliency measure by Druckman and Warwick (2005), whether the party also controls the corresponding ministerial portfolio, and the extent to which the party is over-or under-represented in the committee rounded to the next whole number. In a second model, the saliency attributed by each party to the committee portfolio is also included following the works of Bäck et al. (2011). The results of the two logistic regressions can be found in Table 2 below. The main results to take home from this analysis are that the number of parliamentary seats matters, hence the more seats a party has the more likely it is that the chair will be held. Secondly, if a party controls the government portfolio, it is more likely that they will control the committee chair, but generally speaking under-represented parties are more likely to control the chair. These results point us in an interesting direction. That the number of parliamentary seats matters is perhaps not surprising. The more seats a party has the more MPs they need to find gainful employment for, this also means that we should expect the major parties to take a larger share of the committee chairs, so this relationship is as we should expect. However, it should also be seen in connection with the representation in the committees and the control of the chair. The larger parties are more likely to not fill all of their seats in all committees and instead give some to their supporters in the distributional blocs formed at the start of each parliamentary year (see Hansen, 2010). This means that a party which would be entitled to six committee seats might instead only take five or even four, and then on the other hand take the chair of that committee. Interestingly, it also seems that if a party controls the relevant government portfolio, they are more likely to take the position of chair. This runs counter to the second hypothesis and suggests that the actions of political parties, at the least in the Danish case, is more alike that of what should be expected if policy responsibility is being distributed in the government formation process and mirrored in the parliamentary process. The evidence presented in this part of the analysis does not support a view of committee chairs being used for bargain policing or even shadowing, indeed the opposite seems to be the case. 7 These findings give rise to the next part of the analysis, which is the extent to which the parties are strategically stacking their membership of some committees and not others, and whether there are different factors in play depending on whether a party is over-or under-represented. For this analysis, the dependent variables can take three values: either a party can have exactly the proportion of members that they should have based on the proportionality distribution, or they can have more or less seats. Instead of using the fractions of the representation, it was chosen to round all values up or down based on the usual mathematical rules for rounding. It is also less interesting to note whether a party is over-represented by two seats or by one seat, but that they are over-represented. Given the nature of the variable, a multi-nomial logistic regression has been chosen with the values of no over-or under-representation as the baseline category that under-and over-representation respectively are compared with. These results can be found in Table 3 below where all parties are included into the estimation.
The results are varied with clear differences depending on whether it is overor under-representation that is observed. Holding the chair means that over-representation is significantly less likely, suggesting that there is a high value attached to the position of chair. On the other hand, if the party controls the particular ministerial portfolio over-representation is significantly more likely. When looking at the interaction between holding the chair and being a government party, we see the strong difference where government parties holding the chair are significantly more likely to be over-represented on the particular committee. This suggests that where opposition parties hold a chair, it comes with the cost of other seats on the committees. Overall, this is strong evidence that it is the logic of a more traditional portfolio allocation model that governs the distribution of committee assignments between the parties. This logic was also confirmed when using the chair position as the dependent variable in the previous analysis.
Other interesting relationships are that government parties are generally either over-or under-represented in the committees compared to having the full proportional relationship. This suggests that the government parties actively seek to manage their committee assignments, for instance through giving more committee seats to some of their support parties, while taking slightly more seats on other types of committees. The number of parliamentary seats variable has the impact 7 Given the analysis is at party-level, it is not possible to say anything on the extent to which chairs are used for mollifying MPs who did not make the cabinet. that larger parties are slightly less under-represented, and slightly more overrepresented, but the magnitude of these relationships is extremely minor.
This analysis includes all parties and while this informs us about some general trends in the assignment of committee seats, it is nevertheless necessary to limit the analysis to only government parties allowing for an understanding of what goes on between the parties engaging in formal coalitions. This means excluding all single-party governments, in effect dropping the years 1973-1977 and 1978-1982. The estimation approach is the same for the previous analysis, except for variable for government party and the interaction not being included as this is limited to government parties, the results can be found in Table 4 below.
The results in this part of the analysis are more mixed, but still points towards the traditional portfolio allocation model being the more important logic behind committee assignments. Only one of the relationships that were found when looking at all parties is present: the impact of the number of parliamentary seats, and it suffers from the same issue as previously in having an extremely limited effect size. The variable for controlling the committee chair is in the same direction as it was for all parties, and the relationship is significant except when it comes to the relationship between over-representation and proportional representation in Model 5. Party control of the government portfolio disappears as a significant variable compared to when all parties are included, though again except in relation to over-representation in Model 5. Overall, the results of the analysis suggest that assignment of committee chairs and seats in the Danish parliament is primarily explained by a logic inspired by the portfolio allocation model, thus it is the distribution of government portfolios that is determining how the committees are distributed. There also is no strong evidence that committee chairs or stacking of seats are used in the shadowing capacity as argued by scholars such as Carroll and Cox (2012). However, it should also be noted that the effect sizes are generally very small, except for the chair variable and the interaction between chair and government party, and it is only evidence from one country. The problem with the effect sizes are of course also a question of using the available data. More than 40 years of observations are included in this analysis, including all the committees from the inception of standing committees in the Danish parliament.

Concluding remarks
General elections present the parliamentary parties with the number of seats and allow for negotiations for government formation. In the case of Denmark, as studied in this article, all governments since 1971 have been minority governments Note: ** p < 0.01, * p < 0.05. either single-party or, since 1982, coalition minority governments. Having no majority requires the government to seek a majority bill-by-bill and negotiate to get legislation through parliament. In these negotiations, it is important to have some control over the committees. However, as there is no majority behind the government, there is also no majority for it in the committees, unless the government engages in some form of strategic stacking of committees. This is possible as committee seats are allocated predominantly between two blocs, created only for the purpose of distributing committee seats. Government parties might prioritise certain committees over others, and therefore increase their representation on those while allowing minor parties, who are not guaranteed a seat on all committees, more committee seats than their seat share otherwise allow them. The findings presented in this article are in line with what should be expected from governments interested in securing their policy in the best possible way through parliament.
Where theories of committee behaviour fail in parliamentary systems, it is in the assumption that individual MPs can go against their party and prioritise particular policies that differ from their party. In those cases, as argued in this analysis, the parties will swiftly remove the offenders from the committee seats and replace them with loyalists. This is another reason why focusing on parties over individual MPs is important in parliamentary systems. That being a factor in the near-perfect level of party cohesion in Danish parliamentary votes with the possibility of changing committee members at will (see Hansen, 2013), we must thus conclude that parties are the dominant actors. For government parties there is also another element at play, whereby the MPs not serving in government will be loyal to the party and to the government in order to maximise their chances of promotion to the government. Therefore it is not from within that government parties should be concerned with getting legislation passed. It is instead two elements that are relevant (i) coalition partners and (ii) opposition parties, some of which are needed for a majority.
Generally, it must be assumed that parties in a coalition function as a coherent body that can count on one another in government, in parliamentary voting and in committees. If this is not the case, then the government is unstable and it is not likely that it will survive. In this respect, treating the government as a unitary actor makes sense.
The second element relates to the negotiations necessary for the policy coalitions formed to make a parliamentary majority bill-by-bill. The coalitions are formed for each bill, although most governments will have one or two favoured parties, so-called support parties. These parties can be those included in the blocs formed for the committee assignments at the beginning of each parliamentary year. As such, it is possible for non-government parties to strategically stack a few particular committees, especially those less prioritised by the government.
This was for instance the case in the Citizenship committee in the 1970s where the Socialist People's Party took two seats where they were only supposed to have less than one. This was allowed by the other parties in the distributional bloc, particularly the government party, the Social Democrats. When the conservativeliberal government was formed in 1982, the Christian People's Party gained one ministerial portfolio, the minister for environment, and at the start of the next parliamentary year they sacrificed seats on several committees to secure the chair of the environment committee, consequently gaining full control over proposals and on their passage through parliament. This is of course only anecdotal in nature, however this suggests that there is a broader story to be told, and especially the latter example is in clear support of the results presented in this article.
Committees are an important part of any modern legislature, and yet they are elements we have relatively little knowledge about, for instance concerning what happens under minority coalition governments. This article has presented a perspective on assignments inspired by one of the most important elements related to parliaments: government formation. This is certainly not the final word on how parties assign members to committees, but it does present a different perspective to the congressional theories which dominate the studies of committee assignments in parliamentary systems. Whether these patterns are also found outside of the more than 40 years' worth data from the Danish parliament remains to be seen; however, there is now an outline and results to allow further work to draw upon.