Select Committees and Brexit : Parliamentary Influence in a Divisive Policy Area

In this first comprehensive study of House of Commons departmental select committees and Brexit, we analyse the incidence of divisions (i.e. formal votes) on Brexit-related issues in select committees, showing that unanimity remains the norm with the exception of the Exiting the EU Committee, which is the most divided select committee of recent times. We add to the literature on select committee influence by adapting Russell and Benton’s (2011) methodology to examine government responses to committee recommendations on Brexit. We show that most have been partially accepted but that the level of policy change, rather than divisions on recommendations, is the main factor explaining variation, with greater success for recommendations suggesting a lower degree of change. Committees have also had indirect influence, shaping the agenda and bringing information into the public domain.

Brexit is a highly divisive issue in British politics.Nevertheless, the presence of a minority government since 2017 offered the UK Parliament a greater opportunity for influence than would otherwise have been the case.Departmental select committees in the House of Commons are potentially an effective tool for achieving such influence (Russell and Benton, 2011).How far has this potential been undermined by the polarisation of views on Brexit among MPs?We assess the effectiveness of select committees in the Brexit process by looking at the extent of divisions on Brexit-related issues and how influential committees have been, via a quantitative and qualitative analysis of their reports and government responses to them.The article also draws upon interviews with committee members and clerks conducted on a non-attributable basis in the 2017-2019 parliamentary session.
Brexit has become a major area of activity for Commons' departmental select committees, with 66 inquiries and 56 of 267 reports published in 2017-2019 (21.0%) having a primary focus on Brexit-related issues. 1By contrast, only 22 of 881 reports (2.5%) published between 2010 and the 2016 referendum were primarily concerned with European Union (EU) issues.Brexit is the core focus of the Exiting the European Union Committee, created in 2016 to scrutinise the work of the new government department.But every departmental select committee has launched a Brexit-related inquiry, with only International Development and Transport yet to produce a major report.
This article is the first comprehensive study of select committees and Brexit, assessing the challenges it has posed for select committees, the scale and nature of committee activity, the incidence of committee divisions on Brexit, and committee influence on the Brexit process.It begins by examining the opportunities and challenges that Brexit created for select committees.Select committees are a consensual element of an otherwise partisan parliament.Most reports are agreed by consensus, but the scale of intra-and inter-party divisions on Brexit in the Commons made achieving consensus more challenging.We provide a detailed analysis of the incidence of divisions (i.e.formal votes) on Brexit-related issues in select committees, showing that unanimity remains the norm.However, the Exiting the EU Committee is highly divided.Here, the main fault line is between MPs who voted Remain and those who voted Leave, with the latter in a minority on most votes.The article also adds to the literature on select committee influence by adapting the methodology used by Russell and Benton (2011) to examine how far committee recommendations on Brexit have been accepted by the government.We find that most recommendations have been accepted, at least partially, but that many of these asked for no change or small policy change.Recommendations on which committees were divided in Brexit-related reports were less likely to be accepted than others but this relationship does not hold up when we take into account the degree of policy change that committees requested.This latter factor is the most important for explaining the success of recommendations, with requests for large change more likely to be rejected.Committees have also had indirect influence, shaping the agenda and bringing information into the public domain.Despite its divisions, the Exiting the EU Committee has achieved some influence.

Brexit and challenges for select committees
Select committees offer MPs a high profile and potentially effective tool (Russell and Benton, 2011) for attempting to influence the Brexit agenda and the course of the Brexit process.Research on the UK Parliament undertaken before the EU referendum showed that it is an influential actor, making more of a difference to government agendas and legislation than often assumed (Russell and Cowley, 2016).The Brexit process has, however, posed problems for select committees.One concerned timing, with inquiries halted because of the 2017 election (some were picked up by successor committees).The subsequent delay in establishing committees meant that UK-EU negotiations were underway before scrutiny began in Commons' committees.
The complexity of the Brexit process also poses challenges.Few select committees had previously held inquiries into EU-related issues.Members needed to develop greater understanding of these issues, and EU experts were appointed to the committee staff.The secrecy of the Brexit negotiations has been another challenge.Brexit Secretary David Davis said the government would 'match, and hopefully, improve on what the European Parliament sees' (House of Lords European Union Committee, 2016, p. 65).But this did not materialise.The government is often unwilling to divulge its negotiating strategy or commercially sensitive material.Securing regular ministerial appearances has also been difficult.The quality of government responses has caused frustration, with the Home Affairs Committee complaining that the government 'has shown so little inclination to engage with scrutiny of its preparations for Brexit' (Home Affairs Committee, 2018, para. 4).The government is expected to respond to reports within two months but has frequently missed this target.
A further challenge arises from the proliferation of Brexit-related inquiries.As one interviewee noted, 'Brexit is the only game in town, and everyone wants to play'. 2 The effectiveness of these inquiries may be improved if committees coordinate their efforts to scrutinise the Brexit process.Informal discussions occur between committee chairs or clerks, but there are limits to what can be achieved outside of formal mechanisms although the appointment of a Brexit coordinator improved communications. 3The Liaison Committee, which considers select committee effectiveness, is a potential venue in which to work out how committees can complement each other and reduce duplication but has shown little inclination to do so.By contrast, the House of Lords established an Informal Brexit Liaison Group.It brings together members of the Lords' Liaison Committee and the Senior Deputy Speaker, hosting outside experts and overseeing committee scrutiny.
Some overlap in the content of inquiries is inevitable given that Brexit covers strategic and technical issues across multiple policy areas.

Select Committees and Brexit
Committee has held wide-ranging inquiries and draws upon evidence from other committees.The Business, Energy and Industrial Strategy Committee (BEIS) adopted a sectoral approach, its inquiry on Brexit's impact on business having five sub-inquiries covering different areas.One, on processed food and drink, aimed to avoid overlap with the Environment, Food and Rural Affairs Committee's (EFRA) inquiry into trade in food, which focused on producers.Some committees have held one-off joint evidence sessions.The International Trade and Treasury committees held a joint session on trade policy in April 2018, and their chairs warned that new trade deals would not offset losses following from Brexit (Morgan and MacNeil, 2018).However, there have been no formal joint inquiries or reports on Brexit.
Over 100 MPs from each of the Conservatives and Labour rebelled on Brexit between June 2017 and April 2019.There has been cross-party support for motions to oppose no deal and allow indicative votes but on the latter, soft Brexit and Remain options were primarily supported by opposition MPs while hard Brexit options were almost exclusively backed by Conservatives.For select committees, forging consensus across parties and the Leave-Remain divide has been challenging.MPs who voted Remain in the 2016 EU referendum are in a majority on every 2017-2019 committee except Northern Ireland Affairs (see Table 1).The proportion of Remainers on committees increased slightly after the 2017 election and Leavers lost their majority position on two committees.

Consensus and divisions
There is an 'engrained culture of consensus in select committees' (Russell and Benton, 2011, p. 37).Select committee reports are viewed as more likely to be influential if they are unanimously agreed (Besly and Goldsmith, 2019, p. 342;Tyrie, 2015).They can be billed as cross-party and are deemed to carry more weight than if a committee is heavily divided.If the government accepts key recommendations, the status of the committee is enhanced.But if consensus cannot be reached, it becomes easier to dismiss a report as partisan.
Divisions (i.e.formal votes) may be held on paragraphs summarising evidence, recommendations, or the whole report.Members may table and request a vote on an amendment, although the vast majority of amendments are either withdrawn or agreed without a vote. 4Some divisions occur because a lone MP wishes to put their concern on the record.Divisions on the final report are significant as they prevent the committee from producing a unanimous report.Divisions on draft reports also signify general dissatisfaction.MPs may table an alternative draft report.There have been 48 divisions on draft and final reports, and three on alternative draft reports since 2010.
From May 2010 to the Easter 2019 recess, there were divisions on 125 of 1,325 departmental select committee reports (9.5%) and a total of 412 divisions.The proportion of reports to have seen a division is lower in 2017-2019 (7.1%) than in the 2015-2017 (8.6%) and 2010-2015 (10.6%) parliaments.The number of divisions per report, calculated by dividing the total number of divisions by the total number of reports, is however marginally higher in 2017-2019 (0.37) than in 2015-2017 (0.32) or 2010-2015 (0.29).Looking at those reports on which divisions occur, the mean number of divisions has increased from 2.8 in 2010-2015 to 3.8 in 2015-2017 and 5.2 in 2017-2019.40% of all divisions occurred on the 11 reports which saw 10 or more votes-three on the Exiting the EU Committee.Division rates vary across committees.There has not been a single division since 2010 on the Education or Treasury committees.
Most divisions in select committees in 2017-2019 were on reports focusing on Brexit, although 68 of the 99 votes occurred on the Exiting the EU Committee.The seven divisions on two Northern Ireland Affairs Committee reports on the Irish border saw a Leave versus Remain split with Labour's Kate Hoey voting with Conservative and Democratic Unionist Party (DUP) Leavers, notably on the conclusion that technological solutions to the border issue could be piloted during the implementation period (Northern Ireland Affairs Committee, 2019, para. 55).12 of 13 divisions in the International Trade Committee occurred on a report on the UK Trade Remedies Authority but there was no Leave-Remain divide, while two divisions in the EFRA Committee report on Brexit and trade in food saw Labour split on support for farmers. 5Excluding the Exiting the EU Committee, 37 of 43 Brexit-related reports achieved unanimity, including on issues such as customs, immigration (Home Affairs), transitional arrangements (Treasury) and the impact of Brexit on business (BEIS).

The Exiting the EU Committee
The Exiting the EU Committee is the most divided select committee, recording divisions on 15 of the 16 reports issued between January 2017 and March 2019.11 saw a division on the final report while another saw a vote on the draft report.Since its creation, the committee has recorded 79 divisions, 38 of them on recommendations.In this section, we examine the causes and features of these divisions.
As lead committee on the most contentious of issues, achieving consensus was always likely to be difficult.Brexit also created tensions over how returned EU competences will be devolved.Five of the 11 divisions in 2016-2017 concerned devolution.Other features of the Exiting the EU Committee that make unanimity more difficult include its size (21 members rather than the usual 11) and the number of parties represented (seven in 2016-2017, six in 2017-2019).
As the 2017 election produced a hung parliament, the Conservatives do not have a majority on any departmental select committees.Together, they and the DUP have a majority on the Northern Ireland and Exiting the EU committees but have rarely been able to take advantage of this.Of 68 divisions on Exiting the EU Committee reports in 2017-2019, 12 saw all Conservative and DUP members in attendance vote together-and they lost six of these.
As detailed with statistical evidence below, we find that the most significant factor explaining disagreements on the Exiting the EU Committee is the fault line between members who supported Leave in the 2016 referendum and those who voted Remain, which evolved into a divide between those favouring a harder Brexit and those supporting a softer Brexit or Remain, with the former in the minority.Of the committee's 21 members in 2017-2019, 14 campaigned for Remain: six Labour, four Conservative, two SNP, one Liberal Democrat and one Plaid Cymru (see Table 2).Five Labour MPs subsequently voted to trigger Article 50, but Stephen Timms voted against.The Conservative Remainers voted to trigger Article 50 but Jonathan Djanogly and Jeremy Lefroy rebelled on the EU (Withdrawal) Bill.Seven committee members voted Leave: six Conservatives and one DUP.The Conservatives are all members of the European Research Group (including its chair, Jacob Rees-Mogg) and supported John Baron's no deal option on the indicative votes.Note: The Chair only votes in the event of a tie.Benn has done so four times, voting with opposition parties three times, and with the Conservatives and DUP once.
The level of division in the Exiting the EU Committee can be seen by assessing the 68 votes taken on 13 reports in 2017-2019 using the Rice Index (Clausen, 1977).This is calculated by finding the absolute value of the difference in the proportion of the committee voting yes and that voting no, and multiplying the result by 100.The index can take on values between zero, where a committee is split 50:50, and 100 where all members vote the same way.As there are only 21 members on the committee, the Rice Index is sensitive to small changes in the numbers voting yes or no.Nevertheless, it provides a broad indication of the level of division.The average Rice Index value for the committee in votes in 2017-2019 is only 30.
Looking at the Rice Index by party and by referendum position, MPs who voted Leave score 98, voting cohesively on 64 of 68 divisions.Those voting Remain have an average Rice Index value of 82.Labour MPs voted entirely cohesively on all but one vote-when Chair Hilary Benn used his casting vote to side with Conservative members after a tie-giving a Rice Index average of 99, while the Conservatives were much more divided with a Rice Index average of 43.Lefroy and Djanogly were the Conservatives most likely to side with the majority but Remain-voter Richard Graham often voted with Eurosceptics.While they tend to vote cohesively, hard Brexiteers were in the majority only six times.
We tested the effect of party and Leave-versus-Remain position in a multivariate analysis of voting on Exiting the EU Committee reports.Our dependent variable was whether an MP voted with the majority or not on each vote in 2017-2019.The resulting dataset includes 1,117 cases, one for each MP present at each vote.We estimated a logistic regression model (as we have a binary dependent variable) with random effects for MPs, given they are repeated in the dataset.Coefficients from this regression are shown in Figure 1.The reference category for parties is the Conservatives.A full table of coefficients from the model is reported in the Supplementary Material.The model shows that party affects whether MPs vote with the minority or not, with Labour and SNP MPs having a greater likelihood than Conservatives of being on the winning side.The largest effect is that of each MP's position on the referendum, with Remainers more likely to be in the majority than Leavers.
Our focus now turns to Leave supporters on the committee as they have been responsible for 73% of all divisions-either by proposing amendments or triggering votes on points they objected to-since its creation.Conservative Leavers have used the committee to put their views on the record and challenge Remain positions.They voted against 12 of 13 reports in 2017-2019.On the third report (Exiting the EU Committee, 2018b), Leavers supported an alternative draft report tabled by Rees-Mogg which was more optimistic on a frictionless border between Northern Ireland and the Republic of Ireland and on a free trade agreement with the EU.Although rejected, it is included in the formal minutes (in effect, as a minority report).This was one of 26 unsuccessful amendments put forward by Leave members, covering issues such as EU law, no deal, the Irish border and the 'meaningful vote'.They also opposed (unsuccessfully) recommendations that the Article 50 period be extended if agreement is not reached (Exiting the EU Committee, 2018b, para. 86, 2018d, paras. 36 and44), the committee's 15 tests for the future relationship (Exiting the EU Committee, 2018c, para. 181), and EEA membership (ibid, para. 114).
Leavers have generally engaged with the committee-although part of the rationale for their approach is preventing it from agreeing unanimous reports.Andrea Jenkyns resigned as a Parliamentary Private Secretary to concentrate on her committee role (Jenkyns, 2018).However, the 2017-2018 attendance record of the Conservative Leavers was 65%, below the committee average of 77%.Relations between committee members have been strained.In March 2017 Conservative Leavers walked out of a committee meeting complaining that a report on the negotiations was too gloomy.Feelings were running high as the meeting occurred the day before Article 50 was triggered (D'Arcy, 2017).When Rees-Mogg tabled his alternative draft report minutes before the start of a meeting, Labour's Stephen Kinnock criticised Eurosceptics for not working 'within the collective spirit of the select committee process' (Kinnock, 2018).Rees-Mogg accused the 'high priests of Remain' on the committee of producing a partisan 'prospectus for a vassal state' and warned that 'select committee reports are only influential if they are unanimous' (Craig, 2018).
Despite the divisions, it is important to recognise that significant efforts have been made to reach consensus, with success on many issues.Of 362 conclusions and recommendations, 324 (90%) were agreed without a division.There have been notable areas of agreement.The tenth report of 2017-2019, on the Withdrawal Agreement, was agreed unanimously.There has also been unanimity on citizens' rights, including a unilateral guarantee on the rights of EU citizens' resident in the UK.

Select committee influence in the Brexit process
Divisions within committees may be damaging, but it does not necessarily follow that they produce ineffectiveness.Select committees can exert direct and indirect influence.Direct influence occurs when the government accepts and implements recommendations for policy change made in committee reports.But select committees cannot require government to accept their recommendations.
One way of measuring the policy influence of select committees is to calculate the number of recommendations made in their reports which are accepted by the government in its formal response and acted upon.Although this offers important insights, there are shortcomings.Most government responses are vague, neither accepting nor rejecting recommendations outright but pointing to action that is or will soon be underway (Hindmoor et al., 2009).A significant caveat applies when the government accepts recommendations: singling out the influence of a select committee from that of other actors is very difficult.We therefore provide evidence of where committees appear to have had some influence, but do not claim that this results only from committees' actions.Nonetheless, this approach has produced important findings on committee influence (Hindmoor et al, 2009), notably in the study by Russell and Benton (2011) whose methodology has been adapted elsewhere (Caygill, 2019) and which we adapt to evaluate reports on Brexit.
We coded the conclusions and recommendations of 16 Exiting the EU Committee reports and nine government responses, one from 2016-2017 and eight from 2017-2019. 6We also coded all primarily Brexit-related reports by other departmental select committees in the 2017-2019 session that had received a government response by April 2019.These come from 11 select committees: BEIS (5 reports); Digital, Culture, Media and Sport (DCMS) (1); EFRA (2); Foreign Affairs (1); Health and Social Care (1); Home Affairs (5); International Trade (4); Northern Ireland (3); Science and Technology (2); Treasury (1) and 6 Further details can be found in the Supplementary Material.
Welsh Affairs (1). 7The data set includes 42 reports containing 535 recommendations, 457 of which the government replied to in a total of 20 responses.
The coding scheme distinguishes between recommendations (where the committee asks for action) and conclusions (descriptive material).Recommendations were coded according to the institution they were aimed at (UK central government, the EU and its Member States, or both), the area targeted (UK government policy and negotiating position, or domestic scrutiny of Brexit), and the extent of change to government policy recommended at the time the report was published (no or small change, medium, and large change).Government responses to reports were also examined, with responses to recommendations aimed at the UK government plus the UK government and EU coded as: fully accepted; partially or implicitly accepted; neither accepted nor rejected (including part accepted but part rejected); partially or implicitly rejected; and rejected outright. 8 We discuss our findings first with regard to descriptive statistics, then via a multivariate analysis of the extent to which the government accepted recommendations.
Just over half of the paragraphs in the conclusions of Exiting the EU Committee reports were coded as recommendations, a majority of which were directed at the UK government.Most recommendations (68%) concerned government policy, with 32% concerning domestic scrutiny of the Brexit process.Other committees had a higher proportion of recommendations (63%), and more covered UK policy (94%).
Most recommendations made in reports to which the government responded were accepted: 55% for the Exiting the EU Committee and 58% for others (see Tables 3-4).This is higher than the 40% acceptance rate reported by Russell and Benton (2011).However, the Exiting the EU Committee (74%) and other committee reports (81%) include a higher proportion of recommendations asking for no change or small change compared to those examined by Russell and Benton (40%).Many recommendations accepted by the government were requests for clarity on policy.Committees often publish reports ahead of key stages in the Brexit process, with the government having provided additional detail by the time it responds.But the number of recommendations seeking further detail also indicates the difficulties committees face conducting scrutiny where information is limited.Several recommendations made by the Exiting the EU Committee (2017a) in its first report of 2016-2017 were taken up by the government.It recommended that the government publish a white paper on its negotiating strategy (para.38), seek a transition period following the UK's departure from the EU (para.163) and that parliament be given a vote on the deal agreed with the EU (para.168).

Government rejection of recommendations
The government was most likely to reject Exiting the EU Committee recommendations for medium and, especially, large-scale policy change (see Table 3).Examples include a unilateral guarantee or stand-alone deal on EU citizens' rights, ruling out no deal and parliament deciding on alternatives should the government lose the 'meaningful vote'.There is a high rejection rate (65%) for recommendations on which the committee was divided (Table 3).This is also the case for other committees' Brexit-related reports (see Table 4), 67% of which were rejected.Nevertheless, for the other committees, this is based on only nine divisions.Overall, the rejection rate for Exiting the EU Committee recommendations (37%) is higher than that for the other committees (25%), a substantial difference, although not as high as might be expected given the perception that divided committees lack influence.Part of the explanation for this is the higher proportion of large-scale changes requested by the Exiting the EU Committee (6%) compared with those from other committees (1%).Five of the committee's seven recommendations for large change provoked a division; the government rejected all seven.Nevertheless, if we also include reports for which there has not yet been a government response, the Exiting the EU Committee made roughly equal proportions of unanimously-agreed and divided recommendations for large change.With only 5% of Exiting the EU Committee recommendations asking for large change, most divisions occurred on other recommendations.
A further part of the explanation for the high rejection rate may be that divisions also occur on the final report and are not captured in our coding scheme, which considers only recommendations.Divisions on the final report indicate that some members object strongly to elements of the report.In our study, all of the reports where a division occurred on the final report also had divisions on recommendations, but this is not always the case (e.g. on the Exiting the EU Committee's 11th and 12th reports of 2017-2019, to which there had been no government response at the time of writing).Six of the nine Exiting the EU Committee reports to receive a response saw a division on the final report.The rejection rate for recommendations made in these six reports (44%) was significantly higher than for the other three (23%).By contrast, only one of the 26 Brexit-related reports from other committees to which there was a government response saw a division on the final report.
Before looking at some government responses, we assess how far the bivariate results stand up in a multivariate context, focusing particularly on our expectations that acceptance levels are likely to be lower on recommendations for large change and those on which divisions occur.Our dependent variable is the level of government acceptance, re-coded such that higher values are associated with higher levels of acceptance and ranging from 1 (rejected outright) to 5 (fully accepted).As this is an ordinal variable, we use ordered logistic regression, which relaxes the assumption required for ordinary least squares regression that points on the scale of our dependent variable are equally spaced.Our independent variables are the degree of change (measured with dummy variables for no or small change and medium change), whether a recommendation is aimed at government (as opposed to the EU or government and the EU), whether the recommendation concerns policy (rather than the Brexit process) and whether there was a vote (division) on the recommendation.Following Benton and Russell (2011) we also test whether the presence of an opposition committee chair has an effect on the success of recommendations.We include dummy variables for committees (with Home Affairs as the reference category) to assess the degree of committeespecific effects which might not be picked up by the other independent variables.
The results (see Table 5) show that only the level of policy change proposed has a reliable effect on variation in government acceptance in a multivariate context.Recommendations for no or small change and for medium levels of change are associated with higher acceptance levels than those for major policy change.This is consistent with our bivariate findings and with Benton and Russell (2011).The model shows that recommendations for no or small change are 84 per cent more likely to be partially or implicitly accepted than are those for large change.Recommendations for medium change are 39 per cent more likely to be partially or implicitly accepted than those for large change. 9The most important factor in explaining the degree to which recommendations are accepted is the magnitude of the change recommended.Once this is taken into account, we find no statistically reliable effect of divisions on the success of recommendations (the coefficient for the divisions variable is negative but not statistically significant).As with Benton and Russell's (2011) multivariate analysis, we find no effect for opposition chairs.Our measures of the actor at whom the recommendation is directed and whether it concerns policy or process, have no effect.
Recommendations rejected by the government include six of ten from the Science and Technology Committee. 10It proposed an immigration policy for 9 This is calculated using the margins command in Stata with other independent variables set at their modal values.The committee is set to Exiting the EU. 10 Additional data is provided in the Supplementary Material.
Select Committees and Brexit scientists (Science and Technology Committee, 2018b) after the government rejected its recommendation that the Migration Advisory Committee's report be brought forward (Science and Technology Committee, 2018a).Other recommendations to be rejected included those by the Home Affairs Committee on various immigration and security issues (e.g.extradition) and International Trade Committee on the appointment of the Chair of the Trade Remedies Authority.Requests for information on alternative arrangements for the Irish border and no deal planning were often dodged or refused due to concerns about disclosure.

Other areas of policy influence
Select committees added to the pressure on the government to give parliament a greater role in the Brexit process.Several examined the EU (Withdrawal) Bill with the Exiting the EU Committee and Procedure Committee (2017), a nondepartmental select committee, criticising its provisions on delegated legislation.The government accepted an amendment from Charles Walker, chair of the latter, which established the European Statutory Instruments Committee.They also pressed the government on the 'meaningful vote', proposing that amendments be made to the motion to approve the Withdrawal Agreement (Exiting the European Union Committee, 2018d;Procedure Committee, 2018).The Procedure Committee was divided on this recommendation, but the government Notes: Committee dummies were included but are not shown here, ***p < 0.01, **p < 0.05, dependent variable: level of government acceptance (higher scores indicate higher levels of acceptance).a: the likelihood ratio chi-squared test compares the log likelihood (the basis on which parameters are chosen in logistic regression -analogous to the sum of squared errors in linear regression) for the model in the table with one that includes no independent variables.When multiplied by -2, the difference between the two can be considered a chi-squared statistic.The p value of less than 0.01 for this likelihood test indicates we can reject the null hypothesis that the independent variables are equal to zero (Menard, 2002, pp. 20-22).
accepted it.Immediately after the first 'meaningful vote', the Exiting the EU Committee (2019, para. 21) recommended holding indicative votes.
Pre-legislative scrutiny offers another route for policy influence.The government was criticised for failing to include in the EU (Withdrawal) Act provisions in the Lisbon Treaty recognising animal sentience.It published a draft bill to bring this into UK law, but the EFRA Committee recommended that sentences for animal cruelty offences and animal sentience be dealt with in separate pieces of legislation (Environment, Food and Rural Affairs Committee, 2018a).The government responded that it was looking at ways to 'directly address the committee's concerns' and did not introduce the bill as drafted (Environment, Food and Rural Affairs Committee, 2018b, p. 4).

Indirect influence
Select committees also exercise indirect influence (Russell and Benton, 2011), shaping the agenda and influencing debate by producing evidence-based reports which are picked up by MPs and the media.They also scrutinise and hold the government accountable by questioning ministers and requesting information.
Select committees have helped shape the agenda and influence debate on Brexit.The absence of clear government positions encouraged committees to put forward recommendations.Opposition to no deal and support for continued regulatory alignment and close links with EU agencies after Brexit were common themes across reports (e.g.Treasury Committee, 2017;Business, Energy and Industrial Strategy Committee, 2018;Health and Social Care Committee, 2018), making it more difficult for the government to dismiss them.In some cases, committees have made differing recommendations, allowing the government to select those closest to its own position.The Treasury Committee (2017, para.150) did not, for example, suggest an extension to the Article 50 process whereas others did.
Select committees have also demonstrated an ability described as 'putting a magnifying glass on an issue'. 11The Exiting the EU Committee has tracked the UK-EU negotiations, explored models for the future relationship, and looked indepth at citizens' rights and data protection.Inquiries by it and the Science and Technology Committee drew attention to the cost of a UK equivalent to the Galileo satellite navigation system.Committee reports also provide evidence that MPs draw upon.The BEIS Committee's report on Brexit and the automotive industry (Business, Energy and Industrial Strategy Committee, 2018), was cited by MPs with car manufacturers in their constituencies in parliamentary debates (House of Commons, 2018).
Finally, select committees have enhanced the scrutiny of government and held ministers to account.In October 2017, Davis told the Exiting the EU Committee that a deal with the EU could be signed before parliament held a vote, prompting an urgent question in the Commons.Two months later, he endured a tough session on the government's economic analysis.The divisions on the committee were apparent in the questioning of witnesses, with Leave-supporting members challenging senior civil servants on the perceived costs of a no deal Brexit.This does not make the committee any less effective-indeed, scrutiny has arguably been enhanced because policy has been probed robustly by both Remain-and Leave-supporting MPs.Committee chairs have followed up issues raised in evidence sessions in letters to ministers, requiring the government to provide additional information.Although a useful tool, these are no substitute for proper government engagement with reports.Following Labour's 'motion for a return' humble addresses, the Exiting the EU Committee (2017bCommittee ( , 2018a) ) published the government's sectoral impact assessments and EU exit economic analysis.The government withheld some content and the Conservatives and DUP won committee votes on removing sections.Nevertheless, these are examples of parliament bringing information into the public domain which would not otherwise have been available.

Conclusions
Brexit has proved a highly divisive issue on the floor of the House of Commons.This divisiveness has not, however, infected select committees to the same extent.The proportion of committee reports seeing a formal vote (division) has not increased and many Brexit-related reports have been agreed unanimously.The norm of consensus-seeking remains strong, and select committees are still the prime example of cross-party working in an adversarial House.The Exiting the EU Committee is an important exception-it is the most divided of committees, with all but one of its reports seeing divisions.The key fault line is that between Remainers and Leavers, and relations between the two have been acrimonious.Yet the Committee has reached consensus on key issues and on most of its recommendations.
The belief that unanimously agreed reports are more likely to influence government than ones on which a committee has been divided is prevalent.Our bivariate analysis suggests that recommendations on which a committee divided are more likely to be rejected -but this finding did not hold up in a multivariate context where the level of policy change was taken into account.The most important factor explaining government rejection of committee recommendations is the magnitude of the change being proposed (with requests for large scale change much more likely to be rejected) rather than whether a recommendation was agreed unanimously.
Our analysis provides contrasting evidence for the claim that divisions harm committee influence.On the one hand, the rejection rate is higher for recommendations on which the Exiting the EU Committee was divided, although much of this is explained by the propensity for the committee to divide on proposals for large scale policy change in those reports to which there has been a government response.Where the committee divided on the final report, the rejection rate is higher than in cases where it did not.On the other hand, the acceptance rate for Exiting the EU Committee recommendations is little different from that for Brexit-related reports agreed unanimously by other committees.This suggests that the polarisation on the Exiting the EU Committee has not undermined significantly its capacity for influence.Compared to the Exiting the EU Committee (8%), other committees also see a higher proportion of recommendations that were neither accepted nor rejected (16%).
Select committees have had some influence during the Brexit process.They have produced speedy but informed analysis of issues, shaping debate, highlighting areas of concern and alternative courses of action, and bringing evidence into the public domain.We have shown that most committee recommendations have been accepted partially or implicitly by the government.But this finding comes with caveats.Recommendations were accepted to a greater degree when they involved no/small change or medium levels of change, and the government may have already intended to take this action.Pressures from other sources, such as the threat of backbench rebellion, may also have been significant.While committees have examined proposals and emerging issues, the lack of clear policy and access to information have complicated scrutiny activities.Committee chairs have criticised the government for failing to engage with their reports.As the Brexit process is unfinished, we have not measured the extent to which recommendations have been implemented by government.This could be an area for future research, including assessing how far recommendations initially rejected are later acted upon (Benton and Russell, 2013).

Figure 1 .
Figure 1.Factors affecting likelihood of being on winning side in votes on the Exiting the EU Committee.Note: Coefficients are from a logistic regression (with random effects for 21 MPs) of being on the winning side in votes on the Exiting the EU Committee.The bars show 95% confidence intervals around the coefficients.The reference category for parties is the Conservatives.

Table 1 .
Select Committee composition on the Leave-Remain divide, 2016-2019 Note: Five MPs who served on committees did not disclose how they voted in the referendum.

Table 3 .
Government acceptance of Exiting the EU Committee recommendations

Table 5 .
Factors affecting government acceptance of select committee recommendations (ordered logistic regression)