The credibility of finance committees and information usage: trustworthy to whom?

IMPACT Local government politicians in Brazil were found to perceive the reports and referrals from legislative finance committees as trustworthy depending on the political scenario in which the report was generated. Information usage by politicians then depended on partisan demands—whether they were in opposition or in government. The quality of the accounting information delivered by finance committees needs to be improved and councillors should be encouraged to consider and use financial information more widely: not just in debates in council. Council finance committees could usefully include external specialists to validate referrals and tighten ties with audit institutions.


Introduction
Information is 'an essential tool for legislators in completing their work' (Feldman, 1987), but how information and legislators interact, supplying and demanding information, is not an easy question to answer. The literature on the use of accounting information by politicians acknowledges that this information is critical to decision-making (van Helden, 2016). Scholars usually debate about the influence of the information content (Hyndman, 2016), the context (Giacomini et al., 2016), psychological factors (George et al., 2020), and the characteristics of the decision-maker (Reichard, 2016) on the potential use of such information.
The academic literature has extensively discussed the legislative deliberative process-in which legislators debate and vote on given issues-as part of sensemaking and bargaining processes. Legislators frequently aim to exhort their peers to a similar understanding (Ringe et al., 2017), usually selecting information to make sense of 'ambiguous and often contradictory indicators' (Maxwell, 2005). Although democracies expect legislators to use available information to support their decisions and accountability through the deliberative process, including but not limited to accounting information, they might use the information with political objectives and influence public opinion (Guarini, 2016), and not necessarily for the common public good (van der Voort et al., 2019;Nielsen & Pedersen, 2014). Legislators might strategically use the information under numerous party pressures-much like any other politician (Ringe et al., 2017). Similarly, the incumbent government usually counts on a network of supporting parties, information and resources, that is often not accessible to legislators from opposition parties (Stapenhurst et al., 2006). Therefore, politicians from opposition parties, particularly legislators, might find it difficult to gather information to contest the incumbent government when there is a partisan climate with divergent supporting and opposition networks (Raile et al., 2010;Junior et al., 2015).
In this article, attending the call to simultaneously consider the demand and supply sides of information (Ouda & Klischewski, 2019), we analyse how municipal council members use supplied information according to how credible they believe the source of such information to be. From a supply perspective, legislative finance committees (henceforth finance committees, or simply committees) are sources of information that deeply influence how legislators make sense about given issues to take a position in the debate, supporting, proposing amendments and voting in council (Gaines et al., 2019). Finance committees are technical advisory committees composed of councillors that might navigate fellow council members (non-members of the committee) by providing a comprehensive view of the information or intending to influence the debate (Jorge et al., 2016). From a demand perspective, scholars acknowledge that motivations for information usage might differ according to the type and intensity of information . For instance, legislators may trust and use accounting information if they perceive that the suppliers of information are like minded (Ouda & Klischewski, 2019).
This article analyses the extent to which municipal council members in Brazil use accounting information depending on the credibility of the information source (in our case, finance committees). Our main argument is that the trustworthiness and consequent usage of accounting information depends on contextual features, like other sorts of relevant information and political or social pressures. Finance committees would be one relevant source to inform council members on accounting (including financial) issues, providing information on expenditures, debts, assets and debating complex financial issues.
We drew on the concept of information trustworthiness from the information science literature (Hovland et al., 1953;Lucassen & Schraagen, 2011 to better capture whether the supplied information (the opinions given in finance committees' referrals) carries and is affected by the credibility of its source (the finance committees themselves). Individuals generally use information only if the source is perceived as credible and trustworthy (Lucassen & Schraagen, 2011. In the analysed context, we found that a councillor's perception of the credibility of the referrals prepared by finance committees varied according to the political position such committees occupy in municipal councils regarding the incumbent government. We sent an electronic survey to municipal council members in Brazil. The Brazilian political dynamic follows a strong mayor model, combined with a fragmented multiparty system (Ribeiro & Fabre, 2019;Zucco & Power, 2020). This system often requires the mayor to maintain a coalition to govern through allocating regional funds and jobs to their allies. Therefore, the presence of a powerful coalition influences the deliberative process and the finance committees' alignment with the mayoral agenda. With a powerful coalition in place, finance committees might be perceived as having been captured by the coalition. We investigated how credibility influences information usage in different political scenarios. Our findings show that credibility is a key ingredient enforcing information usage, but its effect might depend on the political scenario.
This article is structured as follows: first, we draw from the previous literature to propose four political scenarios considering whether councillors would use financial information supplied by the finance committee. Second, we describe our empirical setting, data and methods. Third, we explain our results and analysis. The final section presents our discussions and conclusions.
Source credibility: the legislative committees as (un)credible sources Legislators both demand and supply information throughout the deliberative process (Ouda & Klischewski, 2019). In this article, we investigate finance committees as a potential source of information (acting as suppliers) and how their fellow council members (acting as users on the demand side) perceive their finance committees as credible sources. We investigated whether councillors might be more likely to use the information provided by the finance committee depending on their own political positioning (either in supporting or opposition parties), as well on the mayor's power in a coalition (either weak or powerful).
Finance committees, present in both emerging and established democracies, are sources of information that deeply influence debate among legislators (Gaines et al., 2019). These committees are technical advisory bodies that supply information to fellow legislators-in other words, they act as information brokers (De Vrieze, 2020;Jorge et al., 2016). They are expected to navigate legislators through sometimes very complex issues (Hay & Cordery, 2020;Yoshinaka et al., 2010). Their main deliveries as concrete supporting actions are referrals on a given topic (Barrett, 2018), which set up bargaining and debate in the deliberative process (Yoshinaka et al., 2010). Referrals are written pieces supplied by these committees; topics range from the very abstract and complex to the daily demands faced by councillors. For example, referrals might refer to such complex issues as the actuarial balance of pension funds, or to such straightforward demands as the impact opening a new hospital will have on current expenditure. Referrals might also guide fellow council members to better understand audit reports (Hay & Cordery, 2020;Yoshinaka et al., 2010).
Furthermore, finance committees, like other legislative committees, include a chairperson, a rapporteur, and other legislators as regular members. The rapporteur is a key member who drafts such referrals to be jointly enacted by the committee's members. Although rapporteurs have no direct parallel in the Anglo-American legislative systems, they are central in the European Parliament and many other legislatures in Europe and worldwide (Yoshinaka et al., 2010). Rapporteurs in the analysed case are designated by the committee chair to offer an initial understanding of the issue and how the incumbent legislature should react to it. Previous studies show that the rapporteur's opinion expressed in a referral frames the early agreements on the issue (for example Costello & Thomson, 2010).
Yet, referral usage by fellow council members should not be considered as given. Supplied information is not necessarily demanded (nor actually used) information. Overall, information usage depends on how credible the source is perceived (Lucassen & Schraagen, 2011. Prior studies on information science show that individuals decide to rely on supplied information once they are perceived as forthright, truthful, and honest (Hovland et al., 1953;Lucassen & Schraagen, 2011. If potential users perceive any sort of undesired biases on the supplied information, they negatively judge the source credibility, no longer considering the arguments in the message, the source qualifications, nor other effortful methods of information evaluation (Flanagin et al., 2018).
As the sources and their supplied information are usually perceived as tightly connected (Hovland et al., 1953;Lucassen & Schraagen, 2011), whether the source of information is perceived as credible influences one's propensity to use it (Hilligoss & Rieh, 2008). The committee's referral (supplied information) would be directly associated by fellow council members to the finance committee which prepared the referral. Therefore, referral usage would depend on the finance committee's credibility. Rapporteurs and the committee chairs, key role holders at the committees, are also regular legislators with a political background like any other legislator. If they lack legitimacy as members, then committees lack credibility as a source of information and legislators will doubt that the referrals are free from their political preferences (Stapenhurst et al., 2006). Indeed, Park (2017) has shown that political polarization can undermine the informativeness of committees.
Legislators also evaluate a source as credible or not depending on their own judgement and prior experiences. For instance, individuals evaluate whether a source is credible considering their domain expertise and prior source experience (Lucassen & Schraagen, 2011), and often rely on simple rules of thumb (i.e. surface heuristics) or systematic processing (Hilligoss & Rieh, 2008). Further, individuals often prefer messages which provide several perspectives on a matter (two-sided messages) to decide on their own, rather than one-sided message perspectives (Flanagin et al., 2018;Mayweg-Paus & Jucks, 2018). Yet, legislators might consider the one-sided perspectives of partisan groups more credible, aiming to support their beliefs and confirming their pre-existing attitudes (Metzger et al., 2020). Politicians, in general, might search sources of information to attack what they consider to be inadequate practices or choices by those they are already predisposed to dislike (Moynihan, 2016).
Considering these intertwined information supply and demand features, one cannot ignore how political scenarios delineate the flow of information among legislators. Whether finance committees are captured by the mayor's coalition or not influences whether they are perceived as trustworthy by the fellow council members who support or oppose such a coalition. On the one hand, these committees can promote consensus among councillors regarding a technical debate (Yoshinaka et al., 2010;Jorge et al., 2016), as they sustain a nonpartisan climate, involve the public and encourage media coverage (Stapenhurst et al., 2006). On the other hand, committees might be driven by a sort of individual interest. They can pursue party goals (Bussu, 2015;Karlsson, 2013), electoral ward claims (Ritchie & You, 2019), and intend to influence the debate instead of providing a comprehensive view of information (Jorge et al., 2016). Eventually, instead of sources of information, these committees might become an undesirable or constrained voice at the deliberative process by political interests . The dominant coalition chairing in the council presidency may use formal rules to constrain the committee's daily work (De Vrieze, 2020), reducing its resources and mandate (Domingos & Aquino, 2019;Gaylord, 2010). Complicating matters, council members are usually laypeople who are not specialists in accounting (Bussu, 2015;Murphy, 2020). A powerful coalition might therefore shut down the already fragile capacity of finance committees. Figure 1 summarizes our propositions based on the literature. As information users, legislators seek information to cope with different objectives, depending on their political position regarding the incumbent government. Therefore, source credibility would count differently depending on the political scenario in place. In our analysis, we articulate the concept of source credibility specifically within the context of a strong mayor model and fragmented multi-party system (Ribeiro & Fabre, 2019;Zucco & Power, 2020). In a political system like Brazil's, the mayor counts on a set of parties to support their coalition.
We considered two features of councillors' positions regarding the incumbent government: mayor's coalition power and a councillor's political position. First, the mayor's coalition power connects the city hall to the municipal council. These coalition ties are not restricted to the main hall where voting takes place but expand to councillors' offices where alliances and support for the government's agenda are constructed. The councillors in the coalition channel a flow of information and interests, eventually connecting people. For instance, council members are often invited in Brazil, after their term is up, to take a seat at the mayor's cabinet as secretaries, managers, or advisors. However, the coalition power varies according to the municipality and the mandate due to local political dynamics. For simplicity, we considered in our political scenarios that the mayor's coalition might be a weak coalition or a powerful coalition. Second, councillors who support the coalition (from the supporting parties) walk around city hall offices, access the mayor's political network, and gain benefits as they orbit the political agenda of the incumbent government. Conversely, councillors who oppose the coalition (from the opposition parties) are not in the mayor's cabinet and find it hard get access to confidential information.
In Figure 1, scenario 1 refers to the council members from the supporting parties in a legislature of the municipal council in which there is a weak coalition in place. Scenario 2 refers to a council member from the opposition parties, but still facing a weak coalition. When facing a weak coalition, councillors from supporting or opposition parties would rely on a credible committee to elucidate questions and inform themselves. However, if a powerful coalition is in place, then the finance committee could be captured, such that supporting party councillors might use the committee's referrals and opinions to support the coalition agenda (scenario 3: councillors in supporting parties, with a powerful coalition in place). Conversely, the council members from the opposition may doubt that the rapporteur just supports and justifies previous positions from the ruling coalition. If so, those contrary to the ruling coalition would use the committee's information to be aware of the mayor's coalition agenda, but only if they consider the committee as a credible source (scenario 4: councillors in opposition parties, with a powerful coalition in place).
Empirical setting: austerity and political struggles The Brazilian political system at the municipal level is typically the case of a strong mayor model and an extremely fragmented party system (Ribeiro & Fabre, 2019;Zucco & Power, 2020). Such systems translate much of the federal level legislation to the municipal level of government (i.e. symmetric federalism), despite a low congruence between national and subnational party systems in elections (Ribeiro & Borges, 2020). This setting is described in the literature as 'coalitional presidentialism' (Junior et al., 2015;Chaisty et al., 2018). In this context, federal, state, and municipal governments overlap jurisdictions to implement public policies, with some degree of fiscal decentralization.
This coalition presidentialism at the municipal level regulates the municipal executive-legislative dynamics. Mayors rely on different mechanisms to establish legislative support in a fragmented multiparty legislature (Chaisty et al., 2012, p. 3), for example by offering jobs to council members from the supporting parties in their cabinet after that councillor's term ends (Raile et al., 2010;Junior et al., 2015). Although the oversight of the Brazilian governments has increased over the past 20 years due to the Fiscal Responsibility Law, public sector reforms did not bring the expected benefits (Aquino & Batley, 2021). Oversight still depends on the autonomous but politically biased regional courts of accounts (Lino et al., 2021;Lino & Aquino, 2020). Frequently, the courts of accounts, governors, and municipal councils reduce the oversight on fiscal rules in a sort of shared political protective behaviour (Lino et al., 2021;Lino & Aquino, 2020;Melo et al., 2014).
Finance committees, which are mandatory for every municipal council in Brazil, are meant to guide fellow council members through complex accounting issues, such as actuarial issues of pension schemes, carry-over debt, fiscal risks, and potential remedies presented in the annual budgetary guidelines. They can also play a more proactive role, for instance collecting information from the municipal agency running the government pension fund and inviting specialists to discuss the issue. The Brazilian municipal councils vary from nine to 50 council members, according to the number of inhabitants in the municipality. They are often laypeople, elected for fouryear terms. The council members vote on a myriad of topics and can simultaneously be a candidate for the council presidency and council committees (including finance committee).
The composition of the committees will reflect the dominant political coalition, as party leaders appoint the members of the committee. Committee members themselves elect the people with the two most valuable committee roles-the chairperson and the rapporteur. However, as the committee composition reflects the coalition in place, this influences the committee political orientation. In the end, fellow councillors might consider the committees as having being captured by the ruling coalition, further influencing how credible councillors perceive the committees to be (Domingos & Aquino, 2019).
At the time of our survey (2017 to 2018), there was strong national debate on fiscal sustainability taking place in Brazil. It was part of an agenda of fiscal austerity imposed by President Temer. The debates included the influence of politicians on civil servants' pension funds (Schettini & Terra, 2019;Lima & Aquino, 2019) and whether this had contributed to the financial crisis. Accounting and public finance specialists did not reach a consensus on whether the public pension funds were the main cause of the poor financial performance; the subject saw a lot of media coverage in 2017 (Silva et al., 2017).

Data and methods
Our survey-based empirical setting arises from a research collaboration between the National Association of Brazilian Municipalities (Confederação Nacional de Municípios in Portuguese) and the University of São Paulo. Our electronic survey was sent to municipal council members all over the country. The questionnaire was composed of two parts: the first capturing the councillors' preferences on how a local pension fund should be adopted and sustained, and the second one about whether they trusted and use the information supplied by the finance committees. We received 810 valid answers covering both parts of the questionnaire. As the questionnaire focused on the national debate about public pension funds and the austerity agenda, probably our respondents were those councillors involved in these issues.
The survey collected personal information, such as political party affiliation, accumulated mandates, age, schooling, gender and self-reported interest in finance and public policies as education and health. Five-point Likert questions identified their deliberative behaviour and perception about the finance committees and the partisan climate at the municipal council. We also asked about the roles respondents had: committee member, committee chair, or council chair. The main questions, variables, and descriptive statistics are shown in Table 1. The analysed empirical setting focused on finance committees. The finance committees were accountable for a variety of accounting and budgetary issues, for instance the fiscal and budgetary impact of civil servants' pension funds. Finance committee members supply referrals to support their fellow councillors' voting on complex financial issues. Our analysis only considered the subsample of respondents who were not members of a finance committee, i.e. the people who would potentially use referrals.
Next, we briefly explain our proxies for the main concepts in our proposal (as previously described in Figure 1). Referral usage is the councillors' propensity to use the referral supplied by their peers from the finance committees ('You use finance committee referral to inform yourself', five-point Likert). However, as politicians' perceived usefulness of financial information is often higher than their actual use (van Helden, 2016), the survey might be overestimating the potential use of such information by municipal councillors. Our estimates might be an optimistic perspective or the upper limit of information usage (Battaglio et al., 2019).
The credibility attributed to the supplier of information (source credibility) was proxied by the respondents' perception of the committee (we reduced five measures to a factor, see Table 1 for further details regarding factoring methodologies and measures 1 to 5). We drew from the information science literature with regards to information trustworthiness (Hovland et al., 1953;Lucassen & Schraagen, 2011 to better capture the importance of source credibility. However, this literature is not specific to the legislative deliberative process. To the best of our knowledge, there is no validated measure of source credibility in such a context. Our five measures followed prior studies that acknowledge that finance committees are more credible when they have the power to make recommendations (measure 1: 'The finance committee always expresses its opinion during plenary sections'); publish conclusions (measure 2: 'The finance committee publishes its agenda and minutes of meetings on the internet', and measure 5: 'Your fellow councillors consider the finance committee referral as an important delivery'); engage the general audience (measure 3: 'The finance committee meetings are open to the general public and the agenda is announced in advance'); and encourage media coverage (measure 4: 'The finance committee meetings are broadcast over the internet or on TV') (Stapenhurst et al., 2006).
The power of the mayor's coalition in place (mayor's coalition power) and the political position of the council member in relation to that coalition (councillor's political position) were proxied also by five-point Likert items. First, regarding the coalition, respondents considered whether they agree with: 'the mayor's cabinet requests are always attended, such that the legislative has a minor role in oversighting the mayor', and 'a few times the councillors influenced the mayor, but often the mayor influences the councillors'. We considered these two measures separately (the first one is our main measure and the other is an alternative measure) instead of relying on a common factor as their answers were not highly correlated (Cronbach alpha 0.60). Second, regarding their political position, respondents considered whether they 'are part of the opposition to the current mayor'.
Before moving to the main analysis, note that the sample comprised respondents with high literacy and willingness to discuss complex issues, for example public pension funds. According to the Brazilian official electoral statistics, fewer than 30% of Brazilian councillors have a college degree and, in our sample, 67% had a college degree. As any layperson can stand as a candidate in a Brazil, possibly, if not probably, these features would stand for a potential upper limit bias for the usage of information in our sample.
Our main analysis was based on differences in means of referral usage according to source credibility per political scenario. Our main proposition was that the political position occupied by the municipal councillors matters for how they interact with the information supplied by the finance committee. The use of committee information (referral usage) depends on whether the council members perceive the committee as a credible source (source Mayor's coalition power Two alternative 5-point Likert measures are presented below. Likert responses of 'strongly disagree' or 'disagree' indicated the presence of a weak coalition, and 'strongly agree' or 'agree' indicated the presence of a powerful coalition. We dropped respondents who 'neither agree nor disagree'. Coalition power (measure 1) 5-point Likert: 'A few times the councillors influenced the mayor, but often the mayor influences the councillors'. This was used as the main measure of coalition power.
3.36 1.75 1 5 Coalition power (measure 2) 5-point Likert: 'As mayoral requests are always attended to, councillors have a minor role in terms of oversight of the mayor'. This was used as an alternative measure of coalition power.
3.13 1.74 1 5 Councillor's political position 5-point Likert: 'You are part of the opposition to the current mayor'. Likert responses of 'strongly disagree' (N = 194) or 'disagree' (N = 24) indicated the respondent was in a supporting party, and 'strongly agree' (N = 190) or 'agree' (N = 27) indicated the respondent was in an opposition party. We dropped respondents who 'neither agree nor disagree'.
2.99 1.91 1 5 credibility), according to the councillor's political position (supporting or opposition) and the mayor's coalition power (powerful or weak). For simplicity, we split our sample of council members into those who considered the finance committees as 'low credibility' and 'high credibility' according to the source credibility median. As the councillors' motivations for information use will differ according to the political scenario in which they are operating, we investigated the relationship per political scenario outlined in Figure 1 (considering the mayor's coalition and the councillor's political position). Table 1 provides details on how we determined each position.

Analysis: source credibility and information usage
We tested the difference in means of referral usage according to the source credibility for the four proposed political scenarios. Figure 2(a) outlines the means of referral usage comparing those who considered the finance committee as more or less credible (above or below the median) according to the political scenarios. Figure 2(b) outlines the difference in means in each scenario. The council members who considered their committees as credible presented higher means in terms of their likelihood of using a committee referral in scenarios one (in supporting parties, weak coalition in place), three (in supporting parties, powerful coalition in place), and four (in opposition parties, powerful coalition in place). Although we expected that higher source credibility would be associated with higher information usage, we think that the underlying motivations for usage would be different in each scenario. In scenario one (the graph in the top left of Figure 2[a]), our findings suggest that councillors in supporting parties in a weak coalition were most likely to use a committee referral when they considered the committee to be a credible source. However, councillors in supporting parties facing a powerful coalition had different usage criteria. Comparing the two scenarios, the supporting party councillors (scenarios one and three) arguably had better access to the mayor's cabinet and political agenda than the opposition councillors (Raile et al., 2010;Junior et al., 2015). If the finance committee is not a credible source, supporting party councillors might simply rely on alternative sources of information from the mayor's cabinet. This would hold for supporting party councillors facing both a weak coalition (scenario one) and a powerful one (scenario three), as politicians are usually overloaded with information (Walgrave & Dejaeghere, 2017).
Alternatively, opposition parties facing a powerful coalition want information in order to understand the coalition's agenda (scenario 4, bottom right). The credibility of the committee was again associated with higher levels of referral usage.
Finally, we expected that opposition councillors facing a weak coalition (scenario 2, top right), would use a referral to inform themselves and as the basis for asking questions. We found that councillors in this scenario were highly likely to use a referral, regardless of their perception of source credibility. The difference in means for referral usage according to source credibility was significant in all scenarios (see Figure 2[b]), other than scenario 2. Source credibility only affected use of a referral by the opposition if they had to question the coalition. Hence, the credibility of the source was not significant for information usage in that political scenario.
Further analyses corroborated our main propositions regarding the context-dependent usage of the referral and provided us with confidence regarding our measures. All findings were the same when different tests were used. For Figure 2(a). Means of referral usage according to committee credibility.
Notes: The figure represents means of referral usage with the respective confidence intervals (95%), according to the level of source credibility (low versus high). The analysis considers only non-members of the financial committee, following our proposal outlined in Figure 1. See Table 1 for details of the measures used.
instance, we used an alternative measure for mayor's coalition power and the results were the same. We also ran sensitivity analyses for our councillor's political position measure, segregating council members into opposition or supporting parties; again, our results were maintained. See Table 1 for further details on our alternative measures.
In addition, our findings were maintained when we ran simple or multivariate OLS regressions instead of analysing the difference in means. Interestingly, these regressions showed that individuals' features (such as education, prior experience, and self-confidence) did not relate to the likelihood of councillors using financial information.
Finally, we ran additional tests on our subsample of finance committee members to validate our measure of source credibility. Initially, we did not use these respondents in our sample, as they were both suppliers and users of the information. As robustness checks, we ran OLS regressions looking for correlations between source credibility and measures of committee members' perceptions regarding their own activity within the finance committee. For instance, we asked them whether their finance committee played an active role in monitoring expenditure and debt, inviting specialists to debate complex issues, and hosting public hearings. Such initiatives would be expected to be features of a credible and trustworthy committee (Stapenhurst et al., 2006). We found that committees that often hosted public hearings were considered highly credible by committee member themselves. The additional analyses proved that our measure of source credibility accurately captured the perceptions of councillors regarding the trustworthiness and legitimacy of committees.

Discussion and conclusions
Political scenarios influence individual councillors' motivations for information usage and their likelihood of using supplied information. We investigated the role of source credibility (Lucassen & Schraagen, 2011) in influencing the use of both financial and non-financial information. We focused on information from finance committees (De Vrieze, 2020; Jorge et al., 2016), whose members are councillors who supply information to their fellow council members (non-members of the committee), rather than an external accounting team supplying financial information. Generally, whether such referrals were considered as useful or not depended on the perceived credibility of the committee.
In our analysis of municipal councils in Brazil, we found that coalition power and political position strongly influenced the perceived credibility of the finance committees and the information they were providing. Source credibility should influence how legislators select information in an information-overloaded setting as they debate and vote on complex issues (Walgrave & Dejaeghere, 2017). Our results suggest that, in the context of weak coalitions, the council members may have extra room to bargain and influence decisions in the deliberative process. Hence, the referral by the finance committee is a valuable piece of information if councillors perceive the committee to be a credible source. We found that not all councillors perceived credibility in the same way-it depended on their political positions in the coalition. Opposition councillors were more likely to use information from committees they perceived as credible, but mainly to understand the coalition's agenda. On the other hand, supporting party councillors used information from credible committees to ask questions in a weak coalition or to support a powerful coalition's choices.
Councillors' perceptions of the credibility of a finance committee's information were, of course, blended with their political interests. As well, usage was dependent on citizen demands, for example, in our case, civil servants' pension funds.
Following our analysis, we see the need for two avenues of future research. First, rather than using such generic and abstract terms as 'decision-making' that are not helpful in understanding of how politicians interact with information in their daily work, we need to be more precise. Instead of looking at 'the use of accounting information to decisionmaking', we need to understand the dynamics of collective political decisions. We need to understand the use of bargaining within the deliberative process, and the role of previous experiences and (in)accurate claims (Ringe et al., 2017;Yoshinaka et al., 2010).
Second, rather than considering the use of an isolated set of information (for example accounting), we should recognize Notes: The figure presents the plots for the difference in means for the referral usage according to source (low versus high committee credibility). The thick and thin lines represent, respectively, the corresponding 90% and 95% confidence intervals. Crossing the horizontal axis (0) means the difference for referral usage is not statistically significant. The difference for referral usage is significant at the 95% levels for all scenarios except scenario 2. that politicians may use multiple sources of information. Accounting information is just part of the set of information politicians consider. For example, politicians might value a piece of information if it helps them to make sense of a new issue (Maxwell, 2005), if it supports and legitimizes a previously selected choice (Moynihan, 2016), or even if it is useful in misleading others (Ringe et al., 2017). Credibility might play an important but different role, depending on the political scenario.
In terms of policy implications, our findings suggest that improving the quality of accounting information delivered by finance committees is necessary but not the whole solution to improving the political debate in a representative democracy. When dealing with a fragmented political system, similar to the Brazilian one, legislators might perceive finance committees as intermediaries of the incumbent coalition, instead of nonpartisan information sources. Initiatives to prevent a coalition from capturing committees include bringing in external professionals to validate referrals prepared by the committee. Complementary governance mechanisms should enforce the committees as nonpartisan sources. For instance, committees should announce their agenda as early as possible, open and broadcast meetings to the public, and provide meeting minutes on request. These measures will engage a broader audience in committee activities and help to prevent political parties from misusing committees.

Disclosure statement
No potential conflict of interest was reported by the author(s).