University of Groningen Literature review and challenging research agenda on politicians’ use of accounting information

Recent research on politicians’ use of financial and performance information is reviewed. Survey-based studies overestimate the frequency of this use; observational studies present a more accurate picture. A new and challenging research agenda is presented that will improve our understanding of the use and usefulness of accounting information. Implications for the real world of practice are discussed.

In a review paper about public sector performance management research, Pollitt (2006) warned us 10 years ago that politicians were the missing link in this field. Although some studies about politicians' use of financial and performance information-or, more generally, accounting information-have been published, the problem is still largely under-researched. In order to promote research in this domain, this essay has two main aims. First, I present a critical reflection on recent academic studies in the field. Second, I suggest some ideas for a challenging research agenda. The essay ends with some practical implications.

A critical review of politicians' use of accounting information
This section analyses four studies: two about politicians' use of financial information; and two about this group's use of performance information. My goal is to understand why the studies show diverging findings and to see what we can learn from them for future research. First, an overview of the research over the past decade is presented.

Overview of the research
Research on politicians' use of financial/ performance information and its perceived usefulness has emerged over the past decade in international accounting and public administration journals.
Financial information typically concerns budgetary issues, such as appropriations and the expected costs of certain services, or accrualbased issues, such as the depreciation of fixed assets and equity relative to debt (van Helden and Hodges, 2015, chapters 3 and 6). Studies by Liguori et al. (2012) about Italian aldermen point to a high appreciation of cash-based budgeting and accrual-based accounting elements; whereas Buylen's (2014) examination of Flemish councillors indicates a very low use of this information in budgetary debates. The latter finding also applies to Hyndman's (in Ezzamel et al., 2008) study of the use of accrual-based budgeting and accounting information among MPs in the Northern Ireland (NI) Assembly. Yamamoto's (2008) investigation of financial information use in Japanese government shows different findings depending on the governmental level; use was very limited at the central level and there was moderate use at the local level.
Performance information generally entails the quantity and quality of government services and activities, or specific aspects of a transformation process, such as efficiency (for example cost per unit of service) and effectiveness (the outcomes related to the goals). Empirical studies on performance information appreciation and use cover a broad spectrum of countries and levels of government. There is, however, a lack of consensus in the findings. Low levels of appreciation and use are found in studies conducted at the central government level in the UK (Johnson and Talbot, 2007); at the state level (Bourdeaux, 2008) and the federal level (Stalebrink and Frisco, 2011) in the USA; in Estonia's central government (Raudla, 2012); in Dutch municipalities (ter Bogt, 2004); and in German and Italian local governments (Grossi et al., 2016). Other studies, however, show opposite, more 'positive' findings, especially Askim (2007) for Norwegian municipalities; Brun and Siegel (2006) for the Swiss central and intermediate governments; Saliterer and Korac (2013) for municipalities in Austria; and Ho (2006) for local governments in the USA's midwest. In addition, some studies show mixed findings, particularly © 2016 CIPFA ter Bogt et al. (2015) for intermediate government in the Netherlands.
So why are the findings so different? Before answering this question, we need to look at the contexts that contribute to the usefulness and use of accounting information. Although there seems to be little consistency in the above studies, some pointers can be given: the quality of the accounting information matters (Ho, 2006;Grossi et al., 2016); the availability of budgetary powers is an enabler for accounting information use (Bourdeaux, 2008;Raudla, 2012); and cash information is preferred to accrual information (Yamamoto, 2008;Liguori et al., 2012;Buylen, 2014). Liguori et al. (2012) investigated the extent to which politicians and managers in Italian local government appreciated three types of partly NPM-related accounting elements, i.e. budgetary cash-based information, accrual information and non-financial performance information. They obtained their data from a survey of aldermen. Buylen (2014; see also Buylen and Christiaens, 2016) studied the same financial information items among Flemish local government councillors, based on observations about the extent to which this type of information was actually used in municipal councils' annual budgetary meetings. A comparison between both studies reveals astonishing differences (see table  1). Aldermen in Italy highly appreciate budgetary cash-based accounting information and they 'approve' of accrual accounting information. In Belgium, however, the actual use of these types of information by Flemish councillors is very low for budgetary items and almost absent for accrual items. Why do these differences occur?

Financial information use
The most likely explanation is based on a combination of two factors: the way in which usefulness and use are measured; and the type of data collection method applied. The Italian study investigates appreciation or usefulness, while the Flemish study examines use. Moreover, the Italian study is based on a survey and the Flemish study on the observation of actual behaviour. Measuring the usefulness of financial items to politicians-or, in other words, their appreciation of them-through a survey significantly overestimates the actual use of these items. This has to do with the risk that surveys can be biased towards desirable behaviour, which especially applies to more or less technical concepts. Moreover, appreciation of a certain type of information does not automatically lead to its use. To start using an item, people need to be triggered to do so. This analysis shows that: •Politicians' appreciation or perceived usefulness of financial items is much higher than their actual use. •In comparison with observational studies, surveys overestimate the potential use of financial items.
These conclusions were corroborated by a recent *This study also investigated the importance politicians attached to non-financial information and managers' appreciation of the same set of financial and non-financial items (not presented here).
© 2016 CIPFA study about improved budgetary information for politicians in a Dutch province (ter Bogt et al., 2015). Here, a new performance budget was found to be highly appreciated, but its actual use was minimal. Research about the actual use of financial information by politicians is scarce. Noteworthy is a study by Ezzamel et al. (2008) about accounting innovations resulting from devolvement projects in the UK, especially the findings of Hyndman about the limited use of accrual accounting information by politicians in the NI Assembly, which are based on case/field research. In addition, Yamamoto's (2008) survey study about Japanese government showed a limited use of financial information at the central level, while at the local level use was higher. Case/ field research, which mostly relies on interviews and documentary studies, enables indepth investigations about the way in which accounting documents are used. However, observations of actual behaviour are often lacking, which possibly places case/field research between surveys and observational studies in accurately depicting politicians' accounting information use.

Performance information use
For the purposes of this essay, two important studies about politicians' use of performance information are examined in depth: ter Bogt (2004) and Askim (2007). Ter Bogt (2004) conducted a survey among aldermen in larger Dutch municipalities about their performance information use. He found that informal, verbal contacts, as well as formal meetings and contacts with top managers, were the relatively mostly used sources of information (see also Daft and Lengel, 1990;ter Bogt, 2001). Relatively less important, but still useful to some degree, were budgetary and interim reports. In addition, the aldermen were critical about the quality of the performance information in the budgetary documents, and did not make much use of it. There were also some differences in the use of the performance information dependent upon the aldermen's portfolios. In general, ter Bogt suggests that aldermen perceive performance information as less important than other information sources.
Contrary to ter Bogt, Askim (2007), who investigated the importance of performance information to local government councillors in Norway, presents more 'positive' findings. Askim's study observes high levels of performance information utilization, although it varies across policy fields. Askim measures performance information using a relative approach, i.e. relative to other sources of information, such as party programmes and information of the administration about specific cases. As in ter Bogt's study, the evidence is based on a survey.
Why are these findings between a Dutch and a Norwegian study so different? The most likely explanation for the observed differences may relate to the way the studies have framed the performance information. Ter Bogt compares it with verbal informal information, while Askim makes a comparison with other sources, such as party programmes, national benchmarks and surveys on user satisfaction. In my opinion, the differences in findings can be explained by the ways in which the interpretations are framed. To put it differently, the more attractive alternative options are, the less attractive the core option is. Moreover, Askim is positive about a 70% score on the usefulness of the information and a 43% score on its use, whereas ter Bogt is negative, or at least critical, about an average score around 2.5 on a scale between 1 and 5 on information use. However, what a suitable reference point is for assessing whether a certain type of information is useful or used, is contestable, and has to be justified. Table 2 summarises the differences between the two studies. Obvious conclusions are: •Usefulness is likely to be higher than actual use. •Findings about the usefulness and use of performance information depend on: the types of alternative information sources; and the appropriate point of reference, i.e. the boundary between 'good' and 'bad'.
Performance information use is a more popular research theme than financial information use. In general, survey-based studies that assess politicians' appreciation of performance information present 'positive' findings (Brun and Siegel, 2006;Ho, 2006). Field case studies about politicians' actual use of performance information have indicated, however, that this is minimal, or even absent (Bourdeaux, 2008;Raudla, 2012;Grossi et al., 2016). These findings seem to support the conclusion about the papers analysed in this section and the previous one. In addition, there are two multi-method studies which combine observations with a survey, one by Johnson and Talbot (2007) and one by ter Bogt et al. (2015). The first study points to an overall low use, while the second corroborates my earlier conclusion that appreciation measured by a survey can be high, whereas actual use-measured through observations-is low. Other survey studies about performance information use have reported diverging findings (Stalebrink and Frisco, 2011;Saliterer and Korac, 2013). © 2016 CIPFA How to conceptualize politicians' use of accounting information? Figure 1 is an attempt to conceptualize politicians' use of accounting information. It distinguishes between information usefulness and information use. Information usefulness is assumed to be driven by user needs which, in turn, are determined by such issues as problems to be solved and challenges to be coped with. Studies about the usefulness and use of accounting information are mostly based on accounting documents: performance budgets or income statements. These documents are impacted by legislation or fashion, for example moving from cash to accrual accounting, or making financial documents more performance-oriented. Figure 1 focuses on the demand side for accounting information. There are also conceptualizations of financial or performance information which distinguish between the demand for and supply of this type of information (for example Van Dooren, 2005, p. 366).
Much of the available research concentrates on the use and usefulness of regular accounting documents, such as budgets, interim reports and annual accounts. Important decisions, however, are often based on more specific policy documents which include both financial and other information. Examples are policy notes on budget cuts or a policy-making re-orientation regarding a policy field, such as education, culture or the environment. This implies that rather than purely focussing on accounting documents, researchers should increase their understanding of how politicians arrive at decisions, and what information is required for making these decisions (see also Wilson, 2006;Demaj and Summermatter, 2012).

The way forward for research
Accounting for decision-making My main suggestion is that rather than concentrating on politicians' potential use of regular accounting documents (budgets, financial statements and performance reports), future research should focus on the information requirements of the decision-making process, including agenda-setting and problem-solving. Agenda-setting includes politicians' actions to bring a new theme to the attention of their fellow politicians, such as combating race discrimination. Problems have to be understood in a broad sense. If a problem is well defined, for example as a choice among several investment options based on clearlydefined criteria, accounting information can support choosing the best option. However, problems are often not well structured, for instance when they involve sentiments of discomfort regarding potential overspending, in the case of a particular investment project (for example a new museum), without clear, justifiable explanations for the overspending. Other examples are plans for future projects which are surrounded by uncertainties, such as one country's programmes for reducing air pollution while the policies of neighbouring countries are unclear. It is never absolutely clear in advance which accounting information will be helpful; politicians have to determine what they want to know and what information is available. Only then can the information be helpful in framing relatively complex problems (including politically-biased accounting information use for the pursuit of influence and power: see, for example, Johnsen, 2005, andJohnson andTalbot, 2007). A third category of problems does not relate to decision-making for the purpose of Table 2

Research study
ter Bogt (2004) Askim (2007 problem-solving, but is directly associated with issues for which politicians are accountable. In these cases, legitimation is the main purpose, for example in a financial scandal (for instance 'Who is to blame for these useless IT investments?'). Many of the above-mentioned categories of problems and related examples do not refer to the regular accounting documents, such as annual budgets or quarterly performance reports. This finding resonates with Demaj and Summermatter's observation (2012, p. 103) that it is the extreme cases, rather than the average cases, that best reveal how and why information is needed and used.
What methods are most appropriate for this research? My literature review suggests conducting observational studies in combination with semi-structured interviews are to be preferred over questionnaire-based research. The latter, i.e. survey studies, over-simplify the complexity of the information-seeking behaviour of politicians, and are generally inclined to overestimate the actual use of the accounting information. In addition, academic researchers are encouraged to conduct more interventionist studies. This research is primarily initiated by the problems of practitioners, for example politicians, while its goal is twofold. First, in addition to focusing on understanding how and why politicians approach certain problems, it is also designed to contribute to solving those issues. Second, it links research findings to a broader spectrum of academic knowledge-building. Further, as public sector accounting researchers, we should stop using an accounting-based vocabulary. We need to talk to politicians in a language that they understand.
Several branches of theories are potentially suitable for approaching this theme. There are currently two dominant ones in public sector performance management research. The first is contingency-based research, which starts from the view that performance management practices depend upon certain contextual circumstances, such as the complexity of processes or environmental uncertainty (see, for example, Cavalluzzo and Ittner, 2004;Van Dooren, 2005;Moynihan and Pandey, 2010;Speklé and Verbeeten, 2014;and Kroll, 2015, for a review on the link between managerial performance information use and organizational performance). The second one is neo-institutional sociology, which highlights how performance management practices have become taken for granted due to certain pressures, and how these practices contribute to habitual types of actions (see Modell, 2009, for a review). Without denying the importance of these theories, some other theories are suggested here because they potentially enable an understanding of how decision-making by politicians drives their accounting information use. Theories should do justice to the complexities and richness of empirical realities, which is an implicit plea to refrain from views that assume straightforward causalities among abstract constructs (see also Tummers and Karsten, 2012). In decision-making, politicians often face complexities, for example in choosing among various options to solve a problem or in adopting the proper information sources. Hence, ideas about bounded rationality and a problem-driven search for information, as developed more than 50 years ago by Simon (1958) and Cyert and March (1963), are potentially of interest. Simon argues © 2016 CIPFA that due to constraints in their cognitive capabilities, decision-makers mostly do not consider all relevant options for solving their problems, but only a limited number of alternatives. Moreover, decision-makers generally do not strive for the best but for what is considered as satisfactory. Cyert and March indicate that people in organizations search for that particular information which pertains to the problems they are faced with. Furthermore, in seeking information, they draw upon their past experiences. This view is in line with ideas about how decision-makers can cope with information overload (Edmunds and Morris, 2000;see Jansen, 2008, p. 11, for a public sector illustration). What are the implications of these theoretical viewpoints for accounting information use? If a problem arises, the available sources of information might be the first things to consider. These can include a budget, a performance report, or some part of the annual accounts; in general, accounting information which forms part of the planning and control cycle. Often, however, this type of information is regarded as insufficient because it lacks necessary detail for obtaining an appropriate insight into the problem at hand. This situation often leads to a desire for more, and especially more detailed information. What happens next is dependent upon how the various actorsmembers of the executive and delegates from the different parties (the coalition and the opposition)-deal with this information in positioning themselves in the debate. If the actors are satisfied with the 'solution' to the problem, or if they feel that, despite their dissatisfaction with it, there is no room for further 'improvement', they have to accept the new status quo, at least for the time being. However, the actors' decisions also depend on what other actors do and say. It is this dynamic interplay between the various actors and the information sources they desire to adopt that shapes the processes of decisionmaking and information use (see also Askim, 2009). Actor-network theory (ANT) can also be considered for investigating this dynamic interplay (see Christensen and Skaerbaek, 2007, for a public sector accounting study based on ANT).
When politicians use information for addressing or solving problems, their effectiveness in convincing other actors primarily depends on the way in which they present their arguments. Politicians' main asset is speech, rather than reading or writing. This is why the rhetorical theory of diffusion could be helpful in understanding how politicians mobilize their information in debates. According to Green (2004), this theory highlights the causal potency of language in shaping organizational life and behaviour. Rhetorical theory portrays managers (or politicians) as 'rhetors' and suggests that their most influential tool is language. In speech, various types of arguments can be used, ranging from emotional (such as 'fear about the unknown'), rational (where efficiency and effectiveness reasoning is at stake), to claims of moral legitimacy (for example impartiality, integrity). The relevance of each type of argumentation may change with the stage of the policy-making lifecycle (pathos may initiate change, logos may implement it, and ethos may sustain it). Financially-oriented argumentation is often assumed to be linked to rational forms of reasoning, whereas emotional pleas (for example about the loss of autonomy if draconic measures for solving financial deficit problems are disregarded), as well as moral standpoints (such as being accountable towards creditors or future generations) can also be relevant.

The role of financial specialists
A second future research theme concerns the role of financial specialists in party delegations. Here Hyndman (in Ezzamel et al., 2008) makes a distinction between 'insider' and 'outsider' politicians, referring to differences in financial expertise. In his studies about the use of resourcebased accounting and budgeting (RAB) in the NI Assembly, insiders, such as members of the budget committee, have a reasonably good understanding of financial figures. Outsiders, usually members of other parliamentary committees, do not have any knowledge about and interest in financial affairs. Future research could focus on how insiders share their knowledge of financial issues with outsiders.
Various roles of financial specialists in party delegations are thinkable, for example, a supportive role towards non-financial colleagues, or a watchdog role which emphasizes financial prudence and countervailing action to 'unnecessary' spending initiatives.
The information exchange among different types of politicians cannot be easily unravelled using surveys or interview-based case studies. In order to assess what is really going on in this process, a non-participant observational approach is desirable. This approach could include both formal and informal party delegation meetings, thereby reaching behind the scenes of the official council or parliamentary committee meetings. Moreover, a longitudinal approach seems appropriate because deliberation processes regarding politically relevant problems, which require financial information, evolve over time. A promising theoretical avenue would be role © 2016 CIPFA theory, in which expectations about roles-for example the expectations of party delegates about the role of financial specialists-can be compared with the actual behaviour (in this case, that of the financial specialists). This theory can also include organizational and personal characteristics that shape possible conflicts between expected and actual roles. The research by Liguori et al. (2009) about the interconnected roles of politicians and managers, as well as Roussy's (2013) study on roles of internal auditors in the public sector, are fertile sources of inspiration in this context.

Challenges
My suggested research agenda is challenging but risky. In comparison with surveys, casebased research, including observations, is a less structured and therefore more complicated method for studying political accounting information use for decision-making. Getting results published via this approach may therefore take longer and might be difficult. However, gaining meaningful academic knowledge and conducting practice-relevant research should remain the principal motivators of scholars.

Implications for practice
At this stage it is not easy to formulate implications for how politicians should or can use accounting information. This essay shows why we need for indepth investigations into the role of accounting information in political agenda-setting and problem-solving. Nevertheless, some tentative pointers for practice can be given. As Pollitt (2006, p. 49) has already indicated, performance reports, or more generally accounting documents, are available and will be picked up especially in those cases where something seems to have gone wrong. Two implications for practice are suggested.
First, standard accounting documents, such as budgets and interim reports, are not likely to be used systematically by politicians, unless there are institutionalized arrangements for doing so. This can be the case when a budget committee of the legislative employs formalized procedures requiring a systematic analysis of budgetary documents, or when the legislative is supported by intermediaries, which have the obligation to enrich the regular accounting documents for use by politicians. These institutionalized arrangements are perhaps only feasible in larger administrations. Smaller ones are dependent upon the coincidental expertise and interests of individual members of the legislative. If institutional arrangements for using standard accounting documents are in place, a rational type of use is likely. Examples are cost-benefit analyses of programmes or projects, or corrective actions based on a comparison between planned and actual figures. Evidently, certain accounting traditions, which can diverge among countries, can have an impact on the use of the regular accounting documents.
Second, when politicians have to participate in debates about controversial projects or programmes, accounting information is likely to be used intensively at the various stages of the decision-making process. Future research has to reveal which contextual circumstances contribute to certain types of use by the various players. For controversial programmes or projects, a rational type of use seems unlikely. We have to accept that, in such circumstances, politicians may be inclined to use accounting information to accommodate their political preferences.