The beauty of constructive culture: planting the seeds for widespread performance information use among councilors

Introduction

Politics and the well-being of citizens often go hand-in-hand (Radcliff, 2001) because local council decisions affect the welfare of municipal residents in many ways. To better understand such council decisions, one should know what drives them. This study examines politicians' use of performance information (PI) in budget meetings taking place in a municipal council. By PI, we mean qualitative and quantitative information about public sector activities, processes, services, service outcomes and transformation processes, such as productivity/efficiency and effectiveness (e.g., Hatry, 2006; van Helden and Hodges, 2015). Politicians can use this information for learning, controlling, communicating, budgeting, motivating, promoting/advocating, evaluating, celebrating, collaborating, contracting, sense-making, decision-making, account-giving, framing, reassuring and preventing issues to name only a few uses (Behn, 2003; Van Dooren and Van de Walle, 2011).

Public sector financial management refers to a sets of activities, such as analyzing, structuring, setting targets and implementing measures in the field of finance (Bergmann, 2009). In short, Bergmann (2009) defines financial management as “planning” and “controlling.” PI use occurs in the ongoing cycle of public financial management with the movement from the strategic and operational target setting to budgeting, operational planning, implementation, controlling, and feedback (Jones and Pendlebury, 1992). From the perspective of this study, PI use has links to all phases of the cycle as presented by Jones and Pendlebury (1992). In this article, we focus on one of the main events in public financial management – budgeting. A budget meeting offers a chance to examine PI use in these many phases of public financial management. In local governments in Finland, strategic and operational goals and means are planned in budget meetings at the same time that past performance results are evaluated. Resources are allocated to meet the new targets in these budget meetings.

The prior research addressing PI use has primarily examined whether or not PI has been used, along with how and why it has been used (e.g., Giacomini et al., 2016; Ter Bogt, 2004). Both high and low levels of information use have been reported, and according to previous research (Buylen and Christiaens, 2016; Raudla, 2012), many individual and contextual factors drive PI use. In terms of information user groups, the research has covered public managers, politicians, and citizens (Askim, 2007; Kroll, 2015), but the focus has been on public managers. Thus, it has been suggested that the imbalance should be remedied with more research addressing politicians' and citizens' PI use (Pollitt, 2006; van Helden, 2016). Because studies have mostly reported limited PI use by political actors (Rajala, 2019), this study aimed to explore what contextual factors provide a platform for high use levels. This means that the individual attributes of politicians were not scrutinized.

This study's research question asks what contextual factors support extensive and widespread PI use among politicians in one Finnish municipal council. The question was answered by conducting an ethnographic study that utilized observations and interviews. In practice, we analyzed politicians' speeches in budget meetings to assess the presence and drivers of the use of PI in political rhetoric. In addition, interviews and content analysis were conducted to report politicians' views on the intensity of and explanations for the observed high use levels. Many phases of the public sector financial management cycle were present in the observed budget meetings. In budget meetings, councillors regularly refer to strategic and operational targets. Moreover, the performance of local government and feedback from organizations and service users are constantly discussed in councillors' speeches.

Because past studies have mostly reported on the reasons for limited use, this study contributes to the literature by identifying the reasons for high levels of politicians' PI use. Hence, our study presents an intriguing story of successful PI use in public sector financial management processes. Our results are also valuable because past studies have primarily used surveys and interviews to study politicians' PI use (see, e.g., Askim, 2007; Ter Bogt, 2004), with a few exceptions (e.g., Buylen and Christiaens, 2016). The problem with survey and interview studies is that they do not necessarily reveal the true behavior of the politicians (van Helden, 2016). For example, data from surveys and interviews can suffer from unrepresentative samples, non-response issues, and response biases (Bethlehem et al., 2011), which in turn mask the politicians' actual use of PI. This justifies our methodological approach to the topic, because observations cannot suffer from response biases and non-response issues. By using robust observation method, the researchers observe what happens instead of having to rely on other people to report what happened.

In the context of the case organization, there were present several contextual factors that were associated with low PI use in previous studies. However, we found politicians' high use of PI. Threats of information overloading, commercial secrets relating privatization, lack of power, low quality of budgeting laws and budget format were all factors predicting low use levels. Our findings highlight the importance of a constructive political culture, functional dialogue between administrators and politicians, high-quality of the PI and information channels, and trust in the information providers as the main building blocks for continuous PI use. For practitioners and academics, we propose preliminary ideas about what elements generate constructive performance dialogues in budget meetings, where political battles can easily break the constructive atmosphere.

The rest of this article is structured as follows: In the next section, we examine relevant research addressing contextual factors that drive PI use among politicians. In the third section, we describe the case organization. The fourth section is reserved for an explanation of the research methodology, after which our empirical analysis is reported. In the last section, we present conclusions and a discussion of our findings.

Contextual determinants of politicians' performance information use

Prior research indicates that PI use is sometimes case-specific, and utilization varies depending on the government level, city and users (Charbonneau and Bellavance, 2015; Johansson and Siverbo, 2009). So far, the research has identified several contextual determinants that might affect political actors' use of PI (see Table 1). However, studies have defined PI in dissimilar manner, which can make the comparisons of past results more difficult. Moreover, there are some studies focusing on the political use of solely financial information. Table 1 summarizes the contextual determinants that drive information use and the types of information studied in prior research. In Table 1, all factors driving information use can be placed in five broader categories identified and proposed by the authors: environmental factors, organizational factors, information providers, information channels and information.

Environmental factors arise from the society and nature surrounding the organization (see Table 1). Environmental factors refer to service users (Bourdeaux, 2006), the service production environment, laws or rules (Lu et al., 2009), and political (Buylen and Christiaens, 2016) and economic environment (Paulsson, 2006). Average unemployment rate and heterogenous preferences of citizens are examples of drivers of use relating to service users (Bourdeaux, 2006; Lu and Willoughby, 2015). It has been argued that heterogenous service users compel actors in the public sector to use PI because, without this information, it becomes more difficult to find better ways to serve different service users requiring personalized services (Moynihan and Hawes, 2012).

In the environmental factors, legislation has been identified as one factor incentivizing PI use (Lu et al., 2011). Raudla (2012) found that low-quality performance-based budgeting laws lowered the use of PI. According to Lu et al. (2011), PI is used when the quality of performance-based budgeting laws is high. These laws both guide use and demand the use of PI (Lu et al., 2011) and for most public organizations, laws come as a given. Fiscal stress is one typical phenomenon describing the economic environment in past studies (Bjørnholt et al., 2016; Paulsson, 2006). In economic prosperity, the public sector has resources for measurement and time for analysis, whereas during economic downturns, the focus is on cutback management (e.g., Bjørnholt et al., 2016). Political conflict and unstable cultures are examples of PI use drivers arising from political environments. For example, political conflicts can encourage PI use because it becomes ammunition for politicians (Giacomini et al., 2016). However, such conflicts can lower PI use if conflicts prevent measurement or joint interpretation (Rajala et al., 2018).

Others have identified organizational factors as barriers to use (see Table 2). Organizational factors refer here to the level of performance (Saliterer and Korac, 2013) and organizational structure (Askim, 2007) and culture (Lu and Willoughby, 2015). Financial health (Saliterer and Korac, 2013) as an indicator of performance can also influence PI use. For example, fiscal deficits may shift the focus to savings and budget cuts affecting measurement. Lu and Willoughby (2015) emphasized performance-oriented culture as a driver of use, and Saliterer and Korac (2013) proposed rational culture as an incentive to increase PI use. In a rational culture, PI is regarded as helping the process of finding the appropriate means that lead to desired ends (Saliterer and Korac, 2013).

The organizational structure as an organizational factor describes the division of labor (i.e., personnel, administrative bodies, levels of hierarchy, organizational tasks and decision-making powers associated with these tasks) and management structures (e.g., management duties, such as budgeting and controlling/monitoring performance). It has been argued that the budget format as an element of organization structure affects the extent to which PI is utilized in budgetary decision-making. For example, Hill and Andrews (2005) noted that, as compared to performance-based budgeting approaches, input-based budgeting processes limited the use of PI by focusing attentions on appropriations. In contrast, when the budget format is grounded in output or outcome result information, politicians may pay more attention to PI (Raudla, 2012).

In addition to the factors already mentioned, information providers (Pollitt, 2006), information channels (Lu and Willoughby, 2015) and information itself (Raudla, 2012) have been identified as important contextual factors in information systems (see Table 1). Because information providers, information channels and information exist outside or inside an organization, it is unclear whether these factors are environmental or organizational factors. Therefore, we formed the following categories for them: information provider (see Table 3), information channel (see Table 4) and information (see Table 5). The qualities of the information provider have been shown to affect PI use of politicians (Raudla amd Savi 2015). For example, studies have shown that for information to be used, the source of the information must be credible (Pollitt, 2006).

Studies have also demonstrated that fine tuning of information channels and information to serve political needs can promote PI use by making information use easy and beneficial (Jorge et al., 2016; Lu and Willoughby, 2015). For example, information type seems to impact the utilization of PI (Ellul and Hodges, 2019). Furthermore, there should be a suitable amount of information, since information overload will deteriorate information use (Caruana and Faruggia, 2018).

Tables 1–5 present subcategories under the broader categories of information providers, information channels, information quality and the environmental and organizational factors explained above. The five categories proposed by the authors provide a starting point for a taxonomy of contextual PI use drivers. In the empirical analysis, we examine whether drivers of PI fall under the five categories we proposed. If not, new categories can be proposed for the taxonomy after the empirical analysis.

Research context

According to the Finnish Constitution (1999/731, § 121), “Finland is divided into municipalities whose administration must be based on the self-government of the inhabitants.” Self-government means that municipalities have the right to decide their own affairs within the boundaries of national laws. However, national laws set forth mandatory tasks for the municipalities.

The municipal council has ultimate authority in the municipality and plays a major role in the exercise of self-government. It approves budgets and financial plans for at least three years and makes decisions on all matters assigned to it by national laws. The municipal budget is where objectives and finances for municipal operations are accepted. The council is also responsible for the activities and finances of the municipality. Lastly, it is the council's duty to appoint members to other municipal bodies, such as the municipal board and committees.

The council is formed of representatives of the public elected every four years. These representatives are not professional politicians but undertake council work alongside their daily occupations. A municipality's officials and employees run operations in an organizational hierarchy under the council. Although the council has ultimate authority, the municipal executive board or committee may have more power to determine budgets, depending municipality's arrangement of functions.

Local governments in Finland apply an accrual-based accounting and a budgeting model introduced by accounting reform of 1997. The Finnish model is based on a revenue/expense model and historical costs (Sinervo and Haapala, 2019). The types of PI observed in Kangasala have included financial and non-financial information. Non-financial performance measures provide PI in non-monetary terms, including customer satisfaction, innovation/new product development, and employee turnover (Verbeeten and Boons, 2009). Financial PI refers to accrual accounting information (such as liability, receivables, assets, cost of activities, cost of service delivery and depreciation), budgetary financial information (such as revenues and funding sources), current expenditures by nature, current expenditures by destination, capital expenditures by nature, capital expenditure by destination, level transfers to other local entities, budgetary surplus or deficit as a financial outcome, and establishment of accounts receivables to be recovered and commitments to be paid (Liguori et al., 2012). Debt and solvency ratios were also identified as financial outcomes.

The case organization and its unencouraging premises for PI use

In our case, city of Kangasala, there are 51 councillors from seven parties. Two parties have 49% representation in total, while the rest of the parties have approximately 10% representation each. The financial state of Kangasala weakened remarkably in 2003–2013 due to negative operating and investment cash flows for 9 of the prior 10 years. As a result, various cutback policies were adopted during 2014–2020. Because past studies (Bjørnholt et al., 2016) have linked fiscal stress to limited PI use, high use would not be expected.

In terms of legislation, performance-based budgeting regulations are embedded in two laws: the Finnish Local Government Act (410/2015) and the Act on Restructuring Local Government and Services (169/2007). Performance-based budgeting law is of low quality in Finland based on the criteria used by Lu et al. (2009), who noted that low-quality budgeting laws disincentivize politicians' PI use.

In Finnish municipalities, the municipal executive board makes proposal for the final budget to council and council decides whether it accepts this proposal. From the perspective of organizational structure, the politicians in Kangasala had limited power to change budgets in the council's final budget meeting because budget proposals of the municipal executive board were in practice prepared and accepted first by the municipal executive board. The board had representatives from six of seven political parties that formed the municipal council, meaning budget proposals were accepted by six parties before final council approval in the budget meeting, which was effectively more an administrative formality than actual debate about municipal resource allocations. Thus, based on past studies, only limited use would be more likely in final budget meetings because absence of political competition (e.g., Giacomini et al., 2016) and lack of power to change the budget has been associated with low use (e.g., Raudla, 2012).

Any PI presented in Kangasala's budget was on top of existing input information, and performance results were, at best, indirectly connected to inputs in graphs, tables, or texts. Therefore, we concluded that Kangasala had partly input-based budgeting and partly performance-informed budgeting. Because input-based budgeting is associated with limited PI use (Raudla, 2012) and performance-informed budgeting is not performance budgeting, Kangasala would likely have more low levels of use. Here, performance-informed-budgeting refers to resources that are indirectly related to past or proposed future performances.

Kangasala produces a substantial amount of data, but there are also other information providers, such as Statistics Finland. With statistics on different aspects of life, information production is less of a problem. The greater problem is how to use all this information because of the information overload. The possible information overload predicts low PI use. However, information channels have been enhanced in Kangasala. The annual budget documents and strategic documents in Kangasala have lessened in the last five years because of problems with the size of documents operating as information channels. These size-related problems also predict lower PI use, but current developments may have improved the usability of the information channels.

Research methods

This is an observational research with a case study approach. We chose one municipality from central Finland, and the decision to focus on the council was based on three reasons:

1. Kangasala has been successful in establishing active PI use in the political circles, and this use has improved the fiscal state of the municipality.

2. The municipal councils play central role in local decision-making.

3. Empirical data is easily accessible.

The council is the highest decision-making body, and it decides the budget. Therefore, the councillors are in a key position to use PI in budget decision-making. Data on budget meetings are readily available because the chosen municipality records budget sessions and posts them online.

This study's empirical data collection method was based on naturalistic observations, which involve studying the spontaneous behavior of participants in natural surroundings. By using an objectivist approach to observations, we recorded councillors' information uses to determine whether they used PI. In an objectivist approach, researchers do not influence or interfere with those under observation, and they follow rigorous rules that prevent bias in the data (Angrosino, 2005). Rather than taking part in the meetings, we instead observed recordings of budget meetings afterward. We did interview a few councillors in 2018, which could have affected behavior in 2018, 2019, and 2020 budget meetings, but in 2017, the councillors did not know that we would interview 10 of them in 2018. And because there was high-level PI use in every observed year, we conclude that the interviews did not cause notable changes in behavior.

To prevent biasing the data, we used structured observations in data collection (e.g., Holtrop et al., 2016). By coding PI use behavior according to previously agreed-upon categories adopted from Rajala (2019), PI use during budget meetings were sorted into seven categories: input information, output information, workload information, process information, outcome information, productivity information, and cost-effectiveness information (for details, see Appendix 1). Use was observed when politicians' used PI in their rhetoric or referred to politicians' or officials' speech that included PI. To determine the level of the PI use, we employed the interpretative scheme presented in Table 6, which was adapted from Rajala (2019). Instructions for the observers were written before the observations to guide the observers on how to detect PI use and code it into the template. Both the template and instructions were pre-tested with other case organizations and iterated before the observations took place. All of these procedures are well-proven procedures in observational studies (Podmore and Luff, 2012).

In addition to coding observations of PI use in the template, non-use of PI was also recorded by writing down the main points of speakers who did not use PI. We also collected the following information: the names of the councillors who spoke, the time-point when the councillor spoke, the section under which the councillor spoke, the type of service that was under discussion, and a key word or quotation from the speech that indicated what information was being used according to the categories presented in Appendix 1.

The length of the observed meetings was 1,413 min in total, or 23 h and 33 min. The observations were conducted over a time span of four years, following the newly elected council from 2017 to the end of their term, which was the end of the year 2020. To support the observations, other empirical data was utilized including the participant lists to check the correct names of the politicians speaking, budget documents to identify the information being referred to in the local council, and minutes of the meetings to collect section numbers.

The degree of agreement among the researchers and their observations was tested with the inter-observer reliability test called as Cohen's kappa (e.g., Gwet, 2008). The Cohen's kappa value was 0.79, indicating excellent agreement among observers (Fleiss, 1981). To evaluate the magnitude of the Cohen's kappa, the bias and prevalence effect was also assessed. The prevalence-adjusted kappa (PAK) was 0.84, which is higher than the non-adjusted Cohen's kappa of 0.79. The prevalence-adjusted and bias-adjusted kappa (PABAK) was also 0.84, which taken all together indicates good reliability of the observation method.

In addition to the observations, we conducted 10 semi-structured interviews with local politicians. The semi-structured interviews enabled more in-depth exploration of drivers of PI use, although the same drivers of use could be observed in the budget meetings. Interviews were conducted from February to May 2018, and each interview lasted from 42 to 94 min. With the consent of interviewees, the interviews were recorded and the data were transcribed, which produced a transcript of 73,211 words in total. Interviewees were interviewed only once and selected to represent different political parties, educational backgrounds, political experience, age groups and genders (see Appendix 2).

The interview data were examined using content analysis. In the first phase of coding, we searched and coded statements describing the intensity of use and the contextual factors explaining PI use by the politicians. We subsequently compared these answers to what we saw in the budget meetings. In this way, we controlled for biases arising from interviewees. Only similar findings arising from observations and interviews are reported in this study. In the second phase of the coding, we created categories from the answers explaining PI use. A researcher not taking part in the interviews evaluated whether the interviewers caused any biases, and no biases described by Brown (2001) were identified.

The main limitations of this study are that not all council members were interviewed, and council meetings do not reveal fully how often PI is used in the municipal executive board, committees or informal gatherings. These limitations are acceptable as participants cannot be forced to be interviewed and one cannot attend to all formal and informal meetings of the municipality. There are some additional limitations in how representative the samples used in this study are (for details, see Appendix 2).

Results

In total, there were 73 speeches by politicians in the 2017 budget meeting. In this meeting, 59 of 73 speeches utilized PI. This means that 80.8% of the speeches contained PI, and we therefore found extensive PI use. Out of 27 speakers in the budget meeting, 26 used PI, meaning approximately 51% of councillors used PI. This indicates high levels of PI use. On average, PI was used 3 times per speech (median = 2; mode = 1). In 2018, 64 speeches were kept and 59 of these included PI. Thus, 92.2% of the speeches demonstrated PI use, indicating extensive use. Of the councillors, 30 used PI which is 58.8%. This is high use according to the analytical framework presented in Table 6. On average, PI was used 3.8 times per speech (median = 3; mode = 2).

In 2019, 113 speeches were kept and 90 (79.6%) included PI. This is extensive use. In total, 34 councillors used PI and this is 66.6%, indicating high levels of PI use. On average, PI was used 2.3 times per speech (median = 1; mode = 1). In 2020, 89 speeches were kept, and 62 included PI. This was 69.7% of the speeches, demonstrating high levels of PI use. From the councillors, 26 used PI, which is 51%. Again, this is high use according to the analytical framework presented in Table 6. On average, PI was used 2.4 times per speech (median = 1; mode = 0). We did not calculate any confidence intervals for these proportions because conditions for valid confidence intervals were not met. The different information types used in the meetings are described in Table 7.

Overall, PI was used in argumentation and/or sensemaking in every observed year. The arguments promoted budget changes or other assignments attached to the list of actions that were approved with the budget. The sensemaking process helped councillors understand the contents of the budget proposal and past performance results.

Explaining the PI use

Based on the analyses of the interviews and observation data, it appears that PI is used and appreciated by the politicians in Kangasala. Many political groups expressed their gratitude to the city officials by thanking them for the given PI. For example, one party leader stated in a budget meeting that “the Social Democrats want to thank the city officials for their good and transparent budget preparation process.” Later in during the same meeting, a member of the Christian Democrats expressed his satisfaction with the information by stating, “Thanks for this executive summary presenting the key information from the budget.” The willingness to use PI and the gratitude of politicians encouraged the public official to give more information to the politicians. Grateful information users and public officials putting their efforts toward knowledge sharing are both important parts of the conversational culture in Kangasala.

Information channels also supported PI use by feeding the necessary information to discussion in a timely manner. Official performance documents, statistical yearbooks, notices and minutes of meetings, among other information, are digitally available to council members, along with printed versions of, for instance, budgets and financial statements. These performance documents include input, process, output, outcome, productivity and cost-effectiveness data. Modern technological solutions play an important part in this dialogue. To avoid interruptions in information sharing arising from compatibility issues in information transferring, the city provided council members tablet computers. Moreover, the information was constantly displayed through PowerPoint presentations and other screens during the budget meeting.

The interviewed politicians were content with the intelligibility, quality, quantity, and easy accessibility of the information. Information overload was avoided in performance dialogues by using summaries of key figures, chosen in collaboration between councillors and public administrators. Lack of information was avoided by ensuring active dialogue and information exchange between the councillors and public administrators. The dialogues also generated intelligibility, whereas information channels, such as tablets, contributed to accessibility. It seems the administrators were able to offer satisfactory information, but also enabled opportunities for further discussion. This can be seen in the following quotations:

I truly cannot blame municipal managers for not providing information. We can always ask for more.

Experienced male councillor, Left Alliance

It is not only the official documents… I feel that the most important thing is that information is orally explained, and backgrounds and effects opened and reasoned.

Experienced female councillor, Social Democrat

A conversational culture flourished in the city thanks to the safe environment providing opportunities to ask questions about unfamiliar aspects of performance and PI. The political culture was more respectful than conflictual, which partly encouraged politicians to ask about what they did not know in the performance data. Interviewed politicians felt free to ask questions of responsible sector managers, and they were mostly satisfied with the answers they received. This can be seen in one dialogue from a budget meeting:

The number of sick days has been increasing. The costs of paid sick leave has risen 27% last year. There is something wrong. Are there chronic diseases or has the straw broken the camel's back?

Newly elected female councillor, Greens

This is a legitimate question. The sick days have increased from last year…there are two reasons for it…mental health problems and musculoskeletal disorders.

Sector manager

Public managers regularly kept politicians updated on current matters in the city as well as showed initiative in bringing urgent but also topical issues to the awareness of the city council. Altogether, interviewed politicians did not believe that information was being withheld by the administration. It seemed that politicians' questions were welcomed and taken seriously, and politicians valued the expertise of the administration and trusted their judgments, as demonstrated by the following quotations:

I have to give credits for our city mayor for doing a great job. During our council year, he gives us constantly updates how things are going. He keeps providing us information, making preparations … so there is rarely surprises.

Experienced male councillor, Christian Democrat

If some exceptions, overspendings, or so occur during council year, they are already commented in the budget documents. They give explanations why things have gone this way. I think that it works rather well; I do not feel that I have to do my own detective work to find out what happened and then demand answers to a bunch of questions

Experienced female councillor, Social Democrat

The performance dialogue between politicians and municipal managers were active, and the organizational structures included processes that encouraged dialogue. Politicians felt there were multiple opportunities to ask questions concerning the performance or finances of the city. The city organized highly appreciated “group forums” before council meetings and “evening classes” for politicians on current matters. Respondents also indicated that there are training opportunities for newly elected council members.

Public managers and the city mayor were key information providers. They were valued sources of information, and politicians repeatedly trusted and relied on their views and knowledge about organizational performance. The politicians also often had the capability of accepting PI and its message, which made it worthwhile to incorporate PI into arguments. In general, politicians felt that the PI made the argumentation more convincing as compared to statements based on opinions.

All in all, Kangasala has succeeded in creating a participatory organizational culture open to well-functioning, two-way performance dialogue between politicians and municipal managers. The four key dialogue principles were usually present (Isaacs, 2001): listening, respecting the experiences and thoughts of others, suspending (e.g., not conducting aggressive argumentation that leaves no room for other opinions), and voicing (i.e., explicitly articulating thoughts). By showing respect for others, political leaders and public managers practiced what they preached. When arguments started to get more heated, the council chair asked politicians to calm down and behave respectfully. This calmed down politicians in opposition and government. Although the politicians in the opposition were active PI users, they mostly behaved in a constructive manner.

In this participatory organization culture, the information channels, such as tablets or information brokers, fed the required PI to decision-making venues in a timely manner. Thus, the information channels supported active dialogue by providing data that generated conversation topics. Moreover, the dialogue generated trust in information providers and supported information users in their efforts to use PI. The enhanced trust seemed to increase the use of constructive language in budget meetings, and the available information supported a constructive culture where politicians did not usually lose their tempers due to missing information.

Discussion and conclusion

This case study examined the contextual factors that promote sustained high PI use by politicians on one Finnish municipal council, the municipality of Kangasala. Based on observations and interviews, we found that a constructive and supportive organizational and political culture promoting active performance dialogues was the main explanation for the observed high use. A constructive political culture did not mean that there were no debates; rather, debates were respectful and constructive. Therefore, this finding does not contradict the findings of Giacomini et al. (2016), who argued that political conflicts increase PI use.

Past studies have explained use with a variety of factors, but only two studies have emphasized all five contextual factors found to be important in this study (see Tables 1–5). For example, Askim (2007) highlighted information channels and organizational and environmental factors, whereas Saliterer and Korac (2014) emphasized the role of information availability and organizational culture. Only Pollitt (2006) and Ezzamel et al. (2004) have emphasized the role of all five contextual factors in manner similar to this case study. However, neither Pollitt (2006) nor Ezzamel et al. (2004) did examine the relationships of the five factors in-depth, as we did. To conclude, both the theoretical section (see Tables 1–5) and the empirical analysis confirmed the importance of information provider (e.g., Jorge et al., 2016; Pollitt, 2006), information channel (e.g., Raudla, 2012; Ter Bogt, 2004), information quality (Ellul and Hodges, 2019; Ezzamel et al., 2007), organizational factors (Lu and Willoughby, 2015; Paulsson, 2006) and environmental factors, such as political culture (e.g., Giacomini et al., 2016). Therefore, we propose that these categories can be applied as main categories under which subcategories describing more specific PI drivers can be placed (see Tables 1–5).

Interestingly, high PI use was observed in Kangasala despite conditions predicting low use. According to many past studies (e.g., Bjørnholt et al., 2016; Lu and Willoughby, 2015; Raudla, 2012), fiscal stress, low-quality budget laws and partly unfavorable task and power structures relate to low PI use, and these factors were present in Kangasala. The results of this study support previous findings that organizational culture emphasizing the importance of PI will drive PI use (e.g., Saliterer and Korac, 2013). Information quality, trust in PI providers and well-functioning information systems also contributed significantly to politicians' PI use in Kangasala. These results correspond with Pollitt (2006) statement that information providers need to be credible if information is to be used. Our findings also support the claim that information use is affected by the quality of information (Heinrich, 2012), where information seen as misleading or inadequate tends to get ignored. However, future studies should test the solutions we found in this case study to determine whether these practices promote PI use in other contexts. Further studies are also necessary to build theory on the drivers of political–administrative performance dialogues (e.g., Laihonen and Rajala, 2020). To date, very little has been known about what promotes successful political–administrative performance dialogues over the long run.

The main contributions of this study is the identification of relationships between constructive political culture, well-functioning political–administrative performance dialogues, active and trustworthy information providers, efficient information channels, and high-quality PI. Overall, Kangasala's success story demonstrates that a constructive culture is an important factor in establishing PI use among politicians, but other factors, such as information quality and organizational structures, also promote a constructive culture. Indeed, the road to a constructive culture was not as straightforward in the examined case. Kangasala struggled with financial difficulties in the past, especially 2013–2017. Many austerity policies were adopted during these difficult years, and cutback management was widely used. One key turning point in the process of changing the negative spiral in Kangasala occurred in 2013. Instead of hiring an external consultant to handle cutback management, which is often the case in Finnish municipalities, Kangasala created a new organizational culture based on extensive cooperation between councillors and the administration. Through this cooperation, Kangasala attempted to find solutions for balancing the fiscal deficits and solving the problems in service production. A new chief executive was hired in 2014 to lead this cultural change, when the new more constructive culture in Kangasala began. This culture encouraged discussion and questions about PI and its meanings in a respectful manner. However, this culture would not been achieved if:

1. Information channels and politico–administrative dialogue did not work properly,

2. Dialogue forums and necessary information production and sharing tasks did not exist in organizational structures,

3. PI was unreliable, invalid, or consistently produced too late,

4. Politicians did not trust and respect the information provider, and

5. Information providers and information brokers did not listen to politicians and respond actively and respectfully to their information needs with additional information and improvements in information quality.

In the time of writing this, Kangasala has turned its course. For example, they have found ways to balance the expenses and incomes. Moreover, several problems in service production have been solved. The PI use has contributed to these positive developments. The success in Kangasala demonstrates how to advance PI use among politicians who have often been reluctant to use such information in decision-making aiming for improvements in public sector services. Hopefully, these encouraging lessons will help others develop PI use in political decision-making in the post-truth era.