Policy preferences in response to negative economic prospects of COVID-19: A survey-experiment among local politicians in four European countries.

This study investigates how the negative economic prospects of the COVID-19 pandemic affect local government politicians' policy preferences in the Netherlands, the United Kingdom, Switzerland, and Spain. The study examines to what extent politicians prefer increasing the role of government (directive state), transferring public tasks to private sector organizations (hollow state), transferring public tasks to third sector organizations (communitarian state), or downsizing and reducing the role of government without transferring tasks (coping state). The experiment primes decision-makers on the pandemic's negative financial and economic prospects vis-à-vis its impact on health and well-being. When negative economic prospects are emphasized, the study finds decreased preferences for a directive state and increased preferences for a coping state. The study concludes that how decision-makers interpret the nature of a crisis determines their preferred response: An emphasis on the negative economic prospects of the COVID-19 pandemic is likely to increase preferences for renewed policies of austerity.

INTRODUCTION

The COVID‐19 pandemic confronts public administrations around the world with a “multilayered crisis.” The pandemic constitutes a health crisis, as evidenced by increased demands on health care systems and high levels of fatalities (Entress et al., 2020; WHO, 2020). The pandemic can also be characterized as a social or humanitarian crisis (Cepiku et al., 2020), in which constituents of public administrations experience a wide range of negative consequences concerning for instance well‐being, opportunities for employment and education, social cohesion, and mental health. An emerging line of public administration scholarship points in particular to social inequities that characterize the distribution of such negative consequences (Gaynor & Wilson, 2020; Martin‐Howard & Farmbry, 2020; Sancino et al., 2020; Wright & Merritt, 2020). The pandemic is also expected to set in motion a global recession through the constraints that the pandemic—as well as policy measures taken to contain the spread of the virus—places on the economy of affected countries. As a consequence, the pandemic has pronounced financial and fiscal consequences for public administrations (Maher et al., 2020; Wilkins et al., 2020).

Initial policy responses of national governments around the world have consisted of unprecedented fiscal, monetary, and macro‐financial measures to limit the human and economic impacts of the pandemic (for a detailed overview, see: IMF, 2020). However, subnational governments are argued to be “at the forefront of the COVID‐19 health crisis and its social and economic consequences” (OECD, 2020) as they have pronounced responsibilities in the policy domains of health and social welfare, as well as the local economy. Furthermore, local governments are expected to experience a scissors effect, brought on by a decrease in financial revenue and a simultaneous increase in expenditure. For instance, after the 2008 financial crisis, social protection spending by local governments in European countries increased by 6.4% in 2009 and by 3.5% in 2010 (OECD, 2020). This forces local government decision‐makers to make “hard choices” concerning the continuation of public services and balancing the budget (Cepiku et al., 2016; Levine, 1978, 1979; Raudla et al., 2015; Van der Voet, 2019). As with prior economic shocks, the anticipated negative economic effects of the COVID‐19 pandemic again make cutback management a key issue for political decision‐making as well as for public administration scholarship.

In the wake of economic recessions in the 1970s and the 2008 financial crisis, public administration researchers have intensively examined processes of austerity management, organizational decline, and cutback management (Bozeman, 2010; Cepiku et al., 2016; Pandey, 2010; Schmidt et al., 2017). This had led to a range of concepts and typologies that have been used to characterize the responses of decision‐makers, including proportional versus targeted strategies (Levine, 1978; Raudla et al., 2015) and open versus closed strategies (Van der Voet, 2019). An alternative typology is put forward in Lodge and Hood's (2012) framework of the directive, hollow, communitarian, and coping state. As will be outlined more extensively in the following section, the directive state is characterized by an increase of public expenditure and government intervention, whereas the other approaches consist of the privatization of public tasks to the private sector (hollow state), the transfer of state responsibilities to nonprofit, volunteer, and charity organizations (communitarian state), and the reduction of public expenditure and government intervention without replacement by private or third sector (coping state). As discussed by Cepiku et al. (2016), this framework may be utilized to examine policy responses to economic downturn, as past external shocks have shown that decision‐makers are likely to update their preferences in response to an economic recession (Lodge & Hood, 2012).

In this article, the framework by Lodge and Hood (2012) is utilized to examine the impact of the negative economic prospects of the COVID‐19 pandemic on the policy preferences of local politicians. Lodge and Hood (2012) show that, in response to the 2008 economic recession, substantial variation can be witnessed between OECD countries. These differences may be interpreted as originating from different degrees to which the countries were afflicted by the economic recession, as well as differences in administrative systems and historical trajectories of administrative reform (Hammerschmid et al., 2016; Pollitt & Bouckaert, 2017). The study therefore examines and compares political decision‐makers in four European countries: the United Kingdom, the Netherlands, Switzerland, and Spain. In examining variation in the policy preferences of politicians on the individual level of analysis, explorative attention is given to how policy preferences in response to negative economic prospects are related to politicians' political ideology. The research question of this article is: How do the negative economic prospects, brought on by the COVID‐19 pandemic, impact the policy preferences of local government politicians, and how are politicians' responses related to their political ideology?

The research question is investigated by means of a survey‐experiment on the level of individual political decision‐makers in local government. As is explained in more detail in the section on methodology, the survey‐experiment utilizes the multilayered nature of the COVID‐19 crisis by priming and informing respondents that are randomly assigned to the financial impact group on the future, negative economic consequences of the pandemic for local government, whereas a health impact group is primed on and informed about the current consequences of the pandemic for citizen health and well‐being. Compared to prior research on cutback management, which draws heavily on qualitative case studies (Cepiku et al., 2016; Overmans & Timm‐Arnold, 2016; Raudla et al., 2017), a survey‐experiment can inform the causal effect of negative economic prospects on austerity preferences, because issues of endogeneity are addressed through random assignment of participants to experimental conditions. As such, the study provides a complimentary type of evidence regarding policy responses to economic downturns.

An analysis on the micro‐level may inform the preferences of a relatively large number of participants. In this article, a total of 2035 local council members from the four countries have participated in the survey‐experiment. Furthermore, by examining preferences in response to future prospects, insights on current, fast‐moving developments amidst the pandemic can be analyzed. The study is thereby able to inform the policy preferences of political decision‐makers in the “heat of the moment,” as the experimental analysis of individual preferences requires a shorter time span than retrospective case studies of organizational policy. Although individual preferences are a key input to group‐level decision‐making (March & Simon, 1993), these may not necessarily directly translate into future organizational decisions and enacted policy.

The article proceeds as follows. The following section addresses how negative economic prospects may shape policy preferences of decision‐makers, and formulate theoretical expectations. The methodology section introduces the design and procedures of the experiment, the characteristics of participants of the experiment, as well as the measurement of politicians' policy preferences. Subsequently, the analysis and findings are discussed. The article concludes by discussing the results in light of the literature on crisis management and cutback management.

THEORETICAL FRAMEWORK AND HYPOTHESES

In their framework of government responses to economic downturn, Lodge and Hood (2012) identify four approaches through which adjustments can be made to cope with economic shocks. These are labeled the directive state, the hollow state, the communitarian state, and the coping state.

First, the directive state reflects a response to an economic shock by increasing, or returning to, a large role of government in the economy and society. In light of economic crises in prior years, governments around the world have frequently relied on rescue packages, economic stimuli and bailouts in order to attenuate negative economic impacts. An example of such a stimulus package is the European Economic Recovery Plan that was implemented in response to the 2008 financial crisis (Coenen et al., 2013). A reliance on hierarchy rather than the market reflects a tendency of public administrations to centralize service provision in times of crisis, in order to guarantee continuity and safeguard the common interest. Second, the hollow state concerns a response which stresses an increased reliance on market actors in the execution of public tasks and the delivery of public services. By privatizing and contracting out public services to private providers, governments can attempt to reduce state‐spending and improve efficiency through which services are delivered (Alonso & Andrews, 2016; Hefetz & Warner, 2004). A third approach is to transfer the provision of public services away from public bureaucracies and private firms to local communitarian organizations, such as nonprofit organizations and charity organizations and the involvement of citizens as co‐producers. Thereby, the communitarian state is an approach that aims to activate local participation and reduce cost‐expenditure of public services, for instance through the initiation of citizen co‐production (Bovaird, 2007; Whitaker, 1980). Fourth and finally, the coping state consists of the abandonment of responsibilities and diminishment of public service provision. The coping state most closely resembles policies of austerity, as it aims to cut public expenditure by means of downsizing and limiting the delivery of public services. Although such an approach may seem like a politically risky strategy, the coping state approach has been in evidence in response to recent economic shocks (Kickert & Randma‐Liiv, 2017; Randma‐Liiv & Kickert, 2017). Lodge and Hood (2012, p. 83) argue that all four policy approaches may potentially be in evidence in response to economic downturn: “Indeed, local and central responses to fiscal stress in the OECD world to date include cases of each of these four scenarios—that is, a “directing state,” a “hollow state” captured by business, a “local communitarian state” moving beyond both public and corporate private sector provision, and a (barely) “coping state.”

Analytically, the framework by Lodge and Hood (2012) is informative because it highlights that multiple policy approaches can be favored in response to an economic downturn. From the perspective of public administration research, the typology highlights that the different policy measures result in a range of possible value trade‐offs (Hood, 1991). For instance, the directive state may result in an increase in resilience and safeguard equitable treatment of citizens, but may fail to meet ambitions to increase cost‐effectiveness. The hollow state advocates greater frugality and value‐for‐money, but may result in rent‐seeking, cream skimming, and reduced service quality (Elkomy et al., 2019; Jilke et al., 2018). The communitarian state likewise may be an attractive approach to reduce government‐spending and improve responsiveness to local needs, but its reliance on local and informal structures may limit equality and pose challenges in terms of accountability, effectiveness, and inclusion (Aschhoff & Vogel, 2018; Campanale et al., 2020; Jaspers & Steen, 2019). At its core, the coping state concerns a trade‐off between efficiency and service quality, which may disproportionately affect citizens from lower socio‐economic backgrounds or otherwise vulnerable populations (e.g., Levine, 1978). This makes preferences for these policy approaches a value trade‐off, which can be expected to be highly dependent on political decision‐makers' political orientation.

On the level of individual decision‐makers, a range of theories provide insights into how negative economic prospects may impact the policy preferences of political decision‐makers. Research on behavioral public administration tends to have a past‐looking perspective, by drawing on theories such as the behavioral theory of the firm (Holm, 2018; Hong, 2019; Nielsen, 2014) and performance feedback theory (Salge, 2011). Such theoretical perspectives argue that decision‐makers rely on performance relative to historical and social aspirations (Greve, 2003). When prior performance fails to meet aspired levels of performance, decision‐makers respond by initiating reform and policy change that are expected to remedy the performance shortfall (Meier et al., 2015).

Alternative theoretical frameworks of decision‐making are explicitly forward‐looking rather than retrospective. Most prominently, prospect theory states that the degree in which decision‐makers take risks is a function of the situation in which decision‐makers find themselves. A situation of impending economic downturn may be interpreted as the “domain of loss,” in which decision‐makers are expected to be more prospective and risk‐taking (Kahneman & Tversky, 1979). Prospect theory may thus be applied to understand why decision‐makers update their policy preferences in the face of diminished future prospects. Strategic reference point strategy extends prospect theory by taking into account organizational rather than individual level reference points (Shinkle, 2011). The theory states that decision‐makers actively and continuously intervene in strategy and policies, in light of external developments and future ambitions (Fiegenbaum et al., 1996). An external shock that brings about negative economic prospects is therefore likely to trigger a response by decision‐makers that results in updated policy preferences.

Lodge and Hood (2012) invoke Wildavsky's (1985) observation that for every reason used to argue why an economic downturn may result in an increasing role of government, there is an equally plausible reason why the role of government should be reduced. Therefore, a hypothesis is specified that negative economic prospects may positively affect any of the four policy positions specified in the framework by Lodge and Hood (2012). The following hypothesis is formulated:

Negative economic prospects positively affect political decision‐makers' preferences for a directive state, hollow state, communitarian state, and coping state.

Lodge and Hood (2012) state that the direction of the effect of negative economic prospects on policy responses is highly context‐dependent. Their analysis of policy responses to the 2008 financial crisis in OECD countries reveals high variation, among others brought on by differences in the extent to which countries were economically affected, administrative systems, and political orientation of the national government. In line with these findings, it is argued here that, on the individual level of analysis, political ideology is major determinant of the preferred policy response to negative economic prospects brought on by the COVID‐19 pandemic.

Political ideology provides a compass through which decision‐makers connect signals of problems to potential solutions. Depending on political ideology, individuals rely on different cognitive and motivational processes to select and justify policy preferences (Skitka & Tetlock, 1993). Studies on politically motivated reasoning provide evidence that political ideology not only determines the preferred policy responses of political decision‐makers, but that decision‐makers interpret information about future prospects in such a way that it congruent with their politically preferred solutions (Baekgaard et al., 2019; Christensen & Moynihan, 2020).

A relatively sizeable body of research has addressed the relationships between political ideology and the implementation of hollow state practices such as contracting out and privatization in local government. Alonso and Andrews (2020) provide an overview of research that indicates that more right‐leaning government coalitions are more likely to contract out services, although there are also several studies that report that ideology makes little difference. Using a quasi‐experimental design, Alonso and Andrews (2020) complement the available evidence with the finding that local governments with a left‐leaning coalition exhibit an aversion to the involvement of private actors. The relationship between political ideology and communitarian state arrangements has not received ample research attention. In a study of local government election programs, Fattore et al. (2012, p. 225) report that more left‐leaning political parties are more likely to refer to governance arrangements that stress the interaction between government and society. The hollow and communitarian state approach both allow local governments to offset reduced financial capacity by utilizing the resources of external actors. Given the hesitation of left‐leaning political decision‐makers to involve market actors in the provision of public services, it is expected that their preference—next to the directive state—is to involve nonprofit actors such as charities, associations and citizens in the service provision process in order to signal that they reject economic efficiency as the primary criterion that underlies public service provision.

It is therefore hypothesized that more right‐leaning politicians are more likely to favor the contracting out and privatization of public services (hollow state) and downsizing, cuts and reductions (coping state) in response to negative economic prospects, whereas more left‐leaning politicians prefer the expansion of state intervention in the economy (directive state) as well as increased collaboration with third sector organizations to maintain public service delivery (communitarian state). The following hypothesis is formulated:

The effect of negative economic prospects on policy preferences is moderated by political ideology, in such a way that a more right‐leaning political ideology weakens the positive effect of negative economic prospects on preferences for the directive and communitarian state, and strengthens the positive effect of negative economic prospects on preferences for the hollow and coping state.

METHODOLOGY

Research design, procedures, and measures

The study is based on an experimental design that was not pre‐registered, and relies on randomized assignment of participants to a baseline condition about health impacts and a treatment condition about the financial impacts of COVID‐19. When faced with a crisis, a first task for decision‐makers is to make sense of the situation: “What is this crisis about?” (Ansell & Boin, 2019). As outlined in the introduction section, the COVID‐19 pandemic can be characterized as health crisis, but the pandemic also has pronounced social and economic consequences. Because of this multilayered nature, the COVID‐19 pandemic is arguably characterized by greater uncertainty and severity than the economic crises of 2008 and the 1970s. This study utilizes the multilayered nature of the COVID‐19 pandemic to prime participants of the experiment on two distinct interpretations. A health impact group is primed on the negative consequences of what is perhaps the main and initial characteristic of the COVID‐19 pandemic: Its negative implications for the health and well‐being of citizens. The financial impact group is primed on the negative effects of the COVID‐19 pandemic for the economic and financial prospects of the councilor's municipality.

An overview of the experimental procedure is given in Figure 1. After providing consent to participate in a scientific study, participants are randomly assigned to the health impact or financial impact condition. At the time the study was conducted (June 2020), relatively little information was available about the economic and financial consequences for local governments, and the implementation of health measures took center stage. The health impact condition therefore serves as a baseline‐control condition to the experimental treatment of financial and economic prospects. The experimental manipulation relies on two steps. First, respondents in the health impact group are primed to consider the pandemic's health and well‐being effects for citizens, whereas respondents in the financial impact group are primed to consider the pandemic's future effects on the economic and financial situation of the municipality. Priming is an experimental technique aimed at activating memory of and association with a concept in a unobtrusive manner (Epley & Gilovich, 1999). Respondents in the health impact group were asked to respond to the following open‐ended question: “In your view, what were the most important ways in which the COVID‐19 pandemic has affected the health and well‐being of the citizens in your municipality in the past months?” Respondents in the financial impact group were asked to respond to the question: “In your view, what are the most important ways in which the COVID‐19 pandemic affects the financial and economic situation of your municipality for the next years?” On average, respondents in the health impact group wrote a response consisting of 172 characters and respondents in the financial impact group wrote a response that averaged 176 characters, indicating that respondents actively considered these questions and associated with the two distinct topics.

FIGURE 1

Overview of experimental procedures

Second, respondents in both experiment groups were given an informative text. Respondents in the health impact group received retrospective factual information about the health and well‐being consequences of the COVID‐19 pandemic, based on information by the World Health Organization (WHO, 2020). Respondents in the financial impact group received forward‐looking information about the anticipated consequences for the economic and financial situation of local governments, based on information from the Organization for Economic Collaboration and Development (OECD, 2020). Both texts are given in the Table 1.

TABLE 1

Text for health impact and financial impact condition

Text for health impact condition	Text for financial impact condition
According to the World Health Organization (WHO), the COVID‐19 pandemic has had important consequences for the health and well‐being of citizens. The WHO indicates that, as a consequence of the COVID‐19 pandemic:	According to the Organization of Economic Cooperation and Development (OECD), local government organizations are at the frontline in bearing the financial and economic challenges of the COVID‐19 pandemic. The OECD predicts that local government organizations, as a consequence of the COVID‐19 pandemic:
1. As of June 2, over 6 million people world‐wide have contracted the virus, with over 373,000 confirmed deaths.	1. Are likely to experience a “scissors effect” of decreasing tax revenues and rising expenditures.
2. Particularly individuals of 60 years of age and higher are at risk, as well as individuals with health conditions such as lung or heart disease, diabetes, or conditions that affect their immune system.	2. Must likely strongly increase expenditures regarding social protection and social welfare. After the 2008 financial crisis, social protection spending by local governments in European countries increased by 6.4% in 2009 and by 3.5% in 2010.
3. In mental health terms, the main psychological impact to date is higher stress or anxiety, and levels of loneliness and depression are expected to rise.	3. Are likely to face rising unemployment, business bankruptcies and increased municipal debt.

It is unlikely that political decision‐makers in the selected countries have remained fully unexposed to information about the COVID‐19 pandemic prior to participation in this experiment. Such prior exposure is inherent to survey‐experiments that address salient societal issues: “either there is a likelihood of contamination from real‐world experience or the survey experiment explores a nonexistent or politically irrelevant phenomenon” (Gaines et al., 2007, p. 12). Such variation is accounted for through random assignment to experimental conditions, and the presence of prior knowledge and views underlines the appropriateness of priming as an experimental technique.

Measurement of policy preferences

After the two interventions a brief description of the four policy responses by Lodge and Hood (2012) was presented to respondents in random order. These are summarized in the Appendix. Respondents were then asked to distribute 100 points across the four responses—again in random order—to indicate their preferred response using a slider scale. The four responses are conceptually distinct, but can in practice be combined into a policy response (Cepiku et al., 2016; Lodge & Hood, 2012). The distribution of a finite number of points was therefore selected as a measure, because it allows respondents to indicate their preference for a policy approach vis‐à‐vis the other approaches. In our view, this is preferable to, for instance, the use of Likert scales in which each of the four approaches can be preferred to a maximum extent, or a measure that merely informs which is a respondent's most preferred policy response.

Covariates

Following the measure for policy preferences, a number of demographic covariates were measured in order to examine randomization across experimental groups. First, gender was measured and transformed into a dummy variable where female gender is coded as “1,” whereas the other options (male gender; other gender; undisclosed) were coded as “0.” Second, age was measured in years. Third, respondents were asked to identify themselves as member of the coalition or majority in the municipality council, vis‐à‐vis the opposition or minority. This variable was coded into a dummy variable where coalition/majority membership was coded as “1.” Fourth, political ideology was measured with the item “What is your political position regarding economic policy issues?” Responses were indicated on a five‐point scale ranging from “1: Strongly left” to “5: Strongly right.”

Research setting, characteristics of respondents, and experimental balance

The study was conducted in four countries: The Netherlands, the United Kingdom, Switzerland, and Spain. By the time the study was conducted, all four countries had experienced severe negative economic impacts of the COVID‐19 pandemic. Compared to gross domestic product in the same quarter of the prior year, Eurostat (2020) reports reductions of 22.1% for Spain, 21.7% for the United Kingdom, 9% for the Netherlands, and 9.4% for Switzerland in the second quarter of 2020.

In many regards, local governments in the four selected countries face many similar constraints and requirements with regards to financial management and financial autonomy. For instance, local governments in all four countries provide a broad range of services, and must comply with the requirement for budget balancing. Although local governments have varying possibilities to utilize loans in order to facilitate service provision or investments, debt‐limits are in place. Following the 2008 financial crisis, strong oversight by higher level governments (provinces, cantons, or the state) is in place. A marked difference between the Netherlands and the United Kingdom vis‐à‐vis Spain and Switzerland is that revenue in the latter countries is for a greater extent based on local taxation rather than state transfers. In recent years, municipalities in the Netherlands and the United Kingdom have faced increased fiscal pressure due to decreasing levels of vertical financial transfers (for a detailed comparison, see: Geißler et al., 2019).

A total of 2035 municipal council members from the Netherlands, the United Kingdom, Switzerland, and Spain fully participated in the experiment. Participants were invited by utilizing publicly available email addresses. The experiment was conducted to participate in the experiment between 21 June 2020 and 28 July 2020, with one reminder sent 2 weeks after the initial invitation. The public availability of email addresses and response rate differs strongly between the four countries. Although the response rate for the Netherlands is comparable with comparable with prior survey research in this target group, the number of responses in the other three countries are significantly lower.

For the Netherlands, an invitation was sent to 8216 available email addresses of the population of 8261 municipal council members. A total of 1333 complete responses were obtained, indicating an effective response rate of 16.2%. Of these responses, 608 respondents were randomly assigned to the health impact group and 725 were assigned to the financial impact group. The main party affiliations of the Dutch respondents are volkspartij voor vrijheid en democratie (VVD) (36.7%), christen‐democratisch appèl (15.1%), partij van de arbeid (9.2%), D66 (8.2%), GroenLinks (8.2%), and local parties (12.0%). This distribution implies a relatively close approximation of the distribution of seats in the population of council members (VNG, 2020), with an overrepresentation of the VVD party and an underrepresentation of council members representing local parties. Female council members account for 26.9% of the sample, and the average age is 58.2 years.

For the United Kingdom, an invitation was sent to the 15,000 available email addresses of English local council members, resulting in 212 complete responses (response rate of 1.4%), of which 101 were assigned to the health impact group and 111 to the financial impact group. The main party affiliations are Labor (34.9%), Liberal Democrats (30.2%), and Conservative (23.6%), signaling an underrepresentation of the Conservative representatives. The average age of respondents is 60.5 years and female representatives account for 23.6% of the sample.

Switzerland consists of 2202 local authorities, with a median of 1060 inhabitants. In order to balance feasibility and comparability, only those 1434 municipalities with an inhabitant number of 1000 or more were sampled. In these municipalities, contact information for 495 municipal council members was available. In addition to these contacts, an invitation was sent to the general email address of all 1434 municipalities. The survey was distributed in French, German, and Italian, corresponding to the three language regions in Switzerland. This resulted in 264 complete responses (137 assigned to the health impact group, 127 assigned to the financial impact group). The main party affiliations in the sample are Social Democratic Party of Switzerland (15.5%), Liberal Party (12.5%), and local parties (36.4%). The average age of respondents is 53.1 years, with 29.7% of respondents are female.

For Spain, public email addresses for municipal councilors are scarcely available for its 8112 municipalities. In all, 4850 individual email addresses were collected for the 335 largest municipalities in Spain, complemented with 209 general email addresses of the municipality. This resulted in a total of 226 responses from Spanish local council members, of which 120 were assigned to the health impact group and 106 to the financial impact group. The average age of Spanish respondents is 47.8 years and 45.7% of respondents are female. The main party affiliations in the sample are epartido socialista obrero español (30.1%), partido popular (10.2%), Ciudadanos (10.2%), and local parties (32.3%).

Given these characteristics, it is important to emphasize that the experiment is thus characterized by methodological trade‐off: On the one hand, it offers a unique insight in how negative economic prospects impact political preferences at the height of the COVID‐19 pandemic using an experimental design that offers high internal validity. On the other hand, given the timing of the experiment during this time period, the number of responses is relatively low and external validity of the findings may not be warranted.

Balance check

In order to check for successful random assignment to experimental condition, the two groups were compared on a number of covariates. This analysis uncovers no statistically significant differences concerning gender composition, age, membership of coalition/majority versus opposition/minority, and political ideology. The balance check thus fails to reject the null hypothesis that the experimental groups are balanced with respect to these observed characteristics (Table 2).

TABLE 2

Balance check

	Health impact group (n = 966)	Financial impact group (n = 1069)	Full sample (n = 2035)	t‐test
Female gender	0.29	0.28	0.29	t = 0.17; df = 1985.70; p‐value = 0.86
Age	57.9	55.5	56.6	t = 1.11; df = 996.43; p‐value = 0.27
Coalition	0.55	0.54	0.55	t = 0.40; df = 1937.30; p‐value = 0.69
Political ideology	2.74	2.75	2.74	t = −0.05; df = 1980.10 p‐value = 0.96

ANALYSIS AND RESULTS

In this section, the results of the analysis are reported. First, descriptive statistics for the preferred policy responses to the COVID‐19 pandemic of local politicians in the four countries are given. Second, the effect of negative economic prospects on policy preferences is estimated. Third, the moderating role of politicians' political ideology is examined.

Table 3 summarizes the mean values and standard deviations for politicians' preferences for the directive state, hollow state, communitarian state, and coping state across the four countries. Overall, politicians favor the directive state (39.4/100) and communitarian state (31.3/100), and indicate a lower preference for the hollow state (17.2/100) and coping state (12.1/100). Substantial differences exist between the four countries concerning the directive state (p < 0.000), the communitarian state (p < 0.000), and the coping state (p < 0.000). Among Dutch politicians, preferences for a directive state are significantly lower than in the other countries, and preferences for a coping state are significantly higher. In Switzerland and Spain, politicians prefer the communitarian state to a lesser extent than politicians in the Netherlands and the United Kingdom. An interesting finding is that the country variable does not explain any differences concerning preferences for the hollow state (p = 0.923). This result suggests that preferences to privatize and contract out public tasks are equally prevalent among local government politicians in the four countries. The mean values for policy preferences are visualized in Figure 2, including 95% confidence intervals for the country differences.

TABLE 3

Policy preferences per country

	Directive state	Hollow state	Communitarian state	Coping state
The Netherlands	34.5 (27.0)	17.3 (18.2)	34.0 (20.4)	14.2 (18.0)
The United Kingdom	43.0 (27.7)	16.1 (18.6)	32.1 (18.4)	8.8 (16.6)
Switzerland	49.4 (29.9)	18.1 (19.1)	23.8 (19.0)	8.7 (13.8)
Spain	53.2 (27.1)	16.8 (19.0)	23.2 (16.2)	6.9 14.7
Total sample	39.4 (28.4)	17.2 (18.4)	31.3 (20.1)	12.1 (17.3)
p‐Value	0.000	0.923	0.000	0.000
Adjusted R 2	0.06	0.00	0.05	0.03

Note: Mean values are reported. Standard deviations are reported in parentheses.

FIGURE 2

Policy preferences per country

In order to examine the effect of negative economic prospects on policy preferences, a regression analysis is conducted. A dummy variable for negative economic prospects is modeled, indicating membership in the experimental condition. As country membership is not equally distributed among the four countries, the analysis controls for nationality by modeling dummy variables for the Netherlands, Switzerland, and Spain. The United Kingdom functions as the reference category. The results are reported in Table 4.

TABLE 4

Regression analysis

	Directive state	Hollow state	Communitarian state	Coping state
Intercept	44.94*** (1.99)	15.81*** (1.34)	31.81*** (1.42)	7.44*** (1.23)
Treatment: Negative economic prospects	−3.76** (1.22)	0.52 (0.82)	0.63 (87)	2.61*** (0.76)
The Netherlands	−8.41*** (2.03)	1.21 (1.36)	1.88 (1.45)	5.33*** (1.26)
Switzerland	6.30* (2.53)	2.02 (1.70)	−8.31*** (1.81)	−0.01 (1.57)
Spain	9.98*** (2.62)	0.71 (1.77)	−8.95*** (1.87)	−1.74 (1.63)
Reference: United Kingdom
Adjusted R 2	0.07	0.00	0.05	0.03

Note: Unstandardized regression coefficients are reported. Standard errors are given in parenthesis.

p < 0.05.

p < 0.01.

p < 0.000.

First, the results indicate that negative economic prospects are negatively related to preferences for a directive state. Relative to the health impact group, respondents in the financial impact group indicate a preference for the directive state that is on average 3.76 points lower. In accordance with the results in Table 3, the regression analysis indicates that politicians in the Netherlands have a lower preference for a directive state relative to the United Kingdom, whereas politicians in Switzerland and Spain indicate a higher preference for a directive state.

Second, Table 4 indicates that negative economic prospects have a positive effect on preferences for a coping state. On average, respondents in the financial impact group allocate 2.61 more points to a coping state policy than respondents in the health impact group. Preference for a coping state is on average 5.33 points higher among Dutch than British politicians, with no differences between the United Kingdom, Switzerland, and Spain.

In order to examine the robustness of the effect of negative economic prospects on preferences for the directive state and the coping state, the effects of negative economic prospects are analyzed separately for the four countries by means of t‐tests. The results are visualized in Figure 3. The upper graph in Figure 3 indicates that the preference of Dutch politicians decreases by 3 points in response to negative economic prospects (t = 2.04, df = 1302.50, p‐value = 0.04). For politicians in the United Kingdom, an 8.7‐point decrease is present (t = 2.32, df = 209.51, p‐value = 0.02), signaling weakened preferences for a directive state in response to negative economic prospects. Although a decreasing preference for the directive state is also witnessed in Switzerland (4.3 points) and Spain (2.8 points), these differences are not statistically significant (for Switzerland: t = 1.18, df = 258.52, p‐value = 0.24; for Spain: t = 0.76, df = 217.12, p‐value = 0.45).

FIGURE 3

The effect of negative economic prospects on policy preferences per country

The lower graph in Figure 3 shows stronger preferences for a coping state approach in the financial impact group vis‐à‐vis the health impact group across the four countries, but similar to the effect of negative economic prospects on preferences for a directing state, these differences are only statistically significant for the Netherlands and the United Kingdom. Dutch politicians' preference for a coping state increases by 2.7 points (t = −2.70, df = 1324.2, p‐value = 0.01), while British politicians show a 5.1 increase (t = −2.28, df = 197.21, p‐value = 0.02). For Switzerland, the increase is 2.2 points but is not statistically significant (t = −1.31, df = 258.79, p‐value = 0.19). The Spanish sample shows the smallest increase in preference for a coping state, which is statistically indistinguishable from 0 (t = −0.24, df = 215.42, p‐value = 0.81).

The conclusion of this part of the analysis is thus that Dutch and British decision‐makers who were assigned to the treatment that highlights negative economic prospects report a lower preference for the directive state and a higher preference for the coping state than their peers who were assigned to the treatment that highlights health and well‐being, but that this effect is absent among Swiss and Spanish political decision‐makers. An interesting result, which is further addressed in the discussion section, is that the negative economic prospects of the COVID‐19 pandemic appear to result in a trade‐off between the directive state and the coping state, but that policy approaches that aim to decrease public expenditure but maintain public value through partnership with private sector or third sector actors are not considered as viable options.

Next, variation on the level of individual council members is explored by bringing in a measure of political ideology, as well as an interaction term for the interaction between negative economic prospects and ideology. This analysis allows to examine if politicians' preferred policy responses to negative economic prospects differ by political ideology. The results of this analysis are given in Table 5.

TABLE 5

Interaction effects of negative economic prospects and political ideology

	Directive state	Hollow state	Communitarian state	Coping state
Intercept	63.76*** (2.60)	3.20 (1.76)	32.71*** (1.96)	0.33 (1.66)
Treatment: Negative economic prospects	−2.88 (2.82)	1.51 (1.91)	1.97 (2.12)	−0.60 (1.80)
The Netherlands	−6.64*** (1.93)	−0.08 (1.31)	1.94 (1.46)	4.78*** (1.24)
Switzerland	10.00*** (2.41)	−0.27 (1.63)	−7.84*** (1.82)	−1.89 (1.54)
Spain	6.02* (2.51)	3.15 (1.70)	−9.21*** (1.89)	0.04 (1.60)
Reference category: the United Kingdom
Political ideology	−7.28*** (0.68)	4.93*** (0.46)	−0.37 (0.51)	2.73*** (0.43)
Interaction effect: Negative economic prospects × political ideology	−0.39 (0.94)	−0.39 (0.63)	−0.44 (0.71)	1.22* (0.60)
Adjusted R 2	0.17	0.09	0.05	0.09

Note: Unstandardized regression coefficients are reported. Standard errors are given in parenthesis.

p < 0.05.

**p < 0.01.

p < 0.000.

Table 5 shows that political ideology is negatively related to preferences for a directive state. This indicates that the more right‐leaning politicians have lower preferences for a state that plays a large, intervening role in the economy. Furthermore, politicians with a right‐leaning political orientation more strongly prefer a hollow state response in which public tasks are privatized and contracted out to private businesses, as well as a stronger preference for a coping state that reduces expenditure and limits public service delivery. Political ideology is unrelated to preferences for a communitarian state approach.

Whereas the abovementioned results inform politicians' their general preferences for the role of government, the interaction term reveals to what extent variation in political ideology explains divergent effects of negative economic prospects on policy preferences. The results in Table 5 indicate that no statistically significant interaction effects exist for the directive state, hollow state, and communitarian state. For the coping state, a statistically significant interaction effect is identified. The marginal effects of this interaction effect are visualized in Figure 4 using the Interflex package for r (Hainmueller et al., 2019). The marginal effects plot indicates that the positive effect of negative economic prospects on the coping state increases as political orientation becomes more right‐leaning. For value “1” (Strong left), the effect of negative economic prospects cannot be distinguished from 0. For a value of “2” (Left), an effect exists that is weaker in strength than the average effect size of 2.61. For the highest value on the political ideology scale (“5: Strong right”), the effect size of negative economic prospects is roughly twice as large as for the politicians with a centrist political orientation (“3”).

FIGURE 4

Marginal effects of negative economic prospects for different levels of political ideology

DISCUSSION AND CONCLUSION

This study has examined the preferred policy responses of local government politicians to the negative economic prospects of the COVID‐19 pandemic in the Netherlands, the United Kingdom, Switzerland, and Spain. In doing so, it complements recent research on how citizens evaluate policy responses taken in response to COVID‐19 (e.g., Heinzel & Liese, 2021). As policy responses to the pandemic are quickly being enacted in countries across the world, this study has attempted to gain insight into these dynamics by collecting data at a time when political decision‐making was taking place. Combined with the experimental design that utilizes the multilayered nature of the COVID‐19 pandemic, this results in a unique dataset that allows for the examination of the effect of negative economic prospects on the policy preferences of local level politicians.

The main finding of the study is that an emphasis on the negative economic prospects brought on by the pandemic result in a trade‐off between a directive state, in which the government plays a large and intervening role in the economy, and a coping state, in which government reduces its expenditure and limits the delivery of public services. This study finds that political decision‐makers are less likely to express preferences for a directive state when exposed to negative economic prospects, and more likely to indicate a preference for a coping state response. Compared to this group, decision‐makers that were primed on the pandemic's impacts on health and well‐being report a greater preference for the directive state. The results thus show that how decision‐makers interpret the nature of a crisis significantly influences their preferred response: politicians that explicitly considered the pandemic's negative economic prospects report a greater willingness to engage in renewed policies of austerity, cutbacks and downsizing in order to cope with the COVID‐19 pandemic.

Although negative economic prospects increase preferences for the coping state, this study indicates that a directive state approach is on average the most preferred policy response across all four countries. This result resonates with actual policy measures that have been witnessed on the (supra‐)national level of government, which have been largely in line with the directive state approach and include measures such as support for (local) businesses, wage compensation for (self‐employed) workers, increasing access to and extension of unemployment benefits, and deferrals of taxation (IMF, 2020). Comparing the four different countries that were included in the study, it is concluded some variation exists concerning the effect of negative economic prospects. The above‐mentioned willingness to decrease the role of government—as signaled by a decreased preference for a directive state and an increased preference for a coping state—is present in the Netherlands and the United Kingdom, but not in Switzerland and Spain. This result suggests that decision‐makers in more neoliberal economies may respond to the negative economic prospects of the COVID‐19 pandemic with a greater willingness to opt for policies of austerity.

Our study reaffirms that political decision‐makers with more right‐leaning political ideologies express greater preferences for the hollow state (c.f., Alonso & Andrews, 2020). To the public administration literature on co‐production, it adds the empirical finding that preferences for the communitarian state are unrelated to political ideology. An interesting result is that negative economic prospects do not appear to affect political decision‐makers' preferences for increased handing over of public tasks to the private sector (hollow state) or delegation of public tasks to third sector organizations (communitarian state). Although these options allow for a decrease in public expenditure while attempting to maintain public value by increased external collaboration, political decision‐makers appear to opt for austerity instead. In general, the policy preferences of politicians across the four countries do not betray a “comeback” of the New Public Management, as politicians across the four countries report relatively limited preference for increased contracting out public tasks or the privatization of public services. One interpretation of the absence of an effect of negative economic prospects on preferences for a hollow or communitarian state is that increased external collaboration may be deemed infeasible due to time‐constraints amidst a turbulent crisis (Van der Voet, 2019).

Although this study provides a unique insight into the changing policy preferences of actual political decision‐makers, the study is subject to several limitations. First, this study addresses policy preferences on the level of local government, and does not empirically examine policy responses that have been in evidence on the (supra‐)national level of analysis. Furthermore, the experiment aimed to solicit the participations of political decision‐makers to the experiment amidst a highly turbulent time period. The timing of the study has arguably limited hindsight biases and rationalization of past decisions, which are important limitations of retrospective research designs. However, the relatively limited number of responses by British, Swiss, and Spanish council members limits the generalizability of the results to the population of local government councilors. Moreover, the study informs individual policy preferences rather than enacted organizational policy. Although these preferences function as the micro‐foundations of actual policy responses to the COVID‐19 pandemic, they are inputs to processes of collective decision‐making that will not necessarily translate in to enacted local government policy.

Notwithstanding its limitations, the study informs the public administration research literature in several ways. First, the study contributes to the public sector literature on cutback management and crisis management. Whereas this literature has a strong emphasis on qualitative case studies and retrospective analysis on the state and municipal level, this study has informed the policy preferences of individual decision‐makers based on an experimental study about forward‐looking economic prospects. As such, the study provides a complementary type of evidence, with relatively high levels of internal validity due to the experimental design. Second, the study informs research on evidence‐based policy making by demonstrating that preferences of political decision‐makers are subject to the provision of factual information about the (anticipated) consequences of the pandemic (Baekkeskov, 2016; Lundin & Öberg, 2014). This study shows that, compared to decision‐makers who were primed on the negative health and well‐being effects of the pandemic, decision‐makers who were primed on negative economic consequences showed a markedly different response. Whereas the literature on crisis management emphasizes that decision‐makers' sense‐making underlies processes of decision‐making (Ansell & Boin, 2019), this suggests that “sense‐giving” through the provision of factual information by interest groups, bureaucrats, and experts may determine how decision‐makers approach policy responses to the pandemic.

By informing shifting policy preferences in response to a rapidly unfolding crisis, the study provides a glimpse of how local governments might attempt to attenuate the negative economic impacts of the COVID‐19 pandemic. The results suggest that, depending on the political orientation of political decision‐makers, the pandemic may give rise to renewed policies of austerity on the local level. For public administration researchers and practitioners, this signals an opportunity as well as a need for a new upstart of research efforts that informs how high quality public administration and good governance remain viable in a renewed era of financial strain (cf. Cepiku et al., 2016; Van der Voet, 2019), as well as how the negative effects of policy responses on citizens, and in particular the most vulnerable populations, may be attenuated (c.f., Gaynor & Wilson, 2020; Martin‐Howard & Farmbry, 2020; Sancino et al., 2020; Wright & Merritt, 2020).