Introduction
The relationship between economic conditions and political support has long been a focal point of scholarly inquiry (Anderson and Guillory, Reference Anderson and Guillory1997; Lewis-Beck and Stegmaier, Reference Lewis-Beck and Stegmaier2013; Stegmaier et al., Reference Stegmaier, Lewis-Beck, Park, Evans, Arzheimer and Lewis- Beck2017). While a strong economy is generally assumed to foster greater political support and a weak economy to erode it, empirical findings regarding this relationship remain notably mixed (Clarke et al., Reference Clarke, Dutt and Kornberg1993; Dalton, Reference Dalton2004; Anderson and Singer, Reference Anderson and Singer2008; Armingeon and Guthmann, Reference Armingeon and Guthmann2014; Quaranta and Martini, Reference Quaranta and Martini2016). In particular, studies examining citizens’ satisfaction with democracy (SWD)—a widely used attitudinal indicator of regime support—have reported inconsistent results regarding the role of economic performance (Linde and Ekman, Reference Linde and Ekman2003; Norris, Reference Norris2011; Claassen and Magalhães, Reference Claassen and Magalhães2022). This enduring puzzle raises an important question: under what conditions does the economy shape SWD, and why does this relationship appear strong in some contexts but weak or absent in others?
One prominent explanation for the mixed findings in the relationship between economic conditions and political support centers on the concept of clarity of responsibility (CoR) (Powell and Whitten, Reference Powell and Whitten1993). CoR refers to the extent to which political institutions and governing arrangements enable citizens to clearly identify which political actors are accountable for policy outcomes, particularly in the economic domain. When responsibility is concentrated—under conditions such as single-party majority governments or strong presidential leadership—citizens can effectively assign credit or blame, reinforcing the evaluative connection between economic performance and political support (Powell and Whitten, Reference Powell and Whitten1993; Stegmaier et al., Reference Stegmaier, Lewis-Beck, Park, Evans, Arzheimer and Lewis- Beck2017). Conversely, when responsibility is diffused—through coalition governments, minority administrations, divided legislatures, weak party cohesion, or complex institutional veto points—this evaluative link weakens, and citizens struggle to hold specific actors accountable (Stegmaier et al., Reference Stegmaier, Lewis-Beck, Park, Evans, Arzheimer and Lewis- Beck2017). While the theoretical importance of CoR is well established in the economic voting literature, empirical studies applying this framework to explain variations in SWD remain limited and often rely on incomplete or coarse measures of institutional clarity (Criado and Herreros, Reference Criado and Herreros2007; Christensen, Reference Christensen2015; Kestilä-Kekkonen and Söderlund, Reference Kestilä-Kekkonen and Söderlund2017). As a result, the literature has yet to fully resolve the puzzle of when and why economic conditions shape citizens’ SWD.
This study directly addresses this gap by employing a more refined and comprehensive measure of CoR, as proposed by Dassonneville and Lewis-Beck (Reference Dassonneville and Lewis-Beck2017). Their framework moves beyond simplistic indicators—such as the number of governing parties or majoritarian versus proportional systems—and instead introduces cumulative indices that capture both Institutional Rules and Power Rules, encompassing a broader array of theoretically grounded elements. This approach offers greater internal validity by accounting for institutional structures and dynamic political configurations that jointly shape citizens’ ability to attribute responsibility. By applying this improved measure across a large-N, longitudinal dataset spanning three decades and 30 democracies, we provide a more rigorous test of how CoR conditions the relationship between macroeconomic performance and citizens’ SWD.
In the following sections, we develop a theoretical framework explaining how and why CoR moderates the effect of economic performance on SWD. We then present empirical evidence from a broad cross-national dataset, demonstrating that when responsibility is clear, economic conditions have a pronounced effect on citizens’ evaluations of democratic performance. In contrast, where responsibility is blurred, the economy’s role in shaping SWD becomes negligible. These findings contribute to resolving longstanding inconsistencies in the literature and suggests the importance of political context in shaping voters perception on functioning democracy.
The economy and satisfaction with democracy
The relationship between the economy and political support has been the subject of extensive research spanning several decades. The underlying logic appears straightforward: a strong economy tends to increase political support, while a weak one often diminishes it. This relationship is well-established in the vote and popularity (VP) function literature (Lewis-Beck and Stegmaier, Reference Lewis-Beck and Stegmaier2013; Stegmaier et al., Reference Stegmaier, Lewis-Beck, Park, Evans, Arzheimer and Lewis- Beck2017; Hellwig and Singer, Reference Hellwig and Singer2023), but similar dynamics come into play across various aspects of political life, including adherence to regime norms and principles, procedural assessments, and the evaluation of political institutions and authorities. Among these, citizens’ SWD stands out as one of the most common indicators used to gauge their support for the regime and its associated procedures (Quaranta and Martini, Reference Quaranta and Martini2016).
Despite its widespread usage, the concept of SWD is far from achieving a consensus in the scholarly literature (Quaranta and Martini, Reference Quaranta and Martini2016; Claassen and Magalhães, Reference Claassen and Magalhães2022). Some researchers argue that SWD represents an evaluation of democracy as an ideal form of government when compared to other regime types (Canache et al., Reference Canache, Mondak and Seligson2001), while others consider it as a summary measure of political support that aligns with citizens’ partisan preferences (Anderson and Guillory, Reference Anderson and Guillory1997).
However, prevailing studies tend to understand SWD as a measure of citizens’ evaluations of the functioning of regime procedures in practice (Linde and Ekman, Reference Linde and Ekman2003; Dalton, Reference Dalton2004; Torcal and Montero, Reference Torcal and Montero2006; Norris, Reference Norris2011; Quaranta and Martini, Reference Quaranta and Martini2017; Christmann and Torcal, Reference Christmann and Torcal2017; Kestilä-Kekkonen and Söderlund, Reference Kestilä-Kekkonen and Söderlund2017; Dassonneville and McAllister, Reference Dassonneville and McAllister2020). More specifically, it is viewed as an assessment of “how democracy works in practice” (Wagner et al., Reference Wagner, Schneider and Halla2009: 32), and as an “indicator of support for the performance of a democratic regime” (Linde and Ekman, Reference Linde and Ekman2003: 396). Similarly, it reflects a rational response to the output of the regime (Quaranta and Martini, Reference Quaranta and Martini2016) or the outcome of policy performance (Torcal and Montero, Reference Torcal and Montero2006; Claassen and Magalhães, Reference Claassen and Magalhães2022), shedding light on the extent to which the system is responsive in the eyes of its citizens (Morlino, Reference Morlino2011, cited in Quaranta and Martini, Reference Quaranta and Martini2016: 165). In other words, SWD tends to capture an instrumental-based appraisal of the regime (Norris, Reference Norris2011).
The theoretical expectation regarding the impact of the economy on SWD is straightforward. Following the principles of economic voting theory (Lewis-Beck and Stegmaier, Reference Lewis-Beck and Stegmaier2013), it is argued that individuals tend to manifest a critical attitude toward the political system during periods of economic downturns. This is primarily because a poor economic performance often induces feelings of insecurity, stemming from reduced disposable incomes and lower living standards (Clarke et al., Reference Clarke, Dutt and Kornberg1993). Consequently, this decline in economic well-being leads to a decrease in SWD. It can also be posited that individuals withdraw their political support when they perceive their political system as incompetent in directing the government, and poor economic conditions send a strong signal of such incompetence (Duch and Stevenson, Reference Duch and Stevenson2008). Hence, a deteriorating economy is likely to result in a decline in SWD (Quaranta and Martini, Reference Quaranta and Martini2016; Christmann and Torcal, Reference Christmann and Torcal2017; Claassen and Magalhães, Reference Claassen and Magalhães2022).
While the theoretical expectation regarding the impact of the economy on SWD is straightforward, the empirical evidence is less clear-cut. Some studies have found the influence of the economy on SWD to be weak, limited, or marginal (Clarke et al., Reference Clarke, Dutt and Kornberg1993; Dalton, Reference Dalton2004; Norris, Reference Norris2011), while others have reported no significant impact of the economy on SWD (Anderson and Singer, Reference Anderson and Singer2008; Huang et al., Reference Huang, Chang and Chu2008; Dahlberg and Holmberg, Reference Dahlberg and Holmberg2014). In contrast, a number of studies have demonstrated a substantial and notable effect of the economy on SWD (Bratton and Mattes, Reference Bratton and Mattes2001; Wagner et al., Reference Wagner, Schneider and Halla2009; Armingeon and Guthmann, Reference Armingeon and Guthmann2014; Quaranta and Martini, Reference Quaranta and Martini2016, Reference Quaranta and Martini2017; Bellucci et al., Reference Bellucci, Memoli, Sanders, To´ka, Sanders and Magalha˜es2012; Christmann and Torcal, Reference Christmann and Torcal2017).
To address the ambiguity in empirical analyses, scholars have adopted various approaches. They have utilized different data sources to measure SWD (Claassen, Reference Claassen2020) and considered various economic contexts, including the impact of the global recession in Europe (Cordero and Simón, Reference Cordero and Simón2015; Quaranta and Martini, Reference Quaranta and Martini2017; Christmann and Torcal, Reference Christmann and Torcal2017). Additionally, researchers have examined different institutional contexts (Wagner et al., Reference Wagner, Schneider and Halla2009) and explored different economic focuses, such as prospective rather than retrospective economic evaluations (Loveless and Binelli, Reference Loveless and Binelli2020).
Furthermore, recent scholarship has explored the idea that the impact of the economy may not be uniform across different economic and institutional contexts. For example, the economy is found to have a significant impact on SWD in poorer countries, while its influence is less pronounced in wealthier counterparts (Daoust and Nadeau, Reference Daoust and Nadeau2021). Research on the impact of economic bailouts in Eurozone countries has pointed to different results. In response to external intervention in the national economy, citizens in bailed-out countries exhibit stronger attachment to democracy (Cordero and Simón, Reference Cordero and Simón2015), but it is also possible for SWD to decline because citizens perceive such intervention as interference with national democratic procedures (Armingeon and Guthmann, Reference Armingeon and Guthmann2014). When considering different party systems, Dassonneville and McAllister (Reference Dassonneville and McAllister2020) find that citizens’ SWD increases when there are more parties for the voters to choose from.
That said, existing studies suggest that the impact of the economy on SWD is influenced by various moderating factors. Although recent studies such as those by Criado and Herreros (Reference Criado and Herreros2007), Christensen (Reference Christensen2015), Kestilä-Kekkonen and Söderlund (Reference Kestilä-Kekkonen and Söderlund2017), and Tang and Huhe (Reference Tang and Huhe2020) have highlighted the importance of CoR, there remains scope for further exploration. In the subsequent sections, we propose a theoretical framework for how CoR moderates this impact. We also demonstrate how our study advances understanding by employing a refined CoR measure and an extensive dataset spanning over three decades from multiple regions.
Why does the clarity of responsibility matter?
For citizens to evaluate governments based on economic conditions, they must be able to clearly identify which political actors are responsible for those outcomes. CoR captures this institutional and political context, referring to the extent to which political institutions and governing arrangements enable voters to assign credit or blame for policy performance, particularly in the economic domain (Powell and Whitten, Reference Powell and Whitten1993). When the political context presents a unified and cohesive government—such as under single-party majority rule in parliamentary systems or strong presidential leadership—responsibility is transparent, allowing citizens to effectively reward or punish incumbents. Conversely, in fragmented systems characterized by coalition governments, minority administrations, divided legislatures, weak party discipline, or complex institutional veto points, responsibility becomes diffused. Under such conditions, citizens struggle to accurately assign blame or credit, weakening the electoral connection between government performance and public approval and diminishing the strength of retrospective economic evaluations (Stegmaier et al., Reference Stegmaier, Lewis-Beck, Park, Evans, Arzheimer and Lewis- Beck2017).
While the moderating role of CoR is well established in the economic voting literature (for a comprehensive review, see Stegmaier et al., Reference Stegmaier, Lewis-Beck, Park, Evans, Arzheimer and Lewis- Beck2017), it has received limited attention in studies of SWD. This gap is both surprising and consequential, as it leaves unaddressed how institutional contexts that blur political accountability may weaken the theoretically expected relationship between macroeconomic conditions and citizens’ SWD. Addressing this gap is essential for advancing our understanding of the conditions under which economic performance shapes attitudes toward functioning democracy.
The motivation for incorporating the concept of CoR in the study of SWD lies in the idea that citizens formulate their SWD based on their “evaluation of policy performance” (Linde and Ekman, Reference Linde and Ekman2003). Put differently, SWD captures an instrumental or performance-based appraisal of the regime (Mattes and Bratton, Reference Mattes and Bratton2007; Wagner et al., Reference Wagner, Schneider and Halla2009; Norris, Reference Norris2011). It is closely related to support for the incumbent, such as the evaluation of the economy (Claassen and Magalhães, Reference Claassen and Magalhães2022), and serves as a signal about whether the regime is responsive in the eyes of the public (Montero et al., Reference Montero, Gunther and Torcal1997; Linde and Ekman, Reference Linde and Ekman2003; Morlino, Reference Morlino2011). In this sense, SWD can be considered a “good proxy for assessing citizens” evaluations of the output of the democratic regime” (Quaranta and Martini, Reference Quaranta and Martini2016: 167). As a result, people tend to lose confidence in the government when policy output is unfavorable, such as in the case of a poor economy, and conversely, their confidence increases when the output is favorable, as with a strong economy (de Blok et al., Reference de Blok, van der Meer and Van der Brug2022).
Since the evaluation of outputs and performance plays a crucial role in shaping citizens’ confidence in their polities (Easton, Reference Easton1975), the relationship between the economy and SWD is essentially evaluative in nature (van der Meer, Reference Van and Uslaner2017; van der Meer and Hakhverdian, Reference Van der Meer and Hakhverdian2017). A good economy is likely to elicit higher levels of SWD from the public, as they perceive it as an indicator of being “well-governed” (Cordero and Simón, Reference Cordero and Simón2015). However, this evaluative logic raises a critical question—not only whether citizens perceive their country as being well- or poorly governed, but by whom. In other words, it is not merely about how a polity performs in terms of economic outcomes, but also about whether citizens can correctly attribute responsibility for that performance. Even when the economy is underperforming, dissatisfaction with democracy may not follow if citizens believe that external forces—such as a global recession—are to blame, rather than domestic actors.
Although the moderating role of CoR has been extensively examined in relation to behavioral outcomes such as vote choice and incumbent approval, it has received comparatively limited attention in the study of attitudinal outcomes like SWD, with only a few recent exceptions (Criado and Herreros, Reference Criado and Herreros2007; Christensen, Reference Christensen2015; Kestilä-Kekkonen and Söderlund, Reference Kestilä-Kekkonen and Söderlund2017; Tang and Huhe, Reference Tang and Huhe2020). Scholars have increasingly called for greater integration of CoR into this area of research, emphasizing its importance for understanding how institutional contexts shape democratic attitudes. As van der Meer (Reference Van and Uslaner2017: 594) notes, “to understand the dynamic, we need to take into account the assignment of political responsibility.” Similarly, Quaranta and Martini (Reference Quaranta and Martini2016: 173) argue that “the institutional structure (i.e., CoR) might moderate the effect of economic dynamics,” and Cordero and Simón (Reference Cordero and Simón2015: 4) highlight that “the impact of the economy is mediated by political elements” and extends “beyond electoral competition.” These perspectives collectively suggest the need to broaden the scope of research from purely behavioral responses to include attitudinal outcomes, thereby deepening our understanding of how institutional and economic factors jointly influence SWD.
How does clarity of responsibility affect the economy-SWD link?
We propose that CoR plays a crucial role in moderating the impact of the economy on SWD. In particular, we contend that low CoR tends to diminish the influence of the economy on SWD by introducing uncertainty regarding responsibility for policy outcomes. This uncertainty, in turn, hinders citizens from employing the evaluative reward punishment mechanism based on the state of the economy. In essence, low CoR obstructs the evaluative mechanism by making it challenging for individuals to discern who is accountable for economic outcomes. In cases where responsibility is shared among multiple government actors, citizens may struggle to pinpoint who bears responsibility for specific policies. This scenario is commonly observed in coalition governments (Kestilä-Kekkonen and Söderlund, Reference Kestilä-Kekkonen and Söderlund2017). When a government comprises several political parties, it becomes unclear to citizens which coalition partner is responsible for each policy. In line with this, Lewis-Beck and Lockerbie (Reference Lewis-Beck and Lockerbie1989: 167) argue that “with more parties in power, there is less responsiveness to voter views on economic policy. Thus, voters who are concerned about the state of the economy may have less certainty when forming their political support as they seek to punish or reward the government.”
When examining political trust, van der Meer (Reference Van and Uslaner2017) frames the process as an evaluation conducted by trusters (e.g., fellow citizens) concerning a specific object (e.g., political authorities). In simpler terms, citizens trust political authorities to perform certain tasks, as discussed by Hardin (Reference Hardin and Warren1999: 26), cited in van der Meer (Reference Van and Uslaner2017), and thus, “what sets political trust apart is that the object is made up by political institutions rather than fellow citizens; that is, trust is vertical rather than horizontal” (van der Meer, Reference Van and Uslaner2017: 585).
Expanding upon this logic, we assert that citizens expect their political system to be capable of producing a good economy rather than a poor one. In countries with high CoR, such as with a majority party government, citizens can easily attribute responsibility for these economic conditions to the governing party. For instance, if the majority party is in power during an economic upswing, citizens can readily credit that party for the improved economic performance. On the other hand, if the economy takes a turn for the worse, it becomes clear to citizens that the governing party bears responsibility for the economic challenges. With low CoR, such as in systems with coalition governments, we agree that the “basic mechanism of vertical democratic accountability” may falter (Carlin et al., Reference Carlin, Love and Martinez-Gallardo2015: 441), because it leaves citizens grappling with uncertainty about who is to blame for poor economic results.
This clarity in assigning responsibility reinforces the evaluative mechanisms in citizens’ minds. They can confidently assess the government’s performance in managing the economy, which influences their level of SWD. In this scenario, the link between the economy and SWD is more straightforward and predictable because it is apparent who is responsible for the economic outcomes.
Indeed, similar arguments have been tested in recent studies. For instance, Kestilä-Kekkonen and Söderlund (Reference Kestilä-Kekkonen and Söderlund2017) proposed that the economy and power-sharing institutional arrangements, such as coalition governments, interact with one another. Their argument centers on the notion that larger coalition governments offer more actors the opportunity to participate in the policymaking process, which potentially blurs the accountability channels regarding economic outcomes. This, in turn, diminishes the impact of economic fluctuations on citizens’ attitudes toward how democracy functions.
When it comes to context-dependent arguments, Christensen (Reference Christensen2015) focuses on the effective number of parties and electoral proportionality, while Criado and Herreros (Reference Criado and Herreros2007) examine the electoral system (majoritarian vs. proportional systems) as measures of power-sharing structures. Specifically, Christensen (Reference Christensen2015) finds that the link between the economy and SWD, albeit less pronounced compared to previous studies such as Powell and Whitten (Reference Powell and Whitten1993), is muted when power sharing is more prevalent. Criado and Herreros (Reference Criado and Herreros2007) also show that the evaluation of the economy has a smaller impact on SWD in proportional democracies than in majoritarian ones since the former encourages the representation of multiple parties in the government, which in turn, creates power-sharing contexts. A recent study by Tang and Huhe (Reference Tang and Huhe2020) shows that the link between SWD and economic perception is more pronounced where government intervention in the market is strong because the public perceives the economic outcomes as the products of active government roles.
That being said, the presence or absence of CoR can significantly influence the strength of the relationship between the economy and citizens’ SWD, shaping the dynamics of how individuals evaluate their political systems based on economic performance. Building on the above discussion, we claim that in political systems characterized by high CoR, the performance of the national economy will exhibit a stronger impact on citizens’ SWD compared to political systems with low CoR, and propose the following hypotheses:
Hypothesis 1: In political systems characterized by high CoR, higher GDP growth rates will exhibit a stronger positive impact on citizens’ SWD compared to political systems with low CoR.
Hypothesis 2: In political systems characterized by high CoR, higher unemployment rates will exhibit a stronger negative impact on citizens’ SWD compared to political systems with low CoR.
Data and variables
We sourced data for our dependent variable, SWD, from Claassen (Reference Claassen2020).Footnote 1 SWD is gauged through a survey question evaluating respondents’ satisfaction with the way democracy works (or practices or functions) in their respective countries. This measurement yields an average level of SWD within a country for a specific year. With 619 observations from 30 countries since 1990, the average SWD stands at 0.11 (standard deviation: 1.11), ranging from a minimum of –2.33 to a maximum of 2.92. We present the change in the value of SWD over time in Figure A1 in Appendix.
Our primary explanatory variables include the Gross Domestic Product (GDP) growth rate and the unemployment rate, which have been frequently registered as important predictors in empirical studies (Clarke et al., Reference Clarke, Dutt and Kornberg1993; Wagner et al., Reference Wagner, Schneider and Halla2009; Ezrow and Xezonakis, Reference Ezrow and Xezonakis2011; Christmann and Torcal, Reference Christmann and Torcal2017). We obtained the data from the World Bank. To capture economic dynamics, we utilized one-year lagged values of these economic indicators.
For the measurement of CoR, we adopt the framework proposed by Dassonneville and Lewis-Beck (Reference Dassonneville and Lewis-Beck2017), which distinguishes between two key dimensions: Institutional Rules and Power Rules. These dimensions are operationalized through cumulative indices composed of theoretically grounded components. We consider this approach well-suited for capturing the concept of CoR, as each element is explicitly defined in the literature and reflects institutional features that influence responsibility attribution. Importantly, following recent advances (Hobolt et al., Reference Hobolt, Tilley and Banducci2013; Dassonneville and Lewis-Beck, Reference Dassonneville and Lewis-Beck2017), we conceptualize CoR as a multidimensional construct. The two-dimensional structure allows for a more nuanced and accurate representation of institutional variation across democracies.
The first index, Power Rules, includes five indicators that reflect how power is distributed and exercised within the government. First, we distinguish between coalition and single-party governments—a factor consistently identified as central to CoR (Powell, Reference Powell2000). Second, we include a variable differentiating majority from minority governments, based on the assumption that majority governments typically enjoy more autonomous decision-making authority, thus enhancing clarity. Third, we incorporate the effective number of political parties, as greater party system fragmentation tends to obscure lines of accountability (Anderson, Reference Anderson2000). Fourth, we consider the openness of the economy, building on the argument that higher global economic integration may reduce governments’ perceived control over outcomes, thereby weakening responsibility attribution (Hellwig and Samuels, Reference Hellwig and Samuels2007; Duch and Stevenson, Reference Duch and Stevenson2010). Finally, we include the duration of time an incumbent has held office, based on the idea that longer tenure increases public familiarity with governing actors, improving the electorate’s capacity to assign credit or blame.
The Institutional Rules dimension in our measure includes four of the five original components identified by Dassonneville and Lewis-Beck (Reference Dassonneville and Lewis-Beck2017). The only deviation from their framework is the exclusion of compulsory voting, which we do not consider essential for capturing the institutional foundations of CoR. We retain the remaining four components, each grounded in established theoretical expectations.
First, we include the electoral system, distinguishing between majoritarian and proportional (or mixed) systems. Consistent with Powell (Reference Powell2000), we expect majoritarian systems—which tend to produce one-party governments—to enhance CoR. Second, we incorporate the federal–unitary distinction. As federal systems distribute authority across multiple levels of government, they are expected to reduce the public’s ability to attribute responsibility clearly (Cutler, 2004; Reference Cutler2008). Third, we account for differences in government structure by distinguishing parliamentary systems from presidential and semi-presidential systems. We expect presidential systems to promote greater CoR, while dual-executive arrangements in semi-presidential systems may dilute it (Hellwig and Samuels, Reference Hellwig and Samuels2008). Finally, we include a measure of democratic longevity, based on the idea that repeated democratic experience fosters a “learning effect” among citizens, thereby improving their ability to assign responsibility (Stegmaier and Lewis-Beck, Reference Stegmaier and Lewis-Beck2009).
In constructing both indices—Institutional Rules and Power Rules—we follow Dassonneville and Lewis-Beck’s (Reference Dassonneville and Lewis-Beck2017) recommendation of assigning equal weights to all components. This approach avoids privileging any single institutional feature and ensures theoretical neutrality. As they argue, cumulative indices allow individual measurement errors to offset one another, making the underlying signal more discernible (Nadeau et al., Reference Nadeau, Niemi and Yoshinaka2002; Dassonneville and Lewis-Beck, Reference Dassonneville and Lewis-Beck2017). Moreover, cumulative indicators provide greater variance and statistical power than dichotomous measures, which often suffer from limited variation.
To operationalize the indices, continuous variables (e.g., the effective number of parties) are dichotomized using median cutoff points, consistent with established practice in the literature (Powell and Whitten, Reference Powell and Whitten1993; Hobolt et al., Reference Hobolt, Tilley and Banducci2013; Dassonneville and Lewis-Beck, Reference Dassonneville and Lewis-Beck2017). The Institutional Rules index ranges from 0 to 4 (reduced from 5 due to the exclusion of compulsory voting), while the Power Rules index ranges from 0 to 5. In both cases, higher scores reflect greater CoR.
We contend that the Dassonneville and Lewis-Beck (Reference Dassonneville and Lewis-Beck2017) measure of CoR offers significant advantages over the more limited approaches employed in prior studies. For instance, Criado and Herreros (Reference Criado and Herreros2007) focus narrowly on the electoral system, Christensen (Reference Christensen2015) emphasizes only the effective number of parties and electoral proportionality, and Kestilä-Kekkonen and Söderlund (Reference Kestilä-Kekkonen and Söderlund2017) primarily consider the number of parties in government. While each of these factors is relevant, they provide only a fragmented perspective on the broader institutional environment that shapes citizens’ ability to attribute political responsibility. In contrast, our measure not only incorporates these dimensions but also includes additional critical factors—such as government status (majority vs. minority), party system fragmentation, the openness of the economy, and the longevity of democratic experience—resulting in a more holistic, multidimensional, and theoretically grounded assessment of CoR.
Moreover, Tang and Huhe (Reference Tang and Huhe2020) employ the extent of government economic engagement as a proxy for responsibility attribution. Although this approach is intuitive, it risks conflating government activity with perceived responsibility. In highly globalized economies or during periods of external shocks—such as the COVID-19 pandemic or international financial crises—high levels of government intervention do not necessarily translate into greater perceived responsibility for economic outcomes. Our approach avoids this conceptual ambiguity by focusing directly on institutional and political structures that condition responsibility clarity, rather than on policy activity levels that may be endogenous to the outcomes being studied.
Given that CoR is a central feature of our article, we also employ alternative measures, as suggested by Kestilä-Kekkonen and Söderlund (Reference Kestilä-Kekkonen and Söderlund2017). They focus on a measure of government fractionalization, indicating the probability that two randomly chosen cabinet members from the government parties will belong to different parties (Beck et al., Reference Beck2001). This measure ranges between 0 and 1, where a lower number indicates fewer parties in the government, and a higher number suggests a larger number of equally-sized parties in the government. For ease of interpretation, we use reversed values, so that a higher number represents a context of clearer responsibility. Similarly but distinctly, we also utilize government seat share, which is operationalized as the fraction of parliamentary seats held by the government parties (Beck et al., Reference Beck2001). While Kestilä-Kekkonen and Söderlund (Reference Kestilä-Kekkonen and Söderlund2017) include this variable separately as a control variable rather than a component of the CoR measure, we believe that both variables together effectively capture the concept of CoR. The former reflects the extent to which the government is composed of a single party or multiple parties, while the latter indicates the extent to which the government controls the majority in the parliament. We calculate the average of the standardized values of these variables for our analysis.
We present the variations in three measures of CoR over time since 1980 in Figure 1. Notably, the index of Power Rules tends to fluctuate across years in most countries, whereas the index of Institutional Rules changes much less over time Footnote 2 in Figure 1(a). A similar pattern is observed for the index of Power-Sharing in Figure 1(b), which ebbs and flows in different places and at different times. Consequently, Dassonneville and Lewis-Beck (Reference Dassonneville and Lewis-Beck2017) identify Institutional Rules as “fairly stable” or “fixed,” while regarding Power Rules as a “dynamic” indicator shaped by actors’ behavior.
Figure 1. Variation in clarity of responsibility over time.
We have included a battery of economic and political control variables based on guidance from existing studies. One economic control accounts for the level of affluence in a country, which we measure as the logarithm of GDP per capita. This control is included because contrasting expectations exist regarding the impact of the economy on SWD. On the one hand, the economy tends to be a pressing issue among the public in poorer societies (Singer, Reference Singer2011; Kestilä-Kekkonen and Söderlund, Reference Kestilä-Kekkonen and Söderlund2017; Daoust and Nadeau, Reference Daoust and Nadeau2021). Consequently, it is likely that the economy matters more in poorer countries (Daust and Nadeau, Reference Daoust and Nadeau2021). On the other hand, some argue the opposite: that the economy matters in wealthier countries because greater wealth leads the public to have higher expectations of their political systems (Norris, Reference Norris1999; Quaranta and Martini, Reference Quaranta and Martini2017).
Another economic control accounts for inequality, which is measured by the Gini index of inequality of net household income for each country-year (Solt, Reference Solt2012). Rising income inequality tends to produce unfavorable sentiments toward the functioning of their political system (Uslaner, Reference Uslaner2008).
Despite the well-established significance of economic factors in shaping political support, it is widely suggested that corruption also has a substantial impact on the level of SWD (Anderson and Tverdova, Reference Anderson and Tverdova2003; Wagner et al., Reference Wagner, Schneider and Halla2009; Norris, Reference Norris2011; Pellegata and Memoli, Reference Pellegata and Memoli2016; Christmann and Torcal, Reference Christmann and Torcal2017; van der Meer, Reference Van and Uslaner2017). To assess the risk of corruption in a country, we use a measure obtained from the International Country Risk Guide.Footnote 3
We have also introduced a dummy variable indicating whether an election for an executive officer was held in the country in that year. Elections can exert control over incumbents and produce spillover effects on satisfaction (Banducci and Karp, Reference Banducci and Karp2003; Anderson et al., Reference Anderson2005; Esaiasson, Reference Esaiasson2011; Blais et al., Reference Blais, Morin-Chasse and Singh2017).
While scholars have shown that the degree of proportionality of the electoral system and the effective number of parties are positively related to SWD (Anderson and Guillory, Reference Anderson and Guillory1997), others find that majoritarian systems promote higher levels of satisfaction (Norris, Reference Norris1999; Criado and Herreros, Reference Criado and Herreros2007; Aarts and Thomassen, Reference Aarts and Thomassen2008). We control for both variables.Footnote 4 Additionally, we consider the level of democracy using a combined score from Polity IV and Freedom House. Recognizing that the dynamics of democratic support may differ between old and new democracies (Waldron-Moore, Reference Waldron-Moore1999), we include this control variable in our analysis. To address the concern of unit specific errors in the composite error term, it also includes fixed effects estimations.
Results and analyses
Table 1 presents the effects of macroeconomic conditions on SWD across various model specifications, without any interaction terms. Model 1 utilizes ordinary least squares (OLS) regression, while M2 employs feasible generalized least squares (FGLS), and M3 is based on panel corrected standard errors (PCSE).Footnote 5 Overall, the findings suggest that the economy exerts the expected influence. Specifically, the GDP growth rate appears to increase the level of SWD, while the unemployment rate appears to decrease it. These effects are statistically significant and are robust across a range of model specifications.Footnote 6 To account for the possibility of delayed economic effects, we also replicate our models using a two-year lag for economic indicators. These results, reported in the online Appendix (Tables A2 and A3), confirm the robustness of our findings.
Table 1 The effect of macroeconomy on satisfaction with democracy
Standard errors (and Panel corrected standard errors in M3) in parentheses: * p < 0.10, ** p < 0.05, *** p < 0.01
Table 2 The effect of macroeconomy on satisfaction with democracy conditional on clarity of responsibility
Standard errors in parentheses: * p < 0.10, ** p < 0.05, *** p < 0.01
In line with widely accepted studies, both Model 2 and Model 3 indicate that SWD tends to be higher in more economically advanced and more equalized societies, as well as in countries with good governance (i.e., lower corruption). It is noteworthy that SWD is lower in countries with a higher number of parties in Model 1, although this effect is not consistently found in the other models. Furthermore, the electoral system, the level of democracy, and the presence of elections in a given year fail to have a statistically significant effect on the level of SWD in any of the models.
Our primary focus centers on assessing the conditional effect of the economy across differing levels of CoR. To investigate this, we introduce the interaction term in the baseline model. If the coefficient of the interaction term between CoR and GDP growth is positive, it would confirm our hypothesized conditional effects in H1. Conversely, given that unemployment is negatively related to satisfaction, the interaction term for CoR and unemployment should be negative to support our conditional hypothesis in H2.
Table 2 reports the estimated effects of macroeconomic conditions on SWD, conditional on CoR. The interaction terms between GDP growth and CoR are positive and statistically significant in the model using Power Rules, and marginally significant in the Institutional Rules model, suggesting that higher CoR may amplify the positive relationship between economic growth and SWD in these institutional contexts. However, the interaction term is not statistically significant in the Power-Sharing model, indicating no clear moderating effect there. In contrast, the interaction between the unemployment rate and CoR is negative and statistically significant in both the Power Rules and Power-Sharing models, implying that in these contexts, higher CoR intensifies the negative relationship between unemployment and democratic satisfaction. No significant interaction is found using the Institutional Rules measure.
To assess the dynamic effects of the interacted variables, it is essential to visualize the marginal effect of the economy on our dependent variable across various levels of the conditioning variable, CoR. Figure 2 displays the marginal effects of GDP growth and unemployment across the different ranges of CoR, while holding other variables at their mean values. The figures on the left depict the marginal effect of GDP growth, while those on the right illustrate the marginal effect of unemployment. Since we employ three types of CoR (Power Rules, Institutional Rules, and Power-Sharing), we present three sets of sub-figures horizontally.
Figure 2. The effect of economy on satisfaction with democracy conditional on CoR (95% CI).
It is evident that the slope of the marginal effect of GDP growth rises consistently across all types of CoR measures, indicating that, all else being equal, higher levels of CoR amplify the positive effect of GDP growth on SWD. For instance, in Figure 2(a), GDP growth has a positive and statistically significant effect on SWD when the value of Power Rules exceeds 3 on a scale from 0 to 5. However, below approximately 2.5, GDP growth appears to have no discernible effect on SWD, as indicated by the 95% confidence intervals around the marginal effect line. Similarly, the effect of GDP growth remains statistically significant when the value of Institutional Rules is above 2 (on a 0 to 4 scale) and when Power-Sharing exceeds 0 (on a –2 to 2 scale).
To provide a clearer sense of the substantive effect of GDP growth, consider a scenario in which the GDP growth rate is 3%. Based on Figure 2(a) and holding all else constant, the effect of GDP growth on SWD differs depending on the level of Power Rules. When Power Rules equals 5 (high CoR), a 3% GDP growth is associated with an increase in SWD of approximately 0.129. In contrast, when Power Rules equals 3 (moderate clarity), the same 3% GDP growth corresponds to a smaller increase in SWD of about 0.057. These calculations, based on the interaction term coefficient (0.012) and the direct effect of GDP growth (–0.017) reported in Table 2, illustrate how the positive effect of economic growth on democratic satisfaction becomes substantially stronger under conditions of higher CoR, all else being equal.
Turning to the unemployment rate, we observe similar patterns. The marginal effect graphs consistently show a downward trend across all three CoR measures, indicating that, all else being equal, higher unemployment tends to lower SWD. This negative association becomes more pronounced as CoR increases. The negative effect of unemployment remains statistically significant across the entire range of Power Rules. However, it becomes statistically insignificant when the value of Institutional Rules falls below 1.5 and when that of Power-Sharing falls below approximately –0.7.
In conclusion, our findings demonstrate that the effect of the economy on SWD tends to be stronger in countries where political power structures provide clearer cues about who holds responsibility for policy outcomes. In these settings, economic growth is more likely to enhance citizens’ SWD, while rising unemployment is more strongly associated with democratic dissatisfaction. This pattern is particularly evident when CoR is measured through Power Rules and Power-Sharing indicators, which capture the dynamics of government composition and political competition. By contrast, the moderating role of CoR is less pronounced when using the Institutional Rules measure, suggesting that not all institutional arrangements equally shape citizens’ perceptions of responsibility. Together, these results highlight that CoR can play an important role in conditioning how economic performance influences democratic satisfaction, though its moderating effects appear more salient in certain institutional configurations than others.
Discussion
Our theoretical framework, rooted in the literature on economic voting and government approval, extends the application of CoR beyond behavioral outcomes to attitudinal evaluations—specifically, SWD. While previous research has primarily examined how CoR influences vote choice and incumbent support, our findings show that CoR also conditions how citizens evaluate the functioning of democracy in light of economic conditions. This suggests that institutional and political contexts that clarify or obscure political responsibility play a crucial role not only in shaping electoral behavior but also in shaping broader democratic attitudes such as SWD.
Recent work by Singh and Mayne (Reference Singh and Mayne2023) identifies two leading interpretations of SWD. The “literal view” treats SWD as a subjective assessment of democratic quality—how well the system lives up to respondents’ normative expectations of democracy (e.g., Cusack, Reference Cusack1999). By contrast, the “encompassing view” interprets SWD as a broader measure of political system performance, incorporating judgments about the everyday functioning of governance (Schmitt, Reference Schmitt1983).
Our findings are more consistent with the encompassing view. That the effect of economic performance on SWD is significantly moderated by CoR implies that citizens are not merely evaluating abstract democratic principles, but are instead responding to how well political institutions manage economic outcomes—particularly when responsibility for those outcomes is clear. In this way, SWD appears to function as a barometer of perceived government competence and accountability within a given institutional context, rather than as a pure expression of democratic ideals.
These results carry important implications. They caution against interpreting SWD solely as a proxy for democratic support or legitimacy. Instead, researchers should be attentive to the institutional and economic contexts that shape how respondents interpret the question. By showing that CoR conditions the responsiveness of SWD to economic signals, our study highlights the value of integrating institutional moderators when using SWD in cross-national analyses of political support.
Conclusion
This article has demonstrated the pivotal role of CoR in moderating the relationship between macroeconomic conditions and citizens’ SWD. Drawing on aggregate-level data from 30 democracies over more than three decades, our analysis shows that the strength of the economy–SWD relationship is not uniform across countries, but instead depends critically on the institutional and political context—specifically, the degree to which responsibility for policy outcomes is clear.
We find that in contexts with higher CoR, economic growth significantly increases, and rising unemployment significantly decreases, SWD. In contrast, in countries with fragmented or opaque political arrangements, these economic signals exert a far weaker influence on public sentiment. These findings highlight that institutional features that enable citizens to attribute responsibility play a crucial role in shaping democratic attitudes.
This study advances existing research in several ways. First, it applies the CoR framework to attitudinal outcomes such as SWD, which have traditionally been examined through a behavioral lens focused on vote choice or incumbent approval. By bridging this gap, we show that CoR matters not only for how people vote but also for how they evaluate democratic performance. Second, we employ a more theoretically grounded and comprehensive measure of CoR, following Dassonneville and Lewis-Beck (Reference Dassonneville and Lewis-Beck2017), which captures both institutional rules and power dynamics through cumulative indices. This approach improves upon earlier studies that relied on narrower or unidimensional indicators.
Third, our analysis leverages a longitudinal dataset covering over 30 years—from the 1990s to the present. This extended time frame allows us to observe long-term trends, institutional variation, and changes in macroeconomic conditions across democracies, offering a more robust and generalizable perspective than studies limited to shorter periods (e.g., Criado and Herreros, Reference Criado and Herreros2007; Christensen, Reference Christensen2015; Tang and Huhe, Reference Tang and Huhe2020).
Finally, we expand the geographic scope of the literature on SWD, which has largely concentrated on Europe. By incorporating a global sample that includes underrepresented regions such as Asia and Oceania, our study contributes to a more inclusive understanding of how institutional and economic contexts interact to shape democratic satisfaction across diverse political systems.
Our findings also speak directly to the enduring puzzle highlighted at the outset of this study: why does the relationship between economic performance and SWD appear strong in some contexts but weak or absent in others? The results suggest that this inconsistency is not a reflection of theoretical shortcomings, but rather of empirical approaches that have overlooked the role of institutional context—specifically, the clarity with which responsibility for economic outcomes can be assigned. By introducing a more valid and multidimensional measure of CoR, our analysis helps explain why previous studies have reported divergent findings: where responsibility is clear, economic conditions have a discernible impact on democratic satisfaction; where it is blurred, that impact diminishes.
While this study advances our understanding of how CoR shapes the relationship between economic performance and SWD, it also opens several promising avenues for future research. A key implication of our findings is that CoR—originally developed to explain behavioral outcomes such as vote choice and incumbent approval—can also be extended to the domain of political attitudes. Future studies could investigate whether CoR similarly moderates the effects of economic or policy performance on other attitudinal outcomes, such as political trust (e.g., trust in government, political parties, or leaders) or diffuse support for democracy more broadly.
In addition, while our analysis captures cross-national and over-time variation in CoR, future research could further unpack which specific institutional components (e.g., electoral rules, federalism, cabinet composition) most strongly shape responsibility attribution in different contexts. Within-country designs or experiments could also help isolate causal pathways through which institutional clarity affects public evaluations.
In sum, this study suggests that SWD is not solely a function of economic performance, but also of whether political institutions enable citizens to hold actors accountable for those outcomes. CoR thus serves as a critical lens through which citizens interpret economic signals—and, by extension, evaluate the functioning and responsiveness of democratic governance in practice.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1755773925100179.
Acknowledgments
We are deeply grateful to the editors and the three anonymous reviewers for their constructive comments and suggestions. Earlier versions of this paper were presented at the Annual Conference of Asian Election Studies at Keio University in 2024 and at the joint conference of the Niehaus Center for Globalization and Governance and the Korean Political Science Association in 2024. We also thank the participants at these events for their valuable feedback.
Funding statement
This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2024S1A5A8026676).
Appendix A
Figure A1. Satisfaction with democracy over time.
Figure A2. The effect of economy on satisfaction with democracy conditional on Gov’t Fractionalization (Kestila¨-Kekkonen and So¨ derlund 2017) (95% CI).
Table A1 The effect of macroeconomy on satisfaction with democracy using IMF information
Standard errors in parentheses: * p < 0.10, ** p < 0.05, *** p < 0.01.
Table A2 The effect of macroeconomy on satisfaction with democracy (t − 2 lag)
Standard errors (and panel-corrected standard errors in M3) in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Table A3 The effect of macroeconomy on satisfaction with democracy conditional on Clarity of responsibility (t − 2 lag)
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Open access
