Political trust and economic development in European regions

Political trust comprises two components: a rational element (Hardin 2002; Van der Meer 2017) and a normative dimension (Hooghe et al. 2017; Warren 2006). Both are necessary but not sufficient conditions for political trust to materialize. Levi and Stoker (2000) posit that political trust involves a commitment rooted in moral values or normative expectations and the trustworthiness demonstrated by the object of trust. While both dimensions matter, Hooghe (2011) contends that normative expectations hold greater significance, especially when citizens lack sufficient information about political actors or the system. Hooghe et al. (2017) suggest that this explains the relatively stable nature of political trust despite changes in the rational components that constitute citizens’ assessments of government performance.

Previous literature has conceptualized political trust in various ways, with some authors (e.g., Rothstein and Stolle 2008) considering it a multidimensional construct with distinct components tied to different types of institutions. Others (e.g., Christensen and Lægreid 2005; Hooghe 2011) adopt a unidimensional approach, demonstrating that a single underlying variable can account for political trust. André (2014), in an empirical study using exploratory factor analysis (EFA) on European Social Survey (ESS) data, found both perspectives plausible, leaving the choice of conceptualization to the researcher. Following Hooghe (2011), we use a unidimensional concept, as our primary interest lies in examining the underlying drivers of political trust as a whole and its implications for economic development.

While political trust has been extensively studied at the national level by political scientists, primarily using economic performance to explain political trust (Kroknes et al. 2015; van Erkel and van der Meer 2016), subnational analyses have been relatively sparse. However, the region is where trust often develops through regular interpersonal interactions, and it is the scale where the mechanisms linking trust to economic development unfold (Pike et al. 2017; Rodríguez-Pose 1998; Trigilia 2001; Trigilia and Burroni 2009). This applies to political trust as well. While political trust is often directed towards national or international institutions, citizens’ experiences with politicians and political institutions at the local and regional levels significantly shape their trust in the broader political system. This is particularly true in decentralized contexts where regional governments hold substantial authority to implement their own policies (Hooghe et al. 2016). Decentralization also empowers citizens by expanding opportunities for inclusive political participation, which fosters political trust (Putnam 1993; Rodríguez-Pose and Di Cataldo 2015). Thus, the manifestation of political trust, through acceptance of government policies and active participation in politics, often occurs at the regional level (Rodríguez-Pose 1998; Rodríguez‐Pose and Storper 2006).

The regional studies literature has increasingly recognized the importance of political conditions for economic development, particularly in the context of decentralization. However, these studies have rarely explored the role of political trust (e.g., Putnam 1993; Rodríguez-Pose and Di Cataldo 2015). Political trust influences economic development through both top-down and bottom-up mechanisms (Rodríguez‐Pose and Storper 2006; Trigilia 2001; Trigilia and Burroni 2009). From a top-down perspective, political trust legitimizes government, leading citizens to accept its authority in implementing policies and programs (Levi and Stoker 2000; Marien and Hooghe 2011; Hooghe et al. 2017). This reduces enforcement and transaction costs, freeing resources for productive economic activities (Farole et al. 2011; Kaasa 2016). It also enables governments to effectively implement policies and mobilize citizen and organizational participation. Thus, regions need a certain level of political trust to ensure broad acceptance of the government’s legitimacy and the rules governing the system. However, when political trust is at a satisfactory level, the top-down perspective suggests diminishing returns to further improvements.

From a bottom-up perspective, political trust encourages political engagement and, consequently, other forms of participation (Newton and Ramón 2007; Hooghe and Marien 2013). This is helpful for the development of vertical links (or linking social capital) between citizens and elites, cumulatively broadening existing networks (Fukuyama 1995; Wollebaek and Selle 2002) and moderating their quality (Boschma 2005; Wollebaek and Selle 2002; Woolcock 2002; Rodríguez‐Pose and Storper 2006). Conversely, a lack of political trust can lead to citizens withdrawing from cooperative activities, fostering clientelism and rent-seeking behaviors that undermine economic development. This perspective suggests constant returns, as citizen involvement gradually increases with higher levels of political trust.

However, empirical studies examining the significance of political trust for economic development have been scarce, with recent studies primarily focusing on its effects on productivity (e.g., Kaasa 2016). Previous research has not explored whether the returns to political trust remain constant or whether there are diminishing returns.

Building upon the rationale presented above, we anticipate a positive association between political trust and economic development and propose the following hypothesis:

Social trust refers to the trust individuals have in other people not familiar to them (Fukuyama 1995; Newton and Zmerli 2011; Tabellini 2010; Uslaner 2008). A general sense of trust within society affects economic development by resolving collective action dilemmas, enabling societies to reach more productive equilibria. It also reduces transaction costs, redirecting resources from contract enforcement toward productive purposes. In essence, organizations in high-trust societies allocate more resources to production and fewer to legal expenses. Trust additionally curbs opportunistic behavior, facilitates information sharing, aids knowledge exchange, and fosters innovation (Beugelsdijk and Van Schaik 2005; Bjørnskov 2012; Fukuyama 1995; Whiteley 2000). In addition, it promotes tolerance, signaling a welcoming culture and attracting human capital to a region (Florida 2002). In summary, social trust facilitates productive economic activities, thereby increasing economic development (Beugelsdijk and Van Schaik 2005; Feldman 2014; Shearmur 2011; Stam and Bosma 2014).

Despite the perceived positive economic impact of trust (Bjørnskov 2012), empirical research on the link between social trust and economic development remains inconclusive, with results ranging between negative and significant associations (Schneider et al. 2000), non-significant associations (e.g. Akçomak and Ter Weel 2009; Beugelsdijk and Van Schaik 2005; Neira et al. 2009), and positive and significant connections (Tabellini 2010). Some suggest that the relationship is non-linear and heterogeneous for different types of regions (Peiró-Palomino 2016). Notably, the mechanisms driving this relationship have not been extensively explored, and empirical studies on the topic remain scarce. Despite the variability of empirical findings, we hypothesize, in line with the theoretical arguments, that:

Secondly, we explore the connection between social trust and political trust. Drawing on theoretical arguments about the relational aspect of political trust (Warren 2006), we theorize that political trust builds upon social trust. In societies with a high general sense of trust, this trust often extends to individuals within the government and the political system (Fukuyama 1995; Putnam 1993; Schneider et al. 2000; Newton et al. 2018). Therefore, individuals who exhibit social trust tend to be cooperative and accept the legitimacy of political institutions (Dellmuth and Tallberg, 2020). At the very least, a reasonable level of social trust is necessary for political trust to develop, while there may be diminishing returns once this level has been reached. Nevertheless, the mechanisms underlying this process remain insufficiently understood (Newton and Zmerli 2011; Newton et al. 2018).

Social capital theory (e.g., Tocqueville, 2000; Putnam 1993) offers a bottom-up perspective on political trust. Voluntary associations are viewed as democratic training grounds where citizens learn to trust political institutions. This theory also posits that the social trust generated by these associations extends to the political system. In contrast, socio-psychological theories, such as Uslaner (2002), argue that, beyond indirect mechanisms based on social capital theory, social trust directly influences political trust, as individuals transfer their general sense of trust in others to the political system.

Empirical studies (e.g., Newton and Zmerli 2011; Newton et al. 2018) reveal a strong correlation between social trust and political trust, indicating that social trust is a necessary but not sufficient condition for political trust. Prior work by Keele (2007) suggests that social trust precedes political trust, although some studies identify the reverse relationship (e.g., Dinesen et al. 2022). These studies primarily focus on national-level or individual-level analyses. We anticipate that the same relationship holds at the subnational level, as social trust arises from local interactions (Newton and Ramón, 2007; Putnam 1993). In a similar regional-level study, Kaasa (2016) identified a high correlation between social trust and political trust but did not delve into the mechanisms governing this relationship. Additionally, the potential for diminishing returns, where a certain level of social trust is required for political trust to flourish, remains unexplored. Accordingly, we propose the following hypothesis:

Quality of government refers to the extent to which governments deliver public goods efficiently, impartially, and free from corruption (Charron et al. 2010, 2014; Muringani et al. 2019; Rodríguez-Pose and Di Cataldo 2015). Government quality exhibits significant regional variation, even within countries, influencing the efficiency with which different regions provide public goods (Putnam 1993; Treisman 2002). High-quality regional governments mitigate opportunism and rent-seeking behaviors, fostering cooperation and effectiveness. Several empirical studies (e.g., Crescenzi et al. 2016; Muringani et al. 2019; Rodríguez-Pose and Di Cataldo 2015; Rodríguez-Pose and Garcilazo 2015; Rodríguez-Pose and Ketterer 2020; Rodríguez-Pose and Ganau 2022) have uncovered a positive association between regional government quality and economic development. This applies both to government quality as a whole and its individual components (efficiency, impartiality, and lack of corruption). At the country level, studies have also shown a non-linear relationship (e.g., Swaleheen 2011; Kim et al. 2018; Ochi et al. 2022), but this has not been examined at the regional level. Building on this body of work, we propose the following hypothesis:

Logically, higher-quality governments are likely to be trusted more, given their competence, effective economic management, and impartiality (Kumagai and Iorio 2020; Norris 2022). At a minimum, a satisfactory level of government quality may be required to gain public trust, while there may be diminishing returns also in this case. Empirical studies have identified specific dimensions of government quality, such as corruption or procedural fairness, as correlated with political trust (e.g., Wang 2016; Grimes 2017; Uslaner 2017). Khan (2016) reports a similar relationship for an integrated measure of government quality. Wang (2016) suggests that increased corruption reduces the effects of government performance on political trust, but government performance does not affect the relationship between corruption and political trust. In addition, there is evidence to suggest that political trust can enhance government performance (van de Walle and Bouckaert 2003).

However, the focus of this research has primarily been at the national level, leaving us with limited knowledge about how these dynamics manifest at the regional level. In contexts involving devolution, individuals have more interaction with regional government services, and their assessment of the quality of these services affects political trust. If a regional government is perceived as untrustworthy, citizens’ trust in politicians and the broader political system may decline, even in a high-trust society. While Kaasa (2016) identified a strong correlation between government quality and political trust at the regional level, the mechanisms underlying this connection have not been thoroughly examined. Furthermore, the potential for a non-linear relationship has been largely unexplored. Drawing from insights from national-level research, we propose the following hypothesis:

The maps in Figs. 2 and 3 display the average intensity of social and political trust across EU regions from 2002 to 2016. Figure 2 shows that social trust is considerably higher in Western Europe than in Eastern Europe. Within Western Europe, it is also higher in Northern Europe compared to Southern Europe. The highest levels of social trust are found in Scandinavia and parts of the Benelux, the British Isles, and Germany, while the lowest levels are in the Balkans, Poland and Southern Italy. However, there is also considerable heterogeneity within many countries and most notably in Italy and Germany.

Figure 3 depicts the geographical distribution of political trust in the EU. It mirrors the patterns seen for social trust. Political trust is highest in Western Europe, especially in Northern regions. Scandinavia, the Benelux, and Germany are prominent for their high levels of political trust. Conversely, political trust is far lower in Eastern Europe, notably in the Balkans, Greece and Poland.

Research on regional government quality in the EU (e.g., Charron et al. 2019) also reveals a divide between in Western and Eastern Europe. Scandinavia and the Netherlands exhibit the highest government quality scores. Within Western Europe, Northern regions outperform Southern regions in government quality, although both fare relatively well compared to most of Eastern Europe and, in particular, the south-eastern corner of Europe, as well as southern Italy. Considerable regional variation persists within many countries, most notably in Italy.

Over time, both trust and government quality exhibit relative stability, although with some gradual changes. Charron et al. (2022) note improvements in government quality in Eastern Europe and declines in Western Europe between 2010 and 2017. Other studies find no consistent patterns of change in political trust (Glatz and Eder 2020; Hooghe et al. 2017).

Overall, there is considerable overlap in the geographical incidence of these variables. However, it is not always the case that regions with high political trust show high levels of social trust and government quality. This observation aligns with findings at the individual level, as reported by Newton et al. (2018), indicating that high political trust does not necessarily imply high social trust, and the same holds true for government quality. This dynamic is evident in certain Italian regions, such as Lazio, where political trust is relatively high while social trust remains comparatively low. According to this perspective, social trust serves as the foundation for political trust, not vice versa. A similar rationale applies to government quality.

In the empirical analysis, we resort to a generalized structural equation model (GSEM). Unlike simple regression analysis, structural equation modelling allows for exploring the structural relationships between variables (Alesina and Giuliano 2015). It accounts for variances and covariances among disturbance terms across multiple equations (Sabatini 2008; 2009). Our primary focus is on the structural model, excluding measurement models because the variables are either precisely observed (economic development and control variables) or derived at the regional level from individual-level data obtained from other datasets (explanatory variables). As extensive research has previously discussed the scale development and validation of the explanatory variables of interest, including social trust and political trust (e.g., Jowell et al. 2007), and the EQI (e.g., Rothstein et al. 2013), we do not consider the measurement models a central contribution of this paper. Instead, our emphasis is on the structural relationships among the four main variables. We use a generalized SEM model to allow for the incorporation of time and region fixed effects.

The GSEM model comprises two equations:

In the first equation, we investigate the relationship between economic development and political trust (H1), social trust (H2a), and government quality (H3a). The second equation examines the relationship between political trust and social trust (H2b) and government quality (H3b). We model economic development and political trust as endogenous variables, as they are influenced by other variables. Conversely, social trust, government quality, and the control variables are modelled as exogenous variables, meaning they are not influenced by other variables in the model.³\({\beta }_{1}\text{ to}\, {\beta }_{6}\) are the coefficients, while \({\gamma }_{1}\) and \({\gamma }_{2}\) are the region fixed effects, \({\tau }_{1}\) and \({\tau }_{2}\) stand for time fixed effects, and ε₁ and ε₂ for the disturbance terms.

As a second step, we introduce quadratic product terms to assess non-linearity:

To address endogeneity concerns, we conduct robustness tests in two ways. First, we incorporate spatial lags into the GSEM model. One model includes lags for all independent and control variables over a two-year period, while the other includes lags for all independent variables except political trust. However, these lagged variables may not fully address endogeneity concerns due to potential dynamic relationships among unobservable variables (Bellemare et al. 2017).

As a more rigorous step from an econometric perspective, we employ a two-stage least squares instrumental variables (2SLS IV) regression to tackle endogeneity. Integrating this approach within GSEM estimation presents complexities since some regressors in GSEM are endogenous. In our case, political trust—the main variable of interest—is itself an endogenous regressor influenced by social trust and government quality. Thus, we can only instrument for one component of the structural equation. We choose to instrument for political trust using an instrument previously used for both government quality and social trust, as these factors logically influence political trust within the model. Specifically, we use literacy rates from the 1880s as the instrument. This instrument has been used to proxy for social trust by Tabellini (2010) and for government quality by Charron and Lapuente (2013). Historical literacy rates are associated with higher levels of social trust due to their role in facilitating communication with strangers. They also correlate with improved government quality, as literacy is essential for government transparency and public accountability. These combined mechanisms are expected to result in higher political trust. In the IV regression, we employ literacy rates from the 1880s as an instrument for political trust and exclude social trust and government quality, as these variables are also influenced by the instrument. Consequently, this model permits an examination of the causal effect of political trust on economic development.

As discussed, a significant concern when estimating these relationships is the potential impact of endogeneity or reverse causality on the findings (e.g., Bargsted et al., 2023; Bjørnskov, 2012; Van de Walle and Bouckaert 2003). To address this issue, we employ two distinct approaches.

First, we incorporate time lags into the analysis, as presented in Appendix Table A5. By introducing time lags, we elucidate how explanatory variables and controls influence GDP per capita in the subsequent period (two years later). The results closely resemble those of the original analysis in Table 3, which lacked time lags. However, this time, the relationship between social trust and economic development emerges as positive and statistically significant. Meanwhile, for political trust, the effect on economic development appears slightly stronger with the inclusion of lags, while for government quality, it appears slightly weaker. In a separate analysis, we also introduce a lag to political trust, allowing social trust and government quality to explain political trust two years later. In this scenario, social trust and government quality maintain their positive and statistically significant relationships with political trust, although with somewhat weaker coefficients. Notably, all three variables remain significantly associated with economic development.

Second, we resort to an instrumental variable regression design, as detailed in Appendix Table A6. Here, we use literacy rates in the 1880s as an instrument to proxy for political trust, following Tabellini (2010). In this 2SLS IV regression design, we exclude social trust and government quality as explanatory variables, as these factors are also influenced by the instrument. The initial two columns of Appendix Table A6 display the results of the first stage of the 2SLS IV regression. Due to a lack of available data on literacy rates in all regions, the analysis is based on a reduced sample of 130 regions. Additionally, region fixed effects cannot be included because literacy rates in 1880 do not exhibit intra-regional variation. To facilitate result interpretation, we present the GSEM results for the same model—one that excludes social trust, government quality, and region fixed effects—for both the full sample of regions and the reduced sample of 130 regions, as a point of comparison.

The Cragg-Donald statistic is 409.447, which is well above the Stock-Yogo critical value (16.38 at 10% maximal IV size), indicating robust instrument strength. The Kleinbergen-Paap rank LM test is significant, indicating no evidence for underidentification. The results of the IV regression reveal a statistically significant and highly positive effect of political trust on economic development. In summary, we conclude that the impact of political trust on economic development remains robust even when taking into account endogeneity concerns. Table A7 presents a comparison of fit indices for all GSEM models.