Abstract

This paper tests the OECD conceptual framework for trust in institutions. We use the survey on trust in institutions run by the OECD in 18 countries in 2021. Employing a structural equation model, we estimate a confirmatory factor model for three underlying dimensions: trust in politics, trust in institutions, and the quality of Governance. These factors scores are then used in a regression analysis in order to explore systematically the socio-economic determinants such as age, gender, and education level as described in the OECD report in 2022 (OECD 2022). Our study corroborates almost all the socio-demographic effects on trust as described in the report. Furthermore, our study highlights the important role of the quality of governance as a robust predictor of overall trust levels. Finally, to check of the robustness of the OLS regressions, we employ also distributional regression approach using non-parametric regressions (RIF). Our findings reveal distinct patterns: at lower quantiles of trust, gender, education level, and income exhibit positive and significant relationship. Conversely, at higher quantiles, particularly the 0.90 quantile, their impact shifts to a negative association. These results underscore the importance of checking non-linearity associations, which may yield markedly different outcomes compared to traditional ordinary least squares (OLS) frameworks.