Generalized trust and happiness:

Happiness has always been the domain of philosophy and psychology, but since the 1970s happiness has become a topic of interest to economists as well. To traditional economists, utility refers to people's preferences over goods and services. The implicit assumption being that maximizing utility will make people happy. The economics of happiness considers utility in a perspective that observes happiness directly, often using data based on large-scale social surveys. In these surveys people are asked the following (or a similar) question: “Taking all things together, how happy would you say you are?”. Several factors have been associated with happiness, such as income and health. In this research I investigate the association between generalized trust – the trust people have in strangers – and happiness. The rationale behind such association is that not trust frees you from the cost of having to deal with risk and uncertainty. For this research I used a dataset that was compiled on the basis of primarily the European Social Survey (2002-2011). The dataset contains data on generalized trust and selfreported happiness, as well as on several other socio-economic variables. The dataset covers 33 European countries, including Turkey and Russia. Using Ordinary Least Square (OLS) regressions I show that there is indeed a strong and statistically significant relationship between generalized trust and happiness. However, research on generalized trust and happiness is plagued by endogeneity. In order to increase the robustness of my findings, I perform additional regressions – 2-Stage Least Square (2SLS) regressions, also called Instrumental Variable (IV) regressions. The 2SLS regressions use so-called instrumental variables to estimate the independent variable of interest (in this case, generalized trust). This estimate is consequently used to estimate the dependent variable (in this case, happiness). Using 2SLS regressions the endogeneity problem is mitigated. The results of the 2SLS confirm the results obtained with OLS regressions, yet are dependent on the exact model specification.