The proposed research will build on research that attempts to replicate the association found between the serotonin transporter gene (SLC6A4) - specifically variation in the transcriptional control region (the 5-HTT-linked polymorphic region (5-HTTLPR)) - and well-being.

If an association is found, this research proposes to investigate whether Mendelian randomisation can be applied to study the effects of well-being on educational outcomes and social networks. There are a number of considerations to take into account before using genetic markers as instruments (see inter alia Lawlor et al. 2007[1], Lawlor et al. 2008[2], von Hinke Kessler Scholder et al. 2010[3]). An in depth study of whether the 5-HTTLPR and rs25531 can plausibly be used as instruments for well-being is warranted given an increasing interest in genes in economics literature.

While many studies investigate the determinants of well-being, positing that income, health and relationships can explain why some people are happier than others but Lyubomirsky et al. (2005)[4] argue that well-being is important in shaping outcomes - happier people are more successful at cultivating personal relationships and do better in the workplace. However, the bulk of evidence suggesting that well-being affects outcomes comes from cross-section or longitudinal studies and experimental studies. Cross-sectional studies document whether a correlation exists but not the direction of causality. On the other hand, if well-being is measured prior to the outcome, a more convincing case can be made that well-being caused the outcome, which is an advantage of longitudinal data. But this approach falters if well-being is highly correlated over time and it says little about what is driving changes in lagged well-being and therefore if lagged well-being is truly exogenous. Experimental studies provide a gold standard for studying the direction of causation, but by design these studies focus only on short-term changes in well-being. A suitable instrument for well-being would shed light on how long-term well-being affects outcomes and therefore assessing whether the 5-HTTLPR and rs25531 can be used as instruments for well-being, focusing on two different applications is a worthwhile exercise.

The focus on educational outcomes is motivated by the fact that educational outcomes are linked to job prospects and earnings and are therefore informative of the likely impact of well-being on economic circumstances. However, it is also an outcome where the exclusion restriction required for instrumental variable estimation may fail. Frisch et al. (2005)[5] previously show that well-being can predict academic retention. This research will examine whether the 5-HTTLPR and rs25531 are potentially useful instruments in an application to national exam scores.

The formation of social networks is an interesting area to verify whether mendelian randomisation is feasible because previous research on homophily shows that people with similar characteristics flock together (McPherson et al. 2001)[6] and research cited in Lyubomirsky et al. (2005)suggests that happier people are more appealing to others. The research will therefore examine whether this approach can be used to answer questions such as are people with similar levels of well-being more likely to befriend each other, does well-being influence the number of friends one has and position in friendship networks?

If an association is not found, the proposed research would investigate whether there is a gene-environment interaction using plausibly exogenous life stresses such as a death in the family.