On the determinants of happiness: a classification and regression tree (CART) approach

Abstract

This article studies the determinants of the individual’s subjective well-being by applying a classification and regression tree (CART) analysis to data from the Survey on Household Income and Wealth (SHIW) provided by the Bank of Italy. The results support the primary importance of economic conditions but show that their effect is heterogeneously dependent on other individual characteristics.

Keywords:

• Happiness
• classification trees
• Italy

JEL Classification:

• I31
• C45
• C49

Acknowledgements

I wish to thank Massimo Bordignon and Ada Ferrer-i-Carbonell for insightful comments.

Notes

¹ For a recent survey, see Stutzer and Frey (2012).

² For a detailed technical explanation, see Breiman et al. (1984).

³ Scoppa and Ponzo (2008) use SHIW from the years 2004 and 2006 producing the first empirical study on happiness focused on Italy.

⁴ Alesina, Tella, and MacCulloch (2004) in Section 2.2 discuss in more detail this point.

⁵ Few monetary variables report negative values. This is due to the way in which the Bank of Italy aggregates the different components. However, it is worth to note that even if these values were misreported data, it would not be an issue because classification trees would handle it efficiently (Breiman et al. 1984).

⁶ All results provided in this article are derived by using the rpart module implemented in R.

⁷ The predicted type is happy if the probability is higher than 50%.

Additional information

Funding

Financial support from the Swiss National Science Foundation (grant Early Postdoc.Mobility - 158603) is gratefully acknowledged.