Household portfolio decisions are rational after all

Laurent Calvet, Professor of Finance - October 14th, 2014
Household portfolio decisions are rational after all by Laurent Calvet

Habit, genetic factors, socieconomic background, wealth, standards of living, family size, location... what parameters are households portfolios based on? Calvet and Sodini’s ground-breaking study on how twins  invest their savings provides clear answers to this complex issue.

Laurent E. Calvet ©HEC Paris

Laurent E. Calvet is a professor of finance at EDHEC. He was research professor at HEC Paris from 2004 to 2016. A graduate of École Polytechnique, Calvet holds a Ph.D. in (...)

See CV

How do individuals make financial decisions? Do some people have a particular innate taste for risk not shared by others? It is difficult to disentangle the factors involved in decision making because we are all so different from one another. Several studies have suggested that the choices made by households are relatively irrational, which does not inspire much confidence in the ability of families to make investment decisions. However, it now seems that these findings are not entirely justified; according to Calvet, “these conclusions are generally the result of the fragmentary nature of the databases that have been used.”

How can investment factors be isolated?

When Sodini (see methodology) identified a database of exceptional quality and value, Calvet decided to focus his research ontwins, who share the same genes, the same family environment, and often the same friends. These similarities make it is possible to control for social, cultural, and genetic parameters and thereby analyze in detail the influence of an individual’s specific characteristics and life events. In other words, t twins permit to evaluate the impact of various criteria that may be involved in their investment decisions while controlling for latent sources of heterogeneity.

Financial wealth is the decisive factor in risk-taking

Calvet and Sodini demonstrate that financial resources are one of the most significant criteria in household risk-taking. The more wealthy individuals are, the more likely they are to invest a substantial part of their income in risky assets. Likewise, the more they expect to receive a high income in the future, the more they take financial risks. By contrast, consumption patterns, expenditure commitments (such as mortgages) and family size (the number of children) make households more cautious. Furthermore, fraternal twins who are in frequent contact make more similar choices than identical twins who interact infrequently. This indicates that, although genetic factors may have an influence, other factors matter just as much. “The choices made by households are not predetermined by their DNA,” says Calvet. “Life circumstances (wealth, expected income, family characteristics, loans, and so on) are extremely important in their decision-making process, which is reassuring news for rational financial theory.” 


Fraternal twins who are in frequent contact make more similar choices than identical twins who interact infrequently

Different sensitivities to price fluctuations 

Household responses to changes in the performance of financial assets are generally consistent with financial theory : they tend to rebalance the proportion of risky assets in their overall portfolio when it rises or falls. Calvet, Campbell and Sodini (“Fight or Flight? Portfolio Rebalancing by Individual Investors”) show that they do not magnify market shocks and that their portfolios are, by and large, relatively well diversified. Changes in income influence households differently depending on their financial resources. If people who make few investments are given an additional sum of money, they make the most of it and select a more aggressive asset allocation. On the other hand, when people who are already wealthy receive additional wealth, they make only small changes to their asset allocation. The sensitivity of asset allocation to wealth changes is therefore weaker at the macro level than at the level of the median investor. This makes sense: it is the wealthiest people (whose risk-taking is relatively less sensitive to changes in liquid financial wealth) who own the majority of financial assets and who, due to the volume of their transactions, determine the aggregate asset allocation of the household sector and asset prices.

Based on an interview with Laurent E. Calvet and his articles “Twin Picks: Disentangling the Determinants of Risk-Taking in Household Portfolios”, co-written with Paolo Sodini (Journal of Finance Vol. LXIX No. 2, April 2014), and “Fight or Flight? Portfolio Rebalancing by Individual Investors”, co-written with John Y. Campbell and Paolo Sodini (Quarterly Journal of Economics 124, pp. 301-348, February 2009).

For this article Laurent E. Calvet will receive the HEC Foundation's best scientific paper award in December 2014.

Practical Applications
Practical Applications

Household decisions are much less irrational than has long been claimed. Wealth levels, together with anticipated income, are the most influential parameters for financial risk-taking. This is an outcome that may interest bankers, who often bemoan the dearth of information about the decision-making criteria of their customers. These results also provide valuable information to governments and various organizations considering what mechanisms to put in place to ensure more comfortable levels of retirement for households. Finally, the study provides a micro perspective on stock market valuations  over the business cycle: the micro data show that risk premia fluctuate with the level of household wealth. Accordingly, in a recession (for example) less wealthy individuals take on fewer risky assets, thus reducing demand and increasing the risk premium. 


Calvet and Sodini analyze the Swedish government's tax database, which includes  demographic information  (family characteristics, education, employment, salary, location) and detailed financial and real estate holdings  (bank balances, marketable securities) of  every taxpayer in the country, and in particular more than 23,000 pairs of twins. As the data is recorded on an annual basis, it allows for multi-year dynamic analyses. Unlike the fragmentary databases commonly used in this field of research, the completeness of the Swedish data makes it possible to undertake detailed and reliable analyses.