Cognitive abilities and portfolio choice
Introduction
Transaction and information costs, broadly interpreted, have been offered as the main reason why many individuals do not invest in either stocks or other financial assets (Haliassos and Bertaut, 1995; Vissing-Jorgensen, 2004). The exact nature of these costs, however, is not well understood, and isolating the factors that prevent large sectors of the population from holding stocks is a challenge for research in household finance. In this paper we focus on the lack of cognitive abilities (as measured by mathematical and verbal skills and memory functioning) as a potential explanation for limited financial market participation.
There are several channels through which cognitive skills might affect the decision to buy stocks and other financial assets. First, the cost of gathering and processing information is lower for skilled individuals; therefore, low skills can act as a barrier preventing stockholding. Second, cognitive skills tend to be associated with preference features, such as risk aversion, that have a major influence on the willingness to bear financial risk. Third, the perception of risk is also likely to depend on cognitive abilities: low cognitive skills can make some investors overestimate the precision of the information that they possess. Overconfident investors trade more and take more financial risk than rational agents with unbiased perceptions, which implies a negative relation between cognitive skills and stockholding.
In order to study the impact of cognitive factors on stockownership, we exploit within and across country variability in cognitive abilities, stockholding and socioeconomic variables found in data from the first wave of the Survey of Health, Ageing and Retirement in Europe (SHARE), which surveys people aged 50 and above in 11 European countries: Austria, Belgium, Denmark, France, Germany, Greece, Italy, the Netherlands, Spain, Sweden, and Switzerland. The survey asks detailed questions on demographics, physical and mental health, employment, income, assets, social activities, and expectations. All questions are standardized across countries, thus allowing consistent international comparisons.
Most importantly for our purposes, SHARE provides detailed information on reading, mathematical, and recall ability. Cognitive abilities are known to vary across countries, even after taking education into account. For instance, evidence from the OECD Programme for International Student Assessment (PISA) shows that among 15 years olds with the same level of schooling there is large international variability in mathematical and science test scores, and that there is a consistent North–South gradient in skills (Hanushek and Woessamn, 2009). Given the age composition of our sample, we need to recognize in our empirical analysis that the elderly face a substantial mortality risk (especially at advanced ages), which reduces their planning horizon. Moreover, while the retired face more limited labor income risk, they also face much higher uncertainty about medical expenditures. As a result, it is particularly important in our context to control for the effects of investors’ horizon, health status and bequest motives on the willingness to bear financial risk.
To the best of our knowledge, our paper is the first one to empirically demonstrate the effect of cognition on stockholding. We indeed find that cognitive impairment, as manifested by a reduced ability to perform numerical calculations, to read and to recall, has a strong negative influence on stock market participation. Furthermore, this result is net of the effects of age, education, health and economic resources, and it is also robust for different specifications of the cognition and other explanatory variables.
Given that the willingness to bear financial risk may change over the life-cycle, in our analysis we need to disentangle the effect of cognition on stockholding from the corresponding effects of age. We therefore split our sample into age groups (younger and older than 65), which allows us to check whether the correlation between cognitive abilities and stockholding is mainly driven by the choice of the oldest segment of the population.3 Our results show that there are no appreciable differences between age groups, which implies that ageing is not the main reason behind the correlation between cognition and stock market participation. On the other hand, we find that the propensity to hold bonds, which require less information than stocks, is affected by cognitive abilities very weakly or not at all; therefore, the correlation between cognition and stock market participation is likely due to informational barriers.
Finally, our paper offers a systematic account of the main factors affecting household portfolio choice in Europe using fully comparable microeconomic data. It represents therefore a significant improvement over previous work based on individual country data.
Understanding the reasons for financial market participation is important for many reasons. Cocco et al. (2005) show that in calibrated life-cycle models the welfare loss from no stockholding is between 1.5 and 2 percent of consumption. Limited participation and changes in participation over time are also relevant for the equity premium puzzle (Mankiw and Zeldes, 1991; Attanasio et al., 2002), the distribution of wealth (Guvenen, 2006), ownership of individual retirement accounts (Bernheim and Garrett, 2003), and wealth effects on consumption (Dynan and Maki, 2001).
From a policy point of view, European pension reforms are likely to increase reliance on individual retirement accounts and thus investors’ exposure to stock market investment. These investment opportunities could bring higher expected returns, but excessive or ill-advised trading of stocks can significantly reduce realized returns. In addition, poor judgment in allocating retirement wealth can create major financial distress at a point in the life-cycle where the potential for offsetting adjustments is quite limited. Therefore, policies that aim to improve the quality of investors’ financial information and awareness depend crucially on the influence of cognitive abilities on financial decisions.
The rest of the paper is organized as follows. Section 2 provides a simple framework to understand how cognition might affect stock market participation. Section 3 describes the microeconomic data and our three indicators of cognitive abilities, and Section 4, the set of variables that will be used in the empirical analysis to explain stockholding decisions. Section 5 presents the empirical results for the probability of investing in stocks, either directly or through mutual funds and investment accounts, performs robustness analysis and compares the results with the determinants of bondholding. The results are summarized in Section 6.
Section snippets
The effect of cognitive abilities on stock market participation
There are at least three different channels through which cognitive abilities might affect the decision to invest in stocks. First of all, managing a portfolio requires a specific human capital investment, in terms of time and effort needed to familiarize oneself with the notions of transaction costs, asset returns, volatility, and covariance between assets returns. Information costs represent therefore a significant barrier to entry in the stock market. Low cognitive abilities are likely to
Measuring cognitive abilities
We study the relation between cognitive abilities and the decision to hold stocks either directly or indirectly, through managed accounts or mutual funds. We use the first wave of SHARE, which is a representative sample of those aged 50 and above in 11 European countries. The interviews took place in 2004 and 2005, and wealth data refer to December of the year prior to the interview.7, 8
Empirical specification
While traditional finance theory predicts that investor's willingness to take financial risks depends only on risk aversion and investment opportunities, dynamic models of portfolio choice emphasize that investment opportunities and wealth itself change over time, that investors usually face other background risks, and that transaction costs, information costs and borrowing constraints limit household financial decisions. Gollier (2002) surveys some of the recent developments. On the empirical
Econometric results
To model the participation decision, households compare the utility gain from owning stocks with the entry cost. We express the net utility gain aswhere h is the household index, and zh are the observable variables affecting the utility gain from owning stocks. Utility also depends on unobservable variables, which enter (1) through εh, a standard normally distributed random term. Household h owns stocks if , which implies that the probability of observing stockownership is
Conclusions
SHARE documents that there is substantial heterogeneity in stock market participation, both within and across countries. As recent evidence for EU countries and the US shows, this is not a pattern particular to SHARE; heterogeneity in stockownership is a widespread phenomenon, with large differences across countries. It bears repeating, however, that SHARE data are collected on a comparable basis for 11 European countries; therefore, they constitute an extremely rich source of information that
Acknowledgments
We thank for helpful comments Dimitris Georgarakos, Luigi Guiso, Michalis Haliassos, Mike Hurd, Anna Sanz de Galdeano, and participants at the Conference on Housing and Household Portfolios (Copenhagen, 9–11 June 2005), CSEF-IGIER Symposium on Economics and Institutions (Anacapri, 22–27 June 2005), Fourth Conference on Research on Economic Theory and Econometrics (Syros, 11–14 July 2005), First Share-ELSA-HRS User Conference (Lund, 26–28 September 2005), Luxembourg Wealth Study Conference
References (56)
- et al.
Participation and investment decisions in a retirement plan: the influence of colleagues’ choices
Journal of Public Economics
(2002) Reconciling conflicting evidence on the elasticity of intertemporal substitution: a macroeconomic perspective
Journal of Monetary Economics
(2006)- et al.
The consumption of stockholders and nonstockholders
Journal of Financial Economics
(1991) - et al.
Portfolio choice and health status
Journal of Financial Economics
(2004) Implications of rational inattention
Journal of Monetary Economics
(2003)- Ameriks, J., Zeldes, S.P., 2004. How do Portfolio Shares Vary with Age? Mimeo. Columbia...
- et al.
Asset holding and consumption volatility
Journal of Political Economy
(2002) - et al.
Boys will be boys: gender, overconfidence, and common stock investment
Quarterly Journal of Economics
(2001) - et al.
Individual preferences, monetary gambles and stock market participation
American Economic Review
(2006) - et al.
Myopic loss aversion and the equity premium puzzle
Quarterly Journal of Economics
(1995)