VoxEU Column Financial Markets

The financial mistakes of households and their social costs

Most households do not diversify but instead invest in only a handful of stocks, typically ones with which they are familiar. Using a new framework for evaluating the welfare losses from this underdiversification, this column argues that when the effect of familiarity biases on a household’s decision to allocate wealth between risky and safe assets and on its consumption-savings decisions are taken into account, the welfare loss is amplified by a factor of four. The impact on household and social welfare of financial policies through innovation, education, and regulation could thus be substantial.

Diversification is the only free lunch in investing (Markowitz 1952, 1959). Yet, empirical evidence shows that most households do not diversify. Instead, they invest in only a handful of stocks, typically ones with which they are familiar (Huberman 2001). For example, Finnish households hold only a few stocks, usually of firms that are located nearby (Grinblatt and Keloharju 2001). An extreme example of investing in familiar assets is the holding of stock in the company by which one is employed. Enron provides one of the most egregious illustrations of this. At the end of 2000, almost two-thirds of Enron employees’ retirement assets were invested in Enron itself; between January 2001 and January 2002, the value of Enron stock fell by over 99%. Enron is not a special case – five million Americans have over 60% of their retirement savings invested in own-company stock (Mitchell and Utkus 2004). Remarkably, only 33% of the households who own company stock realise that it is riskier than a diversified fund with many different stocks (Benartzi et al. 2007).

Investing in underdiversified portfolios that are biased toward a few familiar assets forces households to bear more financial risk than is optimal. Calvet et al. (2007) empirically study the importance of household portfolio diversification for welfare using data on Swedish households. They find, within a static mean-variance framework, that the welfare costs for individual households arising from underdiversified portfolios are equivalent to a reduction of about 1% per year in a household’s portfolio return.

Evaluating the welfare losses from underdiversificiation: A new framework

In a recent paper (Bhamra and Uppal 2018), we revisit this question under a richer framework that allows underdiversification to also impact asset allocation (i.e. the share of wealth to allocate to risky assets and the share to the safe asset), intermediate consumption (i.e. how much to consume at each date over a household’s lifetime), and aggregate growth. The advantage of this framework is that it provides additional channels through which underdiversification can lead to welfare losses. We consider a model of a production economy with a large number of firms whose physical capital is subject to exogenous shocks. We also allow for heterogeneous households with preferences that exhibit familiarity biases. Each household is assumed to be more familiar with a small subset of firms, with this subset varying across households. Familiarity biases create a desire to concentrate investments in a few familiar firms rather than holding a portfolio that is well diversified across all firms. Importantly, we specify the model so that familiarity biases cancel out across households, implying that our results are not a consequence of aggregate familiarity biases.

We now explain the economic mechanism driving our main results. Because of the familiarity-induced tilt, each household’s portfolio return is excessively risky relative to the return of the optimally diversified portfolio absent familiarity biases. This excessively risky portfolio is the first source of welfare loss for the household. In response to this excessively risky portfolio, the household pulls money out of risky assets and moves it into the safe asset. The misallocation across risky and safe assets decreases the expected return on the household’s investment, which leads to the second source of welfare loss. The reduction in expected return on each household’s investment distorts its consumption-saving decision in the same direction, which is the third source of welfare loss. Upon aggregation across all the households in the economy, the biased consumption-saving decisions of individual households add up to distort the aggregate growth rate and reduce social welfare.

Our findings

Our work delivers two key insights: 

  • First, even if the welfare loss to a household from investing in an underdiversified portfolio is modest, once we incorporate the effect of familiarity biases on the household’s decision to allocate wealth between risky and safe assets and on its consumption-savings decision, the welfare loss to the household is amplified by a factor of four. 
  • Second, we show that even if one were to force the familiarity biases in portfolios to cancel out across households, their implications for consumption and investment choices would not cancel out. This is because holding a portfolio that is excessively risky leads all households to distort their consumption-savings decision by reducing how much they save. Consequently, household-level distortions to individual consumption upon aggregation have a substantial effect on aggregate growth and social welfare in general equilibrium.

Overall, combining the impact of underdiversification on intertemporal consumption and aggregate growth amplifies social welfare losses by five to six times that of the direct losses from underdiversification. These potential gains are equivalent to an increase in the return on aggregate wealth of about 5% per year. Most of this gain arises from a multiplier effect applied to the gains for a household with utility defined over risk and expected return (mean-variance utility). This multiplier effect is driven by the impact of improved portfolio diversification on intertemporal consumption smoothing at the microeconomic level and on aggregate growth at the macroeconomic level.

Policy implications

These insights suggest that the impact on household and social welfare of financial policies through innovation, education, and regulation can be substantial.

One such policy measure would be to ‘nudge’ households toward default portfolios that are well diversified (Thaler and Sunstein 2003). For instance, households could be offered a small number of portfolios to choose from, with the portfolios having different levels of risk, but all of them being well diversified.

Another way to reduce the welfare consequences of underdiversification is to offer financial education to households. For example, households could be educated about the benefits of diversification that result from investing in broadly diversified funds, such as mutual funds and ETFs. Empirical evidence suggests that financial literacy can play an important role in improving decisions made by households (Bayer et al. 2008, Dimmock et al. 2014).

A third measure would be to introduce financial regulation to curtail the tendency of households to bias portfolios toward a few familiar assets. For example, financial regulation could be introduced to prohibit companies from providing employees own-company stock when matching the pension contributions of employees.

Concluding remarks

The benefits of diversification were recognised in both ancient times and the middle ages – diversification is mentioned in the Bible (in the book of Ecclesiastes), in the Talmud, and also in literary work as early as the 16th century (Merchant of Venice by Shakespeare). What our work shows is that, in addition to the direct benefits of diversification, there are substantial indirect benefits both for individual households and for the aggregate economy. Therefore, encouraging households to hold better-diversified portfolios would not only benefit individual households but also boost economic growth.


Bayer, P J, D B Bernheim, and J K Scholz (2008), “The Effects of Financial Education in the Workplace: Evidence from a Survey of Employers”, Working Paper, Duke University. 

Benartzi, S, R H Thaler, S P Utkus, and C R Sunstein (2007), “The Law and Economics of Company Stock in 401(k) Plans”, Journal of Law and Economics 50: 45–79. 

Bhamra, H S, and R Uppal ( 2018), “Does Household Finance Matter? Small Financial Errors with Large Social Costs”, American Economic Review, forthcoming. 

Calvet, L E, J Y Campbell, and P Sodini (2007), “Down or Out: Assessing the Welfare Costs of Household Investment Mistakes”, Journal of Political Economy 115(5): 707–747. 

Campbell, J Y (2006), “Household Finance”, Journal of Finance 61(4): 1553–1604.

Dimmock, S G, R Kouwenberg, S O Mitchell, and K Peijnenburg (2014), “Ambiguity Aversion and Household Portfolio Choice Puzzles: Empirical Evidence”, Netspar Discussion Paper No. 09/2012-035. 

Grinblatt, M, and M Keloharju (2001), “How Distance, Language, and Culture Influence Stockholdings and Trades”, Journal of Finance 56(3): 1053–1073. 

Guiso, L, and P Sodini (2013), “Household Finance: An Emerging Field”, in G M Constantinides, M Harris and R M Stulz (eds), Handbook of the Economics of Finance, Elsevier. 

Guiso, L, M Haliassos, and T Jappelli (2002), Household Portfolios, MIT Press. 

Haliassos, M ( 2002), “Stockholding: Recent Lessons from Theory and Computations”, Working Papers in Economics, University of Cyprus. 

Huberman, G (2001), “Familiarity Breeds Investment”, Review of Financial Studies 14(3): 659–680. 

Markowitz, H M (1952), “Portfolio Selection”, Journal of Finance 7: 77–91. 

Markowitz, H M (1959), Portfolio Selection: Efficient Diversification of Investments, Wiley. 

Mitchell, S O, and S Utkus (2004), “The Role of Company Stock in Defined Contribution Plans”, in S O Mitchell and K Smetters (eds), The Pension Challenge: Risk Transfers and Retirement Income Security, Oxford University Press. 

Thaler, R H and C R Sunstein (2003), “Libertarian Paternalism”, American Economic Review 93(2): 175–179. 

Vissing-Jorgensen, A (2003), “Perspectives on Behavioral Finance: Does “Irrationality” Disappear with Wealth? Evidence from Expectations and Actions”, NBER Macroeconomics Annual: 139–193.

1,365 Reads