Household stock portfolios under sequential search: implications for diversification puzzle and household risk

We develop a theory of sequential stock search to explain U.S. household portfolio composition, focusing on the ‘diversification puzzle’ and its relationship with wealth, search costs, and portfolio risk. In this approach, investors only include a stock in their portfolio if it improves expected utility, accounting for informal effort and professional search costs. Using data from the Survey of Consumer Finances, we find that under rational expectations, investors are incentivized to search for and rank stocks by expected return, ultimately selecting only a few with the highest potential. This behavior helps explain why most households do not follow simple diversification strategies, such as investing in index funds. Our model aligns with empirical evidence: the median stockholding household owns three stocks, and 33.2% own just one. Simulations calibrated to SCF data show a hump-shaped pattern in total financial risk, which increases with wealth before declining, highlighting the complex dynamics that drive household investment decisions.