The portfolio choices of investors over the stock market and life cycle.

Summary Do investors in risky financial securities tend to invest against the grain in the stock market, buying at the highs and selling at the lows? Do they reduce their risk exposure as they age and especially as they approach retirement? We answer these two questions using data from a major French insurer listing Madelin contract subscriptions between 2002 and 2009. Subscribers can invest their savings in two types of funds: a euro fund composed essentially of money market securities that are virtually risk-free, and unit-linked funds representing mutual fund shares invested in securities with risky returns. We show that the part of the capital invested in units of account is sensitive to the stock market situation, but essentially to the date of subscription of the contract. Once the initial share has been selected, a strong inertia of portfolio choices is observed, since investors very rarely change the decision they made at the beginning of the contract. We observe a strong procyclicality of investment choices which is explained by an extrapolation of recent stock market performance. New subscribers buy risky assets when the stock market rises and stop buying them when it falls. This leads them to hold a minimum share of risky assets in 2004, at the beginning of a four-year bull market, and a maximum share in 2008, at the beginning of the stock market decline due to the financial crisis. We also find that the risky share declines steadily with age once we control for time effects and exclude generation effects. The profile by age also declines in the opposite configuration (taking into account generation effects and excluding time effects) but the decline is less pronounced. It is in fact the product of two phenomena. First, the number of savers investing in risky assets tends to increase with age. On the other hand, conditional on investing, the risky share decreases with age. After a discussion of the plausibility of the different effects, we estimate a probability of holding units of account that increases with age by about 12 percentage points between the ages of 40 and 60, and a share invested in units of account conditional on holding a positive share that decreases with age by about 6 percentage points between 40 and 60. The decrease is too small to bring the invested portion to zero as retirement approaches.