Portfolio Choice with Puts: Evidence from Variable Annuities

This study investigated the asset allocation behavior of individuals who select an out-of-the-money long-dated longevity-put option on their investment funds. The asset allocations of these people within their variable annuity subaccounts are 5–30 percent more risky than the allocations of those who do not choose this protection. Investors who do not choose the longevity-put option follow the classic life-cycle, age-phased reduction in equity. A rudimentary model of utility-maximizing behavior is suggested that justifies the increased allocation to risk as long as the investor understands the payoff structure of the longevity put and is willing and able to exercise the annuity option if and when it matures in the money.In the study reported here, we investigated the actual asset allocation behavior of individuals who purchase portfolio insurance in the form of an out-of-the-money long-dated longevity-put option on their investment funds. We compare their asset allocations in these special accounts with those of investors who do not elect to pay for any additional insurance protection. The empirical evidence comes from variable annuity (VA) policies made available by insurance companies in the United States. Using a unique database of nearly 1 million (anonymous) policyholders contributed by seven insurance companies, we found that, all else being equal, investors assume 5–30 percent more risky-asset (equity) exposure when they have selected a longevity-put option. Furthermore, when they are “protected,” their exposure to risky assets is relatively constant across ages of policyholders in our dataset. But when this longevity put is not purchased—so that the economics of the investment portfolio resembles a conventional (tax-sheltered) mutual fund—we confirmed the age-phased reduction in risky-asset exposure. We also found a strong “intermediary effect” whereby the allocation percentages depend on the individual or institution that intermediated the transaction between the insurer and the investor. In addition, we offer a rudimentary model of utility-maximizing behavior in the presence of this longevity-put option that justifies the observed increased allocation to risky assets, provided that the investor truly understands the payoff structure of the longevity put and is willing and able to exercise the annuity option if and when it matures in the money. The assumption that these options will be exercised if they expire in the money is debatable, of course, in light of the long-standing body of evidence that individuals dislike annuitization.We believe that our study is the first to examine actual asset allocations within VA policies. We expect a continued interest in this topic, partly because the VA market is a $1.5 trillion dollar market in the United States and is expected to grow as aging Baby Boomers take control of their retirement assets and try to generate their own pensions.