Michaela Pagel is an Associate Professor (without tenure) at Columbia Business School. She received her Ph.D. from the Economics Department at UC Berkeley and works on topics in behavioral economics, household finance, and macroeconomics. Her dissertation focused on the consumption and investment implications of non-standard preferences. More specifically, she theoretically studied how decision-making is affected by people’s beliefs about their consumption. Her current work analyzes transaction-level data on income, spending, balances, credit limits, and logins stemming from a financial aggregation app. Furthermore, she is working with bank account data linked to individual investors’ security trades and portfolios.
We use transaction-level data of portfolio trades and holdings linked to checking, savings, and settlement account transactions and balances to explore how individuals respond to realized capital gains and losses. To identify the effects of realized gains and losses, we exploit plausibly exogenous mutual fund liquidations. Specifically, we estimate the marginal propensity to reinvest one dollar received from a forced sale event, when the investor either achieved a capital gain or a loss relative to his or her initial investment. Theoretically, if individuals held optimized portfolios, the marginal propensity to reinvest out of forced liquidations should be 100% independent of realizing a gain or a loss. Individuals should just reinvest all of their liquidity immediately into a fund with similar characteristics. Empirically, individuals keep a share of their newly found liquidity in cash, save it, consume it, or reinvest it into different funds, stocks, or bonds. Moreover, individuals reinvest 80% if the forced sale resulted in a capital gain, but only 40% in the event of a loss. Such differential treatment of gains and losses is inconsistent with active rebalancing or tax considerations, but consistent with mental accounting and the idea that individuals treat realized losses differently from paper losses providing evidence for realization utility and effects (Barberis and Xiong, 2012; Imas, 2016).