Ebook Survival Analysis and Individual Trading Behavior
How long do investors typically hold stock positions? In an investor has held a stock for three months, is she equally likely to sell the stock next Monday as she is to sell the stock two Mondays from now? How can financial econometricians incorporate holding times into studies of investor decision making and investor behavior?
To answer the three questions posed above, we use survival time (hazard rate) analysis. Not only does this econometric technique allow us to account for the time dimension inherent in an individual’s stock trades, it allows us to carry out cross-sectional comparisons that are not possible with previously used techniques. Our analysis shows that the conditional probability of a sale is not constant over time. Thus, most studies of investors’ decisions to buy, hold, and then sell financial assets need to account for holding times.
To highlight the benefits of using survival analysis, we focus on a well documented behavioral bias—the propensity of individuals to realize gains and avoid losses—a behavior known as the “disposition effect.” We ask specific questions: How pervasive is the disposition effect? Does it affect all investors to the same degree? If, as Odean (1998) argues, the bias is not value-maximizing, do investors correct this bias over time?
We show large cross-sectional differences in investors’ propensities to realize gains and avoid losses. Sophisticated investors are less apt to display this bias. Certain demographic groups do not exhibit the disposition effect. We also show that investors become less prone to the disposition effect over time. In other words, investors appear to be learning (or self-correcting) over time. The ability to self-correct behavior is valuable to individual investors. Odean (1998) shows that when individual investors in the United States sell a stock, the stock price tends to go up. When they hold a stock, the stock price tends to go down. Chen, Kim, Nofsinger, and Rui (2004) confirm the same timing pattern between individual decisions and stock returns for Chinese investors.
The finding that investors become less prone to the disposition effect over time has important implications for recent work regarding the interaction of behavioral biases and asset prices. The results also pose a serious challenge for financial economists who wish to model asset prices in a world with behavioral biases. For example, Barberis, Huang, and Santos (2001) model an economy that has a representative agent with a utility function that fits empirical findings at an individual level. We show a high degree of heterogeneity in the degreeof the disposition effect. Investors reduce this bias over time and appear to be learning. Our results highlight two problems. It is not clear how to aggregate the utility functions of diverse individuals whose behaviors are changing over time. More importantly, if the marginal investor is a professional trader at Goldman Sachs, we may not want to model an economy with a representative agent who exhibits a bias associated with unsophisticated individuals.
Download
PDF Ebook Survival Analysis and Individual Trading Behavior
Posted in :
