Celebrating Success: Prof Darren Duxbury
Congratulations to Professor Darren Duxbury on the recent publication of the paper: Riley, C., Summers, B., & Duxbury, D. (2020) Capital Gains Overhang with a Dynamic Reference Point. Management Science.
The summary:
Investors who hold a share have what is called a reference point – the price at which they are indifferent between selling it and holding it (reference points are an idea brought out in Prospect Theory which underlies a lot of work in behavioral finance and behavioral decision making more widely). Models of financial markets typically assume that reference points stay at the purchase price – the price at which the share was bought – but we demonstrate experimentally that the reference point changes over time as the share price changes. Using the results from our experiment we show that including reference point adjustment can improve the predictive power of the Capital Gains Overhang model in market data, an important determinant of the predictability of stock returns.
Our experimental results show the importance of highs and lows in the stock price path in determining reference points, in addition to the purchase and final prices. We then take the insights we learn from the controlled conditions of the experiment and apply these to the US stock market, using the Capital Gains Overhang model.
We show that the purchase price is not the only reference point that is predictive of one month ahead returns when plugged into the Capital Gains Overhang model. In fact, alternative Capital Gains Overhang variables based on the maximum, minimum, 52-week maximum or 52-week minimum are equally good predictors.
We create two Capital Gains Overhang Composite variables formed from weighting different salient points in the stock price path to calculate a reference point, using coefficients determined in the experiment and show that the traditional Capital Gains Overhang variable is no longer a positive predictor of returns when either of the Capital Gains Overhang Composite variables are included in the regression.
Our results also suggest that future volume, as well as future returns, is more responsive to Capital Gains Overhang variables based on a composite reference point than Capital Gains Overhang based on a purchase-price based reference point.
The implication of our results is that investors take multiple points into consideration when forming a reference point, so that reference points do adjust over time, and adjusted Capital Gains Overhang variables are therefore a better predictor of future returns and trading volume than the traditional Capital Gains Overhang variable.
Full article available here.