Photo: John Moss, EAM Investors
Individual investors make mistakes, especially new ones. But do they learn over time?That’s the question in a new behavioural finance paper from Maximilian Koestner, Steffen Meyer, and Andreas Hackethal.
Using a massive data set containing the trading history of 19,487 German investors over seven years, they examined if and how investors improve.
Here’s what their results told them:
Investors do not learn from underdiversification
A possible culprit? Stock picking. Citing Hirschleifer (2010), the authors explain that social interactions can drive underdiversifaction. Investors like to tell peers about successful trades. Picking a few volatile stocks is more likely to lead to a notably successful investment than replicating the market. That leads to active investment behaviour and underdiversification.
Investors learn from overconfidence
The proxy for overconfidence in the model was trading frequency. When investors first start, they take too much credit for successful investments, and tend to be overconfident and over trade. As time passes, they gain a more realistic view of their skills and the transaction costs associated with over trading. There was a significant reduction of trading activity as investors gained more experience.
Experience does not decrease the disposition effect
The disposition effect is the tendency, described in behavioural finance, for investors to sell shares that rise in value and hold shares that fall. Basically, they are (irrationally) more willing to recognise gains than losses. The costs of over trading are easy to detect with every statement of trading fees. The disposition effect is a quieter and more difficult mistake to catch.
The study found that, overall, investors do improve their returns with experience. That finding is more robust for overall portfolio returns over time (calculated with the Modified Dietz method) than risk adjusted returns (calculated via the Sharpe Ratio). The authors ascribe the improvement in returns to investor learning from overconfidence. One of their specifications found that 100 additional active trades are associated with a portfolio performance increase of 0.15 per cent per month. Their results were significant at both the one and five per cent confidence interval.
Read the full paper here.