Testing the Implications of Overconfidence for Intraday Trading
40 Pages Posted: 7 Mar 2005
Date Written: January 28, 2005
We test the implications of overconfidence behavior using U.S. intraday trading data. We propose several testable hypotheses for return autocorrelations, trading volume, return volatility, and for the causal interrelations between volume and volatility. As predicted by overconfidence behavior, return autocorrelations are positive for short lags and then gradually decline as lags lengthen. Also consistent with the prediction of overconfidence together with biased self-attribution, return volatility is higher during periods containing public news signals compared with volatility during periods without public news signals. To differentiate between the overconfidence hypothesis and the sequential information arrival hypothesis, we test the lead-lag links between trading volume and return volatility during periods without public news. After necessary Bayesian adjustments to avoid large sample biases, we find evidence that volume Granger-causes volatility but without feedback during the periods without public news. The results lend support to the overconfidence hypothesis as opposed to the sequential information arrival hypothesis and suggest that investors trade according to their private signals but are reluctant to close their positions afterwards.
Keywords: Trading volume, return volatility, behavioral finance, overconfidence
JEL Classification: G12, G14
Suggested Citation: Suggested Citation