Determinants of the component structure of intraday return distributions

Abstract

There is existing evidence of equity returns having a mixture distribution with multiple component structures. Following the increasing interest in intraday trading, this article examines determinants of intraday equity return distributions and finds that greater information flow and stock liquidity reduce the number of components while greater heterogeneity amongst traders increases the number of components. These results show that when empirically modelling intraday return distributions allowance needs to be made for their time varying nature and the fact that they are conditional on the liquidity and information flow of a stock. In addition, the results reinforce the recent emphasis by market regulators on information flow and liquidity being key to increasing transparency and reducing uncertainty (multiple components in the return distribution).

Notes

¹ Component structure is used to refer to the number of mixtures in a mixed normal distribution.

² Results of fortnightly estimates are reported in this article. The results of weekly data are qualitatively the same and available on request.

³ The price variable is included to control for the market microstructure effect of low stock price.

⁴ It has been proved that the BIC criterion is consistent and efficient in model selections (Schwarz, 1987).

⁵ This high frequency is an artefact of the limitation in the empirical estimation where only a maximum of nine distributions can be estimated by the programme.

⁶ We thank the referee for pointing out this difference.

⁷ Panel data analysis is also performed. The Hausman test suggests a random effect to be fitted. The result is qualitatively the same. All coefficients are with same sign and similar significant level except for the order imbalance variable which is not statistically significant in the random effect analysis.