On the correlation analysis of stocks with zero returns

Abstract

The purpose of this article is to study serial correlations, allowing for unconditional heteroscedasticity and time-varying probabilities of zero financial returns. Depending on the set-up, we investigate how the standard autocorrelations can be accommodated to deliver an accurate representation of the serial correlations of stock price changes. We shed light on the properties of the different serial correlations measures by means of Monte Carlo experiments. Theoretical results are also illustrated on shares from the Chilean stock market and Facebook stock intraday data.