Decimalization, Realized Volatility, and Market Microstructure Noise

This paper studies empirically the effect of decimalization on volatility and market microstructure noise. We apply several non-parametric estimators in order to accurately measure volatility and market microstructure noise variance before and after the final stage of decimalization which, on the NYSE, took place in January, 2001. We find that decimalization decreased observed volatility by decreasing noise variance and, consequently, increased the significance of the true signal especially in the trade price data for the high-activity stocks. In general, however, most of the found increase in the signal-to-noise ratio is explainable by confounding and random effects. We also find that although allowing for dependent noise can matter pointwisely, it does not appear to be critical in our case where the estimates are averaged over time and across stocks. For that same reason rare random jumps are not critical either. It is more important to choose a proper data type and prefilter the data carefully.