On estimating market microstructure noise variance

•Propose modified estimates of market microstructure noise (MMN) variance based on Hansen and Lunde (2006).•Report Monte Carlo results of comparison of different estimates of MMN variance.•Report better performance of modified estimates at different frequencies.•Report empirical estimates of MMN variance for some liquid NYSE stocks in 2010-2013.

We study the market microstructure noise-variance estimation of high-frequency stock prices. Based on the Hansen and Lunde (2006) approach, we propose estimates using subsampling method at different time scales. We conduct a Monte Carlo study to compare our method against others in the literature. Our results show that our proposed estimates have lower (absolute) mean error and root mean-squared error, and their performance is quite stable at different time scales.