On the Computation of LOT Liquidity Measure

We clarify that the widely used estimation method for the LOT liquidity model in the market microstructure literature is improper in the sense of econometric inference. Based on an extensive simulation study and a real data analysis, we show that this method not only overestimates the true spread parameter severely, but also the bias and RMSE do not converge as a consistent estimator should do. We suggest one should use the Y-split estimation method proposed by Goyenko et al. (2009) instead when applying the LOT model to measure transaction cost