Abstract

The purpose of this thesis is to apply liquidity proxies from the microstructure literature based on daily data to estimate the liquidity cost within a monthly parametric Value-at-Risk (VaR) model to create a parsimonious yet relevant measure of both market and liquidity risk. The proxies examined in this thesis include the ‘Roll Proxy’ of Roll (1984), Zeros by Lesmond, Ogden and Trzcinka (1999), the Holden, Fong and Trzcinka (2011, FHT henceforth) proxy and the Shane and Schultz (2012) High-Low Spread proxy (HLS henceforth) as they are computationally less onerous, do not necessitate intraday data and were proven to be reliable in the literature. We find that the FHT, Roll and HLS proxies are indeed good estimators of effective costs making them appropriate additions to a VaR model, while the Zero proxy performs very poorly due to its overestimation of effective costs. With high correlations and similar distributional properties relative to the Hasbrouck effective costs, the FHT proxy performs best when back-tested against liquidity-adjusted returns relative to all other proxies.