Two essays on market liquidity

This dissertation consists of two parts and focuses on the equity based market liquidity in Borsa Istanbul, as an order driven electronic market. In the first part, we empirically analyze the agent based relationship between liquidity flow and downside price formation based on the trading network topologies. We apply PageRank Algorithm to extract daily centrality degree in liquidity demand of domestic financial institutions classified as informed traders and use intraday maximum drawdown to capture intraday liquidity shocks. We find evidence that 1) Maximum cumulative loss for a given day, deepens with the increasing liquidity demand of informed traders. 2) The uncertainty in the centrality degree of informed trading on the sell side of the transaction book is overtime positively related with the uncertainty regarding the highest level of intraday loss. 3) Time Patterns are significant: Drawdown depth is highest on Thursdays and lowest on Mondays. Highest (lowest) drawdowns on May (March) indicate the existence of Sell-in-May effect (earnings announcement effect). In the second part, we focus on the market liquidity risk, which is often ignored in conventional Value at Risk (VaR) Metrics. We propose two alternative parametric methods to the existing literature on Liquidity Adjusted Value at Risk. The first model is based on the uncertainty in depth whereas the second model includes the tightness dimension to the latter. Our validation results indicate that both parametric approaches are strong alternatives to existing tightness based models and strictly superior to conventional VaR, with respect to performance related to regulatory compliance, statistical coverage, predictive quantile loss and overall relative cost of liquidity vs. loss.