Macroeconomic Patterns in System-Wide Liquidity Regimes

This research seeks statistical commonalities in market liquidity across a range of corporate equities and bond markets and commodity futures markets, with an aim to better understanding system-level patterns in aggregate or funding liquidity that arise from broad patterns in market liquidity. We present a Bayesian estimation of hidden Markov chain (HMC) models to measure the latent structure of liquidity. Starting with granular measures of market liquidity for equity, bond, and futures markets, we use Markov chain Monte Carlo (MCMC) analysis to estimate the latent structure governing liquidity at the system-wide level. Our input measures of market liquidity take advantage of recent work by Kyle and Obizhaeva [2014], who propose “invariant” price-impact measures of responses to order flow that are readily comparable across a broad range of financial markets and conditions. We find that three latent liquidity regimes—corresponding to high, medium, and low price-impact—are adequate to describe all of the markets we consider. In a more focused examination of the equities markets alone, we test the ability of several macroeconomic time series to recover the estimated liquidity dynamics. This exercise has significant explanatory power and allows for an economically meaningful attribution of the estimated latent liquidity