Editor’s note: We will take the other site of the bet – your move OFR…
Systemwide Commonalities In Market Liquidity by OFR
Mark D. Flood
Office of Financial Research
John C. Liechty
Office of Financial Research and Pennsylvania State University
Thomas Piontek
Office of Financial Research
Abstract
We explore statistical commonalities among granular measures of market liquidity with the goal of illuminating systemwide patterns in aggregate liquidity. We calculate daily invariant price impacts described by Kyle and Obizhaeva [2014] to assemble a granular panel of liquidity measures for equity, corporate bond, and futures markets. We estimate Bayesian models of hidden Markov chains and use Markov chain Monte Carlo analysis to measure the latent structure governing liquidity at the systemwide level. Three latent liquidity regimes — high, medium, and low price-impact — are adequate to describe each of the markets. Focusing on the equities subpanel, we test whether a collection of systemwide market summaries can recover the estimated liquidity dynamics. This allows an economically meaningful attribution of the latent liquidity states and yields meaningful predictions of liquidity disruptions as far as 15 trading days in advance of the 2008 financial crisis.
Systemwide Commonalities In Market Liquidity – Introduction
Liquidity is a central consideration for market quality in general, and for financial stability in particular. We present a new approach to the study of liquidity that identifies broad-based patterns in daily data for individual markets to help explain aggregate, systemwide liquidity conditions. Although funding liquidity in the wholesale markets for institutions is the most immediate concern for systemwide conditions Brunnermeier and Pedersen 2009) there are many more individual markets for financial assets than for intermediaries’ liabilities. It is an empirical question whether there is additional information in the much larger panel of asset markets that might help to explain liquidity conditions in the funding markets. We expand on previous studies of commonalities in liquidity e.g., Chordia et al. [2000] and Karolyi et al. [2012], who find that there are indeed significant patterns in the detailed data by analyzing liquidity in a range of asset classes, including equities, bonds, and financial futures. We also extend the commonalities approach with a novel methodology for connecting aggregate liquidity patterns to a panel of systemwide market summaries.
Starting with granular measures of market liquidity, based on the recent invariant price-impact measures of Kyle and Obizhaeva [2014], we estimate a daily panel of liquidity conditions across a broad range of markets. Specifically, our initial implementation considers volatility index futures, oil futures, and sector portfolios for the Center for Research in Securities Prices (CRSP) universe of U.S. equities and the Transaction Reporting and Compliance Engine (TRACE) universe of corporate bonds over the decade 2004-14. The
market-invariant approach of Kyle and Obizhaeva [2014], carefully normalizes for local volume and volatility conditions to produce liquidity measures that are directly comparable across markets and order-flow conditions. Comparability is crucial for aggregating local liquidity conditions to support systemwide analysis. Using this panel of daily liquidity measurements, we estimate Bayesian hidden Markov chains (HMC) models, using Markov chain Monte Carlo (MCMC) inference methods, to capture the latent structure of each series, and assess the latent structure governing liquidity at the systemwide level. The HMC approach posits that the dynamics of each daily price impact measure (33 in our sample) are determined by an underlying variable that alternates among several liquidity states to drive sudden changes in the observed levels of price impact. The underlying states are latent i.e., not directly observable — and must be inferred from the dynamics of daily price-impact measurements. In the initial analysis, we estimate each price impact series independently; that is, we assume no coordination between the dynamics of the latent liquidity states across markets. Nonetheless, we find surprising consistency in the dynamics of market liquidity across all of these markets. From the perspective of a policy maker who seeks to identify, or even predict, turbulent episodes in the financial system, we find that three liquidity regimes are adequate to describe each market: high, intermediate, and low. Moreover, we find that the low liquidity regime afflicts all markets roughly simultaneously during the financial crisis of 2008.
See full PDF below.
