Ronald Wafula

Abstract

Ronalds thesis is a collection of three essays that empirically examine liquidity resiliency from a market microstructure perspective in the Eurozone sovereign bond markets. Firstly, I offer new insights into liquidity resiliency by constructing Ordinary Least Squares (OLS) and Least Absolute Shrinkage and Selection Operator (LASSO)-based resiliency measures applying the mean-reverting model of Kempf et al. (2015), and this is presented in Chapter 2 (first essay). Second, in Chapter 3 (second essay), I examine the commonality of liquidity resiliency at the GIIPS, non-GIIPS, and pan-European levels. In the final chapter (Chapter 4), I examine whether liquidity resiliency as a separate dimension of liquidity is a priced variable by applying the liquidity-capital asset pricing model advanced by Acharya and Pedersen (2005).

In the first essay (Chapter 2), I present novel insights into liquidity resiliency, a relatively neglected dimension of liquidity. I empirically analyse the liquidity resiliency of the Eurozone sovereign bond markets by applying 2-, 5-, 10-, and 30-year bond maturities. Three periods were applied in this study: the pre-crisis period (January 2008-October 2009), the Eurozone crisis period (November 2009-December 2013), and the full period (January 2008-December 2013). I employed the Kempf et al. (2015) OLS-based approach and a combination of an OLS approach and the LASSO machine learning approach to compute resiliency from relative spreads (RS) and quoted depths (QD). The findings reveal that RS and QD resiliency are negatively correlated with relative spreads and positively correlated with quoted depth. The correlations between RS resiliency, QD resiliency, relative spreads, and quoted depths are generally low. This indicates that resiliency is a separate liquidity dimension and provides information that cannot be subsumed from the relative spreads or quoted depth dimensions. I provide evidence that OLS- and LASSO-based resiliency measures should be used interchangeably. Moreover, I study the interrelationships between resiliency and volatility, returns, and credit risk and find that intertemporal relationships exist.

Chapter 3 examines the commonality in liquidity resiliency of the ten eurozone sovereign bond markets across the maturity spectrum. I applied the principal component analysis (PCA) to compute the market-wide (common) resiliency factor. Using OLS-based regressions, I then tested the sensitivities of individual country bond resiliency to the market-wide resiliency factor. This is analysed at the peripheral (GIIPS) countries, core (non-GIIPS) countries, and pan-European levels. I examine the strength of the relationship between GIIPS and non-GIIPS countries' resiliency using the canonical correlation analysis (CCA) approach. Finally, supply-based variables (funding liquidity constraints), demand-based variables (investor sentiment, economic policy uncertainty, and exchange rate fluctuations), and market variables are used to explain the drivers of commonality in resiliency in the Eurozone sovereign bond markets. I find evidence of commonality in the resiliency of GIIPS and non-GIIPS countries in the pre-crisis and crisis periods; however, this is more apparent and pervasive in the GIIPS region during the Eurozone crisis period. I find a significant Eurozone liquidity resiliency effect, though economically small but statistically significant, as the countries react differently to the Eurozone resiliency factor, which is congruent with the fragmented nature of the Eurozone countries. I find that funding liquidity constraints, as supply-side drivers of commonality in RS and QD resiliency, impact both GIIPS and non-GIIPS countries.

In Chapter 4, I examine whether liquidity resiliency risk is a priced variable in Eurozone sovereign bond returns beyond the level of resiliency, credit risk, and market risk. I apply a variant of the Liquidity-Adjusted Capital Asset Pricing Model developed by Acharya and Pedersen (2005) and further enhanced by O'Sullivan and Papavassiliou (2020). I apply Fama and Macbeth's (1973) two-step regression to estimate the coefficients for return premiums attributable to credit risk, market risk, and the three resiliency risk factors. The first step computes the four risk factors, and the second step runs univariate and multivariate OLS regressions, enabling the investigation of the relationship between returns and credit risk, market risk, and liquidity resiliency risk. I find that the level of liquidity resiliency, credit risk, market risk, and some subsets of liquidity resiliency risk are priced in Eurozone sovereign bond returns; however, the effects depend on the Eurozone region (GIIPS or non-GIIPS), period considered, and the liquidity resiliency measure applied.