THE COMOVEMENTS OF STOCK, BOND, AND CDS ILLIQUIDITY BEFORE, DURING, AND AFTER THE GLOBAL FINANCIAL CRISIS

• Date: 01 December 2020
• Author: Xinjie Wang,Yangru Wu,Zhaodong (Ken) Zhong
• Published date: 01 December 2020
• DOI: http://doi.org/10.1111/jfir.12230

The Journal of Financial Research Vol. XLIII, No. 4 Pages 965–998 Winter 2020

DOI: 10.1111/jfir.12230

THE COMOVEMENTS OF STOCK, BOND, AND CDS ILLIQUIDITY BEFORE,

DURING, AND AFTER THE GLOBAL FINANCIAL CRISIS

Xinjie Wang

Southern University of Science and Technology

Yangru Wu and Zhaodong (Ken) Zhong

Rutgers University

Abstract

Using both marketwide and firm‐level illiquidity measures of the stock, bond, and

credit default swap markets, we find that comovements of illiquidity across markets

increase significantly during the recent global financial crisis. Moreover, the degree

of comovement remains significantly higher in the postcrisis period and regulatory

period than in the precrisis period. Specifically, the distribution of firm‐level

comovements is notably different before and after the crisis (e.g., a much larger

portion of firms with positive pairwise correlations between illiquidity measures in

the postcrisis period than in the precrisis period). Our results provide suggestive

evidence of the financial crisis effects and the subsequent postcrisis regulations on

the comovements of illiquidity across markets.

JEL Classification: G14, G21, G24, G28

I. Introduction

The degree of comovement in (il)liquidity across markets has important implications for

asset valuation, portfolio selection, and risk management. Even well‐diversified

institutions are affected by the illiquidity comovements among different asset classes in

their portfolios. Therefore, understanding the drivers of time‐series variations in illiquidity

comovements is an important topic (e.g., Chordia, Roll, and Subrahmanyam 2000;

Hasbrouck and Seppi 2001; Coughenour and Saad 2004; Chordia, Sarkar, and

Subrahmanyam 2005). Moreover, disruption to common factors for liquidity across

markets is considered by many to be one of the major reasons for the recent financial

crisis and could possibly lead to future financial crises. How do illiquidity across markets

comove with each other during the global financial crisis? How do comovements of

illiquidity across markets change after the crisis? In this article, we answer these research

We thank Jie Cao, Jia Chen, Ren‐Raw Chen, Jia He, Shufang Lai, Peipei Li, Andy Liu, Bianxia, Sun,

Dragon Tang, Ju Xiang, Hongjun Yan, Shuo Yan, Ti Zhou, and conference and seminar participants at the 29th

Australasian Finance and Banking Conference, the Fourth PKU‐NUS Annual International Conference on

Quantitative Finance and Economics, and Southern University of Science and Technology for their helpful

comments. Special thanks go to an anonymous referee and Erik Devos (the editor) for helpful comments and

suggestions that significantly improved the paper. All remaining errors are our own. Xinjie Wang acknowledges

financial support from Southern University of Science and Technology (Grant No. Y01246210, Y01246110).

965

© 2020 The Southern Finance Association and the Southwestern Finance Association

questions by analyzing the comovements of illiquidity measures of the stock, bond, and

credit default swap (CDS) markets before, during, and after the global financial crisis.

Stocks, bonds, and CDSs are all financial claims on the same firm. Changes in

the illiquidity of the stock, bond, and CDS markets are expected to be closely related to

each other even before the crisis. One channel for these comovements is through the

shared capital and information among common market makers. Market makers,

typically dealer banks, often provide liquidity in multiple markets. For example,

corporate bonds and CDSs are traded in the over‐the‐counter (OTC) markets in which

dealer banks are the main market makers (see, e.g., Chen et al. 2011; Eisfeldt

et al. 2018). Dealer banks also are main market makers in stock markets and provide

prime brokerage services to their institutional clients (e.g., mutual funds, pension

funds, and hedge funds).

1

Following a similar line of reasoning as in Coughenour and

Saad (2004), we expect that illiquidity in the stock, bond, and CDS markets is

correlated with one another.

2

For instance, market makers that need to absorb a

negative liquidity shock in the bond market may do so by increasing their holding of

bonds, which crowds out capital for holding stocks and CDSs and results in a lower

liquidity in the stock and CDS markets.

We expect that a sudden drying up of capital for market makers in (such as

what happened during the financial crisis) will lead to greater comovements of

illiquidity among different markets. Moreover, comovements are likely to remain at a

high level after the crisis because market makers are still recovering from the huge

capital loss incurred during the crisis and are under stringent scrutiny to preserve

adequate capital. The postcrisis regulations may also have some potential effects on

the comovements of illiquidity among different markets. For example, Basel III

significantly increases capital and liquidity requirements for banks.

3

Therefore, bank‐

affiliated market makers that are subject to these postcrisis capital and liquidity

requirements face a significant increase in the cost of maintaining inventories of risky

assets. The effect of these regulatory changes is to reduce the inventory capacity of

market makers. A subsequent result (which has not been noticed in the literature) is the

high comovements in illiquidity across different financial markets.

To study comovements in illiquidity among the stock, bond, and CDS markets,

we first examine the marketwide illiquidity measures. We construct marketwide

illiquidity measures using a portfolio of firms that are constituents of two popular

credit indices: CDX.NA.IG and CDX.NA.HY. This allows us to construct marketwide

1

For example, the 2013 annual report of Goldman Sachs shows that a significant portion of revenue is from

market making in equity products, and commissions and fees from executing and clearing institutional client

transactions on stocks. Retrieved from https://www.goldmansachs.com/investor-relations/financials/current/

annual-reports/2013-annual-report.html.

2

Coughenour and Saad (2004) find that stock liquidity comoves with the liquidity of other stocks handled

by the same New York Stock Exchange (NYSE) specialist firm and that liquidity covariation increases with the

risk of providing liquidity.

3

Basel III requires banks to maintain 4.5% of the common equity (up from 2% in Basel II) of risk‐weighted

assets (RWAs) at all times. Minimum Tier 1 capital increases from 4% in Basel II to 6% over RWAs. In

addition, the new “Liquidity Coverage Ratio”standards require a bank to hold sufficient high‐quality liquid

assets to cover its total net cash outflows over 30 days.

966 The Journal of Financial Research

illiquidity measures using firms with the most liquid bonds and CDS contracts. These

firms represent a significant portion of trading activities in the bond and CDS markets,

accounting for more than 40% of total trading volume in both the CDS market and the

corporate bond market.

4

Marketwide illiquidity is widely perceived as a state variable

that determines asset prices (e.g., Pastor and Stambaugh 2003). It is critical to construct

accurate and sensible measures for marketwide illiquidity. The marketwide illiquidity

measures we construct are based on firm‐level bid–ask spreads or effective bid–ask

spreads (i.e., the Roll, 1984, measure). These measures are representative and capture

common variations in illiquidity among firms.

To understand how comovements in illiquidity change over time, we divide

our sample period into four periods: precrisis, crisis, postcrisis, and regulatory. We find

a number several interesting empirical results using the marketwide illiquidity

measures. First, as expected, the marketwide illiquidity levels in all three markets

increase dramatically with the onset of the global financial crisis. Second, and more

important, the marketwide illiquidity measures of the stock, bond, and CDS markets

exhibit a high degree of comovement during the global financial crisis.

5

Finally, the

degree of comovement in the postcrisis and regulatory periods remains considerably

higher than in the precrisis period.

To test whether the increases in illiquidity comovement are statistically

significant after the global crisis, we conduct regression analyses of the comovements

among the marketwide illiquidity measures. We regress the marketwide illiquidity

measures on other marketwide illiquidity measures and define three time dummy

variables: during the crisis, after the crisis but before the postcrisis regulations took

effect, and after the postcrisis regulations took effect. The interaction terms between

the time dummy variables and the illiquidity measures capture the increases in

illiquidity comovements (compared to the precrisis period). The regression results

show that all the interaction terms are positive and statistically significant, indicating

that the increases in illiquidity comovements are statistically significant in the crisis,

postcrisis, and regulatory periods.

After examining comovements among the marketwide illiquidity measures, we

turn to firm‐level data, which allow us to conduct more in‐depth analyses to provide

new insights into how illiquidity comovements have changed over time. First, we

examine whether and how the distributions of firm‐level pairwise correlations between

the three markets have changed over time. The histogram of firm‐level pairwise

correlations shows that a majority of the pairwise correlations between stock and CDS

in the precrisis period are concentrated around 0. However, after the crisis, there is a

noticeable increase in the right tail of the distribution (e.g., there are more firms with a

4

For CDS, trading volume is based on notional in DTCC, a posttrade financial services company providing

clearing and settlement services. It provides comprehensive reports on CDS trading activities. For bond, the

trading volume is based on the number of trades in Trade Reporting and Compliance Engine (TRACE).

5

In addition to pairwise correlations, we investigate commonality in illiquidity measures using principal

component analysis (PCA) and examine the percentages of variance explained by the first three principal

components of the marketwide illiquidity measures. A high percentage in the first principal component indicates

a high level of commonality. Overall, the PCA results are similar to pairwise correlation results.

967Stock, Bond, and CDS Illiquidity