Asymmetric and nonlinear dynamics in sovereign credit risk markets

• Author: Geoffrey M. Ngene,Allen K. Lynch,Parker Benefield
• Date: 01 May 2018
• Published date: 01 May 2018
• DOI: http://doi.org/10.1002/fut.21896

Received: 2 June 2017

|

Accepted: 18 November 2017

DOI: 10.1002/fut.21896

RESEARCH ARTICLE

Asymmetric and nonlinear dynamics in sovereign credit

risk markets

Geoffrey M. Ngene

|

Parker Benefield

|

Allen K. Lynch

Stetson School of Business and Economics,

Mercer University, Macon, Georgia

Correspondence

Geoffrey M. Ngene, Stetson School of

Business and Economics, Mercer

University, 1501 Mercer University Drive,

Macon, GA 31207.

Email: ngene_gm@mercer.edu

We employ asymmetric and nonlinear error correction models to characterize the

price discovery and volatility interactions between the sovereign CDS and bond

spreads for 22 reference entities. We find asymmetric, nonlinear, and bidirectional

short and long-run information flow in the first and second moments. The flow from

the CDS to the bond market is stronger than in the reverse direction, demonstrating

that CDS market is the more effective vehicle for price discovery. The persistence of

volatility implies that informed trading occurs in the CDS markets. Both markets

seem to converge to an equilibrium relationship when the basis is large.

KEYWORDS

asymmetric, CDS and bonds, information flow, nonlinear, sovereign

JEL CLASSIFICATION

G14, G15

1

|

INTRODUCTION

Credit default swaps (CDS) have been popular derivative instruments of hedging credit risk since their inception in the 1990's.

Typically, the bondholder or lender buys insurance against default by the borrower (usually referred to as the reference entity or

the obligor) or on the occurrence of other specified credit event. The insurance premium is called the CDS spread, or simply

CDS, and is usually expressed as a percentage of the par value of the debt. When the bondholder holds a bond with T-years to

maturity and BY yield to maturity and buys protection or insurance equal to the CDS spread (CDS), the net BY should ideally be

equivalent to the risk free yield (RFY) of a Treasury bond with T-years to maturity (See footnote 6 for more information), since

the bondholder has eliminated underlying credit risk. Therefore, BY-CDS = RFY or BY-RFY = CDS. The BS-RFY is the bond

spread (BS). In the absence of market frictions and counterparty risk, the BS = CDS for any reference entity. Duffie (1999) and

Hull and White (2000, 2001) have formalized the parity relationship between CDS and BS.

In relation to sovereign reference entities, the parity between sovereign CDS and BS may not hold due to institutional factors

such as the cost of short selling in the bond market,

1

country specific and global factors, liquidity, speculation in the CDS market,

transaction costs, counterparty risk, and diverse beliefs held by investors regarding the probability of default by a sovereign

entity, among other factors (Chan-Lau & Kim, 2004; Foley-Fisher, 2010; Fontana & Scheicher, 2010; Küçük, 2010; Levy,

2009). Therefore, there exists non-zero basis as long as CDS and BS are misaligned.

1

Unlike bonds, investors can short credit risk using CDS since it does not require any funding cost. Investors buy (sell) CDS as the credit quality of the

reference entity deteriorates (ameliorates). In naked CDS market, buying (selling) the CDS is equivalent to going “short”(“long”) on the CDS and credit

risk since the investor does not own the bond.

J Futures Markets. 2018;38:563–585. wileyonlinelibrary.com/journal/fut © 2017 Wiley Periodicals, Inc.

|

563

The study of the time series behavior and features of the basis is important for three main reasons: First, the mere existence of

basis may induce investors to engage in profit-seeking arbitrage activities. Therefore, basis intensifies trading activities in both

the CDS and bond markets. The misalignment of CDS and BS will afford the arbitrageurs the economic incentive to engage in

basis trade premised on the relative pricing of CDS and cash bond. Specifically, whenever CDS > BS, a positive basis exists

where the bond is overvalued. Arbitrageurs should short sell the overvalued bond, write a protection in the CDS market, and go

long on the risk free bond. This would generate riskless profit. However, since it is problematic and costly to engage in short

trading in the bond market, traders are generally unable to take advantage of positive basis which requires concurrent shorting of

the bond and selling the CDS protection. This phenomenon, as Diamond and Verrecchia (1987) argue, makes the cash (bond)

market inefficient and lags the derivative market in the price discovery process. Conversely, a negative basis exists whenever

CDS < BS, suggesting that the bond is undervalued. Arbitrageurs could generate riskless profits by simultaneously buying the

risky bond, buying insurance in the CDS market, and taking a short position in the risk-free bond. Second, the size of the basis

may be a pointer to the degree of market liquidity. This would be particularly true during periods of turmoil in financial markets

(Akdoğan & Chadwick, 2013). Third, the speed of adjustment to a long-run equilibrium position or the co-integration

relationship between sovereign CDS and bond markets is a potential bellwether of risk premium implied by the worth of

underlying sovereign assets. Regulators and investors can use information from well-functioning CDS and bond markets to

effectively assess the credit health of a reference entity.

2

The CDS, as a market-based measure of credit risk, has been shown to be faster and more reliable in incorporating

macroeconomic news and anticipatory firm-specific credit-related information (such as credit rating downgrades) than stock

prices (Hull, Predescu, & White, 2004; Kou & Varotto, 2008; Norden & Weber, 2009, inter alia.) Also, CDS is seen as a

“cleaner”measure of credit risk (Longstaff, Pan, Pedersen, & Singleton, 2011) than BS since CDS markets have a higher number

of participants, lower funding requirements, flexibility in short selling, and bidirectional liquidity flow.

Studies on sovereign CDS and BS can be categorized into two broad categories: (i) Determinants of sovereign CDS focusing

on domestic versus global determinants

3

and (ii) Price discovery between sovereign BS and CDS. Our study generally falls in the

second category. Proponents of the CDS market argue that short selling in the CDS market improves liquidity and hedging

activities in the marketplace, improves information flow, increases the depth of the market, facilitates hedging and transfer of

risk, acts as an important vehicle of price discovery, and may actually ease volatility in the cash bond market. Moreover, active

trading by speculators guarantees the availability of sellers of CDS at competitive prices. These proponents justify their

arguments by providing empirical evidence using the lead–lag relationship between changes in CDS and changes in BS.

4

A

common feature of past studies on price discovery of sovereign CDS and bond markets is the focus on (i) modeling the

conditional mean (first moment) and (ii) linear pricing discovery. However, financial time series exhibit certain stylized features,

namely volatility clustering, leptokurtic distribution, and heteroskedasticity, inter alia, which may weaken the effectiveness of

modeling the conditional mean. Second, linear pricing discovery assumes that the sign and size of the basis is irrelevant in the

price discovery process.

Volatility is widely used as a measure of risk (Poon & Granger, 2003). Therefore, modeling the second moments, particularly

incorporating short term and long term influences, information flow between sovereign CDS and BS could generate additional

insights on the dynamics and volatility transmissions between the two markets.

In this study, we argue that the failure to incorporate the long run equilibrium relationship that exists between sovereign CDS

and BS and the nonlinear and asymmetric nature of underlying error correction (EC) process results in the development of a

myopic view of the influence of CDS trading on changes in bond spreads volatility. Further, challenges are likely to emerge on

the interpretation of the lead–lag results relationship. Unlike past studies, we also focus on second moment dynamics of CDS and

BS and adequately incorporate nonlinearities,

5

asymmetries and time-variation in volatility. This will enhance accounting for

structural shifts in volatility of CDS and bond prices returns.

Our study bears two principal distinguishing features. First, we employ a unified econometric approach that significantly

increasesthe statistical efficiency of our results throughconcurrent testing of the effects of asymmetryand nonlinearity of the basis

2

A significant portion of the CDS market is dominated by naked CDS market where buyers of CDS have no insurable interest on the bond (They are not

bondholders.) They simply insure against default on a bond they do not own. The basis of such naked CDS is speculation on the credit quality of the

reference entity. Naked CDS are popularly used by hedge funds. Trading of naked sovereign CDS was banned in EU on November 2012.

3

See for example: Abid and Naifar (2006), Küçük (2010), and Longstaff et al. (2011) to mention just but a few of the studies.

4

We use percent return in CDS and BS to denote changes in CDS and BS, respectively.

5

The only studies that have attempted to model nonlinearities in sovereign credit risk are Belke and Gokus (2014) and Coudert and Gex (2010). The two

studies focus on EU countries and first moment.

564

|

NGENE ET AL.