• No results found

Drivers of Sovereign Credit Default Swap (CDS) spreads: The cases of Western and Southern European regions


Academic year: 2023

Share "Drivers of Sovereign Credit Default Swap (CDS) spreads: The cases of Western and Southern European regions"


Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst


University of Amsterdam Amsterdam Business School

BSc Economics and Business Economics – Finance specialization Bachelor Thesis

Drivers of Sovereign Credit Default Swap (CDS) spreads:

The cases of Western and Southern European regions

Investigation on the differentiation between determinants of credit default swaps in the Western and Southern European regions

Author: Ahmed Mohamed Abdelalim Nawara Student number: 12221708

Thesis supervisor: Dr. Tandogan Polat Finish date: June 2021


Statement of Originality

This document is written by student Ahmed Mohamed Abdelalim Nawara, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The faculty of Economics and Business is responsible solely for the supervision of completion of the work, not the for the contents.



The sovereign credit default swaps (CDS) market logically is affected by its economy’s macroeconomic figures, and considering the interconnectedness of the world’s financial markets, several global uncertainty covariates are bound to impact the derivatives market also. Credit default swaps were innovated to offset credit risks, such as payment default, to third party investors, institutions, or sovereign nations. This insurance contract comes at a price: the CDS premium or spread. This study fills the gaps in the existing literature on determinants of CDS spreads by conducting a panel data analysis on seven European countries (France, Belgium, Netherlands, Germany, Italy, Portugal, and Spain) – where the former four represent the Western European region, and the latter three represent the Southern region. The period of study was March 2009 till December 2019. The main empirical findings deduced that macroeconomic factors play a larger role in explaining CDS spreads relative to global factors, financial crisis periods intensify the effect of factors on CDS spreads but does not add explanatory power, and Southern CDS spreads were more sensitive to both global and macroeconomic fluctuations. Hence, global and macroeconomic covariates play a substantial role in determining CDS spreads and should be considered in further research.

Keywords: Credit default swaps, credit risk, European CDS market, macroeconomic factors, panel data analysis, global factors


Table of Contents

Abstract: ... 3

Table of Contents ... 4

List of Tables & Figures ... 5

1. Introduction ... 6

2. Theoretical Framework ... 7

2.1. Credit Events ... 8

2.2. Mechanisms of Credit Default Swaps (CDS) ... 8

2.3. European sovereign debt crisis: ... 10

3. Literature review ... 12

3.1. Global financial/economic factors ... 12

3.2. Macroeconomic determinants ... 16

4. Empirical methodology ... 19

4.1. Research data ... 19

4.2. Econometric models ... 20

4.3. Research design & statistical tests ... 22

5. Data ... 25

5.1. Sample ... 25

5.2. Credit default swaps ... 26

5.3. Global financial and economic variables ... 29

5.4. Country-specific macroeconomic factors ... 30

5.5. Preliminary analysis (descriptive statistics) ... 32

6. Estimation Results ... 35

7. Conclusion & Discussion ... 40

References ... 43

Appendix A: Regression Results table ... 45


List of Tables & Figures

Figure 1: CDS contract structure………...9

Table 1: STATA output: Breusch-Pagan Heteroskedasticity test………23

Table 2: STATA output: Breusch-Pagan LaGrange Multiplier (LM) test………...24

Table 3: Unit root tests (LLC)………..………24

Figure 2a: CDS time pattern for Western group………...27

Figure 2b: CDS time pattern for Southern group………..27

Figure 2c: CDS time pattern for both groups………...………….28

Figure 3: Time trend for global variables…..………...…30

Table 4: Sovereign Credit ratings……….32

Table 5a: Summary statistics: Western Europe………..………..32

Table 5b: Summary statistics: Southern Europe………..……….33

Table 6: Regressions outputs………45


1. Introduction

Having endured a diverse series of radical developments over the past two decades, the corporate and sovereign credit default swaps market has been a central focal point in the financial and economic scope. Credit default swaps (CDS hereafter) are financial derivatives that allow risk- averse investors to offset their exposure with a third party by entering into a debt insurance contract.1 CDS entails a range of various implications, such as providing an instrument for risk management, an index for credit exposure, and a signal for market volatility are among the multitude of reasons why politicians, investors, and financial analysts place large emphasis on the importance of this market.2

Following its inception in 1994 by J.P. Morgan, the CDS market witnessed immense surges in traded volumes and industry growth.3 Market value peaked in December 2007, accumulating a total outstanding notional amount (definition: nominal value insured by CDS) of $61.2 trillion.4 Following the onset of the 2008 global financial crisis and the subsequent Eurozone sovereign debt crisis, whereby credit default swaps played a substantial role, represents two significant milestones within the market’s history.5 Thereafter, in response to the crisis, there was a collective call for improving transparency and bringing the trade under control, which subjected the CDS market to a dynamic paradigm shift in terms of its structural framework, size, and regulation. 6 Progressive decline in outstanding notional amounts of CDS contracts followed, and twelve years later, as of Q2 2019, the CDS market accumulated market activity worth $5.8 trillion.7 Therefore, the notion of “credit risk” has been under pivotal scrutiny, which has provoked interest to assess the determinants that drive CDS spread levels.

1 Nader Naifar, “What explains the sovereign credit default swap spreads changes in the GCC region”, Journal of Risk and Financial management, 2020

2 Doshi, Jacobs, and Zurita, “Economic and Financial Determinants of Credit Risk Premiums in the Sovereign CDS market”, Review of Asset Pricing Studies (2015), 43-80

3 Aldasoro, and Ehlers, “The credit default swap market: what a difference a decade makes”, BIS Quarterly Review, 2018

4 Anton & Nucu, “Sovereign Credit Default Swaps and Stock Markets in Central and Eastern European countries: Are Feedback Effects at Work?”, Entropy, 2020

5 Aldasoro & Ehlers, 2018

6 Aldasoro & Ehlers, 2018

7 “Global Credit Default Swap Market Study”, International Swaps and Derivatives Association, 2019


Following the amplified availability of market data on CDS, numerous literatures have been published with regards to the determinants of CDS, however, the majority place focus on corporate CDS, with few focusing on sovereign CDS. This paper aims to contribute by addressing sovereign CDS by evaluating the global financial variables, country-specific macroeconomic indicators, and crisis periods, to discover which show potential explanatory powers in driving CDS spreads.

Through employing a panel data analysis framework, this study investigates seven countries, within two European regions, Western and Southern Europe, using a wide range of CDS, global, and macroeconomic data from 2009 to 2019. This leads to the research question: To what extent do the determinants of sovereign CDS spreads differentiate between the Western and Southern regions of Europe?

Our empirical estimation results delineated that: (i) country-specific macroeconomic variables are extensively more sufficient and effective predictors of CDS spread levels relative to global financial and uncertainty factors, (ii) the Southern European CDS spreads entail larger elasticity rates (responsiveness) to both exogenous global and country-specific fluctuations and shocks, and (iii) periods of financial crises effectuate volatilities in CDS spreads across both groups, but at a trivial level, hence does not add significant explanatory power.

The remainder of this paper is structured as follows: section two provides an informative theoretical background on credit default swaps, section three inspects previous literature and derives the hypotheses, section four presents the empirical methodology which outlines the data, econometric models, and research design used, section five will deduce an empirical layout of the data with explanations on descriptive statistics, section six will present the main estimation results of the panel data analysis and discuss the findings, and test the hypothesis, , finally, section 7 will deduce the conclusions and provide in-depth evaluation of the strengths and limitations of the study.

2. Theoretical Framework

This section aims to provide a robust foundation of the fundamentals of credit default swaps (CDS) in order to contribute insightful background knowledge of this financial instrument.


2.1. Credit Events

Firstly, in order to illustrate the mechanisms and features of a CDS, the origins of how a CDS contract came to fruition, by means of investors and institutions aiming to offset their share of the risk of a credit event from their debtor, will be demonstrated.

A credit event refers to any sudden adverse development in a borrower’s financial capacity to fulfill its debt obligation.8 They generally result from an array of factors such as high interest rates, lack of proper monitoring, diversion of funds, illiquidity and current account issues, and can occur on an individual, firm, and country level basis.9 Defaulting, which refers to when debtors fail to pay their obligations, whether it is interest or principal payments, on their loan or financial security, along with bankruptcy, debt restructuring, and repudiation/moratorium are the most noteworthy examples of a “credit event”.10 A prominent consequence of a higher credit event probability is a lower credit score (rating), which prompts creditors to view an account (individual, firm, or nation) as a “credit risk”. Essentially, this reduces the debtor’s chance of securing a loan or security or burdens the creditor with a higher risk of reclaiming their money, which leads to higher interest rates, and stringent conditions within the deal between creditors and debtors.11 As a result, credit default swaps (CDS) were innovated in 1994 by J.P. Morgan to transfer its credit exposure from its balance sheets to “protection sellers”.12

2.2. Mechanisms of Credit Default Swaps (CDS)

Credit default swaps (CDS) are financial contracts that permit an investor or financial institution to “swap” their credit risk with a third-party investor or institution.13 They

8 Houweling & Vorst, “Pricing default swaps: Empirical Evidence”, Journal of International Money and Finance, 2005

9 Amadei, Di Rocco, Gentile, Grasso, & Siciliano, “Credit Default Swaps: Contract characteristics and interrelations with the bond market”, Commissione Nazionale per le Società e la Borsa, 2011

10 Vogel, Bannier, & Heidorn, “Functions and characteristics of corporate and sovereign CDS”, Frankfurt Working School Paper Series, Frankfurt School of Finance and Management, 2013

11 Amadei et al., “Credit Default Swaps: Contract characteristics and interrelations with the bond market”, p. 6-11

12 Augustin, Subrahmanyam, Tang, & Wang, “Credit Default Swaps: Past, Present, and Future”, Annual Review of Financial Economics, 2016

13 Amadei et al., “Credit Default Swaps: Contract characteristics and interrelations with the bond market”, p. 6-11


systematically function as insurance contracts, as it permits the CDS buyer to mitigate the default risks of the reference entity with financial coverage from the CDS seller against a credit event (Figure 1). Likewise, as a compensation for accepting to bear the credit exposure of the reference entity and reimburse the CDS buyer in the case of a credit event, the CDS seller will receive a periodic “premium” (“fee” “spread” are also synonymous terms) for ensuring the creditworthiness of the underlying loan or security.14 The CDS premium is computed as an annual percentage or basis points of the notional value of the underlying asset, and can be paid periodically (monthly, quarterly, semi-annual, annual, etc.) depending on the preferences of the contract players.15 Figure 1 below summarizes the structure of CDS contracts, as explained above, in a visual manner.

Figure 1: CDS mechanism & structure Case: Structure if NO credit event takes place

interest (or principal) payments for

a bond or loan. Periodic premiums

Case: Structure if credit event takes place

Payments stop due to Compensation: notional amount credit event

Source: originally created

CDS contracts entail various applications and mechanisms, the most pivotal of which is hedging, whereby financial institutions hedge their risks by shorting the credit exposure of their reference entities. As aforementioned, this is utilized to shift the risk of a credit event from the CDS buyer to the CDS seller.16 Moreover, a secondary use is speculation, whereby individuals or entities that have an appetite for profit undergo risky bets regarding the creditworthiness of reference entities. For instance, if a hypothetical investor positively believes in the prospect of the credit quality of a reference entity, it can sell a CDS and collect periodic premiums. On the contrary, if the investor’s view of the creditworthiness of the reference entity is inadequate or poor,

14 Naifar, “What explains sovereign CDS spread changes in the GCC region”

15 Vogel, Bannier, & Heidorn, 2013

16 Amadei et al., “Credit Default Swaps: Contract characteristics and interrelations with the bond market”

Reference Entity


Buyer CDS



entity CDS

Buyer CDS



it can buy a CDS for a proportionally small fee and receive a larger payoff when the reference entity defaults to profit from the transaction.17

Furthermore, CDS contracts are traded over-the-counter (OTC) vis-à-vis a “broker-dealer network” instead of being verified by an official centralized exchange due to the dynamically complex nature of their usage (hedging, speculation, and trading; as explained above).1819 However, the International Swaps and Derivatives Association (ISDA) was established in order to stabilize the CDS market system through implementing standardized protocols and practices to fulfill CDS transactions, facilitate management and monitoring, and avoid disputes.2021 Also, since CDS contracts are sold OTC, their underlying details (such as maturity, currency, settlement method, etc.) are more flexible and rely on the demands of the contract players.22

A sovereign credit default swap (SCDS) is a credit default swap derivative, where the reference entity is a government.23 SCDS and CDS spreads provide insightful information into credit risk and are generally sufficient indicators of credit ratings due to the highly liquid nature and extensive volume of their trade.24

2.3. European sovereign debt crisis:

An imperative series of events that proved to be a decisive force in reshaping the Eurozone’s financial, economic, and political scope was the European sovereign debt crisis (EDC hereafter).

The EDC refers to a financial crisis period that displayed high levels of government debt, financial system collapse, among other adversities that exhausted several European nations, between 2008 and 2012. 25 The most preeminent impetus behind the emergence of the EDC was the 2008 global financial crisis. 26 Prompted by the US subprime mortgage crisis, the 2008 financial global crisis

17 Augustin, Subrahmanyam, Tang, & Wang, “Credit Default Swaps: A Survey”, Foundations and Treends in Finance, 2014, 1-196

18 Vogel, Bannier, & Heidorn, “Functions and characteristics of corporate and sovereign CDS”, 2013

19 Augustin et al., “Credit Default Swaps: A Survey”, 1-196

20 Amadei et al., “Credit Default Swaps: Contract characteristics”, 11-14

21 Vogel, Bannier, & Heidorn, “Functions of corporate and sovereign CDS”, 21-28

22 Vogel, Bannier, & Heidorn, “Functions of corporate and sovereign CDS”, 11

23 Naifar, “What explains sovereign CDS spread changes in the GCC region”

24 Augustin et al., “Credit Default Swaps: A Survey”, 1-196

25 Beirne and Fratzscher, “The pricing of sovereign risk and contagion during the European sovereign debt crisis”, Journal of International Money and Finance (2013), 60-82

26 Lane, “The European sovereign debt crisis”, Journal of economic perspectives (2012), 49-68


was set in motion, through instigating a ripple effect of economic breakdowns in the form of financial institutions collapses, housing bubbles bursts, investor pessimism and skyrocketing interest rates that spurred the cost of borrowing drastically.27

Moreover, Greece’s economy, which relied excessively on debt, on account of their common currency and monetary policy with the EU, independent fiscal policy, coupled with low interest rates, which incentivized them to borrow beyond their capacity. They accumulated a sovereign debt to GDP ratio of 113%, which exceeded the 60% limit implemented by the Eurozone.28 Additionally, fear of a currency (euro) collapse spread across the region as revelations were made on under reporting activities within Greece’s deficit spending, where the true value of their fiscal deficit was 13.6%, contrary to their reported 6.7%, provoking fear of financial contagion to cause unsustainable levels of credit risks from high dollar value debt levels that deteriorated with higher interest rates.29 Subsequently, by 2009’s end, the member states of Greece, Spain, Italy, Ireland, and Portugal, whom were dubbed the offensive appellation, “PIIGS”, as they experienced the most profound detrimental effects of the crisis, among the Eurozone states.30 They found themselves in a position where refinancing their debt or employing a bailout strategy was unattainable without external aid from the European Central Bank, IMF, or European Financial Stability Facility (EFSF).31

Bierne and Fratzscher (2013) studied the extent to which financial markets have been overpricing sovereign risk in the Eurozone area during the EDC, and how contagion plays a role.32 Through critically investigating 31 European nations (advanced and market economies) between 2000 and 2011, they analyzed the relationship between sovereign risk instruments (long-term government spreads, CDS spreads, and sovereign credit ratings) and economic fundamentals (common determinants), and concluded that deterioration in macro fundamentals and contagion were leading explanatory variables of CDS spreads.33 This verifies the findings of Broto and Pérez-

27 Lemieux, “Why Greece defaulted and others will follow”, Regulation (2011), 5

28 Lemieux, “Why Greece defaulted and others will follow”, p.5

29 Monokroussos, Platon, “The challenge of restoring debt sustainability in a deep economic recession:

the case of Greece”, A Financial Crisis Manual (2015), 170-188

30 Brandorf and Holmberg, "Determinants of sovereign credit default swap spreads for PIIGS - A macroeconomic approach." (2010).

31 Beirne and Fratzscher, “The pricing of sovereign risk”, 60-63

32 Brandorf and Holmberg, “"Determinants of sovereign credit default swap spreads for PIIGS”

33 Beirne and Fratzscher, “The pricing of sovereign risk”, 60-63


Quirós (2015) who deduced that on the emergence of the EDC, financial contagion played a systematically active role in those peripheral countries (“PIIGS”), which illuminated on the economic connection between the underlying nations.34

3. Literature review

In this section, we inspect the existing literature for potential determinants of CDS premiums, however, to ensure proper efficient analysis of the determinants outlined within the literature, this section will be divided according into the following categories: (i) global determinants, (ii) specific national-level macroeconomic determinants.

3.1. Global financial/economic factors

A multitude of research papers have accentuated the substantial implications global financial and economic factors bear on to CDS spreads, especially for emerging markets: Augustin et al.

(2014), Hilschier and Nosbusch (2010), Naifar (2020), Anton and Nucu (2020), Doshi et al.

(2015), Yang et al. (2018), Dailami et al. (2008), Longstaff et al. (2011), and Galil et al. (2013), among others.

First of all, Hilschier and Nosbusch (2010) investigated the extent to which macroeconomic fundamentals can justify fluctuations in CDS spreads. They initially evaluated the influence of country-specific factors and global determinants on sovereign debt prices using a dataset of 31 emerging market economies from 1994 to 2007. The relevant global factors they utilized within their regressions were: volatility of the S&P 500 index (VIX), the US default yield spread (computed as the difference between corporate bonds with AAA and BAA Moody’s ratings), 10- year US treasury yield as a proxy for world interest rate, and TED spread (to operationalize change in liquidity through the difference between 3-month LIBOR rates and 3-month treasury rates).35 The study incorporated several marginal specifications for the regression, one of which accounted solely for the global factors mentioned, without the macroeconomic fundamentals and control variables (will be discussed in the macroeconomic determinants subsection). The results illustrated a low explanatory power rate, as delineated by an adjusted R-squared value of just 0.13, however,

34 Broto and Perez-Quiros, “Disentangling contagion among sovereign CDS spreads during the European debt crisis”, Journal of empirical finance (2015), 165-179

35 Hilscher, Jens and Nosbuch, Yves. “Determinants of sovereign risk: Macroeconomic fundamentals and the pricing of sovereign debt”, Review of Finance (2010), 235-262


the baseline regression, which employed both macroeconomic and global factors generated a higher R-squared value of 0.53, which signified the important influence global factors have on CDS spreads.36 Moreover, the coefficient on the VIX index was positive and statistically significant at 1%, which was in line with the findings of Longstaff et al. (2007) that observed indications of conformed movements of sovereign CDS spreads resulting from the extensive reach of these global factors’ impact, and likewise, conformed with the results of Pan & Singleton (2008) that found the VIX index statistically significant in justifying CDS spreads in three countries:

Mexico, Turkey, and South Korea.3738 Additionally, the findings suggested a positive coefficient for the US default yield spread and the 10 year US treasury yield, and a negative coefficient for TED spread.39

Furthermore, using a quantile regression model, Naifar (2020) expanded upon the existing literature by adding further meaningful global factors, and accounted for different market circumstances (bearish, normal, bullish). The study analyzed the drivers of sovereign credit risk spreads within four Gulf Cooperation Council (GCC) countries, specifically Saudi Arabia, United Arab Emirates (UAE), Qatar, and Bahrain, utilizing weekly data from April 2013 to January 2020.40 The underlying global financial market regressors were: volatility of S&P 500 index (VIX), Merrill Lynch Option Volatility Estimate (MOVE) index to instrumentalize the global bond market uncertainty, the crude oil volatility index (OVX) established by the Chicago Board Options Exchange (CBOE) to represent oil price uncertainty (market expectations of the 30 day volatility of oil prices), and GVZ index to measure the effect of commodity market (market expectation of the 30 day volatility of gold prices).41 Moreover, the global uncertainty (risk) factor was the US economic policy uncertainty (EPU) index, the 10 year US treasury rate as a proxy for world interest rate, the TED spread, and crude oil prices.42 The quantile regressions suggested that the most substantial explanatory variables were the global financial uncertainties instilled in the VIX and MOVE indexes, however, the MOVE index’s effect was evident only when the market was bullish

36 Hilscher and Nosbusch, “Determinants of sovereign risk”, 247

37Hilscher and Nosbusch, “Determinants of sovereign risk”, 248-249

38Naifar, “What explains sovereign CDS spread changes in the GCC region”

39Hilscher and Nosbusch, “Determinants of sovereign risk”, 251-258

40Naifar, “What explains sovereign CDS spread changes in the GCC region”, 7

41Naifar, “What explains sovereign CDS spread changes in the GCC region”, 5-9

42Naifar, “What explains sovereign CDS spread changes in the GCC region”, 5-9


(when CDS spreads and credit are surging). 43 There was no evidence of influence between the EPU and GVZ on CDS spreads; however, the OVX displayed weak explanatory power.

Using daily data between January 2008 and April 2018, Anton and Nucu (2020) employed a vector autoregressive (VAR) model to inquire into the interrelation between sovereign CDS and stock markets in nine emerging economies from Central and Eastern Europe, with Granger causality as the focal point. The findings indicated a two-way duplex relationship between the CDS spreads and stock markets in the countries, which emphasized evidence of the presence of bidirectional feedback between sovereign CDS and stock markets in CEE countries. The results highlight an unpredictable, random shift of risk from the private to public during the European sovereign debt crisis. Additionally, the results delineated that the association between the CDS spreads and stock markets varies over time, and can be the subject of a paradigm shift, determined by global financial circumstances, such as debt crises.44 This corresponds with the conclusions of Augustin et al. (2014) that recommend that sovereign CDS spreads “co-move” and “jump”

together over time considerably, especially, in the presence of global events that influence risk spreads such as the sovereign debt crisis.45 Similarly, this is further corroborated by the empirical results of Eyssel et al. (2013) which regressed global financial and risk factors (the nature of China’s CDS spreads, VIX, default spread, and slope of the term structure), and country-specific variables (which will be outlined in the subsequent subsection) on Chinese sovereign CDS using vector autoregressive models performed on a sample of Chinese spreads from January 2001 to December 2010, controlling for exogenous covariates.46 They divided the sample into two periods, and likewise, discovered greater levels of co-movement in years aligning with the global financial crisis.47 However, country-specific covariates (Chinese stock market index, real interest rate, etc.) displayed a superior explanatory power on CDS spreads in terms of absolute levels and volatilities.48

43Naifar, “What explains sovereign CDS spread changes in the GCC region”, 11-19

44 Anton & Nucu, “Sovereign Credit Default Swaps and Stock Markets”, 338

45 Augustin et al., “Credit Default Swaps: A Survey”, 1-196

46 Eysell, Fung, and Zhang. “Determinants and price discovery of China sovereign credit default swaps”, China Economic review (2013), 1-15

47 Eysell et al. “Determinants and price discovery of China sovereign CDS”, 1-15

48 Eysell et al. “Determinants and price discovery of China sovereign CDS”, 7-15


Dailami et al. (2008) scrutinized the effects of global monetary conditions (“push” factors) versus country-specific determinants (“pull” factors) in pricing emerging market debt spreads using panel regressions of emerging market interest rate spreads over US treasuries.49 Their framework of analysis utilizes the US monetary policies, due to America’s leading figure in the global economy, and they specify it in long-run relationships using US treasury bill rates (short rates), 10 year treasury bond rate, and interest rate spread among investment grade and speculative US corporate bonds.50 As mentioned previously, these variables are incorporated to express global risk appetite, resulting from the USA’s dominance in the world economy. The conclusions drawn suggest that interest rates in the international capital markets are indeed heavily influenced by US rates; however, the extent of the influence depends on the country’s debt levels. For instance, a sovereign nation with debt levels on the merge of solvency would receive steeper surges in their spreads.51 This verifies other sources of existing literature, such as Wang and Moore (2012), which supposed US interest rates as a proxy for global interest rate was the prominent driver behind high levels of correlation among CDS spreads.52

With the combination of a no-arbitrage reduced form framework based on default intensities that are driven by observable economic and financial predictors, Doshi et al. (2015) conducted a complex inspection on 28 countries, three CDS maturities (1 year, 5 years, and 10 years), using more than a decade of daily data in order to generate a robust, good fit model that corresponds with economic intuition.53 The empirical findings delineated a substantial disparity within the impact of the observable covariates on CDS spreads across countries and over time.54 In their benchmark model, CDS spreads displayed a positive relationship with stock market and exchange rate volatilities, yet an inverse relationship with interest rates and stock market yields. A

49 Dailami, Masson, and Padou. “Global monetary conditions versus country-specific factors in the determination of emerging market debt spreads”, Journal of International Money and Finance (2005), 1326-1336

50 Dailami et al. “Global monetary conditions versus country-specific factors”, 1329-1335

51 Dailami et al. “Global monetary conditions versus country-specific factors”, 1331-1336

52 Wang and Moore, “The integration of the credit default swap markets during the US subprime crisis:

Dynamic correlation analysis”, Journal of International Financial Markets (2012), 1-15

53 Doshi et al. “Economic and Financial Determinants of Credit Risk Premiums in the Sovereign CDS market”, 43-45

54 Doshi et al. “Economic and Financial Determinants of Credit Risk Premiums in the Sovereign CDS market”, 66-80


noteworthy finding is that risk premiums peaked during the 2008 financial crisis for all countries in their sample, and the Eurozone nations’ premiums were also excessive during the Eurozone debt crisis, which conforms with previous literature, as well as economic intuition, in the sense that in risky, uncertain periods of economic turmoil, risk aversion increases at an exponential rate, hence risk premiums increase.55

3.2. Macroeconomic determinants

Recalling the aforementioned study conducted by Hilschier and Nosbusch (2010), the country-specific macroeconomic variables incorporated were: change in terms of trade, terms of trade volatility, history of default, debt/GDP ratio, and reserves/GDP ratio.56 Terms of trade refers to an economy’s exports relative to its imports, and fluctuations in terms of trade hinders an economy’s ability to repay external debts, and the volatility of terms of trade is essential for deciphering aggregate national production (GDP) variability, and entail detrimental effects on long-term economic growth.5758 Furthermore, the regression on 31 emerging market economies from 1994 to 2007 generated the following outcomes: the macroeconomic factors display a substantial increase in adjusted R-squared values relative to specifications that included only global variables and credit ratings. Moreover, it was concluded that terms of trade and CDS spreads entail an inverse relationship, as terms of trade deterioration causes higher spreads, and vice versa.

Additionally, CDS spreads are larger for nations that have recently endured through default, which logically makes sense, as history of default implies a worsen credit risk. This is consistent with the findings of Reinhart et al. (2003), which emphasized that history of default is a key driver to future default.59

Brandorf and Holmberg (2010) undertook a focused macroeconomic approach to exploring the determinants of sovereign CDS for the PIIGS block, which refer to Portugal, Italy, Ireland, Greece, and Spain, who were the weakest economies during the European debt crisis. They utilized the following variables: GDP growth rate, sovereign gross debt, inflation rate, and unemployment

55 Doshi et al. “Economic and Financial Determinants of Credit Risk Premiums in the Sovereign CDS market”, 66-80

56Hilscher and Nosbusch, “Determinants of sovereign risk”, 248-249

57 Mendoza, “Terms of trade uncertainty and economic growth”, Journal of development economics (1997), 323- 356

58Hilscher and Nosbusch, “Determinants of sovereign risk”, 235-238

59Hilscher and Nosbusch, “Determinants of sovereign risk”, 237-242


rate to perform regressions on CDS spreads of the aforementioned countries individually, against Germany as a benchmark.60 The dataset ran from Q1 2004 to Q3 2009 and was systematically split into two sup-periods with differing characteristics: financial stability vs financial distress. Their multiple regressions indicated a better fit for the latter period, as the former displayed low levels of explanatory power and statistical significance.61 With regards to the significance of the regressors; sovereign gross debt is consistently statistically significant, unemployment rate is the most regularly (across time-periods) significant, GDP growth rate displayed adequate levels of significance, despite not being as robust as the findings of Tang and Yan (2009), and inflation was discovered to be quite insignificant relatively.62

Furthermore, the study conducted by Galil et al. (2013) examined various macroeconomic and corporate-level factors that might be impetuses for valuing CDS premiums. These factors were categorized into four groups: firm level, macroeconomic or common, Fama & French factors, CRR (Chen, Roll, and Ross) factors. We will mainly focus on the results and analysis of the macroeconomic (common) factors and the CRR factors, as they are the ones relevant to the purpose of this paper. The macroeconomic determinants utilized consisted of the following: (i) spot rate, (ii) term structure slope, (iii) market condition, and (iv) market volatility. Moreover, the CRR determinants consisted of the following: (i) industrial production growth rate, (ii) unexpected inflation/change in expected inflation, (iii) term structure, and (iv) risk premium. The researchers collected the data for 718 American businesses during the period of January 2002 to February 2013, and through utilizing both time-series analysis and cross-sectional analysis, they concluded that the strongest explanatory variables are stock return (firm-level factor), stock volatility (firm- level), and MRI (median CDS spread of firms in the same rating class; common), however, through controlling them, alternative factors were revealed to have explanatory power in their absence.

Moreover, a significant revelation they discovered is that their result suggests their model better explains the period during and after the global financial crisis (July 2007 – June 2009, July 2009 – February 2013) than the period beforehand. This was due to an evident structural change in the valuations of CDS premiums, primarily among investment-grade firms.63 The findings regarding

60Brandorf and Holmberg, “"Determinants of sovereign credit default swap spreads for PIIGS”

61Brandorf and Holmberg, “"Determinants of sovereign credit default swap spreads for PIIGS”

62 Brandorf and Holmberg, “"Determinants of sovereign credit default swap spreads for PIIGS”

63 Galil, Shapir, Amiram, and Ben-Zion. “The Determinants of CDS spreads”, Journal of Banking &

Finance (2014), 271-282


the CRR factor of industrial production growth rate (MP) are consistent with that of Gerhard and Dieter (2016) that outlined a robust correlation between “peripheral” sovereign CDS premiums and industrial production in the euro-area.64

Benbouzid et al. (2017) contributes towards existing literature by targeting bank-level CDS spreads, and how country-level financial structures can shape them. By employing a linear regression controlled for heterogeneity by using random effects and fixed effects, they analyzed 58 banks from 15 nations between 2004 and 2011, accounting for bank level characteristics (leverage, regulatory capital, asset quality, liquidity, and operations income), country-level determinants (economic, financial, and political risk rating indexes), and financial structure measured at the national aggregate level (financial stability, depth, access, and efficiency).65 The regression synthesized a vector called “Bank”, consisting of all previously mentioned bank-level characteristics, a vector called “country” consisting of the indexes, and a vector called “Finstruc”

that includes the four financial structure variables.66 The approach taken within this study is actually quite compelling, as they provide a perceptive measure to study the financial framework of an economy.

Finally, a unique approach taken to investigate European credit default swaps, Calice et al.

(2015) conducted a region-focused scrutiny of short-term components of idiosyncratic sovereign risk premiums.67 Using the difference between 5-year and 10-year CDS spreads as representative of the dynamics of sovereign CDS term premiums, they utilized a “Markov-switching unobserved component model” to break down the CDS spreads of five European countries into two statistically distinct parts, and subsequently, connect them through a vector autoregression to several financial market covariates.68 They concluded that the covariates which generated the most robust outcomes were local stock returns, risk aversion, and CDS market liquidity.69 Moreover, CDS spreads’

64 Gerhard & Dieter, “European CDS Premiums and Industrial Production”, Journal of US-China Public Administration (2016), 256-262

65 Benbouzid, Mallick, and Sousa, “Do country-level financial structures explain bank-level CDS spreads”, Journal of International Financial Markets, Institutions, and Money (2017), 135-145

66 Benbouzid et al. “Do country-level financial structures explain bank-level CDS spreads”. 139

67 Calice, Mio, Sterba, and Vasicek. “Short-term determinants of the idiosyncratic sovereign risk

premium: A regime-dependent analysis for European credit default swaps”, Journal of Empirical Finance (2015), 174-189

68 Calice et al. “Short-term determinants of the idiosyncratic sovereign risk premium”. 182-184

69 Calice et al. “Short-term determinants of the idiosyncratic sovereign risk premium”. 183-188


response (elasticity) to shocks is “regime-dependent” and 10 times stronger during periods of high volatility, which verifies the multitude of previous studies mentioned thus far.70

In light of these findings derived from the previous literature explained above, the following hypothesis were derived:

Hypothesis 1: The Western European CDS spreads will exhibit ampler responsiveness rates to global determinants (VIX, MOVE, GVZ, US10Y, GEPU, and TED spread) due to their enhanced status in the world economy, which emanates their sensitive elasticity to global shocks and fluctuations. The Southern European region will display comparable reactions, but at lower magnitudes (smaller beta coefficients).

Hypothesis 2: The Southern European CDS spreads will, to a greater extent, delineate elastic sensitivity (larger beta coefficients) to country-specific macroeconomic variables relative to the Western region, due to the Southern region’s less stable and prosperous economic conditions, which accommodates their under-supported exposure to exogenous shocks.

Hypothesis 3: During times of economic turmoil such as the financial crisis (2008) and Eurozone debt crisis (2009-2012), CDS spreads will progressively upsurge for both groups due to higher investor risk aversion rates, lower credit ratings, and higher interest rates. Hence, the crisis variable will retain effective explanatory power, which in turn, will improve the R-squared statistic intensely.

Hypothesis 4: Country-specific macroeconomic variables will effectuate substantial explanatory power comparative to global variables in analyzing CDS spreads. This is by virtue of macroeconomic factors’ highly intensified relevance towards credit risks, compared with those of global financial market volatilities. This hypothesis will be assessed by comparing each individual set of regressions, using the R-squared statistic.

4. Empirical methodology

This section outlines the panel data analysis set up, demonstrates the data needed to produce the research, provides a fundamental depiction of the empirical econometric model specifications that will be employed, and their respective regression equations.

4.1. Research data

70 Calice et al. “Short-term determinants of the idiosyncratic sovereign risk premium”. 188


As aforementioned in the introduction, the sample nations that will be investigated are France, Belgium, Germany, Netherlands, Portugal, Italy, and Spain. The former four countries, who represent the Western region of Europe, portray the prosperous and highly advanced economies, while the latter three, who represent the Southern European region, depict the less developed economies. Initially, it was intended that Greece be incorporated as part of the Southern region, as it an important economy in terms of CDS spreads analysis, as explained by their extreme adversity with debt levels and credit risk during the European debt crisis. However, their data was dropped, as it was apparent that their CDS patterns are outliers, and are not representative of the Southern region, thus including it would have been problematic through inflating regression coefficients.

The type of data needed to accomplish the investigation is a set of global financial covariates (VIX, long-term US treasury rate, MOVE index, GVZ index, Global EPU, and TED spread), and a set of country-specific macroeconomic factors (current account, inflation rate, federal reserves, gross external debt, unemployment rate, exchange rate, industrial production, GDP, terms of trade, economic policy uncertainty, and credit ratings).

The motive behind choosing such an extensive list of determinants is threefold: (1) to limit the possibility of omitted variable bias, as the panel data model of choice is a pooled OLS estimator, (2) to achieve a high R-squared statistic to prove effective explanatory power, and (3) to effectively test the hypothesis whether macroeconomic factors entail econometric significance relative to global covariates.

In the study, we control for the global financial covariates (VIX, MOVE, GVZ, TED, US10Y, and EPU), as well as, the foreign exchange rate (FOREX), since they do not differ in value between the Southern and Western region, which will be demonstrated further in the descriptive statistics section, however, their influence on their respective CDS spreads is bound to show clear differences.

4.2. Econometric models

The empirical model that will build our framework of econometric analysis regarding the impact of global and country-specific factors on sovereign CDS spreads in the western and southern European regions is represented by the following baseline regression equation:


𝐶𝐷𝑆$,& = 𝛼$+ 𝛽+𝑉𝐼𝑋&+ 𝛽/𝑀𝑂𝑉𝐸&+ 𝛽3𝐺𝑉𝑍&+ 𝛽6𝐸𝑃𝑈&+ 𝛽9𝑈𝑆10𝑌&+ 𝛽=𝐶𝐴$&+ 𝛽?𝐹𝑂𝑅𝐸𝑋$,&

+ 𝛽B𝐷𝐸𝐵𝑇$,&+ 𝛽E𝑃𝑅𝑂𝐷$,&+ 𝛽+F𝐼𝑁𝐹𝐿𝐴𝑇𝐼𝑂𝑁$,&+ 𝛽++𝐺𝐷𝑃$,&+ 𝛽+/𝑇𝐸𝐷$,&

+ 𝛽+3𝑇𝑂𝑇$,&+ 𝛽+6𝑈𝑁𝐸𝑀𝑃$,&+ 𝛽+9𝑅𝐸𝑆𝐸𝑅𝑉𝐸𝑆$,&+ 𝛽+=𝑟𝑎𝑡𝑖𝑛𝑔$,& + 𝜀$,& (1)

where 𝐶𝐷𝑆$,& depicts the sovereign CDS spreads for country (i) at time (t), 𝑉𝐼𝑋&is the implied volatility on S&P 500 index, 𝑀𝑂𝑉𝐸&refers to the Merrill Lynch Option Volatility Estimate index,

𝐺𝑉𝑍& delineates the volatility for the gold index, 𝐸𝑃𝑈& is the economic policy uncertainty global

index, 𝑈𝑆10𝑌& is the long-term (10-year) US treasury rate, 𝐶𝐴$& is the national current account level, 𝐹𝑂𝑅𝐸𝑋$,& is the foreign exchange rate between Euro and USD, 𝐷𝐸𝐵𝑇$,& refers to the gross external debt, 𝑃𝑅𝑂𝐷$,& is the national industrial production, 𝐼𝑁𝐹𝐿𝐴𝑇𝐼𝑂𝑁$,& is the inflation rate,

𝐺𝐷𝑃$,& is the gross domestic product, 𝑇𝐸𝐷$,& is the Treasury-Euro-Dollar rate, 𝑇𝑂𝑇$,& depicts the

national terms of trade, 𝑈𝑁𝐸𝑀𝑃$,& is the unemployment rate, 𝑅𝐸𝑆𝐸𝑅𝑉𝐸𝑆$,& delineates the international federal reserves, 𝑟𝑎𝑡𝑖𝑛𝑔$,& depicts a sovereign nation’s credit rating, and finally,

𝜀$,& refers to the regression error term. The alpha term (𝛼$) represents the regression constant, and

the betas (𝛽) are the coefficient terms of the regression, which measure the sensitivity of CDS to the underlying covariates.

The regressions that will be conducted within this research will be done twice, once for the Western group, and once for the Southern group. The empirical methodology will incorporate four model specifications (including the baseline) of Equation (1), to produce a total of 8 regression outputs.

The first model specification is the baseline regression, encompassing all explanatory variables, global and country-specific, on West and South groups. Refer to equation (1) for specifics.

The second and third model specification will regress CDS, once only using the six global covariates (VIX, MOVE, GVZ, EPU, US10Y, and TED), and once using the 10 country-specific macroeconomic factors to assess their independent explanatory power and influence on CDS spreads. These are modelled by the following respective regressions:

𝐶𝐷𝑆$,& = 𝛼$ + 𝛽+𝑉𝐼𝑋&+ 𝛽/𝑀𝑂𝑉𝐸&+ 𝛽3𝑇𝐸𝐷&+ 𝛽6𝐺𝑉𝑍&+ 𝛽9𝐸𝑃𝑈&+ 𝛽=𝑈𝑆10𝑌&+ 𝜀$,& (2)


𝐶𝐷𝑆$,& = 𝛼$ + 𝛽+𝐶𝐴$&+ 𝛽/𝐹𝑂𝑅𝐸𝑋$,&+ 𝛽3𝐷𝐸𝐵𝑇$,&+ 𝛽6𝑃𝑅𝑂𝐷$,&+ 𝛽9𝐼𝑁𝐹𝐿𝐴𝑇𝐼𝑂𝑁$,&+

𝛽=𝐺𝐷𝑃$,&+ 𝛽?𝑇𝑂𝑇$,&+ 𝛽B𝑈𝑁𝐸𝑀𝑃$,&+ 𝛽E𝑅𝐸𝑆𝐸𝑅𝑉𝐸𝑆$,& + 𝛽+F𝑟𝑎𝑡𝑖𝑛𝑔$,&+ 𝜀$,& (3)

The final model specification will incorporate a dummy variable called “crisis” depicting whether a period is during a financial crisis, whereas in this case, crisis will equal to 1 for the years up to and including 2012 (2009-2012), as the Eurozone debt crisis began to unwind by 2012, compared to the years prior. It will include all variables, global and country-specific, as the focus is solely on the difference between crisis and non-crisis periods. The equation is as follows:

𝐶𝐷𝑆$,& = 𝛼$+ 𝛽+𝑉𝐼𝑋&+ 𝛽/𝑀𝑂𝑉𝐸&+ 𝛽3𝐺𝑉𝑍&+ 𝛽6𝐸𝑃𝑈&+ 𝛽9𝑈𝑆10𝑌&+ 𝛽=𝐶𝐴$&+

𝛽?𝐹𝑂𝑅𝐸𝑋$,& + 𝛽B𝐷𝐸𝐵𝑇$,& + 𝛽E𝑃𝑅𝑂𝐷$,&+ 𝛽+F𝐼𝑁𝐹𝐿𝐴𝑇𝐼𝑂𝑁$,&+ 𝛽++𝐺𝐷𝑃$,&+ 𝛽+/𝑇𝐸𝐷$,&+

𝛽+3𝑇𝑂𝑇$,& + 𝛽+6𝑈𝑁𝐸𝑀𝑃$,& + 𝛽+9𝑅𝐸𝑆𝐸𝑅𝑉𝐸𝑆$,&+ 𝛽+=𝑟𝑎𝑡𝑖𝑛𝑔$,& + 𝛽+?𝑐𝑟𝑖𝑠𝑖𝑠&+ 𝜀$,& (4)

4.3. Research design & statistical tests

The research design that was employed within this investigation was panel data analysis, as it combines elements of both cross-sectional and time-series studies, which is the most appropriate for our set of data in order to analyze our range of determinants of CDS spreads within different countries across time. The panel data is balanced, as all seven countries have observed variables in all time periods.

Panel data analysis includes a dynamic range of models, the most common of which are pooled OLS estimation, fixed effects, and random effects. After careful evaluation of all three models, it was decided that pooled OLS estimation was the best fitting one due to the abundance of explanatory variables, it expected that there will little or none omitted variable bias (OVB), and exceptionally high R-squared values, which makes pooled OLS a good fit model with unbiased, consistent estimates. However, in order to verify this decision, a series of statistical tests were conducted.

Firstly, a Breusch-Pagan (1979) and Cook-Weisberg test for heteroskedasticity was run in STATA to ascertain whether the variance of the error terms are dependent on the regressing variables. Table 1 below exhibits the heteroskedasticity test results, where the null hypothesis (H0) is constant variance (homoskedasticity), and the outcome produced a significantly small p-value


of 0.0000, indicating the null hypothesis of homoskedasticity is rejected, and we concluded that the error terms are heteroskedastic.

Table 1: STATA output for the Breusch-Pagan/ Cook-Weisberg test:

H0: Constant variance (homoskedasticity) Chi-Squared P-value

18.95 0.0000

Due to heteroskedasticity, we utilize robust standard errors in order to obtain unbiased standard errors for OLS coefficients under heteroskedasticity.

Next, a Hausman specification test (1978) was conducted in order to be able to select between the fixed effects and random effect models. However, it was not possible due to the use of robust standard errors. A further attempt at a Hausman test was done, this time without robust standard errors, which is statistically not valid, however, it revealed that random-effects model is preferred, as the null hypothesis (H0) is to select random effects, and a large p-value of 0.8443 was derived, hence, we accept the null hypothesis.

Alternatively, due to the statistically invalid approach of the Hausman test without robust standard errors, an Over-Identification test, described by Arellano (1993) and Wooldridge (2002) was performed. Fixed effects use the orthogonality condition that suggests regressors are not correlated with the idiosyncratic error. Contrarily, random effects utilize the orthogonality condition that suggests regressors are not correlated with the group-specific error term (hence the name “random effect”). The over-identification test implements an artificial regression, whereby the random effects equation is reapproximated through “augmentation” with supplementary variables consisting of the original regressing variables, modified into deviations from the mean.

The result yielded a statement informing that random effects will produce equivalent results to the pooled OLS estimator.

Henceforth, a Breusch-Pagan LaGrange-Multiplier (LM) test was incorporated to test for random effects model based on the OLS residual, by assessing whether the variance of the error term is significantly different from zero. Table 2 below exhibits the test results, which includes the null hypothesis (H0) that pooled OLS estimation is preferred, the chi bar-squared statistic, and its


relevant p-value. The test evidently generated a p-value of 1.000, thus, we accept the null hypothesis and choose a pooled OLS analysis.

Table 2: STATA output for Breusch-Pagan LaGrange-Multiplier (LM) test H0: Pooled OLS is preferred

Chi-Bar- Squared


0.00 1.0000

Therefore, all the tests performed above (heteroskedasticity, Hausman, Over-Identification, and Breusch-Pagan LM) corroborate the decision to utilize pooled OLS estimation as our panel data model for our analysis on the global and country-specific determinants of sovereign CDS spreads.

Moreover, two more diagnostic statistical tests were performed to ensure that the data is reliable: (i) unit root test to assess stationarity, and (ii) Variance inflation factor (VIF) test to check for multicollinearity between the regressors.

The Levin-Lin-Chu unit root test was employed, taking into account time trends, to assess whether the independent variables are stationary. Table 3 below outlines the unit root test results for the following hypotheses: the null hypothesis (H0) suggests that the panels contain unit roots (non-stationary), and the alternative hypothesis (Ha) suggests panels are stationary.

Table 3: Unit root tests (LLC) – STATA output H0: Panels contain unit roots

Ha: Panels are stationary

Independent Variable P-value

VIX 0.0018

MOVE 0.0001

GVZ 0.0000

EPU 0.0083

US10Y 0.0000

TED 0.0047


Independent Variable P-value

CA 0.0000

FOREX 0.0000

DEBT 0.0003

PROD 0.0000


GDP 0.0140

TOT 0.0000

UNEMP 0.0000


Hence, as depicted in Table 3 above, all variables are stationary, meaning that their statistical properties do not vary overtime, which allows our results and analytical tests to be reliable.

With regards to the multicollinearity check, the VIF (Variance Inflation Factor) command was utilized in STATA. The test yielded a mean VIF of 5.18, which is below the universally acknowledged limit of 10, indicating that multicollinearity is not an issue in our data. However, in the beginning of the study, there were three extra variables in our regression (OVX: crude oil market volatility, Eurostoxx50: European stock market return index, and VSSTOXX: European stock market volatility) that were later dropped as they posed high correlations with other variables, which gave rise to multicollinearity issues, and high VIF values. However, once removed, as stated above, the VIF value decreased to the appropriate level, and there was no issue in removing these variables, as there are plenty of variables in our regression already that estimate market volatilities.

5. Data

This section will provide detailed reasoning and interpretation behind the selection of all variables of our panel data, and subsequently, summary statistics will be demonstrated.

5.1. Sample

As previously mentioned, the study will be conducted on two groups of European regions (West and South), totaling seven countries: France, Belgium, Germany, and Netherlands represent


the West group, while Portugal, Italy, and Spain represent the South region. The motive behind choosing these specific regions is that during the Eurozone crisis, the southern region experienced immense economic adversity in the form of high government debt, deteriorating credit ratings, investor pessimism, surging interest rates, among other. All these issues skyrocketed their sovereign CDS levels. They were dubbed an offensive moniker, “PIIGS”, for their lack of economic strength during the crisis, which makes them important for the study of CDS determinants in Europe, as they constitute the less-developed economies in the region. Hence, in order to create a balance, coupled with a dynamically compelling investigation, it was decided that the opposing group should constitute the more advanced economies in Europe, who are mainly situated in the West.

Daily, quarterly, and annual data was collected for 10 years, 2009 until 2019, in order to:

(1) capture the effects of crisis periods (2009-2012) versus non-crisis periods (2013-2019), and (2) retain a large number of observations to ensure statistical effectiveness.

5.2. Credit default swaps

Our dependent variables consisted of Sovereign Credit default swaps (CDS) data with 5- year maturities, as the 5-year contracts are systematically the most liquid and readily available in the market. Daily CDS data was collected from the Eikon/DataStream database from 2 March 2009, to 31 December 2019. The CDS spreads are instrumentalized in basis points (bps).

Moreover, there are three forms of restructuring that the ISDA approves of in CDS contracting:71 (i) full restructuring (FR)

(ii) modified restructuring (MR)

(iii) modified-modified restructuring (MMR) (iv) no restructuring (XR)

Each restructuring form has their impact on the level of CDS quotes, which differs by industry and country. According to Berndt et al. (2007), modified-modified restructuring (MMR) is primarily the most collectively popular restructuring type in European financial markets, followed by full restructuring (FR).72 On the Eikon database, modified-modified restructuring

71 Berndt, Jarrow, and Kang. “Restructuring risk in credit default swaps: An empirical analysis”, Stochastic processes and their applications (2007), 1724-1749

72 Berndt et al. “Restructuring risk in credit default swaps: An empirical analysis”, 1726-1727


CDS spreads data was missing for five out of the seven countries (Germany, Netherlands, France, Portugal, and Spain), thus, we opted for the alternative best option, which is full restructuring.

Figure 2a, 2b, and 2c below display the CDS spread patterns (time trend) within the seven nations during the period of study:

Figure 2a: CDS spreads for the Western group: 2009 – 2019

Figure 2b: CDS spreads for the Southern group: 2009 – 2019


Figure 2c: CDS spreads for both groups: 2009 – 2019

Figure 2a delineates that the SCDS spreads of Belgium began exhibiting growth starting 2010, which is at the onset of the Eurozone debt crisis and peaked towards the end of 2011 reaching CDS spread levels of approximately 399 bps. For the remainder of the Western nations, all CDS spreads yielded similar patterns and reactions, as displayed by the co-movement of their CDS changes, however, at distinct intensities, whereby Belgium displayed the most prominent surges in CDS spreads, followed by France, Netherlands, and then Germany. The effects of the Eurozone debt crisis began to diminish by mid-2012 as displayed by progressive, consistent drops in CDS spread levels.

On the contrary, Figure 2b reveals that the SCDS spreads of Portugal began exhibiting growth by 2010, at the emergence of the Eurozone debt crisis, however, peaked around early 2012, accumulating average CDS spreads of 1,600 bps. The other Southern countries however display signs of harmonized co-movement, as Spain and Italy are almost identical. The higher levels of CDS patterns in this group is attributed to, as mentioned previously, immense levels of accumulated government debt, coupled by loss of confidence from investors, widened cost of insurance (CDS), and inflated interest rates.

Figure 2c is added just to provide a visual illustration of the CDS patterns of both groups together. As exhibited by each country’s colored line, the largest three patterns are those economies in the Southern region (Portugal, Italy, and Spain), followed by the four Western region


economies. It can be also deduced that resulting from the interconnectedness of the global economy, economic coordination between the EU member states, and structural frameworks such as the 2008 financial crisis, all seven countries display co-movements (harmonized patterns), yet at distinct intensities.

5.3. Global financial and economic variables

To assess the effect of global financial and economic variables on CDS spreads, a dynamic range of determinants, namely: VIX, MOVE, GVZ, GEPU, US long term interest rate, and TED spread, were collected on a daily frequency from March 2, 2009 till December 31, 2019.

Firstly, in order to quantify pessimism and financial uncertainty to instrumentalize market risk, the VIX index was collected on a daily frequency from the Eikon database. The VIX provides an index which measures the implied volatility of the S&P 500 index through generating a 30-day forward volatility estimation.73 It is a widely used index, created by the CBOE (Chicago Board Options Exchange) to measure financial market volatility and risk aversion. Secondly, the MOVE index, likewise created by the CBOE, was collected on a daily frequency from Eikon to account for global bond market’s uncertainty, as it is relevant to inspect contagion between bond market investor sentiments and that of credit default swaps. Furthermore, the GVZ index was gathered from Eikon on a daily frequency, to operationalize the market’s expectation of the 30-day volatility for gold. Fifthly, through Nicholas Bloom’s database, the index GEPU, which quantifies global economic policy uncertainty, was gathered on a monthly frequency. This index is a GDP-weighted average of national EPU indices for 21 nations (five of whom are part of our study: France, Germany, Italy, Netherlands, and Spain), which reflects the comparative persistence of “economic policy uncertainty” in daily newspaper articles.74 Additionally, to capture the global interest rate, the United States long-term interest rate was utilized as a proxy, due to the USA’s and the dollar’s profound command in the global economy. The long-term interest rate was gathered on a quarterly frequency from OECD and is primarily determined by the prices levied on government bonds with 10-year maturities traded on financial markets, borrower risk, and capital value reduction. Finally, TED spread, which measures the difference between the 3-month LIBOR (London Interbank

73 Naifar, “What explains sovereign CDS spread changes in the GCC region”, 11-19

74 Baker, Bloom, and Davis. “Measuring Economic Policy Uncertainty”, Quarterly Journal of Economics (2016), 1593-1636



The hypotheses build on the main aspects gained from the literature: (1) There is a GFC in capital flows, credit growth, leverage and asset prices, (2) US monetary policy

Based on these findings, guidelines are drawn up for the three main parties involved in the CDS market. The first involved party, the ISDA, is the party that has the most influence

It can be concluded that a bond issue during a low business cycle is a valuable addition to the model explaining the credit default swap spread since the coefficient is significant

Looking at the total impact of the studied explanatory variables it is clear that macro-economic variables do indeed influence sovereign CDS spreads, which is positive for

In doing so, the answer is sought to the question of whether investors in the bond market have changed their focus towards Long Term Issue Credit Ratings (LTRs)

As mentioned above the sovereign spread of GIIPS countries do not react different to changes in the debt to GDP ratio, credit rating or US yield compared to the other

When the quality of the portfolio of a bank lowers, the ratio of the variable increases and investors will be buying more credit default swaps which results in a higher spread

As we can see in Table 3, the pricing effects on average are very small and vary distinctly among dealers: (1) the average coefficient of 13 dealers is 0.01038, which