G. G. Postma The pricing of sovereign risk in the Eurozone By

The pricing of sovereign risk in the Eurozone

By

G. G. Postma

University of Groningen

Faculty of Economics and Business

Msc International Economics and Business

2

The pricing of sovereign risk in the Eurozone

Table of content

Introduction

3

Literature review

4

Hypotheses

6

CDS spreads

7

Methodology, data and analysis

8

Correlation matrices

14

Regression analysis

17

Conclusion

22

References

24

Appendix

25

The purpose of this Master thesis is to examine whether the Credit Default Swap (CDS) market in the Eurozone is influenced by internal and external factors. The research is conducted using a panel data set of 14 Eurozone countries and by estimating a random effects regression model with quarterly CDS data. The internal factors of the research are external debt to GDP and official reserve assets to GDP and the external factors are VIX, Treasury rate and TED spread. I find that the CDS market in the Eurozone is influenced by internal as well as external factors.

Credit Default Swaps, Eurozone, Sovereign risk, Sovereign default Supervisor: prof. dr. J. de Haan

Co-assessor: dr. M.J. Gerritse

3

Introduction

In the past the sovereign credit default swap market was almost only used for developing countries. Not a surprising phenomenon, since these developing countries were most likely to default on their debt. Sovereign credit default swaps are financial contracts offering insurance against losses from credit events on outstanding debt issued by sovereign entities (Fender et al., 2012). The government bonds of developed countries were used by investors as a proxy for the long-horizon risk-free rate. Empirical modelling was also mainly orientated towards interest rate risk or liquidity risk, instead of default risk. However, this all changed after the global crisis. Before the global crisis a default of debt issued by developed countries was treated as a very low probability event (Fontana and Scheicher, 2016). Walter Wriston in 1983, then Chairmen of Citibank famously said “countries don’t go bust” (Guill, 2009). However, countries do go bust, in the sense of refusing or being unable to meet their financial obligations (Fender et al., 2012). This became painfully clear during the European sovereign debt crisis, when the Greek government had to restructure its debt. Sovereign defaults mostly involve external debt, and tend to occur in periods of extreme stress. Thus, country risk is an

important factor in the pricing of sovereign debt. Assuming rational investors, one would expect that the credit spreads on sovereign debt instruments reflect these country-specific risks. However, evidence with regard to country-specific risks in the pricing of sovereign debt is rather mixed (Fender et al., 2012). The spreads of a CDS represent the quarterly payment that must be paid that must be paid by the buyer of CDS to the seller for the contingent claim in the case of a credit event (Aizeman et al., 2013). Longstaff et all. (2011) found that CDS spreads are also related to global and especially U.S. financial factors.

I examine what factors play a role in the pricing of the European sovereign credit risk by analysing the CDS spreads of Eurozone countries. I focus my work on the Eurozone because relatively little

4 research towards the pricing of sovereign risk. This study differentiate itself by looking at the

internal aspects as well as the external aspects that are involved with the pricing of sovereign risk. Also, only developed countries are used. Furthermore, all the countries used in the analysis belong the same currency area.

The next section discusses the literature related to sovereign CDS and the pricing of sovereign risk. The third section discusses the hypotheses. The fourth section discusses CDS spreads. The fifth section discusses the Methodology, data and analysis. The sixth section shows the correlation matrices. The seventh section presents the empirical results. Finally, section eight concludes the thesis.

Literature review

The sovereign credit default swap market has grown significantly in the past decade. In Europe the CDS market gained more influence during the European sovereign debt crisis. It is not surprising that this happened since the sovereign credit default swap market can be seen as a proxy for market based sovereign default risk pricing. The total CDS market grew from 10 trillion US dollars in 2004 to 60 trillion US dollars in 2008 (Aizenman et al., 2013). The buying and selling activity on developed sovereign CDS is a relatively new phenomenon. The main reason for this is that prior to the crisis participants had few incentives to negotiate CDS on developed countries. Sovereign risk was not considered significant for developed countries with a high rating, and therefore there is no need for a CDS (Delatte et al., 2012). Trading activity in the sovereign CDS spreads for developed countries was low (Arce et al., 2013).

The credit default swap market and the sovereign bond market go hand in hand. A credit default swap is a bilateral contract between a buyer and seller under which the seller sells protection against the credit risk of the reference entity of the same maturity. A CDS premium is commonly expressed is basis points per annum. These basis points are an underlying fracture of the costs of a sovereign default (Fontana and Scheicher, 2016).

5 multiple highly related financial assets are traded on more than one market, each market may be involved in the price discovery process, but the one that provides a combination of the greatest liquidity, lowest execution costs and greatest leverage opportunities should dominate (Booth et al., 1999). The characteristics of the CDS market are the reason that the CDS market is used rather than the underlying bond market in times of distress. A CDS contract can be negotiated at any time in contrary to bonds. A buyer of a CDS contract puts upward pressure on the premium of the CDS, which exerts a downward pressure on the underlying bond prices. Also in general a CDS contract is not bought because the investor expects a default, but the investor anticipates that the sovereign spreads will increase further implying a rise in the CDS premia (Delatte et al., 2012).

Coudert and Gex (2010) argue that for low-yield countries the price discovery takes place in the bond market and that for high-yield countries the direction changes. Delatte et al. (2012) found with their nonlinear model that in times of distress the information transmission between the CDS and bond markets is dominated by de CDS market. Even more, in high-yield countries the CDS market

dominates the information transmission at all levels. The results stipulate the role of the CDS market as an instrument to speculate against deteriorating conditions of sovereigns.

Aizenman et al. (2013) found that the market price risk of sovereign debt measured by CDS spreads is partly explained by fiscal space and other economic determinants. Fiscal space flexibility in the government spending choices and the overall financial well-being of a government (Heller, 2005). The countries they used in their research were located in- and outside Europe. By comparing the European periphery countries to other similar countries they found that the pricing of sovereign risk of the European periphery countries is not accurately predicted. They found that countries with similar fiscal conditions to European periphery countries had lower CDS spreads.

Fender et al. (2012) studied the determinants of daily spreads for emerging markets sovereign CDSs over the period April 2002 until December 2011 using GARCH models, generalized autoregressive conditional heteroscedasticity models. This type of model has been introduced by Bollerslev (1986). In their research Fender et al. find that country-specific risk factors are not so much the drivers for emerging markets CDS spreads instead the spreads are more related to global and regional risk premiums.

.

6 corresponding stock index returns. Furthermore, they found that especially US financial markets had a great impact on the CDS spreads. Longstaff et al. (2011) followed the research of Pan and Singleton (2008) by using an affine sovereign credit valuation model in order to decompose sovereign CDS spreads into their risk-premium and default risk components.. Affine models has proven to be a remarkably flexible structure for examining the dynamics of default-risk free bonds, and as a result affine modeling has become the dominant framework for term structure research since the early 1980s (Piazzesi, 2010). For more background and theory of the affine models I refer the reader to Duffie et al. (2003). Longstaff et al. (2011) found that both components are strongly related to global macroeconomic factors. The relation between default risk and global macroeconomic factors was in their research much stronger compared to the relation with the risk-premium component of the spread. This highlights the important role of international and especially the role of U.S. financial variables in determination of non-U.S. sovereign CDS spreads. This indicates the spillover effect of the US market on the European sovereign CDS market.

The research of Pan and Singleton (2008) show how the VIX index that represents the US stock market volatility shows a strong relation with the CDS spreads of Mexico, Turkey and South-Korea. This again points out that there is a strong influence of factors outside the own economy that have an influence on the spreads of sovereign countries. The VIX index used by Pan and Singleton (2008) was introduced by the Chicago Board Option Exchange (CBOE) in 1993. The VIX index was originally designed to measure the market’s expectation of 30-day volatility implied by at-the-money S&P 100 Index (OEX Index) option prices. The nowadays VIX index is based on the S&P 500 index, the core index for U.S. equities, and estimates expected volatility by averaging the weighted prices of SPX puts and calls over a wide range of strike prices. The VIX index is seen as the benchmark for US stock market volatility (Exchange, 2009).

Focusing more on the European debt crisis, Fontana and Scheicher (2016), argue that the high CDS premiums in the late 2010 of the Eurozone crisis may to some extent be caused by loss of appetite in high-risk investments and the limited market liquidity instead by macroeconomic factors such as outstanding debt. This phenomenon is also described as the flight-to-quality. The flight-to-quality also plays an important role in the observed CDS spreads.

Hypotheses

7 countries as well as European countries fiscal space and other economic determinants play a role in the pricing of CDS. They compared European and non- European countries. This research does not make a comparison between European and non- European countries. I focus on Eurozone countries and look if the sovereign CDS market is influenced by internal or external factors. Aizenman and Pasricha (2009) argue that additional reserves may allow a country to avoid costly liquidation and therefore lower the cost of the probability of default.

Hypothesis 1: The Eurozone sovereign credit default swap market is influenced by internal factors

such as the fiscal condition and reserves position of the Eurozone countries.

Besides the internal factors that might have an influence on the Eurozone sovereign CDS market external factors are also of importance. Previous studies have shown that there is a significant relation between the sovereign CDS market and global macroeconomic factors. Especially the U.S. financial market has a great influence on the CDS spreads of other countries. The research by Longstaff et al. (2011) and Pan and Singleton (2008), for example, has shown that the CDS spreads of emerging markets are to a large extent driven by US equity and US stock market volatility. To extend this view, I will be looking at the influence of the most important global macroeconomic factors on the European sovereign CDS market. By analyzing the impact of external effects on the European CDS market one can see to what extent the high spreads are driven by factors that are not in control of the local policy makers. Derived from this is the following hypothesis.

Hypothesis 2: The Eurozone sovereign credit default swap market is influenced by external factors

from the US such as the US treasury rate, VIX index and the TED spread.

CDS spreads

8 German sovereign debt. If the German government does not default than the buyer pays this for the full 5 year of the contract. If the events that are mentioned in the CDS contract would happen then the seller of the contract will pay the losses made by the buyer.

Methodology, data and analysis

This section describes how the hypotheses of this research are empirically analyzed. All data used is obtained from DataStream. The dependent variable is sovereign CDS spreads of 5 year maturity of 14 Euro countries. CDS spreads of 5 year maturity are used because these contracts are the most traded in the market. Quarterly observations between the 4th Quarter of 2007 and 1st quarter of 2016 are used and the CDS spread is denominated in US Dollars.

The focus of my work is to determine if the European sovereign CDS market is influenced by internal and external factors.

The internal perspective looks at the variables Externaldebt(%ofGDP) and

Officialreserveassets(%ofGDP), for the external perspective I will look at the variables VIX, Treasuryrate and TEDspread.

The variables used in this research are based on previous literature that also looked at the pricing of sovereign risk. All the variables used in the analysis are briefly explained in table 1. Fender et al. (2012), found that for emerging market borrowing conditions are influenced by both domestic as well as international economic conditions, with the external factors dominating. A higher debt to GDP ratio should increase sovereign borrowing costs. This conjecture is based on the assumption that a rise in the ratio increases the likelihood of a debt crisis.

9 The data is used in a panel regression. The model is as follows:

CDSspreadit = β1VIXit + β2Treasuryrateit + β3TEDspreadit + β4Externaldebt(%ofGDP)it +

β5Officialreserveassets(%ofGDP)it + β6X’it +αi + eit

In this model CDSspread is the depended variable, i stands for country and to for quarter; X is the vector of controls. The unobserved individual effect is captured by α and e is the error term . The control variables used in the analysis are Imports%ofGDP, Taxesonproductionandimports(%ofGDP),

HICP, Euribor3monthrate and Realeffectiveexchangerate. Expected is that HICP will lead to a higher

CDS spread as discussed by Aizenman et al. (2013). A higher Imports%ofGDP is expected to lower the CDS spread of a country since the country has more open economy.

Taxesonproductionandimports(%ofGDP) is also expected to lower the CDS spreads since higher tax

income improves the government budget. The Euribor3monthrate is expected to have a negative effect on the CDS spreads. Realeffectiveexchangerate is also expected to have a negative effect on the CDS spreads.

The In order to find out which regression model suits the data best, I preformed the Hausman test. The Hausman did not reject the null hypothesis therefore the preferred model is the random effects model. All the regressions are therefore performed with the random effects model. The difference between a fixed effects model and a random effects model is that a fixed effects model assumes that the individual specific effect is correlated to the independent variable and a random effects model assumes the individual specific effects are uncorrelated with the independent variables (Fixed and Random effects models, 2016).

In the appendix a fixed effects regression model with the internal and external variables can be found. The direction of the variables does not change when using a fixed effects model. There are however some differences between the fixed effects model and the random effects model. In the fixed effects model the values of the coefficients and standard errors are in some cases higher and in other cases lower compared to the random effects model. Furthermore, the variable

Realeffectiveexchangerates is in the fixed effects model significant at the 5% level. The variable Externaldebt(%ofGDP) is in the fixed effects model significant at the 5% level instead of 1% in the

10

Table 1

Definition of variables

VARIABLES Unit of Measurement Description

CDS spread Number Sovereign credit default swaps of

Euro area countries with a 5 year maturity denominated in US dollars.

VIX Number Volatility index that shows the

market expectations of the upcoming 30 days volatility of S&P 500 companies.

Treasury rate Number The three month Treasury rate

TED spread Number The difference between the

month US$ LIBOR and the 3-month US treasury bill (%)

External debt (% of GDP) Percentage The amount of external debt as a

percentage of GDP denominated in US dollars.

Official reserve assets (% of GDP) Percentage The amount of official reserve

assets as a percentage of GDP denominated in US dollars.

Imports % of GDP Percentage The amount of imports as a

percentage of GDP denominated in Euros.

Taxes on production and imports (% of GDP) Percentage The taxes on production and

imports as a percentage of GDP denominated in Euros.

HICP Quarterly change (%) The harmonized index of

consumer prices. An indicator of inflation and price stability.

Euribor 3 month rate Number The average interest rate of

interbank lending of European banks on 3 month loans.

Real effective exchange rate Index The weighted average of a

12

13 Figure 1 shows the graphs of the CDS spread of the countries used in the dataset. The spreads of most countries show a similar pattern with two peaks in the CDS spread. The first peak occurs somewhere around the last quarter of 2008 and the second is around the first quarter of 2012. Although, most of the countries show a similar pattern in the CDS spread, the height of the spreads differ a lot per country. The graphs make clear that one country is special, namely Greece. After the second quarter of 2011 the CDS spread of Greece goes through the roof and goes above 37000 basis points. After the fourth quarter of 2014 the CDS spread of Greece sharply drops. I expect that the spread of Greece will have a strong impact on the analysis therefore I will provide summary statistics and a correlation matrix with and without Greece in order to decide to include Greece in the analysis.

Table 2

Summary statistics with Greece

VARIABLES N mean sd min max

CDS spread 495 1,032 5,458 5.450 37,030

VIX 510 22.03 7.638 12.75 45.15

Treasury rate 510 0.293 0.651 -0.0100 3.290

TED spread 510 47.24 40.47 19 173

External debt (% of GDP) 472 1,304 1,174 182.4 4,746

Official reserve assets (% of GDP) 463 17.83 8.665 2.438 45.78

Imports (% of GDP) 495 58.09 31.88 22.30 163.8

Taxes on production and imports (% of GDP) 495 13.09 1.978 7.100 19.80

HICP 499 1.641 1.627 -2.800 6.500

Euribor 3 month rate 510 1.137 1.518 -0.240 5.280

Real effective exchange rate 508 99.87 3.583 88.58 113.9

14 Table 3 shows the summary statistics without Greece. The summary statistics of the dependent variable are completely different when Greece is not included. The mean drops from 1032 to 161.9 basis points and the max value is now 1672 instead of 37020. The independent variables show little changes when Greece is not included. The variables VIX, Treasuryrate, TEDspread and

Euribor3monthrate do not change; these variables are the same for every country.

Correlation matrices

The next page shows the correlation matrix of the data with Greece (table 5) and the correlation matrix of the data without Greece (Table 6). The direction of the correlation some of the

independent variables have with the dependent variable in the correlation matrix with Greece is not in line with the literature. An example is the variable HICP that shows a negative relation with

CDSspreads when Greece is included. However previous literature finds that a higher inflation leads

to a higher CDS spread. When looking at the correlation matrix without Greece than the relation between HICP and CDSspread is positive. Other examples are Externaldebt(%ofGDP) and VIX these two variables have a negative relation with the dependent variable in the correlation matrix with Greece in table 5. In the correlation table without Greece both these variables have a positive relation with the dependent variable CDS spread. Based on the graphs, summary statistics and the correlations matrices I decide not to include Greece in the regressions.

Table 3

Summary statistics without Greece

VARIABLES N mean sd min max

CDS spread 461 161.9 219.6 5.450 1,672

VIX 476 22.03 7.638 12.75 45.15

Treasury rate 476 0.293 0.651 -0.0100 3.290

Ted spread 476 47.24 40.47 19 173

External debt (% of GDP) 440 1,341 1,207 182.4 4,746

Official reserve assets (% of GDP) 433 18.45 8.582 2.438 45.78

Imports (% of GDP) 462 59.91 32.23 22.30 163.8

Taxes on production and imports (% of GDP) 462 13.04 1.986 7.100 18.40

HICP 466 1.654 1.560 -2.800 6.500

Euribor 3 month rate 476 1.137 1.518 -0.240 5.280

15 The correlation matrix without Greece still shows multicollinearity between some of the independent variables. The Euribor3monthrate shows multicollinearity with the Treasuryrate and the TEDspread. Also the variable Externaldebt(%ofGDP) shows multicollinearity with Import(%ofGDP).

The regression analysis of this research structured as follows. First, the internal factors are analyzed in a random effects model and the control variables are added stepwise(table 7). Secondly, the external factors are analyzed in a random effects model with the control variables added

16

Table 5

Correlation matrix with Greece

CDS spread VIX Treasury rate TED spread External debt (% of GDP) Official reserve assets (% of GDP) Imports as % of GDP Taxes on production and imports (% of GDP) HICP Euribor 3 month rate Real effective exchange rate CDS spread 1.0000 VIX -0.0863 1.0000 Treasury rate -0.0482 0.1613 1.0000 TED spread -0.0601 0.6613 0.5331 1.0000 External debt (% of GDP) -0.0352 -0.1191 -0.1080 -0.1507 1.0000

Official reserve assets (% of GDP) -0.0996 -0.1842 -0.1667 -0.2193 -0.0163 1.0000

Imports (% of GDP) -0.1060 -0.1424 -0.0830 -0.1492 0.7506 -0.0903 1.0000

Taxes on production and imports (% of GDP) 0.0984 -0.1172 -0.0004 -0.0532 -0.1356 0.3237 -0.1677 1.0000

HICP -0.1228 0.2790 0.3341 0.3509 -0.1544 0.0346 0.0073 0.0101 1.0000

Euribor 3 month rate -0.0718 0.5248 0.7676 0.7930 -0.1396 -0.2177 -0.1193 -0.0629 0.5588 1.0000

Real effective exchange rate -0.1621 0.4289 0.3139 0.4176 -0.2457 -0.1332 -0.1485 -0.1498 0.1863 0.4788 1.0000

Table 6

Correlation matrix without Greece

CDS spread VIX Treasury rate TED spread External debt (% of GDP) Official reserve assets (% of GDP) Imports as % of GDP Taxes on production and imports (% of GDP) HICP Euribor 3 month rate Real effective exchange rate CDS spread 1.0000 VIX 0.0269 1.0000 Treasury rate -0.1602 0.1682 1.0000 TED spread -0.0959 0.6627 0.5374 1.0000 External debt (% of GDP) 0.2288 -0.1135 -0.0994 -0.1452 1.0000

Official reserve assets (% of GDP) 0.1266 -0.1689 -0.1379 -0.2006 -0.0535 1.0000

Imports (% of GDP) 0.0640 -0.1401 -0.0777 -0.1493 0.7492 -0.1600 1.0000

Taxes on production and imports (% of GDP) 0.0651 -0.1029 0.0045 -0.0400 -0.1358 0.3533 -0.1642 1.0000

HICP 0.1658 0.2597 0.3204 0.3454 -0.1535 0.0782 0.0165 0.0482 1.0000

Euribor 3 month rate -0.0753 0.5323 0.7657 0.7933 -0.1302 -0.1915 -0.1157 -0.0528 0.5483 1.0000

17

Regression analysis

This table shows the regression analysis using the random effects model with the internal factors, the control variables are added stepwise. The dependent variable are CDS spreads with a 5 year maturity denominated in US dollars. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

Table 7

Random effects model – Internal factors

VARIABLES

(1)

(2)

(3)

(4)

(5)

External debt (% of GDP)

0.0597*

0.0578*

0.0808***

0.0870***

0.0885***

(0.0313)

(0.0317)

(0.0307)

(0.0312)

(0.0304)

Official reserve assets (% of GDP)

11.10***

11.68***

11.31***

6.771***

8.168***

(1.972)

(1.994)

(1.887)

(2.477)

(2.484)

Taxes on production and imports (% of GDP)

-14.04*

-6.431

-6.977

-4.189

(8.174)

(7.804)

(7.746)

(7.703)

HICP

40.09***

54.50***

59.02***

(5.704)

(7.570)

(7.620)

Euribor 3 month rate

-30.58***

-43.44***

(10.73)

(11.35)

Real effective exchange rates

10.71***

(3.408)

Constant

-104.5

69.80

-114.0

-26.41

-1,154***

(70.69)

(124.5)

(121.2)

(125.3)

(379.7)

Observations

400

400

399

399

399

18 The results of the first regression analysis are shown in table 7. In the first model the two

independent variables Externaldebt(%ofGDP) and Officialreserveassets(%ofGDP) are added and both variables show a positive significant relation with the dependent variable. The control variables are stepwise added to the regression. The fifth model is the model with all the control variables added. The variable Imports(%ofGDP) is left out of the regression due to multicollinearity with the variable

Externaldebt(%ofGDP). The variable Externaldebt(%ofGDP)has as expected a positive relation with

the dependent variable. Meaning that a higher debt to GDP ratio will lead to a higher CDS spread of a sovereign country. The variable Officialreserveassets(%ofGDP)also has a positive relation with the dependent variable. However, this is not in line with the expectations. Aizenman and Pasricha (2009) found a negative relation between the reserves of country and its CDS spread. Their reasoning behind the negative relation is that countries with additional reserves may allow a country to avoid costly liquidations and therefore lower the costs of the probability of default. The variable

Taxesonproductionandimports(%ofGDP) is not significant with the dependent variable. The control

variable HICP is in line with the expectations. Previous studies such as Aizenman et al. (2013) also found that higher inflation leads to higher CDS spreads. The Euribor3monthrateis negatively correlated with the depended variable this also in line with the expectations. The

Realeffectiveexchangerates is positively correlated with the dependent variable. Most of the

19

Table 8

Random effects model – External factors

VARIABLES

(1)

(2)

(3)

(4)

(5)

VIX

3.700**

3.437**

3.043*

1.893

2.458

(1.521)

(1.570)

(1.584)

(1.489)

(1.522)

Treasury rate

-76.79***

-77.98***

-75.41***

-117.5***

-113.3***

(20.00)

(20.28)

(20.28)

(20.36)

(20.44)

TED spread

-0.699**

-0.661*

-0.639*

-0.988***

-0.903***

(0.346)

(0.351)

(0.350)

(0.333)

(0.335)

Imports % of GDP

-0.0715

-0.191

-1.184

-1.568*

(0.934)

(0.972)

(0.873)

(0.912)

Taxes on production and imports % of GDP

-12.40

-10.34

-11.58

(7.643)

(7.091)

(7.123)

HICP

49.09***

48.63***

(6.036)

(6.026)

Real effective exchange rates

-4.848*

(2.770)

Constant

127.8***

138.1*

314.1**

318.6***

825.8***

(41.47)

(74.35)

(131.9)

(120.8)

(309.7)

Observations

461

447

447

446

446

Number of Countries

14

14

14

14

14

20 The results of the second regression analysis are shown in table 8. Table 8 shows the regression with the external factors VIX, Treasuryrate and TEDspread. In the first model of table 8 all three variables are significant at the 5% and 1% level. The relation of the variable VIX with the dependent variable

CDSspread is positive. The variables Treasuryrate and TEDspread are both negative related to the

dependent variable. The control variables are again stepwise added to the regression. The variable

Euribor3monthrate is left out of the regression due to multicollinearity with the variables

Treasuryrate and TEDspread. When all the control variables are included the variable VIX is no longer

significant. In the model with all the control variables the variables Treasuryrate and TEDspread are significant at the 1% level. The variable Imports(%ofGDP) is significant at the 10% level and negatively related to the dependent variable. This is in line with the expectations, namely that more trade openness leads to a lower CDS spread. The variable Taxesonproductionandimports(%ofGDP) is again not significant. HICP has again a positive coefficient and is highly significant. The variable

Realeffectiveexchangerates shows in this model with the external factors a negative sign and is

21

Table 9

Random effects model – External and internal factors

VARIABLES

(1)

(2)

(3)

(4)

VIX

5.803***

5.636***

4.055***

3.495**

(1.560)

(1.572)

(1.506)

(1.542)

Treasury rate

-36.87

-36.07

-88.95***

-95.08***

(24.77)

(24.71)

(26.46)

(26.73)

TED spread

0.00649

0.0358

-0.376

-0.378

(0.381)

(0.380)

(0.366)

(0.364)

External debt % of GDP

0.0606**

0.0599**

0.0748***

0.0786***

(0.0259)

(0.0272)

(0.0246)

(0.0259)

Official reserve assets % of GDP

12.80***

13.39***

7.947***

9.095***

(2.254)

(2.292)

(2.268)

(2.357)

Taxes on production and imports % of GDP

-7.361

-1.400

-0.492

(7.993)

(7.548)

(7.590)

HICP

45.82***

46.48***

(6.347)

(6.337)

Real effective exchange rates

5.400

(3.334)

Constant

-256.2***

-168.2

-176.2

-740.5**

(75.95)

(131.4)

(122.6)

(365.2)

Observations

400

400

399

399

Number of Countries

14

14

14

14

22 The final regression model is shown in table 9. This regression model includes both the internal factors as well as the external factors. Again, the control variables are added stepwise to the regression. The final model includes all independent variables and all the control variables except

Euribor3monthrate and Imports(% of GDP) due to multicollinearity. In the final model the VIX is

significant at the 5% level. The variables Externaldebt(%ofGDP), Officialreserveassets(%ofGDP) and

Treasuryrate are significant at the 1% level. The control variable HICP is also a significant at the 1%

level. The variable TEDspread is no longer significant. Also the control variables

Taxesonproductionandimports(%ofGDP) and Realeffectiveexchangerates are not significant in the

final model. From the analyses I can conclude that the first hypothesis can be accepted. The

Eurozone credit default swap market is influenced by internal factors such as the fiscal condition and reserve position. The variables for the first hypothesis are both highly significant. The second

hypothesis can also be accepted namely that there is a spillover effect from external factors from US such as the treasury rate, VIX index and TED spread towards the Eurozone sovereign credit default swap market. The variables for the second hypothesis are not all significant in the final model only the variable VIX and Treasuryrate are significant and the variable TEDspread is not. The findings of this research are in line with the findings of other researches done towards the pricing of sovereign CDS. Namely, the research of Aizenman et al. (2013) found that the market priced risk of sovereign debt as measured by CDS spreads is partly explained by fiscal space and other economic

determinants. Furthermore, Longstaff et al. (2011) already found that the CDS spread of sovereign countries was heavily influenced by external factors especially by US financial factors.

Conclusion

This master thesis looks at the pricing of sovereign risk of 14 Eurozone countries. By analyzing the credit default swap spreads of these countries between the period of the 4th Quarter of 2007 and the 1st quarter of 2016. The CDS spreads of the Euro area countries really started to peak around the beginning of 2011.The main focus of this research is to determine if the CDS spread is determent by internal or external factors or by both. For the internal factors I looked at the external debt as percentage of GDP and the amount of official reserve assets as percentage of GDP. For the external factors I looked at the VIX index, the US treasury rate and the TED spread. The findings of this research are that the market priced risk of Euro area countries sovereign debt, as measured by CDS spreads is explained by internal as well as external factors. Most of the findings are in line with other researches done towards the pricing of sovereign risk. However the variable

Officialreserveassets(%ofGDP) has a different direction, in the outcome of this research higher

24

References

Anderson, R. W. and K. McKay (2008), Derivatives Markets, in : X. Freixas, P. Hartmann, and C. Mayer (eds.), Handbook of European Financial Markets and Institutions, Oxford University Press, 568-596

Aizenman, J., Hutchison, M., & Jinjarak, Y. (2013). What is the risk of European sovereign debt defaults? Fiscal space, CDS spreads and market pricing of risk. Journal of International Money and Finance, 34, 37-59.

Aizenman, J., & Pasricha, G. K. (2010). Selective swap arrangements and the global financial crisis: Analysis and interpretation. International Review of Economics & Finance, 19(3), 353-365.

Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity.Journal of econometrics, 31(3), 307-327.

Booth, G. G., So, R. W., & Tse, Y. (1999). Price discovery in the German equity index derivatives markets. Journal of Futures Markets, 19(6), 619-643.

Coudert, V., & Gex, M. (2010). Credit default swap and bond markets: which leads the other. Financial Stability Review, Banque de France, 14(2010), 161167.

De Haan, J., Oosterloo, S., & Schoenmaker, D. (2015). Financial Markets and Institutions: A European Perspective. Cambridge University Press.

Delatte, A. L., Gex, M., & López-Villavicencio, A. (2012). Has the CDS market influenced the

borrowing cost of European countries during the sovereign crisis?. Journal of International Money and Finance, 31(3), 481-497.

Duffie, D., Filipović, D., & Schachermayer, W. (2003). Affine processes and applications in finance. Annals of applied probability, 984-1053.

Exchange, C. B. O. (2015). The CBOE volatility index-VIX. White Paper, 1-23.

Fender, I., Hayo, B., & Neuenkirch, M. (2012). Daily pricing of emerging market sovereign CDS before and during the global financial crisis. Journal of Banking & Finance, 36(10), 2786-2794.

Fixed and Random effects models, 2016 June 10, Retrieved from:

http://courses.washington.edu/pbafadv/student%20presentations/Fixed%20Effects,%20Random%20E ffects%20Model%20Cheat%20Sheet.pdf

Fontana, A., & Scheicher, M. (2016). An analysis of euro area sovereign CDS and their relation with government bonds. Journal of Banking & Finance, 62, 126-140.

Guill, G. D. (2009). Bankers Trust and the birth of modern risk management. Wharton School, University of Pennsylvania, April.

Heller, P. (2005). Back to Basics-Fiscal Space: What it is and how to get it. Finance and Development-English Edition, 42(2), 32-33.

Longstaff, F. A., Pan, J., Pedersen, L. H., & Singleton, K. J.. (2011). How Sovereign Is Sovereign Credit Risk? American Economic Journal: Macroeconomics, 3(2), 75–103.

Pan, J., & Singleton, K. J. (2008). Default and recovery implicit in the term structure of sovereign CDS spreads. The Journal of Finance, 63(5), 2345-2384.

25

Appendix

Table 10

Fixed effects model – External and internal factors

VARIABLES

(1)

(2)

(3)

(4)

VIX

6.262***

6.027***

4.323***

3.661**

(1.536)

(1.558)

(1.494)

(1.524)

Treasury rate

-25.67

-26.42

-81.03***

-87.64***

(24.60)

(24.62)

(26.72)

(26.82)

TED spread

0.159

0.169

-0.280

-0.282

(0.376)

(0.377)

(0.362)

(0.360)

External debt % of GDP

0.0690

0.0687

0.140**

0.129**

(0.0569)

(0.0569)

(0.0546)

(0.0546)

Official reserve assets % of GDP

16.80***

16.86***

10.18***

11.78***

(2.575)

(2.576)

(2.644)

(2.751)

Taxes on production and imports % of GDP

-7.624

-1.910

-0.329

(8.359)

(7.917)

(7.924)

HICP

45.58***

46.01***

(6.408)

(6.386)

Real effective exchange rates

6.765**

(3.354)

Constant

-369.0***

-265.6*

-313.8**

-1,010***

(92.25)

(146.1)

(138.4)

(371.8)

Observations

400

400

399

399

R-squared

0.163

0.164

0.262

0.270

Number of countries

Fixed effects

14

YES

14

YES

14

YES

14

YES

26 The fixed effects model is shown in table 10. This regression model includes both the internal factors as well as the external factors. The control variables are added stepwise to the regression. The final model includes all independent variables and all the control variables except Euribor3monthrate and

Imports(% of GDP) due to multicollinearity. In the final model the VIX is significant at the 5% level.

The variables Officialreserveassets(%ofGDP) and Treasuryrate are significant at the 1% level. The control variable HICP is also a significant at the 1% level. The variable Externaldebt(%ofGDP) is significant at the 5% level. Also the control variable Realeffectiveexchangerates is significant at the 5% level stating that a higher real effective exchange rate leads to higher CDS spreads. The variable

TEDspread is not significant. Also the control variable Taxesonproductionandimports(%ofGDP) is not