Sovereign Credit Rating Contagion in the EU

Available online at www.sciencedirect.com

ScienceDirect Procedia Economics and Finance 24 (2015) 218 – 227

International Conference on Applied Economics, ICOAE 2015, 2-4 July 2015, Kazan, Russia

Sovereign credit rating contagion in the EU Leila Fouriea, Ilse Botha b* a

Department of Economics and Econometrics, University of Johannesburg,PO Box 524, Aucklandpark, 2006 Department of Finance and Investment management, University of Johannesburg,PO Box 524, Aucklandpark, 2006

b

Abstract The recent persistent and synchronised deterioration in the euro zone had severe consequences for the euro community, the effects of which have been felt by the global community. This study proved that sovereign rating contagion existed between euro countries during the two recent windows of crises, namely the Lehman and sovereign debt crisis. Compelling evidence from the analysis provided a clear indication of contagion during the two periods of crisis. Results indicated a higher vulnerability to shocks and a higher degree of connection during the windows of crises than during the tranquil periods. Notable was that the European Union (EU) sovereign debt crisis experienced a more pronounced degree of contagion than the Lehman crisis period did. During the sovereign debt crisis window, a dominant theme was the highly integrated connection between the Portugal, Italy, Greece and Spain (PIGS) group of countries. © This is an open access article under the CC BY-NC-ND license © 2015 2015 The TheAuthors. Authors.Published PublishedbybyElsevier ElsevierB.V. B.V. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Selection and/or peer-review under responsibility of the Organizing Committee of ICOAE 2015. Selection and/or peer-review under responsibility of the Organizing Committee of ICOAE 2015. Keywords: : Sovereign credit ratings; contagion; cointegration; causality; EU countries; PIGS; variance decomposition

1. Introduction The persistent and synchronised deterioration of the economic health in the euro countries since the global financial crisis has had far reaching economic and social consequences that will create a drag on the economic growth of these nations for generations to come (Stracca 2013:5[1]; Caporin et al. 2012:2[2]; Gentile and Giordano 2012:1[3]). Contagion is not a new phenomenon. In his speech on sovereign contagion in Europe, European Central

* Corresponding author. Tel.: +27 11 559 2056; fax: +27 11 559 3753. E-mail address: [email protected]

2212-5671 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Selection and/or peer-review under responsibility of the Organizing Committee of ICOAE 2015. doi:10.1016/S2212-5671(15)00651-6

Leila Fourie and Ilse Botha / Procedia Economics and Finance 24 (2015) 218 – 227

219

Bank (ECB) board member Gonzalez-Paramo (2011:1[4]) points out that reference to contagion were made as early as 1871 by Walter Bagehot in the ‘Lombard Street’ speech around the panic caused by the failure of Overend, Gurney and Co in 1866. According to Constancio (2012:110 [5]), the banking crisis that was triggered in August 2007 reached a climax in September 2008 when Lehman defaulted. This culminated in a sovereign crisis in the spring of 2010. The sovereign crisis was triggered by the announcement of a referendum in Greece in November 2010, following the decision to consider a rescue program at an EU summit the previous week (Gonzalez-Paramo 2011:1[4]). Although Germany and France normalised soon after the announcement, this triggered the start of a persistent decline in Italian and Spanish sovereign debt. In addition to the weakening in government bond spreads, a synchronised deterioration in sovereign ratings of the euro zone as well as numerous large bank and corporate rating downgrades ensued. Aside from the dislocation experienced in the euro zone during the sovereign debt crisis, it should be noted that the integration effect of the EU increases the interconnectedness and therefore exacerbates the negative effects of the instability caused by the Greek rescue package. This integration effect has been the subject of numerous studies (Bekaert et al. 2010:1[6]; Beber et al. 2009:1[7]; Gomez-Puig 2008:455[8]; Borensztein et al. 2007:1[9]; Rowland and Torres 2004:3[10]; Amira 2004:1[11]). The synchronised decline in the health of sovereigns in the euro zone and the deterioration of the global economy prompts a number of important questions. What is the nature of the connection between euro countries and their sovereign ratings? Does this relationship change during periods of crisis? This research investigates the presence and directionality of contagion between countries sovereign credit ratings in the euro zone. The remainder of the paper is structured as follows, in section two the literature review will be discussed, in section three the data and method will be discussed briefly, section four present the empirical findings and the discussion of the results and section five conclude this paper. 2. Literature review Gentile and Giordano (2012:8[3]) describe “contagion” as “the amount of co-movement among asset prices which exceeds what is explained by fundamentals”. These authors argue that a degree of extreme connection or asymmetry that goes beyond interdependencies must be present in order for contagion to be present. Research on contagion range from conditional correlation to contagion of bond spreads, equity market stock returns, differences in interest rates, monetary policy and currency market variance (Gentile and Giordano 2012:8[3]; Andenmatten and Brill 2011:1[12]; Saghaian 2010:1[13]; Pontines and Siregar 2007:1[14]; Campbell et al. 2006:2[15]; Suleimann 2003:1[16]; Forbes and Rigobon 2002:2225[17]; Butler and Joaqin 2002:981[18]; Ang and Chen 2002:444[19]; Longin and Solnik 2001:4[20]; and Favero and Giavazzi 2000:3[21}, Beirne and Gieck (2012:1[22]). A wide range of empirical techniques has been used to quantify correlation and contagion. Dungey et al. (2004:4[23]) note that tools to measure contagion range from correlation and covariance analysis undertaken by, for example, Forbes and Rigobon (2002:2225[17]) to probability models by, for example, Eichengreen, Rose and Wyplosz (1995:250[24]) to latent factor structure by Dungey and Martin (2001:2[25]), principle components by Kaminsky and Reinhart (1999:474[26]), multiple equilibria and finally VAR. Vector Autoregression (VAR) cointegration and VECM / Granger causality tests have been used to measure contagion in numerous studies. Beirne and Gieck (2012:1[22]) use a global VAR to measure interdependence and contagion across bonds, stocks and currencies for over 60 economies during periods of crisis. These authors’ analysis reveals that shocks to equity markets typically originate in the US and that bond market shocks tend to originate in the euro zone. Gentile and Giordano (2012:7[3]) use cointegration and VECM / Granger causality tests to measure the existence and direction of contagion in European countries during the Lehman default and sovereign debt crisis. Gentile and Giordano (2012:7[3]) use sovereign bond spreads and stock returns as a country risk indicator. Andenmatten and Brill (2011:1[12]) use a bi-variate test to prove contagion in emerging and industrialised countries using credit default swap (CDS) spreads during the Greek debt crisis of 2009. Saghaian (2010:1[13]) uses Granger causality to create contemporaneous contagion links between agriculture and energy markets within the commodity sector. Suleimann (2003:1[16]) uses VAR to measure contagion in technology stock prices between Europe and the United States. Pontines and Siregar (2007:1[14]) use Markov Switching and VAR to measure contagion in East Asian markets using stock exchange returns during periods of turbulence. Favero and Giavazzi (2000:3[21]) use VAR to test contagion using money market spreads across European Monetary Union countries.

220

Leila Fourie and Ilse Botha / Procedia Economics and Finance 24 (2015) 218 – 227

Baig and Goldfajn (1999:168[28]) use dummy variables in VAR and correlation models to prove contagion during the East Asian crisis. Deaton and Miller (1995:10) use VAR to connect the relationship between commodity prices and the GDP growth of African countries. In this paper, the Gentile and Giordano (2012:8[3]) approach will be adapted by measuring the presence of contagion transmitted by sovereign ratings during windows of distress and tranquility. While research measuring contagion between countries exists, the variables that are included in the analysis do not include a composite variable that captures a multi-dimensional view of the economy such as sovereign ratings. The literature reviewed above indicates that contagion research is typically limited to variables such as equity prices and bond spreads. This narrow focus on isolated variables fails to provide a composite connection between crosscountry macroeconomics. Sovereign ratings use a composite of variables and therefore provide a broad-based indicator of macroeconomic health. Various potential predictive variables that could provoke a sovereign default event through the economic cycle range from political, fiscal and monetary to those exogenous to the general macro economy. The type and intensity of variables work together through the cycle to contribute to a specific type of crisis, ranging from local to external debt, interest rate, currency and financial crises. A sovereign rating default is a common trigger that results from deterioration in the predictive factors and gives rise to a spillover event that could culminate in contagion (Cecchetti, Mohanty and Zampolli 2010:1[29], Reinhart and Rogoff 2009:1[30]; Kindleberger and Aliber 2005:1[31]; Manasse and Roubini 2005:4[32]; Rowland and Torres 2004:3[10]; Manasse, Roubini and Schimmelpfennig 2003:4[33]; Mellios and Paget-Blanc 2003:1[34]; and Reinhart 2002a:2[35]). 3. Data and methodology The model tests for contagion based on the connection between sovereign ratings in eight countries, namely France, Greece, Hungary, Ireland, Italy, Poland, Portugal and Spain. The model uses S&P sovereign ratings recorded on a monthly basis between 2004 and 2013. This period of analysis is used because it covers important periods of distress for the euro area, characterised by trigger default events for the euro area. Constancio (2012:110[5]) and Caporin et al. (2012:3[2]) maintains that a contagious event cannot occur in the absence of a shock, a large shock should occur. Gentile and Giordano (2012:8]3]) identify contagion windows for the euro-contagion analysis based on the Lehman default and sovereign debt crisis. Other researchers using similar windows include studies by Andenmatten and Brill (2011:1[27]), Fofana and Seyte (2012:1[12]) and Caballero, Farhi and Gourinchas (2008:1[27]). The Argentine crisis or the Asian crisis were not possible to model since the sovereign ratings of the euro countries displayed very little movement in their ratings. The windows for this study are the tranquil period between 2004 and 2008, the credit crisis of 2007 triggered by the Lehman default and the euro zone sovereign debt crisis triggered by the Greek sovereign debt restructure in December 2010. The sovereign credit ratings are mapped to a 21-point scale. This is based on an adaptation of the mapping used by Reinhart (2002b:10[37]), Rowland (2004:26[38]), and Cantor and Packer (1996:78[39]) whereby a linear transformation is used to assign numerical values to credit ratings from highest to lowest credit rating. Non-linear transformations were avoided since non-linear transformation found not materially different results (Afonso, Gomes and Rother (2007:23[40]). The model followed a three-step approach, similar to that used by Beirne and Gieck (2012:1[22]), Gentile and Giordano (2012:8[3]), and Baig and Goldfajn (1999:168[28]). Each step compared results between periods of tranquillity and periods of distress. The first step involved analysis of long run equilibrium relations using bi-variate dynamic cointegration. The second step applied Granger causality tests to measure the number of short run relations that existed between countries during each period of time. Lastly, the innovation accounting technique, variance decomposition was applied. Results from each of the three steps were compared for each window of time to evaluate the prevalence and nature of sovereign rating contagion in the Euro area; therefore determining whether the connection between countries was higher or asymmetrical during crisis periods; exceeding the movement caused by fundamentals. 4. Empirical findings and discussion of results Stationarity testing with the Augmented Dickey Fuller and Phillips Peron tests confirmed that all the series were I(1). The optimal lag length, using the Schwartz information criterion (SC) and Hannan-Quin criterion (HQ), was

221

Leila Fourie and Ilse Botha / Procedia Economics and Finance 24 (2015) 218 – 227

one. Using a lag of one period, the Johansen’s cointegration test was applied to determine pairwise long-run relations during each time window. 4.1. Pinpointing contagion windows using pairwise cointegration Pairwise cointegration test results are recorded for the tranquil window between May 2004 and August 2008. The cointegration tests measure the number of long run relations that exist between sovereign ratings of countries. The long run measure is particularly pertinent since sovereign ratings are expected to have a stronger long-run relationship than a short-run relationship. Studies that use variables that respond more rapidly to shocks, on the contrary, such as, for example, equity or bond prices, are likely to display more short-run relations than long-run. Table 1 indicates four long run relations, mainly involving Portugal, Italy, Greece, Spain (PIGS) and Hungary. The relatively higher number of PIGS relationships indicate a higher level of integration during tranquil times and a fundamental connection. This connection becomes asymmetrical during periods of distress as is borne out by a materially higher number of cointegrating relations during the contagion windows displayed in Table 2 and 3. France does not enter into the cointegrating equation due to this country’s strong fundamentals and therefore a high credit rating during the tranquil period. Table 1: Cointegration during tranquil window: May 2004 – August 2008 France Greece France Greece Hungary 0 Ireland Italy 1 Poland 0 Portugal 1 Spain Total Source: Eviews estimates

Hungary

Ireland

Italy

Poland

Portugal

Spain

Total

1 0 0

0 0 0

1 4

The Lehman and sovereign debt contagion windows contain cointegration results in Table 2 and 3 respectively that reflect seven and twelve cointegrating relations respectively. This is far higher than the four relations reflected in the tranquil window. The cointegration results in Table 2 and 3 strongly support the idea that contagion involves externalities and is “distinct because it reflects a market failure and a dangerously amplified transmission of instability” (GonzalezParamo 2011:1[4]). Moreover, the results support Constancio’s (2012:110[5]) view that the spread of instability is abnormal and amplified, that it goes beyond the bounds of normality. These results support the findings by Gentile and Giordano (2012:1[3]) that prove the existence of contagion in the euro countries during the Lehman and sovereign debt crises. Moreover, the results are congruent with those by Andenmatten and Brill (2011:1[12]) and Gray (2009:1[41]), both of whom prove contagion using bond spreads and equity prices during the recent global financial crisis. This study expands upon the contagion results of previous euro studies by proving that contagion can be expanded to sovereign credit ratings. The PIGS countries feature sharply during both contagion windows, with Greece reflecting more than double the number of cointegrating relations during the Lehman window as compared to the tranquil period. The leading position that Greece takes is not surprising given the deterioration of Greece’s economy during this period. France features more significantly during the euro debt crisis window, indicating a higher degree of integration with other euro countries during the sovereign debt crisis. Moreover, the fundamentals of France and the health of its banks were seriously compromised by the exposure of this country to the sovereign debt problems being experienced by the other euro countries.

222

Leila Fourie and Ilse Botha / Procedia Economics and Finance 24 (2015) 218 – 227

Table 2: Cointegration during the Lehman contagion window: September 2008 – December 2009 France Greece France Greece Hungary 1 Ireland 1 Italy Poland Portugal 1 Spain 1 Total Source: Eviews estimates

Hungary

Ireland

Italy

Poland

Portugal

Spain

Total

1 1 1 7

The more meaningful cointegration results are indicated by a higher number of cointegrating relations during the sovereign debt crisis. This indicates a more widespread contagion effect in the number of countries involved, and also involves the downgrade of France, a significant event. The close proximity of these two windows of contagion and the connection with a banking panic begs the question of how long the negative effects of the protracted downturn might last. Table 3: Cointegration during the sovereign debt contagion window: December 2010 – January 2013 France Greece France Greece Hungary Ireland Italy Poland Portugal Spain Total Source: Eviews estimates

Hungary

Ireland 1 1

Italy 1 1 1 1

Poland

Portugal 1 1 1 1

Spain

Total

1

1 12

In summary, the results indicating the abnormally high presence of long run relations in the contagion windows relative to the tranquil window are supportive of the initial notion that the scale and reach of the recent crises has created an externalised impact that caused much higher connectivity or contagion during the crisis periods than during tranquil periods. Moreover, it is notable that this high degree of contagion is relevant during a period that coincides with a banking crisis. The banking crisis that was triggered during the Lehman window continued during the euro debt crisis window because many euro banks were downgraded due to their exposures to sovereign debt issued by troubled countries. These results support findings by Reinhart and Rogoff (2008:1[30]) and Bernanke and Lown (1991:206[42]) that a banking panic is typically associated with a more disorderly and protracted disturbance. This cross-market and cross-border effect makes the transmission effect of shocks complex and poses a seriously multifaceted problem for policy makers. According to Dungey, Fry, Gonzalez-Hermosillo and Martin (2011:192), a complex problem that this situation presents is that the weight of each contagion channel will vary and the impact of policy response will not be linear. Policy makers will therefore need to develop a combination of micro-structure and country-specific policies while also contributing to the macro, cross-jurisdictional measures (Dungey et al. 2011:192). Co-ordinated policy response by countries during the crisis as well as the active ECB and US FED stimulus packages will have contributed to an increased level of integration and co-movement between the countries during this period.

Leila Fourie and Ilse Botha / Procedia Economics and Finance 24 (2015) 218 – 227

4.2. Leaders and followers defined by Granger causality results Granger causality tests were applied to each of the three windows. The previous section presented evidence of contagion in the euro area, based on cointegration tests that indicated long run relations between countries during each window. The results presented in this section, by contrast, provide an indication of the short run relations that exist between the countries during each time period. Short run relations are identified as those countries that reject the null hypothesis at a 95 percent confidence interval. Gentile and Giordano (2012:9[3]) maintain that the Granger causality tests indicate which variables, in our case country sovereign ratings, take on a leader and a follower role. Leaders will be those countries that shape the transmission of short run relations to other sovereign ratings. Follower countries are those that are vulnerable or receptive to innovations or shocks in sovereign ratings. During the first window, that is the tranquil window between 2004 and 2008, in Table 4, only three short run relations are evident. During this window, Hungary causes Italy, Italy causes Poland and Spain causes Portugal. Hungary, Italy and Spain therefore take on leading roles during this period and Italy, Poland and Portugal are respective followers. These follower countries indicate their growth vulnerability. These results are consistent with findings from Gentile and Giordano (2012:32[3]). Table 4: Granger results for the tranquil window Null Hypothesis: F-Statistic PHUNGARY does not Granger Cause 16.091 *0.000 ITALY ITALY does not Granger Cause POLAND 4.398 *0.018 SPAIN does not Granger Cause 3.281 *0.047 PORTUGAL *Reject null hypothesis at 95% confidence level Source: Eviews estimates The Lehman window of contagion in Table 5 indicates five short-run relations between the countries. During this period, Hungary, Portugal and Spain take on lead roles. Ireland is a follower country in most of these relations. This follower status shows that Ireland was particularly vulnerable during this period. This vulnerability is consistent with the distress that this country experienced when Ireland negotiated economic support of 85 billion euros from the EU and IMF (EU 2012:1). According to the EU (2012:1), this crisis was precipitated by a housing bubble that was fueled by risky lending, growing public spending and a blanket guarantee that the government issued to protect distressed banks. These fueling factors caused the sovereign debt of Ireland to reach 9 percent, which caused Ireland to be excluded from the global bond market. Ireland was rescued by the EU and IMF package and also by an agreement by EU members to restructure interest and extend maturity of EU loans (EU 2012:1). Portugal and Spain also took on follower positions indicating a nascent trend of growing debt that would manifest during the sovereign debt crisis. The Granger results can be seen in the table below. Table 5: Granger results for Lehman contagion window Null Hypothesis: F-Statistic P-value. HUNGARY does not Granger Cause IRELAND 6.352 *0.019 HUNGARY does not Granger Cause PORTUGAL 4.500 *0.044 HUNGARY does not Granger Cause SPAIN 4.500 *0.044 PORTUGAL does not Granger Cause IRELAND 7.071 *0.014 SPAIN does not Granger Cause IRELAND 7.071 *0.014 * Reject Null Hypothesis at 95% level

223

224

Leila Fourie and Ilse Botha / Procedia Economics and Finance 24 (2015) 218 – 227

Source: Eviews estimates The sovereign debt window of contagion indicates seven short-run relations. During this window, the effects of the downgrade of France by S&P in January 2012 are clear as France becomes a follower in Table 6 (Adam and Deen 2013:1). In this table, Hungary and Italy both cause France. During this period, France faced a budget deficit and its attempts to resolve this problem by increasing taxes was negatively accepted (Adam and Deen 2013:1). Italy and France are both large, developed G10 countries. A sovereign downgrade in these countries will have a negative impact on the business confidence and market sentiment. During this period, Hungary also causes Greece and Portugal while Spain causes Hungary and Portugal causes Ireland. Granger results can be seen in Table 6 below. Table 6: Granger results for sovereign debt crisis window of contagion Null Hypothesis: F-Statistic P-value HUNGARY does not Granger Cause FRANCE 9.676 *0.001 ITALY does not Granger Cause FRANCE 3.563 *0.049 HUNGARY does not Granger Cause GREECE 3.412 *0.054 HUNGARY does not Granger Cause ITALY 6.354 *0.008 HUNGARY does not Granger Cause 4.467 *0.026 PORTUGAL SPAIN does not Granger Cause HUNGARY 4.693 *0.022 PORTUGAL does not Granger Cause 7.125 *0.005 IRELAND * Reject Null Hypothesis at 95% level Source: Eviews estimates Both cointegration and Granger results indicate that there is materially more correlation in the contagion windows than in the tranquil windows. This supports the theory that contagion exists when “co-movement exceeds that explained by the fundamentals” (Gentile and Giordano 2012:8). Second, the degree of contagion during the sovereign debt crisis exceeds that of the Lehman contagion window. Third, when comparing short- and long-run relations, it is noteworthy that more long-run relations exist per period than short-run relations. This is understandable since sovereign ratings are designed by rating agencies to capture a longer term state. This contrasts with the outcomes in the Gentile and Giordano (2012:4) study, which uses bond spreads and equity prices. Bond spreads and equity prices are shorter-term and more responsive measures. 4.3. Country shock results using variance decomposition The third step in our estimation procedure involves variance decomposition. Variance decomposition is applied to each window of time to indicate how much each of the country sovereign ratings accounts for forecast variance over time. Contagion exists where the degree of variance increases over time and the system’s vulnerability to shocks increases (Gentile and Giordano 2012:31[4]). An increase in variance points to an externality of an event where the co-movement is attributed to something more than fundamentals. All variables were Cholesky ordered for each window according to the degree of exogeneity. The variance decomposition tables are available on request. The table 7 to 9 only represents the variance decompositions for Greece as an example. Looking at the results, we see in the tranquil window (Table 7) the VDs indicate that each country contributes a much higher degree of variance to itself during the tranquil window. Most countries begin in period one to explain more than 95 percent of its own forecast variance and this reduces to around 65 percent after 10 periods. This indicates a lower co-movement between the countries in the tranquil window.

225

Leila Fourie and Ilse Botha / Procedia Economics and Finance 24 (2015) 218 – 227

Table 7: Variance decomposition - Tranquil window: May 2004 – August 2008 Variance Decomposition of GREECE: Period

S.E.

GREECE

ITALY

POLAND

PORTUGAL

SPAIN

HUNGARY

1

0.127

100.000

0.000

0.000

0.000

0.000

0.000

5

0.236

77.390

17.033

3.324

0.002

0.000

2.250

10

0.308

58.861

31.383

7.518

0.117

0.000

2.122

Source: Eviews estimates The Lehman contagion window shown in Table 8 highlights a high degree of connectivity with France entering the system more prominently. The forecast variance of France is explained by 32 percent of Hungary in the fifth period. France is explained by 11 percent of Hungary in period 10 and by 31 percent of the Spanish variance after 10 periods. Greece plays a principal role and its forecast variance is explained by 41 percent of Portugal in period two. Greece is also explained by 39 percent of Ireland and 11 percent of Spain during period 10. France has an almost immediate impact on Greece, causing 24 percent of its variance over the first period before tapering off. During this period, Italy’s forecast variance is explained by 55 percent of France in period five and 18 percent of Hungary and Italy respectively. Spain explains 40 percent of the forecast variance of Italy in period 10. The prominent role that France plays in Italy’s forecast variance continues for Portugal and Spain. France explains 26 percent of Portugal and over 50 percent of Spain. This can be attributed to the high degree of financial integration between these countries. Many French banks were heavily invested in Italian, Spanish and Portuguese sovereign debt. Table 8 Contagion window: September 2008 – December 2009 Variance Decomposition of GREECE: Period

S.E.

FRANCE

GREECE

HUNGARY

IRELAND

ITALY

PORTUGAL

SPAIN

1

1.085

23.779

76.221

0.000

0.000

0.000

0.000

0.000

5

1.920

9.869

31.472

5.254

0.815

0.790

40.750

11.050

10

2.731

5.149

15.597

5.206

39.274

0.728

22.810

11.237

Source: Eviews estimates The sovereign debt window in Table 9 indicates a strong Portugal, Italy, Greece, Spain (PIGS) theme. The results indicate that the forecast variance of Greece is explained by itself and from the fifth period, 35 percent is explained by Portugal’s variance and up to 20 percent by Spain. Once again, Ireland is explained by itself, perhaps indicating that most European countries and financial institutions had reduced its exposure to Ireland after the Irish bailout in the previous window. Italy’s variance is explained by up to 22 percent of Greece in period five and 14 percent of Hungary. Spain’s forecast variance is explained by up to 46 percent of Italy and 10 percent of Greece. During period 10, Spain explains 32 percent of Italy’s forecast variance. This circular and mutually reinforcing pattern supports the theory that unifying, externalised factors are at play, all working together to increase the contagion experienced by the PIGS countries. Table 9: Contagion window: December 2010 – January 2013 Variance Decomposition of GREECE: Period

S.E.

GREECE

HUNGARY

IRELAND

ITALY

PORTUGAL

SPAIN

1

1.097

100.000

0.000

0.000

0.000

0.000

0.000

226

Leila Fourie and Ilse Botha / Procedia Economics and Finance 24 (2015) 218 – 227

5

1.886

44.294

2.976

1.001

1.326

34.063

16.340

10

2.748

21.398

3.703

34.605

1.252

18.954

20.088

Source: Eviews estimates 5. Conclusion The empirical model used a three-step process involving cointegration, Granger causality and variance decomposition tests. The main concept that the tests aimed to prove was that contagion existed in the sovereign ratings during the two recent windows of contagion, namely the Lehman and sovereign debt crisis. The results gave conclusive evidence of contagion during the two crisis windows when compared with the previous tranquil window. The tranquil window indicated four cointegrating relations, as compared to the seven and twelve relations during the Lehman- and sovereign debt crisis periods. This strongly supported the theory that an abnormally high presence of long-run relations exists in the windows of contagion when compared to the tranquil period. These results were validated by the Granger causality tests. Granger results indicated the short-run relations between the euro countries. Once again, many more short-run relations were evidenced during the contagion windows than during the tranquil period. The tranquil window identified only four short-run relations where the contagion windows measured five and seven short run relations respectively. The third and final procedure using variance decomposition further confirmed the findings from the first two tests. The variance decomposition tests were applied to each window of time to indicate how much each of the country sovereign ratings accounts for variance over time. Contagion existed where the degree of variance increased over time and the system’s vulnerability to shocks increased. The variance decomposition results strongly corroborate the findings from the previous two tests, indicating that contagion was present during the Lehman crisis and the sovereign debt crisis period. Test results indicated a higher vulnerability to shocks and a higher degree of connection during the windows of contagion than during the tranquil periods. Notable was that the sovereign debt crisis experienced a more pronounced degree of contagion than the Lehman crisis period did. During the sovereign-debt crisis window, a dominant theme was the highly integrated connection between the PIGS countries. The dominant PIGS theme in the sovereign debt window corroborates the idea that the movement in these countries goes beyond an isolated cause and effect relationship. These results indicate that a dynamic relationship is at play, which is caused by something more than the fundamentals. This gives a strong indication that shocks have been externalised during this period. References [1] Stracca, L. (2013). The global effects of the Euro debt crisis. European Central Bank Working Paper Series, No. 1573. [2] Caporin, M., Pelizzon, L., Ravazzolo, F. & Rigobon, R. (2012). Measuring sovereign contagion in Europe. Norges Bank Working Paper, 2012/05. [3] Gentile, M. & Giordano, L. (2012). Financial contagion during Lehman default and sovereign debt crisis. Comissione Nazionale Per LE Societa E LA Borsa. [4] Gonzalez-Paramo, J. (2011). Sovereign contagion in Europe. Speaker Seminar of the European Economics and Financial Centre, London, 25 November 2011. [5] Constancio, V. (2012). Contagion and the European debt crisis. Financial Stability Review, 16:109-123. [6] Bekaert, G., Harvey, C.R., Lundblad, C.T. & Siegel, S. (2010). The European Union, the euro, and equity market integration. National Bureau of Economic Research Working Paper, No. 16583. [7] Beber, A., Brandt, M.W. & Kavajecz, K.A. (2009). Flight-to-quantity or flight-to-liquidity? Evidence from the Euro-area bond market. Review of Financial Studies, 22:925-957. [8] Gomez-Puig, M. (2008). Monetary integration and the cost of borrowing. Journal of International Money and Finance, 27:455-479. [9] Borensztein, E., Cowan, K. & Valenzuela, P. (2007). Sovereign ceilings “lite”? The impact of sovereign ratings on corporate ratings in emerging market economies. International Monetary Fund Working Paper, No. 07/75. [10] Rowland, P. & Torres, J.L. (2004). Determinants of Spread and Creditworthiness for Emerging Market Sovereign Debt: A Panel Data Study. Banco de la Republica de Colombia Working Paper. [11] Amira, K. (2004). Determinants of sovereign eurobonds yield spread. Journal of Business Finance and Accounting, 31(5-6):795-821. [12] Andenmatten, S. & Brill, F. (2011). Measuring comovements of CDS premia during the Greek debt crisis. Discussion Paper, No. 11-04. University of Bern, Department of Economics. [13] Saghaian, S.H. (2010). The impact of the oil sector on commodity prices: Correlation or causation? Journal of Agricultural and Applied Economics, 42(3):477-485. [14] Pontines, V. & Siregar, R.Y. (2007). Tranquil and crisis windows, heteroscedasticity, and contagion measurement: MS-VAR application of the DCC procedure. University of Adelaide School of Economics Research Paper, No. 2007-02. [15] Campbell, R., Forbes, C.S. & Koedijk, K. (2006). Diversification meltdown or just fat tails? University of Maastricht. 1-38. [16] Suleimann, R. (2003). New technology stock market indexes contagion: A VAR-dccMVGARCH Approach. IDHE-MORA No. 2003-03.

Leila Fourie and Ilse Botha / Procedia Economics and Finance 24 (2015) 218 – 227

227

Available from: http://128.118.178.162/eps/em/papers/0307/0307003.pdf [17] Forbes, K. & Rigobon, R. (2002). No contagion, only interdependencies: Measuring stock market co-movements. Journal of Finance, 57(5):2223-2261. [18] Butler, K.C. & Joaqin, D.C. (2002). Are the gains from international portfolio diversification exaggerated? The influence of downside risk in bear markets. Journal of International Money and Finance, 21(7):981-1011. [19] Ang, A. & Chen, J. (2002). Asymmetric correlations of equity portfolios. Journal of financial Economics, 63(3):443-494. [20] Longin, F. & Solnik, B. (2001). Extreme correlation in international equity markets. Journal of Finance, 56(2):649-676. [21] Favero, C.A. & Giavazzi, F. (2000). Looking for contagion: Evidence from the ERM. National Bureau of Economic Research Working Paper, No. 7797. [22] Beirne, J. & Gieck, J. (2012). Interdependence and contagion in global asset markets. European Central Bank Working Paper Series, No. 1480. [23] Dungey, M., Fry, R. & Martin, V.L. (2004). Currency market contagion in the Asia-pacific region. Australian Economic Papers, 43(4):379395. [24] Eichengreen, B., Rose, A.K. & Wyplosz, C. (1995). Exchange market mayhem: The antecedents and aftermath of speculative attacks. Economics Policy, 21:249-312. [25] Dungey, M. & Martin, V.L. (2001). Contagion across financial markets: An empirical assessment. Canberra: Australian National University. [26] Kaminsky, G. & Reinhart, C.M. (1999). The twin crises: The cause of banking and balance of payment problems. American Economic Review, 89(3):473-500. [27] Caballero, R.J., Farhi, E. & Gourinchas, P.O. (2008). Financial crash, commodity prices and global imbalances. Brookings Papers on Economic Activity, 2:1-55. [28] Baig, T. & Goldfajn, I. (1999). Financial market contagion in the Asian crisis. International Monetary Fund Staff Papers, 46(2):167-195. [29] Cecchetti, S.G., Mohanty, M.S. & Zampolli, F. (2010). The future of public debt: Prospects and implications. Bank for International Settlements Working Paper, No. 300. [30] Reinhart, C.M. & Rogoff, K.S. (2008). This time is different: A panoramic view of eight centuries of financial crises. National Bureau of Economic Research Working Paper. [31] Kindleberger, C.P. & Aliber, R. (2005). Manias, panics and crashes. A history of financial crises. Hoboken: John Wiley & Sons. [32] Manasse, P. & Roubini, N. (2005). ‘Rules of thumb’ for sovereign debt crises. International Monetary Fund Working Paper, No. 05/42. [33] Manasse, P., Roubini, N. & Schimmelpfennig, A. (2003). Predicting sovereign debt crises. International Monetary Fund Working Paper, No. 03/221. [34] Mellios, C. & Paget-Blanc, E. (2003). Which factors determine sovereign credit ratings? Thema University, France. Discussion Paper. [35] Reinhart, C.M. (2002a). Default, currency crises and sovereign credit ratings. National Bureau of Economic Research Working Paper, No. 8738. [36] Fofana, L. & Seyte, F. (2012). Modeling financial contagion: Approach-based on asymmetric cointegration. Laboratoire Montpellierain d’Economie theorique et Appliquee. [37] Reinhart, C.M. (2002b). Sovereign credit ratings before and after financial crises. Munich Personal RePEc Archive. New York UniversityUniversity of Maryland Project. MPRA Paper, No. 7410. Available from: http://mpra.ub.uni-muenchen.de/7410/ [38] Rowland, P. (2004). Determinants of spread, credit ratings and creditworthiness for emerging market sovereign debt: A follow-up study using pooled data analysis. Banco de la Republica. Available from: http://www.banrep.gov.co/docum/ftp/borra296.pdf [39] Cantor, R. & Packer, F. (1996). Determinants and impacts of sovereign credit ratings. The Journal of Fixed Income, 6(3):76-91. [40] Afonso, A., Gomes, P. & Rother, P. (2007). What “hides” behind sovereign debt ratings? European Central Bank Working Paper Series, No. 711 /January 2007. [41] Gray, D. (2009). Financial contagion among members of the EU-8: A cointegration and Granger causality approach. International Journal of Emerging Markets, 10:1108. [42] Bernanke, B.S. & Lown, C. (1991). The credit crunch. Brookings Papers on Economic Activity, 199(2):205-239. [43] Dungey, M., Fry, R., Gonzalez-Hermosillo, B. & Martin, V.L. (2011). Transmission of financial crises and contagion. New York: Oxford University Press.