Contagion from the crises in the Euro-zone: where, when and why?

ABSTRACT

The prevalence of contagion between the Euro-zone countries and other European countries since the Greek crisis of 2009 is now well – known, but the factors that influence the pattern of this contagion are not well understood. We investigate this question both within Europe and beyond to the USA and Japan, using an asymmetric M-GARCH model that focuses on extreme values of the risk premia on government bonds. We compare these extreme values with news of major events and find that they are highly correlated. We find a different pattern of contagion emanating from Ireland compared to the other crisis countries of Greece, Italy, Portugal and Spain. We also examine the factors that have made countries vulnerable to contagion and find that financial factors are more important than trade ones. However, intra-Euro-zone trade has also been a significant factor between the major Euro-zone economies. There is little evidence that global factors affect contagion between EU member states, but some evidence that nominal exchange rate movements offer a degree of insulation from contagion for the non-Euro zone states.

KEYWORDS:

• Contagion
• extreme values
• M-eGARCH models
• quantile regression

JEL CODES:

• G01
• G15
• C58

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 For an analysis of the various types of contagion see Forbes and Rigobon (2002), Forbes (2012), Metiu (2012), Ludwig (2014), Mink and de Haan (2013), and Cronin, Flavin, and Sheenan (2016).

2 Since the model is only estimated over the crisis period, the average level of risk is likely to be higher than in a non-crisis period and so the value of ${\beta }_{0i}$, is also likely to be a little higher. This, however, is not important to the argument here.

3 This assumes that flows out of the crises countries into Germany do not have a significant effect on German interest rates, which are used to capture global capital market factors.

4 1st October 2009 is selected as the start date because it marked the beginning of the permanent deviation of Greek rates from German rates. From the start of the Euro until 2009 the average deviation between Greek and German 10-year bond rates was about 0.26%. In October 2009 the deviation was about 1.36% as the sharp upward trend in Greek rates began, which we take to mark the beginning of the first Greek crisis.

5 We proxy changes in the expected exchange rate in two ways: as either static (no expected change) or perfect foresight (instantly and fully adjusted). If daily exchange rate expectations are constant then x_t =0, whereas in the case of perfect foresight expectations x_t =((e_t+1 /e_t)-1)x 100, where $e_t$ is the spot exchange rate.

6 We also experimented with extreme values of two standard deviations from the mean, but this left us with very few values.

7 A (0, 1) dummy variable was created, in which the one’s denoted each individual news event. This series was then correlated with each of the five the countries series of values of their extreme risk premium. Table 1 reports the simple correlation coefficients from this exercise.

8 See Higgins and Bera (2017) for an alternative analysis of contagion that also discusses ‘flight to quality’ effects.

9 The very similar results with perfect foresight exchange rate expectations are presented in Tables 7A and 7B in the on-line appendix.

10 The statement by Mario Draghi in July 2012 that the ECB would do ‘whatever it takes' to save the Euro seemed to bring the acute phase of the first crisis to an end. In April 2013 the Greek parliament approved further economic reforms and in November Moody’s upgraded Greece’s crediting rating as did Fitch in May 2014. However the growing popularity of the anti-austerity Syriza party sowed the seeds for a second crisis. This can be timed from mid-September 2014 when the leader of the Syriza party announced that he would bring austerity to an end. The crisis peaked in July 2015 but abated as tensions between Greece and its creditors eased and the Greek parliament approved new austerity measures as a precursor to further bailout funds. Detailed statistical support for the identification and timing of the two crises is provided by Bird et al. (2017a).