School of Business and Economics

Research blog from the School of Business and Economics


blog photo

Contagion within and without of the Euro zone

One of the claims of Euro zone membership for the southern European countries was that by joining the Euro their governments could borrow at low (northern European) rates of interest. The reason for this was that there would be no exchange rate risk as all borrowing would be in Euros. This argument, however, ignored the remaining very different country risks, in that Greek debt should still carry a higher interest rate if Greece was believed more risky, than say, the Netherlands. In the early years of the Euro such different country risks were ignored by the markets and so bond yields were generally stable and low averaging between 4.37 per cent and 4.73 per cent. Greece, Portugal and Spain could all borrow at the same rate as Germany and the Netherlands.

After the global financial crisis in 2008, however, it became evident that country risks within Euro member states could not be ignored as was manifest in the Greek crises. Between end-2009 and end-2015 Greece experienced two discernible financial crises. The first was marked by the provision of an initial bail-out package in May 2010, with a second package worth €246 billion agreed in February 2012. In March 2012, continued sovereign debt restructuring in Greece and, perhaps more strategically, a statement by the President of the European Central Bank, Mario Draghi, in July 2012, that the ECB would do “whatever it takes” to save the euro, were associated with a general reduction in bond yields. As indicated by Chart 1, the acute phase of the crisis seemed to be over. The second crisis was associated with the rise to power of the Syriza party in 2015, and as Chart 1 shows the risk premia on Greek 10-year government bonds rose sharply again at this time.

The research project had two main aims. First, to see if there was contagion from the two Greece crises to the other Euro area members and the extent to which the degree of contagion was different between the two crises. The second aim of the project was to examine the extent of contagion from both crises to the Non-Euro members of the EU, all of whom operated some kind of managed floating exchange rate regime with the Euro, which we divided up into two groups of countries – the Czech Republic, Hungary and Poland – which all have aspirations to join the Euro-zone and Great Britain, Sweden and Switzerland who are financially more developed countries who have no plans to ever join the Euro area. This would enable to us test the idea that some exchange rate flexibility might offer these member states some protection from financial contagion within the Euro zone.

The methodology employed was to compare the Greek risk premium – measured as the difference between the Greek 10-year bond rate and the German 10-year bond rate – with the other countries risk premium over the German bond rate, using daily data. For the non-Euro zone countries we also took account of the expected changes in the bilateral nominal exchange rates against the Euro – although as it turned out such changes were not important.  The comparisons involved the use of correlation measures – some of which are reported below – but also regression analysis using a multivariate GARCH dynamic conditional correlation model – to which reference should be made to the working papers.

Figure 1. Long-Term Government Bond Yields of Greece, Ireland, Italy, Portugal, Spain, France, the Netherlands, Germany, Switzerland, U.K., Sweden, Poland, Czech Republic and Hungary from 1st October 2009, to 12th August 2016. Source: Stock Market Quotes & Financial News

With regards to the effect of the Greek crises on the risk premium in other Euro zone economies – a positive and statistically significant correlation between the risk premia – is taken as evidence of contagion. Table 1 shows positive and statistically significant correlation coefficients between Greece and other Euro area countries, which is indicative of contagion.  A commonly expressed view is that the effects of the second crisis were more muted since the systemic risks were seen by markets as being lower.

However, using a rolling correlation model we find that this is not the case. In both the 15-day and 5-day roiling windows the correlation coefficients for all Euro zone countries are larger in the 2nd crisis period than in the first and these differences are statistically significant for France, Netherlands and Portugal, as Table 1 below shows.

Table 1:  Correlations between Changes in risk premia of Greece and the Other Selected Countries in the First and the Second Greek Crisis

1st crisis 2nd crisis difference between the 1st and 2nd crisis (in abs)
15-day rolling window (mean) R 5-day rolling window (mean) R 15-day rolling window (mean) R 5-day rolling window (mean) R 15-day rolling window (mean) R 5-day rolling window (mean) R
Ireland 0.41*** 0.34** 0.42** 0.36** 0.01 0.02
Portugal 0.36*** 0.28** 0.46** 0.40** 0.10*** 0.12***
Spain 0.39*** 0.33** 0.40** 0.34** 0.01 0.01
Italy 0.39*** 0.30** 0.39** 0.34** 0.00 0.04
France 0.26*** 0.19** 0.30** 0.26** 0.04** 0.07**
Netherlands 0.18*** 0.09** 0.26** 0.25** 0.08*** 0.16***

        Note: *** and ** indicates the significant levels of 1% and 5% respectively.

With regards to the non Euro zone countries we found very different results, although not due to expected changes in the bilateral nominal exchange rates as we initially expected. The results, shown in Table 2,  show that there was contagion to the three acceding members of the Euro zone (Czech Republic, Hungary and Poland), exactly as there was with full Euro zone members, so the exchange rate flexibility they had in fact offered them very little independence from Euro zone financial shocks. On the other hand, for the three financially developed economies – Switzerland, Sweden and GB – there was little evidence of contagion, but instead rather strong evidence of safe haven effects, probably due to a ‘flight to safety’ from Euro-denominated bonds, including German bonds. That is a rise in the Greek risk premium lead to a fall in the risk premium (a negative correlation) over Germany of the three economies – suggesting that investors chose to switch funds to these economies to avoid Euro zone risks. In this case remaining outside the Euro zone did not insulate GB from the shocks, but they had the opposite effect to that of other countries in that financial capital flowed into rather than out of the country.

Future work on this topic is ongoing as we try to improve the modelling of the interdependencies between the countries risk premia, and in particular combine the Euro zone and non-zone results into a common model framework.

Table 2: Dynamic Conditional Correlations between risk premia of selected EU countries (Greece exogenous)

Switzerland Czech Republic GB Hungary Poland
Czech Republic -0.008
GB 0.240*** -0.221***
Hungary -0.111** 0.674*** -0.259***
Poland -0.139*** 0.627*** -0.181*** 0.747***
Sweden 0.076** -0.000 0.207*** 0.104** 0.121***

Note: *** and ** indicates the significant levels of 1% and 5% respectively.

This Blog post was written by Professor Eric Pentecost, member of the Economics discipline group and of the Money and Developing Economies Research Interest Group. Eric can be contacted on