The Determinants of Government Yield Spreads in the Euro Area

Abstract

This paper analyses the determinants of sovereign spreads in the euro area from January 2002 to May 2012. The objective is to disentangle the role of country-specific fundamentals, driven by fiscal and macroeconomic factors, from what is referred to as contagion. Following the existing empirical literature, the work estimates a model of the determinants of 10-year yield spreads relative to Germany for ten euro zone countries. The results show that since the eruption of the 2007-2008 financial crisis, sovereign spreads have shown a time-dependent contagion component. On average, such a component explains almost one third of the spreads dynamic in 2009-2010 and almost 10 per cent since 2011. However, results at the country level are quite different between core and peripherals. As shown by the analysis, core countries (excluding Germany, which is our benchmark to measure spreads) were not affected by contagion till 2011; since the worsening of the sovereign debt crisis they seem to have benefited from a flight-to-quality effect. For example, in the first months of 2012, France shows spreads lower than what implied by fundamentals by an amount ranging from roughly 50 to 90 basis points, depending on the model specification, while for Netherlands such a “discount” can be as high as roughly 60 basis point. Peripheral countries, which at the onset of the European Monetary Union took advantage from a mispricing of their actual economic and fiscal fragility, since 2009 have suffered from the abrupt revision of market expectations, showing spreads on average significantly higher than what justified by macroeconomic and fiscal factors. In 2012, for most of these countries contagion has a role comparable to fundamentals in explaining the level of the spreads. For example, it accounts for an amount ranging from roughly 170 to 240 basis points for Spain, while for Italy – probably penalized by its historically highest debt to GDP ratio – contagion explains something between roughly 150 and 180 basis points of the spread, depending on the model specification.