“Contagion” is today’s buzzword (de Haan and Mink 2012, Manasse and Trigilia 2011).

• The troika’s bailout of the Greek Government and the heavy haircut imposed on private bond holders have failed to reassure markets about Greece’s permanence in the Eurozone.
• The relief brought about by the recent election results waned in a matter of hours.
• Similarly, the decision by the EU to pour about €100 billion into Spanish banks has not proved sufficient to convince investors that the umbilical cord between the State’s and the banks’ balance sheets have been severed.
• In the meantime, confidence on peripheral sovereign bonds and banks has been crumbling, as interest rates and CDS spreads rose in the past weeks.

While proposals for a banking union, Eurobonds (Manasse 2010) and fiscal union still belong to the realm of dreams (or nightmares, depending on who should be footing the bill), the question is whether the flight out of Europe’s periphery will become a flight out of the euro full stop.

Irrationality or delayed realisation? Empirical evidence on contagion

But are financial markets behaving “irrationally” or – following a long period of benign neglect – are they simply rediscovering market fundamentals (Manasse 2011b)? And crucially, what policies should southern Europeans, Italy in particular, implement to maintain market access?

In order to address some of these questions we look at the recent evidence on contagion across EU sovereigns CDS spreads. Our empirical model builds on Bekaert et al. (2009) by using time-varying factor loadings and market indexes in order to proxy the dynamics of common and specific risk factors. In the empirical model the daily change in a country sovereign’s spread depends on four elements:

• The change in a global risk factor, measured by an index of the most important (non-European) sovereigns’ CD.
• The change in a European risk factor, measured by an index of Western European sovereigns’ CDS.
• The change in a financial risk factor, measured by an index of the CDS on private European Financial Institution.
• A (time-varying) idiosyncratic component captures the market participants’ assessment of the individual country sovereign risk.

We proceed in two steps. First we estimate the model on daily observations (1630) from January 1, 2006 to March 29, 2012, separately for 15 European countries, 11 of which belong to the Eurozone (Germany, France, Italy, Spain, Belgium, Greece, Portugal, Ireland, Netherland, Austria and Finland), and four of which who do not (Sweden, Norway, UK and Denmark). Unlike the previous literature, we run a series of rolling regressions, on a moving window of 200 observations, estimating 1,427 regressions for each country and retaining the time series of parameters relating the spread change to the Global, European and Banking indexes.

The evolution of these parameters is quite interesting in itself, as it reveals comparatively when and to what extent each country reacted to “external risks” through the recent waves of crises (we do not discuss these results here for reasons of space). In addition, our approach allows us to trace the systematic movements in the country-specific risk, the drift in the CDS spread daily change. This captures the idiosyncratic component that is unrelated to the remaining “external” factors accounted by the model. When this parameter spikes simultaneously for many countries, we have an indication of “pure contagion” in the sovereign CDS market, possibly resulting from herd behaviour, a rise in risk aversion, agents’ coordination on a “bad equilibrium”.

“Pure” idiosyncratic sovereign risks during the US sub prime and the Greek crises

Figure 1 (see below) plots the evolution of the systematic change in the idiosyncratic risk component (only when significantly different from zero at 5%) for the countries in the sample. The US subprime crisis (September 2008 and March 2009) and the Greek crisis (around November 2009) are evident in the data, as jumps in risks are clustered around these episodes. As regards to the crises’ impact, countries naturally divide themselves into three sizes: Smalls (Finland, Germany and Norway), Mediums (Sweden, Denmark, the Netherlands, Belgium, France, the UK and Austria on the high side) and Larges (the “periphery”).

More fundamentally, while the US sub prime earthquake affected all Europeans, albeit with different magnitudes (Ireland, followed by Austria and the UK being the most affected for reasons due to the role of their financial institutions), the Greek crisis is largely a matter for the Eurozone. Norway, Sweden, the UK and Denmark, which do not belong to the Eurozone, were hardly affected. But differences inside the Eurozone were at least as remarkable, with France, Belgium, Italy, Spain, Ireland and Portugal showing large and recurrent spikes in idiosyncratic risk. Thus an explanation requires more than the one-size-fits-all corset provided by the euro.

Figure 1. “Smalls”

Figure 2. “Mediums”

Figure 3. “Larges”

The role of fundamentals

In order to understand what makes a country vulnerable to changes in “market sentiment”, we perform a second round analysis. We use panel estimation in order to explain cross-country differences in the model parameters on the basis of each country’s (lagged) economic fundamentals (trade openness, the public debt/GDP ratio, the budget deficit/GDP ratio, the current account balance as percentage of GDP, the rate of unemployment, the monthly change in industrial production, the country’s sovereign rating (Moody’s), an index of market volatility (the VIX), of liquidity-shortage in the inter-bank market (the TED spread). We also include a “crisis” dummy which takes the value of one from November 2009, when the numbers for the Greek 2009 budget deficit was revised from 5% to 12.7%, and a Eurozone dummy.

Table 1 presents the results.

The first column shows the effects of the countries’ lagged economic fundamentals on the time-varying risk coefficients, outside the crisis. The second column shows the additional effects of each variable during the Greek crisis (eg. the coefficient of the interaction of each variable with the crisis dummy). It appears that before the outset of the European debt crisis (first column) three variables significantly affect the idiosyncratic risk: the rate of growth in industrial production (which significantly reduces the risk), the budget deficit/GDP (which significantly raises it), and the volatility index (which significantly raises it).

However, things change dramatically during the crisis. First, markets sentiments shift against the Eurozone countries, as documented by the fact that the Eurozone dummy becomes significant during the Greek crisis with positive sign; second, the systematic change in the spread is higher in countries where the public debt-to-GDP ratio is higher, as this variable now becomes positively associated to the systematic risk. Moreover, the real economy matters more inside the crisis: idiosyncratic risks respond positively and significantly to the rate of unemployment, and the coefficient for the rate of growth of industrial production becomes larger during the recent crisis.

Another difference with the pre-crisis period is that countries’ systemic risks become sensitive to credit rating: this variable, which was not significantly different from zero outside the crisis, becomes positive and significant. Overall, our measures for economic fundamentals explain more than 50% of the cross-country variation in idiosyncratic risks.

Conclusions

Our evidence supports the conclusion that after a long period of benign neglect in the Eurozone, financial markets have rediscovered that fundamentals and structural fragilities matter for sovereign risks.

• In the crisis, markets have increasingly focused on the public debt, on the real economy, and on the labour market in particular.

The implications for the appropriate policy response required in order to become more resilient are straightforward:

• More emphasis should be placed upon the employment and growth so that fiscal consolidation does not backlash by plunging the economy into recession.

Here the reason it is not because the recession widens the deficit through the automatic stabiliser. The effect works via a direct link from lower employment and growth to spreads.

• Although not necessarily the panacea for solvency (Bekaert et al. 2011), privatisations should be accelerated, not only because they do not adversely affect the economy, but also because they may well work to calm fears of debt insolvency;
• Labour market reforms that reduce unemployment should have priority.

However, structural reforms aimed at increasing flexibility and reducing firing costs – such as the one currently being voted by the Italian Parliament – may backlash if they raise unemployment in the short run. They should be accompanied by reforms in the wage bargaining system in order to make real wages more flexible (Manasse 2011a).

• The evidence seem to reject the naïve “credibility” view of multiple equilibria and sunspots, according to which “credible promises” of future reforms may be all that is needed to select the “good equilibrium” of low interest rates and solvency (Manasse 2012).

If this were true, past economic fundamentals should not explain cross-country differences in “fear perceptions”. However, to a large extent, they do.

This column is based on Luca Zavalloni’s MSc Economics (Lmec) Dissertation at the University of Bologna and on work in progress with Paolo Manasse.

This article first appeared on Vox.