This is the largest divergence between the two in the past five years and comes at a time when the economic situation in Europe is deteriorating while ECB is becoming increasingly dovish, at least in rhetoric, if not in action.
We build a simple model to explain variation in credit spreads over time using the ZEW surveys and the V2X equity volatility index. Such a model can explain significantly more than 70% of variation in credit spreads over a five-year period and indicates that if iTraxx Main priced off ZEW and V2X alone, spreads should be materially wider than presently.
We interpret this as a willingness of markets to “look through” recent economic weakness as long as ECB stays true to its dovish rhetoric. In a sense, this is likely to show similar dynamics to that of the US leading into Fed QE, where “bad news is good news,” increasing the probability of further accommodative action.
Although we applied this framework to credit spreads, we note that similar conclusions can be derived from analyzing equity markets (SX5E for instance) instead – broadening the generic conclusions to markets beyond credit, although as a general statement, it appears that credit markets are the most willing to buy into ECB rhetoric.
Beyond looking at the optics of Main vs ZEW (Figure 1), we build a “model” to explain iTraxx Main using three variables: V2X and the surveys conducted by ZEW for “Eurozone Assessment of Current Situation” and “Eurozone Expectation of Economic Growth” (tickers GRZEEUCU and GRZEEUEX on Bloomberg, respectively). We use a five-year horizon, going back to November 2009, using monthly data, reflecting the frequency of release of the ZEW surveys. Estimating the model, we have an acceptable fit with highly significant parameter estimates with intuitive coefficients.
The fitted model has a good fit to actual spread data over the 5yr horizon (Figure 3), capturing broad market moves. As the initial chart of Main vs ZEW (Figure 1) indicates, the modelled level for iTraxx Main is 132bp – compared to the Friday 17 October level of 73bp – almost 60bp difference. Bearing in mind that the standard error of the model is 16.6bp, this is a 3.5x standard deviation move – much larger than previously experienced over the past five years.
Where does the divergence between actual and modelled spread come from? The divergence between the model and actual spreads started in August 2014 and was largely driven by deteriorating ZEW surveys – with the last published numbers on 14 October confirming this trend – but in the past few days, the model-implied spread has also risen as the V2X index has increased significantly.
When seeing divergences of this kind, it is worthwhile taking a step back and considering whether the explanatory variables are demonstrating an unusual behaviour. It is fair to say that the ZEW surveys show very bearish readings presently – and running a similar model using European Commission indicators gives qualitatively similar, but less extreme, results. Further, the equity volatility market appears to have experienced strong technicals, potentially causing an increase beyond what credit spreads “care” about.