Patterns of financial crisis: EUR/USD, CAC40 and the Dow 
Written by Forex Automaton  
Wednesday, 04 March 2009 16:05  
From the two cases considered so far in the index1/index2 vs currency1/currency2 correlation technique (correlating FTSE/DJIA and DAX/DJIA with EUR/USD), one common feature is evident: oneday lag correlation "prefers" to have an opposite sign to the "instantaneous" correlation. Today's post is about correlations between CAC/DJIA (the ratio of France's and US' blue chip indexes, Cotation Assistee en Continu and the Dow) and EUR/USD. With the ratios costructed this way (symmetrically), the null hypothesis would be to expect a negative correlation: the value of dividendgenerating assets such as stocks is negatively correlated with the cost of money, even without stock market bubbles in the picture, but especially when such bubbles are part of the picture. However, if international capital flows are taken into account, one would have to distinguish between lowinterest and highinterest currencies. A scenario is possible where both value of the stock market in country A and the national currency of the country can be inflated by cheap money originating in country B. The case in hand may not be a good example of this. The point however is that such a mechanism would invalidate the reasoning behind the "null hypothesis". As always on this site (and in contrast with most of the available literature) the emphasis is on correlation values at nonzero time lags, since these, when nonzero in a regular way, indicate technical predictability of the markets  at least in principle. As before, I work with logarithmic returns, therefore it's irrelevant in what units the quantities are expressed (that is how these indexes are defined and what absolute magnitude they typically assume).
The time evolution of both time series under study, EUR/USD and CAC/DJIA, is show in Fig.1. As before, I define the visible phase of the present financial crisis to begin on August 16, 2007, the day of Countrywide Financial nearbankruptcy event, followed by an extraordinary halfpercent Fed discount rate cut next day. This study covers 80 weeks from August 2, 2007 through February 12, 2009. Fig.2 shows the time evolution of the correlation peak by plotting the correlation values for the negative and positive lags in separate panels.
Overall for the period, looking at same day (zero timelag) data, CAC/DJIA is seen to be positively correlated with EUR/USD, similarly to DAX/DJIA, and in contrast to FTSE/DJIA. The main events contributing to make this correlation positive are the stock market minicrash of January 2008 (much stronger in Europe) and the volatile second half of September 2008. The time lag is defined as t_{d} = t_{1}  t_{2}, where index "1" denotes EUR/USD and index "2" denotes CAC/DJIA. Therefore, positive correlation values at negative lags mean that movements of the same direction in EUR/USD and CAC/DJIA happen at an earlier time in EUR/USD, or EUR/USD is a leading indicator for CAC/DJIA. Negative correlation values at negative lags (top panel, January 2008) mean that movements of the opposite direction in EUR/USD and CAC/DJIA happen at an earlier time in EUR/USD, or EUR/USD is a leading indicator for DJIA/CAC. For the forex trader, the bottom panel of Fig.2 is more interesting as it deals with potential leading indicator for EUR/USD. Negative correlation values at positive lags (look at the bottom panel of Fig.2, Fall 2008) mean that movements of the opposite direction in EUR/USD and CAC/DJIA happen at an earlier time in CAC/DJIA, or CAC/DJIA is a leading indicator for USD/EUR. Speaking of stable patterns, clearly the negative bunch of 1day lag correlations in the bottom panel during the Fall of 2008 is more than a random coincidence. So is the tendency for the 1day lag correlation value to be of an opposite sign to the "instantaneous" (lag zero) correlation value. 

Last Updated ( Monday, 04 January 2010 12:39 ) 