Just a quick note here … the Kansas City Fed published a paper by Craig S. Hakkio and William R. Keeton titled Financial Stress: What Is It, How Can It Be Measured and Why Does It Matter?.
One of the coefficients is the stock/bond correlation.
Correlation between returns on stocks and Treasury bonds. In normal times, the returns on stocks and government bonds are either unrelated or move together in response to changes in the risk-free discount rate. In times of financial stress, however, investors may view stocks as much riskier than government bonds. If so, they will shift out of stocks into bonds, causing the returns on the two assets to move in opposite directions. A number of studies, some for the United States and some for other countries, confirm that the correlation between stock returns and government bond returns tends to turn negative during financial crises (Andersson and others; Baur and Lucey; Connolly and others; Gonzalo and Olmo). Thus, the stock-bond correlation provides an additional measure of the flight to quality during periods of financial stress. This correlation is computed over rolling three month periods using the S&P 500 and a 2-year Treasury bond index. Also, the negative value of the correlation is used in the KCFSI, so that increases in the measure correspond to increases in financial stress.
The authors are somewhat critical of the Bank of Canada Stress Index:
It includes some variables, such as exchange rate volatility, that are more important for a small open economy like Canada’s than for the United States. It includes the slope of the yield curve, which likely reveals more about the stance of monetary policy than financial stress. And it fails to include any measures of investor uncertainty about bank stock prices.
There appears to be some predictive value in the index:
As shown in the accompanying box, high values of the KCFSI have tended to either coincide with or precede tighter credit standards over the last 20 years. This evidence suggests that changes in credit standards provide an additional channel through which financial stress may affect economic activity.
Regretably, the authors do not dicuss whether these changes in credit standards can be better predicted by other methodologies. I have the same problem with their analysis of the predictive power of the KCFSI on the value of the Chicago Fed National Activity Index.
The authors suggest that the KCFSI could be used to help time the Fed’s exit strategy for the current crisis, but are, frankly, rather unconvincing.
Anyway, the index is down again for September, after a huge decline from the October 2008 peak, but still above July 2007.