
Daily Speculations 
4Jun2006
Correlated Assets, by
Kim Zussman
Recent discussions about tandem movement of various asset classes prompts the question as to what, if any, effect this has on stocks. One way to look at this is to check whether correlation of bond and stock moves has an effect on subsequent returns for stocks.
TNX (10 yr treasury yield) weekly closes since 1962 were inverted to give a proxy for bond price, and from this weekly change in bonds was calculated. The SP500 index weekly closes over the same period were used for stock weekly change. At the end of each year, correlation was calculated for change in bonds with change in stocks, looking back over the prior 50 weeks. The return of stocks over each subsequent year was regressed against the prior year's correlation of change in stocks and change in bonds:
Regression Analysis: sum nxt y versus cor st int The regression equation is sum nxt y = 0.0529 + 0.330 cor st int Predictor Coef SE Coef T P Constant 0.05291 0.02645 2.00 0.052 cor st int 0.3304 0.1675 1.97 0.055 S = 0.145077 RSq = 8.7% RSq(adj) = 6.4%
Higher correlation last year between change in stocks and change in bonds predicts higher return in this year's stocks, though the relationship was just shy of significance. 2005 correlation was about 0.08, so with any decent threats from the turbanbeard crowd stocks should be up in 2006.
Here are the years and correlations between weekly change bonds and stocks:
year cor st int 2005 0.077 2004 0.039 2003 0.037 2002 0.040 2001 0.026 2000 0.022 1999 0.053 1998 0.088 1997 0.013 1996 0.297 1995 0.007 1994 0.003 1993 0.000 1992 0.023 1991 0.190 1990 0.190 1989 0.067 1988 0.145 1987 0.332 1986 0.102 1985 0.102 1984 0.009 1983 0.259 1982 0.254 1981 0.184 1980 0.307 1979 0.152 1978 0.095 1977 0.049 1976 0.122 1975 0.280 1974 0.164 1973 0.147 1972 0.086 1971 0.213 1970 0.334 1969 0.073 1968 0.081 1967 0.207 1966 0.110 1965 0.013 1964 0.026 1963 0.066 1962 0.256