How do I test whether the Dow Jones price is associated with its earnings and dividends?

Question

I have average annual data going back to 1960 for all data. I have run multiple regressions in Excel,but I am not sure of what the data means on the Excel print out. I get conflicting answers from people I know who are familiar with statistics.

Answer 1

If you've run the multiple regression, then you should have beta statistics for each independent variable. Now you need to test to see if your beta stats are statistically different from zero. Your statistical output should give you t stats for the betas. If the betas are different from zero at a staistically significant level then the price is associated with the variable represented by the beta coefficient.

Once you have determined if your variables have explanatory power, then look at the adjusted r-squared for the equation. it will tell you to what degree the price is explained by the earnings and dividends. But if your betas are insignificant, you can't trust the r-squared.

Answer 2

Look at the R^2. This tells you how much of the variation in the dependent variable is explained by variation in the independent variables.

Next, look at the coefficients on the independent variables. Are they statistically different than 0? If so, and the coefficient is positive, then an increase in the independent variable is associated with an increase in the dependent variable- on average and holding all else constant. If it is negative, the the opposite holds. Multiply a value of the independent variable by its coefficient to get the predicted dependent variable value for the given value of the independent variable.

You can also calculate the correlation coefficient w/o running a regression. This tells you how strong the correspondence is between two variables.