Multiple Regression, Computing Expected Return

The predicted annual return for a stock that is included in the DJIA, has a market capitalization of $52 billion and a book-to-market ratio of 0.45 is closest to:
A. – 9.08%.
B. 3.59%.
C. 7.33%.

ri = b0 + b1 Sizei+ b2 BMRatioi + b3 InDowi + εi
where Sizei is the natural log of the market capitalization of company i in billions of dollars. The output from the regression model is shown in Exhibit 1.

The answer is A instead of C.

Question: If you are using the regression equation to estimate, do you have to use every variable, even when the variable is not significantly different than zero? The answer includes for dummy variable lnDow, even though it is not statistically significant.

Yes.
If you want to remove a variable you need to re-run the regression.

Understood. Thanks!