sfsu fin822 project 1online.sfsu.edu/donglin/project_822.pdf · 1 sfsu fin822 project 1 this...

17
1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied by printouts of programming outputs. You could use any software to solve the problems. I recommend SPSS or GRETL. Answers to the questions should be typed. No late reports are acceptable. It is very important for every student to know in advance that before the project is discussed, the instructor cannot provide help to any individual student outside the classroom. This is to be fair to the whole class. However, the instructor reserves right to provide hints to ALL students inside the classroom, where you may ask questions. Part I On the course web page there is an Excel workbook (project_data1.xls) that contains a worksheet with monthly returns from October 1990 through July 2005 for the following: • An aggregate value-weighted index of U.S.-listed stocks. The index is constructed based on all companies traded on the NYSE, Amex, and Nasdaq. • Four individual assets: Target, CISCO, Amgen and Best Buy. • A short-maturity Treasury bill, which is used a risk free asset. Capital Asset Pricing Model First, we need to compute the excess returns of the CRSP value-weighted index, Target, CISCO, Amgen, and Best Buy. "Excess returns" here refer to raw returns over and above the short-maturity Treasury bill returns (risk free rates). The 3rd column in the data spread sheet is the monthly Treasury bill returns. (a) Calculate the mean excess returns to the CRSP index and the four assets, and the mean raw returns of the short-maturity Treasury bill (risk free rates). Calculate the standard deviation of excess returns for the four assets and the raw returns of the T-bill. (b) Draw a Q-Q plot to test whether the excess returns to the CRSP index and excess returns for the four assets follow normal distribution. (This can only be done in SPSS, not available in GRETL. If you work with GRETL, draw histograms instead of Q-Q plots.) (c) Report the correlation and the spearman (rank) correlation between these excess returns. (d) Estimate betas for each asset by regressing each asset's excess return on the market's excess return assuming that the CRSP value-weighted index tracked the market portfolio perfectly. Do these regressions give us reason to reject the assumptions of the Capital Asset Pricing Model? (An implication of the Capital Asset Pricing Model is that the Alpha coefficients (i.e., intercepts) should be zero. From the regression output, are the Alpha coefficients statistically different from zero?)

Upload: others

Post on 27-Jun-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

1

SFSU FIN822 Project 1

This project can be done in a team of up to 3 people. Your project report must be accompanied byprintouts of programming outputs. You could use any software to solve the problems. I recommendSPSS or GRETL. Answers to the questions should be typed. No late reports are acceptable. It is veryimportant for every student to know in advance that before the project is discussed, the instructorcannot provide help to any individual student outside the classroom. This is to be fair to the whole class.However, the instructor reserves right to provide hints to ALL students inside the classroom, whereyou may ask questions.

Part IOn the course web page there is an Excel workbook (project_data1.xls) that contains a worksheetwith monthly returns from October 1990 through July 2005 for the following:

• An aggregate value-weighted index of U.S.-listed stocks. The index is constructed based onall companies traded on the NYSE, Amex, and Nasdaq.

• Four individual assets: Target, CISCO, Amgen and Best Buy.

• A short-maturity Treasury bill, which is used a risk free asset.

Capital Asset Pricing ModelFirst, we need to compute the excess returns of the CRSP value-weighted index, Target, CISCO,Amgen, and Best Buy. "Excess returns" here refer to raw returns over and above the short-maturityTreasury bill returns (risk free rates). The 3rd column in the data spread sheet is the monthly Treasurybill returns.

(a) Calculate the mean excess returns to the CRSP index and the four assets, and the mean rawreturns of the short-maturity Treasury bill (risk free rates). Calculate the standard deviation ofexcess returns for the four assets and the raw returns of the T-bill.

(b) Draw a Q-Q plot to test whether the excess returns to the CRSP index and excess returns for thefour assets follow normal distribution. (This can only be done in SPSS, not available in GRETL.If you work with GRETL, draw histograms instead of Q-Q plots.)

(c) Report the correlation and the spearman (rank) correlation between these excess returns.

(d) Estimate betas for each asset by regressing each asset's excess return on the market's excess returnassuming that the CRSP value-weighted index tracked the market portfolio perfectly. Do theseregressions give us reason to reject the assumptions of the Capital Asset Pricing Model? (Animplication of the Capital Asset Pricing Model is that the Alpha coefficients (i.e., intercepts) shouldbe zero. From the regression output, are the Alpha coefficients statistically different from zero?)

Page 2: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

2

(e) Suppose the risk free rate today is the mean short-maturity Treasury bill return from 1990 through2005 computed in (a), and risk premia are constant. Using the Capital Asset Pricing Model, whatis the expected return of each risky asset today?

(f) Which of the four assets has the greatest firm-specific risk? Which has the greatest market risk?For which asset does market movement explain a greater fraction of return variability?

(g) Verify that the squared correlation (from (c)) between market excess return and stock excessreturn is equal to R-square (from (d)).

Part IIYou can download from my web another dataset (project_data2.xl) that contains information aboutthousands of US public companies ’ financials—items from firms’ balance sheets, incomestatements,and statements of cash flows.

Let’s define variables as the following:

ROA=earningsBeforeInterestAndTaxes/totalassetsnext_roa= next year’s ROAdebtratio=totalLiabilities/totalassetsif totalStockholderEquity>0 thenroe=netIncome/totalStockholderEquityaccrual= (operatingIncomeorLoss-

totalCashFlowFromOperatingActivi)/totalassetsReturn_sameyear=stock return over the calendar year.

! For your convenience, I have calculated the above variables for you and set outliers’ valuesto missing. (An outlier is an observation which has very extreme value relative to the sampledistribution.)

For each of the following questions, in addition to answering questions, please also report estimatedcoefficients (including intercepts), p-values (or t-values), and R-squared, if applicable.

Page 3: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

3

2. (a) Find the relation between this year’s accrual and next year ROA. Is the coefficient onACCRUAL negative or positive? Is it significant?

NEXT_ROA = a + c*ACCRUAL

2 (b) Find the relation between this year’s accrual and next year ROA after controlling for thisyear’s ROA (This means we also include this year’s ROA in the regression). Is the coefficient onROA significantly positive? Is the coefficient on ACCRUAL negative or positive? Is it significant?

NEXT_ROA = a + b*ROA + c*ACCRUAL

2 (c) Find the relation between ROA, ACCRUAL and the contemporaneous stock returns overthe calendar year (Return_sameyear). Is the coefficient on ROA significantly positive? Why? Is thecoefficient on ACCRUAL significant? What does the result mean?

Return_sameyear = a + b*ROA + c*ACCRUAL

Project Tips below:

1. Load in project_data1.xls. Click Cancel when you see below.

Page 4: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

4

2. Open the data file.3. When opening the file, make sure you specify Files of type, “All Files”. Otherwise you may

not be able to see the excel file.

4. Click OK when you see this.

5. You should now see two windows that look like this.

Page 5: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

5

6. To compute the excess returns, pull down the menu “Transform” and “Compute variable”.

7. Compute the market excess return and excess returns of the four stocks.

Page 6: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

6

8. Then pull down the menu “Analyze”, “Descriptive Statistics”, “Descriptive…”9. Choose variables of our interest. Then click OK.

10. Now you get results in the output window. Right click the results and paste it in to aword file for your report.

11. To draw the Q-Q plot, click the pull down menu “Analyze”, “Descriptive Statistics”, “Q-Qplots…”Add variables, and then click “OK”.

Page 7: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

7

12. Paste the results to a word file.

13. To get correlations, pull down “Analyze”, “Correlates”, “Bivariate”. When you seethe window below after adding relevant variables, make sure you also choose“Spearman”. Spearman means rank correlation.

Page 8: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

8

14. You should see 2 tables looking like this. In each cell the first number is the correlation.

15. To estimate beta, pull down “Analyze”, “Regression”, “Linear…”, Add dependentand independent variable, and click “OK”.

16. In the output, which should look like this, ignore the column “Standardized coefficient”.Also, “Constant” means “intercept.” “Sig.” means p-value. From “Model summary”, report

Page 9: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

9

“R square”. Report the estimated beta for each stock.

Examine whether the intercepts (that is: estimates of Alphas, the coefficient before “const” inthe output) are significantly different from zero. Comment on whether the data support orreject CAPM. (If significant, the evidence would reject CAPM. Otherwise it supportsCAPM.)

17. Calculate the expected return (manually) of each stock as:

Mean raw returns of the Treasury bill+

beta estimates from step above * Mean excess return of the CRSP value-weighted index

18.Firm specific risk can be measured by Standard error of residuals. It is reported in “Modelsummary”, “Std. error of the estimate”Market risk should be measured by ?The fraction of stock return variability that can be explained by market movement is thedegree of Goodness-of-Fit. (Comparing R-square.)

19. Verify whether R-square= correlation (you got them earlier) squared. Not spearmancorrelation.

20. Load in another dataset for part II. Again, choose “All Files” in “Files of Type”.Otherwise you may not see the excel file.

Page 10: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

10

21. Run three separate regressions.

22. Add your explanations in your report.

Page 11: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

11

Below are some tips for GRETL software.1. Load in project_data1.xls

Make sure you choose “All Files” in “Files of type”; otherwise the excel files may not bedisplayed.

2. Click Ok when you see this.

Page 12: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

12

After loading in the data, you will see a window looking like this:

3. Let’s define some excess returns:

Then click OK.

Continue to define all excess returns…until finished.

Page 13: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

13

4.

Highlight all relevant variables and pull down the menu as show in this graph.

5. Copy and paste your output

Page 14: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

14

You can copy and paste to your project report. Please highlight the relevant portion of theoutput. Summarize your answer.

6. Gretl cannot draw Q-Q plot. So let’s draw the histograms instead.

Page 15: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

15

7. To get correlations, highlight all relevant variables and click the following.Unfortunately GRETL cannot compute Spearman correlation. So you just report theordinary correlation.

8. To estimate stock beta...

Page 16: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

16

Click OK. Then copy and paste your result:

Page 17: SFSU FIN822 Project 1online.sfsu.edu/donglin/Project_822.pdf · 1 SFSU FIN822 Project 1 This project can be done in a team of up to 3 people. Your project report must be accompanied

17

Report the estimated beta for each stock.

Examine whether the intercepts (that is: estimates of Alphas, the coefficient before “const” inthe output) are significantly different from zero. Comment on whether the data support orreject CAPM. (If significant, the evidence would reject CAPM. Otherwise it supportsCAPM.)

9. Calculate the expected return (manually) of each stock as:

Mean raw returns of the Treasury bill+

beta estimates from step (b) * Mean excess return of the CRSP value-weighted index

10.Firm specific risk can be measured by standard error of residuals in the regression outputs.Market risk should be measured by ?The fraction of stock return variability that can be explained by market movement is thedegree of Goodness-of-Fit. (Comparing unadjusted R-squared.)