Rhonda Hamilton Case Scenario

Rhonda Hamilton manages the Select Electric Fund. Hamilton is reviewing a research report
written by her colleague Brian Ender about the US electric utility industry. Ender’s report
includes the results of a regression of the monthly return for an electric utility equity index for
the previous 203 months (the dependent variable) against the monthly returns for the S&P 500
Index and the difference between the monthly returns on long-term US government bonds and
one-month US Treasury bills (SPREAD) (the two independent variables).

Hamilton has reviewed Ender’s regression results. She agrees that the S&P 500 and SPREAD
are reasonable independent variables, but she is not convinced of the validity of Ender’s model.
Using Ender’s data, Hamilton tested for and confirmed the presence of conditional
heteroskedasticity. She then ran a regression similar to that run by Ender and corrected for
conditional heteroskedasticity using robust standard errors (i.e., Hansen’s method). Hamilton’s
regression model and relevant statistics are presented in Exhibit 1.

EXHIBIT 1
HAMILTON’S REGRESSION MODEL FOR ELECTRIC UTILITY INDUSTRY

Variable Coefficient t-Statistic p-Value

Constant −0.000069 −0.013 0.99

S&P 500 0.3625 6.190 <0.01

SPREAD 1.0264 4.280 <0.01

R2 0.40

Durbin–Watson statistic 0.84

Correlation between SPREAD and S&P 500 0.30

Hamilton is interested in how closely the S&P 500 predicts the electric utility index returns and
wants to use the regression results as a basis for her prediction. Also, she wants to determine
whether the model has serial correlation. Selected values of the t-distribution are shown in
Exhibit 2.
EXHIBIT 2
SELECTED VALUES OF THE t-DISTRIBUTION (DEGREES OF FREEDOM = DF,
ONE-TAILED PROBABILITIES = p)

DF p = 0.05 p  = 0.025

100 1.660 1.984

110 1.659 1.982

 DF p = 0.05 p  = 0.025

120 1.658 1.980

200 1.653 1.972

∞ 1.645 1.960

Q1. Given Hamilton’s finding regarding heteroskedasticity, the most appropriate conclusion is

that the variance of the error term is correlated with:
1. the independent variables only.
2. both the dependent and independent variables.
3. the dependent variable only.
Q2. If Hamilton assumes that the monthly value for SPREAD is 1.5% and the monthly
value for the S&P 500 is −1.0%, the predicted monthly return for the electric utility equity
index is closest to:
1. −0.49%.
2. 1.17%.
3. 1.89%.
Q3. Based on Exhibits 1 and 2, if the standard error of the coefficient is 0.055 and the
degrees of freedom is 200, the 95% confidence interval for the coefficient on the S&P
500 is closest to:
1. −1.61 to 2.33.
2. 0.25 to 0.47.
3. 0.27 to 0.46.
Q4. Given the information in Exhibit 1, Hamilton’s conclusion that multicollinearity is not
a problem, is most likelybased on the observation that the:
1. model F-value is high and the p-values for the S&P 500 and SPREAD are low.
2. correlation between the S&P 500 and SPREAD is low.
3. model R2 is relatively low.
Q5. The most appropriate conclusion Hamilton can make about whether the model has
serial correlation is that the model errors appear to have:
1. negative serial correlation.
2. no significant serial correlation.
3. positive serial correlation.
Paul Charlent Case Scenario
Paul Charlent works for a London-based merchant bank that specializes in assisting small- and
medium-sized companies in developing markets to place debt and equity issues with US and UK
investors. Charlent is conducting exploratory analysis regarding possible relationships between
developing market equity returns and various US and UK macroeconomic variables. He
regresses monthly total returns of the Bangkok SET Index on one-month Libor (for a US dollar–
denominated contract). The period of the study is from July 2006 to December 2013. To improve
the statistical validity of the variables, for both the SET Index and Libor, Charlent uses the
natural logarithms of one plus the monthly returns in the regression calculation. The results of
the regression are shown in Exhibit 1 and Exhibit 2.

EXHIBIT 1
REGRESSION OF SET INDEX ON LIBOR, SUMMARY OUTPUT

ln(1 + SET) = α + β × ln(1 + Libor) + ε

Regression Statistics  

Multiple R 0.1623

R2 0.0263

Adjusted R2 0.0152

Standard error 0.0729

Observations 89
EXHIBIT 2
REGRESSION OF SET INDEX ON LIBOR, ANALYSIS OF VARIANCE

ln(1 + SET) = α + β × ln(1 + Libor) + ε

Degrees of Sum of Mean F- Significance

  Freedom Squares Square Statistic of F

Regression 1 0.0125 0.0125 2.355 0.1285

Residual 87 0.4624 0.0053    

Total 88 4499.91      
 Standard t- Lower Upper
  Coefficients Error Statistic p-Value 95% 95%

Intercept 0.031 0.015 1.997 0.0489 0.01 0.061

Libor –0.732 0.477 –1.535 0.1285 –1.679 0.216

Charlent suspects that his regression equation might not be well specified. In particular, he is
concerned with the possibility that one or both of the time series in the regression exhibit a unit
root. Using the Engle–Granger approach, he tests the residuals from the regression and rejects
the null hypothesis that the error term has a unit root.

Charlent next regresses the natural logarithm of one plus the SET Index monthly returns on the
natural logarithm of one plus Libor, the natural logarithm of one plus the effective Fed funds
rate, and the $/£ exchange rate. The results are reported in Exhibit 3 and Exhibit 4. Charlent
recalls that the null hypothesis of no positive serial correlation is rejected if the calculated
Durbin–Watson (DW) statistic is less than the lower critical value and that the null hypothesis of
no negative serial correlation is rejected if the calculated DW statistic exceeds 4 minus the lower
critical value.

Exhibit 5 reports the pairwise correlations of the variables used in the second regression.

EXHIBIT 3
REGRESSION OF SET INDEX ON LIBOR, THE FED FUNDS RATE, AND $/£
EXCHANGE RATE, SUMMARY OUTPUT

ln(1 + SET) = α + β × ln(1 + Libor) + β2 × ln(1 + Fed Funds) + β3 × £ + ε

Regression Statistics  

Multiple R 0.5544

R2 0.3073

Adjusted R2 0.2829

Standard error 0.0622

DW statistic 1.9566

DW upper critical value 1.73

DW lower critical value 1.59

Observations 89
EXHIBIT 4
REGRESSION OF SET INDEX ON LIBOR, THE FED FUNDS RATE, AND $/£
EXCHANGE RATE, ANALYSIS OF VARIANCE

ln(1 + SET) = α + β × ln(1 + Libor) + β2 × ln(1 + Fed Funds) + β3 × $/£ + ε

Degrees of Sum of Mean F- Significance

  Freedom Squares Square Statistic of F

Regression 3 0.146 0.0486 12.572 7.03E-07

Residual 85 0.3289 0.0039    

Total 88 0.4749      

Standard t- Lower Upper

  Coefficients Error Statistic p-Value 95% 95%

Intercept 0.152 0.077 1.977 0.0512 –0.001 0.304

Libor –11.199 1.966 –5.697 1.71E- –15.107 –7.291

07

Fed funds 11.07 1.92 5.765 1.28E- 7.252 14.888

07

$/£ –0.063 0.048 –1.293 0.199 –0.159 0.034

EXHIBIT 5
PAIRWISE CORRELATIONS

Variable Libor Fed Funds $/£

Libor 1    

Fed funds 0.9814 1  

$/£ 0.6872 0.6798 1

Geoffrey Small, a colleague of Charlent, comments on the results of the two regressions. Small
states that the highly significant F-statistic of the second regression along with the
increased R2 of the second regression means that the addition of the Fed funds rate and the $/£
exchange rate to the analysis provides more reliable estimates of linear associations than the first
regression.

Q6. Based on the results in Exhibits 1 and 2, the most appropriate interpretation is that:

1. the variation in Libor does not explain the variation in SET Index returns.
2. Libor has a statistically significant linear relationship with returns of the SET Index.
3. there is a small but positive correlation between the SET Index and Libor.
Q7. Using Exhibit 2 and two-tail t-tests to determine whether the coefficients are equal
to zero at the 5% significance level, the null hypotheses are most likely:
1. not rejected for the intercept and rejected for the slope.
2. rejected for the intercept and not rejected for the slope.
3. rejected for both the intercept and the slope.
Q8. Using the regression equation results reported in Exhibit 2, if the value for Libor is
3%, the point estimate of the associated return on the SET Index is closest to:
1. −2.16%.
2. 0.94%.
3. 0.90%.
Q9. The most appropriate conclusion that follows from the result of the Engle–Granger
test is that the two time series are:
1. cointegrated and tests of the estimates of the intercept and slope are thus valid.
2. not cointegrated and tests of the estimates of the intercept and slope are thus valid.
3. cointegrated and tests of the estimates of the intercept and slope are thus not valid.
Q10. Based on Exhibits 3 and 4 and the reported Durbin–Watson (DW) statistic,
the most appropriate conclusion is:
1. significant serial correlation is present and the standard errors are likely to be
underestimated.
2. significant serial correlation is present and the standard errors are likely to be
overestimated.
3. serial correlation is not significant and the standard errors are unbiased.
Q11. Regarding Geoffrey Small’s statement about the second regression, which of the
following is most accurate?
A. It is true that the second regression has substantially greater explanatory power than the
first regression.
B. The second regression displays multicollinearity.
C. The F-statistic of the second regression is likely underestimated.

Eduardo DeMolay Case Scenario

Eduardo DeMolay, a research analyst at Mumbai Securities, is studying the time-series
behavior of price-to-earnings ratios (P/Es) computed with trailing 12-month earnings
(Etrailing). He and his assistant, Deepa Kamini, are reviewing the results of the ordinary
least squares time series regression shown in Exhibit 1.
EXHIBIT 1
RESULTS OF REGRESSION OF P/E ON LAGGED P/E (P/Et = b0 + b1P/Et−1 + εt)
Standard
  Coefficient Error t Significance of t

Constant (b0) 0.143 0.153 0.935 0.176

Lagged 0.991 0.003 292.958 0

P/E (b1)

Standard Error of the Durbin– Significance

R 2
Estimate Watson F of F

0.075 1.48978 2.094 130.066 0

DeMolay states: “This regression is a special case of a first-order autoregressive
[AR(1)] model in which the value for b0 is close to zero and the value of b1 is close to 1.
These values suggest that the time series is a random walk.”
Kamini replies: “I’m convinced the P/E series based on trailing earnings truly is a
random walk.”

Kamini and DeMolay next examine the behavior of P/Es calculated using forward 12-
month earnings (Eforward). Kamini estimates another AR(1) model but uses the forward P/E
values this time. She denotes the errors from this second regression as h t. She states:
“The presence of first-order autoregressive conditional heteroskedasticity [ARCH(1)]
errors in this regression is highly likely given the results reported in Exhibit 2.”
EXHIBIT 2
RESULTS OF REGRESSION OF SQUARED RESIDUALS, η2tηt2, ON LAGGED
SQUARED RESIDUALS, η2t−1ηt−12
(η2t=c0+c1η2t−1+ut)(ηt2=c0+c1ηt−12+ut)

Standard
  Coefficient Error t Significance of t

Constant 0.339 0.039 8.768 0

(c0)

Lag 1 (c1) 0.273 0.024 11.405 0

Standard Error of the Durbin– Significance

R 2
Estimate Watson F of F
 (η2t=c0+c1η2t−1+ut)(ηt2=c0+c1ηt−12+ut)

Standard
  Coefficient Error t Significance of t

0.075 1.48978 2.094 130.066 0

After further discussion, DeMolay proposes that he and Kamini incorporate more
variables into the analysis. He suggests they use a variation of the Fed model, in which
the earnings-to-price ratio (E/P) is regressed on long-term interest rates.

DeMolay cautions Kamini: “Remember that when we analyze two time series in
regression analysis, we need to ensure that

1. neither the dependent variable series nor the independent variable series has a unit root,
or
2. that both series have a unit root and are not cointegrated.
Unless Condition 1 or Condition 2 holds, we cannot rely on the validity of the estimated
regression coefficients.”

Q12. DeMolay’s statement that the coefficients depicted in Exhibit 1 are consistent with
a random walk is most likely:
1. correct.
2. incorrect because b1 should be close to 0.
3. incorrect because b0 should be close to 1.
Q13. If Kamini is correct regarding the trailing P/E time series, the best forecast of next
period’s trailing P/E is most likely to be the:
1. current period’s trailing P/E.
2. forecast derived from applying the AR(1) model depicted in Exhibit 1 to the data.
3. average P/E of the time series.
Q14. The results depicted in Exhibit 2 are best described as consistent with a regression
that has ARCH(1) errors because:
1.  c1 is significantly different from 0.
2.  c1 is significantly different from 1.
3.  c0 is significantly different from 0.
Q15. Based on the results depicted in Exhibit 2, DeMolay and Kamini should most
likely model the forward P/E data using a(n):
1. generalized least squares model.
2. AR(1) model.
3. random walk model.
Q16. DeMolay’s caution given in Condition 1 is best described as:
1. correct.
2. incorrect because only the independent variable series needs to be tested for the absence
of a unit root.
3. incorrect because only the dependent variable series needs to be tested for the absence of
a unit root.
Q17. DeMolay’s caution given in Condition 2 is best described as:
1. incorrect because if both series have unit roots, they must exhibit cointegration for the
results of the regression to be valid.
2. incorrect because the regression results are valid whether cointegration exists or does not
exist.
3. correct.

Jordan Garfield Case Scenario

Jordan Garfield, an analyst for a firm that specializes in international equities, is investigating the
behavior of HighTech Inc., a technology stock. He believes its returns should be influenced by
the return on the NASDAQ index, as many analysts suggest. Garfield collects five years of
monthly returns from 2005 to 2009 for the NASDAQ index.

Garfield estimates a simple linear regression using the NASDAQ return to explain the variation
in HighTech’s return. The summary output from this analysis is shown in Exhibit 1.

EXHIBIT 1
GARFIELD’S FIRST REGRESSION MODEL, SUMMARY OUTPUT, REGRESSION
OF HIGHTECH RETURNS ON NASDAQ INDEX RETURNS, 2005–2009

Regression Statistics  

Multiple R 0.737399823

R2 0.543758499

Standard error of estimate  

Observations 60

  Coefficient Standard Error p-value

Intercept 0.001795002 0.007209589 0.804260285

NASDAQ return 1.086005661 0.130620835 0.000000000

 ANOVA Degrees of Freedom (DF) Sum of Squares (SS) Mean Square (MS)

Regression 1 0.214743645 0.214743645

Residual 58 0.180181024 0.003106569

Total 59 0.394924669  
Garfield presents the regression results to the investment committee with the following three
conclusions:

1. The regression intercept is statistically significant.

2. The model explains more than half of the variation in HighTech’s returns.
3. The NASDAQ index return and the HighTech return are positively correlated.
The committee asks Garfield whether he can use the model to predict the return on HighTech’s
stock. Ram Gupta, a committee member, asks: “What would HighTech’s return be in a month
when the return on the NASDAQ index is 0.05633?”

Another committee member, Riko Samora, thinks that the simple regression model omits
important factors that might affect HighTech’s performance. Samora believes that because more
than 40% of HighTech’s customers are in Tokyo, the value of the Japanese currency should
influence HighTech’s sales and that the model’s significance would considerably improve if
Garfield considers this fact.

Following Samora’s suggestion, Garfield runs a multiple linear regression adding the change in
the JPY/USD exchange rate as a second independent variable. The results from this regression
are shown in Exhibit 2.

EXHIBIT 2
GARFIELD’S SECOND REGRESSION MODEL, SUMMARY OUTPUT, REGRESSION
OF HIGHTECH RETURNS ON NASDAQ INDEX RETURNS AND JPY/USD
CHANGES, 2005–2009
 

Regression Statistics  

Multiple R 0.753840729

R2 0.568275844

Adjusted R2 0.553127628

Standard error of estimate 0.054691883

 Regression Statistics  

Durbin–Watson (DW) 2.02

Observations 60

  Coefficients Standard Error t-Statistic p-Value

Intercept 0.000214 0.007127649 0.030025 0.976152

NASDAQ return 1.122096 0.130216256 8.617173 0.000000

JPY/USD change 0.2864262 0.291700144 0.981919 0.330289

ANOVA Degrees of Freedom (DF) Sum of Squares (SS) Mean Square (MS)

Regression 2 0.224426149 0.112213075

Residual 57 0.170498519 0.002991202

Total 59 0.394924669  
Garfield presents the new results to Samora, who asks him two questions:

1. Are the results of this second regression significant?

2. Do you suspect that the model has problems with multicollinearity or serial correlation?
Garfield responds to the Samora’s questions by examining the F-, t-, and DW statistics in the
regression output to see whether they are significant.

Q18. The standard error of estimate of the regression model shown in Exhibit 1 is closest to:
1. 0.0031.
2. 0.1802.
3. 0.0557.
Q19. Which of Garfield’s conclusions to the investment committee about the findings
from his first model (Exhibit 1) is least likely correct? Conclusion:
1. 1
2. 2
3. 3
Q20. In response to Gupta’s question about predicting HighTech’s return, Garfield’s prediction
(in decimal form) will be closest to:
1. 0.06118.
2. 0.04154.
3. 0.06297.
Q21. Using the results shown in Exhibit 2, the value of the F-statistic is closest to:
1. 9.63.
2. 37.51.
3. 16.76.
Q22. Based on the results of the regression model shown in Exhibit 2,
the best  conclusion Garfield can make about a hypothesis that the coefficient JPY/USD
change is zero is to:
1. reject the alternative hypothesis.
2. reject the null hypothesis.
3. fail to reject the null hypothesis.
Q23. In preparing his response to Samora’s second question,
Garfield’s most appropriate conclusion is that the model:
1. has multicollinearity but not serial correlation.
2. has serial correlation but not multicollinearity.
3. does not have either multicollinearity or serial correlation.

Jorge Reyes Case Scenario

Jorge Reyes is a financial analyst with Valores de Playa SA de CV, located in a suburb of
Mexico City, Mexico. Two nights a week he works as an adjunct professor at a local technical
institute, lecturing on investments and serving as a consultant in statistics and related fields.

During one of his lectures, Reyes points out that regression plays an important part in many
empirical studies in finance. As an exercise, Reyes presents the results of a regression of returns
(Rt) on the company that owns the Mexican stock exchange (ticker symbol BOLSAA.MX)
against the US dollar/Mexican peso exchange rate (Et). The data cover the period from late 2011
through early 2012. There are 64 daily observations in the study. Exhibit 1 reports the results of
the regression.
Exhibit 1
Regression Results: Rt = b0 + b1Et + εt

  Standard
Coefficient Error
Constant (b0) 0.0011 0.0019
USD/MXN exchange rate (b1) −0.5789 0.2221
Number of observations used in the regression 64
Critical t-value at the 5% level of significance (two-tailed test that the 2
coefficient equals zero)
 Standard Error of the Durbin– F- Significance of F-
R 2
Adjusted R 2
Estimate Watson Value Value
0.098 0.0842 0.0153 2.3434 6.7927 0.0114
7
One of the students asks Reyes about the adjusted R2 reported in Exhibit 1. Reyes explains that
adjusted R2 adjusts for the effects of serial correlation in the data.
A second student recalls that the presence of heteroskedasticity affects interpretation of the test
statistics computed by a regression. Reyes confirms that that is true and suggests the students
examine a plot of the predicted BOLSAA return values minus their actual values (the BOLSAA
residuals) against the independent variable (USD/MXN exchange rate). Exhibit 2 provides such
a graph.

Exhibit 2
Plot of BOLSAA Residuals against the USD/MXN Exchange Rate

Interpreting the graph, Reyes states:

The presence of heteroskedasticity is indicated when there is a systematic relationship between

the values of the residuals and the independent variable. As shown in Exhibit 2, there is no
systematic relationship between the BOLSAA residuals and the USD/MXN exchange rate.
Therefore, heteroskedasticity does not appear to be a problem in this regression.
In a later exercise, Reyes asks his students to consider a time series of weekly prices of Maya 22
crude oil. A substantial proportion of Mexico’s oil production is Maya 22 heavy crude. The
period of the study is from January 1997 to July 2008. At Reyes’s suggestion, the students first
model the prices as an exponential trend (log-linear model). They test for correlated errors from
the model using the Durbin–Watson statistic. The results are reported in Exhibit 3.

Exhibit 3
Durbin–Watson Test: Log-Linear Model

Durbin–Watson test statistic 3.97

Durbin–Watson critical values for the null hypothesis that:
• there is no positive serial correlation (at the 5% level) 1.6 1.69
5
• there is no negative serial correlation (at the 5% level) 2.35
Reyes next suggests they use a first-order autoregressive model [AR(1)]. To reduce the effect of
the exponential trend, the students continue to use the natural logarithms of the prices, but now
they also take the first differences of these logarithms of the prices (xt). They fit an AR(1) to the
differences of logs. The results of the regression are reported in Exhibit 4.
Exhibit 4
Regression Results: xt+1 = b0 + b1xt + εt+1 where xt = ln(Pt+1) – ln(Pt)

 
Coefficient Standard Error t-Statistic
Constant (b0) 0.002381 0.002056 1.1582
xt (b1) 0.235546 0.039778 5.9214
Number of observations used in the 599
regression
Because nonstationarity or heteroskedasticity would negatively affect use of the AR(1) model,
Reyes asks the students to test for the presence of each. Results of the unit root test for
nonstationarity and of a test for the presence of heteroskedasticity are reported in Exhibit 5.

Exhibit 5
Unit Root Test for Nonstationarity and the Test for Heteroskedasticity

Unit root test statistic −18.7402

Unit root test critical value at the 5% level of significance −2.89
Heteroskedasticity test statistic 2.016733
Heteroskedasticity test critical value at the 5% level of significance 1.96

Q24. In the regression of the returns of BOLSAA.MX against the USD/MXN exchange rate
(Exhibit 1), the coefficient of the USD/MXN exchange rate is most accurately described as:
1. indeterminate because Exhibit 1 provides insufficient information.
2. significantly different from zero.
3. not significantly different from zero.
Q25. Reyes’s explanation regarding Adjusted R2 is best characterized as:
1. correct.
2. incorrect, because adjusted R2 adjusts for the loss of degrees of freedom when additional
independent variables are added to a regression.
3. incorrect, because adjusted R2 adjusts for heteroskedasticity in the independent variables.
Q26. Reyes’s interpretation of the graph in Exhibit 2 is best described as:
1. correct.
2. incorrect, because the effects of heteroskedasticity are, in a regression such as this one,
hidden by the negative slope of the regression line.
3. incorrect, because heteroskedasticity is indicated when there is not a systematic
relationship between the residuals and the independent variable.
Q27. The Durbin–Watson test reported in Exhibit 3 is most accurately interpreted as
indicating that the correlation in the errors is:
1. insignificant.
2. significantly positive.
3. significantly negative.
Q28. Based on the regression results reported in Exhibit 4, the mean-reverting level of
the differences of logarithms of the Maya 22 prices [i.e., the time series as modeled in
the AR(1) model] is closest to:
1. 0.30812.
2. 0.00239.
3. 0.00311.
Q29. Based on the results reported in Exhibit 5, the AR(1) model is best described as
having:
1. a unit root.
2. heteroskedasticity in the error term variance.
3. reliable standard errors.

Julie Moon is an energy analyst examining electricity, oil, and natural gas consumption
in different regions over different seasons. She ran a regression explaining the variation
in energy consumption as a function of temperature. The total variation of the
dependent variable was 140.58, the explained variation was 60.16, and the unexplained
variation was 80.42. She had 60 monthly observations.

Q30. Compute the coefficient of determination.

Q31. What was the sample correlation between energy consumption and temperature?
Q32. Compute the standard error of the estimate of Moon’s regression model.
Q33. Compute the sample standard deviation of monthly energy consumption.
You are examining the results of a regression estimation that attempts to explain the
unit sales growth of a business you are researching. The analysis of variance output for
the regression is given in the table below. The regression was based on five
observations (n = 5).
 d Significance 
ANOVA f SS MSS F F
Regression 1 88.0 88.0 36.667 0.00904
Residual 3 7.2 2.4    
Total 4 95.2      
Q34. How many independent variables are in the regression to which the ANOVA
refers?
Q35. Define Total SS.
1. The sum of the squared deviations of the dependent variable Y about its mean.
2. The part of total sum of squares explained by the regression.
3. The fraction of total variation that is explained by the independent variable.
4. The ratio of the average regression sum of squares to the average sum of squared errors.

Q36. Calculate the sample variance of the dependent variable using information in the
above table.
Q37. Define Regression SS.
Q38. Explain how its value of 88 is obtained in terms of other quantities reported in the above
table.
Q39. What hypothesis does the F-statistic test?
1. The F-statistic tests whether all the slope coefficients in a linear regression are equal to 0.
2. The F-statistic tests whether all the slope coefficients in a linear regression are equal to
0.5.
3. The F-statistic tests whether all the slope coefficients in a linear regression are equal to 1.
4. The F-statistic tests whether all the slope coefficients in a linear regression are equal to
1.5

Q40. Explain how the value of the F-statistic of 36.667 is obtained in terms of other quantities
reported in the above table.
Q41. Is the F-test significant at the 5 percent significance level?
1. Yes.
2. No.
3. There's not enough information to determine this.

An economist collected the monthly returns for KDL’s portfolio and a diversified stock
index. The data collected are shown below:

Mont
h Portfolio Return (%) Index Return (%)
1 1.11 −0.59
2 72.10 64.90
Mont
h Portfolio Return (%) Index Return (%)
3 5.12 4.81
4 1.01 1.68
5 −1.72 −4.97
6 4.06 −2.06
The economist calculated the correlation between the two returns and found it to be
0.996. The regression results with the KDL return as the dependent variable and the
index return as the independent variable are given as follows:

Regression    
Statistics

Multiple R 0.996  

R-squared 0.992  

Standard error 2.861  

Observations 6  

Significance 
ANOVA df SS MSS F F
Regression 1 4101.62 4101.62 500.79 0
Residual 4 32.76 8.19    
Total 5 4134.38      

  Coefficient
s Standard Error t-Statistic p-Value
Intercept 2.252 1.274 1.768 0.1518
Slope 1.069 0.0477 22.379 0
When reviewing the results, Andrea Fusilier suspected that they were unreliable. She
found that the returns for Month 2 should have been 7.21 percent and 6.49 percent,
instead of the large values shown in the first table. Correcting these values resulted in a
revised correlation of 0.824 and the revised regression results shown as follows:
Regression    
Statistics

Multiple R 0.824  

R-squared 0.678  

Standard error 2.062  

Observations 6  

Significance 
ANOVA df SS MSS F F
Regression 1 35.89 35.89 8.44 0.044
Residual 4 17.01 4.25    
Total 5 52.91      

  Coefficient
s Standard Error t-Statistic p-Value
Intercept 2.242 0.863 2.597 0.060
Slope 0.623 0.214 2.905 0.044
Q42. Explain how the bad data affected the results.

Kenneth McCoin, CFA, is a fairly tough interviewer. Last year, he handed each job applicant a
sheet of paper with the information in the following table, and he then asked several questions
about regression analysis. Some of McCoin’s questions, along with a sample of the answers he
received to each, are given below. McCoin told the applicants that the independent variable is the
ratio of net income to sales for restaurants with a market cap of more than $100 million and the
dependent variable is the ratio of cash flow from operations to sales for those restaurants. Which
of the choices provided is the best answer to each of McCoin’s questions?

Regression Statistics  

Multiple R 0.8623

R-squared 0.7436

Standard error 0.0213

Observations 24
 ANOVA df SS MSS F Significance F

Regression 1 0.029 0.029000 63.81 0

Residual 22 0.010 0.000455    

Total 23 0.040      

  Coefficients Standard Error t-Statistic p-Value

Intercept 0.077 0.007 11.328 0

Slope 0.826 0.103 7.988 0

Q43. What is the value of the coefficient of determination?

1. 0.8261.
2. 0.7436.
3. 0.8623.

Q44. Suppose that you deleted several of the observations that had small residual values. If you
re-estimated the regression equation using this reduced sample, what would likely happen to the
standard error of the estimate and the R-squared?
  Standard Error of the Estimate   R-Squared  

A Decrease   Decrease  

B Decrease   Increase  

C Increase   Decrease

Q45. What is the correlation between X and Y?

1. −0.7436.
2. 0.7436.
3. 0.8623.

Q46. Where did the F-value in the ANOVA table come from?

1. You look up the F-value in a table. The F depends on the numerator and denominator
degrees of freedom.
2. Divide the “Mean Square” for the regression by the “Mean Square” of the residuals.
3. The F-value is equal to the reciprocal of the t-value for the slope coefficient.

Q47. If the ratio of net income to sales for a restaurant is 5 percent, what is the predicted
ratio of cash flow from operations to sales?
1. 0.007 + 0.103(5.0) = 0.524.
2. 0.077 − 0.826(5.0) = −4.054.
3. 0.077 + 0.826(5.0) = 4.207.

Q48. Is the relationship between the ratio of cash flow to operations and the ratio of net
income to sales significant at the 5 percent level?
1. No, because the R-squared is greater than 0.05.
2. No, because the p-values of the intercept and slope are less than 0.05.
3. Yes, because the p-values for F and t for the slope coefficient are less than 0.05.

Howard Golub, CFA, is preparing to write a research report on Stellar Energy Corp. common
stock. One of the world’s largest companies, Stellar is in the business of refining and marketing
oil. As part of his analysis, Golub wants to evaluate the sensitivity of the stock’s returns to
various economic factors. For example, a client recently asked Golub whether the price of Stellar
Energy Corporation stock has tended to rise following increases in retail energy prices. Golub
believes the association between the two variables to be negative, but he does not know the
strength of the association.

Golub directs his assistant, Jill Batten, to study the relationships between Stellar monthly
common stock returns versus the previous month’s percent change in the US Consumer Price
Index for Energy (CPIENG), and Stellar monthly common stock returns versus the previous
month’s percent change in the US Producer Price Index for Crude Energy Materials (PPICEM).
Golub wants Batten to run both a correlation and a linear regression analysis. In response, Batten
compiles the summary statistics shown in Exhibit 1 for the 248 months between January 1980
and August 2000. All of the data are in decimal form, where 0.01 indicates a 1 percent return.
Batten also runs a regression analysis using Stellar monthly returns as the dependent variable and
the monthly change in CPIENG as the independent variable. Exhibit 2 displays the results of this
regression model.

Exhibit 1
Descriptive Statistics

  Lagged Monthly
Change
Monthly Return Stellar Common
 
Stock CPIENG PPICEM
Mean 0.0123 0.0023 0.0042
  Lagged Monthly
Change
Monthly Return Stellar Common
 
Stock CPIENG PPICEM
Standard Deviation 0.0717 0.0160 0.0534
       
Covariance, Stellar vs.
−0.00017    
CPIENG
Covariance, Stellar vs.
−0.00048    
PPICEM
Covariance, CPIENG vs.
0.00044    
PPICEM
Correlation, Stellar vs.
−0.1452    
CPIENG
Exhibit 2
Regression Analysis with CPIENG

Regression Statistics

Multiple R 0.1452  

R-squared 0.0211  

Standard error of the 0.0710  

estimate

Observations 248  

Coefficient
 
s Standard Error t-Statistic
Intercept 0.0138 0.0046 3.0275
Slope coefficient −0.6486 0.2818 −2.3014

Q49. Did Batten’s regression analyze cross-sectional or time-series data, and what was the
expected value of the error term from that regression?

  Expected Value of Error  

Data Type   Term
 A Time-series   0  

B Time-series   εi  

C Cross-sectional   0
Q50. Based on the regression, which used data in decimal form, if the
CPIENG decreases by 1.0 percent, what is the expected return on Stellar common stock
during the next period?
1. 0.0073 (0.73 percent).
2. 0.0138 (1.38 percent).
3. 0.0203 (2.03 percent).
Q51. Based on Batten’s regression model, the coefficient of determination indicates that:
1. Stellar’s returns explain 2.11 percent of the variability in CPIENG.
2. Stellar’s returns explain 14.52 percent of the variability in CPIENG.
3. Changes in CPIENG explain 2.11 percent of the variability in Stellar’s returns.

Q52. Based on Batten’s regression model, the coefficient of determination indicates that:

1. Stellar’s returns explain 2.11 percent of the variability in CPIENG.
2. Stellar’s returns explain 14.52 percent of the variability in CPIENG.
3. Changes in CPIENG explain 2.11 percent of the variability in Stellar’s returns.
Q53. For Batten’s regression model, the standard error of the estimate shows that the
standard deviation of:
1. the residuals from the regression is 0.0710.
2. values estimated from the regression is 0.0710.
3. Stellar’s observed common stock returns is 0.0710.

Q54. For the analysis run by Batten, which of the following is an incorrect conclusion

from the regression output?
1. The estimated intercept coefficient from Batten’s regression is statistically significant at
the 0.05 level.
2. In the month after the CPIENG declines, Stellar’s common stock is expected to exhibit a
positive return.
3. Viewed in combination, the slope and intercept coefficients from Batten’s regression are
not statistically significant at the 0.05 level.
Anh Liu is an analyst researching whether a company’s debt burden affects investors’ decision to
short the company’s stock. She calculates the short interest ratio (the ratio of short interest to
average daily share volume, expressed in days) for 50 companies as of the end of 2016 and
compares this ratio with the companies’ debt ratio (the ratio of total liabilities to total assets,
expressed in decimal form).
Liu provides a number of statistics in Exhibit 1. She also estimates a simple regression to
investigate the effect of the debt ratio on a company’s short interest ratio. The results of this
simple regression, including the analysis of variance (ANOVA), are shown in Exhibit 2.

In addition to estimating a regression equation, Liu graphs the 50 observations using a

scatterplot, with the short interest ratio on the vertical axis and the debt ratio on the horizontal
axis.

Exhibit 1
Summary Statistics

Statisti Debt Ratio Short Interest Ratio

c Xi Yi

Sum 19.8550 192.3000

Averag 0.3971 3.8460

e

Sum of
squared
deviatio ∑i=1n(Xi−X⎯⎯⎯)2=2.2225∑i=1n(Xi−X ∑i=1n(Yi−Y⎯⎯⎯)2=412.2042∑i=1n(Yi−Y¯
ns from ¯)2=2.2225 )2=412.2042
the
mean

Sum of
cross-
product
s of
∑i=1n(Xi−X⎯⎯⎯)(Yi−Y⎯⎯⎯)=−9.2430∑i=1n(Xi−X¯)(Yi−Y¯)=−9.2430
deviatio
ns from
the
mean
 

Exhibit 2
Regression of the Short Interest Ratio on the Debt Ratio

ANOVA Degrees of Freedom (df) Sum of Squares (SS) Mean Square (MS)

Regression 1 38.4404 38.4404

 ANOVA Degrees of Freedom (df) Sum of Squares (SS) Mean Square (MS)

Residual 48 373.7638 7.7867

Total 49 412.2042  
 

Regression Statistics     

Multiple R 0.3054    

R2 0.0933    

Standard error of estimate 2.7905    

Observations 50    

       

  Coefficients Standard Error t-Statistic

Intercept 5.4975 0.8416 6.5322

Debt ratio –4.1589 1.8718 –2.2219

 

Liu is considering three interpretations of these results for her report on the relationship between
debt ratios and short interest ratios:

Interpretation 1 Companies’ higher debt ratios cause lower short interest ratios.
Interpretation 2 Companies’ higher short interest ratios cause higher debt ratios.
Interpretation 3 Companies with higher debt ratios tend to have lower short interest ratios.

She is especially interested in using her estimation results to predict the short interest ratio for
MQD Corporation, which has a debt ratio of 0.40.

Q55. Based on Exhibits 1 and 2, if Liu were to graph the 50 observations, the scatterplot
summarizing this relation would be best  described as:
1. horizontal.
2. upward sloping.
3. downward sloping.
Q56. Based on Exhibit 1, the sample covariance is closest to:
1. −9.2430.
2. −0.1886.
3. 8.4123.
Q57. Based on Exhibit 1, the correlation between the debt ratio and the short interest
ratio is closest to:
1. −0.3054.
2. 0.0933.
3. 0.3054.
Q58. Which of the interpretations best describes Liu’s findings for her report?
1. Interpretation 1
2. Interpretation 2
3. Interpretation 3
Q59. The dependent variable in Liu’s regression analysis is the:
1. intercept.
2. debt ratio.
3. short interest ratio.
Q60. Based on Exhibit 2, the degrees of freedom for the t-test of the slope coefficient in
this regression are:
1. 48.
2. 49.
3. 50.
Q61. The upper bound for the 95% confidence interval for the coefficient on the debt
ratio in the regression is closestto:
1. −1.0199.
2. −0.3947.
3. 1.4528.
Q62. Which of the following should Liu conclude from these results shown in Exhibit 2?
1. The average short interest ratio is 5.4975.
2. The estimated slope coefficient is statistically significant at the 0.05 level.
3. The debt ratio explains 30.54% of the variation in the short interest ratio.
Q63. Based on Exhibit 2, the short interest ratio expected for MQD Corporation is closest
to:
1. 3.8339.
2. 5.4975.
3. 6.2462.
Q64. Based on Liu’s regression results in Exhibit 2, the F-statistic for testing whether the
slope coefficient is equal to zero is closest  to:
1. −2.2219.
2. 3.5036.
3. 4.9367.
Elena Vasileva recently joined EnergyInvest as a junior portfolio analyst. Vasileva’s supervisor
asks her to evaluate a potential investment opportunity in Amtex, a multinational oil and gas
corporation based in the US. Vasileva’s supervisor suggests using regression analysis to examine
the relation between Amtex shares and returns on crude oil.

Vasileva notes the following assumptions of regression analysis:

Assumption 1 The error term is uncorrelated across observations.

Assumption 2 The variance of the error term is the same for all observations.
Assumption 3 The expected value of the error term is equal to the mean value of the dependent variable.

Vasileva runs a regression of Amtex share returns on crude oil returns using the monthly data she
collected. Selected data used in the regression are presented in Exhibit 1, and selected regression
output is presented in Exhibit 2.

Exhibit 1
Selected Data for Crude Oil Returns and Amtex Share Returns

Predict
ed
Amtex Amtex Regression Squared
Oil Retur Cross-Product Return Residual Residual
Return n (Xi−X⎯⎯⎯)(Yi−Y⎯⎯⎯) (Yˆ) (Yi−Yˆ) (Yi−Yˆ)2(Yi−
 
(Xi) (Yi) (Xi−X¯)(Yi−Y¯) (Y^) (Yi−Y^) Y^)2
Mont −0.0320 0.0331 −0.000388 0.0020 −0.031134 0.000969
h1 00 45 11
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
Mont 0.02863 0.0623 0.002663 0.0162 −0.046053 0.002121
h 36 6 34 82
Sum     0.085598     0.071475
Avera −0.0180 0.0052        
ge 56 93

Exhibit 2
Selected Regression Output
Dependent Variable: Amtex Share Return

Coefficien
  t Standard Error
Intercept 0.0095 0.0078
Oil return 0.2354 0.0760
Note: The critical t-value for a one-sided t-test at the 5% significance level is 1.691.

Vasileva expects the crude oil return next month, Month 37, to be −0.01. She computes the
variance of the prediction error to be 0.0022.
Q65. Which of Vasileva’s assumptions regarding regression analysis is incorrect?
1. Assumption 1
2. Assumption 2
3. Assumption 3
Q66. Based on Exhibit 1, the standard error of the estimate is closest to:
1. 0.044558.
2. 0.045850.
3. 0.050176.
Q67. Based on Exhibit 2, Vasileva should reject the null hypothesis that:
1. the slope is less than or equal to 0.15.
2. the intercept is less than or equal to 0.
3. crude oil returns do not explain Amtex share returns.
Q68. Based on Exhibit 2, Vasileva should compute the:
1. coefficient of determination to be 0.4689.
2. 95% confidence interval for the intercept to be –0.0037 to 0.0227.
3. 95% confidence interval for the slope coefficient to be 0.0810 to 0.3898.
Q69. Based on Exhibit 2 and Vasileva’s prediction of the crude oil return for month 37,
the estimate of Amtex share return for month 37 is closest to:
1. –0.0024.
2. 0.0071.
3. 0.0119.
Q70. Using information from Exhibit 2, Vasileva should compute the 95% prediction
interval for Amtex share return for month 37 to be:
1. –0.0882 to 0.1025.
2. –0.0835 to 0.1072.
3. 0.0027 to 0.0116.
Doug Abitbol is a portfolio manager for Polyi Investments, a hedge fund that trades in the United
States. Abitbol manages the hedge fund with the help of Robert Olabudo, a junior portfolio
manager.

Abitbol looks at economists’ inflation forecasts and would like to examine the relationship
between the US Consumer Price Index (US CPI) consensus forecast and actual US CPI using
regression analysis. Olabudo estimates regression coefficients to test whether the consensus
forecast is unbiased. Regression results are presented in Exhibit 1. Additionally, Olabudo
calculates the 95% prediction interval of the actual CPI using a US CPI consensus forecast of
2.8.

Exhibit 1
Regression Output: Estimating US CPI

     
Regression Statistics
Multiple R 0.9929    
R-squared 0.9859    
      
Regression Statistics
Standard error of estimate 0.0009    
Observations 60    

 
Coefficients Standard Error t-Statistic
Intercept 0.0001 0.0002 0.5351
US CPI consensus forecast 0.9830 0.0155 63.6239
Notes:

1.  
The absolute value of the critical value for the t-statistic is 2.0 at the 5% level of significance.

2.  
The standard deviation of the US CPI consensus forecast is sx = 0.7539.
3. The mean of US CPI consensus forecast is X⎯⎯⎯X¯ = 1.3350.

To conclude their meeting, Abitbol and Olabudo discuss the limitations of regression analysis.
Olabudo notes the following limitations of regression analysis:

 Limitation 1: Public knowledge of regression relationships may negate their future

usefulness.
 Limitation 2: Hypothesis tests and predictions based on linear regression will not be valid
if regression assumptions are violated.

Q71. Based on Exhibit 1, Olabudo should:

1. conclude that the inflation predictions are unbiased.
2. reject the null hypothesis that the slope coefficient equals 1.
3. reject the null hypothesis that the intercept coefficient equals 0.
Q72. Based on Exhibit 1, Olabudo should calculate a prediction interval for the actual
US CPI closest to:
1. 2.7506 to 2.7544.
2. 2.7521 to 2.7529.
3. 2.7981 to 2.8019.
Q73. Which of Olabudo’s noted limitations of regression analysis is correct?
1. Only Limitation 1
2. Only Limitation 2
3. Both Limitation 1 and Limitation 2
With many US companies operating globally, the effect of the US dollar’s strength on a US
company’s returns has become an important investment issue. You would like to determine
whether changes in the US dollar’s value and overall US equity market returns affect an asset’s
returns. You decide to use the S&P 500 Index to represent the US equity market.
Q74. Select the correct multiple regression equation that tests whether changes in the value of
the dollar and equity market returns affect an asst's returns. Use the notations below.
Rit = return on the asset in period t
RMt = return on the S&P 500 in period t
ΔXt = change in period t in the log of a trade-weighted index of the foreign exchange value of
US dollar against the currencies of a broad group of major US trading partners.
1.   Rit  =  b0  +  b1RMt  +  b2ΔXt  + εit

2.   Rit  =  b0  × (b1RMt/b2ΔXt) + εit

3.   Rit  =  b1RMt  +  b2ΔXt  + εit

You estimate the regression for Archer Daniels Midland Company (NYSE: ADM). You
regress its monthly returns for the period January 1990 to December 2002 on S&P 500
Index returns and changes in the log of the trade-weighted exchange value of the US
dollar. The table below shows the coefficient estimates and their standard errors.
Coefficient Estimates from Regressing ADM’s Returns: Monthly Data, January 1990–
December 2002

 
Coefficient Standard Error
Intercep 0.0045 0.0062
t
RMt 0.5373 0.1332
ΔXt −0.5768 0.5121
n = 156    
Source: FactSet, Federal Reserve Bank of Philadelphia.

Q75. Determine whether S&P 500 returns affect ADM’s returns. Use a 0.05 significance
level to make your decision.
1. S&P 500 Index returns do not affect ADM’s returns
2. S&P 500 Index returns do affect ADM’s returns
3. There is not enough information to determine whether S&P 500 returns affect ADM’s
returns.
Q76. Determine whether changes in the value of the US dollar affect ADM’s returns.
Use a 0.05 significance level to make your decision.
1. Changes in the value of the US dollar do not affect ADM’s returns.
2. Changes in the value of the US dollar do affect ADM’s returns.
3. There is not enough information to determine whether changes in the value of the US
dollar affect ADM’s returns.
With many US companies operating globally, the effect of the US dollar’s strength on a US
company’s returns has become an important investment issue. You would like to determine
whether changes in the US dollar’s value and overall US equity market returns affect an asset’s
returns. You decide to use the S&P 500 Index to represent the US equity market.
Q77. Based on the estimated coefficient on RMt, is it correct to say that “for a 1 percentage point
increase in the return on the S&P 500 in period t, we expect a 0.5373 percentage point increase
in the return on ADM”?
1. Yes
2. No
One of the most important questions in financial economics is what factors determine the cross-
sectional variation in an asset’s returns. Some have argued that book-to-market ratio and size
(market value of equity) play an important role.

Q78. Select the correct multiple regression equation to test whether book-to-market ratio and
size explain the cross-section of asset returns. Use the notations below.
(B/M)i = book-to-market ratio for asset i
Ri = return on asset i in a particular month
Sizei = natural log of the market value of equity for asset i
1. Ri  =  b0  +  b1(B/M)i  +  b2Sizei

2.  Ri = b0 + b1(B/M)i + b2Sizei + εi.
3.   Ri  =  b1(B/M)i  +  b2Sizei  + εi

One of the most important questions in financial economics is what factors determine the cross-
sectional variation in an asset’s returns. Some have argued that book-to-market ratio and size
(market value of equity) play an important role.

The table below shows the results of the linear regression for a cross-section of 66 companies.
The size and book-to-market data for each company are for December 2001. The return data for
each company are for January 2002.

Results from Regressing Returns on the Book-to-Market Ratio and Size

 
Coefficient Standard Error
Intercep 0.0825 0.1644
t
(B/M)i −0.0541 0.0588
Sizei −0.0164 0.0350
n = 66    
Source: FactSet.

Q79. Determine whether the book-to-market ratio and size are each useful for explaining the
cross-section of asset returns. Use a 0.05 significance level to make your decision.
Select all that apply.

1. Asset size is not useful in explaining the cross-sectional variation of asset returns in this
sample.
2. Asset size is useful in explaining the cross-sectional variation of asset returns in this
sample.
3. The book-to-market ratio is useful in explaining the cross-sectional variation in returns
for this sample.
4. The book-to-market ratio is not useful in explaining the cross-sectional variation in
returns for this sample.

There is substantial cross-sectional variation in the number of financial analysts who follow a
company. Suppose you hypothesize that a company’s size (market cap) and financial risk (debt-
to-equity ratios) influence the number of financial analysts who follow a company. You
formulate the following regression model:(Analyst following)i = b0 + b1Sizei + b2(D/E)i + εi
where

(Analyst following)i = the natural log of (1 + n), where ni is the number of analysts following
company i
Sizei = the natural log of the market capitalization of company i in millions of dollars
(D/E)i = the debt-to-equity ratio for company i
In the definition of Analyst following, 1 is added to the number of analysts following a company
because some companies are not followed by any analysts, and the natural log of 0 is
indeterminate. The following table gives the coefficient estimates of the above regression model
for a randomly selected sample of 500 companies. The data are for the year 2002.
Coefficient Estimates from Regressing Analyst Following on Size and Debt-to-Equity
Ratio

  Standard
Coefficient Error t-Statistic
Intercep −0.2845 0.1080 −2.6343
t
Sizei 0.3199 0.0152 21.0461
(D/E)i −0.1895 0.0620 −3.0565
n = 500      
Source: First Call/Thomson Financial, Compustat.

Q79 . Consider two companies, both of which have a debt-to-equity ratio of 0.75. The first
company has a market capitalization of $100 million, and the second company has a market
capitalization of $1 billion. Based on the above estimates, how many more analysts will follow
the second company than the first company?
Fill in the following blank: _____ more analysts.

Q80. Suppose the p-value reported for the estimated coefficient on (D/E)i is 0.00236. State the
interpretation of 0.00236.
In early 2001, US equity marketplaces started trading all listed shares in minimal
increments (ticks) of $0.01 (decimalization). After decimalization, bid–ask spreads of
stocks traded on the NASDAQ tended to decline. In response, spreads of NASDAQ
stocks cross-listed on the Toronto Stock Exchange (TSE) tended to decline as well.
Researchers Oppenheimer and Sabherwal (2003) hypothesized that the percentage
decline in TSE spreads of cross-listed stocks was related to company size, the
predecimalization ratio of spreads on NASDAQ to those on the TSE, and the
percentage decline in NASDAQ spreads. The following table gives the regression
coefficient estimates from estimating that relationship for a sample of 74 companies.
Company size is measured by the natural logarithm of the book value of company’s
assets in thousands of Canadian dollars.

Coefficient Estimates from Regressing Percentage Decline in TSE Spreads on

Company Size, Predecimalization Ratio of NASDAQ to TSE Spreads, and Percentage
Decline in NASDAQ Spreads
 
Coefficient t-Statistic
Intercept −0.45 −1.86
Sizei 0.05 2.56
(Ratio of spreads)i −0.06 −3.77
(Decline in NASDAQ 0.29 2.42
spreads)i
n = 74    
Source: Oppenheimer and Sabherwal (2003).

Q.81 The average company in the sample has a book value of assets of C$900 million
and a predecimalization ratio of spreads equal to 1.3. Based on the above model, what
is the predicted decline in spread on the TSE for a company with these average
characteristics, given a 1 percentage point decline in NASDAQ spreads?
Fill in the following blank: _____ percent decline.

The “neglected-company effect” claims that companies that are followed by fewer analysts will
earn higher returns on average than companies that are followed by many analysts. To test the
neglected-company effect, you have collected data on 66 companies and the number of analysts
providing earnings estimates for each company. You decide to also include size as an
independent variable, measuring size as the log of the market value of the company’s equity, to
try to distinguish any small-company effect from a neglected-company effect. The small-
company effect asserts that small-company stocks may earn average higher risk-adjusted returns
than large-company stocks.

The table below shows the results from estimating the model Ri = b0 + b1Sizei + b2(Number of
analysts)i + εi for a cross-section of 66 companies. The size and number of analysts for each
company are for December 2001. The return data are for January 2002.
Results from Regressing Returns on Size and Number of Analysts
 
Coefficient Standard Error t-Statistic
Intercept 0.0388 0.1556 0.2495
Sizei −0.0153 0.0348 −0.4388
(Number of analysts)i 0.0014 0.0015 0.8995

ANOVA df SS MSS
Regression 2 0.0094 0.0047
Residual 63 0.6739 0.0107
Total 65 0.6833  
Residual standard 0.1034    
error
R-squared 0.0138    
Observations 66    
Source: First Call/Thomson Financial, FactSet.

Q82. What test would you conduct to see whether the two independent variables
are jointly statistically related to returns (H0: b1 = b2 = 0)?
1.  t-test
2.  z-test
3.  F-test
4.  p-test
5. Two-tailed test
Q83. What information do you need to conduct the appropriate test?
Select all that apply

1. total number of observations,  n

2. sum of squared errors or residuals, ∑i=1n(Yi−Yˆi)2∑i=1n(Yi−Y^i)2 abbreviated SSE, and
3. total number of regression coefficients to be estimated, k + 1
4. regression sum of squares, ∑i=1n(Yˆi−Y⎯⎯⎯)2∑i=1n(Y^i−Y¯)2 abbreviated RSS

Q84. Determine whether the two variables jointly are statistically related to returns at the 0.05
significance level.
Q85. Explain the meaning of adjusted R2.
Q86. Select whether the adjusted R2 for the regression would be smaller than, equal to,
or larger than 0.0138.
1. Smaller than 0.0138.
2. Equal to 0.0138.
3. Larger than 0.0138.
Some developing nations are hesitant to open their equity markets to foreign investment because
they fear that rapid inflows and outflows of foreign funds will increase volatility. In July 1993,
India implemented substantial equity market reforms, one of which allowed foreign institutional
investors into the Indian equity markets. You want to test whether the volatility of returns of
stocks traded on the Bombay Stock Exchange (BSE) increased after July 1993, when foreign
institutional investors were first allowed to invest in India. You have collected monthly return
data for the BSE from February 1990 to December 1997. Your dependent variable is a measure
of return volatility of stocks traded on the BSE; your independent variable is a dummy variable
that is coded 1 if foreign investment was allowed during the month and 0 otherwise.

You believe that market return volatility actually decreases with the opening up of equity
markets. The table below shows the results from your regression.
Results from Dummy Regression for Foreign Investment in India with a Volatility
Measure as the Dependent Variable

  Standard
Coefficient Error t-Statistic
Intercep 0.0133 0.0020 6.5351
t
Dummy −0.0075 0.0027 −2.7604
n = 95      
Source: FactSet.

Q87. State null and alternative hypotheses for the slope coefficient of the dummy variable that
are consistent with testing your stated belief about the effect of opening the equity markets on
stock return volatility.