CFA

CHAPTER 1

MULTIPLE REGRESSION

1. (A) 5.
Explanation
The F-statistic is equal to the ratio of the mean squared regression to the mean
squared error.
F = MSR / MSE = 20 / 4 = 5.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

2. (B) Homoskedasticity
Explanation
Homoskedasticity refers to the basic assumption of a multiple regression model
that the variance of the error terms is constant.
(Module 1.1, LOS 1.c)
Related Material
SchweserNotes - Book 1

3. (C) Incorrectly pooling data.

Explanation
The relationship between returns and the dependent variables can change over
time, so it is critical that the data be pooled correctly. Running the regression for
multiple sub-periods (in this case two) rather than one time period can produce
more accurate results.
(Module 1.3, LOS 1.g)
Related Material
SchweserNotes - Book 1

4. (C) R = a + bM + c1D1 + c2D2 + , where D1 = 1 if the return is from the first

manager, and D2 = 1 if the return is from the third manager.
Explanation
The effect needs to be measured by two distinct dummy variables. The use of
three variables will cause collinearity, and the use of one dummy variable will not
appropriately specify the manager impact.
(Module 1.4, LOS 1.l)
Related Material
SchweserNotes - Book 1

Quantitative Methods 1 Multiple Regression

CFA
5. (B) regression should have higher sum of squares regression as a ratio to the total
sum of squares.
Explanation
The index fund regression should provide a higher R2 than the active manager
regression. R2 is the sum of squares regression divided by the total sum of
squares.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

6. (C) multicollinearity
Explanation
An indication of multicollinearity is when the independent variables individually are
not statistically significant but the F-test suggests that the variables as a whole do
an excellent job of explaining the variation in the dependent variable.
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1

Using a recent analysis of salaries (in $1,000) of financial analysts, Timbadia runs a
regression of salaries on education, experience, and gender. (Gender equals one
for men and zero for women.) The regression results from a sample of 230
financial analysts are presented below, with t-statistics in parenthesis.
Salary = 34.98 + 1.2 Education + 0.5 Experience + 6.3 Gender
(29.11) qw (8.93) (2.98) (1.58)

Timbadia also runs a multiple regression to gain a better understanding of the

relationship between lumber sales, housing starts, and commercial construction.
The regression uses a large data set of lumber sales as the dependent variable
with housing starts and commercial construction as the independent variables. The
results of the regression are:
Coefficient Standard Error t-statistics
Intercept 5.337 1.71 3.14
Housing starts 0.76 0.09 8.44
Commercial construction 1.25 0.33 3.78

Quantitative Methods 2 Multiple Regression

CFA
Finally, Timbadia a regression between the returns on a stock and its industry
index with the following results:
Coefficient Standard Error
Intercept 2.1 2.01
Industry index 1.9 0.31
• Standard error of estimate = 15.1
• Correlation coefficient = 0.849

7. (B) 59.18
Explanation
34.98 + 1.2(16) + 0.5(10) = 59.18
(Module 1.2 LOS 1.f)
Related Material
SchweserNotes - Book 1

8. (B) 9.7%
Explanation
Y = b0 + bX1
Y = 2.1 + 1.9(4) = 9.7%
(Module 1.2 LOS 1.f)
Related Material
SchweserNotes - Book 1

9. (B) 72.1%
Explanation
The coefficient of determination, R2, is the square the correlation coefficient.
0.8492 = 0.721.
(Module 1.2 LOS 1.d)
Related Material
SchweserNotes - Book 1

Ben Sasse is a quantitative analyst at Gurnop Asset Managers. Sasse is

interviewing Victor Sophie for a junior analyst position, Sasse mentions that the
firm currently uses several proprietary multiple regression models and wants
Sophie’s opinion about regression models.
Sophie makes the following statements:
Statements 1: Multiple regression models can be used to forecast independent
variables.
Statement 2: Multiple regression models can be used to test existing theories of
relationships among variables.
Quantitative Methods 3 Multiple Regression
CFA
Sasses then discusses a model that the firm used to forecast credit spread on
investment-grade corporate bonds. Sasse states that while the current model
parameters are a secret, the following is an older version of the model.
CSP = 0.22 + 1.04  DSC – 0.32  index + 1.33  D/E
Where:
CSP = credit spread (%)
DSC = EBITDA/unsecured debt.
Index = 1 if the issuer is part CDX index; 0 otherwise
D/E = long-term debt/equity

10. (C) only Statement 2 is correct.

Explanation
Multiple regression models can be used to identify relations between variables,
forecast the dependent variable, and test existing theories. Statement 1 is
inaccurate in because it mentions forecast independent (and not dependent)
variables.
(Module 1.2 LOS 1.a)
Related Material
SchweserNotes - Book 1

11. (B) The credit spread on the firm's issue will decrease by 32 bps.
Explanation
The coefficient on the index dummy variable is –0.32, and if the variable takes a
value of 1 (inclusion in the index), the credit spread would decrease by 0.32%, or
32 bps.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

12. (B) The dependent variable is not serially correlated.

Explanation
The assumption calls for the residual (or errors) to be not serially correlated. The
dependent variable can have serial correlation. Other assumptions are accurate.
(Module 1.1, LOS 1.c)
Related Material
SchweserNotes - Book 1

Quantitative Methods 4 Multiple Regression

CFA
13. (B) Error term is normally distributed.
Explanation
A normal QQ plot of the residuals can visually indicate violation of the assumption
that the residuals are normally distributed.
(Module 1.1, LOS 1.c)
Related Material
SchweserNotes - Book 1

A real estate agent wants to develop a model to predict the selling price of a
home. The agent believes that the most important variables in determining the
price of a house are its size (in square feet) and the number of bedrooms.
Accordingly, he takes a random sample of 32 homes that has recently been sold.
The results of the regression are:

Coefficient Standard Error t-statistics

Intercept 66,500 59,292 1.12
House Size 74.30 21.11 3.52
Number of Bedrooms 10306 3230 3.19
R2 = 0.56; F = 40.73
Selected F-table values for significance level of 0.05:
1 2
28 4.20 3.34
29 4.18 3.33
30 4.17 3.32
32 4.15 3.29
(Degree of freedom for the numerator in columns; Degree of freedom for the
denominator in rows)
Additional information regarding this multiple regressions.
1. Variance of error is not constant across the 32 observations.
2. The two variables (size of the house and the number of bedrooms) are highly
correlated.
3. The error variance is not correlated with the size of the house nor with the
number of bedrooms.

14. (B) $256,000.

Explanation
66,500 + 74,30(2,000) + 10,306(4) = #256,324
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1
Quantitative Methods 5 Multiple Regression
CFA
15. (B) be rejected as the calculated F of 40.73 is greater than the critical value of 3.33.
Explanation
We can reject the null hypothesis that coefficients of both independent variables
equal 0. The F value for comparison is F2,29 = 3.33. The degrees of freedom in the
numerator is 2; equal to the number of independent variables. Degrees of freedom
for the denominator is 32 – (2 + 1) = 29. The critical value of the F-test needed
to reject the null hypothesis is thus 3.33. The actual value of the F-test statistic is
40.73, so the null hypothesis should be rejected, as the calculated F of 40.73 is
greater than the critical value of 3.33.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

16. (B) Multicollinearity

Explanation
Multicollinearity is present in a regression model when some linear combination of
the independent variables are highly correlated. We are told that the two
independent variables in this question are highly correlated. We also recognize
that unconditional heteroskedasticity is present — but this would not pose any
major problems in using this model for forecasting. No information is given about
autocorrelation in residuals, but this is generally a concern with time series data
(in this case, the model uses cross-sectional data).
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1

Consider a study of 100 university endowment funds that was conducted to

determine if the funds’ annual risk-adjusted returns could be explained by the size
of the fund and the percentage of fund assets that are managed to an indexing
strategy. The equation used to model this relationship is:
ARARi = b0 + b1Sizei + b2Indexi + ei
Where:
ARARi = the average annual risk-adjusted percent returns for the fund i over the
1998-2002 time period.
Sizei = the natural logarithm of the average assets under management for fund i.
Indexi = the percentage of assets in fund i that were managed to an indexing
strategy.

Quantitative Methods 6 Multiple Regression

CFA
The table below contains a portion of the regression results from the study.
Partial Results from Regression ARAR on Size and Extent of indexing
Coefficients Standard Error t-statistic
Intercept ??? 0.55 –5.2
Size 0.6 0.18 ???
Index 1.1 ??? 2.1

17. (B) will change by 0.6% when the natural logarithm of assets under management
changes by 1.0, holding index constant.
Explanation
A slope coefficient in a multiple linear regression model measures how much the
dependent variable changes for a one-unit change in the independent variable,
holding all other independent variables constant. In this case, the independent
variable size (= In average assets under management) has a slope coefficient of
0.6, indicating that the dependent variable ARAR will change by 0.6% return for a
one-unit change in size, assuming nothing else changes. Pay attention to the units
on the dependent variable.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

18. (B) 0.52.

Explanation
The t-statistic for testing the null hypothesis H0: i = 0 is t = (bi –; 0) / i, where i
is the population parameter for independent variable i, bi is the estimated
coefficient, and i is the coefficient standard error. Using the information provided,
the estimated coefficient standard error can be computed as bIndex / t = Index = 1.1
/ 2.1 = 0.5238.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

19. (B) 3.33.

Explanation
The t-statistic for testing the null hypothesis H0: i = 0 is t = (bi – 0) / i, where i is
the population parameter for independent variable i, bi is the estimated coefficient,
and i is the coefficient standard error. Using the information provided, the t-
statistic for size can be computed as t = bSize/Size = 0.6/0.18 = 3.3333.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 7 Multiple Regression

CFA
20. (A) –2.86.
Explanation
The t-statistic for testing the null hypothesis H0: i; = 0 is t = (bi – 0) / i, where i
is the population parameter for independent variable i, bi is the estimated
parameter, and i is the parameter's standard error. Using the information
provided, the estimated intercept can be computed as b0 = t x 0 = – 5.2 x 0.55
= – 2.86.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

21. (B) All of the parameter estimates are significantly different than zero at the 5% level
of significance.
Explanation
At 5% significance and 97 degrees of freedom (100 – 3), the critical t-value is
slightly greater than, but very close to, 1.984. The t-statistic for the intercept and
index are provided as –5.2 and 2.1, respectively, and the t-statistic for size is
computed as 0.6 / 0.18 = 3.33. The absolute value of the all of the regression
intercepts is greater than tcritical = 1.984. Thus, it can be concluded that all of the
parameter estimates are significantly different than zero at the 5% level of
significance.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

22. (C) The error term is linearly related to the dependent variable.
Explanation
The assumptions of multiple linear regression include: linear relationship between
dependent and independent variable, independent variables are not random and
no exact linear relationship exists between the two or more independent variables,
error term is normally distributed with an expected value of zero and constant
variance, and the error term is serially uncorrelated.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

23. (B) TEEN only.

Explanation
The critical t-values for 40-3-1 = 36 degrees of freedom and a 5% level of
significance are  2.028. Therefore, only TEEN is statistically significant.
(Module 1.1. LOS 1.b)
SchweserNotes - Book 1

Quantitative Methods 8 Multiple Regression

CFA
24. (A) 1 1
Explanation
Assigning a zero to both categories is appropriate for someone with neither
degree. Assigning one to the business category and zero to the engineering
category is appropriate for someone with only a business degree. Assigning zero
to the business category and one to the engineering category is appropriate for
someone with only an engineering degree. Assigning a one to both categories is
correct because it reflects the possession of both degrees.
(Module 1.4, LOS 1.I)
Related Material
SchweserNotes - Book 1

25. (B) The adjusted-R2 is greater than the R2 in multiple regression.

Explanation
The adjusted-R2 will always be less than R2 in multiple regression.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

26. (B) Omit one or more of the collinear variables.

Explanation
The first differencing is not a remedy for the collinearity, nor is the inclusion of
dummy variables. The best potential remedy is to attempt to eliminate highly
correlated variables.
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

Binod Salve, CFA, is investigating the application of the Fama-French three-factor

model (Model 1) for the Indian stock market for the period 2001-2011 (120
months). Using the dependent variable as annualized return (%), the results of the
analysis are shown in Indian Equities—Fama-French Model

Indian Equities-Fama-French Model

Factor Coefficient P-value VIF
Intercept 1.22 < 0.001
SMB 0.23 < 0.001 3
HML 0.34 0.003 3
Rm-Rf 0.88 < 0.001 2
Quantitative Methods 9 Multiple Regression
CFA
R-squared 0.36
SSE 38.00
BG (lag 1) 2.11
BG (lag 2) 1.67

Partial F-Table (5% Level of Significance)

Degrees of Freedom Denominator Degree of Freedom Numerator
1 2 3
112 3.93 3.08 2.69
113 3.93 3.08 2.68
114 3.92 3.08 2.68
115 3.92 3.08 2.68
116 3.92 3.07 2.68
117 3.92 3.07 2.68

Partial Chi-Square Table (5% Level of Significance)

Degrees of Freedom Critical Value
1 3.84
2 5.99
3 7.81
4 9.49
5 11.07
6 12.59

27. (B) Because the test statistic of 7.20 is lower than the critical value of 7.81, we fail to
reject the null hypothesis of no conditional heteroskedasticity in residuals.
Explanation
The chi-square test statistic = n x R2 = 120 x 0.06 = 7.20.
The one-tailed critical value for a chi-square distribution with k = 3 degrees of
freedom and  of 5% is 7.81. Therefore, we should not reject the null hypothesis
and conclude that we don't have a problem with conditional heteroskedasticity.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

Quantitative Methods 10 Multiple Regression

CFA
28. (C) Data improperly pooled.
Explanation
Out of the four forms of model misspecifications, serial correlation in residuals may
be caused by omission of important variables (not an answer choice) and by
improper data pooling.
(Module 1.3, LOS 1.g)
Related Material
SchweserNotes - Book 1

29. (B) No.

Explanation
The BG test statistic has an F-distribution with p and n – p – k – 1 degrees of
freedom, where p = the number of lags tested. Given n = 120 and k = 3, critical
F-values (5% level of significance) are 3.92 (p = 1) and 3.08 (p = 2). BG stats in
Indian Equities—Fama-French Model are lower than the critical F-values; therefore,
serial correlation does not seem to be a problem for both lags.
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

30. (C) No.

Explanation
Multicollinearity is detected using the variance inflation factor (VIF). VIF values
greater than 5 (i.e., R2 > 80%) warrant further investigation, while values above 10
(i.e., R2 > 90%) indicate severe multicollinearity. None of the variables have VIF > 5.
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1

31. (C) Unconditional heteroskedasticity.

Explanation
Unconditional heteroskedasticity does not impact the statistical inference
concerning the parameters. Misspecified models have inconsistent and biased
regression parameters. Multicollinearity results in unreliable estimates of
regression parameters.
(Module 1.3, LOS 1.h)
(Module 1.3, LOS 1.i)
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1
Quantitative Methods 11 Multiple Regression
CFA
32. (B) heteroskedasticity.
Explanation
Heteroskedasticity is present when the variance of the residuals is not the same
across all observations in the sample, and there are sub-samples that are more
spread out than the rest of the sample.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

34. (B) This model is in accordance with the basic assumptions of multiple regression
analysis because the errors are not serially correlated.
Explanation
One of the basic assumptions of multiple regression analysis is that the error
terms are not correlated with each other. In other words, the error terms are not
serially correlated. Multicollinearity and heteroskedasticity are problems in multiple
regression that are not related to the correlation of the error terms.
(Module 1.3, LOS 1.i)
SchweserNotes - Book 1

George Smith, an analyst with Great Lakes Investments, has created a

comprehensive report on the pharmaceutical industry at the request of his boss.
The Great Lakes portfolio currently has a significant exposure to the
pharmaceuticals industry through its large equity position in the top two
pharmaceutical manufacturers. His boss requested that Smith determine a way to
accurately forecast pharmaceutical sales in order for Great Lakes to identify further
investment opportunities in the industry as well as to minimize their exposure to
downturns in the market. Smith realized that there are many factors that could
possibly have an impact on sales, and he must identify a method that can quantify
their effect. Smith used a multiple regression analysis with five independent
variables to predict industry sales. His goal is to not only identify relationships
that are statistically significant, but economically significant as well. The
assumptions of his model are fairly standard: a linear relationship exists between
the dependent and independent variables, the independent variables are not
random, and the expected value of the error term is zero.
Smith is confident with the results presented in his report. He has already done
some hypothesis testing for statistical significance, including calculating a t-
statistic and conducting a two-tailed test where the null hypothesis is that the
regression coefficient is equal to zero versus the alternative that it is not. He feels
that he has done a thorough job on the report and is ready to answer any
questions posed by his boss.
However, Smith's boss, John Sutter, is concerned that in his analysis, Smith has
ignored several potential problems with the regression model that may affect his
conclusions. He knows that when any of the basic assumptions of a regression
Quantitative Methods 12 Multiple Regression
CFA
model are violated, any results drawn for the model are questionable. He asks
Smith to go back and carefully examine the effects of heteroskedasticity,
multicollinearity, and serial correlation on his model. In specific, he wants Smith to
make suggestions regarding how to detect these errors and to correct problems
that he encounters.

34. (B) the variance of the error term is correlated with the values of the independent
variables.
Explanation
Conditional heteroskedasticity exists when the variance of the error term is
correlated with the values of the independent variables.
Multicollinearity, on the other hand, occurs when two or more of the independent
variables are highly correlated with each other. Serial correlation exists when the
error terms are correlated with each other.
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1

35. (C) Type I error by incorrectly rejecting the null hypotheses that the regression
parameters are equal to zero.
Explanation
One problem with conditional heteroskedasticity while working with financial data,
is that the standard errors of the parameter estimates will be too small and the t-
statistics too large. This will lead Smith to incorrectly reject the null hypothesis
that the parameters are equal to zero. In other words, Smith will incorrectly
conclude that the parameters are statistically significant when in fact they are not.
This is an example of a Type I error: incorrectly rejecting the null hypothesis when
it should not be rejected.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

36. (B) The R2 is high, the F-statistic is significant and the t-statistics on the individual
slope coefficients are insignificant.
Explanation
Multicollinearity occurs when two or more of the independent variables, or linear
combinations of independent variables, may be highly correlated with each other. In
a classic effect of multicollinearity, the R2 is high and the F-statistic is significant, but
the t-statistics on the individual slope coefficients are insignificant.
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1
Quantitative Methods 13 Multiple Regression
CFA
37. (C) the error terms are correlated with each other.
Explanation
Serial correlation (also called autocorrelation) exists when the error terms are
correlated with each other.
Multicollinearity, on the other hand, occurs when two or more of the independent
variables are highly correlated with each other. One assumption of multiple
regression is that the error term is normally distributed.
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

Manuel Mercado, CFA has performed the following two regressions on sales data
for a given industry. He wants to forecast sales for each quarter of the upcoming
year.
Model ONE
Regression Statistics
Multiple R 0.941828
R2 0.887039
Adjusted R2 0.863258
Standard Error 2.543272
Observations 24

Durbin-Watson test statistics = 0.7856

ANOVA
df SS MS F Significance F
Regression 4 965.0619 241.2655 37.30006 9.49E-09
Residual 19 122.8964 6.4685
Total 23 1087.9583

Coefficients Standard Error t-statistics

Intercept 31.40833 1.4866 21.12763
Q1 –3.77798 1.485952 –2.54246
Q2 –2.46310 1.476204 –1.66853
Q3 –0.14821 1.470324 –0.10080
TREND 0.851786 0.075335 11.20848

Quantitative Methods 14 Multiple Regression

CFA

Model ONE
Regression Statistics
Multiple R 0.941796
R2 0.886979
Adjusted R2 0.870026
Standard Error 2.479538
Observations 24

Durbin-Watson test statistic = statistic = 0.7860

df SS MS F Significance F
Regression 3 964.9962 321.6654 52.3194 1.19E–09
Residual 20 122.9622 6.14811
Total 23 1087.9584

Coefficients Standard Error t-statistics

Intercept 31.32888 1.228865 25.49416
Q1 –3.70288 1.253493 –2.95405
Q2 –2.38839 1.244727 –1.91881
TREND 0.85218 0.073991 11.51732
The dependent variable is the level of sales for each quarter, in $ millions, which
began with the first quarter of the first year. Q1, Q2, and Q3 are seasonal dummy
variables representing each quarter of the year. For the first four observations the
dummy variables are as follows: Q1:(1,0,0,0), Q2:(0,1,0,0), Q3:(0,0,1,0). The
TREND is a series that begins with one and increases by one each period to end
with 24. For all tests, Mercado will use a 5% level of significance. Tests of
coefficients will be two-tailed, and all others are one-tailed.

38. (B) Model TWO because it has a higher adjusted R2.

Explanation
Model TWO has a higher adjusted R2 and thus would produce the more reliable
estimates. As is always the case when a variable is removed, R2 for Model TWO is
lower. The increase in adjusted R2 indicates that the removed variable, Q3, has
very little explanatory power, and removing it should improve the accuracy of the
estimates. With respect to the references to autocorrelation, we can compare the
Durbin-Watson statistics to the critical values on a Durbin-Watson table.
Since the critical DW statistics for Model ONE and TWO respectively are 1.01
(> 0.7856) and 1.10 (> 7860), serial correlation is a problem for both equations.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

Quantitative Methods 15 Multiple Regression

CFA
39. (B) $51.09 million.
Explanation
The estimate for the second quarter of the following year would be (in millions):
31.4083 + (– 2.4631) + (24 + 2) x 0.851786 = 51.091666.
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1

40. (C) Inappropriate variable scaling.

Explanation
Inappropriate variable scaling may lead to multicollinearity or heteroskedasticity in
residuals. Omission of important variable may lead to biased and inconsistent
regression parameters and also heteroskedasticity/serial correlation in residuals.
Inappropriate variable form can lead to heteroskedasticity in residuals.
(Module 1.3, LOS 1.g)
Related Material
SchweserNotes - Book 1

41. (B) Regression coefficients will be unbiased but standard errors will be biased.
Explanation
Presence of conditional heteroskedasticity will not affect the consistency of
regression coefficients but will bias the standard errors leading to incorrect
application of t-tests for statistical significance of regression parameters.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

42. (B) the intercept is essentially the dummy for the fourth quarter.
The fourth quarter serves as the base quarter, and for the fourth quarter, Q1 = Q2
= Q3 = 0. Had the model included a Q4 as specified, we could not have had an
intercept. In that case, for Model ONE for example, the estimate of Q4 would have
been 31.40833. The dummies for the other quarters would be the 31.40833 plus
the estimated dummies from the Model ONE. In a model that included Q1, Q2, Q3,
and Q4 but no intercept, for example:
Q1 = 31.40833 + (–3.77798) = 27.63035
Such a model would produce the same estimated values for the dependent
variable.
(Module 1.4, LOS 1.l)
Related Material
SchweserNotes - Book 1

Quantitative Methods 16 Multiple Regression

CFA
43. (C) grow, but by less than $1,000,000.
Explanation
The specification of Model TWO essentially assumes there is no difference
attributed to the change of the season from the third to fourth quarter. However,
the time trend is significant. The trend effect for moving from one season to the
next is the coefficient on TREND times $1,000,000 which is $852,182 for
Equation TWO.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Vijay Shapule, CFA, is investigating the application of the Fama-French three-factor

model (Model 1) for the Indian stock market for the period 2001-2011 (120
months). Using the dependent variable as annualized return (%), the results of the
analysis are shown in Indian Equities-Farma-French Model.

Indian Equities-Farma-French Model

Factor Coefficient P-value
Intercept 1.22 <0.001
SMB 0.23 <0.001
HML 0.34 0.003
Rm-Rf 0.88 <0.001
R-squared 0.36
SSE 38.00
AIC –129.99
BIC –118.84

Shapule then modifies the model to include a liquidity factor. Results for this four-
factor model (Model 2) are shown in
Revised Fama-French Model With Liquidity Factor
Revised Fama-French Model With Liquidity Factor
Factor Coefficient P-value
Intercept 1.56 <0.001
SMB 0.22 <0.001
HML 0.35 0.012
Rm-Rf 0.87 <0.001
LIQ –0.12 0.02
R-squared 0.39
SSE 34.00
AIC –141.34
BIC –127.40
Quantitative Methods 17 Multiple Regression
CFA
44. (B) 0.37.
Explanation
Given n = 120 months, k = 4 (for Model 2), and R2 = 0.39:
 120 −1  
R2a = 1 -   × (1 − 0.39 )  = 0.37
 120 − 4 − 1  
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

45. (B) Model 2 because it has a lower Akaike information criterion.

Explanation
The Akaike information criterion (AIC) is used if the goal is to have a better
forecast, while the Bayesian information criterion (BIC) is used if the goal is a
better goodness of fit. Lower values of both criteria indicate a better model. Both
criteria are lower for Model 2.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

46. (B) 13.33.

Explanation
( SSER - SSEU ) / q
F=
( SSEU ) / (n - k -1)
where n = 120, k = 4, and q = 1.
(38 - 34 ) / 1
= 13.33
(34 ) / (120 - 4 -1)
Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

47. (C) 6.80%.

Explanation
Model 1:
Return = 1.22 + 0.23 x SMB + 0.34 x HML + 0.88 Rm-Rf
= 1.22 + 0.23 x 3.30 + 0.34 x 1.25 + 0.88 x 5 = 6.80%.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

Quantitative Methods 18 Multiple Regression

CFA
48. (A) The variance of the error terms is not constant (i.e., the errors are heteroskedastic).
Explanation
The variance of the error term IS assumed to be constant, resulting in errors that
are homoskedastic.
(Module 1.1, LOS 1.c)
Related Material
SchweserNotes - Book 1

49. (A) SALES =  + 1 POP + 2 INCOME + 3 ADV + .

Explanation
SALES is the dependent variable. POP, INCOME, and ADV should be the
independent variables (on the right hand side) of the equation (in any order).
Regression equations are additive.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

50. (A) The F-statistic suggests that the overall regression is significant, however the
regression coefficients are not individually significant.
Explanation
One symptom of multicollinearity is that the regression coefficients may not be
individually statistically significant even when according to the F-statistic the
overall regression is significant. The problem of multicollinearity involves the
existence of high correlation between two or more independent variables. Clearly,
as service employment rises, construction employment must rise to facilitate the
growth in these sectors. Alternatively, as manufacturing employment rises, the
service sector must grow to serve the broader manufacturing sector.
• The variance of observations suggests the possible existence of
heteroskedasticity.
• If the Durbin—Watson statistic may be used to test for serial correlation at a
single lag.
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1
Lynn Carter, CFA, is an analyst in the research department for Smith Brothers in
New York. She follows several industries, as well as the top companies in each
industry. She provides research materials for both the equity traders for Smith
Brothers as well as their retail customers. She routinely performs regression
analysis on those companies that she follows to identify any emerging trends that
could affect investment decisions.

Due to recent layoffs at the company, there has been some consolidation in the
research department. Two research analysts have been laid off, and their workload
will now be distributed among the remaining four analysts. In addition to her
Quantitative Methods 19 Multiple Regression
CFA
current workload, Carter will now be responsible for providing research on the
airline industry. Pinnacle Airlines, a leader in the industry, represents a large
holding in Smith Brothers' portfolio. Looking back over past research on Pinnacle,
Carter recognizes that the company historically has been a strong performer in
what is considered to be a very competitive industry. The stock price over the last
52-week period has outperformed that of other industry leaders, although
Pinnacle's net income has remained flat. Carter wonders if the stock price of
Pinnacle has become overvalued relative to its peer group in the market, and
wants to determine if the timing is right for Smith Brothers to decrease its position
in Pinnacle.

Carter decides to run a regression analysis, using the monthly returns of Pinnacle
stock as the dependent variable and monthly returns of the airlines industry as the
independent variable.
Analysis of Variance Table (ANOVA)
df SS Mean Square
Source
(Degree of Freedom) (Sum of Squares) (SS/df)
Regression 1 3,257 (RSS) 3,257 (MSR)
Error 8 298 (SSE) 37.25 (MSE)
Total 9 3,555 (SS Total)

51. (C) The independent variable is correlated with the residuals.

Explanation
Although the linear regression model is fairly insensitive to minor deviations from
any of these assumptions, the independent variable is typically uncorrelated with
the residuals.
(Module 1.1, LOS 1.c)
Related Material
SchweserNotes - Book 1

52. (C) 0.916, indicating that the variability of industry returns explains about 91.6% of
the variability of company returns.
Explanation
The coefficient of determination (R2) is the percentage of the total variation in the
dependent variable explained by the independent variable.
The R2 = (RSS / SS) Total = (3,257 / 3,555) = 0.916. This means that the
variation of independent variable (the airline industry) explains 91.6% of the
variations in the dependent variable (Pinnacle stock).
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

Quantitative Methods 20 Multiple Regression

CFA
53. (C) predicted value of the independent variable equals 15.
Explanation
Note that the easiest way to answer this question is to plug numbers into the
equation.
The predicted value for Y = 1.75 + 3.25(15) = 50.50.
The variable X1 represents the independent variable.
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1

54. (C) Points 2, 3, and 4.

Explanation
One of the basic assumptions of regression analysis is that the variance of the
error terms is constant, or homoskedastic. Any violation of this assumption is
called heteroskedasticity.
Therefore, Point 1 is incorrect, but Point 4 is correct because it describes
conditional heteroskedasticity, which results in unreliable estimates of standard
errors. Points 2 and 3 also describe limitations of regression analysis.
(Module 1.1, LOS 1.c)
Related Material
SchweserNotes - Book 1

Raul Gloucester, CFA, is analyzing the returns of a fund that his company offers.
He tests the fund's sensitivity to a small capitalization index and a large
capitalization index, as well as to whether the January effect plays a role in the
fund's performance. He uses two years of monthly returns data, and runs a
regression of the fund's return on the indexes and a January-effect qualitative
variable. The "January" variable is 1 for the month of January and zero for all other
months. The results of the regression are shown in the tables below.

Regression Statistics
Multiple R 0.817088
R2 0.667632
Adjusted R2 0.617777
Standard Error 1.655891
Observations 24

ANOVA
df SS MS
Regression 3 110.1568 36.71895
Residual 20 54.8395 2.741975
Total 23 164.9963

Quantitative Methods 21 Multiple Regression

CFA

Coefficient Standard Error t-Statistics

Intercept –0.23821 0.388717 –0.61282
January 2.560552 1.232634 2.077301
Small Cap Index 0.231349 0.123007 1.880778
Large Cap index 0.951515 0.254528 3.738359

Exhibit 1: Partial F-Table (5% Level of Significance)

Degree of Freedom Denominator Degree of Freedom Numerator
1 2 3
18 4.41 3.55 3.16
19 4.38 3.52 3.13
20 4.35 3.49 3.10
21 4.32 3.47 3.07
22 4.30 3.44 3.05
23 4.28 3.42 3.03
Gloucester plans to test for serial correlation and conditional and unconditional
heteroskedasticity.

55. (A) 66.76%.

Explanation
The R2 tells us how much of the change in the dependent variable is explained by
the changes in the independent variables in the regression: 0.667632.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

56. (A) No, because the BG statistic is less than the critical test statistic of 3.55, we don't
have evidence of serial correlation.
Explanation
Number of lags tested = p = 2. The appropriate test statistic for BG test is F-stat
with (p = 2) and (n – p – k – 1 = 18) degrees of freedom. From the table, critical
value = 3.55.
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

Quantitative Methods 22 Multiple Regression

CFA
57. (A) of 1.30 indicates that we cannot reject the hypothesis that the coefficient of small-
cap index is not significantly different from 0.
Explanation
SSER = SST – RSSR = 164.9963 – 106.3320 = 58.6643
F = [(SSER – SSEU) / q] / [SSEU / (n – k – 1)] = [(58.6643 – 54.8395) / 1] /
(54.8395 / 20) = 3.8248 / 2.742 = 1.30
Critical F(1, 20) = 4.35 (from Exhibit 1)
Since the test statistic is not greater than the critical value, we cannot reject the
null hypothesis that b2 = 0.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

58. (C) neither the Durbin-Watson test nor the Breusch-Pagan test.
Explanation
Breusch-Godfrey and Durbin-Watson tests are for serial correlation. The Breusch-
Pagan test is for conditional heteroskedasticity; it tests to see if the size of the
independent variables influences the size of the residuals. Although tests for
unconditional heteroskedasticity exist, they are not part of the CFA curriculum, and
unconditional heteroskedasticity is generally considered less serious than
conditional heteroskedasticity.
(Module 1.3, LOS 1.h)
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

59. (A) 2.322.

Explanation
The forecast of the return of the fund would be the intercept plus the coefficient
on the January effect: 2.322 = –0.238214 + 2.560552.
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1

60. (B) multicollinearity

Explanation
When the F-test and the t-tests conflict, multicollinearity is indicated.
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1

Quantitative Methods 23 Multiple Regression

CFA
61. (A) 10.00
Explanation
The F-statistic is equal to the ratio of the mean squared regression to the mean
squared error,
F = MSR/MSE = 20/2 = 10.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

62. (A) heteroskedasticity

Explanation
The residuals appear to be from two different distributions over time. In the earlier
periods, the model fits rather well compared to the later periods.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

63. (A) Breusch-Godfrey test

Explanation
The Breusch-Godfrey test is used to detect serial correlation. The Breusch-Pagan
test is a formal test used to detect heteroskedasticity while a scatter plot can give
visual clues about presence of heteroscedasticity.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

64. (C) Model misspecification.

Explanation
When data are improperly pooled over multiple economic environments in a
multiple regression analysis, the model would be misspecified.
(Module 1.3, LOS 1.g)
Related Material
SchweserNotes - Book 1
Autumn Voiku is attempting to forecast sales for Brookfield Farms based on a
multiple regression model. Voiku has constructed the following model:
Sales = b0 + (b1  CPI) + (b2  IP) + (b3  GDP) + t
sales = $ change in sales (in 000’s)
CPI = change in the consumer price index.
IP = change in industrial production (millions)

Quantitative Methods 24 Multiple Regression

CFA
GDP = Change in GDP (millions)
All changes in variables are in percentage terms.

Voiku uses monthly data from the previous 180 months of sales data and for the
independent variables. The model estimates (with coefficient standard errors in
parentheses) are:
SALES = 10.2 + (4.6  CPI) + (5.2  IP) + (11.7  GDP)
(5.4) (3.5) (5.9) (6.8)
The sum of squared errors is 140.3 and the total sum of squares is 368.7.
Voiku calculates the unadjusted R2, the adjusted R2, and the standard error of
estimate to be 0.592, 0.597, and 0.910, respectively.
Voiku is concerned that one or more of the assumptions underlying multiple
regression has been violated in her analysis. In a conversation with Dave Grimbles,
CFA, a colleague who is considered by many in the firm to be a quant specialist.

Voiku says, "It is my understanding that there are five assumptions of a multiple
regression model:"
Assumption 1: There is a linear relationship between the dependent and
independent variables.
Assumption 2: The independents variables are not random, and there is zero
correlation between any two of the independent variables.
Assumption 3: The residual term is normally distributes with an expected value
of zero.
Assumption 4: The residual are serially correlated.
Assumption 5: The variance of the residuals id constant.
Grimbles agrees with Miller's assessment of the assumptions of multiple regression.

Voiku tests and fails to reject each of the following four null hypotheses at the
99% confidence interval:
Hypothesis 1: The coefficient on GDP is negative.
Hypothesis 2: The intercept term is equal to – 4
Hypothesis 3: A 2.6% increase in the CPI will result in an increases in sales
of more than 12.0%
Hypothesis 4: A 1% increase in industrial production will result in a 1%
decrease in sales.
Figure 1: Partial table of the Student's t-distribution (One-tailed probabilities)
df p = 0.10 p = 0.05 p = 0.025 p = 0.01 p = 0.005
170 1.287 1.654 1.974 2.348 2.605
176 1.286 1.654 1.974 2.348 2.604
180 1.286 1.653 1.973 2.347 2.603

Quantitative Methods 25 Multiple Regression

CFA
Figure 2: Partial F-Table critical values for right-hand tail area equal to 0.05
df1 = 1 df1 = 3 df1 = 5
df2 = 170 3.90 2.66 2.27
df2 = 176 3.89 2.66 2.27
df2 = 180 3.89 2.65 2.26

Figure 3: Partial F-Table critical values for right-hand tail area equal to 0.025
df1 = 1 df1 = 3 df1 = 5
df2 = 170 5.11 3.19 2.64
df2 = 176 5.11 3.19 2.64
df2 = 180 5.11 3.19 2.64

65. (C) incorrect to agree with Voiku's list of assumptions because two of the assumptions
are stated incorrectly.
Explanation
Assumption 2 is stated incorrectly. Some correlation between independent
variables is unavoidable; and high correlation results in multicollinearity. However,
an exact linear relationship between linear combinations of two or more
independent variables should not exist.
Assumption 4 is also stated incorrectly. The assumption is that the residuals are
serially, uncorrelated (i.e., they are not serially correlated).
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

66. (B) Hypothesis 2.

Explanation
The critical values at the 1% level of significance (99% confidence) are 2.348 for a
one-tail test and 2.604 for a two-tail test (df = 176).
The t-values for the hypotheses are:
Hypothesis 1: 11.7 / 6.8 = 1.72
Hypothesis 2: 14.2 / 5.4 = 2.63
Hypothesis 3: 12.0 / 2.6 = 4.6, so the hypothesis is that the coefficient is greater
than 4.6, and the t-stat of that hypothesis is (4.6 – 4.6) / 3.5 = 0.
Hypothesis 4: (5.2 + 1) / 5.9 = 1.05
Hypotheses 1 and 3 are one-tail tests; 2 and 4 are two-tail tests. Only Hypothesis
2 exceeds the critical value, so only Hypothesis 2 should be rejected.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 26 Multiple Regression

CFA
67. (A) reject the the null hypothesis because the F-statistic is larger than the critical F-
value of 2.66.
Explanation
RSS = 368.7 – 140.3 = 228.4, F-statistic = (228.4 / 3) / (140.3 / 176) = 95.51.
The critical value for a one-tailed 5% F-test with 3 and 176 degrees of freedom is
2.66. Because the F-statistic is greater than the critical F-value, the null hypothesis
that all of the independent variables are simultaneously equal to zero should be
rejected.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

68. (B) incorrect in her calculation of both the unadjusted R2 and the standard error of
estimate.
Explanation
SEE = 140.3/(180 – 3 – 1) = 0.893
unadjusted R2 = (368.7 − 140.3) / 368.7 = 0.619
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

69. (C) multicollinearity

Explanation
The regression is highly significant (based on the F-stat in Part 3), but the
individual coefficients are not. This is a result of a regression with significant
multicollinearity problems. The t-stats for the significance of the regression
coefficients are, respectively, 1.89, 1.31, 0.88, 1.72. None of these are high
enough to reject the hypothesis that the coefficient is zero at the 5% level of
significance (two-tailed critical value of 1.974 from t-table).
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

70. (A) 0.5 to 22.9

Explanation
A 90% confidence interval with 176 degrees of freedom is coefficient ± tc (se)=
11.7 ± 1.654 (6.8) or 0.5 to 22.9.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 27 Multiple Regression

CFA
71. (A) multicollinearity
Explanation
When we use dummy variables, we have to use one less than the states of the
world. In this case, there are three states (groups) possible. We should have used
only two dummy variables. Multicollinearity is a problem in this case. Specifically, a
linear combination of independent variables is perfectly correlated. X1 + X2 + X3 =
1.
There are too many dummy variables specified, so the equation will suffer from
multicollinearity.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

72. (B) INCOME only.

Explanation
The calculated test statistic is coefficient/standard error. Hence, the t-stats are 0.8
for POP, 3.059 for INCOME, and 0.866 for ADV. Since the t-stat for INCOME is the
only one greater than the critical t-value of 2.120, only INCOME is significantly
different from zero.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

73. (B) Rejected at 2.5% significance and 5% significance.

Explanation
The F-statistic is equal to the ratio of the mean squared regression (MSR) to the
mean squared error (MSE).
RSS = SST – SSE = 430 – 170 = 260
MSR = 260 / 5 = 52
MSE = 170 / (48 – 5 – 1) = 4.05
F = 52 / 4.05 = 12.84
The critical F-value for 5 and 42 degrees of freedom at a 5% significance level is
approximately 2.44. The critical F-value for 5 and 42 degrees of freedom at a
2.5% significance level is approximately 2.89. Therefore, we can reject the null
hypothesis at either level of significance and conclude that at least one of the five
independent variables explains a significant portion of the variation of the
dependent variable.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

Quantitative Methods 28 Multiple Regression

CFA
74. (C) no evidence that there is conditional heteroskedasticity or serial correlation in the
regression equation.
Explanation
The test for conditional heteroskedasticity involves regressing the square of the
residuals on the independent variables of the regression and creating a test
statistic that is n x R2, where n is the number of observations and R2 is from the
squared-residual regression. The test statistic is distributed with a chi-squared
distribution with the number of degrees of freedom equal to the number of
independent variables. For a single variable, the R2 will be equal to the square of
the correlation; so in this case, the test statistic is 60 x 0.22 = 2.4, which is less
than the chi-squared value (with one degree of freedom) of 3.84 for a p-value of
0.05. There is no indication about serial correlation.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

75. (C) Heteroskedasticity only occurs in cross-sectional regressions.

Explanation
If there are shifting regimes in a time-series (e.g., change in regulation, economic
environment), it is possible to have heteroskedasticity in a time-series.
Unconditional heteroskedasticity occurs when the heteroskedasticity is not related
to the level of the independent variables. Unconditional heteroskedasticity causes
no major problems with the regression. Breusch-Pagan statistic has a chi-square
distribution and can be used to detect conditional heteroskedasticity.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1
76. (B) The variable X3 is statistically significantly different from zero at the 2%
significance level.
Explanation
The p-value is the smallest level of significance for which the null hypothesis can
be rejected. An independent variable is significant if the p-value is less than the
stated significance level. In this example, X3 is the variable that has a p-value less
than the stated significance level.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Dave Turner is security analyst is using regression analysis to determine how well
two factors explain returns for common stocks. The independent variables are the
natural logarithm of the number of analysis following the companies. Ln(no. of
analysis), and the logarithm of the market value of the companies, Ln(market
value). The regression output generated from a statistical program is given in the
following tables, Each p-value correspondence to a two-tail test.

Quantitative Methods 29 Multiple Regression

CFA
Turner plants to use the result in the analysis of two investments. WLK Corp. has
twelve analysts following it and a market capitalization of $2.33 billion, NGR Corp,
has two analysts following it and a marker capitalization of $47 million.
Table 1: Regression Output
Standard Error of
Variable Coefficient t-statistic p-value
the Coefficient
Intercept 0.043 0.01159 3.71 <
0.001
Ln (No. of Analysts) –0.027 0.00466 –5.80 <
0.001
Ln (Market Value) 0.006 0.00271 2.21 0.028

Table 2: ANOVA
Degrees of Freedom Sum of Squares Mean Square
Regression 2 0.103 0.051
Residual 194 0.559 0.003
Total 196 0.662

77. (B) The intercept and the coefficient on In(no. of analysts) only.
Explanation
The p-values correspond to a two-tail test. For a one-tailed test, divide the
provided p-value by two to find the minimum level of significance for which a null
hypothesis of a coefficient equaling zero can be rejected. Dividing the provided p-
value for the intercept and In(no. of analysts) will give a value less than 0.0005,
which is less than 1% and would lead to a rejection of the hypothesis. Dividing
the provided p-value for In(market value) will give a value of 0.014 which is
greater than 1 %; thus, that coefficient is not significantly different from zero at
the 1% level of significance.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

78. (A) 0.011 to 0.001.

Explanation
The confidence interval is 0.006 ± (1.96)(0.00271) = 0.011 to 0.001
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 30 Multiple Regression

CFA
79. (C) –0.019.
Explanation
Initially, the estimate is 0.1303 = 0.043 + In(2)(–0.027) + In(47000000)(0.006)
Then, the estimate is 0.1116 = 0.043 + In(4)(-0.027) + In(47000000)(0.006)
0.1116 – 0.1303 = –0.0187, or –0.019
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

80. (A) 15.6% of the variation in returns.

Explanation
R2 is the percentage of the variation in the dependent variable (in this case,
variation of returns) explained by the set of independent variables. R2 is calculated
as follows: R2 = (SSR / SST) = (0.103 / 0.662) = 15.6%.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

81. (C) F = 17.00, reject a hypothesis that both of the slope coefficients are equal to
zero.
Explanation
The F-statistic is calculated as follows: F = MSR / MSE = 0.051 / 0.003 = 17.00;
and 17.00 > 4.61, which is the critical F-value for the given degrees of freedom
and a 1% level of significance. However, when F-values are in excess of 10 for a
large sample like this, a table is not needed to know that the value is significant.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

82. (B) At least one of the t-statistics was not significant, the F-statistic was significant,
and a positive relationship between the number of analysts and the size of the
firm would be expected.
Explanation
Multicollinearity occurs when there is a high correlation among independent
variables and may exist if there is a significant F-statistic for the fit of the
regression model, but at least one insignificant independent variable when we
expect all of them to be significant. In this case the coefficient on In(market value)
was not significant at the 1% level, but the F-statistic was significant. It would
make sense that the size of the firm, i.e., the market value, and the number of
analysts would be positively correlated.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 31 Multiple Regression

CFA
83. (A) Intercept term.
Explanation
The intercept term is the value of the dependent variable when the independent
variables are set to zero.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

84. (B) R&D, COMP, and CAP only.

Explanation
The critical t-values for 40-4-1 = 35 degrees of freedom and a 5% level of
significance are ± 2.03.
The calculated t-values are:
t for R & D = 1.25/0.145 = 2.777
t for ADV = 1.0/2.2 = 0.455
t for COMP = –2.0/0.63 = –3.175
t for CAP = 8.0/2.5 = 3.2
Therefore, R&D, COMP, and CAP are statistically significant.)
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

85. (C) The Durbin-Watson statistic.

Explanation
The Durbin-Watson statistic is the most commonly used method for the detection
of serial correlationat the first lag, although residual plots can also be utilized. For
testing of serial correlation beyond the first lag, we can instead use the Breusch-
Godfrey test (but is not one of the answer choices).
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

86. (A) coefficient on each dummy tells us about the difference in earnings per share
between the respective quarter and the one left out (first quarter in this case).
Explanation
The coefficients on the dummy variables indicate the difference in EPS for a given
quarter, relative to the first quarter.
(Module 1.4, LOS 1.l)
Related Material
SchweserNotes - Book 1

Quantitative Methods 32 Multiple Regression

CFA
87. (A) Based on the following executive-specific and company-specific variables, how
many shares will be acquired through the exercise of executive stock options?
Explanation
The number of share can be a broad range of values and is, therefore, not
considered a qualitative dependent variable.
(Module 1.3 LOS 1.h)
Related Material
SchweserNotes - Book 1

88. (B) logistic regression model.

Explanation
The only one of the possible answers that estimates a probability of a discrete
outcome is logit or logistic modeling.
(Module 1.4, LOS 1.m)
Related Material
SchweserNotes - Book 1

89. (A) 14.10.

Explanation
= 10 + 1.25 (4) + 1.0 (0.30) – 2.0 (0.6)
= 10 + 5 + 0.3 – 1.2
= 14.10
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1

90. (A) The assumption of linear regression is that the residuals are heteroskedastic.
Explanation
The assumption of regression is that the residuals are homoskedastic (i.e., the
residuals are drawn from the same distribution).
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

Quantitative Methods 33 Multiple Regression

CFA
91. (B) multicollinearity
Explanation
Multicollinearity refers to the condition when two or more of the independent
variables, or linear combinations of the independent variables, in a multiple
regression are highly correlated with each other. This condition distorts the
standard error of estimate and the coefficient standard errors, leading to problems
when conducting t-tests for statistical significance of parameters.
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1

92. (B) If R&D and advertising expenditure are $1 million each, there are 5 competitors,
and capital expenditure are $2 million, expected Sales are $8.25 million.
Explanation
Predicted sales = $10 + 1.25 + 1 – 10 + 16 = $18.25 million.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

93. (A) $509,980,000.

Explanation
Predicted sales for next year are:
SALES =  + 0.004 (120) + 1.031 (300) + 2.002 (100) = 509,980,000.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Werner Baltz, CFA, has regressed 30 years of data for forecast future sales for
National Motor Company based on the percent change in gross domestic (GDP)
and the change in retail price of a U.S. gallon of fuel. The results are presented
below.
Predictor Coefficient Standard Error of the Coefficient
Intercept 78 13.170
 GDP 30.22 12.120
 $ Fuel –412.39 183.981

Analysis of Variance Table (ANOVA)

Source Degrees of Freedom Sum of Squares
Regression 291.30
Error 27 132.12
Total 29 423.42

Quantitative Methods 34 Multiple Regression

CFA
94. (C) $206.00.
Explanation
Sales will be closest to $78 + ($30.22 x 2.2) + [(–412.39) x (–$0.15)] = $206.34
million
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1

95. (C) at least one of the independent variables has explanatory power, because the
calculated F-statistic exceeds its critical value.
Explanation
MSE = SSE / [n – (k + 1)] = 132.12 + 27 = 4.89. From the ANOVA table, the
calculated F-statistic is (mean square regression / mean square error) = 145.65
/4.89 = 29.7853. From the F-distribution table (2 df numerator, 27 df
denominator) the F-critical value may be interpolated to be 3.36. Because
29.7853 is greater than 3.36, Baltz rejects the null hypothesis and concludes that
at least one of the independent variables has explanatory power.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

96. (B) coefficient estimates.

Explanation
Conditional heteroskedasticity results in consistent estimates, but it biases
standard errors, affecting the computed t-statistic and F-Statistic.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

97. (C) The regression will still exhibit multicollinearity, but the heteroskedasticity and serial
correlation problems will be solved.
Explanation
The correction mentioned solves for heteroskedasticity and serial correlation.
(Module 1.3, LOS 1.h)
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

98. (A) Multicollinearity does not seem to be a problem with the model.
Explanation
Multicollinearity occurs when an independent variable is highly correlated with a
linear combination of the remaining independent variables. VIF values exceeding 5
need to be investigated while values exceeding 10 indicate strong evidence of
multicollinearity.
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1
Quantitative Methods 35 Multiple Regression
CFA
99. (A) 11.
Explanation
The appropriate number of dummy variables is one less than the number of
categories because the intercept captures the effect of the other effect. With 12
categories (months) the appropriate number of dummy variables is 11 = 12 – 1. If
the number of dummy variables equals the number of categories, it is possible to
state any one of the independent dummy variables in terms of the others. This is a
violation of the assumption of the multiple linear regression model that none of
the independent variables are linearly related.
(Module 1.4, LOS 1.I)
Related Material
SchweserNotes - Book 1

100. (B) The R2 is the ratio of the unexplained variation to the explained variation of the
dependent variable.
Explanation
The R2 is the ratio of the explained variation to the total variation.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

101. (A) Transforming a variable.

Explanation
The four types of model specification errors are: omission of an important
independent variable, inappropriate variable form, inappropriate variable scaling
and data improperly pooled. Transforming an independent variable is usually done
to rectify inappropriate variable scaling.
(Module 1.3, LOS 1.g)
Related Material
SchweserNotes - Book 1

William Brent, CFA, is the chief financial officer for Mega Flowers, one of the
largerest producers of flowers and bedding plants in the Western United States.
Mega Flowers its plants in three large nursery facilities located in California. Its
products are sold in its company-owned retail nurseries as well as large, home and
garden “super centers”. For it retail stores, Mega Flowers has designed and
implemented marketing plans each season that are aimed at its consumers in
order to generate additional sales for certain high-margin products. To fully
implement the marketing plain, additional contract salespeople are seasonally
employed.

Quantitative Methods 36 Multiple Regression

CFA
For the past several years, these marketing plans seemed to be successful,
providing a significant boost in sales to those specific products highlighted by the
marketing efforts. However, for the past year, revenues have been flat, even
through marketing expenditures increased slightly. Brent is concerned that the
expensive seasonal marketing campaigns are simply no longer generating the
desired returns, and should either be significantly modified or eliminated
altogether. He proposes that the company hire additional, permanent salespeople
to focus on selling Mega Flowers’ high-margin products all year long. The chief
operating officer, David Johnson, disagrees with Brent. He believes that although
last year’s results were disappointing, the marketing campaign has demonstrated
impressive result for the past five years, and should be continued. His belief is that
the prior years’ performance can be used as a gauge for future results, and that a
simple increase in the sales force will not bring about the desired results.
Brent gathers information regarding quarterly sales revenue and marketing
expenditures for the past five years. Based upon historical data, Brent derives the
following regression equation for Mega Flowers (states in million of dollars):
Expected Sales = 12.6 + 1.6 (Marketing Expenditures) + 1.2 (# of Salespeople)

Brent shows the equation to Johnson and tells him, “This equation shown that a
$1 million increase in marketing expenditures will increase the independent
variable by $1.6 million by $1.6 million, all other factors being equal.” Johnson
replies, “It also appears that sales will equal $12.6 million if all independent
variables are equal to zero”.

102. (B) Brent's statement is incorrect; Johnson's statement is correct.

Explanation
Expected sales is the dependent variable in the equation, while expenditures for
marketing and salespeople are the independent variables. Therefore, a $1 million
increase in marketing expenditures will increase the dependent variable (expected
sales) by $1.6 million. Brent's statement is incorrect.
Johnson's statement is correct. 12.6 is the intercept in the equation, which means
that if all independent variables are equal to zero, expected sales will be $12.6
million.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

103. (C) both independent variables are statistically significant.

Explanation
Using a 5% significance level with degrees of freedom (df) of 17 (20 – 2 – 1),
both independent variables are significant and contribute to the level of expected
sales.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 37 Multiple Regression

CFA
104. (B) 15.706.
Explanation
The MSE is calculated as SSE / (n – k – 1). Recall that there are twenty
observations and two independent variables. Therefore, the MSE in this instance
[267 / (20 – 2 – 1)] = 15.706.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

105. (C) 2 of Brent's points are correct.

Explanation
The statements that if there is a strong relationship between the variables and the
SSE is small, the individual estimation errors will also be small, and also that any
violation of the basic assumptions of a multiple regression model is going to affect
the SEE are both correct.
The SEE is the standard deviation of the differences between the estimated values
for the dependent variables (not independent) and the actual observations for the
dependent variable. Brent's Point 1 is incorrect.
Therefore, 2 of Brent's points are correct.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

106. (C) $24,200,000.

Explanation
Using the information provided, expected sales equals 12.6 + (1.6 x 3.5) + (1.2 x 5)
= $24.2 million. Remember to check the details - i.e. this equation is denominated
in millions of dollars.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

107. (B) The F-statistic.

Explanation
To determine whether at least one of the coefficients is statistically significant, the
calculated F-statistic is compared with the critical F-value at the appropriate level
of significance.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 38 Multiple Regression

CFA
108. (A) R2 = 0.20 and F= 10.
Explanation
R2 = RSS / SST = 20 / 100 = 0.20
The F-statistic is equal to the ratio of the mean squared regression to the mean
squared error.
F= 20 /2 = 10
(Module 1.2, LOS 1.d)
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1
Toni Williams, CFA, has determined that commercial electric generator sales in the
Midwest U.S. for Self-Start Company is a function of several factors in each area:
the cost of heating oil, the temperature, snowfall, and housing starts. Using data
for the most currently available year, she runs a cross-sectional regression where
she regresses the deviation of sales from the historical average in each area in the
deviation of each explanatory variable from the historical average of that variable
for that location, She feels this is the most appropriate method since each
geographic area will have different average values for the inputs, and the model
can explain how current conditions explain how generator sales are higher or
lower from the historical average in each area. In summary, she regresses current
sale for each area minus its respective historical average on the following variables
for each area.
 The difference between the retail price of heating oil and its historical
average.
 The mean number of degrees the temperature is below normal in Chicago.
 The amount of snowfall above the average.
 The percentage of housing starts above the average.
Williams use a sample of 26 observation obtained from 26 metropolitan areas in
the Midwest U.S. The results are in the tables below. The dependent variable is in
sales of generation is million of dollars.
Coefficient Estimates table
Variable Estimated Coefficient Standard Error of the Coefficient
Intercept 5.00 1.850
$ Heating Oil 2.00 0.827
Low Temperature 3.00 1.200
Snowfall 10.00 4.833
Housing Starts 5.00 2.333

Quantitative Methods 39 Multiple Regression

CFA
Analysis of Variance Table (ANOVA)
Source Degrees of Freedom Sum of squares Mean Square
Regression 4 335.20 83.80
Error 21 606.40 28.88
Total 25 941.60

Table of the F-Distribution

Critical values for right-hand tail area equal to 0.05
Numerator: df1 and Denominator: df2
df1
df2 1 2 4 10 20
1 161.45 199.50 224.58 241.88 248.01
2 18.513 19.000 19.247 19.396 19.446
4 7.7086 6.9443 6.3882 5.9644 5.8025
10 4.9646 4.1028 3.4780 2.9782 2.7740
20 4.3512 3.4928 2.8661 2.3479 2.1242
One of her goals is to forecast the sales of the Chicago metropolitan area next
year. For that area and for the upcoming year, Williams obtains the following
projections: heating oil prices will be $0.10 above average, the temperature in
Chicago will be 5 degrees below normal, snowfall will be 3 inches above average,
and housing starts will be 3% below average.

In addition to making forecasts and testing the significance of the estimated

coefficients, she plants to perform diagnostic tests to verify the validity of the
model’s results.

109. (B) $35.2 million above the average.

Explanation
The model uses a multiple regression equation to predict sales by multiplying the
estimated coefficient by the observed value to get:
[5 + (2 x 0.10) + (3 x 5) + (10 x 3) + (5 x (-3))] x $1,000,000 = $35.2 million.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

Quantitative Methods 40 Multiple Regression

CFA
110. (C) at least one of the independent variables has explanatory power.
Explanation
From the ANOVA table, the calculated F-statistic is (mean square regression /
mean square error) = (83.80 / 28.88) = 2.9017. From the F distribution table (4
df numerator, 21 df denominator) the critical F value is 2.84. Because 2.9017 is
greater than 2.84, Williams rejects the null hypothesis and concludes that at least
one of the independent variables has explanatory power.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

111. (B) both a Breusch-Godfrey test and a Breusch-Pagan test.

Explanation
Since the model utilized is not an autoregressive time series, a test for serial
correlation is appropriate so the Breusch-Godfrey test would be used. The
Breusch-Pagan test for heteroskedasticity would also be a good idea.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

112. (C) adjusted R2 value.

Explanation
This can be answered by recognizing that the unadjusted R-square is (335.2 /
941.6) = 0.356. Thus, the reported value must be the adjusted R2. To verify this
we see that the adjusted R-squared is: 1– ((26 – 1) / (26 – 4 – 1)) x (1 – 0.356) =
0.233. Note that whenever there is more than one independent variable, the
adjusted R2 will always be less than R2.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

113. (A) There is a linear relationship between the independent variables.

Explanation
Multiple regression models assume that there is no linear relationship between
two or more of the independent variables. The other answer choices are both
assumptions of multiple regression.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1
Quantitative Methods 41 Multiple Regression
CFA
114. (A) Multiple regression model
Explanation
Fye wants to test a theory of January effect on stock returns (dependent variable)
using a dummy (January = 1, other months = 0), market cap, and beta
(independent variables). A multiple regression model would be most appropriate.
Because the dependent variable (stock returns) is not a qualitative variable, a
logistic regression would not apply.
(Module 1.1, LOS 1.a)
Related Material
SchweserNotes - Book 1

Quin Tan Liu, CFA, is looking at the retail property sector for her manager. She is
undertaking a top down review as the feels this is the best way to analyse the
industry segment. To predict U.S. property starts (housing), she has used
regression analysis.
Liu included the following variables in her analysis:
 Average nominal interest rates during each year (as a decimal)
 Annual GDP per capita in $’000
Given these variables the following output was generated from 30 years of data:

Exhibit 1- Result from Regressing Housing Starts (in Millions) on Interest Rates and
GDP Per Capita
Coefficient Standard Error T-statistic
Intercept 0.42 3.1
Interest rate –1.0 –2.0
GDP per capita 0.03 0.7
ANOVA df SS MSS F
Regression 2 3.896 1.948 21.644
Residual 27 2.431 0.090
Total 29 6.327
Observations 30
Durbin-Watson 1.22
Exhibit 2: Critical Values for F-Distribution at 5% Level of Significance
Degrees of Freedom for the Degrees of Freedom (df) for the Numerator
Denominator 1 2 3
26 4.23 3.37 2.98
27 4.21 3.35 2.96
28 4.20 3.34 2.95
29 4.18 3.33 2.93
30 4.17 3.32 2.92
31 4.16 3.31 2.91
32 4.15 3.30 2.90
The following variable estimates have been made for 20X7:
GDP per capita = $46,700
Interest rate = 7%

Quantitative Methods 42 Multiple Regression

CFA
115. (B) 1,751,000
Explanation
Housing starts = 0.42 – (1 x 0.07) + (0.03 x 46.7) = 1.751 million
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1

116. (C) The independent variables explain 61.58% of the variation in housing starts.
Explanation
The coefficient of determination is the statistic used to identify explanatory power.
This can be calculated from the ANOVA table as 3.896/6.327 x 100 = 61.58%.
The residual standard error of 0.3 indicates that the standard deviation of the
residuals is 0.3 million housing starts. Without knowledge of the data for the
dependent variable it is not possible to assess whether this is a small or a large
error.
The F-statistic does not enable us to conclude on both independent variables. It
only allows us the reject the hypothesis that all regression coefficients are zero
and accept the hypothesis that at least one isn't.
(Module 1.2, LOS 1.d)
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

117. (A) Adjusted R-square is a value between 0 and 1 and can be interpreted as a
percentage.
Explanation
Adjusted R-square can be negative for a large number of independent variables
that have no explanatory power. The other two statements are correct.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

118. (A) slope coefficient in a multiple regression is the value of the dependent variable for
a given value of the independent variable.
Explanation
The slope coefficient is the change in the dependent variable for a one-unit
change in the independent variable.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1
Quantitative Methods 43 Multiple Regression
CFA
In preparing an analysis of HB Inc., Jack Stumper is asked to look at the company’s
sales in relation to broad based economic indicators, Stumper’s analysis indicates
that HB’s monthly sales are related to changes in housing starts (H) and changes
in the mortgage interest rate (M). The analysis covers the past ten years for these
variables. The regression equation is:
s = 1.76 + 0.23H – 0.08M
Number of observations: 123
Unadjusted R2: 0.77
F statistic: 9.80
Durbin Watson statistic 0.50
p-value of Housing Starts 0.017
t-stat of Mortgage Rates –2.6
Variables Descriptions
S = HB Sales (in thousands)
H = Housing starts (in thousands)
M = mortgage interest rate (in percent)

November 20X6 Actual Data

HB’s monthly sales: $55,000
Housing starts: 150,000
Mortgage interest rate (%): 7.5

Critical values for student’s t-Distributions

Level of significance for one-tailed test
Degrees of 10% 5% 2.5% 1% 0.5% 0.05%
Freedom Level of significance for two-tailed test
20% 10% 5% 2% 1% 0.1%
10 1.372 1.812 2.228 2.764 3.169 4.587
20 1.325 1.725 2.086 2.528 2.845 3.850
30 1.310 1.697 2.042 2.457 2.750 3.646
40 1.303 1.684 2.021 2.423 2.704 3.551
120 1.289 1.658 1.980 2.358 2.617 3.373

119. (A) $36,000.

Explanation
1.76 + 0.23 * (150) – 0.08 * (7.5) = 35.66.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1
Quantitative Methods 44 Multiple Regression
CFA
120. (B) different from zero; sales will rise by $23 for every 100 house starts.
Explanation
A p-value (0.017) below significance (0.05) indicates a variable which is
statistically different from zero. The coefficient of 0.23 indicates that sales will rise
by $23 for every 100 house starts.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

121. (C) yes, because 2.6 > 1.98.

Explanation
The correct degrees of freedom for critical t-statistic is n-k-1 = 123-2-1 = 120.
From the t-table, 5% L.O.S., 2-tailed, critical t-value is 1.98. Note that the t-stat
for the coefficient for mortgage rate is directly given in the question (-2.6).
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

122. (A) the joint significance of the independent variables.

Explanation
The F-statistic indicates the joint significance of the independent variables. The
deviation of the estimated values from the actual values of the dependent variable
is the standard error of estimate. The degree of correlation between the
independent variables is the coefficient of correlation.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

123. (C) 77.00.

Explanation
The question is asking for the coefficient of determination.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 45 Multiple Regression

CFA
124. (A) standard errors are too low but coefficient estimate is consistent.
Explanation
Positive serial correlation does not affect the consistency of coefficients (i.e., the
coefficients are still consistent) but the estimated standard errors are too low
leading to artificially high t-statistics.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Miles Mason, CFA, works for ABC Capital, a large money management company
based in New York. Manson has several years of experience as a financial analyst,
but is currently working in the marketing department developing materials to be
used by ABC’s sales team for both existing and prospective clients. ABC Capital’s
client base consists primarily of large net worth individuals and Fortune 500
companies. ABC invests its clients’ money in both publicly traded mutual funds as
well as its own investment funds that are managed in-house, Five years ago,
roughly half of its assets under management. Currently, approximately 75% of
ABC’s assets under management are invested in publicly traded funds, with the
remaining 25% being distributed among ABC’s private funds. The managing
partners at ABC would like to shift more of its client’s assets way from publicly
traded funds into ABC’s proprietary funds, ultimately returning to 50/50 spilt of
assets between publicly traded funds and ABC funds. There are three key reasons
for this shift in the firm’s asset base. First, ABC’s in-house funds have
outperformed other funds consistently for the past five years. Second, ABC can
offer its clients a reduced fee structure on funds managed in-house relative to
other publicly traded funds, Lastly, ABC has recently hired a top fund manager
away from a can offer its clients a top fund manager away from a competing
investment company and would like to increase his assets under management.

ABC capital’s upper management requested that current clients be surveyed in

order to determine the cause of the shift of assets away from ABC funds. Results
of the survey indicated that clients feel there is a lack of information regarding
ABC’s funds. Clients would like to see extensive information about ABC’s past
performance, as well as a sensitivity analysis showing how to the funds will
perform in varying market scenarios, Mason is part of a team that has been
charged by upper management to create a marketing program to present to both
current and potential clients of ABC. He needs to be able to demonstrate a history
of strong performance for the ABC funds, and, while nor promising any measure of
future performance, project possible return scenarios. He decides to conduct a
regression analysis on all of ABC’ in-house funds. He is going to use 12
independent economic variable in order to predict each particular fund’s return.
Mason is very aware of the many factors that could minimize the effectiveness of
his regression model, and if any are present, he known he must determine if any
corrective actions are necessary. Mason is using a sample size of 121 monthly
returns.

Quantitative Methods 46 Multiple Regression

CFA
125. (C) Breusch-Pagan.
Explanation
Durbin-Watson and Breusch-Godfrey test statistic are used to detect
autocorrelation. The Breusch-Pagan test is used to detect heteroskedasticity.
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

126. (C) use robust standard errors.

Explanation
Using generalized least squares and calculating robust standard errors are
possible remedies for heteroskedasticity. Improving specifications remedies serial
correlation. The standard error cannot be adjusted, only the coefficient of the
standard errors.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

127. (B) multicollinearity

Explanation
Common indicators of multicollinearity include: high correlation (>0.7) between
independent variables, no individual t-tests are significant but the F-statistic that
are opposite of what is expected.
(Module 1.3, LOS 1.j)
Related Material
SchweserNotes - Book 1

128. (C) $320.25 million.

Explanation
Predicted sales
= $10 + 1.25 (5) + 1.0 (4) –2.0 (10) + 8 (40)
= 10 + 6.25 + 4 – 20 + 320 = $320.25
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1

Quantitative Methods 47 Multiple Regression

CFA
129. (C) If the p-value of a variable is less than the significance level, the null hypothesis
can be rejected.
Explanation
The p-value is the smallest level of significance for which the null hypothesis can
be rejected. Therefore, for any given variable, if the p-value of a variable is less
than the significance level, the null hypothesis can be rejected and the variable is
considered to be statistically significant.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Peter Pun, an enrolled candidate for the CFA Level II examination, has decided to
perform a calendar test to examine whether there is any abnormal return
associated with investments and disinvestments made in blue-chip stocks on
particular days of the week. As a proxy for blue-chips, he has decided to use the
S&P 500 Index. The analysis will involve the use of dummy variables and is based
on the past 780 trading days. Here are selected findings of his study.
RSS 0.0039
SSE 0.9534
SST 0.9573
R-squared 0.004
SEE 0.035
Jessica Jones, CFA, a friend of Peter, overhears that he is interested in regression
analysis and warms him that whenever heteroskedasticity is present in multiple
regression, it could undermine the regressions results. She mentions that one easy
way to spot conditional heteroskedasticity it through a scatter plot, but she adds
that there is a more formal test.
Unfortunately, she can’t quire remember its name. Jessica believes that
heteroskedasticity can be rectified using White-corrected standard errors. Her son
Jonathan who has also taken part in the discussion, hears this comment and
argues that White corrections would typically reduce the number of Type II error in
financial data.

130. (C) The return on a particular trading day.

Explanation
The omitted variable is represented by the intercept. So, if we have four variables
to represent Monday through Thursday, the intercept would represent returns on
Friday. Remember when we want to distinguish between "n" classes we always use
one less dummy variable the number of classes (n – 1).
(Module 1.4, LOS 1.l)
Related Material
SchweserNotes - Book 1
Quantitative Methods 48 Multiple Regression
CFA
131. (B) There is no value to calendar trading.
Explanation
This question calls for a computation of the F-stat for all independent variables
jointly. F = (0.0039 / 4) / (0.9534 / (780 – 4 – 1) = 0.79. The critical F is
somewhere between 2.37 and 2.45 so we fail to reject the null that the coefficient
are equal to zero.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

132. (C) Breusch-Pagan, which is a one-tailed test.

Explanation
The Breusch-Pagan is used to detect conditional heteroskedasticity and it is a one-
tailed test. This is because we are only concerned about large values in the
residuals coefficient of determination.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

133. (A) Both are correct.

Explanation
Jessica is correct. White-corrected standard errors are also known as robust
standard errors. Jonathan is correct because for financial data, generally, White-
corrected errors are higher than the biased errors leading to lower computed t-
statistics and, therefore, less frequent rejection of the null hypothesis (remember
incorrectly rejecting a true null is Type I error).
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

134. (A) Positive serial correlation.

Explanation
Positive serial correlation is the condition where a positive regression error in one
time period increases the likelihood of having a positive regression error in the
next time period. The residual terms are correlated with one another, leading to
coefficient error terms that are too small.
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

Quantitative Methods 49 Multiple Regression

CFA
135. (A) If a company spends $1 more on R&D (holding everything else constant), sales are
expected to increase by $ 1.5 million.
Explanation
If a company spends $1 million more on R&D (holding everything else constant),
sales are expected to increase by $1.5 million. Always be aware of the units of
measure for the different variables.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

In preparing an analysis of Treefell Company, Jack Lumber is asked to look at the

company’s relation to broad-based economic indicators. Lumber’s analysis
indicates that Treefell’s monthly sales are related to changes in housing starts (H)
and changes in the mortgage interest rate (M). The analysis covers the past
10 years for these variables. The regression equation is:
S = 1.76 + 0.23H – 0.08M
Number of observations: 123
2
Unadjusted R : 0.77
F-statistic: 9.80
Durbin-Watson statistic: 0.50
p-value of Housing Starts: 0.017
t-stat Mortgage Rates: –2.6

Variable Descriptions
S = Treefell Sales (in thousands)
H = housing starts (in thousands)
M = mortgage interest rate (in percent)

November 20X6 Actual data

Treefell’s monthly sales: $55,000
Housing starts: 150,000
Mortgage interest rate(%): 7.5

Partial Chi-Square Table (5% Level of significance)

Degrees of Freedom Critical Value
1 3.84
2 5.99
3 7.81
4 9.49
5 11.07
6 12.59

Quantitative Methods 50 Multiple Regression

CFA
136. (B) $36,000.
Explanation
1.76 + 0.23 x (150) – 0.08 x (7.5) = 35.66.
(Module 1.2, LOS 1.f)
Related Material
SchweserNotes - Book 1

137. (B) Different from zero; sales will rise by $23 for every 100 house starts.
Explanation
A p-value (0.017) below significance (0.05) indicates a variable that is statistically
different from zero. The coefficient of 0.23 indicates that sales will rise by $23 for
every 100 house starts.
Remember the rule p-value < significance, then reject null
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

138. (A) the joint significance of the independent variables.

Explanation
The F-statistic is for the general linear F-test to test the null hypothesis that slope
coefficients on all variables are equal to zero.
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

139. (A) 77.00.

Explanation
The question is asking for the coefficient of determination.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

140. (B) With a test statistic of 13.53, we can conclude the presence of conditional
heteroskedasticity.
Explanation
Chi-square = n x R2 = 123 x 0.11 = 13.53. Critical Chi-square (degree of freedom
= k = 2) = 5.99. Because the test statistic exceeds the critical value, we reject the
null hypothesis (of no conditional heteroskedasticity).
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

Quantitative Methods 51 Multiple Regression

CFA
141. (B) R2 = 0.25 and F = 13.333.
Explanation
R2 = RSS / SST = 100 / 400 = 0.25
The F-statistic is equal to the ratio of the mean squared regression to the mean
squared error. F = 100 / 7.5 = 13.333
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

142. (C) three dummy variables.

Explanation
Three. Always use one less dummy variable than the number of possibilities. For a
seasonality that varies by quarters in the years, three dummy variables are needed.
(Module 1.4, LOS 1.l)
SchweserNotes - Book 1

Damon Washburn, CFA, is currently enrolled as a part-time graduate student at

State University. One of his recent assignment for his course on Quantitative
Analysis is to perform a regression analysis utilizing the concepts covered during
the semester. He must interpret the results of the regression as well as the test
statistics, Washburn is confident in his ability to calculate the statistics because the
class is allowed to use statistical software. However, he realize that the
interpretation of the statistics will be the true test of his knowledge of regression
analysis. His professor has given to the students a list of questions that must be
answered by the results of the analysis.

Washburn has estimated a regression equation in which 160 quarterly return on the
S&P are explained by three macroeconomic variables: employed growth (EMP) as
measured by nonfarm payrolls, gross domestic product (GDP) growth, and private
investment (INV). The results of the regression analysis are as follows:

Coefficient Estimates
Parameter Coefficient Standard Error of Coefficient
Intercept 9.50 3.40
EMP – 4.50 1.25
GDP 4.20 0.76
INV – 0.30 0.16
Other Data:
 Regression sum of squares (RSS) = 126.00
 Sum of squared errors (SSE) = 267.00
 Durbin-Watson statistic (DW) = 1.34
Quantitative Methods 52 Multiple Regression
CFA

Abbreviated Table of the Student’s t-distribution (One-Tailed Probabilities)

df p = 0.10 p = 0.05 p = 0.025 p = 0.01 p = 0.005
3 1.638 2.353 3.182 4.541 5.841
10 1.372 1.812 2.228 2.764 3.169
50 1.299 1.676 2.009 2.403 2.678
100 1.290 1.660 1.984 2.364 2.626
120 1.289 1.658 1.980 2.358 2.617
200 1.286 1.653 1.972 2.345 2.601

Critical Values of Durbin-Watson Statistics ( = 0.05)

K=1 K=2 K=3 K=4 K=5
n dl du dl du dl du dl du dl du
20 1.20 1.41 1.10 1.54 1.00 1.68 0.90 1.83 0.79 1.99
50 1.50 1.59 1.46 1.63 1.42 1.67 1.38 1.72 1.34 1.77
> 100 1.65 1.69 1.63 1.72 1.61 1.74 1.59 1.76 1.57 1.78

143. (B) Two of the three are statistically significant.

Explanation
To determine whether the independent variables are statistically significant, we
use the student's t-statistic, where t equals the coefficient estimate divided by the
standard error of the coefficient. This is a two-tailed test. The critical value for a
5.0% significance level and 156 degrees of freedom (160-3-1) is about 1.980,
according to the table.
The t-statistic for employment growth = –4.50/1.25 = -3.60.
The t-statistic for GDP growth = 4.20/0.76 = 5.53.
The t-statistic for investment growth = –0.30/0.16 = -1.88.
Therefore, employment growth and GDP growth are statistically significant
because the absolute values of their t-statistics are larger than the critical value,
which means two of the three independent variables are statistically significantly
different from zero.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 53 Multiple Regression

CFA
144. (A) not rejected because the t-statistic is equal to 0.92.
Explanation
The hypothesis is:
H0: bGDP = 3.50
Ha: bGDP  0 3.50
This is a two-tailed test. The critical value for the 1.0% significance level and 156
degrees of freedom (160 – 3 – 1) is about 2.617. The t-statistic is (4.20 –
3.50)/0.76 = 0.92. Because the t-statistic is less than the critical value, we cannot
reject the null hypothesis. Notice we cannot say that the null hypothesis is
accepted; only that it is not rejected.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

145. (A) 32%.

Explanation
The R2 is the percentage of variation in the dependent variable explained by the
independent variables. The R2 is equal to the SSRegression/SSTotal, where the SSTotal is
equal to SSRegression + SSError. R2 = 126.00/ (126.00 + 267.00) = 32%.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

146. (A) significant positive serial correlation in the residuals.

Explanation
The Durbin-Watson statistic tests for serial correlation in the residuals. According
to the table, dI = 1.61 and du = 1.74 for three independent variables and 160
degrees of freedom. Because the DW (1.34) is less than the lower value (1.61), the
null hypothesis of no significant positive serial correlation can be rejected. This
means there is a problem with serial correlation in the regression, which affects
the interpretation of the results.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

147. (B) 5.0%.

Explanation
Predicted quarterly stock return is 9.50% + (–4.50)(2.0%) + (4.20)(1.0%) +
(–0.30)(–1.0%) = 5.0%.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Quantitative Methods 54 Multiple Regression

CFA
148. (C) 1.31.
Explanation
The standard error of the estimate is equal to [SSE/(n – k – 1)]1/2
= [267.00/156]1/2 = approximately 1.31.
(Module 1.1, LOS 1.b)
Related Material
SchweserNotes - Book 1

Jessica Jenkins, CFA, is looking at the retail property sector for her manager. She
in undertaking a top down review as she feels this is the best way to analyse the
industry segment. To predict U.S. property starts (housing), she has used
regression analysis.
Jessica included the following variables in her analysis:
 Average nominal interest rates during each year (as a decimal)
 Annual GDP per capita in $’000
Given these variables, the following output was generated from 30 years of
data:

Exhibit 1 – Results from regressing housing starts (in millions) on interest rates
and GDP per capita
Coefficient Standard Error T-statistic
Intercept 0.42 3.1
Interest rate – 1.0 – 2.0
GDP per capita 0.03 0.7
ANOVA df SS MSS F
Regression 2 3.896 1.948 21.644
Residual 27 2.431 0.090
Total 29 6.327
Observations 30
Durbin-Watson 1.27

Quantitative Methods 55 Multiple Regression

CFA
Exhibit 2 – Critical Values for F-Distribution at 5% Level of significance
Degrees of Freedom (df) for the
Degrees of Freedom for the Denominator Numerator
1 2 3
26 4.23 3.37 2.98
27 4.21 3.35 2.96
28 4.20 3.34 2.95
29 4.18 3.33 2.93
30 4.17 3.32 2.92
31 4.16 3.31 2.91
32 4.15 3.30 2.90
The following variable estimates have been made for 20X7.
GDP per capita = $46,700
Interest rate = 7%

149. (B) 1,751,000.

Explanation
Housing starts = 0.42 – (1 x 0.07) + (0.03 x 46.7) = 1.751 million
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

150. (A) The independent variables explain 61.58% of the variation in housing starts.
Explanation
The coefficient of determination is the statistic used to identify explanatory power.
This can be calculated from the ANOVA table as 3.896 / 6.327 x 100 = 61.58%.
The residual standard error of 0.3 indicates that the standard deviation of the
residuals is 0.3 million housing starts. Without knowledge of the data for the
dependent variable, it is not possible to assess whether this is a small or a large
error.
The F-statistic does not enable us to conclude on both independent variables. It
only allows us the reject the hypothesis that all regression coefficients are zero
and accept the hypothesis that at least one isn't.
(Module 1.2, LOS 1.d)
(Module 1.2, LOS 1.e)
Related Material
SchweserNotes - Book 1

Quantitative Methods 56 Multiple Regression

CFA
151. (C) Adjusted R-square can be higher than the coefficient of determination for a model
with a good fit.
Explanation
Adjusted R-squared cannot exceed R-squared (or coefficient of determination) for
a multiple regression.
(Module 1.2, LOS 1.d)
Related Material
SchweserNotes - Book 1

Philip Lee works for Song Bank as a quantitative analyst. He is currently working
on a model to explain the returns (in %) of 20 hedge funds for the past year. He
includes three independent variables:
 Market return = return on a broad-based stock index (in %)
 Closed = dummy variable (= 1 if the fund is closed to new investors; 0
otherwise)
 Prior period alpha = fund return for the prior 12 months – return on market
(in %)
Estimated model: hedge funds return = 3.2 + 0.22 market return + 1.65 closed –
0.11 prior period alpha

Less is concerned about the impact of outliers on the estimated regression model
and collects the following information:
Observation 1 2 3 4 5 6 7 8 9 10
Cook’s D 0.332 0.219 0.115 0.212 0.376 0.232 0.001 0.001 0.233 0.389
Observation 11 12 13 14 15 16 17 18 19 20
Cook’s D 0.089 0.112 0.001 0.001 0.219 0.001 0.112 0.044 0.517 0.212
Additionally, Lee wants to estimate the probability of a hedge fund closing to new
investors, and he uses tow variables:
 Fund size = log of assets under management.
 Prior period alpha (defined earlier)

Results are shown as follows:

Variable Coefficient
Intercept – 3.76
Fund size – 2.98
Prior period alpha – 2.99

Quantitative Methods 57 Multiple Regression

CFA
152. (B) A closed fund is estimated to have an extra return of 1.65% relative to funds that
are not closed.
Explanation
The interpretation of the coefficient is the extra return relative to the alternative
outcome.
(Module 1.4, LOS 1.I)
Related Material
SchweserNotes - Book 1

153. (C) Studentized residuals.

Explanation
Studentized residuals are used to identify outliers (in the dependent variable).
Leverage is used to identify high-leverage observations (in the independent
variable), while Cook's D is a composite measure (combines both independent and
dependent variables) to identify influential observations.
(Module 1.4, LOS 1.k)
Related Material
SchweserNotes - Book 1

154. (A) Observations 10 and 19.

Explanation
Influential observations are those that, when excluded, cause a significant change
to the model coefficients.
k 3
Observations where Cook’s D > = = 0.3873.
n 20
Observations 10 (D = 0.389) and 19 (D = 0.517) satisfy this criteria.
(Module 1.4, LOS 1.k)
Related Material
SchweserNotes - Book 1

155. (C) 4.83%.

Explanation
Odds = ecoeff (fund size) = e–2.98 = 0.0508.
Probability = odds / (1 + odds) = 0.0483 = 4.83%.
(Module 1.4, LOS 1.m)
Related Material
SchweserNotes - Book 1

Quantitative Methods 58 Multiple Regression

CFA
156. (C) Serial correlation occurs least often with time series data.
Explanation
Serial correlation, which is sometimes referred to as autocorrelation, occurs when
the residual terms are correlated with one another, and is most frequently
encountered with time series data. Positive serial correlation can lead to standard
errors that are too small, which will cause computed t-statistics to be larger than
they should be, which will lead to too many Type I errors (i.e. the rejection of the
null hypothesis when it is actually true). Serial correlation however does not affect
the consistency of the regression coefficients.
(Module 1.3, LOS 1.h)
Related Material
SchweserNotes - Book 1

157. (C) The Breusch-Pagan test.

Explanation
The Breusch-Pagan test is a test of the heteroskedasticity and not of serial
correlation.
(Module 1.3, LOS 1.i)
Related Material
SchweserNotes - Book 1

158. (B) incorrect variable form.

Explanation
Incorrect variable form misspecification occurs if the relationship between
dependent and independent variables is nonlinear.
(Module 1.3, LOS 1.g)
Related Material
SchweserNotes - Book 1

Quantitative Methods 59 Multiple Regression