Question 1

Two time series are being analyzed, one as the dependent variable and one as the independent variable in a linear regression. Both time series have a unit root. The relationship between the two variables:

Group of answer choices

A- cannot be analyzed using linear regression.

B- can only be analyzed using linear regression if the two series are cointegrated.

C- can only be analyzed using linear regression if the two series are not cointegrated.

Question 2

1pts

An analyst estimates the following AR(1) model for the earnings per share of a company:

x

t

=0.55+0.87x

t1

+

t

xt=0.55+0.87xt1+t

If the company's earnings per share currently stands at $2.55, which of the following values is closest to the twoperiodahead forecast of the model?

Group of answer choices

A) $2.7685.

B) $2.9586.

C)Unknown since the values of tand t+1need to be observed in order to make the forecast.

Question 3

1pts

An analyst has constructed a log linear time series model based on monthly sales (in millions of dollars) as follows: ln(Salest) =b0+b1t. Her regression output indicates thatb0= 0.02 andb1= 0.022. If time 1 is January 2009, the predicted value of sales for December 2010 isclosest to:

Group of answer choices

A) $0.58 million.

B) $1.09 million.

C) $1.73 million.

Flag question: Question 4

Question 4

1pts

A time series model is based on a linear trend,y=b0+b1t, using quarterly data. The intercept and the independent coefficient are estimated to beb0= 2.48 andb1= 1.44, respectively. If time 1 is March 2010, then the predicted value ofyfor March 2011 isclosestto:

Group of answer choices

A) 5.36

B) 8.24

C) 9.68

Flag question: Question 5

Question 5

1pts

George Clinton, CFA, is a quantitative analyst using autoregressive time series models to forecast macroeconomic data. He notes the following results from two models he has been testing:

ModelStandard ErrorRMSEAR(1)3.30533.5992AR(2)2.98763.8012

Which of the following statements regarding the accuracy of the two models ismost likelyto be accurate?

Group of answer choices

A) The AR(1) model is more accurate at forecasting than the AR(2) model.

B) The AR(2) model is more accurate at forecasting than the AR(1) model.

C) Neither model has an adequate level of forecasting accuracy.

Flag question: Question 6

Question 6

1pts

Consider the following statements:

Statement 1: In the DickyFuller test for unit root, the dependent variable is the firstdifference of the time series, while the independent variable is the first lag of the time series.

Statement 2: The null hypothesis for the DickyFuller test is that g1equals 0.

Which of the following ismost likely?

Group of answer choices

A) Both statements are correct.

B) Only Statement 2 is correct.

C) Both statements are incorrect.

Flag question: Question 7

Question 7

1pts

Serena Robins, CFA, is modeling the monthly returns of the FTSE 100 share index as the following AR(1) model:

x

t

=0.001+0.5x

t1

+

t

xt=0.001+0.5xt1+t

Which of the following values is closest to the mean reverting level of this model?

Group of answer choices

A) 0.001.

B) 0.002.

C) 0.5.

Flag question: Question 8

Question 8

1pts

Consider the following statements regarding regressions with more than one time series:

Statement 1: If neither of the time series has a unit root, linear regression cannot be used to test the relationship between the two series.

Statement 2: If both the time series have unit roots, linear regression can only be used if they are cointegrated.

Which of the following ismost likely?

Group of answer choices

A) Only Statement 1 is correct.

B) Only Statement 2 is correct.

C) Both statements are correct.

Flag question: Question 9

Question 9

1pts

The table below shows the autocorrelations of residuals from an AR(1) model designed to fit changes in the net profit margin (NPM) of XYZ Company. The data covers monthly profitability data for three years.

LagAutocorrelation10.023120.022130.601240.095750.010860.021970.001580.021390.0295100.0585110.0348120.4512

The table below shows the output for a regression on changes in NPM for XYZ, where the specifications of regression have been changed:

CoefficientStandard ErrortStatIntercept0.00090.00110.8182NPMt10.05810.05811.0000NPMt120.72640.058112.5026

The change that has been made to the regression model ismost likely:

Group of answer choices

A) Independent variables for the third and twelfth lags of the dependent variable have been added to the regression.

B) A seasonal lag has been added to the model.

C) The form of the independent variable has been changed from NPM to lnNPM.

Flag question: Question 10

Question 10

1pts

Consider the following statements:

Statement 1: Unlike a simple random walk, a random walk with a drift has an intercept term that is different from 0.

Statement 2: Similar to a simple random walk, a random walk with a drift is not covariance stationary.

Which of the following ismost likely?

Group of answer choices

A) Only Statement 1 is correct.

B) Only Statement 2 is correct.

C) Both statements are correct.

Flag question: Question 11

Question 11

1pts

If a time series follows a random walk without drift, it ismost likelythat the intercept for an AR(1) model for the series is:

Group of answer choices

A) zero and its lag coefficient is one.

B) one and its lag coefficient is zero.

C) zero and its lag coefficient is zero.

Flag question: Question 12

Question 12

1pts

An analyst hypothesizes that there exists an exponential relationship between a variableyand timetas follows:

y

t

=e

b

a+b

1t

yt=eba+b1t

Which of the following linear regression models will correctly reflect this relationship?

Group of answer choices

A) A linear regression of the natural log ofyversus the natural log of timet.

B) A linear regression of the natural log ofyversus timet.

C) A linear regression of theyversus the natural log of timet.

Flag question: Question 13

Question 13

1pts

Which of the following tests ismost likelyused to determine whether two time series are cointegrated?

Group of answer choices

A) DickyFuller test

B) EngleGranger test

C) DurbinWatson test

Flag question: Question 14

Question 14

1pts

An analyst has constructed an autoregressive time series model of order 1 based on 40 observations. The model is designed to forecast quarterly earnings per share and is based on the first difference of the natural logarithms of earnings per share. Details of the model are as follows.

Regression StatisticsR20.39Observations40DurbinWatson2.02CoefficientStandard ErrortStatisticIntercept0.060.023.0Lag 10.210.082.6Autocorrelations of the ResidualLagAutocorrelationStandard ErrortStatistic10.210.1581.32920.110.1580.69230.190.1581.20340.350.1582.215

If significance testing is conducted at the 5% error level, it ismost likelythat:

Group of answer choices

A) the model needs to be adjusted for seasonality.

B) there is no seasonality because the Durbin Watson coefficient is approximately 2.

C) there is no seasonality because none of the autocorrelations of the residuals are significant.

Flag question: Question 15

Question 15

1pts

Consider the following statements:

Statement 1: A loglinear trend model predicts that the dependent variable will grow by a constant amount equal to the slope coefficient each period.

Statement 2: The DurbinWatson test can be used to test for serial correlation in the residuals in a loglinear trend model.

Which of the following ismost likely?

Group of answer choices

A) Only Statement 1 is correct.

B) Only Statement 2 is correct.

C) Both statements are correct.