TRẮC NGHIỆM KTLTC
TRẮC NGHIỆM KTLTC
TRẮC NGHIỆM KTLTC
Continuous number
1. If a researcher uses daily data to examine a 6. In a cross-country study, a researcher codes the
particular problem and creates a variable that US as “1”, Europe as “2” and the rest of the world as
assigns a numerical value of 1 to Monday “3”. This is best described as a/an:
observations, what term would best describe this
type of number? A. Cardinal number
4. Suppose that observations are available on the B. They can be validly averaged over time
monthly bond prices of 100 companies for 5 years.
C. They can be validly averaged cross-sectionally
What type of data are these?
D. They can be expressed in proportion or percentage
A. Cross-sectional
terms
B. Time-series
10. Which of the following statements is TRUE
C. Panel concerning simple returns?
C. Nominal number 11. Suppose that the simple returns on a stock for
each of four years are 10%, -6%, 13% and -8%.
The appropriately calculated aggregate return
over the whole four-year period to the nearest 1% 4. The point at which a function crosses the x-axis
is: is called:
A. 9% A. The intercept
B. 2% B. The slope
C. 7% C. A root
Suppose that the log returns on a stock for each of 5. If the relationship between two variables is y =
four years are 10%, -6%, 13% and -8%. The 3x3 + 2x2 + x – 6, what is the functional form that
appropriately calculated aggregate return over the links them?
whole four-year period to the nearest 1% is:
A. Linear
A. 9%
B. Non-linear
B. 2%
C. Quadratic
C. 7%
D, Exponential
D. 1.8%
6. If the relationship between two variables is y = –
CHƯƠNG 2 4x + 2, what is the functional form that links them?
A. 86% D. Cubic
C. 74 A. a = 0
D. 1 B. b = 0
C. 1.12 A. a = 1
D. 5.6 B. a = b
C. b = 1 15. Another way of writing log(x + y) is:
D. b = 0 A. log(x) + log(y)
A. 3 D. Undefined
B. At most three 17. Writing out all the terms in the expression
would lead to:
C. 0
(i) 3x3
D. At least one
(ii) x3
12. Which shape will be the function y = –3 + 2x –
x2? (iii) 27x3
A. (i) D. (iv)
A. 0 D. A matrix
D. 3x + 4-4y A. 1 x 4
B. 3 x 2 𝟏 𝟑
29. The inverse of the matrix is:
𝟐 𝟓
C. 2 x 3
1 3
(i)
D. Square 2 5
32. What are the eigenvalues of the matrix D. For an adequate model, the residual (u-hat) will be
−𝟑 −𝟔 zero for all sample data points.
?
𝟐 𝟒
4. Which of the following statements is TRUE
A. 0 and 0 concerning OLS estimation?
(i) The standard error will be positively related to the model , estimated using 100
the residual variance observations, and where standard errors are
15. Suppose that a hypothesis test is conducted (iv) There is insufficient information given in the
using a 5% significance level. Which of the question to reach a conclusion.
following statements are correct?
A. (i)
(i) The significance level is equal to the size of the
test B. (ii)
B. 0.013 (1)
6. Consider the following statistics calculated from (i) The restricted regression would be the one
the raw data: labelled as equation (1) above
(ii)
8. Suppose that 100 separate firms were tested to A. (ii) and (iv) only
determine how many of them “beat the market”
using a Jensen-type regression, and it is found that B. (i) and (iii) only
3 fund managers significantly do so. Does this
C. (i), (ii) and (iii) only
suggest prima facie evidence for stock market
inefficiency? D. (i), (ii), (iii) and (iv)
A. Yes 11. Consider the following regression equation
estimated using 1,000 daily observations.
B. No
Model 1:
(1)
(iii) β1 = 0 and β2 = 0 and β3 = 0 and β4 = 0 (ii) Model 2 must have an adjusted R2 at least as
high as that of model 1
(iv) β1 = 0 or β2 = 0 or β3 = 0 or β4 = 0
(iii) Models 1 and 2 would have identical values of
A, (i) R2 if the estimated coefficient on α3 is zero
B. (ii) (iv) Models 1 and 2 would have identical values of
adjusted R2 if the estimated coefficient on α3 is
C. (iii)
zero.
D. (iv)
A. (ii) and (iv) only
14. Which one of the following is examined by
B. (i) and (iii) only
looking at a goodness of fit statistic?
C. (i), (ii), and (iii) only
A. How well the population regression function fits the
data D. (i), (ii), (iii), and (iv)
17. Suppose that, for the models in question 16, 20. What does a quantile regression measure?
the R2 is higher for model 2 but the adjusted R2 is
lower for model 2. Which one of the following is A. The entire distribution of y given the distributions
the most plausible explanation? of the explanatory variables
i) The coefficient estimate on α3 is zero B. The fifth and ninety fifth percentiles of y only given
the explanatory variables
(ii) The coefficient estimate α3 is non-zero but not
significant C. The median of y only given the explanatory
variables
(iii) The variable x3t is highly correlated with the
variable x2t D. The relationship between the mean of y and the
mean of the explanatory variables
(iv) The researcher must have made a mistake
since the situation described in the question could 21. The parameters of a quantile regression
not happen. function are estimated by:
18. Suppose that the two models in question 16 D. Minimising the weighted sum of squared residuals
have identical R2 values. Which one of the
following statements is true? CHƯƠNG 5
i) The two models will also have identical values of 1. Which of the following assumptions are
adjusted R2 required to show the consistency, unbiasedness
and efficiency of the OLS estimator?
(ii) Model 2 must have a higher value of adjusted
R2 (i) E(ut) = 0
A. Quantile regressions are more robust to outliers (i) The coefficient estimates are not optimal
B.Quantile regressions do not require the (ii) The standard error estimates are not optimal
homoscedasticity assumption
(iii) The distributions assumed for the test
C. Quantile regressions can capture non-linear statistics are inappropriate
relationships between variables
(iv) Conclusions regarding the strength of
D. Quantile regressions can be used to capture tail relationships between the dependent and
behaviour independent variables may be invalid.
A. (ii) and (iv) only B. 118.50
D. (i), (ii), (iii), and (iv) 6. What would be then consequences for the OLS
estimator if heteroscedasticity is present in a
3. What is the meaning of the term regression model but ignored?
“heteroscedasticity”?
A. It will be biased
A. The variance of the errors is not constant
B. It will be inconsistent
B. The variance of the dependent variable is not
constant C. It will be inefficient
C. The errors are not linearly independent of one C. All of (a), (b) and (c) will be true
another
7. Which of the following are plausible approaches
D. The errors have non-zero mean to dealing with a model that exhibits
heteroscedasticity?
4. Consider the following regression model (2)
(i) Take logarithms of each of the variables
B. White’s test
Suppose that model (2) is estimated using 100
quarterly observations, and that a test of the type C. The RESET test
described in question 4 is conducted. What would
be the appropriate 2 critical value with which to D. The Breusch-Godfrey test
compare the test statistic, assuming a 10% size of
test? 10. If a Durbin Watson statistic takes a value close
to zero, what will be the value of the first order
A. 2.71 autocorrelation coefficient?
A. Close to zero C. (i), (ii), and (iii) only
C. The test result is inconclusive (iv) Try a model in first differenced form rather
than in levels.
12. Suppose that a researcher wishes to test for
autocorrelation using an approach based on an A. (ii) and (iv) only
auxiliary regression. Which one of the following
auxiliary regressions would be most appropriate? B. (i) and (iii) only
(i) Coefficient estimates may be misleading D. (i), (ii), (iii), and (iv)
(ii) Hypothesis tests could reach the wrong 16. Including relevant lagged values of the
conclusions dependent variable on the right hand side of a
regression equation could lead to which one of the
(iii) Forecasts made from the model could be following?
biased
A. Biased but consistent coefficient estimates
(iv) Standard errors may inappropriate
B. Biased and inconsistent coefficient estimates
A. (ii) and (iv) only
C. Unbiased but inconsistent coefficient estimates
B. (i) and (iii) only
D.Unbiased and consistent but inefficient coefficient B. The coefficient estimates will be biased but
estimates consistent
17. Near multicollinearity occurs when C. The coefficient estimates will be biased and
inconsistent
A. Two or more explanatory variables are perfectly
correlated with one another D. Test statistics concerning the parameters will not
follow their assumed distributions.
B. The explanatory variables are highly correlated
with the error term 22. A leptokurtic distribution is one which
C. The explanatory variables are highly correlated with A. Has fatter tails and a smaller mean than a normal
the dependent variable distribution with the same mean and variance
D. Two or more explanatory variables are highly B. Has fatter tails and is more peaked at the mean than
correlated with one another a normal distribution with the same mean and
variance
18. Which one of the following is NOT a plausible
remedy for near multicollinearity? C. Has thinner tails and is more peaked at the mean
than a normal distribution with the same mean and
A. Use principal components analysis variance
B. Drop one of the collinear variables D. Has thinner tails than a normal distribution and is
skewed
C. Use a longer run of data
23. Under the null hypothesis of a Bera-Jarque test,
D. Take logarithms of each of the variables the distribution has
19. What will be the properties of the OLS A. Zero skewness and zero kurtosis
estimator in the presence of multicollinearity?
B. Zero skewness and a kurtosis of three
A. It will be consistent, unbiased and efficient
C. Skewness of one and zero kurtosis
B. It will be consistent and unbiased but not efficient
D. Skewness of one and kurtosis of three
C. It will be consistent but not unbiased
24. Which one of the following would be a
D It will not be consistent plausible response to a finding of residual non-
normality?
20. Which one of the following is NOT an example
of mis-specification of functional form? A. Use a logarithmic functional form instead of a linear
one
A. Using a linear specification when y scales as a
function of the squares of x B. Add lags of the variables on the right hand side of
the regression model
B. Using a linear specification when a double-
logarithmic model would be more appropriate C. Estimate the model in first differenced form
C. Modelling y as a function of x when in fact it scales D. Remove any large outliers from the data
as a function of 1/x
25. A researcher tests for structural stability in the
D. Excluding a relevant variable from a linear following regression model:
regression model
D. The sum of the RSS for the first and second sub- (i) The standard errors would be biased
samples
(ii) If the excluded variable is uncorrelated with all
26. Suppose that the residual sum of squares for of the included variables, all of the slope
the three regressions corresponding to the Chow coefficients will be inconsistent.
test described in question 25
(iii) If the excluded variable is uncorrelated with
all of the included variables, the intercept
[ ] are 156.4, 76.2 and coefficient will be inconsistent.
61.9. What is the value of the Chow F-test statistic?
(iv) If the excluded variable is uncorrelated with
A. 4.3 all of the included variables, all of the slope and
intercept coefficients will be consistent and
B. 7.6
unbiased but inefficient.
C. 5.3
A. (ii) and (iv) only
D. 8.6
B. (i) and (iii) only
27. What would be the appropriate 5% critical
C. (i), (ii), and (iii) only
value for the test described in questions 25 and
D. (i), (ii), (iii), and (iv)
26? [ ]
31. A parsimonious model is one that
A. 2.6
A. Includes too many variables
B. 8.5
B. Includes as few variables as possible to explain the
C. 1.3
data
D. 9.2
C. Is a well-specified model
28. Suppose now that a researcher wants to run a
D. Is a mis-specified model
forward predictive failure test on the last 5
observations using the same model and data as in 32. An overparameterised model is one that
question 25 [ ]. A. Includes too many variables
Which would now be the unrestricted residual
sum of squares? B. Includes as few variables as possible to explain the
data
A. The RSS for the whole sample regression
C. Is a well-specified model
B. The RSS for the long sub-sample regression
D. Is a mis-specified model
C. The RSS for the short sub-sample regression
33. Which one of the following is a disadvantage of
D. The sum of the RSS for the long and short sub- the general to specific or “LSE” (“Hendry”)
sample regressions approach to building econometric models, relative
to the specific to general approach?
29. If the two RSS for the test described in question
28 are 156.4 and 128.5, what is the value of the A. Some variables may be excluded at the first stage
test statistic? leading to coefficient biases
A. 13.8 B. The final model may lack theoretical interpretation
B. 14.3 C. The final model may be statistically inadequate
C. 8.3 D. If the initial model is mis-specified, all subsequent
steps will be invalid
D. 8.6
34. Which of the following consequences might A. They are not theoretically motivated
apply if an explanatory variable in a regression is
measured with error? B. They cannot produce forecasts easily
(i) The corresponding parameter will be estimated C. They cannot be used for very high frequency data
inconsistently
D. It is difficult to determine the appropriate
(ii) The corresponding parameter estimate will be explanatory variables for use in pure time-series
biased towards zero models
(iii) The assumption that the explanatory 3. Which of the following conditions are necessary
variables are non-stochastic will be violated for a series to be classifiable as a weakly stationary
process?
(iv) No serious consequences will arise
(i) It must have a constant mean
A. (i) only
(ii) It must have a constant variance
B. (i) and (ii) only
(iii) It must have constant autocovariances for
C. (i), (ii), and (iii) only given lags
(ii) The corresponding parameter estimate will be D. (i), (ii), (iii), and (iv).
biased towards zero
4. A white noise process will have
(iii) The assumption that the explanatory
variables are non-stochastic will be violated (i) A zero mean
B. (i) and (ii) only (iv) Autocovariances that are zero except at lag
zero
C. (i), (ii), and (iii) only
A. (ii) and (iv) only
D. (iv) only
B. (i) and (iii) only
CHƯƠNG 6
C. (i), (ii), and (iii) only
1. Which of the following is a typical characteristic
of financial asset return time-series? D. (i), (ii), (iii), and (iv)
(iv) The acf and pacf will be the same at lag two for
an MA(1) model
A. An AR(1)
A. (ii) and (iv) only
B. An ARMA(2,1)
B. (i) and (iii) only
C. An MA(2)
C. (i), (ii), and (iii) only
D. An AR(2)
D. (i), (ii), (iii), and (iv).
21. Consider the following picture and suggest the
17. An ARMA(p,q) (p, q are integers bigger than model from the following list that best characterises
zero) model will have the process:
(iv) If the model suggested at the identification (ii) An ARIMA(p,1,q) model estimated on a series
stage is appropriate, the coefficients on the of logs of prices is equivalent to an ARIMA(p,0,q)
additional variables under the overfitting model estimated on a set of continuously
approach will be statistically insignificant compounded returns
A. (ii) and (iv) only (iii) It is plausible for financial time series that the
optimal value of d could be 2 or 3.
B. (i) and (iii) only
(iv) The estimation of ARIMA models is
C. (i), (ii), and (iii) only incompatible with the notion of cointegration
A. The current value of y 33. Which of the following statements are true
concerning the estimation and forecasts of an
B. zero exponential smoothing model,
D. The average value of y over the in-sample period (i) Using the standard notation, the larger the
value of a, the less weight is attached to more
29. If a series, y, follows a random walk with drift recent observations
b, what is the optimal one-step ahead forecast of
the change in y? (ii) If a = 0, there will be no updating as new
observations become available
A. The current value of y
(iii) The one-step ahead forecast only from an
B. zero exponential smoothing model will be the most
recently available smoothed value
C. one
(iv) If a = 1, the model is equivalent to a random
D. The average value of the change in y over the in- walk for the series y
sample period
A. (ii) and (iv) only
30. An “ex ante” forecasting model is one which
B. (i) and (iii) only
A. Includes only contemporaneous values of variables
on the RHS C. (i), (ii), and (iii) only
B. Includes only contemporaneous and previous D. (i), (ii), (iii), and (iv)
values of variables on the RHS
34. Which one of the following statements is true
C. Includes only previous values of variables on the concerning alternative forecast accuracy
RHS measures?
D.Includes only contemporaneous values of exogenous A. Mean squared error is usually highly correlated
variables on the RHS with trading rule profitability
31. Consider the following MA(2) model B. Mean absolute error provides a quadratic loss
function
𝒚𝒕 = 𝟎. 𝟑 + 𝟎. 𝟓𝒖𝒕−𝟏 − 𝟎. 𝟒𝒖𝒕−𝟐 + 𝒖𝒕
C. Mean absolute percentage error is a useful measure
What is the optimal two-step ahead forecast from for evaluating asset return forecasts
this model, made at time t, if the values of the
residuals from the model at time t and t-1 were 0.6 D. Mean squared error penalises large forecast errors
and –0.1 respectively and the values of the actual disproportionately more than small forecast errors
series y at time t-1 was –0.4?
35. Which one of the following factors is likely to B. (i) and (iii) only
lead to a relatively high degree of out-of-sample
forecast accuracy? C. (i), (ii), and (iii) only
A. A model that is based on financial theory D. (i), (ii), (iii), and (iv)
B. A model that contains many variables 4. Consider the following system of equations
(with time subscripts suppressed and using
C. A model whose dependent variable has recently standard notation)
exhibited a structural change
CHƯƠNG 7
1. In the context of simultaneous equations According to the order condition, the first equation
modelling, which of the following statements is is
true concerning an endogenous variable?
A. Unidentified
A. The values of endogenous variables are determined
outside the system B. Just identified
D. Reduced form equations will contain only 5. Consider again the system of equations in
endogenous variables on the RHS question 4. According to the order condition, the
second equation is
2. If OLS is applied separately to each equation that
is part of a simultaneous system, the resulting A. Unidentified
estimates will be
B. Just identified
A. Unbiased and consistent
C. Over-identified
B. Biased but consistent
D. It is not possible to tell whether the equation is
C. Biased and inconsistent identified since the question does not give the reduced
form models
D. It is impossible to apply OLS to equations that are
part of a simultaneous system 6. Consider again the system of equations in
question 4. Which estimation method, if any, can
3. Which of the following statements are true be used for the third equation in the system:
concerning a triangular or recursive system?
(i) OLS
(i) The parameters can be validly estimated using
separate applications of OLS to each equation (ii) 2SLS
(iv) The independent variables may be correlated C. (i), (ii), and (iii)
with the error terms in the equations in which
D. The coefficients cannot be validly estimated using
they appear as independent variables
any method
A. (ii) and (iv) only
7. The order condition is
A. A necessary and sufficient condition for B. (i) and (iii) only
identification
C. (i), (ii), and (iii) only
B. A necessary but not sufficient condition for
identification D. (i), (ii), (iii), and (iv)
C. A sufficient but not necessary condition for 11. How many parameters will be required to be
identification estimated in total for all equations of a standard
form, unrestricted, tri-variate VAR(4), ignoring the
D. A condition that is nether necessary nor sufficient intercepts?
for identification
A. 12
8. A Hausman test would be used for
B. 4
A. Determining whether an equation that is part of a
simultaneous system is identified C. 3
D. Determining whether the structural form equations A. The coefficient estimates have intuitive theoretical
can be obtained via substitution from the reduced interpretations
forms
B. The coefficient estimates usually have the same sign
9. Which of the following estimation techniques for all of the lags of a given variable in a given equation
are available for the estimation of over-identified
systems of simultaneous equations? C. VARs often produce better forecasts than
simultaneous equation structural models
(i) OLS
D. All of the components of a VAR must be stationary
(ii) ILS before it can be used for forecasting
(iii) 2SLS 13. Suppose that two researchers, using the same 3
variables and the same 250 observations on each
(iv) IV variable, estimate a VAR. One estimates a VAR(6),
while the other estimates a VAR(4). The
A. (iii) only determinants of the variance-covariance matrices
of the residuals for each VAR are 0.0036 and
B. (iii) and (iv) only 0.0049 respectively. What is the values of the test
statistic for performing a test of whether the
C. (ii), (iii), and (iv) only
VAR(6) can be restricted to a VAR(4)?
D. (i), (ii), (iii) and (iv)
A. 77.07
10. Which of the following are advantages of the
B. 0.31
VAR approach to modelling the relationship
between variables relative to the estimation of full C. 0.33
structural models?
D. 4.87
(i) VARs receive strong motivation from financial
and economic theory 14. Consider again the VARs that were discussed in
question 13. What is the number of degrees of
(ii) VARs in their reduced forms can be used easily freedom for the critical value for testing the
to produce time-series forecasts restriction?
(iii) VAR models are typically highly parsimonious A. 3
(iv) OLS can be applied separately to each B. 6
equation in a reduced form VAR
C. 9
A. (ii) and (iv) only
D. 18 18. Which of the following statements is true
concerning variance decomposition analysis of
15. Suppose now that a researcher wishes to use VARs?
information criteria to determine the optimal lag
length for a VAR. 500 observations are available (i) Variance decompositions measure the impact of
for the bi-variate VAR, and the values of the a unit shock to each of the variables on the VAR
determinant of the variance-covariance matrix of
residuals are 0.0336, 0.0169, 0.0084, and 0.0062 (ii) Variance decompositions can be thought of as
for 1, 2, 3, and 4 lags respectively. What is the measuring the proportion of the forecast error
optimal model order according to Akaike’s variance that is attributable to each variable
information criterion?
(iii) The ordering of the variables is important for
A. 1 lag calculating impulse responses but not variance
decompositions
B. 2 lags
(iv) It is usual that most of the forecast error
C. 3 lags variance for a given variable is attributable to
shocks to that variable
D. 4 lags
A. (ii) and (iv) only
16. Consider the following bivariate VAR(2) model:
B. (i) and (iii) only
B. The d coefficients significant and the b coefficients B. The tests may be oversized
insignificant
C. The tests may fail to reject the null hypothesis when
C. The a coefficients significant and the c coefficients it is incorrect
insignificant
D. All of (a) to (c) could potentially apply
D. The c coefficients significant and the a coefficients
insignificant CHƯƠNG 8
17. Consider again the VAR model of equation 16. 1. Which one of the following would NOT be a
consequence of using non-stationary data in levels
form?
Which of the following conditions must hold for it B. Test statistics may not follow standard distributions
to be said that there is bi-directional feedback? C. Statistical inferences may be invalid
A. The b and d coefficients significant and the a and c
D. Parameter estimates may be biased
coefficients insignificant
2. For a stationary autoregressive process, shocks
B. The a and c coefficients significant and the b and d
will
coefficients insignificant
A. Eventually die away
C. The a and c coefficients significant
B. Persist indefinitely
D. The b and d coefficients significant
C. Grow exponentially
D. Never occur
(i)
Which one of the following most accurately
describes the process for yt? (ii)
A. A unit root process
(iii)
B. A stationary process
(iv)
C. A deterministic trend process
A. (i)
D. A random walk with drift
B. (ii)
4. If a series, yt is said to be integrated of order 2,
which of the following statements is INCORRECT? C. (iii)
(i) It requires differencing twice to generate a D. (iv)
stationary series
7. Note that statistical tables are not necessary to
(ii) It contains exactly two unit roots answer this question. For a sample of 1000
observations, the Dickey-Fuller test statistic values
(iii) If the series is differenced three times, the
are
resulting series will be stationary
A. More negative than (i.e. bigger in absolute value
(iv) A plausible model for the series would be
than) those in the left hand tail of a normal
distribution
What is the appropriate conclusion? 12. If the Engle-Granger test is applied to the
residuals of a potentially cointegrating regression,
(i) yt is stationary what would be the interpretation of the null
hypothesis?
(ii) yt contains exactly one unit root
A. The variables are cointegrated
(iii) yt contains at least one unit root
B. The variables are not cointegrated
(iv) yt contains exactly two unit roots
C. Both variables are stationary
A. (i)
D. Both variables are non-stationary
B. (ii)
13. Consider the following model for yt:
C. (iii)
D. (iv)
Which of the following statements are true?
10. Suppose that the following Dickey-Fuller test
(i) The gamma terms measure the long-run
regression is conducted relationship between y and x
and the value of the test statistic is –6.3.
(ii) The gamma terms measure the short-run
What is the appropriate conclusion? relationship between y and x
(iv) yt contains exactly two unit roots A. (ii) and (iv) only
C. The variables are treated asymmetrically in the 21. If a Johansen “trace” test for a null hypothesis
cointegrating tests of 2 cointegrating vectors is applied to a system
containing 4 variables is conducted, which
D. It is not possible to perform tests about the eigenvalues would be used in the test?
cointegrating relationship
A. All of them
17. What are the characteristic roots of the
B. The largest 2
C. The smallest 2
following matrix ?
D. The second largest
A. 0 and 8
22. What problems may arise if the Perron (1989)
B. 0 and 4 procedure that allows for a known structural
C. 2 and 4 break is used when testing for a unit root?
C. The number of parameters to estimate may be large, B. The least squares dummy variables approach
resulting in a loss of degrees of freedom
C. The random effects model
D. The fixed effects approach can only capture cross-
D. Heteroscedasticity and autocorrelation consistent
sectional heterogeneity and not temporal variation in
the dependent variable 6. Which of the following is a disadvantage of the
random effects approach to estimating a panel
2. The “within transform” involves
model?
A. Taking the average values of the variables
A. The approach may not be valid if the composite
B. Subtracting the mean of each entity away from each error term is correlated with one or more of the
observation on that entity explanatory variables
C. Estimating a panel data model using least squares B. The number of parameters to estimate may be large,
dummy variables resulting in a loss of degrees of freedom
D. Using both time dummies and cross-sectional C. The random effects approach can only capture
dummies in a fixed effects panel model cross-sectional heterogeneity and not temporal
variation in the dependent variable.
3. Which of the following are advantages of the use
of panel data over pure cross-sectional or pure D. All of (a) to (c) are potential disadvantages of the
time-series modelling? random effects approach.
(i) The use of panel data can increase the number 7. In order to determine whether to use a fixed
of degrees of freedom and therefore the power of effects or random effects model, a researcher
tests conducts a Hausman test. Which of the following
statements is false?
(ii) The use of panel data allows the average value
of the dependent variable to vary either cross- A. For random effects models, the use of OLS would
sectionally or over time or both result in consistent but inefficient parameter
estimation
(iii) The use of panel data enables the researcher
allows the estimated relationship between the B. If the Hausman test is not satisfied, the random
independent and dependent variables to vary effects model is more appropriate
either cross-sectionally or over time or both
C. Random effects estimation involves the construction
A. (i) only of “quasi-demeaned” data
B. (i) and (ii) only D. Random effects estimation will not be appropriate if
the composite error term is correlated with one or
C. (ii) only more of the explanatory variables in the model
8. Which of the following statements is false A. A logit model
concerning the linear probability model?
B. A multinomial logit
A. There is nothing in the model to ensure that the
estimated probabilities lie between zero and one C. A tobit model
B. Even if the probabilities are truncated at zero and D. An ordered logit model
one, there will probably be many observations for
which the probability is either exactly zero or exactly 12. A dependent variable whose values are not
one observable outside a certain range but where the
corresponding values of the independent variables
C. The error terms will be heteroscedastic and not are still available would be most accurately
normally distributed described as what kind of variable?
10. Which of the following is correct concerning B. (i) and (iii) only
logit and probit models?
C. (i), (ii), and (iv) only
A. They use a different method of transforming the
model so that the probabilities lie between zero and D. (i), (ii), (iii), and (iv)
one
14. Under which of the following situations would
B. The logit model can result in too many observations bootstrapping be preferred to pure simulation?
falling at exactly zero or exactly one
(i) If it is desired that the distributional properties
C. For the logit model, the marginal effect of a change of the data in the experiment are the same as those
in one of the explanatory variables is simply the of some actual data
estimate of the parameter attached to that variable,
whereas this is not the case for the probit model (ii) If it is desired that the distributional
properties of the data in the experiment are
D. The probit model is based on a cumulative logistic known exactly
function
(iii) If the distributional properties of the actual
11. Suppose that we wished to evaluate the factors data are unknown
that affected the probability that an investor
would choose an equity fund rather than a bond (iv) If the sample of actual data available is very
fund or a cash investment. Which class of model small
would be most appropriate?
A. (ii) and (iv) only D. Each series has a unit root under the alternative
hypothesis
B. (i) and (iii) only