London School of Economics Econometrics Final Assessment exam with
questions and answers with rationales 2026
London School of Economics (LSE)
Econometrics Final Assessment Exam 2026
Questions with Answers and Rationales
.
Section 1: Model Specification & Methodology (Questions 1–20)
Question 1
In the LSE "general-to-specific" modeling approach, the first step is to:
A) Estimate a parsimonious model and then add variables
B) Specify a General Unrestricted Model (GUM) that encompasses the data
generation process
C) Perform unit root tests on all variables
D) Run a simple bivariate regression
Answer: B
,Rationale: The LSE methodology (associated with Denis Sargan and David Hendry)
prioritizes starting with a General Unrestricted Model (GUM) that includes
relevant lags and variables. This GUM is then simplified ("reduced") through
testing to arrive at a congruent, parsimonious final model .
Question 2
What is meant by "marginalisation" in the context of the LSE dynamic modeling
tradition?
A) Excluding variables with low statistical significance
B) Eliminating variables that are irrelevant to the parameters of interest
C) Adding irrelevant regressors to the model
D) Testing for unit roots in the residuals
Answer: B
Rationale: Marginalisation refers to the process of eliminating variables from the
model that do not affect the parameters (coefficients) of interest for the specific
hypothesis being tested. Combined with conditioning (focusing on relevant
variables), this allows for a reduction in model complexity .
Question 3
The "Encompassing Principle" in econometrics states that:
,A) A model should include all possible variables
B) A model must be able to explain the results of rival models
C) Only models with the highest R-squared are valid
D) Models should never be simplified
Answer: B
Rationale: The Encompassing Principle requires that a proposed econometric
model should be able to account for (or "encompass") the findings of alternative,
competing models. If a new model can explain the results of the old model, it is
considered superior .
Question 4
Which of the following is NOT a primary goal of the LSE general-to-specific
methodology?
A) Congruence (the model matches the data evidence)
B) Parsimony (the model is as simple as possible)
C) Data mining (searching for any significant result)
D) Encompassing (the model explains other models)
Answer: C
, Rationale: The LSE tradition explicitly guards against "data mining." While it
involves extensive testing, the statistical properties of the tests are maintained by
fitting a general model first. Data mining implies searching data for patterns to
achieve significance, which violates the principle of pre-specifying the GUM.
Question 5
The rational for an Error Correction Model (ECM) is that:
A) Short-run dynamics are independent of long-run equilibrium
B) Changes in variables adjust to past disequilibrium (deviations from long-run
relationship)
C) Only first differences of variables should be used
D) It does not require stationarity of variables
Answer: B
Rationale: The ECM captures both short-run dynamics and long-run equilibrium.
The "Error Correction" term (the lagged residual from the cointegrating
regression) measures the deviation from long-run equilibrium and forces the
model to adjust back toward that equilibrium over time .
Section 2: Bivariate & Multivariate Regression Theory (Questions 21-50)
Question 6
questions and answers with rationales 2026
London School of Economics (LSE)
Econometrics Final Assessment Exam 2026
Questions with Answers and Rationales
.
Section 1: Model Specification & Methodology (Questions 1–20)
Question 1
In the LSE "general-to-specific" modeling approach, the first step is to:
A) Estimate a parsimonious model and then add variables
B) Specify a General Unrestricted Model (GUM) that encompasses the data
generation process
C) Perform unit root tests on all variables
D) Run a simple bivariate regression
Answer: B
,Rationale: The LSE methodology (associated with Denis Sargan and David Hendry)
prioritizes starting with a General Unrestricted Model (GUM) that includes
relevant lags and variables. This GUM is then simplified ("reduced") through
testing to arrive at a congruent, parsimonious final model .
Question 2
What is meant by "marginalisation" in the context of the LSE dynamic modeling
tradition?
A) Excluding variables with low statistical significance
B) Eliminating variables that are irrelevant to the parameters of interest
C) Adding irrelevant regressors to the model
D) Testing for unit roots in the residuals
Answer: B
Rationale: Marginalisation refers to the process of eliminating variables from the
model that do not affect the parameters (coefficients) of interest for the specific
hypothesis being tested. Combined with conditioning (focusing on relevant
variables), this allows for a reduction in model complexity .
Question 3
The "Encompassing Principle" in econometrics states that:
,A) A model should include all possible variables
B) A model must be able to explain the results of rival models
C) Only models with the highest R-squared are valid
D) Models should never be simplified
Answer: B
Rationale: The Encompassing Principle requires that a proposed econometric
model should be able to account for (or "encompass") the findings of alternative,
competing models. If a new model can explain the results of the old model, it is
considered superior .
Question 4
Which of the following is NOT a primary goal of the LSE general-to-specific
methodology?
A) Congruence (the model matches the data evidence)
B) Parsimony (the model is as simple as possible)
C) Data mining (searching for any significant result)
D) Encompassing (the model explains other models)
Answer: C
, Rationale: The LSE tradition explicitly guards against "data mining." While it
involves extensive testing, the statistical properties of the tests are maintained by
fitting a general model first. Data mining implies searching data for patterns to
achieve significance, which violates the principle of pre-specifying the GUM.
Question 5
The rational for an Error Correction Model (ECM) is that:
A) Short-run dynamics are independent of long-run equilibrium
B) Changes in variables adjust to past disequilibrium (deviations from long-run
relationship)
C) Only first differences of variables should be used
D) It does not require stationarity of variables
Answer: B
Rationale: The ECM captures both short-run dynamics and long-run equilibrium.
The "Error Correction" term (the lagged residual from the cointegrating
regression) measures the deviation from long-run equilibrium and forces the
model to adjust back toward that equilibrium over time .
Section 2: Bivariate & Multivariate Regression Theory (Questions 21-50)
Question 6
