Econometric Analysis II: Practice Questions

^ = Ȳ
β
1. Consider the model: Y i =βX i +ui If β has three alternative estimators – (i)
1
X̄ ,

β^ 2 =
∑ Xi Y i
β^ 3 =
∑ ( X i − X̄ )(Y i−Ȳ )

(ii) ∑ , and (iii)

X 2i ∑ ( X i − X̄ )2 , examine if they are unbiased.
Derive the expression for the variance of each of these estimators. Which of the
following estimators has the minimum variance?

σ 2u ^ σ 2u ^ σ 2u
var ( β^ 1 )= var ( β 2 )= var ( β 3 )=
n X̄ 2 ; (ii) ∑ X 2i , and (iii) ∑ x2i
2. Consider the following two regression models:

Y i =α 0 +α 1 D 1i + α 2 X i +α 3 ( D1 i× X i ) +ui
Y i = Monthly per capita
Model I: with
X D =0 for the APL
consumption expenditure; i = Monthly per capita income; 1 i
D =1 for the BPL households; and
households, and 1 i
Y i =α 0 +α 1 D1i + α 2 X i +α 3 ( D1 i× X i ) +ui
Y i = Monthly per capita
Model II: with
X D =0 for the BPL
consumption expenditure; i = Monthly per capita income; 1 i
D =1 for the APL households
households, and 1 i
How will the regression results differ between these two models in respect of
estimates of the coefficients, their standard errors, goodness-of-fit, and statistical
significance of the model? Justify your answer.

3. If the model
Y i =γ+δX 1i +v i is estimated instead of the true model
Y i =α+ β1 X 1i +β 2 X 2i +ui , examine if the OLS estimator of the intercept will be
unbiased.
3
4. If the model
Y i =α+ β1 X 1i +β 2 X 2i +ui is estimated instead of the true model
Y i =γ+δX 1i +v i , will there be any consequence on statistical significance of the
estimated model? Justify your answer.

5. Using a suitable set of regression results, explain why specification/selection of

appropriate model is important in econometric modelling. Also, discuss how you will
select the appropriate model(s) in such cases.
6. While estimating a distributed lag model, Almon transformation process should be
preferred to the Koyck transformation procedure. Comment on the statement with
justification/proof.

7. Why do we introduce lags in econometric models? How does the adaptive expectation
model differ from the partial adjustment model? Can one estimate both the models by
applying the OLS method? Justify your answer.

8. Assume that the true regression model is

Y i =β 1 + β 2 X 2 i + β 3 X 3 i +ui . But, a student

has estimated the model i 1 Y =α +α x +α x + v

2 2i 3 3i i (here, x2 and x3 are deviations of
X2 and X3 from their respective arithmetic means). Will the regression results differ?
Justify your answer.

ut
Q t = ALαt K tβ e
9. Consider the production function . How can one examine if the
production function follows constant returns to scale?

Y i =α+ β1 X 1i + β 2 X 2i +ui
10. Prove that for regression the model ,
2
−r 12 σ
cov ( β^ 1 , β^ 2 )=
2
(1−r 12 ) √∑ x21 i ∑ x 22i
where r12 stands for the pair-wise correlation
coefficient between X1 and X2. How will you interpret this expression?