College of Business and Economics

Department of Economics

MSC. Program in Development Economics

Advanced Econometrics Course Individual Assignment 1

Important Instruction: Read the questions and instructions carefully. Answer all questions.
Be specific when you put on the answers of each question.

1. Derive the variance of the OLS estimator for an intercept ( ˆ0 ) of a simple linear regression.
2. A regression model is specified as Yi = βXi + εi, where ε and X satisfy all of the assumptions
of the classical linear regression model. The following estimator has been proposed:
Y
̂ 
X
Note that if Yi = βXi + εi, then Y   X   .
A. Prove that this estimator is unbiased and consistent.
B. What steps would you take to determine whether or not the estimator is efficient? You do
not have to actually determine whether or not it is efficient. Just outline the procedure
you would follow.

3. Coach has come up with a new estimator for the linear model: y = Xβ + μ. His estimator is
1
coach’s estimator for   ˆ  1 where ˆ is the usual OLS estimator, n = sample size (the
n
number of observations) and 1 is a “k by 1” vector of one’s (and ˆ is a “k by 1” vector).
A. Is his estimator (given the usual assumptions) unbiased? Prove it.
B. Is his estimator (given the usual assumptions) consistent? Prove it.
4. A researcher is using data for a sample of 32 companies to investigate the relationship
between annual spending Yi (measured in millions of dollars per year) and annual firm
profits Xi (measured in millions of dollars per year). Preliminary analysis of the sample data
produces the following sample information:
 N N N
N = 32,  Yi  4,917.8,
i 1
 X i  11,856.1,
i 1
Y
i 1
i
2
 4,022,814.0

N N N

X
i 1
i
2
 25,796,522.5, X Y
i 1
i i  9,785,312.0, x y
i 1
i i  7,963,252.0

N N N

y
i 1
i
2
 3,267,040.6 , x
i 1
i
2
 21,403,801.0 ,  uˆ
i 1
i
2
 304,324.7

where xi  X i  X , y i  Yi  Y uˆ i  Yi  Yˆi  Yi  ˆ0  ˆ1 X i for i = 1, ..., N. Use the

above sample information to answer all the following questions. Show explicitly all
formulas and calculations.
(a) Use the above information to compute OLS estimates of the intercept coefficient β 0 and
the slope coefficient β1.
(b) Interpret the slope coefficient estimate you calculated in part (a) -- i.e., explain what the
numeric value you calculated for ˆ1 means.
(c) Compute the value of R2, the coefficient of determination for the estimated OLS
sample regression equation. Briefly explain what the calculated value of R 2 means.
N
(d) What is the value of  X uˆ
i 1
i i for the estimated sample regression equation? Explain

briefly how you obtained your answer.

(e) Calculate the estimated variance of ˆ1 .
(f) Perform a test of the null hypothesis H0: β1= 0 against the alternative hypothesis H1: β1 ≠ 0 at
the 5% significance level (i.e., for significance level α = 0.05). State the decision rule you use,
and the inference you would draw from the test. Would you draw the same inference if you
performed the test at the 1% significance level (i.e., for significance level α = 0.01)?
(g) Compute the two-sided 95% confidence interval for the slope coefficient β 1.
(h) Perform a test of the proposition that β1 > 0.30 at the 5% significance level (i.e., for
significance level α = 0.05). State the null and alternative hypotheses, and show how you
calculated the test statistic. State the decision rule you use, and the inference you would draw
from the test. Would you draw the same inference if you performed the test at the 1%
significance level (i.e., for significance level α = 0.01)?

5. Suppose you estimate the following regression model by OLS:

 HWAˆ GE = 5.00 + 0.5YEARS - 0.01YEARS2 - 3.00FEMALE- 2.5PARTIME
(2.00) (0.10) (0.002) (2.00) (3.25)

where the numbers in brackets are estimated standard errors

a. Interpret the regression output.

b. Hourly wages reach a maximum at how many years of work experience?

c. Test the hypothesis that the female dummy variable has no explanatory power in
the model at the 95% level. Test the hypothesis that the true value of the female coefficient is -4.
How do you reconcile the two results?

d. Test the hypothesis that regression as a whole has no explanatory power.

6. Suppose Model A: Yi   0   1 X 1i   2 X 2i  u i

Model B: Yi   0 *   1 * X 1i   i

(a) What are the consequences of estimating Model B if Model A is the true model?

(b) What does the coefficient 1 * measure if Model A is the true model? Show this as precisely
as possible.

(c) What are the consequences of estimating Model A if Model B is the true model?

7. Suppose the estimation of the model Yt  β 0  β1X1t  β 2 X 2t  β 3 X 3t  ε t from a sample of

104 observations using OLS yields:
 5.0  1.4  2.0 4.8
XX    1.4 20.0  7.5 , εε  20 ,
1 ˆβ  4.0
 
 2.0  7.5 45.0  3.6

a. Which of the explanatory variables has a statistically significant impact on Y based on a t-test
criterion. Show how you arrive at your conclusions.

b. Test the hypothesis that 2 = 3.

 c. Construct and interpret a 95% confidence interval for 2 + 3.

d. With the information provided, is it possible to test the hypothesis that Ho: 1 = 2 = 3 =

0? Explain.

8. Consider the regression of Y on X variables (with a constant) Y   X  U .

Consider an alternative regression of Z = XP where P is a nonsingular matrix of order K.

a) Prove that the residual vector in the regressions of Y on X and Y on Z are identical.

b) Define a transformation matrix that expresses Y and X variables in mean deviation form and
label them as y and x. Define another transformation matrix that will convert the y and x
variables in standardized form (i.e., each of the variables will have zero mean and unit
variance).

c) Show that the effect of such transformation has no effect on R 2 , standard errors of 
and their t values.

d) Show that the residual sum of squares, êe in the OLS regression of Y on X and the regression
of y on x are the same.

9. Show the effects of classical measurement error (in the explanatory variable) on estimates in
each of the following situations:
a. OLS

b. IV

10. Assume a model of the following form: Yi = α + βXi + ui which has a heteroskedastic
disturbance generated by the following specification:
 i 2  kX i 2
a. What procedure would you use to estimate the model?
b. Compare the results of estimating the model with OLS versus the method you described in a,
above, with regard to both parameter estimates and estimated parameter standard errors. Explain.