Scribd
Upload
4 views3 pages

Problem Set 9

The document is a problem set for an econometrics course, due on May 21, containing various exercises related to regression models and hypothesis testing. It includes tasks involving the estimation of parameters, testing hypotheses, and interpreting regression coefficients using a provided dataset. The problems focus on concepts such as consistency, unbiasedness, variance, and significance levels in econometric analysis.

Uploaded by

uditm
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views3 pages

Problem Set 9

The document is a problem set for an econometrics course, due on May 21, containing various exercises related to regression models and hypothesis testing. It includes tasks involving the estimation of parameters, testing hypotheses, and interpreting regression coefficients using a provided dataset. The problems focus on concepts such as consistency, unbiasedness, variance, and significance levels in econometric analysis.

Uploaded by

uditm
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Econometrics A

Problem Set #5
Date Due: Thursday, May 21

1. Stock and Watson, Exercises 7.9, E7.3, E8.1

2. Consider the following model:

Y = β0 + β1 X1 + β2 X2 + β3 X3 + U .

Let X = (1, X1 , X2 , X3 )0 . Suppose

E[XU ] = 0

and that X is not perfectly colinear. Suppose further that E[Y 4 ] < ∞ and
E[Xj4 ] < ∞ for 1 ≤ j ≤ 3. Using a large sample of i.i.d. observations from
(Y, X1 , X2 , X3 ), you estimate this equation using OLS. Let β̂ = (β̂0 , . . . , β̂3 ) de-
note the resulting estimate of β = (β0 , . . . , β3 ).

(a) Let θ = 2β1 − β3 . Is θ̂n = 2β̂1 − β̂3 a consistent estimate of θ? Is it an


unbiased estimate of θ?
(b) Express Var[θ̂n ] in terms of Var[β̂1 ], Var[β̂3 ], and Cov[β̂1 , β̂3 ].
(c) How would you test the null hypothesis H : θ = 1 versus the alternative
K : θ 6= 1 at the 5% significance level? In particular, write down your test
statistic, your critical value, and rule you would use to determine whether
or not to reject the null hypothesis.
(d) What is the p-value for your test?

3. Consider the following regression:

Y = X 0β + U ,

where E[U |X] = 0 (remember that this is a stronger assumption than our usual
assumption that E[XU ] = 0). Suppose Y is binary.

(a) What is P {Y = 1|X}?


(b) What is Var[U |X]? Is it reasonable to assume that U is homoskedastic?
(c) Is it possible for the fitted value of Y to lie outside [0, 1]?

1
4. Download the dataset for this problem from the TA’s webpage. The dataset has
4 variables in the following order: college (an indicator for completing college), hs
(an indicator for completing high school, but not college), wage (average hourly
wage) and fem (an indicator for female). In order to load the data into Stata,
you may need to increase the memory. To do this, type

set mem 10m

in Stata. This will increase the memory from the default of 1 mb to 10 mb.

(a) Consider the following model of the determinants of income:

wage = β0 + β1 fem + β2 college + β3 hs + U .

Suppose that the regressors are not perfectly colinear.


i. What is the interpretation of U ? Is it uncorrelated with the regressors?
ii. Interpret each of the coefficients in the model.
iii. For the rest of part (a) and part (b) below, suppose U is uncorrelated
with the regressors and that the fourth moments of the dependent vari-
able and each of the regressors exist. Estimate the model using OLS
and construct a 95% confidence interval for the effect of being female
on income. What coverage property does the interval satisfy?
iv. Let θ = β2 − β3 and θ̂n = β̂2 − β̂3 .
A. Interpret θ.
B. Under what conditions will β̂2 and β̂3 be unbiased estimators for β2
and β3 , respectively? Under these conditions, will θ̂n be an unbiased
estimator of θ?
C. Are β̂2 and β̂3 consistent estimators of β2 and β3 , respectively? Is
θ̂n a consistent estimator of θ?
D. Derive an expression for the variance of θ̂n in terms of Var[β̂2 ],
Var[β̂3 ] and Cov[β̂2 , β̂3 ]. Using your results, what is your estimated
standard deviation (i.e., standard error) for θ̂n ?
v. Without running a new regression, test H0 : β2 = β3 versus H1 : β2 > β3
at the 5% significance level. Be careful to explain how you construct
the test. Explain the null and alternative hypotheses in words.
(b) Use the dataset to estimate the following model,

wage = γ0 + γ1 fem + γ2 college + γ3 (hs + college) + U.

2
To do this regression, you will have to create a new variable that is the sum
of hs and college.
i. How do the coefficients γ0 , . . . , γ3 relate to the coefficients β0 , . . . , β3
defined in part (a)?
ii. Test H0 : γ2 = 0 versus H1 : γ2 > 0 at the 5% significance level. How do
the results of your test compare to what you found in part (v) of part
(a) above? Explain briefly.
(c) Consider the following model of the determinants of income:

wage =
β0 + β1 fem + β2 college + β3 fem × college + β4 hs + β5 fem × hs + U .

Suppose that the regressors are not perfectly colinear.


i. Interpret each of the coefficients in the model.
ii. For the rest of part (c), suppose that U is uncorrelated with each of the
regressors and that the fourth moments of the regressors and dependent
variable exist. Estimate the model using OLS. According to your results,
what is the estimated effect on income of college = 1 versus hs = 1 for
men? Construct a 95% confidence interval for this effect. What coverage
property does the interval satisfy?
iii. Consider the null hypothesis that the effect on income of college = 1
versus hs = 1 is the same for men and women, versus the alternative
that they are different for men and women. Formally state the null and
alternative hypotheses. Conduct the test at the 10% level.
iv. Consider the null hypothesis that there are no interactions between gen-
der and education. Formally state the null and alternative hypotheses.
Conduct the test at the 10% level.

You might also like