Econometrics 2 Homework #1

Due in class on March 27

Attention: Please keep in mind the following two rules that will apply to all homework
assignments in this course. (Rule #2 also applies to exams.)
(1) Please place your homework, on the day it is due, on my desk in the classroom before
the lecture starts.
(2) Please make sure that your work is neat, perfectly legible, and stapled in correct order.
In case our TA and I have trouble reading or locating your work, it may end up ignored in
grading.

1 (10pts). A classmate is interested in estimating the variance of the error term in the
equation yi = 0 + 1 xi + "i with homoskedasticity assumed and xi being endogenous.
(1) Suppose that she uses the estimator from the second-stage regresssion of 2SLS:

1 Pn b2SLS b2SLS X
bi
2
b2a = i=1 Yi 0 1
n 2

where X bi is the …tted value from the …rst-stage regresssion. Is this estimator consistent?
(You don’t need to formally prove, use heuristic argument)
(2) Is the estimator below consistent?

1 Pn b2SLS b2SLS Xi
2
b2a = i=1 Yi 0 1
n 2

2 (10pts). Consider the structural equation y = 0 + 1 x + 2 z1 + ", where x is endogeous

while z1 is exogeneous and Cov (x; z1 ) 6= 0: To identify the structural equation, we …nd two
valid instrumental variables, z2 and z3 ; and use 2SLS method to estimate the structural
equation. Suppose in the …rst stage, we regress x on (z2 ; z3 ) to obtain the predicted value
b and use it in the second stage regression, i.e. regress y on (b
x x; z1 ) : Does this procedure
generate consistent estimator for 1 ? Prove your claim.
3 (15pts). (This is Question 15.2 on Page 537 in the textbook by Wooldridge (4th edition))
Suppose that you wish to estimate the e¤ect of class attendence on college student perfor-
mance. A basic model is

stndf nl = 0 + 1 atndrte + 2 priGP A + 3 ACT +u

where stndf nl is the standardized score on a …nal exam, atndrte is the attendence rate,
priGP A is the prior-college GPA, and ACT stands for ACT (American College Testing)
score.
(1) Let dist be the distance from the students’living quarters to the lecture hall. Do you
think dist is uncorrelated with u?
(2) Assuming that dist and u are uncorrelated, what other assumption must dist satisfy to
be a valid IV for atndrte?
(3) Suppose we add the interaction term priGP A atndrte to the regression

stndf nl = 0 + 1 atndrte + 2 priGP A + 3 ACT + 4 priGP A atndrte + u

1
If atndrte is correlated with u, then, in general, so is priGP A atndrte: What might be a
good IV for priGP A atndrte? (Hint: suppose priGP A; ACT; dist are all exogeneous, i.e.
E [ujpriGP A; ACT; dist] = 0)

4 (10pts). (This is Question 16.4 on Page 568 in the textbook by Wooldridge (4th edition))
Suppose that annual earnings and alcohol consumption are determined by the SEM

log (earnings) = 0+ 1 alcohol + 2 edu + u1

alcohol = 0+ 1 log (earnings) + 2 edu + 3 log (price) + u2

where price is a local price index for alcohol, which includes state and local taxes. Assume
that edu and price are exogeneous. If 1 ; 2 ; 1 ; 2 and 3 are all di¤erent from zero, which
equation is identi…ed? How would you estimate that equation?

5 (15pts). With a single explanatory variable, the equation used to obtain the between
estimator is
y i = 0 + 1 x i + ai + u i
where the overbar represents the average over time for each individual. We can assume that
E [ai ] = 0: Suppose that ui is uncorrelated with xi , but Cov (xit ; ai ) = xa for all t and all
individual.
(1) Letting e1 be the between estimator, that is, the OLS estimator using time averages,
show that
plime1 = 1 + xa =V ar (xi )
where the probability limit (plim) is de…ned as N ! 1 (the number of individual gets
larger).
(2) Assume further that the xit , for all t = 1; :::; T are uncorrelated with constant variance
2 e 2
x : Show that plim 1 = 1 + T ( xa = x ) :
(3) If the explanatory variables are not very highly correlated across time, what does part
(2) suggest about whether the inconsistency in the between estimator is smaller when there
are more time periods? And how does the answer change if the explanatory variable are
indeed highly correlated across time (consider the extreme case where xit does not chage
over time)?

Empirical Exercise:

6 (20pts). How does fertility a¤ect labor supply? That is, how much does a woman’s labor
supply fall when she has an additional child? In this exercise you will estimate this e¤ect using
data for married women from the 1980 U.S. Census. Use the dataset fertility_small.dta
for this problem. The data set contains information on married women aged 21-35 with two
or more children.
(1) Regress weeksm1 on the indicator variable morekids using OLS. On average, do women
with more than two children work less than women with two children? How much less?
(2) Explain why the OLS regression estimated in (1) is inappropriate for estimating the
causal e¤ect of fertility (morekids) on labor supply (weeksm1 ).
(3) The data set contains the variable samesex, which is equal to 1 if the …rst two children
are of the same sex (boy-boy or girl-girl) and equal to 0 otherwise. Are couples whose …rst

2
two children are of the same sex more likely to have a third child? Is this e¤ect statistically
signi…cant?
(4) Explain why samesex is a valid instrument for the IV regression of weeksm1 on morekids.
(5) Estimate the regression of weeksm1 on morekids using samesex as an instrument. How
large is the fertility e¤ect on labor supply?

7 (20pts). For this exercise, use the data PENSION.dta which contains information on
participant-directed pension plans for U.S. workers. Some of the observations are for couples
within the same family, so this data set consititutes a small cluster sample (with cluster size
of two).
(1) Ignoring the clustering by family, use OLS to estimate the model

pctstck = + 1 choice + 2 prf tshr + 3 f emale + 4 age + 5 educ

0
+ 6 f inc25 + 7 f inc35 + 8 f inc50 + 9 f inc75 + 10 f inc100
+ 11 f inc101 + 12 wealth89 + 13 stckin89 + 14 irain89 + u

where the variables are de…ned in the data set. The variable of most interest is choice,
which is a dummy variable equal to one if the worker has a choice in how to allocate pension
funds among di¤erent investments. What is the estimated e¤ect of choice? Is it statistically
signi…cant?
(2) Are the income, wealth, stock holding, and IRA holding control variables important?
Explain.
(3) Determine how many di¤erent families there are in the data set. Now obtain the standard
errors for OLS that are robust to cluster correlation within a family. Do they di¤er much
from the usual OLS standard errors?
(4) Estimate the equation by di¤erencing across only the spouses within a family. Why do
the explanatory variables asked about in part (2) drop out in the …rst-di¤erenced estimation?
(5) Are any of the remaining explanatory variables in part (4) sigini…cant?