ECON704 Discussion 3: Answer Key
Wooyoung Kim
Sep 26, 2014
1. (a) 0 + 201 + 2
(b) The expected test score of Maya would increase by |151 |, all else
unchanged.
2. (a) (see Lecture Note 3, p.10) The error term can be understood as
summarizing the factors t
ECON704 Discussion 12
Wooyoung Kim
Dec 5, 2014
1
STATA: ivregress
How can we apply instrument variable approach in STATA? You can use the
command ivregress.
Use CARD.DTA. We want to estimate the following causal model:
log(W age) = 0 + 1 Educ + 2 Exper +
Reading Instruction:
Focus on Introduction, Section I and III.
American Economic Association
American Economic Association
http:/www.jstor.org/stable/30034396 .
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, avai
Econ 704: Econometrics I
Instructor: Xiaoxia Shi
TA: Wooyoung Kim
Lectures: MW 2:30pm-3:45pm SS 5231.
TA sessions: TBA
Instructor Office Hours: MW 3:45pm 4:45pm.
TA Office Hours: M 1:00pm-2:00pm, Th 10:30am 11:30am.
Course webpage: We will be using Piazza
Quiz 4 12/01
Let Y be a dependent variable, X1 and X2 be two treatment variables. Let the
causal model for Y be:
Y = 0 + 1 X1 + 2 X2 + U,
(1)
where U summarizes other factors that contributes to the generation of Y . Assume
that E(U |X1 , X2 ) = E(U ). Un
Problem Set 5, due 12/03, before class starts.
1. Exercise Questions 1 in Lecture 9
(a) Answer: The paper studies the causality between a fully anticipated temporary wage increase and 1) working time choice and 2) the eort during
working time.
(b) Answer:
Problem Set 4, due 11/10, before class starts.
1. Let Y be a dependent variable and X = (1, D) , where D is an indicator
variable. Consider the estimation of the model E(Y |X) = 0 + 1 D using an
i.i.d. sample cfw_(Yi , Di )n of (Y, D). Let (0 , 1 ) .
i=1
Problem Set 6, due 12/16 1pm in Wooyoungs mailbox
1. Exercise Questions 1 in Lecture 10
Table 1: Estimation
(a)
coef.(s.e.)
cigs
-0.222
(-1.538)
cig0
5.881*
(2.430)
parity
fatheduc
white
male
constant
114.180*
(48.660)
results
d (b)
coef.(s.e.)
-0.425*
(-
Problem Set 4, due 11/10, before class starts.
1. Let Y be a dependent variable and X = (1, D) , where D is an indicator
variable. Consider the estimation of the model E(Y |X) = 0 + 1 D using an
i.i.d. sample cfw_(Yi , Di )n of (Y, D). Let (0 , 1 ) .
i=1
1
Exercises
1. Review Example 3 in Lecture 4. Suppose that you want to estimate the production function of an industry of a certain product (say, cement). Assume that
the production function is of the Cobb-Douglas form described in the question.
Also supp
ECON704 Discussion 13
Wooyoung Kim
Dec 12, 2014
1
Final Information
Hours: Dec 18 (Next Thursday) 5:05-7:05 PM (There will be some extra time if you want)
Location: 5106 Social Science
Checking Grade / Claim : Hopefully Dec 20 (Saturday) morning but no
Large-Sample Approximations
Wooyoung Kim
University of Wisconsin-Madison
October 16, 2014
Wooyoung Kim
Large-Sample Approximations
Introduction
Lets begin with a benchmark case as below:
Suppose you are risk-neutral.
You ip the coin: you can get nothing w
ECON704 Discussion 3
Wooyoung Kim
Sep 26, 2014
1
Problem Set 1 Review
a
, where a is a
scalar, is a k 1 vector and is a k k matrix. Find the element
at the rst row and rst column of A1 . You can assume invertibility of
anything that needs to be inverte
ECON704 Discussion 6 Answer Key
Wooyoung Kim
Oct 17, 2014
1
Exercises
1. (From Lecture Note 5 Ex.1(d) Let Xi = (1, Xi ) for all i. Write down
n
1
E[n
i=1 Xi Yi ] in terms of Xi and Yi .
Answer:
n
n
1
Xi
Xi Yi ] = E[n1
E[n1
i=1
i=1
n
= n1
= n1
i=1
= n1
Yi
ECON704 Discussion 2
Wooyoung Kim
Sep 19, 2014
Before start: Linear algebra and probability are not the focus of this class.
DONT spend too much time to understand every small feature about them.
Instead, use it as a reference when we advance to econometr
Final Exam (12/18/2013, 5:05pm - 7:05pm)
1. Please do not turn this page over until you are instructed otherwise.
2. Try to follow the suggested time for each question. Each minute corresponds
to one point. There are 100 points in total. If you nish in su
ECON704 Discussion 1
Wooyoung Kim
Sep 12, 2014
1
General Information
My name is Wooyoung Kim([email protected]), 2nd year Ph.D. student
in economics.
Oce Hours : Monday 1-2PM, Thursday 10:30-11:30AM @ SS7218
Please send your questions before visiting my
ECON704 Discussion 9
Wooyoung Kim
Nov 7, 2014
1
Midterm II Notication
Date: Nov 12 (next Wednesday) in class (no extra time)
Location: EDU L196 (Same as Midterm I)
TA Oce hours: Mon 1-2pm, Tue 10:30-11:30am, 4-5pm (No oce hour on Thursday)
2
How to per
Problem Set 5, due 12/03, before class starts.
1. Exercise Questions 1 in Lecture 9
2. Exercise Questions 3 in Lecture 9
3. Exercise Questions 4 in Lecture 9
4. Exercise Questions 5 in Lecture 9
5. Exercise Questions 6 in Lecture 9
6. Exercise Questions 9
Problem Set 3 Solution, due 10/29, before class starts
1. Suppose that you have two independent unbiased estimators 1 and 2 of the
same parameter . Suppose that V ar(1 ) = v1 and V ar(2 ) = v2 . What linear
combination (c) = c1 1 +c2 2 makes the minimum v
Xiaoxia Shi
Econ 704: Econometrics I
Lecture 10:
Estimating the Linear Causal Model - II
In Lecture 9, we focused on the simple causal model (Y = 0 +1 X +U ) with one
dependent variable X and one explanatory variable Y and one causal eect parameter
to est
Lecture 8: Hypothesis Testing and Model
Specication
1
Hypothesis Testing
For an unknown parameter of interest, e.g., the population mean of test score for
school districts with student-teacher ratio less than 20, we have learned how to answer
the question
Xiaoxia Shi
Econ 704: Econometrics I
Lecture 5. Estimating the Linear Conditional
Mean Model - I
So far we have talked extensively about how to interpret the coecients in a
conditional mean model, taking them as known. But in practice, we do not know
thos
Xiaoxia Shi
Econ 704: Econometrics I
Lecture 6. Estimating the Linear Conditional
Mean Model II
1
Sample Analogue Estimator for Conditional Mean
Model
Now consider the general linear conditional mean model:
Y = X + U,
(1)
where U = Y E(Y |X), Y is a scala
Lecture 9: Estimating the Linear Causal Model - I
Recall (from Lecture 3) that we divide economic models into three types: descriptive, causal and forecasting. For descriptive models, we focused on the linear
conditional mean models, and discussed the int
Lecture 7: Standard Error and Condence
Interval Based on Asymptotically Normal
Estimators
1
Standard Error
Previously we showed that the OLS estimator of in the conditional mean model:
E(Y |X) = X is asymptotically normal under the four basic assumptions.