Economics 140A
Winter 2015
Jones/White
Homework 3
Note: In this assignment, CPS data refers to the file cps_march_2014.wf1
1.
a. Find the formula for var(1 2 ).
b. Find the formula for var(1 1 + 2 2 )
2. Use the data set cdcosts2010 to estimate the margin
Fiona Jiayuan Li
Homework 4
Econ 140A
11/12/2014
series male = 1- Fe
ls wsal_val c ed age age^2 fe black male*black fe*ed black*ed male*black*ed
Dependent
Variable:
WSAL_VAL
Method: Least
Squares
Date: 11/12/14
Time: 14:55
Sample: 1 89984
Included
observa
10/9/2014
Goals
Idea of estimator as a random variable.
Sampled over repeated draws.
Review/introduce expected value, variance,
covariance
Sample Mean and Related Issues
Rules from mathematical statistics for
E , var , cov()
Derive distribution of sa
11/21/2014
Goals
What happens when error terms are
correlated with one another?
Quasi-differencing and Cochrane-Orcutt.
Testing for serial correlation.
Lagged dependent variables and serial
correlation.
Serial correlation
Copyright Dick Startz
1
Copyr
Economics 140A
Winter 2015
Jones/White
Final
Answer all 6 questions. Each question is worth 15 points, although some questions are easier
than others. Be sure to show your work for each question.
Write your answers on the pages provided separately from th
Economics 140A
Winter 2015
Jones/White
Final
Answer all 6 questions. Each question is worth 15 points, although some questions are easier
than others. Be sure to show your work for each question.
Write your answers on the pages provided separately from th
Economics 140A
Winter 2015
Jones/White
Quiz 2
Answer all 4 questions. Each question is worth 15 points, although some questions are easier
than others. Be sure to show your work for each question.
Write your answers on the pages provided separately from t
Economics 140A
Winter 2015
Jones/White
Quiz 2
Answer all 4 questions. Each question is worth 15 points, although some questions are easier
than others. Be sure to show your work for each question.
Write your answers on the pages provided separately from t
Economics 140A
Winter 2015
Jones/White
Quiz 1
Answer all 4 questions. Each question is worth 15 points, although some questions are easier
than others. Be sure to show your work for each question.
Write your answers on the pages provided separately from t
Economics 140A
Fall 2014
Jones/White
Quiz 1
Answer all 4 questions. Each question is worth 15 points, although some questions are easier
than others. Be sure to show your work for each question.
Write your answers on the pages provided separately from the
11/25/2014
Goals
Individual fixed effects
Control for unobservables at individual level
Panel data
Fixed Effects and
Difference-in-Differences
Difference-in-differences
Control for unobservables at group level
Repeated cross-section data
*Both are
10/9/2014
Goals
Interpret regression coefficient as marginal
effect.
Error terms versus residuals.
Derive least squares estimator.
Simple Regression
Copyright Dick Startz
1
What is the MC and FC of producing
CDs?
Generic regression model
= + +
CD e
10/14/2014
Goals
Derive distribution of regression coeffficient.
Expected value, variance, normal distribution
Regression continued
Copyright Dick Startz
Which assumptions are needed for which
results?
Gauss-Markov theorem
t-tests and z-tests
1
Copyr
10/28/2014
Goals
Extend simple regression to right hand side
variables.
Multicollinearity.
t-tests involving more than one coefficient.
F-tests involving more than one restriction.
Multiple Regression
Copyright Dick Startz
1
2
What determines pass rat
11/6/2014
Goals
Omitted variable.
Correlated with included variables or not.
Theils misspecification theorem.
Misspecification
Included extraneous variable.
Correlated with error or not.
Copyright Dick Startz
1
Copyright Dick Startz
2
Problem: correl
11/4/2014
Goals
Dummy (0/1) variables.
Interpretation.
Multiple ways of writing.
Dummy variable trap.
Multiple categories.
Dummy variables and functional
forms
Functional forms
Logs
Power series
Copyright Dick Startz
1
Whats nonlinear?
Copyright D
9/5/2014
Goals
Individual fixed effects
Control for unobservables at individual level
Panel data
Fixed Effects and
Difference-in-Differences
Difference-in-differences
Control for unobservables at group level
Repeated cross-section data
*Both are ju
11/21/2014
Goals
What happens when error terms are
correlated with one another?
Quasi-differencing and Cochrane-Orcutt.
Testing for serial correlation.
Lagged dependent variables and serial
correlation.
Serial correlation
Copyright Dick Startz
1
Copyr
9/5/2014
Goals
Linear Probability Model
Probit Regression
Limited Dependent Variables
Questions to be answered
What traits get you into law school?
Who tends to have extramarital affairs?
Binary Dependent Variable
UWashington Law School Admissions
T
11/13/2014
Goals
What happens when error variances differ
across observations?
Generalized least squares (GLS).
Heteroskedasticity
Weighted least squares (WLS).
Testing for heteroskedasticity.
Robust standard errors.
ARCH (if time)
Copyright Dick St
Economics 140A
Winter 2015
Jones/White
Homework 6
Open up the workfile called HW6_Data. Were going to try estimating some supply and
demand curves. Suppose the system is
= 3 2 +
= 3 + + 3 +
(demand)
(supply)
where and are price and quantity (perhaps t
Economics 140A
Winter 2015
Jones/White
Homework 4
Use our CPS data for this homework.
Consider the regression
= 0 + 1 + 2 + 3 2 + 4 + 5 + 6
+ 7 + 8 + 9
1. Give point estimates for salaries for the following (you might want to use Excel to save
time on
Economics 140A
Winter 2015
Jones/White
Homework 5
2
1. Consider the regression = + . There are = 1 + 2 observations and () = 1
2
for the first = 1,2, , 1 observations, while () = 2 for the last = 1 + 1, 1 +
2, , 1 + 2 .
a. Derive the variance of . (You ma
2/9/2015
Goals
Omitted variable.
Correlated with included variables or not.
Theils misspecification theorem.
Misspecification
Included extraneous variable.
Correlated with error or not.
Copyright UCSB Economics
1
Copyright UCSB Economics
2
Problem: c
5/12/2015
Goals
Weighted averages
What happens when error variances differ
across observations?
Generalized least squares (GLS).
Weighted Least Squares and
Heteroskedasticity
Weighted least squares (WLS).
Robust standard errors.
Copyright UCSB Econom
5/7/2015
Last Time
Omitted Variables Bias (Theils Misspecification Theorem)
True DGP:
You Estimate:
Fixed Effects and
Difference-in-Differences
= + +
= +
If you want the effect of on , youre trying to estimate !
But you estimate , and with some math:
Micro Paper B: Forecasting Consumer Expenditure
The US Bureau of the Census has asked you to analyze data on consumer expenditure to help
forecast the aggregate demand for certain goods and services. You are given aggregate annual
data for 1959-2003 from
1)
ls cost c cds
2)
ls cost c cds cds^2
3)
smpl if ca_res = 1
series male = 1 - fe
LS LNWAGE FE MALE
LS LNWAGE C FE
LS LNWAGE C MALE
b)
ls wsal_val c ed age fe 1-fe
What happens and why?
4)
LS LNWAGE C FE
smpl if ca_res=1
ls lnwage c age
5)
smpl if ca_res
Causality
and Endogeneity
Goals
Causality
Endogeneity and Exogeneity
Omitted Variables Bias
Simultaneity Bias
Solution: Instrumental Variables
Two Stage Least Squares
Causality
= + +
Often (but not always), were interested in estimating the
causa
Economics 140A
Winter 2016
Bana/White
Final
Answer all 6 questions. Each question is worth 15 points, although some questions are easier than
others. Be sure to show your work for each question.
Write your answers on the pages provided separately from the
Economics 140A
Winter 2016
Bana/White
Quiz 1
Answer all 4 questions. Each question is worth 15 points, although some questions are easier than
others. Be sure to show your work for each question.
Write your answers on the pages provided separately from th