V31.0402 Notes on Algebra for the Two-Stage Least
Squares Estimator
Konrad Menzel
February 24, 2014
We want our instrumental variables model to allow for multiple regressors and multiple
instrumental variables. Suppose we have n i.i.d. observations of a d
Matrix Algebra Cheat Sheet
Konrad Menzel
January 27, 2014
For our purposes, matrices are a notational trick to organize a large number of
variables.
An n k matrix A is an array of n rows and k columns that has the form
a11 a12 a1k
a21 a22
A = [aij ] = .
V31.0402 Notes on Measurement Error in Linear
Regression
Konrad Menzel
February 5, 2014
1
Measurement Error in Survey Data
Survey data used for estimation can often be very messy, mainly for the following reasons:
1. the survey doesnt really measure the e
V31.0402 Notes on Partitioned/Residual Regression
Konrad Menzel
February 9, 2014
1
Motivation
Wed like to understand better how introducing additional regressors aects the OLS
coecients in the linear regression framework.
Suppose we want to estimate a Phi
V31.0402 Notes on Random Coecient Models
Konrad Menzel
March 10, 2014
1
Motivation
In the empirical literature on education and earnings, the main object of interest is the
human capital earnings function (HCEF)
Yi = + Si + 1 Xi + 2 Xi2 + i
Typically, IV
P1: JYD/.
CB495-03Drv
CB495/Train
3
3.1
KEY BOARDED
May 25, 2009
15:49
Char Count= 0
Logit
Choice Probabilities
By far the easiest and most widely used discrete choice model is logit.
Its popularity is due to the fact that the formula for the choice proba
DISCURSO DIRECTO
DISCURSO INDIRECTO*
PRESENTE
-Maana ir a Canad.
I will go to Canada
tomorrow.
Dice que ir a Canad
maana.
He says that he will go to
Canada tomorrow.
PASADO
Dijo que ira a Canad al da
siguiente.
He said he would go to Canada
the following