ECON2209 Lecture 34: 24April_EconStats_Karagodsky
Document Summary
An unbiased estimator: e[ ] = : y = + x + , y = + x, e[ ] = , e[ ] = . An estimator is consistent if : n > as sample size increases, > the estimator becomes closer to being accurate. If something like household income were there this would be an issue. Hhincome might be correlated to education, so you are double counting for education. Causes of indogeneity: omitted variable bias (forgetting hhincome, measurement error, reverse causality (cov(x, ) 6= 0) 1 wage = + m ale: you collect data on income between males and females and you nd out that some of the data for males was entered as female and vice versa. There would be a downward bias on : price = + quantity + can"t be done, p e ects q and q e ects p. they are set at the same time.
