This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Econ 508: Simple Regression—Inference and prediction Juan Fung MSPE February 25, 2011 Juan Fung (MSPE) Econ 508: Simple Regression—Inference and prediction February 25, 2011 1 / 19 From estimation to inference Simple regression model Recall from last time that the objective of an econometric model, Y = α + β X + is to explain the association between two variables, X and Y , by a deterministic component, E [ Y  X ] = α + β X , and a stochastic component, = Y E [ Y  X ] — which accounts for deviations from mean behavior. Juan Fung (MSPE) Econ 508: Simple Regression—Inference and prediction February 25, 2011 2 / 19 From estimation to inference OLS The method of ordinary least squares (OLS) finds α,β to minimize ESS = X i 2 i = X i ( Y i ( α + β X i )) 2 . The OLS estimators for the single regression model are ˆ β = S xy S xx , ˆ α = Y ˆ β X , where S xy = ∑ i ( X i X )( Y i Y ) and S xx = ∑ i ( X i X ) 2 . The next step is to use the OLS estimators to make inferences about α,β . Juan Fung (MSPE) Econ 508: Simple Regression—Inference and prediction February 25, 2011 3 / 19 From estimation to inference Test statistics In order to test hypotheses, or construct confidence intervals, we must characterize the sampling distributions for ˆ α, ˆ β . What distribution do ˆ α, ˆ β follow? Normal. Why? We assume i ∼ N . What are their expected values? They are unbiased estimators : E [ ˆ β ] = β, E [ˆ α ] = α . You also derived their variance: Var ( ˆ β ) = σ 2 S xx Var (ˆ α ) = σ 2 ∑ i X 2 i S xx . Juan Fung (MSPE) Econ 508: Simple Regression—Inference and prediction February 25, 2011 4 / 19 From estimation to inference Then since ˆ β ∼ N ( β,σ 2 / S xx ), Z = ˆ β β p σ 2 / S xx ∼ N (0 , 1) . What is the problem with using Z as a test statistic? σ 2 unknown!...
View
Full
Document
This note was uploaded on 05/23/2011 for the course ECON 508 taught by Professor Staff during the Spring '08 term at University of Illinois, Urbana Champaign.
 Spring '08
 Staff

Click to edit the document details