1 e2 2 yi 1 2 xi 2 ie normal so the only difference

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 1 + β2 Xi β2 f (Z ) is now the change in the probability of choosing 1 given a unit change in X f(Z) is the derivative of F(Z) i.e. f (Z ) = e −Z (1+e−Z )2 f(Z) is usually evaluated at the mean of all the independent variables B KOO EC220 Revision Lectures Introduction MLE LDV Problems Introduction Binary Choice Tobit Sample Selection Bias Probit Probit: Another way to overcome the problems with LPM is to use a probit function: Pr (Yi = 1|Xi ) = F (Z ) 1 where F (Zi ) = f (Zi ) = density! − 1 √ e2 σ 2π Yi −β1 −β2 Xi σ 2 i.e. Normal So the only difference between logit and probit is the choice of the function F. B KOO EC220 Revision Lectures Introduction MLE LDV Problems Introduction Binary Choice Tobit Sample Selection Bias Tobit Unlike the previous section of LPM, Logit and Probit where the output variable was 0 or 1, we can think of a situation where the output variable can be continuous or a single value. This for example could be the wages in the labour market. Given a person’s characteristics, the person can either earn the minimum wage (normalized to 0) or something greater. Our model now is: Y= Y 0 if Y>0 otherwise where Y = β1 + β2 X + ε B KOO EC220 Revision Lectures Introduction MLE LDV Problems Introduction Binary Choice Tobit Sample Selection Bias Question - Why can’t we just do OLS? B KOO EC220 Revision Lectures Introduction MLE LDV Problems Introduction Binary Choice Tobit Sample Selection Bias Sample Selection Bias Going back to the labour market model, but this time with no minimum wage. Instead a person makes a choice. If she chooses to enter the labour market, then a wage is observed which is based on some characteristics. Otherwise, there is NO OBSERVATION. In other words the model is a two step process: decision to enter: Bi = 1 if Bi∗ = δ1 + δ2 Qi + εi > 0 and 0 otherwise wage: Yi = Yi∗ = β1 + β2 X2i + β3 X3i + ui if Bi = 1 and there is NO OBSERVATION otherwise. B KOO EC220 Revision Lectures Introduction MLE LDV Problems Introduction Binary Choice Tobit Sample Selection Bias P6 2005 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P7 2009 B KOO EC220 Revision Lectures Introduction MLE LDV Problems P6 2007 also covers LDV and MLE in a similar fashion. For LDV, Drawbacks of LPM Probit, Logit as an alternative to LPM Tobit, Sample Selection Bias For MLE, Construction of (Log)likelihood function MLE estimator as a maximizer of the (log)likelihood function : FOC, SOC B KOO EC220 Revision Lectures Introduction Measurement Error Simultaneous Equations EC220 Revision Lectures Lecture 3 Can Celiktemur London School of Economics 7 May 2010 Past Exam Practice Question Introduction Measurement Error Simultaneous Equations Setup Of Review Lectures Revision Lecture Topics 1 April 27 (Tuesday 15:00-17:00): Omitted Variable Bias, Including Irrelevant Variables 2 April 28 (...
View Full Document

Ask a homework question - tutors are online