悉尼大学计量经济学原理课件lec11 ECMT5001 sem1 .pptVIP

ECMT5001 Principles of Econometrics Lecture 11 Semester 1, 2010 Heteroskedasticity The White Test With six independent variables in the original model, the White regression could involve 27 regressors. This abundance of regressors is a weakness of the White test: it uses many degrees of freedom for models with just a moderate number of independent variables. * Example 3: House prices in Boston In an attempt to better understand the determinants of value of owner occupied homes, consider the following specification of a demand function: Valuei = b0 + b1Lcrimei + b2Roomsi  + b3Accessi + b4Ltaxi + b5PTRatioi + ui This model is estimated using a sample of 506 individuals. Data definitions and associated summary statistics are given in Table 1.1 * Example 3: House prices (Table 1.1) * Example 3: House prices in Boston The following are some examples of questions that might be asked in an exam. They will give you an idea of the kind of information you might need to interpret. A) Do you think heteroskedascity is likely to be a problem? B) What problems might occur if the incorrect standard errors are used (i.e. if hetero exists)? C) What output should you be using and why? * Example 3: House prices in Boston * Example 3: House prices in Boston Note the robust se Then Why? * Example 3: House prices in Boston D) One possible option for the form of σi2 is σi2 = σ2Taxi. Redefine the model so that the variances of the errors are constant and then estimate the corrected model: * Example 3: House prices in Boston where * Example 3: House prices in Boston Computing the new model: Create the new variables in Gretl using the “Define new variable” from the “Add” tab in the main menu:Const1=1/(Tax^0.5) Lcrime1=Log(Crime)/(Tax^0.5) rooms1= Rooms/(Tax^0.5) access1=Access/(Tax^0.5) Ltax1=log(Tax)/(Tax^0.5)ptratio1=PTRatio/(Tax^0.5) * Example 3: House prices in Boston (GLS/WLS) * Example 3: House prices in Boston (WLS in Gretl) * Breusch-Pagan Test The Breusch-Pagan tes