3 Use the data in Housepricexls price house price 1000s lot

3. Use the data in House_price.xls: price = house price, $1000s lotsize = size of lot in square feet sqrft = size of house in square feet bdrms = number of bedrooms price = 0 + 1*lotsize + 2*sqrft + 3*bdrms

a. Apply the White test for heteroskedasticity to the above model (you can use the STATA command for this). Write down the chi-square statistic and the p-value. What do you conclude? (4 points) Now generate the natural log of several of the variables using the following commands: gen lnprice = ln(price) gen lnlotsize = ln(lotsize) gen lnsqrft = ln (sqrft) *Note: We are not taking the natural log of bdrms* Our new model will be: lnprice = 0 + 1*lnlotsize + 2*lnsqrft + 3*bdrms

b. Apply the White test for heteroskedasticity to this new model. Write down the chi-square statistic and the p-value. What do you conclude? (4 points)

4. What explanation might there be for the results that you saw in a and b above? (Hint: How would taking the natural log of variables tend to affect the White test results?) (2 points)

Solution