Reference no: EM132271320

Intermediate Econometrics Assignment -

For one part of the coursework you will need Stata package whitetst.ado (available on Moodle). The dataset housing.dta contains 546 observations on sales prices of houses sold during July, August and September, 1987, in the city of Windsor, Canada on the following variables:

Variable name	Variable description
logprice	log of sale price of a house (in C$)
Lotsize	lot size of a property in square feet
bedrooms	number of bedrooms
gashw	dummy variable, 1 if the house uses gas for hot water heating
bathrms	number of full bathrooms
airco	dummy, 1 if there is central air conditioning
stories	number of stories excluding basement
Table 1: Variables description

1. Write down a linear regression model whereby logprice depends on a constant, the lot size of a property, the number of bedrooms, the number of bathrooms, and whether the house uses gas for hot water heating or not.

2. Estimate the model in part (1), report the estimation output and comment on the overall goodness of fit of the model using TWO criteria.

3. Carefully interpret the coefficients estimated in part (2). Next, carefully explain how you conduct an hypothesis test on the individual statistical significance of the estimated coefficients. Last, say if the estimated coefficients are statistically significant at a 5% significance level.

4. Explain what heteroskedastic errors are and what their consequences are as regards the ordinary least square (OLS) estimator.

5. Save regression residuals. Check whether there is an association between their squares and the lot size of a property and/or the number of bedrooms and/or the number of bathrooms and/or whether the house uses gas for hot water heating or not. Explain.

6. Run a regression where you try to explain the squared residuals from the explanatory variables of the model in part (1). What do you conclude?

7. Explain how a Breusch-Pagan test for heteroskedasticity is constructed in this case. Then, per-form the test assuming the heteroskedasticity may be related to all explanatory variables of the model in part (1). What do you conclude?

8. Explain how a White test for heteroskedasticity is constructed in this case. Then, perform the test. What do you conclude?

9. Run a regression with White standard errors and compare them with the routinely computed standard errors. Does your interpretation of the regression output change?

10. Estimate a larger linear regression model that includes also the number of stories excluding basement and whether there is central air conditioning or not as explanatory variables. Report the estimation results.

11. Explain whether and why you prefer the model estimated in part (2) to the model estimated in part (10) using THREE criteria.

12. Do the estimated coefficients attached to the lot size of a property, the number of bathrooms, and whether the house uses gas for hot water heating or not considerably change across the models estimated in part (10) and in part (2). Why or why not? What about their significance? Why?

Instructions: To be solved using Stata.

Attachment:- Assignment Files.rar