tut3
ECON 3208
Econometrics Methods
Tutorial 3
(Tutorials to be held in week 4)
The following questions are based on the data in file jtrain2.dta (variables are described in file jtrain2.des).
a) Using a LPM regress the probability of unemployment in 1978 against the training dummy and age, educ, black, hisp, married, and nodegree. Describe the effects of the training on the probability of unemployment.
b) Re-estimate the model described in part a) using a logit model. Compare predicted probabilities between the logit and the LPM.
c) Calculate and compare the effects of the training program on the probability of unemployment for both the LPM and the logit model.
d) Add the two variables unem74 and unem75 to the logit model and re-estimate it. Do the effects of training vary for individuals depending on their history? Compute some partial effects of training to show this.
e) Suppose that you want to allow for shifts in the coefficient on training depending on the unemployment history of the person, how would you do this? Run this new regression. Calculate the effects of training on the probability of unemployment making sure you interpret exactly what you are calculating.
f) Re-estimate the model described in part a) using a probit model.
g) Compare the predicted probabilities between the logit and the LPM to those implied by your probit results.
h) Are the results of significance tests on individual coefficients different in the three models?
i) Calculate and compare the effects of the training program on the probability of unemployment from your probit to those from the LPM and the logit model.
j) Add the two variables unem74 and unem75 to the probit model and re-estimate it. Do the effects of training vary for individuals depending on their history in the probit model? Compute some partial effects of training to show this.
k) Perform joint significance tests on the coefficients of the variables unem74 and unem75 in all three models (LPM, logit and
Econometrics Methods
Tutorial 3
(Tutorials to be held in week 4)
The following questions are based on the data in file jtrain2.dta (variables are described in file jtrain2.des).
a) Using a LPM regress the probability of unemployment in 1978 against the training dummy and age, educ, black, hisp, married, and nodegree. Describe the effects of the training on the probability of unemployment.
b) Re-estimate the model described in part a) using a logit model. Compare predicted probabilities between the logit and the LPM.
c) Calculate and compare the effects of the training program on the probability of unemployment for both the LPM and the logit model.
d) Add the two variables unem74 and unem75 to the logit model and re-estimate it. Do the effects of training vary for individuals depending on their history? Compute some partial effects of training to show this.
e) Suppose that you want to allow for shifts in the coefficient on training depending on the unemployment history of the person, how would you do this? Run this new regression. Calculate the effects of training on the probability of unemployment making sure you interpret exactly what you are calculating.
f) Re-estimate the model described in part a) using a probit model.
g) Compare the predicted probabilities between the logit and the LPM to those implied by your probit results.
h) Are the results of significance tests on individual coefficients different in the three models?
i) Calculate and compare the effects of the training program on the probability of unemployment from your probit to those from the LPM and the logit model.
j) Add the two variables unem74 and unem75 to the probit model and re-estimate it. Do the effects of training vary for individuals depending on their history in the probit model? Compute some partial effects of training to show this.
k) Perform joint significance tests on the coefficients of the variables unem74 and unem75 in all three models (LPM, logit and