ECMT2150 INTERMEDIATE ECONOMETRICS Week 12 Tutorial
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
ECMT2150 INTERMEDIATE ECONOMETRICS
Week 12 Tutorial – Natural Experiments (and IV) + Panel Data Models
Q1. Natural Experiments
Read the following paper (available via our Reading List on Canvas – in the folder “Journal Articles” ) and answer the questions below :
Leigh, Andrew and Ryan, Chris (2008) “Estimating Returns to Education Using Different Natural Experiment Techniques” Economics of Education Review, 27, pp. 149 – 160.
a) What is the authors’ research question? Why is this an important question?
b) What effect does omitting ability from the model have on the estimated effect of education on wages?
c) How do “natural experiments” deal with the omitted variable bias problem?
d) Would you expect the estimated effect of education on wages to be similar in all of the natural experiments the authors consider? Why or why not? What’s the take-away message?
Q2. Wooldridge Chp 13 Q5
Suppose that we want to estimate the effect of several variables on annual saving and that we have a panel data set on individuals collected on Jan 31st 2010 and Jan 31st 2012. If we include a year dummy for 2012 and use first differencing, can we also include age in the original model? Explain.
Stata 1 (Wooldridge Chp 13 Computer Exercise 7)
Use GPA3.dta for this exercise.
• The data set is for 366 student-athletes from a large university for fall and spring semesters. [A similar analysis is in Maloney and McCormick (1993), but here we use a true panel data set.]
• Because you have two terms of data for each student, an unobserved effects model is appropriate.
• The primary question of interest is this: Do athletes perform more poorly in school during the semester their sport is in season?
a) What variable is the index for observations i=1, …,n and what variable is the time index?
b) Use pooled OLS to estimate a model with term GPA (trmgpa) as the dependent variable. The explanatory variables are spring, sat, hsperc, female, black, white, frstsem, tothrs, crsgpa, and season.
c) Interpret the coefficient on season. Is it statistically significant? Is it economically significant?
d) Most of the athletes who play their sport only in the fall are football players. Suppose the ability levels of football players differ systematically from those of other athletes. If ability is not adequately captured by SAT score (sat) and high school percentile rank (hsperc), explain why the pooled OLS estimators will be biased.
e) Now, use estimate the model in first differences (i.e with the data differenced across the two terms). Which variables drop out/must you leave out?
f) Using the estimates from the model in first difference test for an in-season effect.
g) Can you think of one or more potentially important, time-varying variables that have been omitted from the analysis? Why should we be worried if these are omitted from the model?
Stata 2 (Wooldridge Chp 14 Computer Exercise 2)
Use CRIME4.dta for this exercise.
This data is panel data on 90 counties in North Carolina in the US for the years 1981- 1987. See Example 13.9 in the textbook for more details on the data.
Our unobserved effects model is:
ln(cTmTteit ) = F0 + F1 ln(pTbaTTit ) + F2 ln(pTbconvit ) + F3 ln(pTbpTisit )
+F4 ln(avgsenit ) + F5 ln(polpcit ) + 61 d82t + … + 66 d87t + ai + uit where i indexes the counties and t the years from 1981- 1987.
Please use cluster robust standard errors with all models here. The cluster variable is the county.
a) What kind of factors are captured in the unobserved effect term ai ? How does using a panel data model such as first differences or fixed effects help with mitigating the endogeneity concerns we have discussed during the semester when estimating a model of crime rates using city or county level data?
b) Estimate the model in first differences. That is, reproduce the results in Example 13.9. Note that your coefficient estimates should match those in the square brackets [ ] as you are using cluster robust standard errors.
c) Now use fixed effects to estimate the model.
d) Compare your findings from (b) and (c). Are there any notable sign or magnitude changes in the coefficients? What about statistical significance? Are the signs of the estimated coefficients on the criminal justice measures consistent with your expectations?
e) Add the logs of each wage variable in the data set (lwcon lwtuc lwtrd lwfir lwser lwmfg lwsta lwloc) and estimate the model by fixed effects. What are we trying to control for with these wage variables? How does including these variables affect the coefficients on the criminal justice variables in part (b)?
f) Do the wage variables in part (e) all have the expected sign? Explain. Are they jointly significant?
2023-06-17
Natural Experiments (and IV) + Panel Data Models