Smith School of Business
MGMT 880/990
Winter 2018
Professor Olena Ivus
Midterm Exam
You have 80 minutes to solve 5 questions, which are worth of a total of 100 points.
Good luck!
1. (15 points)
Please state if the following statement is true or false and explain: “Estimation of
a linear probability model (LPM) is more robust than probit or logit because the
LPM does not assume homoskedasticity or a distributional assumption.”
2. (30 points)
Please state if the following statement is true or false and explain: “More controls
is always better to include in a regression.” In your answer, discuss the following
concepts: the omitted variable bias, confounding variables, bad controls.
3. (25 points)
Consider the Stata output in Appendix A. In the first regression output, the fol-
lowing state-level equation is estimated for 1987 and 1990:
lowbrthit =θ1+θ2d90 + β1af dcprcit +β2log(phypcit ) + β3log(bedspcit)+
β4log(pcincit) + β5log(populit ) + ci+uit
where idenotes a state and tdenotes time; lowbrth is the percentage of births that
are classified as low birth weight; afdcprc is the percentage of the population in
the welfare program, Aid to Families with Dependent Children (AFDC); d90 is the
indicator variable for the year 1990. The other variables, which act as controls for
quality of health care and income levels, are physicians per capita, hospital beds
per capita, per capita income, and population. In the second regression output, one
additional regressor, afdcprc2
it, is added.
(a) Interpret the magnitude and statistical significance of the highlighted three
coefficients in the first regression output.
(b) Interpret the magnitude of the highlighted two coefficients in the second re-
gression output. What is the estimated turning point in the quadratic?
1
