Study Guides (400,000)
US (230,000)
OSU (2,000)
ECON (100)
Final

# ECON 5420 Study Guide - Final Guide: Instrumental Variable, Econometrics, Statistical InferenceExam

Department
Economics
Course Code
ECON 5420
Professor
Ryan Ruddy
Study Guide
Final

This preview shows pages 1-2. to view the full 8 pages of the document.
Exam 2, Version 1 Econ 5420, Econometrics II, Spring 2016
Section: Name:
1 Multiple Choice
Choose the answer that best answers the question. Each question is worth 3 points.
1. If I estimate the equation W ageit =β0+β1xit +αf+it, where αfis a family
ﬁxed eﬀect, what x variable cannot be included.
(a) mother’s education
(b) education
(c) age
(d) None of the above
2. What problem can be caused by a weak ﬁrst stage in an IV regression?
(a) The exclusion restriction is violated.
(b) The order condition is violated.
(c) Statistical inference based on t-stats is misleading.
(d) The IV estimates are biased.
3. Suppose we estimate the DD equation yit =β0+β1T reati+β2Aftert+β3T reat
Afterit +it. Which represents ¯yfor the treated group before treatment?
(a) β3
(b) β2+β3
(c) β1+β2+β3
(d) β0+β1+ +β2+β3
4. Which best describes the order condition for instrumental variables?
(a) We must have at least as many instruments as endogenous variables.
(b) Instruments must not be correlated with the structural error.
(c) There must be a strong ﬁrst stage.
(d) Endgoenous variables must not be correlated with the error term.

Only pages 1-2 are available for preview. Some parts have been intentionally blurred.

5. Which scenarios(s) apply if the parallel trend assumption is violated in a diﬀerence
in diﬀerences estimation?
(a) The DD estimator is still unbiased.
(b) We could ﬁnd that treatment has zero eﬀect when there actually is an eﬀect.
(c) We could ﬁnd that treatment has an eﬀect when there actually is not an
eﬀect.
(d) B & C.
6. Which of the following types of variables cannot be included in a ﬁxed eﬀects
model?
(a) Discrete dependent variable
(b) Time-varying independent variable
(c) Time-constant independent variable
(d) Dummy variables of any type
7. Let year2011 be a dummy variable equal to one in 2011. Suppose we have data
from 2011 and 2012. If we estimate the equation wageit =β0+β1year2011t+
β2educationit +β3year2011 educationit +eit, what does ˆ
β3tell us?
(a) The returns to eduction in 2011
(b) The diﬀerence between returns to education in 2011 and 2012
(c) The diﬀerence in wages in 2011 relative to 2012
(d) The returns to education
8. Which of the following statements is true about probits:
(a) ˆ
βis the increase in p(y= 1) for a marginal increase in x
(b) Predicted probabilities can be negative.
(c) ˆ
βis estimated through OLS.
(d) The error term is assumed to follow a normal distribution.
9. Suppose we have data on wage and education where wage has been censored at
\$30 per hour and education positively aﬀects wages. If ˆ
βrepresents the estimated
returns to education with OLS and ˆαestimates the estimated returns of education
with a censored regression model, which is true?
(a) ˆ
β= ˆα
(b) ˆ
β < ˆα
(c) ˆ
β > ˆα

Unlock to view full version

Only pages 1-2 are available for preview. Some parts have been intentionally blurred. 10. Which of the following assumptions is required to obtain a ﬁrst-diﬀerenced esti-
mator in a two-period panel data analysis?
(a) The explanatory variable does not change over time for any individuals.
(b) The explanatory variable changes by the same amount in each time period.
(c) The variance of the error term in the regression model is not constant.
(d) The explanatory variable changes over time for some individuals.
11. Suppose I want to estimate the equation yi=β0+β1x1i+β2x2i+β3x3ii, but
x1iis endogenous. What is required for a consistent IV estimation using zias an
instrument?
(a) Cov(i, zi) = 0
(b) Cov(i, x2i)=0
(c) Cov(x1i, zi)6= 0
(d) All of the above
12. Which is true of a balanced panel?
(a) Each individual is surveyed every year .
(b) Each year a new set of individuals are surveyed.
(c) There are the same number of time periods as individuals.
(d) All of the above.
13. The model is designed to model corner solution dependent vari-
ables.
(a) linear probability
(b) logit
(c) probit
(d) Tobit
14. Bias caused by individuals leaving a panel.
(a) Attrition bias
(b) Omitted Variable bias
(c) Misspeciﬁcation bias
(d) None of the above