Practice questions econometrics II

Chapter 1: regression with qualitative variables (3 questions in exit exam)

1. Which of the following variables are qualitative? A)Age B)Educational level C)Participation in equp D)CGPA
E)Sales
2. Bereket wants to investigate the effect of transport mode and transport costs on transportation time. There are
four modes of transport: car, bajaj, taxi or by foot. Transport costs are expressed in birr. How many dummy
variables should be included in the regression? ______________
3. When estimating the effect of departments (ecs, acc, mngt, tourism) and gender on starting salary, Adanech
estimated the following model:
𝑆𝑖 = 2304 + 786𝑒𝑐𝑠 − 324𝑚𝑛𝑔𝑡 + 20𝑡𝑜𝑢𝑟𝑖𝑠𝑚 + 900𝑚𝑎𝑙𝑒 + 𝑒
t=2.5** t=6.3*** t=1.3* t=0.05 t=9.8***
Which answers are correct?
A) It is not possible to predict the starting salary of accounting graduates based on this model.
B) The highest average starting salary is for economics students, after controlling for gender.
C) Management students get a significantly lower starting salary than tourism students at 5%, after
controlling for gender.
D) Male graduates are getting a significantly higher salary than female graduates, after controlling for
department.
4. Based on Adanech’ model estimation, fill in the blanks:
The predicted value of starting salary for female accounting students is ______
The predicted value of starting salary for female economics students is _______
The predicted value of starting salary for male tourism students is _________
5. Adanech also runs a Linear Probability Model that predicts the probability of finding work within three
months after graduation.
𝑃(𝑌𝑖 = 1) = 0.3 + −0.02𝑒𝑐𝑠 + 0.04𝑚𝑛𝑔𝑡 + −0.2𝑡𝑜𝑢𝑟𝑖𝑠𝑚 + 0.1𝑚𝑎𝑙𝑒 + 𝑒
t=3.5*** t=0.04 t=1.4* t=1.05 t=4.8***
What is the expected probability of a female accounting student to find work within three months after
graduation?__________________
What is the expected probability of a male management student to find work within three months after
graduation?___________________
6. Which of the following answers is not a limitation of the Linear Probability Model (LPM)?
A) Predicted probabilities may be >1
B) Predicted probabilities may be <0
C) The residual term is related to X
D) Heteroscedasticity
E) A change in X has a constant effect on P(Y=1)
F) All the answers are limitations of the LPM
7. Why does the LPM suffer from heteroscedasticity?
A) Because the residual is either 1 − 𝑌̂ or 0 − 𝑌̂
B) Because the residual is either 1 − 𝛽𝑋 or 0 − 𝛽𝑋
C) Because the residual is a function of X
D) A, B and C are all correct
8. Identify the incorrect statement
A) Probit regression is based on the cumulative normal probability distribution function
B) Logit regression is based on the logistic cumulative probability distribution function
C) Z can exceed 1, both in probit and in logit regression
D) The predicted P(Y=1) can exceed 1, both in probit and in logit regression
E) All statements are correct
9. Hiwot runs a probit model and she finds that the LR test outcome is chi2=5.45, p=0.0000. What can she
conclude?
A) All regressors are good predictors for the regressand
B) At least one regressor is a good predictor for the regressand
C) The model is significant
D) The model is insignificant
E) Both A, B and C
F) Both B and C
 10. In Hiwot’s probit model the Pseudo R2 is 0.35. What does that mean?
A) The log likelihood improved by 35% after introducing the regressors
B) 35% of the variation in Y is explained by the model
C) The model is useless
D) The model is extremely useful
11. Hiwot’s probit model’s dependent variable is car ownership. Independent variables are gender
(1=male,0=female) and income. Their probit coefficients are as follows:
income =0.0004***
gender=0.911
Which answer(s) are/is correct?
A) Income has a positive effect on the probability to have a car
B) If income increases by 1 birr, the probability to have a car increases by 0.004%
C) Gender has no effect on the probability to have a car
D) Males are 91.1% more likely to have a car than females
12. Hiwot also runs the logit model with car ownership as dependent variable, and income and gender as
independent variable. The odds ratios are as follows:
income=1.0007
gender=6.03
which answer(s) are/is correct?
A) If income increases by 1 birr, the probability of owning a car increases by 0.07%
B) If income increases by 1 birr, the person is 1.0007 times more likely to own a car
C) If gender increases by 1, the person is 6.03 times more likely to own a car
D) Men are 6.03 times more likely to own a car than women
E) Women are 6.03 times more likely to own a car than men
F) Men are 6.03% more likely to own a car than women
G) Income has a positive effect on owning a car
13. Lastly, Hiwot obtains the marginal effects, which are:
income=1.99e-09
gender=4.73e-06
which answer(s) are/is correct?
A) If income increases by 1 birr, an average respondent is 1.99e-07% more likely to own a car
B) If gender increases by 1, an average respondent is 4.37e-04% more likely to own a car
C) Compared to average females, average males are 4.73e-04% more likely to own a car

About chapter 2: Structural Equation Modeling (exam only 1 question, comprehension)

Question 1: In this system of equations

1. What are the endogenous variables, what are the exogenous variables?
2. Which equations are identified? Which ones are not identified? Use the order condition to determine.
3. Explain how simultaneity bias can be a problem in this system of equations.
4. Write the model in reduced form equations
Question 2: for the following titles, identify if simultaneity bias can be present.
A. Predict the effect of educational level on family income level. YES / NO
B. Predict the effect of inflation on unemployment YES / NO
C. Predict the effect of gender on wages YES / NO
D. Predict the effect of income on car ownership YES / NO
E. Predict the effect of the price of juice on the demand for juice YES / NO
Question 3: use the order condition
 Are these equations identified?
 Demand Function: 𝑄𝑡𝑑 = 𝛼0 + 𝛼1 𝑃𝑡 + 𝑈1𝑡 ; 𝛼1 < 0 (7)
 Supply Function: 𝑄𝑡𝑠 = 𝛽0 + 𝛽1 𝑃𝑡 + 𝑈2𝑡 ; 𝛽1 > 0 (8)
 Are these equations identified?
 Demand Function: 𝑄𝑡𝑑 = 𝛼0 + 𝛼1 𝑃𝑡 + 𝛼2 𝑌𝑡 + 𝑈1𝑡 (9)
 Supply Function: 𝑄𝑡𝑠 = 𝛽0 + 𝛽1 𝑃𝑡 + 𝑈2𝑡 (10)
Question 4: Is the first equation identified, according to the rank condition?
 Demand Function: 𝑄𝑡𝑑 = 𝛼0 + 𝛼1 𝑃𝑡 + 𝛼2 𝑌𝑡 + 𝛼3 𝑃𝑆𝑡 + 𝑈1𝑡 (11)
 Supply Function: 𝑄𝑡𝑠 = 𝛽0 + 𝛽1 𝑃𝑡 + 𝛽2 𝑃𝑡−1 + 𝑈2𝑡 (12)
Question 5: Which answer describes best the ILS method?
A. First estimate the structural equation, next use these estimates to get the reduced form equations
B. First estimate the reduced form equation, next use these estimates to get the structural equation
C. Using OLS to estimate a structural equation leads to unbiased parameter estimates
D. OLS must be used in ILS to estimate the reduced form equations
E. Both A and C
F. Both B and D
Question 6: Consider the following system of 2 equations. Are the statements true or false?

A. If we estimate the money supply function as it is, there is simultaneity bias

B. In this case, to estimate the money supply with 2SLS, the first stage would be to run OLS for 𝑌1𝑡 = 𝜋0 +
𝜋1 𝑋1𝑡 + 𝜋2 𝑋2𝑡 + 𝑒1𝑡
C. The structural equation for money supply will be estimated by𝑌2𝑡 = 𝛽0 + 𝛽1 𝑌̂ 1𝑡 + 𝑢2𝑡
D. The instrumental variables in this case are Y1 and Y2
Chapter 3 Time Series (3 questions in exit exam, one psychomotoric)
Question 1: Which of the following datasets is time series data?

Question 2: Which data can be collected as time series data? (multiple answers are possible)
A. Military expenditure of Russia over the past decade
B. Coffee production of farmer Tilahun who lives close to Woleita
C. Number of students present in Main Campus
D. Grades of section A for the intermediate test of Econometrics II 2015 EC

Question 3: Spurious regression (answer the questions)

A. What, do you expect, is the trend in access to internet over the past 20 years in Ethiopia?___
B. What, do you expect, is the trend in import of second hand clothing over the past 20 years in
Ethiopia?_________________________
C. If we regress internet access (regressand) based on second hand clothing import (regressor) for the period 2000-
2020, do you expect a significant or an insignificant model?
Why?_____________________________________________________
D. Does it mean that internet access increases if Ethiopia imports more second hand
clothes?_______________________
Question 4:
Indicate which graphs display stationary time series, and which ones are non-stationary. For non-stationarity, indicate
the reason(s)

Question 5: Which answer is correct about trends in time series?

A. In the case of a deterministic trend, the data does not vary around the trendline.
B. A deterministic trend depends on the lagged value.
 C. Time series data with a deterministic trend becomes stationary if detrended
D. A and B
Question 6: Suppose your time series has unit root. What procedure can you use best to make the series stationary?
A. Detrending C. Demeaning
B. Differencing D. Deseasoning
Question 7: GDP per capita is non-stationary. What should be done to make it surely stationary?
A. Detrending C. Differencing once
B. Detrending until it becomes stationary D. Differencing until it becomes stationary
Question 8: Enrollment rates are I(3), what does that mean?
A. The first, second and third differences of C. Enrollment rates are non-stationary
enrollment rates are non-stationary D. B and C
B. The first and second differences of enrollment
rates are non-stationary
Question 9 (see correlogram): Are expenses stationary? Why
(not)?
Question 10: The Dickey Fuller test’s H0 is in line with all
answers, except
A. Random walk
B. Stationary
C. Stochastic trend
D. Unit root
Question 11: Does this output of the augmented Dickey Fuller test (including trend, constant and lags) state that the
variable lnWiki is stationary or not? __________

Ch 4 panel data modeling (1 question in exit exam)

1. Which of the following data is panel data?

2. For the data in A, assume that profit is the dependent variable, and turnover the independent variable. Write the
mathematical specification of the pooled OLS regression model_______________
3. For the same variables, write the mathematical specification of the FEM.___________________
4. For the same variables, write the mathematical specification of the REM.___________________
5. Which of the following is a disadvantage of the FEM?
a. It controls for individual differences.
b. It leads to a large degree of freedom loss for models with a large N.
c. It allows regressors that are time-invariant.
d. There are no disadvantages of the FEM.
6. Which of the following is an advantage of the REM?
a. It is more efficient than the FEM if the individual error term is not related to the regressors
b. It solves the endogeneity problem
c. It is unable to include individual differences
d. All three are advantages of the REM
Consider the following research case: The dependent variable is FDI inflow, and the explanatory variables are market
size, trade openness, infrastructure, inflation rate, interest rate (IR), research and development (R&D) and human capital
(HC). The cross-sectional units are 18 countries in South- East- and South-East Asia with different levels of
development. The authors have estimated the Fixed Effects and the Random Effects Model.
Independent variables Fixed effects (within) regression Random effects GLS
regression
Constant -7.84 (-3.58) 5.05 (5.96)
Market size 1.63 (7.39)** 0.33 (3.85)**
Trade openness 0.63 (1.17) -1.108 (-2.19)
Infrastructure -0.00012 (-2.22)** 0.000072 (1.48)**
Inflation -0.00055(-3.67)** 0.0044 (4.66)
Interest rate -0.0084 (-3.67)** -0.0096 (-3.91)**
Research and development -1.09e-06 (-1.27) 5.47e-07 (0.64)
Human capital 0.0037 (2.82)** 0.0046 (3.33)
No of countries 18 18
No of observations 304 304
Overall R squared 0.54 0.48
F-statistic: 46.88 (p:0.000) Wald chi-square: 262.88
(p:0.000)
Hausman p-value 0.000
t-statistics are in parentheses. ** significant at the 5% level.
7. Fill in the blanks
a. How many dummy variables are included in the FEM?______________________________
b. Specify the REM model_______________________________________________________
8. Indicate which answers are true:
a. The unobserved characteristics of the 18 countries are related to the regressors.
b. This panel data is balanced
c. The REM is biased because of endogeneity.
d. Market size has a significant positive effect on FDI inflow
e. After controlling for the other variables, human capital does not significantly affect FDI inflow.
f. Inflation has a negative significant effect on FDI inflow.
9. Under what conditions should we use pooled regression for this case?
a. If in the FEM all country intercepts are 0
b. If in the REM all country-specific errors terms are 0
c. If all time-variant variables are stationary
d. All answers are correct
10. In which case do you expect a higher rho?
a. Dependent variable: GDP per capita, entities: China, Eritrea, UK, USA, Ethiopia
b. Dependent variable: Birth growth, entities: SNNPR, Amhara, Oromia, Afar