SVKM'S NMIMS

SARLA ANIL MODI SCHOOL OF ECONOMICS

Academic Year: 2018-.2019

Program: B.Sc Economics Year: Ill Semester: VI

Subject: Advanced Econometrics Batch: 2016-19
Date: loth April, 2019 Time: 01.00 pin -03.00 pin (2 hrs)
Marks: 50 No. of pages:rs
Final Examination

Gen;Hal insfroctions :

1. Answer any five out of six questions. Each question is of 10 marks.

2. Figures in parenthesis indicate' the marks for each sub-pat.
3. Answer all the sub-pats of a question together.
4.. Create only.one R file for the data related questions.

(i) Using simulated data show that Augr}ented Dickey-Fuller test has low power and
we may not be able to reject the null of unit roots in an otherwise stationary AR(1)
model with a coefficient close to unity. (5)

(ii) consider. the NelflQ ¢at?set .Soxp tserie§ ... package which.. contains the
maeroeconomic variab.1es .cousidere¢ by Nelson and Plosser (1982). Consider cpi
.. (columnl), gnp (column 3), ve.I (velocity.of money: column 4) and money.stock

(column 9) variables. Can we say that there exists a long run relation among the
variables based 'on the regression of log(gnp) on log(money.stock), log(vel), and
log(cpi).? Justify ysur answer... (5)

Consider that growl in money. supply follows ARMA(1,1) .

Eil
mjj7rowffa€ = o.8quLj7row€fef_1 + tit + o.8ttt_1; ttt~wIV(o,1)

(i) Solie the underlying difference equation and comment whether the process is
stationary or not. (3)

(ii) Obtain the mean, variance and ACE using 'Yule Wancer equa:tious. Find the
autocorrelation for lagsl -3. (7)

(i) Using the panel data PSID7682 from AER.papkage and answer the foHowing:
(a) Create a subset of data consisting of all the variables for the years 1976 and
1982. (You may use rbind to combine. the data together). (1)

Page 1 Of 3
 a) Now define a variable efrfe which takes the value one for Ofam (Affican
American) and zero otherwise. Also, define j;82 demmy which takes the
value one for year 1982 and zero otherwise. (2)

(c) Assume that in 1979 the Government reduced the minimum wages.. You
intend to analyze whether the reduction in minimum wages was associated
with fall in wages for Affican Americans using difference-in-difference
method. Regress wfl!ge ap j/82,. effe and `interaction of]/8? and effe .and intepret
the results. Did 'the policy have an adverse effect on wages of Affican
Americans? What could be one qualitative` reason for the result? .(4)

(ii) Suppose logarithm of wages and logarithm of price levels both follow AR(1). Wall
` real wages also follow an AR(1) process? Clearly show your workings. (3)

Using the panel data PSD7682 from AER pack.age y.ou wish to analyze whether
log(wage)dapendsaponexpedenc6andsquareofbapedence.
(i) Estimate the pooling and within estimator and test whether you additionally need
to account for time invariarit. unobserved effects using F-test.. C.16aldy mention the
null and the alternative hypotheses. (3)

(ii) htapret the model that you cfroose based on yoin results in (i) above. (2)
(iii)Suppose we wish to study the relation between ethnicity and log(wage) @o not
estimate the model). Show why FE model can't be used for this purpose. Also,

propose a model that will give us the relation between ethnicity and log(wage)
.. after controlling for tine invarian:t qnobservable factors. Mention the proposed
m6dsl' s RSs degrees of freedom (5)

5.

(i) Consider the followifig ARCH(3) ortyut based on German DAX retuns:
Estimate Standard EgroF
•0.00007 I
Alpha_0 (Constant) 0.000003

Alpha_1 0..0492 0.02

Alpha_2 0.0735 0.02

Page 2 Of 3
 Alpha_3 0.1519 0.03

(a) Inteapret the regression output. (2)

a) What is the estimated conditional volathity for t+1 and t+2 if the shock for period
t, t-1 and t-2 were -1,+1 and -2 respectively. (2)

(c) What is the estinated unconditional volatility in returns? (1)

(d) Explain how we can visually test whether ARCH(3) is a good approxination for
the data using the ACE plot. (2)

(ii) Show that h-period ahead forecasts of AR(1) model tends to its uncoflditional mean as
h increases. (3)

iiii 6.
\

(i) Show that in SLRM the IV method will yield consistent estimator when the
explanatory variable is endogenous. (4)

(fi) The MASchools dataset ip AER package contains data on test performance, school
characteristics and student demographic backgrounds for school districts in
Massachusetts. The Score8 variable gives the 8th grade score in Maths, English and
Science, £frofro gives student-teacher ratio and 7.7!coREe gives per capita income. Using
the data answer the following:

(a) You wish to regress /ogrscore8/ on /og/z.z!canleej and sgiv"fro (Do not estimate the
model). Give one reason why g."coffle variable can be endog`enous. (2)

a) Now consider J#z!ch variable defined as the percent of students receiving a

subsidized lunch in the school as an IV for income variable. Estimate the linear
TegFessirca Of log(score8) on log(iveome) and stratto ustun8 lunch as an TV.
?

Inteapret the results. (2)

(c) Test whether i.7Scoffle is endogenous or not. Explain the null and the alternative
hypotheses. (2)

•--------------\ -.---..-------

Page 3 of 3