Faculty of Social Studies and Humanties

Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 16

UNIVERSITY OF MAURITIUS

FACULTY OF SOCIAL STUDIES AND HUMANTIES

SECOND SEMESTER EXAMINATIONS

May 2014

PROGRAMME BSc. Economics and Finance Level II/ BSc Economics with
International Relations Level II/ BSc Economics with
Management II

MODULE NAME Principles of Econometrics

DATE (to be filled by MODULE CODE ECON 2003Y


Exam Section)

TIME (to be filled by DURATION 3 hours


Exam Section)

NO. OF 7 NO. OF QUESTIONS TO 4


QUESTIONS SET BE ATTEMPTED

INSTRUCTIONS TO CANDIDATES

1. This exams paper consists of SEVEN questions and THREE sections


2. Candidates should answer THREE questions.
3. Section A is COMPULSORY
4. Answer at least ONE question from Section B
5. Answer at least ONE question from Section C
6. Candidates are strongly advised to divide their time accordingly.
SECTION A

COMPULSORY

Question One

(a) A researcher estimates the following two econometrics models:

Where the error terms ut and vt are normally distributed and x3t is an
irrelevant variable which does not enter into the data generating process
for Yt. Will the value of (a) R2, (b) Adjusted R2, be higher for the second
model than the first? Explain your answers
[7 marks]
Answer

The value of R2 will almost always be higher for the second model since it has
another variable added to the regression. The value of R2 would only be identical
for the two models in the very, very unlikely event that the estimated coefficient
on the x4t variable was exactly zero. Otherwise, the R2 must be higher for the
second model than the first.

The value of the adjusted R2 could fall as we add another variable. The reason for
this is that the adjusted version of R2 has a correction for the loss of degrees of
freedom associated with adding another regressor into a regression. This implies
a penalty term, so that the value of the adjusted R2 will only rise if the increase in
this penalty is more than outweighed by the rise in the value of R2.

(b) Consider the following Cobb-Douglas function:

Suppose you suspect that there is presence of constant returns to scale.


Describe a test that can be used to test for the presence of constant returns.
[10 marks]
Answer
Applying log

(1)
Now if there are constant returns to scale, economic theory will suggest b1+
b2 =1 or b1= 1 – b2 or b2=1 – b1

(2)

Equation (1) is called the unrestricted regression (UR) and (2) the restricted
regression (R). To compare the unrestricted regression and restricted
regression, we apply the F – test.

The F test is denoted

m denotes number of linear restrictions;


k denotes the parameters in the unrestricted regression and
n is the number of observations.

(b) Two researchers, using the same set of data but working independently,
arrive at different lag lengths for the VAR equation. Describe and evaluate
two methods for determining which of the lag lengths is more
appropriate.
[8 marks]

Answer
The two methods that we have examined are model restrictions and information
criteria.. Briefly, the model restrictions approach involves starting with the larger
of the two models and testing whether it can be restricted down to the smaller
one using the likelihood ratio test based on the determinants of the variance-
covariance matrices of residuals in each case. The alternative approach would be
to examine the value of various information criteria and to select the model that
minimises the criteria. Since there are only two models to compare, either
technique could be used. The restriction approach assumes normality for the
VAR error terms, while use of the information criteria does not. On the other
hand, the information criteria can lead to quite different answers depending on
which criterion is used and the severity of its penalty term. A completely
different approach would be to put the VARs in the situation that they were
intended for (e.g. forecasting, making trading profits, determining a hedge ratio
etc.), and see which one does best in practice.
SECTION B
(Answer at least one question)
Question Two

Consider the following regression model

where N(0, )

(a) Show that


[15 marks]
(b) Consider the following regression model

Derive the OLS estimator of .


[10 marls]

Total: 25 marks
Question Three

(a) What is heteroscedasticity? (5 marks)

Answer
One of the assumptions of the classical linear regression model is that the
disturbance term ut in the population model has a constant variance 2, for all t.
When the disturbance term has a variance which differs from observation to
observation, so that Var (ut) = 2t, the t indicates that the variance is not constant,
we say that there is heteroscedasticity. If the variance of the disturbance term is
constant, we say that there is homoscedasticity.

(b) Describe the Goldfeld – Quandt testing procedure for heteroscedasticity.


(10 marks)

Answer

Goldfeld-Quandt test assumes that the observation can be divided into two
groups (the first group with large variances and the second with small
variances), then checks whether or not disturbance variances of the groups are
different systematically. So, we have to identify a variable to be used to separate
data. F distribution requires that disturbance variances are normally distributed.

For simplicity consider


It is assumed that

This test involves the following procedures:

First, sort the observation based on Xi (beginning with the lowest value).
Second, separate the observations into two groups so that the first group has a
bigger variance. The central value is c. Divide the (n-c) observations into two
groups each of (n-c)/2 observations.

Third, run OLS separately to RRS1 and RSS2 with each degree of
freedom.

Fourth, compute the ratio

If  is greater than the critical value of F at the chosen level of significance, we


say that heteroscedasticity is very likely to be present.

Note that: a number of observations in the middle of the sample may be omitted
in order to increase the power of the test by highlighting the difference. But no
more than a third of the observations should be dropped (Harvey and Phillips
1974), since smaller degree of freedom may diminish the power of the test.

(c) Consider a two-variable model

Given , prove that var (10 marks)

Answer

where
Thus,

var
= since =

From above

This implies

var cross product terms)

since the expectations of the cross product terms are zero

var

since

var =

= since

Question Four

(a) Explain how dummy variable can be used to test for a ‘structural change’

[12 marks]

Answer

Consider the following regression model.


(1)

The regression is estimated for n number of observation (1970 – 2000).

Suppose that there is a structural change at a particular date. We can divide the
sample into two where n = n1 + n2

with n1 number of observation (1970 – 1985) (2)

with n2 number of observation (1986 – 2000) (3)

Referring to the above regressions, we have four possibilities:

1. Both the intercept and slope coefficients in (2) and (3) are the same (No
structural change)
2. Only the intercepts in the two regressions, (2) and (3), are different but the
slope is different. This is the case of parallel regressions.
3. The intercepts in the two regressions, (2) and (3) are the same, but the
slopes are different. This is the case of concurrent regressions.
4. Both the intercepts and slopes in the two regressions (2) and (3) are
different. This is the case of dissimilar regressions.

Using dummy variables we can test whether there has been a ‘change in
structure’. The source of difference, if any, can be pinned down by pooling all
the observations (n = 30) and running one regression.

(4)

where D = 1 for observation 1970 – 1985


= 0 otherwise (for observation 1986 – 2000)

Assuming that E (ut) = 0

The mean function for the period 1970 – 1985 is:

The mean function for the period 1986 – 2000 is:


These are the same as functions as (2) and (3), with , , and
. Therefore estimating (4) is the same as estimating the two
regressions, (2) and (3).

In (4), is the differential intercept and is the differential slope coefficient,


indicating how much the slope coefficient of (2) differs from that of (1). The
standard t-test is used for either a change in intercept or a change in slope. The
test on both simultaneously can be performed by using the F-test.

(b) Give details of an alternative test for a ‘structural change’.


[10 marks]

Answers

The alternative test is known as the Chow test. Consider equation (1), (2) and (3)
in section (b).

The chow test assume the following

1.  N (0, 2) and  N (0, 2)


2. The 2 error terms are independently distributed

The mechanics of the Chow test are as follows:

1. Estimate regression (1) and obtain RSS1 with degrees of freedom, df = n1 +


n2- k (where k is the parameter estimated). RSS1 is also known as the
Restricted Residual Sum of Squares (RSSR). The restrictions are and
= .
2. Estimate regression (2) and obtain RSS2 with degrees of freedom, df = n1 -
k (where k is the parameter estimated).
3. Estimate regression (3) and obtain RSS3 with degrees of freedom, df = n2 -
k (where k is the parameter estimated).
4. Since the 2 sets of examples are deemed independent, we can add RSS1
and RSS2 to obtain the Unrestricted RSS (RSSUR). Therefore, RSSUR =RSS2 +
RSS3.
5. The Chow test assume that there is no structural change (regression (1)
and (2) should be identical). In this case RSSR = RSSUR. The F-statistics is
computed as:


6. We do not reject the null hypothesis of structural change if the computed
F-value does not exceed the critical F-value at the chosen level of
significance.

(c) What other uses, apart from a test of change in structure or seasonal
correction, might an econometrician employ dummy variables?
[3 marks]

Other uses for dummy variables are: (i) the inclusion of qualitative information
(such as a change in government or the effects of currency devaluation at a
certain point in time); (ii) in cross-section survey data we often have information
on such things as educational qualifications (namely whether the individual has
a degree or some sort of school leaving certificate, etc. ) This information can be
introduced by the use of a dummy variable.

SECTION C
(Answer at least one question)
.
Question Five

(a) Why is it important to test for non-stationarity in time series data before
attempting to build an empirical model?
(5 marks)
Answer
Non-stationarity can be an important determinant of the properties of a series.
Also, if two series are non-stationary, we may experience the problem of
“spurious” regression. This occurs when we regress one non-stationary variable
on a completely unrelated non-stationary variable, but yield a reasonably high
value of R2, apparently indicating that the model fits well.

Most importantly therefore, we are not able to perform any hypothesis tests in
models which inappropriately use non-stationary data since the test statistics
will no longer follow the distributions which we assumed they would (e.g. a t or
F), so any inferences we make are likely to be invalid.

(b) Given the following model (random walk with drift), prove that it is non –
stationary.
(10 marks)
Answer

A random walk with drift evolves according to the


where d is the drift parameter
The model can be written as

Mean

Now =0 since u is white noise €

Therefore

Variance

substituting in for Yt and E (Yt)

var (Yt) =
Therefore var (Yt) =
This demonstrates that in the case of a simple random walk, the variance of Y is
not a constant and is not independent on time. It is in fact a linear and increasing
function of time. So we conclude that the random walk with drift is not
stationary.

(c) Define the following terms and describe the processes that they represent:

(i) Pure Random Walk


(ii) Deterministic Trend
(iii) Random Walk with Drift and Deterministic Trend
(iv) Deterministic Trend with Stationary AR (1) component
(10 marks)
Answer

If the trend in a time series is completely predictable and not variable, we call it a
deterministic trend, whereas if it is not predictable, we call it a stochastic trend.
To make the definition more formal, consider the following model of the time
series Yt.

We have the following possibilities

Pure Random Walk:

The model is non-stationary but is stationary in first difference (a difference


stochastic process)

Random Walk with Drift:

which can be written as , this means that Yt will exhibit a positive


trend ( or ). Such a trend is called a stochastic trend. It is a difference
stochastic process.
Deterministic Trend:

which is a trend stationary process (TSP). The mean is not a constant but the
variance is. Once the values of and are known, the mean can be forecast
perfectly. Therefore, if we subtract the mean of Yt from Yt, the resulting series
will be stationary, hence the name trend stationary. The procedure of removing
the (deterministic) trend is called detrending.

Random Walk with Drift and Deterministic Trend:

which is of course not stationary.

Deterministic Trend with Stationary AR (1) component:

which is stationary around the deterministic trend

Question Six

(a) Clearly illustrate the properties of an integrated time series?


(7 marks)
Answer
Let xt ,yt , zt be three time series. Then, the following properties of integrated time
series can be noted.

1. If xt ~I (0) and yt ~ I (1), then zt = (xt + yt ) = I (1); that is a linear


combination or sum of stationary and nonstationary time series is
nonstationary.
2. If xt ~I (d), then zt = (a + bxt) = I (d), where a and b are constants. That is, a
linear combination of an I (d) series is also I (d). Thus, if xt ~I (0), then Zt =
(a + bxt) = I (0).

Thus, one have to pay careful attention in combining two or more times series
that are integrated of different order. Consider the following model

Suppose Yt is I (0), but xt is I (1); that is the former is stationary and the latter is
nonstationary. Since xt is nonstationary, its variance will increase indefinitely,
thus he estimator will converge to zero asymptotically and it will not even
have an asymptotic distribution.
Granger points out some essential differences between I (0) and I (1) series:

 An I (0) series has a constant mean and the series tends to return to the
mean provided that it fluctuates around the mean value. Autocorrelation
declines rapidly as lag increases and the process gives low weights to
events in the medium to distant past and thus has a finite memory.

 An I (1) series without drift will be relatively smooth, wander widely and
rarely return to an earlier value. Autocorrelations are near one in
magnitude even for large samples and the process has an indefinitely long
memory.

(b) Consider a series of values for the spot and future prices of a given
commodity. In the context of these series, explain the concept of
cointegration.
(9 marks)
Answer
If two or more series are cointegrated, in intuitive terms this implies that they
have a long run equilibrium relationship that they may deviate from in the short
run, but which will always be returned to in the long run. In the context of spot
and futures prices, the fact that these are essentially prices of the same asset but
with different delivery and payment dates, means that financial theory would
suggest that they should be cointegrated. If they were not cointegrated, this
would imply that the series did not contain a common stochastic trend and that
they could therefore wander apart without bound even in the long run. If the
spot and futures prices for a given asset did separate from one another, market
forces would work to bring them back to follow their long run relationship given
by the cost of carry formula.

(c) Explain, with reference to the implication of cointegration, why non –


cointegration between series might be expected.
(9 marks)
Answer
There are many examples that one could draw from financial or economic theory
of situations where cointegration would be expected to be present and where its
absence could imply a permanent disequilibrium. It is usually the presence of
market forces and investors continually looking for arbitrage opportunities that
would lead us to expect cointegration to exist. Good illustrations include equity
prices and dividends, or price levels in a set of countries and the exchange rates
between them. The latter is embodied in the purchasing power parity (PPP)
theory, which suggests that a representative basket of goods and services should,
when converted into a common currency, cost the same wherever in the world it
is purchased. In the context of PPP, one may expect cointegration since again, its
absence would imply that relative prices and the exchange rate could wander
apart without bound in the long run. This would imply that the general price of
goods and services in one country could get permanently out of line with those,
when converted into a common currency, of other countries. This would not be
expected to happen since people would spot a profitable opportunity to buy the
goods in one country where they were cheaper and to sell them in the country
where they were more expensive until the prices were forced back into line.
There is some evidence against PPP, however, and one explanation is that
transactions costs including transportation costs, currency conversion costs,
differential tax rates and restrictions on imports, stop full adjustment from taking
place. Services are also much less portable than goods and everybody knows that
everything costs twice as much in the UK as anywhere else in the world.

Question Seven

You are interested in analyzing the relationship between real GDP and human capital in
Mauritius. The following equation is specified in a VAR setting.

Where GDPPC denotes GDP per capita; FDI denotes FDI as a percentage
of GDP; M2 denotes M2 as a percentage of GDP; OPEN denotes openness
and u is the error term. Note that all variables are in form of log.

(a) The following table shows ADF results of certain variables from STATA.
By analysing the table, explain which variables are stationary and which
ones are not.

Critical Value
Variable t-statistics 1% 5% 10%
LPDINV -3.811 -3.71 -2.983 -2.623
LOPEN -1.295 -3.71 -2.983 -2.623
LMIGDP -2.745 -3.71 -2.983 -2.623
LHUMAN -0.695 -3.71 -2.983 -2.623
LINF -3.445 -3.71 -2.983 -2.623
[5 marks]

(b) To check for the number of lags, you have invoked the VARSOC command in
STATA which yield the following result:
Lag LL LR df p-value AIC HQIC SBIC
0 126.025 -7.33484 -7.25855 -7.1081
1 275.527 293 25 0.0000 -14.6986 -14.2409 -13.3381
2 319.791 94.526 25 0.0004 -16.0479 -15.2087 -13.5537
3 345.307 51.023 25 0.0000 -16.0792 -14.8585 -12.4513
4 420.55 150.49 25 0.0000 -19.1243 -17.5221 -14.3626

What is the most appropriate lag number to be selected? Illustrate your answer.
[5 marks]

(b) You were interested in the causality among variables. You got the following
results after invoking the VARGRANGER command in STATA:

Equation Excluded Chi2 df Prob>chi2


LRDGP LFDI 2.2268 1 0.136
LRDGP LHUMAN 14.844 1 0.000
LRDGP LM2GDP 0.3583 1 0.009
LRDGP LOPEN 0.3247 1 0.569
LRDGP ALL 48.07 4 0.000
LFDI LRGDP 12.836 1 0.000
LFDI LHUMAN 20.533 1 0.000
LFDI LM2GDP 0.1474 1 0.701
LFDI LOPEN 3.3871 1 0.070
LFDI ALL 47.42 4 0.000
LHUMAN LRGDP 2.0149 1 0.006
LHUMAN LFDI 3.9037 1 0.048
LHUMAN LM2GDP 4.5486 1 0.033
LHUMAN LOPEN 19.247 1 0.000
LHUMAN ALL 63.59 4 0.000
LM2GDP LRGDP 13.414 1 0.000
LM2GDP LFDI 8.6085 1 0.003
LM2GDP LHUMAN 2.0224 1 0.155
LM2GDP LOPEN 0.8591 1 0.354
LM2GDP ALL 39.604 4 0.000
LOPEN LRGDP 26.931 1 0.000
LOPEN LFDI 4.4927 1 0.034
LOPEN LHUMAN 49.245 1 0.000
LOPEN LM2GDP 0.1858 1 0.666
LOPEN ALL 94.852 4 0.000

Clearly interpret the results from the table


[15 marks]
END OF EXAMS PAPER

You might also like