TOPICS IN TIME SERIES ECONOMETRICS

Phùng Thanh Bình

[email protected]

UNIT ROOT TESTS, COINTEGRATION,

ECM, VECM, AND CAUSALITY MODELS
Compiled by Phung Thanh Binh1

(SG - 30/11/2013)

“EFA is destroying the brains of current generation’s researchers in

this country. Please stop it as much as you can. Thank you.”

The aim of this lecture is to provide you with the key

concepts of time series econometrics. To its end, you are
able to understand time-series based researches,
officially published in international journals2 such as
applied economics, applied econometrics, and the likes.
Moreover, I also expect that some of you will be
interested in time series data analysis, and choose the
related topics for your future thesis. As the time this
lecture is compiled, I believe that the Vietnam time
series data3 is long enough for you to conduct such
studies. This is just a brief summary of the body of
knowledge in the field according to my own understanding.

1
School of Economics, University of Economics, HCMC. Email: [email protected].
2
Selected papers were compiled by Phung Thanh Binh & Vo Duc Hoang Vu (2009). You
can find them at the H library.
3
The most important data sources for these studies can be World Bank’s World
Development Indicators, IMF-IFS, GSO, and Reuters Thomson.

1
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Therefore, it has no scientific value for your citations.

In addition, researches using bivariate models have not
been highly appreciated by international journal’s
editors and my university’s supervisors. As a researcher,
you must be fully responsible for your own choice in this
field of research. My advice is that you should firstly
start with the research problem of your interest, not
with data you have and statistical techniques you know.
At the current time, EFA becomes the most stupid
phenomenon of young researchers that I’ve ever seen in my
university of economics, HCMC. They blindly imitate
others. I don’t want the series of models presented in
this lecture will become the second wave of research that
annoys the future generation of my university. Therefore,
just use it if you really need and understand it.

Some topics such as serial correlation, ARIMA models,

ARCH family models, impulse response, variance
decomposition, structural breaks4, and panel unit root and
cointegration tests are beyond the scope of this lecture.
You can find them elsewhere such as econometrics
textbooks, articles, and my lecture notes in Vietnamese.

The aim of this lecture is to provide you:

 An overview of time series econometrics

 The concept of nonstationary
 The concept of spurious regression

4
My article about threshold cointegration and causality analysis in growth-energy
consumption nexus (www.fde.ueh.edu.vn) did mention about this issue.

2
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

 The unit root tests

 The short-run and long-run relationships
 Autoregressive distributed lag (ARDL) model and error
correction model (ECM)
 Single-equation estimation of the ECM using the
Engle-Granger 2-step method
 Vector autoregressive (VAR) models
 Estimating a system of ECMs using vector error
correction model (VECM)
 Granger causality tests (both cointegrated and non-
cointegrated series)
 Optimal lag length selection criteria
 ARDL and bounds test for cointegration
 Basic practicalities in using Eviews and Stata
 Suggested research topics

1. AN OVERVIEW OF TIME SERIES ECONOMETRICS

In this lecture, we will mainly discuss single equation

estimation techniques in a very different way from what
you have previously learned in the basic econometrics
course. According to Asteriou (2007), there are various
aspects to time series analysis but the most common theme
to them is to fully exploit the dynamic structure in the
data. Saying differently, we will extract as much
information as possible from the past history of the
series. The analysis of time series is usually explored
within two fundamental types, namely, time series
forecasting and dynamic modelling. Pure time series
forecasting, such as ARIMA and ARCH/GARCH family models,

3
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

is often mentioned as univariate analysis. Unlike most

other econometrics, in univariate analysis we do not
concern much with building structural models,
understanding the economy or testing hypothesis, but what
we really concern is developing efficient models, which
are able to forecast well. The efficient forecasting
models can be empirically evaluated using various ways
such as significance of the estimated coefficients
(especially the longest lags in ARIMA), the positive sign
of the coefficients in ARCH, diagnostic checking using
the correlogram, Akaike and Schwarz criteria, and
graphics. In these cases, we try to exploit the dynamic
inter-relationship, which exists over time for any single
variable (say, asset prices, exchange rates, interest
rates, ect). On the other hand, dynamic modelling,
including bivariate and multivariate time series
analysis, is mostly concerned with understanding the
structure of the economy and testing hypothesis. However,
this kind of modelling is based on the view that most
economic series are slow to adjust to any shock and so to
understand the process must fully capture the adjustment
process which may be long and complex (Asteriou, 2007).
The dynamic modelling has become increasingly popular
thanks to the works of two Nobel laureates in economics
2003, namely, Granger (for methods of analyzing economic
time series with common trends, or cointegration) and

4
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Engle (for methods of analyzing economic time series with

time-varying volatility or ARCH)5. Up to now, dynamic
modelling has remarkably contributed to economic policy
formulation in various fields. Generally, the key purpose
of time series analysis is to capture and examine the
dynamics of the data.

In time series econometrics, it is equally important that

the analysts should clearly understand the term
stochastic process. According to Gujarati (2003), “a
random or stochastic process is a collection of random
variables ordered in time”. If we let Y denote a random
variable, and if it is continuous, we denote it a Y(t),
but if it is discrete, we denote it as Yt. Since most
economic data are collected at discrete points in time,
we usually use the notation Yt rather than Y(t). If we let
Y represent GDP, we have Y1, Y2, Y3, …, Y88, where the
subscript 1 denotes the first observation (i.e., GDP for
the first quarter of 1970) and the subscript 88 denotes
the last observation (i.e. GDP for the fourth quarter of
1991). Keep in mind that each of these Y’s is a random
variable.

In what sense we can regard GDP as a stochastic process?

Consider for instance the GDP of $2873 billion for
1970Q1. In theory, the GDP figure for the first quarter
of 1970 could have been any number, depending on the
economic and political climate then prevailing. The

5
http://nobelprize.org/nobel_prizes/economics/laureates/2003/

5
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

figure of $2873 billion is just a particular realization

of all such possibilities. In this case, we can think of
the value of $2873 billion as the mean value of all
possible values of GDP for the first quarter of 1970.
Therefore, we can say that GDP is a stochastic process
and the actual values we observed for the period 1970Q1
to 1991Q4 are a particular realization of that process.
Gujarati (2003) states that “the distinction between the
stochastic process and its realization in time series
data is just like the distinction between population and
sample in cross-sectional data”. Just as we use sample
data to draw inferences about a population; in time
series, we use the realization to draw inferences about
the underlying stochastic process.

The reason why I mention this term before examining

specific models is that all basic assumptions in time
series models relate to the stochastic process
(population). Stock & Watson (2007) say that the
assumption that the future will be like the past is an
important one in time series regression. If the future is
like the past, then the historical relationships can be
used to forecast the future. But if the future differs
fundamentally from the past, then the historical
relationships might not be reliable guides to the future.
Therefore, in the context of time series regression, the
idea that historical relationships can be generalized to
the future is formalized by the concept of stationarity.

6
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

2. STATIONARY STOCHASTIC PROCESSES

2.1 Definition

According to Gujarati (2003), a key concept underlying

stochastic process that has received a great deal of
attention and scrutiny by time series analysts is the so-
called stationary stochastic process. Broadly speaking,
“a time series is said to be stationary if its mean and
variance are constant over time and the value of the
covariance6 between the two periods depends only on the
distance or gap or lag between the two time periods and
not the actual time at which the covariance is computed”
(Gujarati, 2011). In the time series literature, such a
stochastic process is known as a weakly stationary or
covariance stationary. By contrast, a time series is
strictly stationary if all the moments of its probability
distribution and not just the first two (i.e., mean and
variance) are invariant over time. If, however, the
stationary process is normal, the weakly stationary
stochastic process is also strictly stationary, for the
normal stochastic process is fully specified by its two
moments, the mean and the variance. For most practical
situations, the weak type of stationarity often suffices.
According to Asteriou (2007), a time series is weakly
stationary when it has the following characteristics:

6
or the autocorrelation coefficient.

7
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

(a) exhibits mean reversion in that it fluctuates

around a constant long-run mean;

(b) has a finite variance that is time-invariant; and

(c) has a theoretical correlogram that diminishes as

the lag length increases.

In its simplest terms a time series Yt is said to be

weakly stationary (hereafter refer to stationary) if:

(a) Mean: E(Yt) = (constant for all t);

(b) Variance: Var(Yt) = E(Yt- )2 = 2

(constant for all
t); and

(c) Covariance: Cov(Yt,Yt+k) = k = E[(Yt- )(Yt+k- )]

where k, covariance (or autocovariance) at lag k, is the

covariance between the values of Yt and Yt+k, that is,
between two Y values k periods apart. If k = 0, we obtain

0, which is simply the variance of Y (= 2); if k = 1, 1

is the covariance between two adjacent values of Y.

Suppose we shift the origin of Y from Yt to Yt+m (say, from

the first quarter of 1970 to the first quarter of 1975
for our GDP data). Now, if Yt is to be stationary, the
mean, variance, and autocovariance of Yt+m must be the
same as those of Yt. In short, if a time series is
stationary, its mean, variance, and autocovariance (at
various lags) remain the same no matter at what point we
measure them; that is, they are time invariant. According
to Gujarati (2003), such time series will tend to return

8
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

to its mean (called mean reversion) and fluctuations

around this mean (measured by its variance) will have a
broadly constant amplitude.

If a time series is not stationary in the sense just

defined, it is called a nonstationary time series. In
other words, a nonstationary time series will have a
time-varying mean or a time-varying variance or both.

Why are stationary time series so important? According to

Gujarati (2003, 2011), there are at least two reasons.
First, if a time series is nonstationary, we can study
its behavior only for the time period under
consideration. Each set of time series data will
therefore be for a particular episode. As a result, it is
not possible to generalize it to other time periods.
Therefore, for the purpose of forecasting or policy
analysis, such (nonstationary) time series may be of
little practical value. Second, if we have two or more
nonstationary time series, regression analysis involving
such time series may lead to the phenomenon of spurious
or nonsense regression (Gujarati, 2011; Asteriou, 2007).

In addition, a special type of stochastic process (or

time series), namely, a purely random, or white noise,
process, is also popular in time series econometrics.
According to Gujarati (2003), we call a stochastic
process purely random if it has zero mean, constant
2
variance , and is serially uncorrelated. This is similar
to what we call the error term, ut, in the classical

9
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

normal linear regression model, once discussed in the

phenomenon of serial correlation topic. This error term
is often denoted as ut ~ iid(0, 2).

2.2 Random Walk Process

According to Stock and Watson (2007), time series

variables can fail to be stationary in various ways, but
two are especially relevant for regression analysis of
economic time series data: (1) the series can have
persistent, long-run movements, that is, the series can
have trends; and, (2) the population regression can be
unstable over time, that is, the population regression
can have breaks. For the purpose of this lecture, I only
focus on the first type of nonstationarity.

A trend is a persistent long-term movement of a variable

over time. A time series variable fluctuates around its
trend. There are two types of trends often seen in time
series data: deterministic and stochastic. A
deterministic trend is a nonrandom function of time (i.e.
Yt = A + B*Time + ut, Yt = A + B*Time + C*Time2 + ut, and
so on)7. For example, the LEX [the logarithm of the
dollar/euro daily exchange rate, TABLE13-1.wf1, Gujarati
(2011)] is a nonstationary seris (Figure 2.1), and its
detrended series (i.e. residuals from the regression of

7
Yt = a + bT + et => et = Yt – a – bT is called the detrended series. If Yt is
nonstationary, while et is stationary, Yt is known as the trend (stochastic)
stationary (TSP). Here, the process with a deterministic trend is nonstationary but
not a unit root process.

10
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

log(EX) on time: et = log(EX) – a – b*Time) is still

nonstationary (Figure 2.2). This indicates that log(EX)
is not a trend stationary series.

.5

.4

.3

.2

.1

.0

-.1

-.2
500 1000 1500 2000

Figure 2.1: Log of the dollar/euro daily exchange rate.

.3

.2

.1

.0

-.1

-.2
500 1000 1500 2000

Figure 2.2: Residuals from the regression of LEX on time.

11
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

In contrast, a stochastic trend is random and varies over

time. According to Stock and Watson (2007), it is more
appropriate to model economic time series as having
stochastic rather than deterministic trends. Therefore,
our treatment of trends in economic time series focuses
mainly on stochastic rather than deterministic trends,
and when we refer to “trends” in time series data, we
mean stochastic trends unless we explicitly say
otherwise.

The simplest model of a variable with a stochastic trend

is the random walk. There are two types of random walks:
(1) random walk without drift (i.e. no constant or
intercept term) and (2) random walk with drift (i.e. a
constant term is present).

The random walk without drift is defined as follow.

Suppose ut is a white noise error term with mean 0 and
2
variance . The Yt is said to be a random walk if:

Yt = Yt-1 + ut (1)

The basic idea of a random walk is that the value of the

series tomorrow (Yt+1) is its value today (Yt), plus an
unpredictable change (ut+1).

From (1), we can write

Y1 = Y0 + u 1

Y2 = Y1 + u2 = Y0 + u1 + u2

Y3 = Y2 + u3 = Y0 + u1 + u2 + u3

Y4 = Y3 + u4 = Y0 + u1 + … + u4

12
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Yt = Yt-1 + ut = Y0 + u1 + … + ut

In general, if the process started at some time 0 with a

value Y0, we have

Yt Y0 ut (2)

therefore,

E(Yt) E(Y0 ut) Y0

In like fashion, it can be shown that

Var(Yt) E(Y0 ut Y0)2 E( ut)2 t 2

Therefore, the mean of Yt is equal to its initial or

starting value, which is constant, but as t increases,
its variance increases indefinitely, thus violating a
condition of stationarity. In other words, the variance
of Yt depends on t, its distribution depends on t, that
is, it is nonstationary.

Interestingly, if we re-write (1) as

(Yt – Yt-1) = ∆Yt = ut (3)

where ∆Yt is the first difference of Yt. It is easy to

show that, while Yt is nonstationary, its first difference
is stationary. And this is very significant when working
with time series data. This is widely known as the
difference stationary (stochastic) process (DSP).

13
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

-4

-8

-12

-16

-20
50 100 150 200 250 300 350 400 450 500

Figure 2.3: A random walk without drift.

.03

.02

.01

.00

-.01

-.02

-.03
500 1000 1500 2000

Figure 2.4: First difference of LEX.

14
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

The random walk with drift can be defined as follow:

Yt = + Yt-1 + ut (4)

where is known as the drift parameter. The name drift

comes from the fact that if we write the preceding
equation as:

Yt – Yt-1 = ∆Yt = + ut (5)

it shows that Yt drifts upward or downward, depending on

being positive or negative. We can easily show that, the
random walk with drift violates both conditions of
stationarity:

E(Yt) = Y0 + t.
2
Var(Yt) = t

In other words, both mean and variance of Yt depends on t,

its distribution depends on t, that is, it is
nonstationary.

Stock and Watson (2007) say that because the variance of

a random walk increases without bound, its population
autocorrelations are not defined (the first
autocovariance and variance are infinite and the ratio of
the two is not well defined)8.

Cov(Yt, Yt 1)
8
Corr(Yt,Yt-1) = ~
Var(Yt)Var(Yt 1)

15
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

30

25

20

15

10

-5

-10
50 100 150 200 250 300 350 400 450 500

Figure 2.5: A random walk with drift (Yt = 2 + Yt-1 + ut).

10

-5

-10

-15

-20

-25
50 100 150 200 250 300 350 400 450 500

Figure 2.6: Random walk with drift (Yt = -2 + Yt-1 + ut).

16
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

2.3 Unit Root Stochastic Process

According to Gujarati (2003), the random walk model is an

example of what is known in the literature as a unit root
process.

Let us write the random walk model (1) as:

Yt = Yt-1 + ut (-1 1) (6)

This model resembles the Markov first-order

autoregressive model [AR(1)], mentioned in the basic
econometrics course, serial correlation topic. If = 1,
(6) becomes a random walk without drift. If is in fact
1, we face what is known as the unit root problem, that
is, a situation of nonstationarity. The name unit root is
due to the fact that = 1. Technically, if = 1, we can
write (6) as Yt – Yt-1 = ut. Now using the lag operator L
so that LYt = Yt-1, L2Yt = Yt-2, and so on, we can write (6)
as (1-L)Yt = ut. If we set (1-L) = 0, we obtain, L = 1,
hence the name unit root. Thus, the terms
nonstationarity, random walk, and unit root can be
treated as synonymous.

If, however, |ρ| 1, that is if the absolute value of

is less than one, then it can be shown that the time
series Yt is stationary.

2.4 Illustrative Examples

Consider the AR(1) model as presented in equation (6).

Generally, we can have three possible cases:

17
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Case 1: < 1 and therefore the series Yt is stationary.

A graph of a stationary series for = 0.67 is

presented in Figure 2.7.

Case 2: > 1 where in this case the series explodes. A

graph of an explosive series for = 1.26 is

presented in Figure 2.8.

Case 3: = 1 where in this case the series contains a

unit root and is non-stationary. Graph of
stationary series for = 1 are presented in
Figure 2.9.

In order to reproduce the graphs and the series which are

stationary, exploding and nonstationary, we type the
following commands in Eviews:

Step 1: Open a new workfile (say, undated type),

containing 200 observations.

Step 2: Generate X, Y, Z as the following commands:

smpl 1 1

genr X=0

genr Y=0

genr Z=0

smpl 2 200

genr X=0.67*X(-1)+nrnd

genr Y=1.26*Y(-1)+nrnd

genr Z=Z(-1)+nrnd

18
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

smpl 1 200

Step 3: Plot X, Y, Z using the line plot type (Figures

2.7, 2.8, and 2.9).

plot X

plot Y

plot Z

-1

-2

-3

-4
25 50 75 100 125 150 175 200

Figure 2.7: A stationary series

19
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

1.6E+19

1.4E+19

1.2E+19

1.0E+19

8.0E+18

6.0E+18

4.0E+18

2.0E+18

0.0E+00
25 50 75 100 125 150 175 200

Figure 2.8: An explosive series

-5

-10

-15

-20

-25
25 50 75 100 125 150 175 200

Figure 2.9: A nonstationary series

20
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

3. THE UNIT ROOTS AND SPURIOUS REGRESSIONS

3.1 Spurious Regressions

Most macroeconomic time series are trended and therefore

in most cases are nonstationary. The problem with
nonstationary or trended data is that the standard
ordinary least squares (OLS) regression procedures can
easily lead to incorrect conclusions. According to
Asteriou (2007), it can be shown in these cases that the
regression results have very high value of R2 (sometimes
even higher than 0.95) and very high values of t-ratios
(sometimes even higher than 4), while the variables used
in the analysis have no real interrelationships.

Asteriou (2007) states that many economic series

typically have an underlying rate of growth, which may or
may not be constant, for example GDP, prices or money
supply all tend to grow at a regular annual rate. Such
series are not stationary as the mean is continually
rising however they are also not integrated as no amount
of differencing can make them stationary. This gives rise
to one of the main reasons for taking the logarithm of
data before subjecting it to formal econometric analysis.
If we take the logarithm of a series, which exhibits an
average growth rate we will turn it into a series which
follows a linear trend and which is integrated. This can
be easily seen formally. Suppose we have a series Xt,
which increases by 10% every period, thus:

Xt = 1.1Xt-1

21
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

If we then take the logarithm of this we get

log(Xt) = log(1.1) + log(Xt-1)

Now the lagged dependent variable has a unit coefficient

and each period it increases by an absolute amount equal
to log(1.1), which is of course constant. This series
would now be I(1).

More formally, consider the model:

Yt = β1 + β2Xt + ut (7)

where ut is the error term. The assumptions of classical

linear regression model (CLRM) require both Yt and Xt to
have zero and constant variance (i.e., to be stationary).
In the presence of nonstationarity, then the results
obtained from a regression of this kind are totally
spurious9 and these regressions are called spurious
regressions.

The intuition behind this is quite simple. Over time, we

expect any nonstationary series to wander around (see
Figure 3.1), so over any reasonably long sample the
series either drift up or down. If we then consider two
completely unrelated series which are both nonstationary,
we would expect that either they will both go up or down
together, or one will go up while the other goes down. If
we then performed a regression of one series on the
other, we would then find either a significant positive

9
This was first introduced by Yule (1926), and re-examined by Granger and Newbold
(1974) using the Monte Carlo simulations.

22
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

relationship if they are going in the same direction or a

significant negative one if they are going in opposite
directions even though really they are both unrelated.
This is the essence of a spurious regression.

It is said that a spurious regression usually has a very

high R2, t statistics that appear to provide significant
estimates, but the results may have no economic meaning.
This is because the OLS estimates may not be consistent,
and therefore all the tests of statistical inference are
not valid.

Granger and Newbold (1974) constructed a Monte Carlo

analysis generating a large number of Yt and Xt series
containing unit roots following the formulas:

Yt = Yt-1 + eYt (8)

Xt = Xt-1 + eXt (9)

where eYt and eXt are artificially generated normal random

numbers (as the same way performed in section 2.4).

Since Yt and Xt are independent of each other, any

regression between them should give insignificant
results. However, when they regressed the various Yts to
the Xts as show in equation (8), they surprisingly found
that they were unable to reject the null hypothesis of β2
= 0 for approximately 75% of their cases. They also found
that their regressions had very high R2s and very low
values of DW statistics.

23
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

To see the spurious regression problem, we can type the

following commands in Eviews (after opening the new
workfile, say, undated with 500 observations) to see how
many times we can reject the null hypothesis of β2 = 0.
The commands are:

smpl @first @first+1 (or smpl 1 1)

genr Y=0
genr X=0
smpl @first+1 @last (or smpl 2 500)
genr Y=Y(-1)+nrnd
genr X=X(-1)+nrnd
scat(r) Y X
smpl @first @last
ls Y c X
An example of a plot of Y against X obtained in this way
is shown in Figure 3.1. The estimated equation between
these two simulated series is:

Table 3.1: Spurious regression

24
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

10

-10

-20
Y

-30

-40

-50
-10 -5 0 5 10 15 20 25

Figure 3.1: Scatter plot of a spurious regression

Granger and Newbold (1974) proposed the following “rule

of thumb” for detecting spurious regressions: If R2 > DW
statistic or if R2 1 then the estimated regression
‘must’ be spurious.

To understand the problem of spurious regression better,

it might be useful to use an example with real economic
data. This example was conducted by Asteriou (2007).
Consider a regression of the logarithm of real GDP (Yt) to
the logarithm of real money supply (Mt) and a constant.
The results obtained from such a regression are the
following:

25
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Yt = 0.042 + 0.453Mt; R2 = 0.945; DW = 0.221

(4.743) (8.572)

Here we see very good t-ratios, with coefficients that

have the right signs and more or less plausible
magnitudes. The coefficient of determination is very high
(R2 = 0.945), but there is a high degree of
autocorrelation (DW = 0.221). This shows evidence of the
possible existence of spurious regression. In fact, this
regression is totally meaningless because the money
supply data are for the UK economy and the GDP figures
are for the US economy. Therefore, although there should
not be any significant relationship, the regression seems
to fit the data very well, and this happens because the
variables used in the example are, simply, trended (i.e.
nonstationary). So, Asteriou (2007) recommends that
econometricians should be very careful when working with
trended variables. You can find more such examples in
Gujarati (2011, pp.224-226)

3.2 Explaining the Spurious Regression Problem

According to Asteriou (2007), in a slightly more formal

way the source of the spurious regression problem comes
from the fact that if two variables, X and Y, are both
stationary, then in general any linear combination of
them will certainly be stationary. One important linear
combination of them is of course the equation error, and
so if both variables are stationary, the error in the
equation will also be stationary and have a well-behaved

26
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

distribution. However, when the variables become

nonstationary, then of course we can not guarantee that
the errors will be stationary and in fact as a general
rule (although not always) the error itself be
nonstationary and when this happens, we violate the basic
CLRM assumptions of OLS regression. If the errors were
nonstationary, we would expect them to wander around and
eventually get large. But OLS regression because it
selects the parameters so as to make the sum of the
squared errors as small as possible will select any
parameter which gives the smallest error and so almost
any parameter value can result.

The simplest way to examine the behaviour of ut is to

rewrite (7) as:

ut = Yt – β1 – β2Xt (10)

or, excluding the constant β1 (which only affects ut

sequence by rescaling it):

ut = Yt – β2Xt (11)

If Yt and Xt are generated by equations (8) and (9), then

if we impose the initial conditions Y0 = X0 = 0 we get
that:

ut Y0 eY1 eY2 ... eYi 2(X0 eX1 eX2 ... eXi)

or
t t
ut eYi 2 eXi (12)
i 1 i 1

27
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

From equation (12), we realize that the variance of the

error term will tend to become infinitely large as t
increases. Hence, the assumptions of the CLRM are
violated, and therefore, any t test, F test or R2 are
unreliable.

In terms of equation (7), there are four different cases

to discuss:

Case 1: Both Yt and Xt are stationary10, and the CLRM is

appropriate with OLS estimates being BLUE.

Case 2: Yt and Xt are integrated of different orders. In

this case, the regression equations are
meaningless.

Case 3: Yt and Xt are integrated of the same order [often

I(1)]and the ut sequence contains a stochastic
trend. In this case, we have spurious regression
and it is often recommended to re-estimate the
regression equation in the semi-difference
methods (such as the FGLS method: Orcutt-Cochrane
procedure, AR(1), and Newey-West standard error).
These methods did mention in my lectures about
serial correlation and a brief review of basic
econometrics.

Case 4: Yt and Xt are integrated of the same order and the

ut is stationary. In this special case, Yt and Xt

10
Based on the statistical tests such as ADF, PP, and KPSS.

28
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

are said to be cointegrated. The concept of

cointegration will be examined in detail later.

4. TESTING FOR UNIT ROOTS

4.1 Graphical Analysis

According to Gujarati (2003), before one pursues formal

tests, it is always advisable to plot the time series
under study. Such a plot (line graph of the level) and
correlogram [of both the level (ACF) and the first
difference (ACF and PACF)] gives an initial clue about
the likely nature of the time series. Such a intuitive
feel is the starting point of formal tests of
stationarity (i.e. choose the appropriate test equation).

4.2 Autocorrelation Function and Correlogram

Autocorrelation is the correlation between a variable

lagged one or more periods and itself. The correlogram or
autocorrelation function is a graph of the
autocorrelations for various lags of a time series data.
According to Hanke (2005), the autocorrelation
coefficients11 for different time lags for a variable can
be used to answer the following questions:

(1) Are the data random? (This is usually used for the
diagnostic tests of forecasting models).

(2) Do the data have a trend (nonstationary)?

11
This is not explained in this lecture. You can make references from either
Gujarati (2003: 808-813), Hanke (2005: 60-74), or Nguyen Trong Hoai et al (2009:
Chapter 3, 4, and 8).

29
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

(3) Are the data stationary?

(4) Are the data seasonal?

Besides, the correlogram is very useful when selecting

the appropriate p and q in the ARIMA models and ARCH
family models12.

(1) If a series is random, the autocorrelations (i.e.

ACF) between Yt and Yt-k for any lag k are close to
zero (i.e. the autocorrelation coefficient is
statistically insignificant). The successive values
of a time series are not related to each other
(Figure 4.1). In other words, Yt and Yt-k are
independent.

(2) If a series has a (stochastic) trend, successive

observations are highly correlated, and the
autocorrelation coefficients are typically
significantly different from zero for the first
several time lags and then gradually drop toward
zero as the number of lags increases. The
autocorrelation coefficient for time lag 1 is often
very large (close to 1). The autocorrelation
coefficient for time lag 2 will also be large.
However, it will not be as large as for time lag 1
(Figure 4.2).

(3) If a series is stationary, the autocorrelation

coefficients for lag 1 or lag 2 are significantly

12
See Nguyen Trong Hoai et al, 2009 and my lecture about ARIMA models.

30
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

different from zero and then suddenly die out as the

number of lags increases (Figure 4.3). In other
words, Yt and Yt-1, Yt and Yt-2, Yt and Yt-3 are weakly
correlated; but Yt and Yt-k [as k increases] are
completely independent.

(4) If a series has a seasonal pattern, a significant

autocorrelation coefficient will occur at the
seasonal time lag or multiples of seasonal lag
(Figure 4.4). This pattern is not important within
this lecture context.

Figure 4.1: Correlogram of a random series

31
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Figure 4.2: Correlogram of a nonstationary series

Figure 4.3: Correlogram of a stationary series

32
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Figure 4.4: Correlogram of a seasonal series

The correlogram becomes very useful for time series

forecasting and other practical (business) implications.
If you conduct academic studies, however, it is necessary
to provide more formal statistics such as t statistic13,
Box-Pierce Q statistic, Ljung-Box (LB) statistic, or
especially unit root tests.

4.3 Simple Dickey-Fuller Test for Unit Roots

Dickey and Fuller (1979, 1981) devised a procedure to

formally test for nonstationarity (hereafter refer to DF
test). The key insight of their test is that testing for
nonstationarity is equivalent to testing for the
existence of a unit root. Thus the obvious test is the
following which is based on the following simple AR(1)
model:

13
See Nguyen Trong Hoai et al, 2009 and my lecture about ARIMA models to understand
the standard error in time series econometrics s.e. = 1/ n .

33
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Yt = Yt-1 + ut (13)

What we need to examine here is = 1 (unity and hence

‘unit root’). Obviously, the null hypothesis is H0: = 1,
and the alternative hypothesis is H1: < 1 (why?).

We obtain a different (more convenient) version of the

test by subtracting Yt-1 from both sides of (13):

Yt – Yt-1 = Yt-1 – Yt-1 + ut

∆Yt = ( - 1)Yt-1 + ut

∆Yt = Yt-1 + ut (14)

where = ( - 1). Then, now the null hypothesis is H0:

= 0, and the alternative hypothesis is H1: < 0 (why?).
In this case, if = 0, then Yt follows a pure random walk
(and, of course, Yt is nonstationary).

Dickey and Fuller (1979) also proposed two alternative

regression equations that can be used for testing for the
presence of a unit root. The first contains a constant in
the random walk process as in the following equation:

∆Yt = + Yt-1 + ut (15)

According to Asteriou (2007), this is an extremely

important case, because such processes exhibit a definite
trend in the series when = 0, which is often the case
for macroeconomic variables.

The second case is also allow, a non-stochastic time

trend in the model, so as to have:

34
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

∆Yt = + T + Yt-1 + ut (16)

The Dickey-Fuller test for stationarity is the simply the

normal ‘t’ test on the coefficient of the lagged
dependent variable Yt-1 from one of the three models (14,
15, and 16). This test does not, however, have a
conventional ‘t’ distribution and so we must use special
critical values which were originally calculated by
Dickey and Fuller. This is also known as the Dickey-
Fuller tau statistic (Gujarati, 2003; 2011). However,
most modern statistical packages such as Stata and Eviews
routinely produce the critical values for Dickey-Fuller
tests at 1%, 5%, and 10% significant levels.

MacKinnon (1991,1996) tabulated appropriate critical

values for each of the three above models and these are
presented in Table 4.1.

Table 4.1: Critical values for DF test

Model 1% 5% 10%

∆Yt = Yt-1 + ut -2.56 -1.94 -1.62

∆Yt = + Yt-1 + ut -3.43 -2.86 -2.57

∆Yt = + T + Yt-1 + ut -3.96 -3.41 -3.13

Standard critical values -2.33 -1.65 -1.28

Source: Asteriou (2007)

35
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

In all cases, the test concerns whether = 0. The DF test

statistic is the t statistic for the lagged dependent
variable. If the DF statistical value is smaller in
absolute terms than the critical value then we reject the
null hypothesis of a unit root and conclude that Yt is a
stationary process.

4.4 Augmented Dickey-Fuller Test for Unit Roots

As the error term is unlikely to be white noise, Dickey

and Fuller extended their test procedure suggesting an
augmented version of the test (hereafter refer to ADF
test) which includes extra lagged terms of the dependent
variable in order to eliminate autocorrelation in the
test equation.

The lag length14 on these extra terms is either determined

by Akaike Information Criterion (AIC) or Schwarz
Bayesian/Information Criterion (SBC, SIC), or more
usefully by the lag length necessary to whiten the
residuals (i.e. after each case, we check whether the
residuals of the ADF regression are autocorrelated or not
through LM tests and not the DW test (why?)).

The three possible forms of the ADF test are given by the
following equations:
p
Yt Yt 1 i Yt i ut (17)
i 1

14
This issue will be discussed later in this lecture.

36
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

p
Yt Yt 1 i Yt i ut (18)
i 1

p
Yt T Yt 1 i Yt i ut (19)
i 1

The difference between the three regressions concerns the

presence of the deterministic elements α and T. The
critical values for the ADF test are the same as those
given in Table 4.1 for the DF test.

According to Asteriou (2007), unless the econometrician

knows the actual data-generating process, there is a
question concerning whether it is most appropriate to
estimate (17), (18), or (19). Daldado, Jenkinson and
Sosvilla-Rivero (1990) suggest a procedure which starts
from estimation of the most general model given by (19)
and then answering a set of questions regarding the
appropriateness of each model and moving to the next
model. This procedure is illustrated in Figure 4.1. It
needs to be stressed here that, although useful, this
procedure is not designed to be applied in a mechanical
fashion. Plotting the data and observing the graph is
sometimes very useful because it can clearly indicate the
presence or not of deterministic regressors. However,
this procedure is the most sensible way to test for unit
roots when the form of the data-generating process is
typically unknown.

In practical studies, researchers mostly use both the ADF

and the Phillips-Perron (PP) tests. Because the

37
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

distribution theory that supporting the Dickey-Fuller

tests is based on the assumption of random error terms
[iid(0, 2)], when using the ADF methodology we have to
make sure that the error terms are uncorrelated and they
really have a constant variance. Phillips and Perron
(1988) developed a generalization of the ADF test
procedure that allows for fairly mild assumptions
concerning the distribution of errors. The regression for
the PP test is similar to equation (15).

∆Yt = + Yt-1 + et (20)

While the ADF test corrects for higher order serial

correlation by adding lagged differenced terms on the
right-hand side of the test equation, the PP test makes a
correction to the t statistic of the coefficient from
the AR(1) regression (semi-difference method) to account
for the serial correlation in et.

So, the PP statistics are just modifications of the ADF

t statistics that take into account the less restrictive
nature of the error process. The expressions are
extremely complex to derive and are beyond the scope of
this lecture. Luckily, since most statistical packages
have routines available to calculate these statistics, it
is good for researcher to test the order of integration
of a series performing the PP test as well. The
asymptotic distribution of the PP t statistic is the same
as the ADF t statistic and therefore the MacKinnon
(1991,1996) critical values are still applicable.

38
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Figure 4.1: Procedure for testing for unit roots

Estimate the model

p
Yt T Yt 1 i Yt i ut
i 1

NO STOP: Conclude
= 0? that there is no
unit root
YES: Test for the
presence of the trend
NO
is = 0? NO STOP: Conclude
given that = 0? that Yt has a
= 0? YES
unit root

YES

Estimate the model STOP: Conclude

p NO
Yt Yt 1 i Yt i ut that there is no
i 1 unit root
is = 0?

YES: Test for the NO

presence of the constant

is = 0? NO STOP: Conclude
given that = 0? that Yt has a
YES
= 0? unit root

YES
STOP: Conclude
that there is no
Estimate the model NO
unit root
p
Yt Yt 1 i Yt i ut
i 1
is = 0? YES STOP: Conclude
that Yt has a
Source: Asteriou (2007) unit root

39
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

As with the ADF test, the PP test can be performed with

the inclusion of a constant and linear trend, or neither
in the test regression.

Dickey-Fuller tests may have low power (H0 of unit root

not rejected, whereas in reality there may be no unit
root) when ρ is close to one. This could be the case of
trend stationarity (H0). An alternative test is KPSS
(Kwiatkowski-Phillips-Schmidt-Shin). Its test procedure
is briefly summarized as:

(1) Regress Yt on intercept and time trend and obtain

OLS residuals et.
t
(2) Calculate partial sums St = es for all t.
s 1

2
T
S2t
(3) Calculate the test statistic KPSS = T 2
, and
s 1 ˆ
compare with critical value.

The critical values are routinely produced by

statistical packages such as Stata and Eviews. The null
hypothesis is rejected if the KPSS test statistic is
larger than the selected critical value.

4.5 Performing Unit Root Tests in Eviews

4.5.1 The DF/ADF test

Table 4.2: DF/ADF test procedure

Step 1 Open the file ADF.wf1 by clicking

File/Open/Workfile and then choosing the file
name from the appropriate path.

40
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Step 2 Let’s assume that we want to examine whether the

series named GDP contains a unit root. Double
click on the series named ‘GDP’ to open the
series window and choose View/Unit Root Test …
In the unit-root test dialog box that appears,
choose the type test (i.e. the Augmented Dickey-
Fuller) by clicking on it.

Step 3 We then specify whether we want to test for a

unit root in the level, first difference, or
second difference of the series. We can use this
option to determine the number of unit roots in
the series. However, we usually start with the
level and if we fail to reject the test in
levels we continue with testing the first
difference and so on.

Step 4 We also have to specify which model of the three

ADF models we wish to use (i.e. whether to
include a constant, a constant and linear trend,
or neither in the test regression). For the
model given by equation (17) click on ‘none’ in
the dialog box; for the model given by equation
(18) click on ‘intercept’ in the dialog box; and
for the model given by equation (19) click on
‘intercept and trend’ in the dialog box.

Step 5 Finally, we have to specify the number of lagged

dependent variables to be included in the model
in order to correct the presence of serial
correlation. In practice, we just click the
‘automatic selection’ on the ‘lag length’ dialog
box.

Step 6 Having specified these options, click <OK> to

carry out the test. Eviews reports the test

41
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

statistic together with the estimated test

regression.

Step 7 We reject the null hypothesis of a unit root

against the one-sided alternative if the ADF
statistic is less than (lies to the left of) the
critical value, and we conclude that the series
is stationary.

Step 8 After running a unit root test, we should

examine the estimated test regression reported
by Eviews, especially if unsure about the lag
structure or deterministic trend in the series.
We may want to rerun the test equation with a
different selection of right-hand variables (add
or delete the constant, trend, or lagged
differences) or lag order.

Source: Asteriou (2007)

Figure 4.2: Illustrative steps in Eviews (ADF)

42
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Table 4.3: ADF test of GDP series

This figure is
positive, so
the selected
model is
incorrect
(see Gujarati
(2003)).

43
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

4.5.2 The PP/KPSS test

Table 4.4: PP/KPSS test procedure

Step 1 Open the file PP.wf1 by clicking

File/Open/Workfile and then choosing the file
name from the appropriate path.

Step 2 Let’s assume that we want to examine whether the

series named GDP contains a unit root. Double
click on the series named ‘GDP’ to open the
series window and choose View/Unit Root Test …
In the unit-root test dialog box that appears,
choose the type test (i.e. the Phillipd-
Perron/Kwiatkowski-Phillips-Schmidt-Shin) by
clicking on it.

Step 3 We then specify whether we want to test for a

unit root in the level, first difference, or
second difference of the series. We can use this
option to determine the number of unit roots in
the series. However, we usually start with the
level and if we fail to reject the test in
levels we continue with testing the first
difference and so on.

Step 4 We also have to specify which model of the three

we need to use (i.e. whether to include a
constant, a constant and linear trend, or
neither in the test regression). For the random
walk model click on ‘none’ in the dialog box;
for the random with drift model click on
‘intercept’ in the dialog box; and for the
random walk with drift and with deterministic
trend model click on ‘intercept and trend’ in
the dialog box.

44
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Step 5 Finally, for the PP/KPSS test we specify the lag

truncation to compute the Newey-West standard
error consistent estimate of the spectrum at
zero frequency.

Step 6 Having specified these options, click <OK> to

carry out the test. Eviews reports the test
statistic together with the estimated test
regression.

Step 7 We reject the null hypothesis of a unit root

against the one-sided alternative if the ADF
statistic is less than (lies to the left of) the
critical value, and we conclude that the series
is stationary.

Source: Asteriou (2007)

Figure 4.3: Illustrative steps in Eviews (PP)

45
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Table 4.5: PP test of GDP series

This figure is
positive, so the
selected model
is incorrect
(see Gujarati
(2003)).

Table 4.6: ADF test of log(EX)

(TABLE13-1.wf1, Gujarati, 2011)

Table 4.7: ADF test of log(EX)

(TABLE13-1.wf1, Gujarati, 2011)

46
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Table 4.8: PP test of log(EX)

(TABLE13-1.wf1, Gujarati, 2011)

5. SHORT-RUN AND LONG-RUN RELATIONSHIPS

5.1 Understanding Concepts

In case of bivariate model, you have once known the

static or short-run causal relationship between two time
series Yt and Xt, where Yt is dependent variable and Xt is
independent variable. The OLS regression often
experiences the serial correlation, and we perform
various remedies such as semi-difference methods
(Cochrane-Orcutt, Prais-Winsten, AR(1)), first difference
method, and Newey-West standard error. By any way, the
purpose of your study is just to know the short-run slope
or elasticity of Yt with respect to Xt [ Yt/ Yt]. However,
the nature of the structural modeling is to discover the
dynamic causal relationship between Yt and Xt. In such
model, you must at least distingush between short-run and

47
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

long-run relationship [slope or elasticity]. To simplify

our analysis, we consider the simple autoregressive
distributed lag model [ARDL(1,1)] in the following form:

Yt = A0 + A1Yt-1 + B0Xt + B1Xt-1 + ut (21)

We can analyse both short-run and long-run effects

(slopes or elasticities) as follows:

(1) Short-run or static effect:

Yt
B0 (22)
Xt

(2) Long-run or dynamic or equilibrium effect:

YT B0 B1
(23)
Xt 1 A1

Proof:

Yt
B0
Xt

Yt 1 Yt
A1 B1 = A1.B0 B1 (why?)
Xt Xt

Yt 2 Yt 1
A1 A1(A1.B0 B1) (why?)
Xt Xt

Yt 3 Yt 2
A1 A12(A1.B0 B1)] (why?)
Xt Xt

Yt 1 Yt
A1 A1(A1.B0 B1)] (why?)
Xt Xt

48
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

If |A1| < 1, the cummulative effect or long-run slope

(Slr) will be the sum of all derivatives:

Slr B0 [A1B0 B1] A1[A1B0 B1] A12(A1.B0 B1)] ... A1(A1.B0 B1)] (24)

Multiply both sides of (24) by A1, we have:

A1Slr A1B0 A1[A1B0 B1] A12(A1.B0 B1)] ... A1(A1.B0 B1)] (25)

By substract (25) from (24), we obtain:

Slr – A1Slr = B0 + B1

B0 B1
Slr = = equation (23)
1 A1

We can also take expectations to derive the long-run

relation between Yt and Xt:

E(Yt) = A0 + A1E(Yt-1) + B0E(Xt) + B1E(Xt-1)

E(Yt) = A0 + A1E(Yt) + B0E(Xt) + B1E(Xt)

E(Yt) - A1E(Yt) = A0 + (B0 + B1)E(Xt)

(1-A1)E(Yt) = A0 + (B0 + B1)E(Xt)

A0 (B0 B1)
=> E(Yt) = E(Xt)
1 A1 (1 A1

= α + βE(Xt)

or simply to write:

Y* = α + βX* (26)

Here, β = (B0+B1)/(1-A1) is the long-run effect of a

lasting shock in Xt. And the short-run effect of a change
in Xt is B0.

In the same token, we can expand the model ARDL(p,q):

49
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

(1) Short-run or static effect:

Yt
B0 (27)
Xt

(2) Long-run or dynamic or equilibrium effect:

YT B0 B1 B2... Bq
(28)
Xt 1 A1 A2... Ap

5.2 ARDL and Error Correction Model (ECM)

By substracting Yt-1 both sides of equation (23), and re-

arraning, we have:

Yt–Yt-1 = A0 + A1Yt-1 – Yt-1 + B0Xt – B0Xt-1 + B0Xt-1 + B1Xt-1 + ut

Yt = A0 – (1–A1)Yt-1 + B0 Xt + (B0+B1)Xt-1 + ut

A0 (B0 B1)
= B0 Xt – (1–A1) Yt 1 Xt 1 + ut
(1 A1) (1 A1)

= B0 Xt – (1–A1) Yt 1 Xt 1 + ut

= B0 Xt – Yt 1 Xt 1 + ut (29a)

= B0 Xt – ECTt-1 + ut (29b)

This is applicable with all ARDL models. Part in brackets

of equation (29a) is error-correction term (equilibrium
error). Equations (29a or 29b) are widely known as the
error correction model (ECM). Therefore, ECM and ARDL are
basically the same if the series Yt and Xt are integrated
of the same order [often I(1)] and cointegrated.

In this model, Yt and Xt are assumed to be in long-run

equilibrium, i.e. changes in Yt relate to changes in Xt

50
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

according B1. If Yt-1 deviates from the optimal value (i.e.

its equilibrium), there is a correction. Speed of
adjustment is given by = (1-A1), which is between > 0
and <1. This will be further discussed later.

6. ENGLE-GRANGER 2-STEP METHOD OF COINTEGRATION

6.1 Cointegration

According to Asteriou (2007), the concept of

cointegration was first introduced by Granger (1981) and
elaborated further Engle and Granger (1987), Engle and
Yoo (1987), Phillips and Ouliaris (1990), Stock and
Watson (1988), Phillips (1986 and 1987), and Johansen
(1988, 1991, and 1995).

It is known that trended time series can potentially

create major problems in empirical econometrics due to
spurious regressions. One way of resolving this is to
difference the series successively until stationary is
achieved and then use the stationary series for
regression analysis. According to Asteriou (2007), this
solution, however, is not ideal because it not only
differences the error process in the regression, but also
no longer gives a unique long-run solution.

If two variables are nonstationary, then we can represent

the error as a combination of two cumulated error
processes. These cumulated error processes are often
called stochastic trends and normally we could expect
that they would combine to produce another non-stationary

51
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

process. However, in the special case that two variables,

say Xt and Yt, are really related, then we would expect
them to move together and so the two stochastic trends
would be very similar to each other and when we combine
them together it should be possible to find a combination
of them which eliminates the nonstationarity. In this
special case, we say that the variables are cointegrated
(Asteriou, 2007). Cointegration becomes an overriding
requirement for any economic model using nonstationary
time series data. If the variables do not co-integrate,
we usually face the problems of spurious regression and
econometric work becomes almost meaningless. On the other
hand, if the stochastic trends do cancel to each other,
then we have cointegration.

Suppose that, if there really is a genuine long-run

relationship between Yt and Xt, the although the variables
will rise overtime (because they are trended), there will
be a common trend that links them together. For an
equilibrium, or long-run relationship to exist, what we
require, then, is a linear combination of Yt and Xt that
is a stationary variable [an I(0) variable]. A linear
combination of Yt and Xt can be directly taken from
estimating the following regression:

Yt = β1 + β2Xt + ut (30)

And taking the residuals:

ût Yt ˆ ˆX (31)
1 2 t

52
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

If ût ~ I(0), then the variables Yt and Xt are said to be

co-integrated.

6.2 An example of cointegration

TABLE14-1.wf1 gives quarterly data on personal
consumption expenditure (PCE) and personal disposible
(i.e. after-tax) income (PDI) for the USA for the period
1970-2008 (Gujarati, 2011: pp.226). Both graph (Figure
6.1) and ADF tests (Table 6.1) indicate that these two
series are not stationary. They are I(1), that is, they
have stochastic trends. In addition, the regression of
log(PCE) on log(PDI) seems to be spurious (Table 6.3).

Since both series are trending, let us see what happens

if we add a trend variable to the model. The elasticity
coefficient is now changed, but the regression is still
spurious (Table 6.4). However, after estimating the
regression of log(PCE) on log(PDI) and trend, we realize
that the obtained residuals is a stationary series
[I(0)]. This implies that a linear combination (et =
log(PCE) – b1 – b2log(PDI) – b3T) cancels out the
stochastic trends in the two variables. Therefore, this
regression is, in fact, not spurious. In other words, the
variables log(PCE) and log(PDI) are cointegrated.

53
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

9.2

9.0

8.8

8.6

8.4

8.2

8.0

7.8

7.6
25 50 75 100 125 150

LOG(PCE) LOG(PDI)

Figure 6.1: Logs of PDI and PCE, USA 1970-2008.

Table 6.1: Unit root tests for log(PCE)

54
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Table 6.2: Unit root tests for log(PDI)

Table 6.3: OLS regression of log(PCE) on log(PDI)

Table 6.4: Regression of log(PCE) on log(PDI) and trend

55
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Economically speaking, two variables will be cointegrated

if they have a long-run, or equilibrium, relationship
between them. In the present context, economic theory
tells us that there is a strong relationship between
consumption expenditure and personal disposible income.

In the language of cointegration theory, the equation

log(PCE) = B1 + B2log(PDI) + B3T is known as a
cointegrating regression and the slope parameters B2 and
B3 are known as cointegrating parameters.

6.3 Engle-Granger Tests for Cointegration

For single equation, the simple tests of
cointegration are DF and ADF unit root tests on
the residuals estimated from the cointegrating
regression. These modified tests by the Engle-
Granger (EG) and Augmented Engle-Granger (AEG)
tests. Notice the difference between the unit root
and cointegration tests. Tests for unit roots are
performed on single time series, whereas simple
cointegration deals with the relationship among a
group of variables, each having a unit root
(Gujarati, 2011).
6.4 Interpretation of the ECM

According to Asteriou (2007), the concepts of

cointegration and the error correction mechanism are very
closely related. To understand the ECM, it is better to

56
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

think first of the ECM as a convenient reparametrization

of the general linear autoregressive distributed lag
(ARDL) model (see Section 5).

Consider the very simple dynamic ARDL model describing

the behaviour of Yt in terms of Xt as equation (21):

Yt = A0 + A1Yt-1 + B0Xt + B1Xt-1 + ut (21)

where ut ~ iid(0, 2).

In this model15, the parameter B0 denotes the short-run

reaction of Yt after a change in Xt. The long-run effect
is given when the model is in equilibrium where:

Y* = α + βX* (26)

Recall that he long-run effect (slope oe elasticity)

between Yt and Xt is captured by β=(B0+B1)/(1-A1). It is
noted that, we need to make the assumption that |A1| < 1
(why?) in order that the short-run model (21) converges
to a long-run solution [equation (26), see Section 6].

The ECM is shown in equation (29a or 29b):

Yt = B1 Xt – Yt 1 Xt 1 + ut (29a)

or

Yt = B1 Xt – ECTt-1 + ut (29b)

According to Asteriou (2007), what is of importance here

is that when the two variables Yt and Xt are cointegrated,

15
We can easily expand this model to a more general case for large numbers of
lagged terms [ARDL(p,q)].

57
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

the ECM incorporates not only short-run but also long-run

effects. This is because the long-run equilibrium [Yt-1 –
α – βXt-1] is included in the model together with the
short-run dynamics captured by the differenced term.
Another important advantage is that all the terms in the
ECM model are stationary and the standard OLS estimation
is therefore valid. This is because if Yt and Xt are I(1),
then ∆Yt and ∆Xt are I(0), and by definition if Yt and Xt
are cointegrated then their linear combination [Yt-1 – α –
βXt-1] ~ I(0).

A final important point is that the coefficient = (1-A1)

provides us with information about the speed of
adjustment in cases of disequilibrium. To understand this
better, consider the long-run condition. When equilibrium
holds, then [Yt-1 – α – βXt-1] = 0. However, during periods
of disequilibrium this term is no longer be zero and
measures the distance the system is away from
equilibrium. For example, suppose that due to a series of
negative shocks in the economy in period t-1. This causes
[Yt-1 – α – βXt-1] to be negative because Yt-1 has moved
below its long-run equilibrium path. However, since =
(1-A1) is positive (why?), the overall effect is to boost
∆Yt back towards its long-run path as determined by Xt in
equation (26). Notice that the speed of this adjustment
to equilibrium is dependent upon the magnitude of =
(1-A1).

58
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

The coefficient in equation (29a,b) is the error-

correction coefficient and is also called the adjustment
coefficient. In fact, tells us how much of the
adjustment to equilibrium takes place each period, or how
much of the equilibrium error is corrected each period.
According to Asteriou (2007), it can be explained in the
following ways:

(1) If ~ 1, then nearly 100% of the adjustment takes

place within the period16, or the adjustment is very
fast.

(2) If ~ 0.5, then about 50% of the adjustment takes

place each period.

(3) If ~ 0, then there seems to be no adjustment.

According to Asteriou (2007), the ECM is important and

popular for many reasons:

(1) Firstly, it is a convenient model measuring the

correction from disequilibrium of the previous period
which has a very good economic implication.

(2) Secondly, if we have cointegration, ECM models are

formulated in terms of first difference, which
typically eliminate trends from the variables
involved; they resolve the problem of spurious
regressions.

16
Depending on the kind of data used, say, annually, quarterly, or monthly.

59
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

(3) A third very important advantage of ECM models is the

ease with they can fit into the general-to-specific
(or Hendry) approach to econometric modeling, which
is a search for the best ECM model that fits the
given data sets.

(4) Finally the fourth and most important feature of ECM

comes from the fact that the disequilibrium error
term is a stationary variable. Because of this, the
ECM has important implications: the fact that the two
variables are cointegrated implies that there is some
automatically adjustment process which prevents the
errors in the long-run relationship becoming larger
and larger.

6.5 Engle-Granger 2-Step Method of Cointegration

Granger (1981) introduced a remarkable link between

nonstationary processes and the concept of long-run
equilibrium; this link is the concept of cointegration.
Engle and Granger (1987) further formulized this concept
by introducing a very simple test for the existence of
co-integrating (i.e. long-run equilibrium) relationships.

This approach involves the following steps:

Table 6.5: Engle-Granger 2-Step Method: Step-by-Step

Step 1 Test the variables for their order of

integration.
The first step is to test each variable to
determine its order of integration. The Dickey-

60
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Fuller and the augmented Dickey-Fuller tests

can be applied in order to infer the number of
unit roots in each of the variables. We might
face three cases:
a) If both variables are stationary (I(0)), it
is not necessary to proceed since standard
time series methods apply to stationary
variables.
b) If the variables are integrated of different
orders, it is possible to conclude that they
are not cointegrated.
c) If both variables are integrated of the same
order, we proceed with step two.

Step 2 Estimate the long-run (possible co-integrating)

relationship.
If the results of step 1 indicate that both Xt
and Yt are integrated of the same order
(usually I(1)) in economics, the next step is
to estimate the long-run equilibrium
relationship of the form:

Yt ˆ ˆX ût
t

and obtain the residuals of this equation.

If there is no cointegration, the results
obtained will be spurious. However, if the
variables are cointegrated, OLS regression
yields consistent estimators for the co-

61
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

integrating parameter ˆ .

Step 3 Check for (cointegration) the order of

integration of the residuals.
In order to determine if the variables are
actually cointegrated, denote the estimated
residual sequence from the equation by ût .
Thus, ût is the series of the estimated
residuals of the long-run relationship. If
these deviations from long-run equilibrium are
found to be stationary, the Xt and Yt are
cointegrated.

Step 4 Estimate the error correction model.

If the variables are cointegrated, the
residuals from the equilibrium regression can
be used to estimate the error correction model
and to analyse the long-run and short-run
effects of the variables as well as to see the
adjustment coefficient, which is the
coefficient of the lagged residual terms of the
long-run relationship identified in step 2. At
the end, we always have to check for the
accuracy of the model by performing diagnostic
tests.

Source: Asteriou (2007)

According to Asteriou (2007), one of the best features of

the Engle-Granger 2-step method is that it is both very

62
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

easy to understand and to implement. However, it also

remains some caveats:

(1) One very important issue has to do with the order

of the variables. When estimating the long-run
relationship, one has to place one variable in the
left-hand side and use the others as regressors.
The test does not say anything about which of the
variables can be used as regressors and why.
Consider, for example, the case of just two
variables, Xt and Yt. One can either regress Yt on
Xt (i.e. Yt = C + DXt + u1t) or choose to reverse
the order and regress Xt on Yt (i.e. Xt = D + EYt +
u2t). It can be shown, which asymptotic theory,
that as the sample goes to infinity the test for
cointegration on the residuals of those two
regressions is equivalent (i.e. there is no
difference in testing for unit roots in u1t and
u2t). However, in practice, in economics we rarely
have very big samples and it is therefore possible
to find that one regression exhibits cointegration
while the other doesn’t. This is obviously a very
undesirable feature of the EG approach. The
problem obviously becomes far more complicated
when we have more than two variables to test.

(2) A second problem is that when there are more than

two variables there may be more than one
integrating relationship, and Engle-Granger 2-step
method using residuals from a single relationship

63
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

can not treat this possibility. So, the most

important problem is that it does not give us the
number of co-integrating vectors.

(3) A third and final problem is that it replies on a

two-step estimator. The first step is to generate
the residual series and the second step is to
estimate a regression for this series in order to
see if the series is stationary or not. Hence, any
error introduced in the first step is carried into
the second step.

The Engle-Granger 2-step method in Eviews

The Engle-Granger 2-step method is very easy to perform

and does not require any more knowledge regarding the use
of Eviews. For the first step, ADF and PP tests on all
variables are needed to determine the order of
integration of the variables. If the variables (let’s say
X and Y) are found to be integrated of the same order,
then the second step involves estimating the long-run
relationship with simple OLS procedure. So the command
here is simply:

ls X c Y

or

ls Y c X

depending on the relationship of the variables. We then

need to obtain the residuals of this relationship which
are given by:

64
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

genr res1=resid

The third step (the actual test for cointegration) is a

unit root test on the residuals, the command for which
is:

adf res1

for no lags, or

adf(4) res1

for 4 lags in the augmentation term, and so on.

6.6 Engle-Granger 2-step Method: An Example

Use the example in Section 6.2, we have the following

error correction model:

Table 6.6: Error correction model of LPCE and LPDI

All coeeficients in the table are individually

statistically significant at 6% or lower level. The

65
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

coefficient of about 0.31 shows that a 1% increase in

log(PDIt/PDIt-1) will lead on average to a 0.31% increase
in ln(PCEt/PCEt-1). This is the short-run consumption-
income elasticity. The long-run value is given by the
cointegrating regression (Table 6.4), which is 0.77.

The coefficient of the error-correction term of about -

0.06 suggests that about 6% of the discrepancy between
long-term and short-term PCE is corrected within a
quarter (quarterly data), suggesting a slow rate of
adjustment to equilibrium.

7. VECTOR AUTOREGRESSIVE MODELS

According to Asteriou (2007), it is quite common in

economics to have models where some variables are not
only explanatory variables for a given dependent
variable, but they are also explained by the variables
that they are used to determine. In those cases, we have
models of simultaneous equations, in which it is
necessary to clearly identify which are the endogenous
and which are the exogenous or predetermined variables.
The decision regarding such a differentiation among
variables was heavily criticized by Sims17 (1980).

According to Sims (1980), if there is simultaneity among

a number of variables, then all these variables should be
treated in the same way. In other words, these should be
no distinction between endogenous and exogenous

17
Nobel prize in economics 2012.

66
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

variables. Therefore, once this distinction is abandoned,

all variables are treated as endogenous. This means that
in its general reduced form, each equation has the same
set of regressors which leads to the development of the
VAR models.

VAR is defined as a system of ARDL equations describing

dynamic evolution of a set of variables from their common
history (here vector implies multiple variables involved.
The VAR model is defined as follow. Suppose we have two
series, in which Yt is affected by not only its past (or
lagged) values but current18 and lagged values of Xt, and
simultaneously, Xt is affected by not only its lagged
values but current and lagged values of Yt. This simple
bivariate VAR model is given by:
p p
Yt = A1 + B1Xt + CjYt j + DjXt j + u1t (32)
j 1 j 1

p p
Xt = A2 + B2Yt + EjYt j + FjXt j + u2t (33)
j 1 j 1

where we assume the u1t and u2t are uncorrelated white-

noise error terms, called impulses or innovations or
shocks in the language of VAR (Gujarati, 2011, pp.266).
Note that these equations are not reduced-form equations

18
Gujarati (2011, pp.266) said that [from the point of view of forecasting] each
equation in VAR contains only its own lagged values and the lagged values of the
other variables in the system. Similarly, Wooldridge (2003, pp.620-621) said that
whether the contemporaneous (current) value is included or not depends partly on
the purpose of the equation. In forecasting, it is rarely included.

67
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

since Yt has a contemporaneous impact on Xt, and Xt has a

contemporaneous impact on Yt.

The bivariate VAR often has the following features

(according to Gujarati, 2011, pp.266):

(1) Although the number of lagged values of each

variable can be different, in most cases we use
the same number of lagged terms in each equation.

(2) The bivariate VAR system given above is known as a

VAR(p) model, because we have p lagged values of
each variable on the right-hand side. If we have
only one lagged value of each variable on the
right-hand side, it would be a VAR(1) model; if
two lagged terms, it would be a VAR(2) model; and
so on.

(3) Although we are dealing with only two variables,

the VAR system can be extended to several
variables.

(4) In the two-variable system, there can be at most

one cointegrating, or equilibrium, relationship
between them. If we have a three-variable VAR
system, there can be at most two cointegrating
relationships between the three variables.

Note that all variables have to be of the same order of

integration. The following cases are distinct:

(1) All the variables are I(0) (stationary): one is in

the standard case, i.e. a VAR in level.

68
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

(2) All the variables are I(d) (non-stationary) with

d > 0 (usually d = 1):

 If there are nonstationary [I(1)] variables, we

estimate a VAR using first differences of
variables [that are I(0)] to remove common
trends.
 The variables are cointegrated: the error
correction term has to be included in the VAR
model. The model becomes a vector error
correction model (VECM) which can be seen as a
restricted VAR. Why (if possible) a VECM instead
of a VAR on differenced variables? VECM gives
long-run structural relations plus information
on adjustment, which provides better insight in
economic processes.
 The variables are not cointegrated: the
variables have first to be differenced d times
and one has a VAR in difference.

According to Asteriou (2007), the VAR model has some good

characteristics. First, it is very simple because we do
not have to worry about which variables are endogenous or
exogenous. Second, estimation is very simple as well, in
the sense that each equation can be estimated with the
usual OLS method separately. Third, forecasts obtained
from VAR models are in most cases better than those
obtained from the far more complex simultaneous equation
models (see Mahmoud, 1984; McNees, 1986). Besides

69
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

forecasting purposes, VAR models also provide framework

for causality tests, which will be presented shortly.

However, on the other hand the VAR models have faced

severe criticism on various different points. According
to Asteriou (2007), the VAR models have been criticised
by the following aspects. First, they are a-theoretic
since they are not based on any economic theory. Since
initially there are no restrictions on any of the
parameters under estimation, in effect ‘everything causes
everything’. However, statistical inference is often used
in the estimated models so that some coefficients that
appear to be insignificant can be dropped, in order to
lead models that might have an underlying consistent
theory. Such inference is normally carried out using what
are called causality tests. Second, they are criticised
due to the loss of degrees of freedom. Thus, if the
sample size is not sufficiently large, estimating that
large a number of parameters, say, a three-variable VAR
model with 12 lags for each, will consume many degrees of
freedom, creating problems in estimation. Third, the
obtained coefficients of the VAR models are difficult to
interpret since they totally lack any theoretical
background.

8. VECM AND COINTEGRATION

8.1 Rank of Cointegrating Matrix

In this section, we extend the single-equation error

correction model to a multivariate one. Let’s assume that

70
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

we have three variables, Yt, Xt and Wt, which can all be

endogenous, i.e. we have that (using matrix notation for
Zt = [Yt,Xt,Wt])

Zt = A1Zt-1 + A2Zt-2 + … + ApZt-p + ut (34)

A VAR(p) can be reformulated in a vector error correction

model as follows:

∆Zt = 1∆Zt-1 + 2∆Zt-2 + … + p-1∆Zt-p+1 + Zt-1 + ut (35)

where the matrix contains information regarding the

long-run relationships. We can decompose = β’ where
will include the speed of adjustment to equilibrium
coefficients, while β’ will be the long-run matrix of
coefficients.

Therefore, the β’Zt-1 term is equivalent to the error

correction term [Yt-1 – α – βXt-1] in the single-equation
case, except that now β’Zt-1 contains up to (p – 1)
vectors in a multivariate framework.

For simplicity, we assume that p = 2, so that we have

only two lagged terms, and the model is then the
following:

Yt Yt 1 Yt 1
Xt 1 Xt 1 Xt 1 ut (33)
Wt Wt 1 Wt 1

or

71
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Yt Yt 1 11 12 Yt 1
11 21 31
Xt 1 Xt 1 21 22 Xt 1 ut (34)
12 22 32
Wt Wt 1 31 32 Wt 1

Let us now analyse only the error correction part of the

first equation (i.e. for ∆Yt on the left-hand side) which
gives:

1Zt-1 = ([ 11β11 + 12β12][ 11β21 + 12β22]

Yt 1

[ 11β31 + 12β32]) Xt 1 (35)

Wt 1

Equation (35) can be rewritten as:

1Zt-1 = 11(β11Yt-1 + β21Xt-1 + β31Wt-1) +

12(β12Yt-1 + β22Xt-1 + β32Wt-1) (36)

which shows clearly the two co-integrating vectors with

their respective speed of adjustment terms 11 and 12.

What are advantages of the multiple equation approach?

(1) From the multiple equation approach we can obtain

estimates for both co-integrating vectors (36), while
the simple equation we have only a linear combination
of the two long-run relationships.

(2) Even if there is only one co-integrating relationship

(for example the first only) rather than two, with
the multiple equation approach we can calculate all
three differing speeds of adjustment coefficients
( 11 21 31)’.

72
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

(3) Only when 21 = 31 = 0, and only one co-integrating

relationship exists, can we then say that the
multiple equation method is the same (reduces to) as
the single equation approach, and therefore, there is
no loss from not modelling the determinants of ∆Xt
and ∆Wt. Here, it is good to mention as well that
when 21 = 31 = 0, is equivalent to Xt and Wt being
weakly exogenous.

Suppose that we have k variables in a VECM, the k k

matrix contains the error correction terms (linear
combinations of k variables in Zt-1 that are I(0).

Table 8.1: Rank of matrix and its implications

Rank of Implications

r = 0 There is no cointegration. No stable long-

run relations between variables. VECM is not
possible (only VAR in first differences).
0 < r < k There are r cointegrating vectors. These
vectors describe the long-run relationships
between variables. VECM is o.k.
r = k All variables are already stationary. No
need to estimate the model as VECM. VAR on
untransformed data is o.k.

8.2 Johansen’s Test for Cointegration

8.2.1 Test Procedure

According to Asteriou (2007), if we have more than two

variables in the model, then there is a possibility of
having more than one co-integrating vector. By this we

73
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

mean that the variables in the model might form several

equilibrium relationships. In general, for k number of
variables, we can have only up to k-1 co-integrating
vectors. To find out how many cointegrating relationships
exist among k variables requires the use of Johansen’s
methodology19.

This method involves the following steps:

Table 8.2: Johansen’s approach

Step 1 Testing the order of integration of all

variables.

Step 2 Setting the appropriate lag length of the

model.
Setting the value of the lag length is affected
by the omission of variables that might affect
only the short-run behaviour of the model. This
is due to the fact that omitted variables
instantly become part of the error term.
Therefore, very careful inspection of the data
and the functional relationship is necessary
before proceeding with estimation in order to
decide whether to include additional variables.
It is white common to use dummy variables to
take into account short-run ‘shocks’ to the
system, such as political events that had
important effects on macroeconomic conditions.
The most common procedure in choosing the

19
Similar to EG approach, the Johansen’s approach also requires all variables in
the system are integrated of the same order 1 [I(1)].

74
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

optimal lag length is to estimate a VAR model

including all variables in levels (non-
differenced). This VAR model should be
estimated for a large number of lags, then
reducing down by re-estimating the model for
one lag less until we reach zero lag.
In each of these models, we inspect the values
of the AIC and the SBC criteria, as well as the
diagnostics concerning autocorrelation,
heteroskedasticity, possible ARCH effects and
normality of the residuals. In general, the
model that minimizes AIC and SBC is selected as
the one with the optimal lag length. This model
should also pass all the diagnostic checks.

Step 3 Choosing the appropriate model regarding the

deterministic components in the multivariate
system.
Another important aspect in the formulation of
the dynamic model is whether an intercept
and/or trend should enter either the short-run
or the long-run model, or both models. The
general case of the VECM including all the
various options that can possibly happen is
given by the following equation:

∆Zt = 1∆Zt-1 + … + k-1∆Zt-p+1 + 1 (Z t 1 1 t) Zt-1

1

+ 2 + 2t + ut (37)
In general five distinct models can be
considered. Although the first and the fifth

75
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

model are not that realistic, we present all of

them for reasons of complementarity.
Model 1: No intercept or trend in CE (co-
integrating equation) or VAR ( 1 = 2 = 1 = 2

= 0).
Model 2: Intercept (no trend) in CE, no
intercept or trend in VAR ( 1 = 2 = 2 = 0).
Model 3: Intercept in CE and VAR, no trend in
CE and VAR ( 1 = 2 = 0).
Model 4: Intercept in CE and VAR, linear trend
in CE, no trend in VAR ( 2 = 0).
Model 5: Intercept and quadratic trend in CE,
intercept and linear trend in VAR.

Step 4 Determining the rank of or the number of co-

integrating vectors.
There are two methods for determining the
number of co-integrating relations, and both
involve estimation of matrix .
(1) One method tests the null hypothesis,
that Rank( ) = r against the hypothesis
that the rank is r+1. So, the null in
this case is that there are co-
integrating vectors and that we have up
to r co-integrating relationships, with
the alternative suggesting there is (r+1)
vectors.
The test statistics are based on the
characteristic roots (also called

76
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

eigenvalues) obtained from the estimation

procedure. The test consists of ordering
the largest eigenvalues in descending
order and considering whether they are
significantly different from zero. To
understand the test procedure, suppose we
obtained n characteristic roots denoted
by 1 > 2 > 3 > … > n. If the variables
under examination are not cointegrated,
the rank of is zero and all the
characteristic roots will equal zero.
Therefore, ( 1 ˆ ) will be equal to 1
i

and since ln(1) = 0. To test how many of

the numbers of the characteristic roots
are significantly different from zero,
this test uses the following statistic:
ˆ
max(r, r 1) T ln(1 r 1) (38)

As we said before, the test statistic is

based on the maximum eigenvalue and
because of that is called the maximal
eigenvalue statistic (denoted by max).

(2) The second method is based on a

likelihood ratio test about the trace of
the matrix (and because of that it is
called the trace statistic). The trace
statistic considers whether the trace is
increased by adding more eigenvalues
beyond the rth eigenvalue. The null
hypothesis in this case is that the
number of co-integrating vectors is less

77
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

than or equal to r. From the previous

analysis, it is clear that when all ˆ i =
0, the trace statistic is equal to zero
as well. This statistic is calculated by:
n
ˆ
trace(r) T ln(1 r 1) (39)
i r 1

Critical values for both statistics are

provided by Johansen and Juselius (1990).
These critical values are directly
provided from Eviews after conducting a
cointegration test.

Source: Asteriou (2007)

8.2.2 A Numerical Example

Remind that we reject the null hypothesis that r, the

number of cointegrating vectors, is less than k if the
test statistic is greater than the critical values
specified.

Table 8.3: Trace test

H0 H1 Statistic 95% Critical Decision

r = 0 r = 1 62.18 47.21 Reject H0

r 1 r = 2 19.55 29.68 Accept H0

r 2 r = 3 8.62 15.41 Accept H0

r 3 r = 4 2.41 3.76 Accept H0

78
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

We conclude that this data exhibits one cointegrating

vector.

8.2.3 The Johansen approach in Eviews

Eviews has a specific command for testing for

cointegration using Johansen approach under group
statistics. Consider the file JOHANSEN.wf1, which has
quarterly data for three macroeconomic variables: X, Y,
and Z.

Step 1: Determine the order of integration for the

variables.

To do this, we apply the unit-root tests on all three

variables. We apply the Doldado, Jenkinson and Sosvilla-
Rivero (1990) procedure for choosing the appropriate
model and we determine the number of lags according to
the SBC criterion.

Table 8.4: Integration of the variables at level

79
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Table 8.5: Integration of the variables at difference

80
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Step 2: Determine the optimal lag length

Unfortunately, Eviews does not allow us to automatically

detect the lag length (in cointegrating test equation),
so we need to estimate the model for a large number of
lags and then reduce down to check for the optimal value
of AIC and SBC. A rule of thumb is to choose the lag
length according to the data interval (year, quarter,
month). By doing this, we found that the optimal lag
length is 4 lags (quarterly data).

In Eviews, we can automatically determine the lag length

as the following steps.

81
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

(1) Estimate unrestricted VAR (using I(0) variables in

the VAR model, i.e. first differences) with the
defaut lag length (routinely 2).

(2) At the ‘VAR estimates’, select View/Lag

Structure/Lag Length Criteria

Figure 8.1: Lag length criteria

Step 3: Perform the test equation for cointegration with

optimal lag length determined above.

We test each one of the models for cointegration in

Eviews by opening Quick/Group Statistics/Cointegration
Test. Then in the series list window, we enter the names
of the series to check for cointegration, for example:

82
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

X Y Z

then press <OK>. The five alternative models explained in

Step 3 above are given under labels 1, 2, 3, 4, and 5.
There is another option (option 6 in Eviews) that
compares all these models together.

Figure 8.2: Johansen test specification

In our case, we wish to estimate models 2, 3, and 4

(because as noted earlier models 1 and 5 occur only very
rarely). To estimate model 2, we select that model, and
specify the number of lags in the bottom-right corner box
that has the (default by Eviews) numbers ‘1 2’ for
inclusion of two lags. We change the ‘1 2’ ro ‘1 4’ for
four lags, and click <OK> to get the results.

83
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Note that there is another box that allows us to include

(by typing their names) variables that will be treated as
exogenous. Here we usually put variables that are either
found to be I(0) or dummy variables that possibly affect
the behaviour of the model.

Doing the same for models 3 and 4 (in the untitled group
window select View/Cointegration Test) and simply change
the model by clicking next to 3 or 4. We get the results
as the following tables.

Table 8.6: Cointegration test results (model 2)

84
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Table 8.7: Cointegration test results (model 3)

Table 8.8: Cointegration test results (model 4)

Step 4: Decide which of the estimated models to choose in

testing for cointegration.

We can use ‘Option 6’ in Figure 8.2 to select the best

model to use in the VECM.

85
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Table 8.9: Johansen Cointegration Test Summary

From this summary, we see that Model 4 (Linear,

Intercept, and Trend) seems to be the best model (one
cointegrating vector).

8.3 Estimation of VECM in Eviews

After determining optimal lag length, and number of

cointegrating vectors, we start estimating the VECM:
(1) Quick/Estimate VAR
(2) In ‘VAR Specification’, choose Vector Error
Correction in ‘VAR type’, add all variables in
‘Endogeneous Variables’, then choose ‘Lag Length’.

86
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Figure 8.3: VAR specification

(3) We then enter number of cointegrating in

‘Cointegration’:
Figure 8.4: Cointegration

87
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Table 8.10: VECM estimates

9. CAUSALITY TESTS

According to Asteriou (2007), one of the good features of

VAR models is that they allow us to test the direction of
causality. Causality in econometrics is somewhat
different to the concept in everyday use (take
examples?); it refers more to the ability of one variable
to predict (and therefore cause) the other.

Suppose two stationary variables, say Yt and Xt, affect

each other with distributed lags. The relationship

88
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

between Yt and Xt can be captured by a VAR model. In this

case, it is possible to have that (a) Yt causes Xt
(Unidirectional Granger causality from Y to X), (b) Xt
causes Yt (Unidirectional Granger causality from X to Y),
(c) there is a bi-directional feedback (causality among
the variables), and finally (d) the two variables are
independent. The problem is to find an appropriate
procedure that allows us to test and statistically detect
the cause and effect relationship among variables.

Granger (1969) developed a relatively simple test that

defined causality as follows: a variable Yt is said to
Granger-cause Xt, if Xt can be predicted with greater
accuracy by using past values of the Yt variable rather
than not using such past values, all other terms
remaining unchanged. This test has been widely applied in
economic policy analysis.

9.1 The Granger Causality Test

The Granger causality test for the of two stationary

variables, say, Yt and Xt, involves as a first step the
estimation of the following VAR model:
p p
Yt = A1 + Cj Yt j + Dj Xt j + u1t (40)
j 1 j 1

p p
Xt = A2 + Ej Yt j + Fj Xt j + u2t (41)
j 1 j 1

where it is assumed that both u1t and u1t are uncorrelated

white-noise error terms, and Yt and Xt are integrated of

89
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

order 1. In this model, we can have the following

different cases:

Case 1 The lagged X terms in equation (40) are

statistically different from zero as a group, and
the lagged Y terms in equation (41) are not
statistically different from zero. In this case,
we have that Xt causes Yt.

Case 2 The lagged Y terms in equation (41) are

statistically different from zero as a group, and
the lagged X terms in equation (40) are not
statistically different from zero. In this case,
we have that Yt causes Xt.

Case 3 Both sets of lagged X and lagged Y terms are

statistically different from zero as a group in
equation (40) and (41), so that we have bi-
directional causality between Yt and Xt.

Case 4 Both sets of lagged X and lagged Y terms are

not statistically different from zero in equation
(40) and (41), so that Xt is independent of Yt.

The Granger causality test, then, involves the following

procedures. First, estimate the VAR model given by
equations (40) and (41). Then check the significance of
the coefficients and apply variable deletion tests first
in the lagged X terms for equation (40), and then in the
lagged Y terms in equation (41). According to the result
of the variable deletion tests, we may conclude about the

90
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

direction of causality based upon the four cases

mentioned above.

More analytically, and for the case of one equation (i.e.

we will examine equation (40)), it is intuitive to
reverse the procedure in order to test for equation (41),
and we perform the following steps:

Step 1 Regress Yt on lagged Y terms as in the

following model:
p
Yt A Cj Yt j u1t (42)
j 1

and obtain the RSS of this regression (which is

the restricted one) and label it as RSSR.
Step 2 Regress Yt on lagged Y terms plus lagged X
terms as in the following model:
p p
Yt A1 Cj Yt j Dj Xt j u1t (40)
j 1 j 1

and obtain the RSS of this regression (which is

the unrestricted one) and label it as RSSU.
Step 3 Set the null and alternative hypotheses as
below:
p
H0 : Dj 0 or Xt does not cause Yt
j 1

p
H1 : Dj 0 or Xt does cause Yt
j 1

Step 4 Calculate the F statistic for the normal Wald

test on coefficient restrictions given by:

91
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

(RSSR RSSU)/ p
F
RSSu /(N k)
where N is the included observations and k = 2p
+ 1 is the number of estimated coefficients in
the unrestricted model.
Step 5 If the computed F value exceeds the critical F
value, reject the null hypothesis and conclude
that Xt causes Yt.

Open the file GRANGER.wf1 and then perform as follows:

Figure 9.1: An illustration of GRANGER in Eviews20

20
However, this is not a good way of conducting Granger causality test (why?)

92
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Why I use the first differenced series?

Note that this ‘lag specification’ is not highly

appreciated in empirical studies (why?).

93
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Table 9.2: A result of the Granger causality tests

9.2 The Sims Causality Test

Sims (1980) proposed an alternative test for causality
making use of the fact that in any general notion of
causality, it is not possible for the future to cause the
present. Therefore, when we want to check whether a
variable Yt causes Xt, Sims suggests estimating the
following VAR model:
p p m
Yt A1 Cj Yt j Dj Xt j i Xt i u1t (43)
j 1 j 1 i 1

p p m
Xt A2 Ej Xt j Fj Yt j i Yt i u2t (44)
j 1 j 1 i 1

Assume that Yt and Xt are nonstationay [I(1)]. The new

approach here is that apart from lagged values of X and
Y, there are also leading values of X included in the

94
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

first equation (and similarly leading values of Y in the

second equation).
Examining only the first equation, if Yt causes Xt, then
we will expect that there is some relationship between Y
and the leading values of X. Therefore, instead of
testing for the lagged values of Xt, we test for
m
i 1 i 0 . Note that if we reject the restriction, then
the causality runs from Yt to Xt, and not vice versa,
since the future cannot cause the present.
To carry out the test, we simply estimate a model with no
leading terms (which is the restricted model) and then
the model as appears in (43), which is the unrestricted
model, and the obtain the F statistic as in the Granger
test above.
It is unclear which version of the two tests is
preferable, and most researchers use both. The Sims test,
however, using more regressors (due to the inclusion of
the leading terms), leads to a bigger loss of degrees of
freedom.

10. LAG LENGTH SELECTION CRITERIA

This section discusses statistical methods for choosing

the number of lags, first in an autoregression, then in a
time series regression model with multiple predictors.

10.1 Determining the Order of an Autoregression

According to Stock and Watson (2007), choosing the order

p of an autoregression requires balancing the marginal
benefit of including more lags against the marginal cost

95
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

of additional estimation uncertainty. On the one hand, if

the order of an estimated autoregression is too low, you
will omit potentially valuable information contained in
the more distant lagged values. On the other hand, if it
is too high, you will be estimating more coefficients
than necessary, which in turn introduces additional
estimation error into your forecasts. Various statistical
methods can be used, but two most important ones are SIC
and AIC21.

The SIC. A way around this problem is to estimate p by

minimizing an ‘information criterion’. One such
information is the Schwarz Information Criterion (SIC),
which is:

RSS(p) ln T
SIC(p) ln (p 1) (45)
T T

where RSS(p) is the sum of squared residuals of the

estimated AR(p). The SIC estimator of p, p̂ , is the value
that minimizes SIC(p) among the possible choices p = 0,
1, …, pmax, where pmax is the largest value of p
considered.

The formula for SIC might look a bit mysterious at first,

but it has an intuitive appeal. Consider the first term
in equation (45). Because the regression coefficients are
estimated by OLS, the sum of squared residuals
necessarily decreases (or at least does not increase)

21
Others including FPE, HQ, and LR are also used in empirical studies.

96
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

when you add a lag. In contrast, the second term is the

number of estimated regression coefficients (the number
of lags, p, plus one for the intercept) times the factor
(lnT)/T. This second term increases when you add a lag.
The SIC trades off these two forces so that the number of
lags that minimizes the SIC is a consistent estimator of
the true lag length.

The AIC. The SIC is not the only information criterion;

another is the Akaike Information Criterion (AIC), which
is:

RSS(p) 2
AIC(p) ln (p 1) (46)
T T

The difference between the AIC and the SIC is that the
term “lnT” in the SIC is replaced by “2” in the AIC, so
the second term in the AIC is smaller (why?). Stock and
Watson (2007) state that the second term in the AIC is
not large enough to ensure that the correct lag length is
chosen, even in large samples, so the AIC estimator of p
is not consistent.

Despite this theoretical shortcoming, the AIC is widely

used in practice. If you are concerned that the SIC might
yield a model with two few lags, the AIC provides a
reasonable alternative.

How to choose the optimal lag length in Eviews?

97
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Figure 10.1: Lag length of a VAR: Step-by-step

98
TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

99
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

10.2 Lag Length in Multiple-Predictor Regression

As in an autoregression, the SIC and AIC can be used to

estimate the number of lags and the variables in the time
series regression model with multiple predictors. After
we determine (and fix) the optimal lags for the
autoregression, we then continue add successive lags of
explanatory variables. If the regression model has K
coefficients (including the intercept), the SIC and AIC
are defined as:

RSS(K) ln T
SIC(K) ln K (47)
T T

RSS(K) 2
AIC(K) ln K (48)
T T

100
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

There are two important practical considerations when

using an information criterion to estimate the lag
length. First, as is the case for the autoregression, all
the candidate models must be estimated over the same
sample. Second, when there are multiple predictors, this
approach is computationally demanding because it requires
many different models. According to Stock and Watson
(2007), in practice, a convenient shortcut is to require
all the regressors to have the same number of lags, that
is, to require that p = q1 = … = qk. This does, therefore,
not provide the best lag length structure.

11. BOUNDS TEST FOR COINTEGRATION

Another way to test for cointegration and causality is

the Bounds Test for Cointegration within ARDL modelling
approach. This model was developed by Pesaran et al.
(2001) and can be applied irrespective of the order of
integration of the variables (irrespective of whether
regressors are purely I(0), purely I(1) or mutually
cointegrated). This is specially linked with the ECM
models and called VECM as specified in Section 5.

11.1 The Model

The ARDL modelling approach involves estimating the

following error correction models:
m m
Yt 0y 1yYt 1 2y Xt 1 i Yt i j Xt j u1t (49)
i 1 j 1

m m
Xt 0x 1xYt 1 2xXt 1 i Xt i j Yt j u2t (50)
i 1 j 1

101
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

11.2 Test Procedure

Suppose we have Yt and Xt are nonstationary.

Step 1: Testing for the unit root of Yt and Xt

(using either DF, ADF, or PP tests)

Suppose the test results indicate that Yt and Xt have

different orders of integration [I(0) and/or I(1)].

Step 2: Testing for cointegration between Yt and Xt

(using Bounds test approach)

For equations 1 and 2, the F-test (normal Wald test) is

used for investigating one or more long-run
relationships. In the case of one or more long-run
relationships, the F-test indicates which variable should
be normalized.

In equation (49), when Y is the dependent variable, the

null hypothesis of no cointegration is H0: 1y = 2y = 0
and the alternative hypothesis of cointegration is H1: 1y

2y 0.

On the other hand, in equation (50), when X is the

dependent variable, the null hypothesis of no
cointegration is H0: 1x = 2x = 0 and the alternative
hypothesis of cointegration is H1: 1x 2x 0.

In this model, we still apply the Granger causality test

following the same procedure presented above.

102
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

12. SUGGESTED RESEARCH TOPICS

From previous studies, I would like to suggest the

following topics that you can consider for your coming
research proposal.

Saving, Investment and Economic Development

 An analysis of the interaction among savings,

investments and growth in Vietnam

 Are saving and investment cointegrated? The case of

Vietnam

 Causal relationship between domestic savings and

economic growth: Evidence from Vietnam

 Does saving really matter for growth? Evidence from

Vietnam

 The relationship between savings and growth:

Cointegration and causality evidence from Vietnam

 The saving and investment nexus for Vietnam: Evidence

from cointegration tests

 Do foreign direct investment and gross domestic

investment promote economic growth?

 Foreign direct investment and economic growth in

Vietnam: An empirical study of causality and error
correction mechanisms

 The interactions among foreign direct investment,

economic growth, degree of openness and unemployment in
Vietnam

103
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Trade and Economic Development

 How trade and foreign investment affect the growth: A

case of Vietnam?

 Trade, foreign direct investment and economic growth in

Vietnam

 A cointegration analysis of the long-run relationship

between black and official foreign exchange rates: The
case of the Vietnam dong

 An empirical investigation of the causal relationship

between openness and economic growth in Vietnam

 Export and economic growth in Vietnam: A Granger

causality analysis

 Export expansion and economic growth: Testing for

cointegration and causality for Vietnam

 Is the export-led growth hypothesis valid for Vietnam?

 Is there a long-run relationship between exports and

imports in Vietnam?

 On economic growth, FDI and exports in Vietnam

 Trade liberalization and industrial growth in Vietnam:

A cointegration analysis

Stock Market and Economic Development

 Causality between financial development and economic

growth: An application of vector error correction to
Vietnam

104
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

 Financial development and the FDI growth nexus: The

Vietnam case

 Macroeconomic environment and stock market: The Vietnam

case

 The relationship between economic factors and equity

market in Vietnam

 Modelling the linkages between the US and Vietnam stock

markets

 The long-run relationship between stock returns and

inflation in Vietnam

 The relationship between financial deepening and

economic growth in Vietnam

 Testing market efficient hypothesis: The Vietnam stock

market

 Threshold adjustment in the long-run relationship

between stock prices and economic activity

Energy and the Economy

 The dynamic relationship between the GDP, imports and

domestic production of crude oil: Evidence from Vietnam

 Causal relationship between gas consumption and

economic growth: A case of Vietnam

 Causal relationship between energy consumption and

economic growth: The case of Vietnam

105
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

 Causality relationship between electricity consumption

and GDP in Vietnam

 The causal relationship between electricity consumption

and economic growth in Vietnam

 A cointegration analysis of gasoline demand in Vietnam

 Cointegration and causality testing of the energy-GDP

relationship: A case of Vietnam

 Does more energy consumption bolster economic growth?

 Energy consumption and economic growth in Vietnam:

Evidence from a cointegration and error correction
model

 The causality between energy consumption and economic

growth in Vietnam

 The relationship between the price of oil and

macroeconomic performance: Empirical evidence for
Vietnam

Fiscal Policy and Economic Development

 A causal relationship between government spending and

economic development: An empirical examination of the
Vietnam economy

 Economic growth and government expenditure: Evidence

from Vietnam

 Government revenue, government expenditure, and

temporal causality: Evidence from Vietnam

106
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

 The relationship between budget deficits and money

demand: Evidence from Vietnam

Monetary Policy and Economic Development

 Granger causality between money and income for the

Vietnam economy

 Money, inflation and casuality: Evidence from Vietnam

 Money-output Granger causality: An empirical analysis

for Vietnam

 Time-varying parameter error correction models: The

demand for money in Vietnam

 Monetary transmission mechanism in Vietnam: A VAR

analysis

Tourism and Economic Development

 Cointegration analysis of quarterly tourism demand by

international tourists: Evidence from Vietnam

 Does tourism influence economic growth? A dynamic panel

data approach

 International tourism and economic development in

Vietnam: A Granger causality test

 Tourism demand modelling: Some issues regarding unit

roots, co-integration and diagnostic tests

 Tourism, trade and growth: the case of Vietnam

107
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Agriculture and Economic Development

 Dynamics of rice prices and agricultural wages in

Vietnam

 Macroeconomic factors and agricultural production

linkages: A case of Vietnam

 Is agriculture the engine of growth?

 The causal relationship between fertilizer consumption

and agricultural productivity in Vietnam

 Macroeconomics and agriculture in Vietnam

Others

 Hypotheses testing concerning relationships between

spot prices of various types of coffee

 The relationship between wages and prices in Vietban

 An error correction model of luxury goods expenditures:

Evidence from Vietnam

 The relationship between macroeconomic variables and

housing price index: A case of Vietnam

 Explaining house prices in Vietnam

 Long-term trend and short-run dynamics of the Vietnam

gold price: an error correction modelling approach

 Macroeconomic adjustment and private manufacturing

investment in Vietnam: A time-series analysis

108
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

 Testing for the long run relationship between nominal

interest rates and inflation using cointegration
techniques

 The long-run relationship between house prices and

income: Evidence from Vietnam housing markets

It is noted that many empirical studies use the

nonstationary panels, typically characterised by panel
unit root tests and panel cointegration tests. However,
they are beyond the scope of this lecture.

109
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

REFERENCES
Asteriou, D. and Hall, S.G. (2007) Applied Econometrics:
A Modern Approach Using Eviews and Microfit, Revised
Edition. Palgrave Macmillan.
Cheung, Y.W. and Lai, K.S. (1995) ‘Lag Order and Critical
Values of the Augmented Dickey-Fuller Test’, Journal of
Business & Economic Statistics, Vol.13, No.3, pp.277-
280.
Dickey, D.A. and Fuller, W.A. (1979) ‘Distribution of the
Estimators for Autoregressive Time Series with a Unit
Root’, Journal of the American Statistical Association,
Vol.74, No.366, pp.427- 431.
Dickey, D.A. and Fuller, W.A. (1981) ‘Likelihood Ratio
Statistics for Autoregressive Time Series with a Unit
Root’, Econometrica, Vol.49, p.1063.
Diebold, F.X. (2004) Elements of Forecasting, 3rd Edition,
Thomson.
Dolado, J., T.Jenkinson and S.Sosvilla-Rivero. (1990)
‘Cointegration and Unit Roots’, Journal of Economic
Surveys, Vol.4, No.3.
Durbin, J. (1970) ‘Testing for Serial Correlation in
Least Squares Regression When Some of the Variables Are
Lagged Dependent Variables’, Econometrica, Vol.38,
pp.410-421.
Engle, R.F. and Granger, C.W.J. (1987) ‘Co-integration
and Error Correction Estimates: Representation,
Estimation, and Testing’, Econometrica, Vol.55, p.251–
276.

110
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Granger, C.W.J. (1981) ‘Some Properties of Time Series

Data and Their Use in Econometric Model Specification’,
Journal of Econometrics, Vol.16, pp.121-130.
Granger, C.W.J. (1988) ‘Some Recent Developments in the
Concept of Causality’, Journal of Econometrics, Vol.39,
pp.199-211.
Granger, C.W.J. (2004) ‘Time Series Analysis,
Cointegration, and Applications’, The American Economic
Review, Vol.94, No.3, pp.421-425.
Granger, C.W.J. and Newbold, P. (1977) ‘Spurious
Regression in Econometrics’, Journal of Econometrics,
Vol.2, pp.111-120.
Griffiths, W.E., R.C.Hill and G.C.Lim. (2008) Using
rd
Eviews for Principles of Econometrics, 3 Edition, John
Wiley & Sons.
Gujarati, D.N. (2003) Basic Econometrics, 4th Edition,
McGraw-Hill.
Gujarati, D.N. (2011) Econometrics by Example, 1st
Edition, Palgrave Macmillan.
Hanke, J.E. and Wichern, D.W. (2005) Business
Forecasting, 8th Edition, Pearson Education.
Holton, W.J. and Keating, B. (2007) Business Forecasting
With Accompanying Excel-Based ForecastXTM Software, 5th
Edition, McGraw-Hill.
Johansen, S. (1991) ‘Estimation and Hypothesis Testing of
Cointegration Vectors in Gaussian Vector Autoregressive
Models’, Econometrica, Vol.59, pp.1551-1580.
Johansen, S. and Juselius, K. (1990) ‘Maximum Likelihood
Estimation and Inference on Cointegration, with

111
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Applications for the Demand for Money’, Oxford Bulletin

of Economics and Statistics, Vol.52, pp.169-210.
Kang, H. (1985) ‘The Effects of Detrending in Granger
Causality Tests’, Journal of Business & Economic
Statistics, Vol.3, No.4, pp.344-349.
Katircioglu, S. (2009) ‘Tourism, Trade and Growth: The
Case of Cyprus’, Applied Economics, Vol.41, pp.2741-
2750.
Li, X. (2001) ‘Government Revenue, Government
Expenditure, and Temporal Causality: Evidence from
China’, Applied Economics, Vol.33, pp.485-497.
Ljung, G.M. and Box, G.E.P. (1978) ‘On a measure of Lack
of Fit in Times Series Models’, Biometrica, Vol.65,
pp.297-303.
Mackinnon, J.G. (1994) ‘Approximate Asymptotic
Distribution Functions for Unit-Root and
Cointegration’, Journal of Business & Economic
Statistics, Vol.12, No.2, pp.167-176.
Mackinnon, J.G. (1996) ‘Numerical Distribution Functions
for Unit Root and Cointegration Tests’, Journal of
Applied Econometrics, Vol.11, No.6, pp.601-618.
Mackinnon, J.G., Alfred A. Haug and Leo Michelis. (1999)
‘Numerical Distribution Functions of Likelihood Ratio
Tests for Cointegration’, Journal of Applied
Econometrics, Vol.14, No.5, pp.563-577.
Mahmoud, E. (1984) ‘Accuracy in Forecasting: A Survey’,
Journal of Forecasting, Vol.3, pp.139-159.
McNees, S. (1986) ‘Forecasting Accuracy of Alternative
Techniques: A Comparison of US Macroeconomic

112
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Forecasts’, Journal of Business and Economic

Statistics, Vol.4, pp.5-15.
Mehrara, M. (2007) ‘Energy-GDP Relationship for Oil-
Exporting Countries: Iran, Kuwait and Saudi Arabia’,
OPEC Review, Vol.3.
Nguyen Trong Hoai, Phung Thanh Binh, and Nguyen Khanh
Duy. (2009) Forecasting and Data Analysis in Economics
and Finance, Statistical Publishing House.
Phillips, P.C.B. (1987) ‘Time Series Regression with a
Unit Root’, Econometrica, Vol.55, No.2, pp.277-301.
Phillips, P.C.B. (1998) ‘New Tools for Understanding
Spurious Regressions’, Econometrica, Vol.66, No.6,
pp.1299-1325.
Phillips, P.C.B. and Perron, P. (1988) ‘Testing for a
Unit Root in Time Series Regression’, Biometrica,
Vol.75, No.2, pp.335-346.
Pindyck, R.S. and Rubinfeld, D.L. (1998) Econometric
th
Models and Economic Forecasts, 4 Edition, McGraw-Hill.
Ramanathan, R. (2002) Introductory Econometrics with
th
Applications, 5 edition, Harcourt College Publisher.
Sims, C.A. (1980) ‘Macroeconomics and Reality’,
Econometrica, Vol.48, No.1, pp.1-48.
Stock, J.H. and Watson, M.W. (2007) Introduction to
Econometrics, 2nd Edition, Pearson Education.
Studenmund, A.H. (2001) Using Econometrics: A Practical
Guide, 4th Edition, Addison Wesley Longman.

Toda, H.Y. and Yamamoto, T. (1995) ‘Statistical Inference

in Vector Autoregressive with Possibly Integrated

113
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Processes’, Journal of Econometrics, Vol.66, No.1,

pp.225-250.

Vogelvang, B. (2005) Econometrics: Theory and

Applications with Eviews, Pearson Education.
Wooldridge, J.M. (2003) Introductory Econometrics: A
Modern Approach, 2nd Edition, Thomson.

114
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

APPENDIX
STATA COMMANDS
Source: Practical exercises, Advanced Econometrics course
2012, Wageningen University, The Netherlands.

EXAMPLE 1
Use the data set WHEATOIL.dta, which contains (nominal)
prices of wheat (pwht), nominal oil prices (poil) and a
time indicator (t). The data are monthly and available
for the period Jan 1990 till December 2008 (19 years*12
months = 228 obs.). In this example, we will investigate
whether there is a long-run relationship between wheat
prices and oil prices. There may be all kinds of reasons
for such a relationship: oil is an important input in
fertilizer production, is used for applying machinery,
drives transportation costs, etc.
Please declare the data to be time-series data using the
following command:
tsset t

1. Create one graph with line plots of both pwht and poil
against t. Considering the line plot for pwht, do you
think this variable is stationary? Motivate your
answer.

twoway (tsline pwht) (tsline poil, yaxis(2))

115
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

100
300

80
250
pwht

poil
200

60
150

40
100

20
0 50 100 150 200 250
t

pwht poil

The graphs for both pwht and poil indicate that there are
stochastic trends (means are not constant) and their
variances are also not constant. For the pwht, it first
increases and highly fluctuates (from observation 1 to
about 70), followed by a declining period (from
observation about 70 to about 120) with less fluctuation,
then it tends to increase and especially decline very
quickly in the last months. Therefore, we might say that
pwht is not stationary.

2. Use appropriate tests to find out the orders of

integration of both pwht and poil?

116
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

In order to check the order of integration for pwht we

perform the Augmented Dickey Fuller (ADF) test and the
KPSS test on pwht until finding a stationary time series.

2.1 The pwht series.

ADF test

H0: The pwht series is non-stationary (the pwht series has

a unit root)

As this is a monthly series, we starts with 12 lags. In

addition, we include the trend in the test equation.

117
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Because the coefficients of trend and lag 12 are not

statistically significant, so we can remove these in the
test equation.

If we choose the 5% significance level, the coefficients

of lag 8 to lag 11 are not significant. Therefore, we try
the test equation with 7 lags.

118
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

As the absolute value of the test statistics (2.579) is

smaller than the absolute value of 5% critical value
(2.882), we cannot reject the null hypothesis at 5%
significance level. Therefore, ADF test suggests that
pwht series is not stationary. To be sure, we apply the
KPSS test (with H0: the pwht series is stationary).

119
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

KPSS test

All test statistics are greater than the 5% critical

values, so we reject the null hypothesis. That means the
pwht series is non-stationary.

We now examine stationarity of the first-differenced

series of pwht without the constant term in the test
equation because there is no trend in the original series
of pwht. Here is the test result.

ADF test

H0: The first-difference of pwht is not stationary.

120
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

As the absolute value of the test statistics (4.796) is

larger than the 5% critical value (1.95), we reject the
null hypothesis. That means the first-differenced series
of pwht is stationary. We now examine the KPSS test for
this first-differenced series.

KPSS test

H0: The first-differenced series of pwht is stationary.

121
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

The KPSS test results indicate that we fail to reject the

null hypothesis.

In conclusion, the pwht series is integrated of order one

[I(1)].

2.2 The poil series.

We proceed in the same way as in the case of pwht.

First I perform the ADF test and the KPSS test on the
values of poil.

ADF test

H0: The poil series is not stationary.

122
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Because the lag 11 and lag 12 are not significant, so we

remove them in the test equation.

123
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

As the absolute value of the test statistics (3.37) is

smaller than the 5% critical value (3.43), we cannot
reject the null hypothesis. This implies that the poil
series is not stationary. To be sure, we apply the KPSS
test.

KPSS test

H0: The poil series is stationary.

All test statistics are greater than the critical values

(even at 1% significance level), so we reject the null
hypothesis. That means the poil series is non-stationary.

We now examine stationarity of the first-differenced

series of poil with the constant term in the test
equation because there is trend in the original series of
poil. Here is the test result.

124
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

ADF test

H0: The first-difference of poil is not stationary.

As the absolute value of the test statistics (4.166) is

larger than the 5% critical value (2.882), we reject the
null hypothesis. The ADF test indicates that the first-
differenced series of poil is stationary. We now confirm
this by applying the KPSS test.

KPSS test

H0: The first-difference of poil is stationary.

125
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

The KPSS test results indicate that the first-differenced

series of poil is stationary.

Therefore, the poil series is integrated of order one

[I(1)].

3. Given your findings at question 2, consider whether it

is possible whether there exists a long-run
(cointegrating) relationship between pwht and poil.

As both series are integrated of order one, there could

exist a long-run relationship between pwht and poil.We
must apply the cointegration tests to see whether there
is really a long-run (or cointegrating) relationship
between them.

126
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

4. Estimate the potential long-run relationship with pwht

as dependent variable and poil as explanatory variable.
Does this regression have the typical characteristics
of a spurious regression or as a cointegrating
relationship.

The OLS estimation results seem to be spurious because of

the following signals: The t-ratio is very high, while
the Durbin-Watson test statistic is very small (0.103).
The graph of residuals from this regression (see below)
show that the residuals seem to be non-stationary. The R2
is low (0.213); this is not really a phenomenon of
spurious regression. This can be a signal of positive
autocorrelation. To be sure, we must apply the
statistical tests.

127
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]
150
100
Residuals

50
0
-50

0 50 100 150 200 250

t

5. Investigate whether pwht and poil are cointegrated

using two different tests. What do you conclude? How
do you judge the estimation results of question 4: as
a spurious regression or as a cointegration
relationship.

We will apply two different tests: (i) residual-based

test for no cointegration; and (ii) CRDW test for no
cointegration. Both tests check for the cointegration
between poil and pwht. poil and pwht are cointegrated if
the residuals of the above estimated model ( are
stationary process.

128
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

5.1 Residual-based test for no cointegration (Engle-

Granger approach)

First test: Dickey Fuller test on the residuals

Null Hypothesis: the residuals are not stationary (no

cointegration)

As coefficients of trend and lag 12 are not significant,

so we remove them from the test equation.

129
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

The test results indicate that the residuals seem to be

not stationary at 5% significance level. However, the
constant in this test equation is not significant, so we
try to remove it from the test equation.

130
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

The test equation without the constant term shows that

the residuals become stationary even at 1% significance
level. This confused results might be due to the less
power of test of the ADF test. To avoid this, we now
apply the KPSS test.

131
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

The KPSS test results we reject the null hypothesis that

the residuals are stationary. Therefore, there seems to
be no-cointegration between pwht and poil.

132
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

5.2 CRDW test for no cointegration

The Durbin-Watson test statistic is 0.103, which is

smaller than the 5% critical value CRDW tests for no
cointegration (~ 0.2, about 200 observations, 2
variables, Table 9.3, Verbeek, 2012). Therefore, we fail
to reject the null hypothesis that the residuals is non-
stationary. In other words, pwht and poil are not
cointegrated.

In conclusion, there is no long-run relationship between

pwht and poil. Therefore, the OLS estimation regression
between pwht and poil is likely to be spurious
regression.

5. Based on your findings at the previous questions,

estimate a simple VAR or VECM (this choice depends upon
finding at question 5). Specify only one lag. Based on
this estimated model indicate how pwht and poil relate.

133
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Because pwht and poil are not cointegrated, so we cannot

apply the VECM model. It is just possible to use the VAR
model for the first-differenced series of pwht and poil.

The VAR model results indicate that the p-values of

coefficients of the variable poilLD. in the first
equation (0.202) and of the variable pwhtLD in the second
equation (0.97) are very high. These suggest that neither

134
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

poil affects pwht, nor pwht affects poil. However, the

cofficients of pwhtLD in the first equation, and poilLD.
in the second equation are highly significant. These
indicate that the first-differenced series follow the AR
process.

EXAMPLE 2

In this example, we analyse the dataset TEXASHOUSING.dta

with monthly housing prices in four major cities in Texas
(USA): Austin, Dallas, Houston and San Antonio. Natural
logarithms of housing prices are available from January
1990 till December 2003 (168 observations). It is
expected that there are regional linkages between these
housing markets. If houses get very expensive in one
city, people may decide to move to another city, creating
upward pressure on housing prices in our cities. In other
words it is assumed that there exist a long-run (spatial)
equilibrium between these four housing prices series.
That is what we will investigate here.

1. Investigate for all four prices series the order of

integration.

We only report the Augment Dickey Fuller test and KPSS

test of the last step.

We recall that:
Null hypothesis in Augmented Dickey Fuller test: time
series is not stationary

Null Hypothesis in KPSS test: the time series is

stationary

135
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

AUSTIN

As the absolute value of the test statistics (1.554) is

smaller than the 5% critical value (2.886), we cannot
reject the null hypothesis. The time series is not
stationary.

136
TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

137
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

The ADF test results indicate that housing price in

Austin is non-stationary. However, the KPSS test
indicates the non-stationarity (at 5% significance level)
up to 5 lags only. When trying to use KPSS test without
trend, the KPSS results turn out to provide a strong
evidence of non-stationarity.

We now test the stationarity of the first-difference of

the housing price in Austin.

As the absolute value of the test statistics (10.964) is

larger than the 5% critical value (2.886), we reject the
null hypothesis. The time series of first differences is
stationary.

138
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Both ADF and KPSS tests indicate that the first-

differenced data of housing price in Austin is
stationary. Therefore, the housing price in Austin is
integrated of order on [I(1)].

139
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

DALLAS

As the absolute value of the test statistics (1.792) is

smaller than the 5% critical value (3.442), we cannot
reject the null hypothesis. The time series is not
stationary.

140
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Both ADF and KPSS tests indicate that the housing price
in Dallas is non-stationary. We now test the stationarity
of its first-differenced data.

141
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

As the absolute value of the test statistics (5.683) is

smaller than the 5% critical value (2.886), we reject the
null hypothesis. The time series is stationary.

142
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Both ADF and KPSS tests indicate that the first-

differenced data of housing price in Dallas is
stationary. Therefore, the housing price in Dallas is
integrated of order on [I(1)].

143
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

HOUSTON

144
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Both ADF and KPSS tests indicate that the housing price
in Houston is non-stationary. We now test the
stationarity of its first-differenced data.

As the absolute value of the test statistics (8.323) is

larger than the 5% critical value (2.886), we reject the
null hypothesis. The time series of first differences is
stationary.

145
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Both ADF and KPSS tests indicate that the first-

differenced data of housing price in Houston is
stationary. Therefore, the housing price in Houston is
integrated of order on [I(1)].

146
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

SAN ANTONIO

147
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

The ADF test results indicate that housing price in San

Antonio is non-stationary. However, the KPSS test

148
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

indicates the non-stationarity (at 5% significance level)

up to 5 lags only. When trying to use KPSS test without
trend, the KPSS results turn out to provide a strong
evidence of non-stationarity. We now test the
stationarity of the first-difference of the housing price
in San Antonio.

149
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

Both ADF and KPSS tests indicate that the first-

differenced data of housing price in San Antonio is
stationary. Therefore, the housing price in San Antonio
is integrated of order on [I(1)].

In conclusion, all housing prices in these cities are

integrated of the same order one. Therefore, there could
be cointegrating relationships among these housing
prices.

150
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

2. For the variables that have the same order of

integration, use Johansen’s rank and maximum
eigenvalues tests to investigate the rank of the
cointegrating matrix. Set the number of lags to 3. The
maximum eigenvalues test can be obtain at the reporting
tab. What do you conclude from these tests?

All the variables have the same order of integration (1),

so they could be all cointegrated. In order to check
that we perform two cointegration tests (Johansen tests
for cointegration). Both test check the rank of the
cointegrating matrix (matrix that contains the
coefficients of cointegrating relationships). The number
of cointegration relationships is equal to the rank of
the matrix. The number of cointegration relationships
cannot be larger than the number of variables minus one
(in this case cannot be larger than 3).

Null hypothesis in both test: r≤r0 where r is the rank of

the matrix of cointegrating relationships.

151
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

In the trace test I reject the null hypothesis up to the

rank of two (test statistics 9.88 smaller than 15.41).
The eigenvalues test rejects the null hypothesis up to
the rank of 2 (test statistics 9.55 smaller than 14.07).
So I conclude that the rank of the cointegrating matrix
is two; This implies that there exist two long-run
relationships among housing prices in these cities.

3. Image that all variables are I(0). Explain how this

would be reflected in the output of Johansen’s rank
and maximum eigenvalues tests.

If four variables were stationary at their original data

[I(0)], the cointegrating matrix would have full rank (r0
= 4). In the output of Johansen’s rank and maximum
eigenvalue tests, we should have found a rank of 4.

4. Estimate a VECM with the appropriate number of

cointegrating relations (again set lag at 3). What
are the long-run cointegrating relationships? Why are
some of the adjustment parameters not significant for
these cointegrating relations?

152
TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

153
TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

154
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

From the cointegrating equations results, (based on the

significance of the estimated coefficients) we realize
that there are two long-run cointegrating relationships
between/among house prices of: (i) Austin and San
Antonio; and (ii) Dallas, Houston, and San Antonio.

The adjustment parameters in the VECM model can be

summarized as followed:

Adj.Parameter Coef. P-value Significance

D_austin 11 -0.148 0.012 Yes

12 -0.040 0.742 No

D_dallas 21 0.073 0.128 No

155
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

22 -0.309 0.002 Yes

D_houston 31 0.190 0.000 Yes

32 0.604 0.000 Yes

D_sa 41 0.283 0.000 Yes

41 -0.178 0.185 No

 For Austin: The adjustment parameter of the second

cointegrating relation is not significant because
Austin is omitted in this relation (see _ce2 in
cointegrating equations).
 For Dallas: The adjustment parameter of the first
cointegrating relation is not significant because
Dallas is omitted in this relation (see _ce1 in the
cointegrating equations).
 For Houston: Both adjustment parameters are highly
significant because Houston exists in both relations
(see _ce1 and _ce2 in the cointegrating equations).
 For San Antonia: The adjustment parameter of the
second cointegrating relation is not significant
(although it is included in both the cointegrating
equations) because of lag selection (maybe). Say, when
we change from lag(3) to lag(4), both adjustment
parameters become significant at 5% significance
level.

156
 TOPICS IN TIME SERIES ECONOMETRICS
Phùng Thanh Bình
[email protected]

vec austin dallas houston sa, trend(constant) rank(2) lag(4)

If the lags are 4, we can see that there are two long-run
cointegrating relationships between/among house prices
of: (i) Austin, Houston and San Antonio; and (ii) Dallas,
Houston, and San Antonio.

157