Lecture 7 VAR, VECM and Multivariate Cointegration

VAR, Multivariate Cointegration and
Vector-error Correction Model

(VECM)
The School of Management

10 – 11 Nov 2014 1
Preview
2
Law, S.H., Department of Economics, UPM
Learning Objectives
1. Describe the VAR model and Causality Test
3
VAR Models
• Quite common in economics to have models

where some variables are not only explanatory
variables for a given dependent variable; but
also explained by 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 endogenous and which are
exogenous or predetermined variables.

VAR Models
• Single equations models of the form y = X + u
• All of the variables contained in the X matrix are

assumed to be Exogenous.
• y is an Endogenous variable.
• Predetermined variables are variables that were
determined prior to the current period.
– The current period error term is uncorrelated with current and
lagged values of the predetermined variable but may be
correlated with future values

VAR Models
Sims (1980) suggests:

• if there is simultaneity among a number of
variables, then all these variables should be
treated in the same way.
• So, all variables are treated as endogenous.
• This means that in its general reduced form
each equation has the same set of
regressors.

VAR Models
• For example, time series yt that is affected by current

and past values of xt and, simultaneously, the time
series xt to be a series that is affected by current and
past values of the yt series.
• In this case, simple bivariate model is given by:
yt = β10 − β12xt + γ11yt−1 + γ12xt−1 + uyt
xt = β20 − β21yt + γ21yt−1 + γ22xt−1 + uxt
• where uit is an iid disturbance term with E(uit)=0, i=y,x;
E(uyt uxt)=0.

VAR Models
This is a first-order VAR model, because the

longest lag length is unity.
Rewriting the system with matrix algebra:
1 12   yt   10   11  12   yt 1  u yt 
          
 21 1   xt    20   21  21   xt 1  u xt 

VAR Models
Advantages of VAR models:
(a) Very simple (no worry about which variables
are endogenous or exogenous)
(b) Estimation is very simple (usual OLS)
(c) Forecasts from VAR models are better than
those obtained from far more complex
simultaneous equation models

VAR Models
Disadvantages of VAR models:
(a) A-theoretic as they are not based on any
economic theory ‘everything causes everything’
(resolved by statistical inference and causality
tests)
(b) Loss of degrees of freedom
(c) Obtained coefficients of VAR models are
difficult to interpret as they lack theoretical
background (overcome by impulse response
functions)

Causality Tests
Suppose two variables, say yt and xt , affect each other

with distributed lags.
The relationship between those variables can be
captured by a VAR model.
In this case it is possible to have that:
(a) yt causes xt
(b) xt causes yt
(c) there is bi-directional feedback (causality among
variables)
(d) the two variables are independent

Causality Tests
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.

Causality Tests
First step is to estimate following VAR model:

n m
yt  a1    i xt i    j yt  j  e1t
i 1 j 1
n m
xt  a2    i xt i    j yt  j  e2t
i 1 j 1

Causality Tests
• Case 1 Lagged x terms in (1) may be statistically different from
zero as a group, and lagged y terms in (2) not statistically different
from zero. In this case we have that xt causes yt.
• Case 2 Lagged y terms in (1) may be statistically different from
zero as a group, and lagged x terms in (1) not statistically different
from zero. In this case we have that yt causes xt.
• Case 3 Both sets of x and y terms are statistically different from
zero in (1) and (2), so that have bi-directional causality.
• Case 4 Both sets of x and y terms are not statistically different from
zero in (1) and (2), so that xt is independent of yt.

Causality Tests
Step 1 Regress yt on lagged y terms and obtain 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 and

obtain RSS of this regression (which now is the
unrestricted one) and label it as RSSU
Step 3 Set the null and the alternative hypotheses:

Ho: coefficients of the lagged terms of x are equal to zero, or x−/→y
Ha: coefficients of the lagged terms of x are not equal to zero, or x→y

Causality Tests
Step 4 Calculate F statistic for normal Wald test on

coefficient restrictions
Step 5 If computed F value exceeds

F-critical value, reject the null hypothesis and
conclude that xt causes yt

Multivariate Cointegration Test
17
More Practical Issues in Cointegration
Testing
• Note: we have focussed on two variables, Y and X.
• In practice, you may have many more variables.
• Example: consider the three variables: income (Y),
consumption (C) and investment (I).
• Some macroeconomists claim that the ratios C/Y and I/Y
are roughly stable in the long-run.
• Common to take logs, so
ln(C) – ln(Y)  constant
and
ln(I) – ln(Y)  constant
There are two separate tests for cointegration, which can give
different results.
18
Testing
• If ln(C), ln(Y) and ln(I) all contain unit roots,
reasoning above suggests that two cointegrating
relationships might occur.
• The variables in the model might form several
equilibrium relationships governing the joint evolution
of all the variables.
• In general, for n number of variables we can have only

up to n – 1 cointegrating vectors
• When n = 2, if cointegration exists then the
cointegrating vector is unique
19
Testing
• [Note: Engle-Granger test (based on a cointegrating
regression involving all three variables), would only find
whether cointegration is/is not present (not tell you how
many cointegrating relationships)]
• What should you do in this case? One option is to use

the Johansen cointegration test for multiple equations
• Using the Johansen Maximum Likelihood (ML)

procedure, it is possible to obtain more then a single
cointegrating relationship.
20
The Johansen ML Cointegration
• Given that Johansen cointegration is a maximum
likelihood based test (Engle-Granger is OLS
based), it requires a large sample.
• The Johansen multivariate cointegration test is
based on a VAR, not a single OLS estimation.
• All the variables are assumed to be endogenous
(although it is possible to include exogenous
variables)
• The test relies on the relationship between the rank
of a matrix and its eigenvalues or characteristic
roots.
21
• The approach to testing for cointegration in a
multivariate system is similar to the ADF test, but
requires the use of a VAR approach:
xt  A1 xt 1  ut
xt  ( A1  I ) xt 1  ut
xt  xt 1  ut
Where :   ( A1  I )
Where in a system of g variables:
xt and ut are g x 1 vectors.
A1 is an g x g matrix of parameters
I is an g x g Identity matrix 22
• The rank of π equals the number of cointegrating
vectors
• If π consists of all zeros, as with the ADF test, the rank
of the matrix equals zero, all of the xs are unit root
processes, implying the variables are not cointegrated.
• As with the ADF test, the equation can also include
lagged dependent variables, although the number of
lags included is important and can affect the result.
This requires the use of the Akaike or Schwarz-
Bayesian criteria to ensure an optimal lag length.
23
The π Matrix
• The number of cointegrating vectors (r), is the rank
of π and determines the cointegration relationship
• When r = 0 there are no cointegrating vectors
• If there are g variables in the system of equations,

there can be a maximum of g-1 cointegrating
vectors.
24
The π Matrix
• π is defined as the product of two matrices: α and β’, of
dimension (g x r) and (r x g) respectively.
• The β gives the long-run coefficients of the cointegrating

vectors, the α is known as the adjustment parameter and
is similar to an error correction term.
• The relationship can be expressed as:

π = αβ’
25
The Johansen Test Statistics
• There are two test statistics produced by the
Johansen ML procedure.
• There are the Trace test and maximal Eigenvalue

test.
• Both can be used to determine the number of

cointegrating vectors present, although they don’t
always indicate the same number of cointegrating
vectors.
26
The Johansen Test Statistics
• The test statistics for cointegration are formulated as
g
trace (r )  T ln(1  ˆi )


i  r 1
and

max (r , r  1)   T ln(1  r 1 )

where r is the estimated value for the ith ordered
eigenvalue from the  matrix.

27
Differences Between the Two Test Statistics
• The Trace test is a joint test, the null hypothesis is
that the number of cointegrating vectors is less than
or equal to r, against a general alternative
hypothesis that there are more then r.
 trace = 0 when all the i = 0, so it is a joint test.
• The Maximal Eigenvalue test conducts separate

tests on each eigenvalue. The null hypothesis is that
there are r cointegrating vectors present against the
alternative that there are (r + 1) present.
28
The Johansen Critical Values
 Johansen & Juselius (1990) provide critical values for the
2 statistics. The distribution of the test statistics is non-
standard.
 The critical values depend on:
1. the value of g-r, the number of non-stationary
components
2. whether a constant and / or trend are included in the
regressions.

 If the test statistic is greater than the critical value from
Johansen’s tables, reject the null hypothesis that there
are r cointegrating vectors in favour of the alternative that
there are more than r.

29
The Johansen Testing Sequence
• The testing sequence under the null is r = 0, 1, ..., g-1
so that the hypotheses for trace are

H0 : r = 0 vs H1: 0 < r  g
H0 : r = 1 vs H1: 1 < r  g
H0 : r = 2 vs H1: 2 < r  g
... ... ...
H0 : r = g-1vs H1: r = g

• We keep increasing the value of r until we no longer reject the
null.

30
Interpretation of Johansen Test
Results
• But how does this correspond to a test of the rank of
the  matrix?
• r is the rank of .
  cannot be of full rank (g) since this would
correspond to the original yt being stationary.
• If  has zero rank, then by analogy to the univariate
case, yt depends only on yt-j and not on yt-1, so that
there is no long run relationship between the elements
of yt-1. Hence there is no cointegration.
• For 1 < rank () < g , there are multiple cointegrating
vectors.
31
Example
• Given the following model of stock prices (s) and income
(y);
st   0  1 yt  ut ut  i.i.d .(0, )
i.i.d  independently and identically distributed.
 - variance - covariance matrix, indicating no heteroskedasticity.
Johansen ML Results (Trace Test)
Null Alternative Trace 5% CV
r=0 r1 40.3 20.2
r1 r=2 7.6 9.16
Johansen ML Results (Maximum Eigenvalue Test)
Null Alternative Eigenvalue 5% CV
r=0 r =1 34.7 15.9
r=1 r=2 8.6 9.2
32
Interpretation of Results
• Given that for both tests, the test statistic exceeds
its critical value (5%) when the null is r = 0, we can
conclude that at least one cointegrating vector is
present.
• For more than one cointegrating vector, the test

statistic is less than the critical value so we conclude
only a single cointegrating vector is present.
33
Eviews Output
Unrestricted Cointegration Rank Test (Trace)
Hypothesized Trace 0.05

No. of CE(s) Eigenvalue Statistic Critical Value Prob.**
None * 0.551568 33.49504 29.79707 0.0179

At most 1 0.206403 7.831124 15.49471 0.4836
At most 2 0.013452 0.433397 3.841466 0.5103
Trace test indicates 1 cointegrating eqn(s) at the 0.05 level

* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values
34
Eviews Output
Unrestricted Cointegration Rank Test (Maximum Eigenvalue)
Hypothesized Max-Eigen 0.05

No. of CE(s) Eigenvalue Statistic Critical Value Prob.**
None * 0.551568 25.66392 21.13162 0.0107

At most 1 0.206403 7.397728 14.26460 0.4432
At most 2 0.013452 0.433397 3.841466 0.5103
Max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.05 level

* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values
35
Normalised Cointegrating Vector
(Long-run β Coefficients)
• The long-run coefficients are normalised, such that we
express the relationship in terms of one of the variables as a
dependent variable:
1 Cointegrating Equation(s): Log likelihood 143.4857
Normalized cointegrating coefficients (standard error in parentheses)

FD FDI Y
1.000000 -0.265277 -0.596661
(0.13271) (0.14183)
FD = 0.2652 FDI + 0.5966 Y
Adjustment coefficients (standard error in parentheses)
D(FD) -0.214006
(0.07411)
D(FDI) 0.104577
(0.17519)
D(Y) -0.092476
(0.06283)
36
The α Adjustment Coefficients
• These can be interpreted in exactly the same way as the
error correction term, asymptotic t-statistics are in
parentheses (interpreted in the same way as t-statistics):
1 Cointegrating Equation(s): Log likelihood 143.4857
Normalized cointegrating coefficients (standard error in parentheses)

FD FDI Y
1.000000 -0.265277 -0.596661
(0.13271) (0.14183)
Adjustment coefficients (standard error in parentheses)

D(FD) -0.214006
(0.07411)
D(FDI) 0.104577
(0.17519)
D(Y) -0.092476
(0.06283)
37
Tests of Specific Restrictions
• The Johansen ML approach, unlike the bi-variate
approach can be used to apply certain restrictions to
the long-run β coefficients.
• This can involve testing if they are significantly

different to zero or not, or equal to one.
38
Main Differences with the Bi-variate Test
for Cointegration
• Using the Johansen Maximum Likelihood (ML)
procedure, it is possible to obtain more then a single
cointegrating relationship, whereas only one can be
obtained with the Engle-Granger test.
• There are two separate tests (Trace & Max Eigenvalue)
for cointegration with the Johansen, but only one with the
Engle-Granger which can give different results.
• Given that the Johansen is a maximum likelihood based
test (Engle-Granger is OLS based), it requires a large
sample.
• The multivariate test is based on a VAR, not a single
OLS estimation as with the Engle-Granger approach.
39
Criticisms of the Johansen Approach
• The result can be sensitive to the number of lags

included in the test and the presence of
autocorrelation
• If there are more than two cointegrating vectors
present, how do we find the most appropriate vector
for the subsequent tests.
• If the two test statistics differ, which one gives the
correct result?
• This is a large sample test.
• The Wickens critique suggests we often find
evidence of cointegration when none exists.
40
Multivariate Cointegration and Vector-
Error Ccorrection Models (VECM)
• Vector Error Correction Models (VECM) are the basic VAR,
with an error correction term incorporated into the model
and as with bivariate cointegration, multivariate
cointegration implies an appropriate VECM can be formed.
• The reason for the error correction term is the same as with
the standard error correction model, it measures any
movement away from the long-run equilibrium.
• These are often used as part of a multivariate test for
cointegration, such as the Johansen ML test, having found
evidence of cointegration of some I(1) variables, we can
then assess the short run and potential Granger causality
with a VECM.
41
The Approach to Multivariate
Cointegration and VECMs
1) Test the variables for stationarity using the usual ADF tests.
2) If all the variables are I(1) include in the cointegrating
relationship.
3) Use the AIC or SBC to determine the number of lags in the
cointegration test (order of VAR)
4) Use the trace and maximal eigenvalue tests to determine
the number of cointegrating vectors present.
5) Assess the long-run β coefficients and the adjustment α
coefficients.
6) Produce the VECM for all the endogenous variables in the
model and use it to carry out Granger causality tests over
the short and long run.
42
Vector Error-correction Modeling (VECM)
Multi-Equation ECM
• It is termed vector error-correction modeling
(VECM).
• The VECM is taken is to investigate dynamic causal

interactions among the variables  among them is
the notion of Granger causality.
• The interest is also to see which variable takes the

burden of adjusting towards the long-run
equilibrium.
43
VECM (3-variable system)
k k k
Yt   1   1i Yt i    1i X t i    1i Z t i  1 t 1  u1t
i 1 i 1 i 1
k k k
X t   2    2i Yt i    2i X t i    2i Z t i   2  t 1  u 2t
i 1 i 1 i 1
k k k
Z t   3    3i Yt i    3i X t i    3i Z t i  3 t 1  u 3t
i 1 i 1 i 1
Yt 1  [  1 X t 1   2 Z t 1 ]   t 1
• What should be the signs of the speed of adjustment

coefficients?
44
• Lag length selection: information criteria and the
requirement that the errors must be non-autocorrelated.
• Note: if the variables are not cointegrated, the error

correction terms must be dropped. In this case, we
have VAR.
• Note: in a system equation like VAR and VECM, the

interest is not in the estimated coefficients [except the
coefficient of the error correction terms]. Accordingly,
they are not normally reported.
45
INTERPRETATION I: GRANGER CAUSALITY
k k k
Yt   1   1i Yt i    1i X t i    1i Z t i  1 t 1  u1t
i 1 i 1 i 1
k k k
X t   2    2i Yt i    2i X t i    2i Z t i   2  t 1  u 2t
i 1 i 1 i 1
k k k
Z t   3    3i Yt i    3i X t i    3i Z t i  3 t 1  u 3t
i 1 i 1 i 1
• The normal channel of causality is the coefficient of lagged first-

differenced variables.
• The error-correction terms add an additional channel of
causality.
• The former is termed “short-run” causality while the latter is
“long-run” causality.
• Thus, omitting the error correction terms from the equation
create bias in the estimation due to omitting a relevant variable.
46
• Another causality is the strong exogeneity test which
imposes stronger restrictions by testing the joint
significance of both the lagged dynamic terms and
ECT (Charemza & Deadman, 1992; Engle, Hendry,
& Richard, 1983).
• Strong causality – joint test to check for strong

causality test, where variables bear the burden of
short-run adjustment to re-establish a long run
equilibrium
47
Example: Results of Vector Error-correction Modeling
48
VAR - Number of cointegrating vector (Eviews)
49
50
51
Summary: Steps in Time Series Econometrics
THEORETICAL UNDERPINNINGS
Variables to be Included? Focus? Well-specified Equation or System Dynamics.
Motivation?
UNIT ROOT TESTS
COINTEGRATION TESTS
One-equation Focus: SYSTEM DYNAMICS

Error Correction Modeling
- Specification Granger VAR or VECM
- Estimation Causality
- Diagnostics Variance
- Inferences Decompositions &
Impulse Responses

Hand on Session
• VECM and Cointegration, Granger
Causality, Impulse Response Functions
and Variance Decomposition
• Datasets: Macro.xls
53

Lecture 7 VAR, VECM and Multivariate Cointegration

Uploaded by

Copyright:

Available Formats

Lecture 7 VAR, VECM and Multivariate Cointegration

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture 7 VAR, VECM and Multivariate Cointegration

Uploaded by

Copyright:

Available Formats

VAR, Multivariate Cointegration and

Vector-error Correction Model

The School of Management

• Quite common in economics to have models

Law, S.H., Department of Economics, UPM

• Single equations models of the form y = X + u

• All of the variables contained in the X matrix are

Law, S.H., Department of Economics, UPM

Sims (1980) suggests:

Law, S.H., Department of Economics, UPM

• For example, time series yt that is affected by current

Law, S.H., Department of Economics, UPM

This is a first-order VAR model, because the

Rewriting the system with matrix algebra:

Law, S.H., Department of Economics, UPM

Law, S.H., Department of Economics, UPM

Law, S.H., Department of Economics, UPM

Suppose two variables, say yt and xt , affect each other

Law, S.H., Department of Economics, UPM

Granger (1969) developed a relatively simple

A variable yt is said to Granger-cause xt, if xt

Law, S.H., Department of Economics, UPM

First step is to estimate following VAR model:

Law, S.H., Department of Economics, UPM

Law, S.H., Department of Economics, UPM

Step 2 Regress yt on lagged y terms plus lagged x and

Step 3 Set the null and the alternative hypotheses:

Law, S.H., Department of Economics, UPM

Step 4 Calculate F statistic for normal Wald test on

Step 5 If computed F value exceeds

Law, S.H., Department of Economics, UPM

• In general, for n number of variables we can have only

• What should you do in this case? One option is to use

• Using the Johansen Maximum Likelihood (ML)

• When r = 0 there are no cointegrating vectors

• If there are g variables in the system of equations,

• The β gives the long-run coefficients of the cointegrating

• The relationship can be expressed as:

• There are the Trace test and maximal Eigenvalue

• Both can be used to determine the number of

trace (r )  T ln(1  ˆi )

max (r , r  1)   T ln(1  r 1 )

• The Maximal Eigenvalue test conducts separate

• For more than one cointegrating vector, the test

Hypothesized Trace 0.05

None * 0.551568 33.49504 29.79707 0.0179

Trace test indicates 1 cointegrating eqn(s) at the 0.05 level

Hypothesized Max-Eigen 0.05

None * 0.551568 25.66392 21.13162 0.0107

Max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.05 level

Normalized cointegrating coefficients (standard error in parentheses)

Normalized cointegrating coefficients (standard error in parentheses)

Adjustment coefficients (standard error in parentheses)

• This can involve testing if they are significantly

• The result can be sensitive to the number of lags

• The VECM is taken is to investigate dynamic causal

• The interest is also to see which variable takes the

• What should be the signs of the speed of adjustment

• Note: if the variables are not cointegrated, the error