Scribd
Upload
23 views6 pages

ECN 5013-Time Series Models-II

The document discusses cointegrated models and error correction models (ECM) in econometrics, focusing on the long-run equilibrium of economic variables and the conditions for cointegration. It outlines the Engle-Granger two-step methodology for estimating long-run relationships and constructing ECMs, emphasizing the importance of testing for integration and assessing model adequacy. The advantages of ECM include addressing spurious correlation, reducing multicollinearity, and enabling classical regression methods for estimation.

Uploaded by

yacoubouikyamohamed
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
23 views6 pages

ECN 5013-Time Series Models-II

The document discusses cointegrated models and error correction models (ECM) in econometrics, focusing on the long-run equilibrium of economic variables and the conditions for cointegration. It outlines the Engle-Granger two-step methodology for estimating long-run relationships and constructing ECMs, emphasizing the importance of testing for integration and assessing model adequacy. The advantages of ECM include addressing spurious correlation, reducing multicollinearity, and enabling classical regression methods for estimation.

Uploaded by

yacoubouikyamohamed
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

ECN 5013- Econometrics-II

Time Series Models

COINTEGRATED MODEL

A set of economic variables are said to be in the long-run


equilibrium when

β1x1t + β2 x 2t + L + βn x nt = 0
[1]
Or if βx t = 0 ,
where β = (β1, β2 ,Lβn ) and x t = (x1t , x 2t ,L x nt )′ .

The deviation from the long run equilibrium is called the


equilibrium error and is denoted as e t = β x t . If the
variables are non-stationary and the equilibrium is
meaningful, then the equilibrium error process is
stationary and the variables in the vector
x t = (x1t , x 2t ,L x nt )′ is said to be cointegrated. Engle
and Granger provide the following definition of
cointegration:

1. All variables in x t = (x1t , x 2t ,L x nt )′ are integrated


of order d;
2. There exists a vector β = (β1, β2 ,Lβn ) such that the
linear combination
β x t = β1x1t + β2 x 2t + L + βn x nt
is integrated of order (d-b), where b > 0 . Then the
vector β = (β1, β2 ,Lβn ) is called the cointegrated
vector.

There are four imported points to be noted regarding the


definition of co-integration:

1. Cointegration refers to the linear combination of


non-stationary variables;
2. All variables must be integrated of the same order;
3. If β = (β1, β2 ,Lβn ) is a cointegrating vector then
for any non-zero value of λ ,
λβ = (λβ1, λβ2 ,L λβn ) is also a cointegrating
vector. Typically one of the variable is used to
normalise the cointegrated vector by fixing its
coefficient as unity. e.g. to normalization the
variable x k its λ is set as 1/ βk . If
x t = (x1t , x 2t ,L x nt )′ has n components, there may
be as many as n-1 linearly independent cointegrated
vectors. The number of cointegrating vectors is
called cointegrating rank of x t . E.g. if
1 −βo −β1 −β2 −β3 
β=  and then the two
 1 −γ o −γ1 −γ 2 0 
linear combination of β x t are stationary and the
cointegrating rank of x t is 2.
4. Mostly cointegration literature relate with variables
with single unit root.

ERROR CORRECTION MODEL


In the error correction model the short term dynamics of
the variables in the system are influenced by the
deviation from equilibrium.

Engle Granger Two Step Methodology


The Engle-Granger two-step procedure first considered
the estimation of the long run from a single equation
regression and then the residual from this model defines
a cointegrating vector. Lagged residuals can be entered
into a dynamic model and this is then described as an
error correction or more precisely the equilibrium
correction term.

Step 1:
For the variables y t and z t pre-test the order of
integration. If the variables are integrated of the same
order, it is possible that they are co-integrated. If both the
variables are stationary, it is not necessary to proceed,
since the standard time series methods apply to the
stationary variables. If the variables are integrated of
different orders, it is most likely possible that they are
not cointegrated.

Step 2:
If the variables y t and z t are I(1) and are co-integrated,
estimate the long run equilibrium relationship

y t = β0 + β1z t + e t [2]

by using OLS including a constant term and possibly a


trend. The estimates of β0 and β1are super-consistent, i.e.
they converge faster towards the true parameter values
than in OLS model using stationary variables. Dickey
Fuller test can be used to test the integration of the
residuals without an intercept. i.e. test is based on
H 0 : a1 = 0 on

∆eˆ t = a1eˆ t −1 + ε t [3]

If the residuals of (2) do not appear to be white noise, an


augmented form of the test can be used on

n
∆eˆ t = a1eˆ t −1 + ∑ a i +1∆eˆ t −i + ε t [4]
i =1

Step3:
The residuals from the equilibrium regression can be
used to estimate the error correction model of the form

∆y t = α1+α y (y t −1 − β1z t −1)

+ ∑ α11(i)∆y t −i + ∑ α12 (i)∆z t −i + ε yt


i =1 i =1
[5]
∆z t = α 2 +α z (y t −1 − β1z t −1)

+ ∑ α 21(i)∆y t −i + ∑ α 22 (i) ∆z t −i + εzt


i =1 i =1
[6]
or
∆y t = α1+α yeˆ1t−1

+ ∑ α11(i)∆y t −i + ∑ α12 (i)∆z t −i + ε yt


i =1 i =1
[7]
∆z t = α 2 +α zeˆ t −1

+ ∑ α 21(i)∆y t −i + ∑ α 22 (i) ∆z t −i + εzt


i =1 i =1
[8]

Step4:
Asses the model adequacy based on the following points:
1. Assess the model adequacy based on the diagnostic
checks. If the residuals of the error-correction model
is serially correlated, re-estimate the model with
adequate lag length that yield serially uncorrelated
errors.
2. Interpret the speed of adjustment coefficients α y
and α z .
3. Impulse response and variance decomposition
analysis can be used to obtain the information
concerning the interaction or contemporaneous
correlation among the variables.

Advantages of the ECM


1. Error Correction Model (ECM) is formulated in
terms of first differences which typically eliminate
the trends from variables and thus can play an
important role in dealing with the problem of
spurious correlation.
2. When ECM is correctly formulated, its
disequilibrium error term can also be regarded as a
stationary variable.
3. Provided sample size is sufficiently large, ECM can
be estimated by the classical regression methods.
4. By the ECM, the problem of multi-collinearity are
greatly reduced.
5. ECM can be fitted through Hendry’s general to
specific methodology for the search of the most
parsimonious model.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy