Scribd
Upload
117 views19 pages

Time Series Analysis With Eviews

Uploaded by

Geetika Jain
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
117 views19 pages

Time Series Analysis With Eviews

Uploaded by

Geetika Jain
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/ 19

Time Series Analysis with Eviews

Dr. Sunita Arora


Associate Professor
Government College for Girls,
Sector 14, Gurugram
Email: [email protected]

1
References
by Damodar N. Gujarati
by
Chris Brooks
by Walter
Enders
•Econometrics by Examples by Damodar
Gujarati

2
Types of Data
•Cross Sectional data: Data on one or more
variables from more than one units collected at the
same point in time.
•Time Series data: Data on one/more variable/s
ordered chronologically from one unit.
•Panel data: Combines features of both cross-
section and time series data

3
Examples of issues addressed by analysing time series data

We may be interested in knowing whether


•Share prices of a company for one day are affected by the
prices of previous day.
•GDP causes money supply or money supply causes GDP.
•Spot prices of an asset lead its futures prices or there may
be reverse relationship i.e. futures prices may lead spot
prices.
•Share prices of one company are more volatile than the
share prices of another company.
•Return or volatility for Monday is more or less than the
return or volatility for other days of the week.
4
Models applicable to answer the above
questions
Question 1 Auto Regressive model

Question 2 Granger Causality

Question 3 Error correction Model

Question 4 ARCH Family of Models

Question 5 Dummy Variable Regression

5
Basic Requirements Before
Analysing Time Series
•Graphically Present the data to detect outliers.
•Check for stationarity : A time series is strictly
stationary if all the moments of its probability
distribution are invariant over time. In most of
the practical situations weakly stationary
suffices. In weakly stationary mean and variance
and auto-covariance(at various lags) remain
constant over time.
6
Graphical Presentation of data to detect outliers

7
Why Stationary Time Series
•Spurious Regression- example Regression
between salary of bank employees in India and
death rate in Iraq.

How to check: R2>Durbin Watson.

•Results of the regression analysis can not be


generalised, i.e. these results are applicable for
that particular time period.

8
Diagnostic Tools for Stationary Time
Series
• Time Series Plot: If the time series is not stationary, it will not
hover around the mean
• Autocorrelation Function (ACF) and Correlogram
The ACF at lag k is defined as:
 


k
k
0

Where γk is the covariance at lag k and γ0 is the variance.


• Unit Root Test
Augmented Dickey Fuller Test
Philips Perron test
KPSS
9
Non Stationary: Another Name- Explosive, Unit Root

FUTURES

4,800

4,400

4,000

3,600

3,200

2,800

2,400

2,000

1,600

0 09 0 10 0 11 0 12 0 13 0 14
2 /2 /2 /2 /2 /2
1/ /1 /1 /2 /1 /1
1/ 1 1 1 1 1

10
Stationary: Other names- Non-explosive, mean reversion

FUTURESRETURN

.06

.04

.02

.00

-.02

-.04

-.06
9

4
0

1
20

20

20

20

20

20
1/

1/

1/

2/

1/

1/
1/

1/

1/

1/

1/

1/
11
Correlogram of a non stationary Time Series

Date: 11/03/14 Time: 18:41


Sample: 1/01/2009 10/24/2014
Included observations: 1752

Autocorrelation Partial Correlation AC PAC Q-Sta... Prob

1 0.998 0.998 1746.4 0.000


2 0.995 -0.00... 3485.3 0.000
3 0.993 -0.02... 5216.3 0.000
4 0.990 0.033 6939.9 0.000
5 0.988 -0.05... 8655.4 0.000
6 0.985 -0.01... 10363. 0.000
7 0.982 -0.01... 12061. 0.000
8 0.979 0.004 13751. 0.000
9 0.976 -0.02... 15432. 0.000
1... 0.974 -0.00... 17104. 0.000
1... 0.971 -0.00... 18767. 0.000
1... 0.968 -0.02... 20421. 0.000
1... 0.965 0.021 22066. 0.000
1... 0.962 -0.00... 23702. 0.000
1... 0.959 -0.01... 25329. 0.000
1... 0.956 0.015 26947. 0.000
1... 0.953 0.005 28556. 0.000
1... 0.950 0.008 30157. 0.000
1... 0.948 0.043 31750. 0.000
2... 0.945 0.012 33335. 0.000

12
Correlogram of a stationary time series.

Date: 11/03/14 Time: 19:08


Sample: 1/01/2009 10/24/2014
Included observations: 1751

Autocorrelation Partial Correlation AC PAC Q-Sta... Prob

1 0.022 0.022 0.8216 0.365


2 0.043 0.042 4.0184 0.134
3 -0.01... -0.02... 4.6524 0.199
4 0.032 0.031 6.4915 0.165
5 -0.02... -0.02... 7.8859 0.163
6 0.057 0.056 13.635 0.034
7 -0.01... -0.01... 14.281 0.046
8 -0.01... -0.02... 14.629 0.067
9 -0.01... -0.00... 14.824 0.096
1... -0.01... -0.01... 15.253 0.123
1... 0.010 0.015 15.415 0.164
1... -0.01... -0.01... 15.834 0.199
1... -0.00... 0.001 15.834 0.258
1... -0.01... -0.00... 16.066 0.309
1... -0.03... -0.03... 17.793 0.274
1... 0.008 0.014 17.906 0.329
1... -0.03... -0.03... 19.777 0.286
1... -0.02... -0.02... 20.768 0.291
1... -0.02... -0.01... 21.716 0.299
2... -0.06... -0.06... 28.350 0.101

13
Regression equation for The Augmented Dickey-Fuller
(ADF)
m
Yt   1   2t  Yt 1  iY t  1  t
i 1

14
Transforming Non Stationary
Time Series:
A non stationary time series may be difference
stationary or trend stationary, if it is differenced
stationary, it may be transformed to make it stationary
by any of the following two ways:
X t  X t  Xt 1

or

 log Xt  log Xt  log Xt  1

15
Null Hypothesis: FUTURES has a unit root
Exogenous: Constant, Linear Trend
Lag Length: 0 (Automatic - based on SIC, maxlag=24)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -1.270099 0.8944


Test critical values: 1% level -3.963292
5% level -3.412377
10% level -3.128130

*MacKinnon (1996) one-sided p-values.

16
Null Hypothesis: FUTURESRETURN has a unit root
Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=24)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -40.86045 0.0000


Test critical values: 1% level -2.566284
5% level -1.941005
10% level -1.616578

*MacKinnon (1996) one-sided p-values.

17
Models applied on stationary and
non stationary time series
Stationary Series Non Stationary Series
• OLS •Cointegration Test
• ARMA ARIMA •Error Correction Model
• Granger Causality Test
• VAR
• ARCH family of models

18
19

You might also like