This document provides an introduction and overview of using the Eviews software platform. It discusses how to create and open work files, generate random and transformed time series data, perform descriptive statistics and correlation analysis, and check for autocorrelation using correlograms. Key aspects of time series properties like stationarity and white noise are also covered. The document demonstrates various commands and functions in Eviews for working with time series data, from importing datasets to generating statistics and exploring the characteristics of the series.
Q3 2024 Earnings Conference Call and Webcast Slides
Introduction to Eviews.pptx
1. Introduction to Eviews
Ms P B Saranya
Assistant Professor, GRGSMS
Dr. N.G.P. Institute of Technology, Coimbatore
Faculty Development Programme
Research Methods In Finance
3. To Create a New Work file
Select
Unstructured
Enter the Number
of observations
Click OK
4. Opening a Data File
An Eviews Work File
• File – Open Eviews Work File – Select the required File – Ok
• Double click the Eviews Work File
An Excel File
• File – Open – Foreign Work File - Select the required File – Ok
• Right Click the Excel File – Choose Open With – Eviews
• Copy the data in the excel sheet – Paste it in the Eviews
window (on the blue area)
Import from File
6. Series Generation
To Generate a random Variable
In the Command Tab
smpl 1 250
(to set a space to generate a random variable, smpl
means Sample, 1 250 represents the number of
observations)
Series y=nrnd
(y represents the variable name and nrnd represents
random)
In the menu
• Quick – Generate Series
• Work file window – Genr tab
7. Series Generation (Commands)
Commands to Generate different Series while using Quick – Generate
Series or Genr Tab
Creating a lagged series
Variable_Name=Variable_Name(-1)
Creating a differenced series
Variable_Name=d(Variable_Name) ---- First Difference
Variable_Name=d((Variable_Name)) --- Second Difference
Creating a log series
Variable_Name=log(Variable_Name)
Creating a difference log series
Variable_Name=dlog(Variable_Name)
Creating a logirthamic return series
Variable_Name=log(Variable_Name/Variable_Name(-1)) (Log Returns)
Variable_Name=dlog(Variable_Name)*100 (Log returns in %)
11. Skewness
For Normal
Distribution it will
zero or near zero,
will be negative
when left skewed
and positive when
right skewed
Kurtosis
For Normal
Distribution it will
3 or near 3, will be
greater than 3
when peaked and
close to zero when
flat
14. Auto Correlation & Correlogram Test
Auto correlation is the correlation of a time series with its own
past and future values, it is also sometimes called as serial
correlation.
PACF takes into consideration the correlation between a time
series and each of its intermediate lagged values. It finds
correlation of the residuals (which remains after removing the
effects which are already explained by the earlier lag(s)) with the
next lag value. Finding the correlation between the present value
and 2nd lag by controlling the effect of the 1st lag.
15. Click on Quick – Series Statistics – Correlogram (or) Open the
variable –Click View - Correlogram
18. Properties of Time Series Data
• Non Stationary – Mean and variance is not constant, varies over time
• The time series variable (for example, the stock price) may have
a trend over time.
• The data may have an irregular component, which is referred to as
the White Noise or a random variation which is not explained by
any factor
• The data will have serial or auto correlation between subsequent
observations
19. Stationarity
Stationary process is in a statistical equilibrium; its
statistical properties remain unchanged as time
passes.
(i) constant Mean: E (yt) = E (yt-j ) =
(ii) constant variance
(iii) Constant covariance
Shocks to the system will gradually die away (i.e.,
effect of a shock at t will have small effect in t+1,
smaller effect in t+2 and so on.
It is a flat looking series.
20. Why stationary time series is important
First, if a time series is non-stationary, we can study its
behavior only for the time period under consideration. As
a result, it is not possible to generalize it to other time
periods. Therefore, for the purpose of forecasting or
policy analysis, such (non-stationary) time series may be
of little practical value.
Second, if we have two or more non-stationary time
series, regression analysis involving such time series may
lead to the phenomenon of spurious or nonsense
regression.
21. Stationarity
Stationarity Non Stationarity
It has a constant mean,
variance and covariance's
They are not constants, but
time dependent
Mean reverting process Not mean reverting as there is
no long run mean to return
Has theoretical correlogram
that diminishes as time lag
increases
Auto correlation coefficients do
not decay
It has limited memory of its
past behaviour implying that
shocks to it are temporary
Such effects are permanent
22. White Noise
• Mean is Zero
• Constant variance
• No auto correlation
• β=0
• Yt is not depending upon Yt-1
• Mean reverting