Scribd
Upload
7 views

Unit Tests (Lab Week of 2-5-2023)

This document provides instructions for conducting unit root tests using the Augmented Dickey-Fuller (ADF) test in EViews. It explains that an ADF test should first be run on the level of a variable to determine if it contains a unit root and is non-stationary (I(1)), then if found to be I(1) the first difference should be tested which should be stationary (I(0)). Precautions are noted such as ADF test statistics must be negative and conflicting results across test specifications should be resolved by the majority outcome.

Uploaded by

Eleanora Lawrence
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
7 views

Unit Tests (Lab Week of 2-5-2023)

This document provides instructions for conducting unit root tests using the Augmented Dickey-Fuller (ADF) test in EViews. It explains that an ADF test should first be run on the level of a variable to determine if it contains a unit root and is non-stationary (I(1)), then if found to be I(1) the first difference should be tested which should be stationary (I(0)). Precautions are noted such as ADF test statistics must be negative and conflicting results across test specifications should be resolved by the majority outcome.

Uploaded by

Eleanora Lawrence
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 5

Brief Recap

 Unit root  non-stationary variable  I(1)


 Stationary variable  I(0)
 See “Lecture 1 (time series)(s 23).pptx”
 Click on variable
 Unit root tests
 ADF test
 Start with “Test for unit root in”: level
 Select different options in ADF test: none, intercept, trend + intercept
 Keep automatic lag length selection (say SIC)
 Null hypothesis: variable has a unit root
 If absolute value of Augmented Dickey-Fuller test statistic < absolute values of critical
values at 1% or 5% or 10% then do not reject null hypothesis  variable is I(1)
 If variable is I(1) then first difference should be I(0)
 Now test for unit root in the first difference of the variable.
 If the variable is I(1) then its first difference will be I(0)  all the ADF test statistics will
have values greater than the absolute values of the critical values
Some precautions

 ADF test statistics have to be negative. Ignore positive values.


 In the “none, intercept, trend + intercept” specifications of the random walk you may get
conflicting results: go with the majority outcome regarding whether or not the variable is
I(1) or I(0).
 Tests for cointegration are only relevant for I(1) variables.

You might also like