Friday, August 30, 2013

Right-Tail Augmented Dickey-Fuller Tests in EViews

A few days ago, I received an email from Itamar Caspi, a regular follower of this blog. Itamar has developed a really nice EViews "Add-In" package that facilitates the application of "Right-Tail Augmented Dickey-Fuller" tests.

His Add-in package, named rtadf, covers four tests:

1. ADF.
2. Rolling ADF.
3. sup ADF (SADF). See Phillips et al. (2011).
4. Generalized SADF (GSADF), see Phillips et al. (2013).

As Itamar explains, the tests are "....designed to detect the presence of an unobserved bubble component in an observed asset price and to date-stamp its occurrence".

In EViews parlance, "Add-In" packages are "..... EViews programs that, when installed, look and feel like built-in EViews procedures. Packages may generally be run from the EViews object and Add-ins menu, or via commands. Once installed, add-in packages should require no user-modification.

(If you use R, then the concept is essentially  the same as with R packages.)

The various Add-In packages that are available for EViews (versions 7.1 and above) can be seen here.

If you're an EViews user, you probably use Add-Ins already. If you haven't tried any of them yet, here's what you do if you're using EViews 8. (EViews 7 users have to download Add-ins manually from the Add-in webpage: http://www.eviews.com/Addins/addins.shtml).
  1. Launch EViews (of course!) 
  2. Select the Add-ins button on the top tool-bar, select Manage Add-ins from the drop-down menu, and you see something like this:


     3. These are the Add-ins that are currently available for you to integrate into your installed version of EViews.

     4.   Highlight the Add-in(s) you want, and click on the Install button. You'll then see: something like:


    5. Click "O.K." at all of the prompts, and then you're ready to go!

    6.  To access your Add-in, make the following selections on the main top tool-bar of the EViews window:
         File         Open       Programs in Add-ins Folder...
         
          and select the Add-in that you want to use; just as you would for any other program.

    7.  To see the documentation for an Add-in, just go back to Step 2. This time, click on the "Installed" tab            and highlight the Add-in of interest, and select the Docs button. For example, if you do this for the                rtadf  Add-in package, this is what you'll see:



If you're not an EViews user, you can still access Itamar's excellent description of his rsadf Add-in package, in pdf format, through DropBox (here).

Nice work, Itamar - I hope we see more contributions like this from you in the future!


References

Phillips, P., S. Shi, and J. Yu, 2013. Testing for Multiple Bubbles 1: Historical episodes of exuberance and collapse in the S&P 500.

Phillips, P., Y. Wu, Y., and J. Yu, 2011. Explosive behavior in the 1990s Nasdaq: When did exuberance escalate asset values? International Economic Review, 201, 201–226.

Phillips, P. C. B. and J. Yu, 2011. Dating the timeline of financial bubbles during the subprime crisis. Quantitative Economics, 2, 455–491.


© 2013, David E. Giles

15 comments:

  1. Quick comment - your instructions on how to install the Add-in are only valid for EViews 8 users. EViews 7 users will have to download it manually from the Add-in webpage:
    http://www.eviews.com/Addins/addins.shtml

    ReplyDelete
    Replies
    1. Gareth - whoops - I have amended the wording slightly above. Thanks for catching this!!!!

      Delete
  2. Has an econometric method ever detected a bubble before it burst? Or are these methods purely retrospective?

    ReplyDelete
    Replies
    1. And how many non-bubbles have they detected?

      Delete
  3. It simply tests for rho > 0 (in the differenced regression). Dickey and Fuller have the critical values in their original paper. Its interpretation is that the time series has an explosive root (root > 1). I do not know why this is a test for bubbles. In economics, bubbles occur when prices increase without a change in fundamentals...

    W

    ReplyDelete
  4. The following recent paper by Philip Hans Franses may also be of interest:
    http://repub.eur.nl/res/pub/39598/

    ReplyDelete
  5. In the real world, it all comes down to one test: How much of your own income/net worth are you willing to bet on your opinion of what is going on in the time series? Do econometricians ever wager their own money on the accuracy/usefulness of their models?

    ReplyDelete
  6. Ok, here is a practical question for Dave and Weshah: Is there a bubble in US ag land right now? If yes, how do we exploit it?

    ReplyDelete
    Replies
    1. Joe - I would examine the elasticity of the supply of land. An inelastic supply curve (say vertical for example) would always result in ever higher price in the face of a demand shock. Restrictions and regulations can also make the supply inelastic and result in very high price. If the supply is very inelastic it is hard to argue for bubbles.

      Delete
  7. I take exception to your "bubbles occur when prices increase without a change in fundamentals." There is ALWAYS a change in fundamentals. Read the old pieces about the housing bubbles. Plenty of economists assured us that there was no bubble precisely because the fundamentals supported the increase in prices. Two of my favorites are here: http://www.nber.org/papers/w11643.pdf and http://ftp.ny.frb.org/research/epr/04v10n3/0412mcca.pdf. Ceteris paribus does not make any sense outside of the econ textbooks.

    ReplyDelete
    Replies
    1. Thanks, what I meant is that the increase in price is unjustified by the change in fundamentals...sorry for that.

      Delete
  8. need help asap
    am using Eview 7 i installed the RTADF add ins but they dnt work ! something about registration and IHS name or somethng !!! Plz how can i get them on my PC asap ! i really need this for my theises

    ReplyDelete
    Replies
    1. Why on earth don't you contact the author of the add-in, or the people at EVIEWS?!?!?!?! I don't work for them.

      Delete
  9. How would you perform these tests on STATA? Specially the GSADF. Thank you very much.

    ReplyDelete
    Replies
    1. You should raise this on one of the Stata user groups, not here.

      Delete