Bounds Testing & ARDL Models - More From the EViews Team

The team at EViews has just released another post about ARDL modelling on their blog. This one is titled, "AutoRegressive Distributed Lag (ARDL) Estimation. Part 2 - Inference". This post is a follow-up to one that they wrote last month, and which I commented on here.

Given by the number of comments and requests that I get about this topic, these two posts from EViews are "must read" items for a lot of you.

And the great news is that there's a third post on the way, and this one will focus on implementing ARDL Modelling/Bounds Testing in EViews,

Great job!

© 2017, David E. Giles

The Stock Market Crash - VECM's & Structural Breaks

A few weeks ago, Roger Farmer kindly drew my attention to a recent paper of his - "The Stock Market Crash Really Did Cause the Great Recession" (here). To whet your appetites, here's the abstract:

"This note shows that a big stock market crash, in the absence of central bank intervention, will be followed by a major recession one to four quarters later. I establish this fact by studying the forecasting ability of three models of the unemployment rate. I show that the connection between changes in the stock market and changes in the unemployment rate has remained structurally stable for seventy years. My findings demonstrate that the stock market contains significant information about future unemployment."

Granger Causality

It's interesting, to me, that the posts on this blog that have received (and continue to receive) the most hits are those relating to Granger causality. Or, more correctly, testing for Granger non-causality.

The top one of all time remains, "Testing for Granger Causality". (Maybe it's the catchy title?) Then, just behind "How Many Weeks Are There in a Year" (which has nothing to do with causality - at least, not in any  obvious sense), comes "VAR or VECM When Testing for Granger Causality?"

Moreover, in addition to the many comments/questions that are published with those posts, I get numerous emails on this topic - almost on a daily basis.

Of course, some of these are pretty predicable - essentially, they are asking me to do give them a research project; tell them how to write their paper; or else they want to me to tell them how to complete an assignment for some course they're taking!

But then there are the many, many thoughtful emails that ask interesting questions, and raise all sorts of issues that get me thinking. I really enjoy responding to as many of these as I can manage.

So, I've been thinking.

Is there a demand for a short monograph on testing for Granger causality, with the emphasis on the practice, not the theory. In other words, a "how to do it properly" book for non-specialists, with lots of real-data examples.

Any thoughts on this?

• Is there a need?
• What format should it take - printed or e-book?
• Does this sounds like something that might interest you and/or your students?

I'll be interested to see your feedback.

© 2013, David E. Giles

My "Must Read" List

I have to confess that the number of items on my list of papers that I really must read (very soon) is rather large. My excuse is the same as everyone else's - too many papers, too little time. However, here's a small selection of of some of the papers that I've added to that list recently:

Surplus-Lag Granger Causality Testing

My previous posts (here, here, and especially here) on Granger causality testing have attracted more interest than I anticipated. One of the things that I've discussed at some length is the "surplus-lag" approach that can be used when the data are possibly non-stationary and possibly cointegrated. In particular I've talked about the Toda and Yamamoto (1995) procedure, but there are alternatives such as those introduced by Dolado and Lütkepohl  (1996) and Saikkonen and Lütkepohl (1996).

These modifications to the standard approach to testing for Granger (non-) causality are needed to ensure that the Wald test statistic has its usual chi-square asymptotic null distribution. You can't just test in the usual way unless the data are stationary. In fact, the "surplus lag" approach has advantages even beyond those that we knew about already.

VECMs, IRFs & gretl

In a comment on my post yesterday, "psummers" kindly pointed out that the free econometrics package, gretl, will also produce confidence intervals for Impulse Response Functions (IRFs) generated by a VECM.

I had an earlier post about gretl, and here is a very brief run-down on using it to produce those VECM-IRF confidence intervals.

Impulse Response Functions From VECMs

In the comments and discussion associated with an earlier post on "Testing for Granger Causality" an interesting question arose. If we're using a VAR model for constructing Impulse Response Functions, then typically we'll want to compute and display confidence bands to go with the IRFs, because the latter are  simply "point predictions". The theory for this is really easy, and in the case of EViews it's just a trivial selection to get asymptotically valid confidence bands.

But what about IRFs from a VECM - how do we get confidence bands in this case? This is not nearly so simple, because of the presence of the error-correction term(s) in the model. EViews doesn't supply confidence bands with the IRFs in the case of VECMs. What alternatives do we have?

VAR or VECM When Testing for Granger Causality?

It never ceases to amaze me that my post titled "How Many Weeks are There in a Year?" is at the top of my all-time hits list! Interestingly, the second-placed post is the one I titled "Testing for Granger Causality". Let's call that one the number one serious post. As with many of my posts, I've received quite a lot of direct emails about that piece on Granger causality testing, in addition to the published comments.

One question that has come up a few times relates to the use of  a VAR model for the levels of the data as the basis for doing the non-causality testing, even when we believe that the series in question may be cointegrated. Why not use a VECM model as the basis for non-causality testing in this case?