Sunday, January 27, 2013

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

21 comments:

  1. I'm sure students here would greatly benefit. I would definitely use this if I were teaching time series.

    ReplyDelete
  2. Does it need to be book length? How about a "practitioners" journal article/working paper?

    ReplyDelete
    Replies
    1. David - perhaps you're right. There's a lot to cover, though. Stationary and non-stationary data; structural breaks; nonlinear causality; indirect causality; instantaneous causality; asymmetric causality; data frequency issues;......
      Something to ponder!

      Delete
  3. Granger causality is used in financial trading strategies -- that may also be why.

    ReplyDelete
  4. ebook sounds good. Also, there is a built in market among undergraduates. Econometrics tends to be the most daunting thing for that group.

    ReplyDelete
  5. "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."

    I must admit, this could apply to a heck of a lot of topics in econometrics and especially in time series. I think there's a lack of good 'applied' econometrics books out there, they seem to be few and far between.

    But to answer your questions:

    (1) Yep I think there's a definite need for a monograph on GC.

    (2) I think an e-book would be perfect, it'll make it more accessible to students and practitioners.

    (3) Absolutely would interest me!

    ReplyDelete
  6. As a practitioner, I would buy this book.

    I think I would prefer printed to e-book. I like having my references on a shelf at my desk.

    Love the blog, by the way.

    ReplyDelete
  7. As a practitioner in the private sector, I'd love to see a short, practical book on the art and science of Granger causality testing. Unlike academics, we generally don't have access to JSTOR and the like; and, as I think a lot of people would agree, reading many papers about a topic and then trying to distill a consensus or set of best practices is not easy unless you've been closely following the literature. A short book that summarizes everything sure would be nice! I'd buy a few copies for myself and my assistants.

    ReplyDelete
  8. The sense I get, as an applied person, is that the need exists insofar as Granger (non)causality purports to offer some manner of establishing causality, it is interesting to support panel results with a weak instrument, especially if the series are sufficiently long. But the literature appears to have definitively moved away from the method, and if I were to include it in a paper, it would almost certainly be a form of alternative estimation approach/robustness test rather than as the baseline methodology. I understand that those who do research on the topic realize that there are many nuances to the issue (as is the case for anyone who does research on any method), but I would agree with David that a "practitioner's"-type article would probably be more welcome (although not so much a survey paper; something more akin to Roodman's very popular xtabond2 paper is what I'm thinking).

    ReplyDelete
  9. Would definitely be a market for such a book amongst grad students and undergrads. Mostly Harmless Econometrics has been very popular because of its more simplified, applied approach.

    ReplyDelete
  10. Maybe blog chapters? With data and code downloads as you are already doing! The format you are using for your blog posts works excellent for me as a practitioner!

    ReplyDelete
  11. I would also buy a dead-tree version. Also, I still have not found an applied time series book that I like.

    ReplyDelete
  12. Yes, a non-specialist's guide to testing for Granger causality would be very useful, for students and practitioners. I would prefer an e-book version :). I have benefited from some earlier posting by Professor Giles on Granger causality. Really looking forward to this.

    ReplyDelete
  13. I would be interested in such a text/file. While it may be a bit more advanced, I'd also be interested if it included some discussion on the topic of Bayesian VAR and/or Qual VAR (http://research.stlouisfed.org/wp/2001/2001-012.pdf).

    ReplyDelete
  14. Hello,

    First of all, thank you for this wonderful blog.

    I wanted to do a VAR model in my next memory since I have not studied the model before I wanted to know if my approach is correct or not?
    1 / Make a stationarity test of two variables (with ADF): the two variables are stationary fortunately
    2 / represent in the VAR model (note that I work with stata)
    3 / when I find p-value = <5% (critical value chosen) of a variable I can say it affects the other
    Could you confirm the accuracy of my approach? or I need another tests for relevant results
    Another question: when I do Granger causality is the same p-value that appears in the step 3 so not need to do it?

    I think I would prefer e-book to printed

    ReplyDelete
  15. Thank you, Dr. Giles. It would be much appreciated! I am sure some advice on the implementation within panel (unbalanced in particular) would be of great help. Beryl

    ReplyDelete
  16. Sure,
    There are not good practical-empirical books using examples with real data. The manual examples you may never see in practice where data has outliers, breaks, nonlinearities, far from normal distributions, etc. Showing examples with real data and open source software like R, would definitely be a great value.
    Good luck

    ReplyDelete
  17. This is an excellent idea, I already suggested in another comment. A book/article would be highly appreciated since there is much confusion around how to test for Granger-causality. I find the already existing blog entries very helpful, and I hope there are many more to come on related issues, or preferably, of course, a book/article on that. I'm looking forward to it!

    ReplyDelete
  18. Having extensively consulted your posts on causality testing for my honours thesis...yes...please write a book. Great help!

    ReplyDelete