Thursday, September 13, 2012

Granger Causality Testing With Panel Data

Some of my previous posts on testing for Granger causality (for example, here, here, and here) have drawn quite a lot of interest. That being the case, I'm sure that readers of this blog will enjoy reading a new paper by two of my colleagues, and a former graduate student of theirs.

The paper, by Weichun Chen, Judith Clark, and Nilanjana Roy is titled "Health and Wealth: Short Panel Granger Causality Tests for Developing Countries". Here's the abstract of their paper:
"The world has experienced impressive improvements in wealth and health, with, for instance, the world’s real GDP per capita having increased by 180% from 1970 to 2007 accompanied by a 50% decline in infant mortality rate. Healthier and wealthier. Are health gains arising from wealth growth? Or, has a healthier population enabled substantial growth in wealth? The answers to these questions have serious policy implications. We contribute to understanding dynamic links between wealth and health by analyzing the relationship between health (as measured by infant mortality rate) and wealth (as measured by GDP per capita) for a panel of 58 developing countries using quinquennial data covering the period 1960 through 2005. We examine for causal rather than associative links between these fundamental macro measures of economic development. The panel enables us to examine for causal links using several methods that differ in how cross-country and temporal heterogeneity is imposed: cross-country homogeneity with temporal heterogeneity and cross-country heterogeneity with temporal homogeneity. Under the latter case, we consider sensitivity to assuming fixed versus random causal coefficients. In addition, we explore robustness of outcomes to level of economic development (as measured by national income) and inclusion of another covariate (education)".
You can download a copy of the full paper from here.

The paper illustrates the application of various ways of testing for Granger non-causality in the context of panel data, so in that sense it makes a second really useful contribution for researchers who want to "get up to speed" on this important topic.

I hope that you enjoy reading this paper, and that the references below are also helpful!


References

Devlin, N., and P. Hansen. 2001. Health care spending and economic output: Granger causality. Applied Economics Letters, 8, 561-4.

Dreger, C., and H.E. Reimers. 2005. Health care expenditures in OECD countries: A panel unit root and cointegration analysis. International Journal of Applied Econometrics and Quantitative Studies, 2, 5-20.

Erdil, E., and I.H. Yetkiner. 2009. The Granger-causality between healthcare expenditure and output: A panel data approach. Applied Economics, 41, 511-8.

Holtz-Eakin, D., W. Newey, and H.S. Rosen. 1988. Estimating vector autoregressions with panel data. Econometrica, 56, 1371-95.

Hurlin, C. 2005. Granger causality tests in panel data models with fixed coefficients. Revue Economique, 56, 1-11.

Hurlin, C. 2008. Testing for Granger non-causality in heterogeneous panels, Working Paper, Laboratoire d’Economie D’Orleans, University of Orleans.

Hurlin, C., and B. Venet. 2003. Granger causality tests in panel data models with fixed coefficients. Mimeograph, Universitie Paris I.

Hurlin, C. and B. Venet. 2008. Financial development and growth: A re-examination using a panel Granger causality test, Working Paper, Laboratoire d’Economie D’Orleans, University of Orleans.


© 2012, David E. Giles

9 comments:

  1. Dear sir,
    i'm performing a granger causality test between FDI and labour productivity,capital productivity & TOtal factor productivity within an economy at sectoral level.(11sectors,period yearly from 2000 to 2012)

    so,sir could you advice me the steps in doing granger causality test for a panel data.
    for example: panel unit root test,etc,,,,

    thks!

    ReplyDelete
  2. dear professor,
    Could you advice me what are the following steps involves in doing granger causality test for panel data?

    for example:panel unit root test,etc,,,

    ReplyDelete
  3. Prof. Giles,

    Just found out your unbelievably wonderful blog.

    Could you please point to me the discussion where it is justified to use two (may be three) time series or panels to establish causality. I am thinking about Health, Income and possibly Education are used to study causality. On the other hand, in cross sectional models we expect to see many, many more controls. For example, health care expenditures, physicians per 1000, tobacco and alcohol consumption, ... Even then critics raise issues with omitted variables. My puzzle is why are there are no controls in the panel or VAR causality studies.

    I am hoping to see your explanation or lead to a source.

    Thank you for your wonderful blog.

    ReplyDelete
    Replies
    1. You can certainly include control variables in VAR causality studies if this seems appropriate.

      Delete
  4. Hello Professor and Audience, I'm wondering whether there is around any R package to deal with Granger causality with panel data.
    Thanks,

    ReplyDelete
  5. Replies
    1. In this post? None!
      In the other posts - EViews - as was clearly discussed.

      Delete
  6. Hello Prof
    I really do appreciate your painstaking effort in analysing causality test. Your blog is simply educating. I'm your core fan here and I have been reading most of the articles which do help me appreciate statistics and econometrics better. I have few questions which I will be glad to have the answers promptly.
    1. How can we carry out Markov switching Granger causality test with eviews 7?
    2. What is the difference between using dummies to account for structural breaks and using Markov switching Granger causality test?
    3. Can dummies be used for regime switching in Granger causality test?
    Thanks in anticipation of your prompt response.
    Nsisong Ekong(Nigeria)

    ReplyDelete