Another look at the Relationship between Telecommunications Investment and Economic Activity in the United States

Abstract

In a recent issue of this journal, Beil et al. (2005, hereafter, BFJ) examined the relationship between telecommunications investment and economic growth in the United States over the period 1947–1996. Based on results from Granger-Sims test, BFJ conclude that ‘… investment by telecommunications firms is caused by, but does not cause, economic activity, and the findings are robust across lag lengths’. However, using another version of the Granger causality test due to Toda & Yamamoto (1995) and the same data set transformed into natural logarithms, the evidence in this paper indicates a feedback where there was a bi-directional causality between telecommunications investment and economic growth. This seems to suggest that policies aimed at stimulating the US economy by accelerating investment in the telecommunications sector may be successful.

Keywords:

• Telecommunications investment
• GDP
• modified Granger causality

Acknowledgements

The author is extremely grateful for the constructive critical comments received from Professors Richard O. Beil, George S. Ford and John H. Jackson for an earlier version of this paper. The author is also grateful to the anonymous referee for constructive comments and suggestions. All referees of this journal including the Editor helped to fundamentally shape the presentation and the contents of the paper. The usual disclaimer applies and the opinions are the sole responsibility of the author.

Notes

¹Another novelty of the Toda & Yamamoto (1995) approach is that Zapata & Rambaldi (1997) have made the procedure easy to apply using Seemingly Uncorrelated Regression (SUR) in several of the available econometric packages in GAUSS, RATS, SAS and SHAZAM. In our calculation, we used TSP, RATS, SHAZAM and Microsoft software packages.

²Before testing for causality we tested the statistical adequacy of our model using several lag selection criteria. For all the lags used (1 to 4) we carried out several misspecification tests (see Pesaran & Pesaran, 1997 and Enders, 2004). These tests indicate that our model was appropriately specified. Results of these tests are available from the author on request.

³Using the logarithms of the first difference of the series and using the conventional Granger causality test used by BFJ we found evidence of a feedback between telecommunications investment and economic activity. Admittedly, the results are not robust to varying the lag length and also are not as robust as when using Toda & Yamamoto (1995) in the logs of the levels of the series.