Defense spending and economic growth across the Taiwan straits: a threshold regression model

Abstract

This paper employs both linear and non‐linear models to investigate the relationship between national defense spending and economic growth for Taiwan and China. Using data from 1953–2000 on defense spending, GDP, import, export and capital, we find that China's defense spending leads that of Taiwan. There exists the phenomenon of an arms race between both countries when official Chinese data are used. On the one hand, feedback relations prevail between economic growth and defense spending growth in Taiwan. On the other hand, China's national defense is found to lead economic growth.

Keywords:

• Arms race
• Multivariate threshold models
• Defense spending
• Economic growth

ACKNOWLEDGEMENTS

An earlier draft of this paper was presented at the Third Annual Conference on Empirical Economics held in Pu‐Li, Taiwan on April 21, 2002. We are grateful to Professor Lee Torng‐Her for his helpful comments. Views expressed in this paper reflect those of the authors, not their respective institutions. We would like to thank three reviewers and editor for detailed comments and suggestions. All remaining errors are our own.

Notes

The suggestion by a reviewer on this issue is greatly appreciated.

† Corresponding author. E‐mail: [email protected]

The advantage of the non‐linear model lies in its ability to detect shifting interactions between defense spending and economic growth when rival's defense spending growth exceeds a threshold value.

According to Waller (1997), China's defense spending may well be underestimated as a result of budget falsification on the PLA's (People Liberation Army) revenue and the purchasing power parity of yuen. The accuracy of the spending from the official publication and SIPRI Yearbook cannot be determined either. As such, we first test the model using the figures from China's yearbook followed by that from SIPRI Yearbooks.

Only three equations (Δgd_t , Δy_t , Δgdr_r ) are reported, in order to analyze the causality between them.

Refer to Tsay (1998) for details of the multivariate threshold regression model.

We thank a reviewer who pointed out an important phenomenon: an arms race is occuring only in cases in which two defense expenditure rates are both on the rise. As such, using the rate of change in defense expenditure is a reasonable choice.

The cointegration result remains unchanged, irrespective of using the SIPRI Yearbook or Chinese Statistics Yearbook: there exists only one set of cointegration.