Chapter 28 The Econometrics of Military Arms Races

Abstract

Abstract Arms races – enduring rivalries between pairs of hostile powers, which prompt competitive acquisition of military capability – appear to be a pervasive phenomenon. From the past Cold War competition, between the US and the USSR, to present regional antagonisms, such as India and Pakistan, arms races remain a matter of continuing concern. This chapter reviews the econometric issues involved in estimating models of military arms races, of the competitive acquisition of military capability by hostile powers. As econometrics involves the synthesis of theory, data and statistical methods, in reviewing the econometrics of arms races, as much attention is paid to theory and data as to statistical methods. After discussing the choice of data and the theoretical issues in specification, we then examine four types of model: time-series estimation of classical Richardson type action-reaction models, using India and Pakistan as an example; Markov switching estimation of game-theory type models, using Greece and Turkey as an example; cross-section models and panel models. Our first general conclusion is that the theory suggests that the parameters of arms race interactions are unlikely to be constant over time and the empirical literature largely confirms this. Nonetheless, cross-section and panel estimates may be useful in that they allow estimation of average interaction effects, which may allow one to calculate the costs of the spill over effects of increases in military expenditure in one country. Our second general conclusion is that globalization means that one cannot confine attention to a bivariate interaction between two countries without taking account of the wider strategic context. The emphasis in this literature is on quantitative-symmetric arms races, because those are easier to estimate, but this emphasis may be misleading, qualitative-asymmetric arms races, particularly between governments and their non-governmental opponents, may be much more important.