The Stochastic Process of Alliance Formation Behavior

This article attempts to examine alliance formation behavior as a stochastic process amenable to systems interpretation. We test two hypotheses linking the systems characteristics of bipolar distribution/tightly knit bloc structure to serial dependence and stability of alliance formation behavior and multipolar distribution/loosely knit bloc structure to randomness and process instability.For testing these hypotheses, we have introduced a modeling strategy whereby identification of serial dependence and parameter estimation are accomplished by means of stochastic difference equations and time series autocorrelation and partial autocorrelation functions, and systems interpretation is facilitated by employing its continuous analogue in differential equation form.We analyze data on three sets of observations corresponding to three distinct periods of international relations: 1815–1914, 1919–1939, 1945–1965. The results do confirm the proposed hypotheses revealing randomness and instability of alliance formation behavior in the multipolar/loosely knit periods of 1815–1914 and 1919–1939 and serial dependence and stability in the bipolar/tightly knit 1945–1965 period. Finally, we consider the implications of these results for the earlier findings, the balance of power model, and the probability of war.