Evaluating Alternative Foreign Policy Decision-Making Models

The focus of this study is the policy-making process which has as its principal output the annual U.S. defense expenditure total. A Richardson-type arms race model and a Davis, Dempster, and Wildavsky-type organizational politics model have been identified as contending but not mutually exclusive models of this process. Underlying each model is the assertion that a series of inputs is translated in a specified manner into the U.S. defense expenditure total. A systematic empirical evaluation is undertaken to determine the ability of the models to forecast by using each of them to generate a series of historical and ex post forecasts of the U.S. defense expenditure time series from 1954 to 1973. Tests are made in order to determine which model generates the more accurate forecasts and whether these forecasts are adequate. The results of the empirical test indicate that the forecasts of the two models are indistinguishable both from one another and from a naive model. Several reasons are given for these results, and the paper closes with suggestions for further research.