This paper estimates the relationships between alternative measures of US defense spending and economic growth and government spending growth. The estimated relationships are derived from a plausible partial adjustment model that is based on the common practice of multi-year defense contracting and budget requests and appropriations. An attempt is made to forecast defense consumption spending, defense investment spending and, hence, aggregate defense spending till 2010, using the alternative forecast methodologies that are validated in-sample prior to their use in out-of-sample predictions. Two major findings emerge from this study. First, defense spending in general, is tied more to trends in government spending than economic growth. Second, a variety of simple linear smoothing moving average forecast models (used individually or in combination) as well as a partial adjustment model do surprisingly well in historical simulations of all types of defense spending.

This study pursues two objectives. First, it estimates relationships between growth of US defense expenditure and economic growth and government spending growth, using quarterly and annual data for the period from 1947 to second quarter of 2007. Particular attention is given to estimation of specifications derived from a plausible partial adjustment model that relates the path of defense spending (aggregate defense spending as well as defense consumption spending and defense investment spending) to economic growth and growth in government spending. Fundamentally, a partial adjustment model of the dynamics of defense spending seems appropriate as it is in line with a common practice of multi-year defense contracting and defense budget appropriation requests by the US government. While there is some evidence in the literature regarding the effects of defense spending on the economy (Atesoglu, 2002; and Gerace, 2002), little or no evidence exists concerning the effects that changes in economic growth and government expenditure have on defense spending. After all, changes in economic activity impact government revenues which, in turn, affect the government expenditure of which defense spending is an important component, particularly in so far as requests for new incremental defense budget are concerned. This study attempts to fill this void by shedding some light on a few basic relationships between defense spending, economic growth and government expenditure.

Second, we conduct a forecasting exercise for two types of defense expenditure, namely defense consumption expenditure and defense investment expenditure (the former constitutes the majority of defense spending), using two types of forecasting models. The first is moving average models of varying lengths, ranging between 5 years and 10 years, with one year increments. Since spending levels are likely to exhibit mean non-stationarity that renders any estimated regression parameters spurious and uninformative, we resort to models other than regression models such as moving average models, when working with dollar spending levels; particularly because these models do not require estimation of any parameters, in addition to having a wide appeal due to their simplicity and usefulness in time series forecasting. Moving averages are estimated on a rolling basis to incorporate more recent information about the series being forecasted.

Applied Economics Journal, Economic Growth, Gross Domestic Product, GDP, Federal Government, Augmented Dickey-Fuller, ADF, Error Correction Model, ECM, Cointegration Tests, Cointegration Analysis, Federal Government, Root Mean Square Error, RMSE.