A Positive Model of Expenditure Growth: Toward Closure of the Organizational Process Theory of Budgeting
Abstract: Optimal control theory and martingale methods are more elegant ways of modeling dynamic, stochastic processes than alternatives such as Monte Carlo simulation methods and equally powerful, but they are nevertheless rarely used in our field, due in part, perhaps, to their unfamiliarity. In this article we illustrate their power by completing one of the more compelling computational models of organizational process extant: the Crecine-Padgett model of the budgetary process. This model very accurately explains spending at the program or agency level in the period in which a budget is executed in terms of last year’s spending for the same purpose and revenue growth. But it lacks predictive power because its treats revenue change as an exogenous variable. Consequently, to close this model, a theory of revenue and/or aggregate spending growth that is endogenous to organizational process theories is needed. This article offers (and tests) just such a theory for jurisdictions that face a hard budget constraint, using optimal control theory and martingale methods. This theory takes a jurisdiction’s existing revenue structure and its fiscal assets (savings) as given, and treats revenue and savings growth as continuous-time, continuous-state stochastic processes. This means that next year’s revenues and, thereby, funds available for spending can be predicted using current (in the period of budget formulation or enactment) state variables only. Combined with Crecine-Padgett model of the budgetary process, this means that we can predict and not merely explain spending changes at the program or agency level. Keywords: Process • Mechanism • Martingale methods • Optimal stopping models JEL Classification Numbers: H71 • H72
