Revenue forecasts yield estimates, reliable or not, that budgetary actors may depend on to determine how best to allocate resources. Bretschneider et. al., argued recently in Policy Sciences (1988; see also Bretschneider and Schroeder, 1985;Bahl and Schroeder, 1984;Schroeder, 1982), that forecasts can 'influence budgetary choices.' The objective of this study is to determine which factors can predict the influence of forecasting on budgetary decisions. Is the responsiveness affected by the process used to generate the forecast, by the complexity of the forecasting techniques (regardless of the real accuracy of their results), or by the budgetary process? In an analysis of fifteen U.S. counties, MacManus and Grothe (1989) argue that the forecast process, forecast method, computer technology and experience with forecasting contribute to improved forecasts. In this study we will test the impact of these factors and the budgetary process not on any one policy objective, such as cutback management (MacManus and Grothe, 1989), but on budgetary decisionmaking in general.Budgetary decisions are often influenced by many factors, including politics, the economy and technology, but the impact of each influence is often unknown until after the fact. Of the above factors, approaches to budget reform have been guided by the notion that technology, or at least more advanced technology, is good for budgeting. While the validity of this hypothesis has been debated since the onset of performance budgeting, which strongly emphasizes the use of technical performance measures (Schick, 1971: Chapter 3; and more recently, Cope, 1987;Grizzle, 1987;and Klay, 1987), GFOA, ICMA and others continue to offer budget technology to governments, including computerized accounting systems, investment management over the wire, and means to study the historical financial condition and estimate future revenues. Whether these 'advances' are better than the old ways is not for certain, but to the degree they are perceived by key budgetary actors as being better, then there is a good chance that their results will be incorporated into the budgeting algorithm. 334