the form of sophisticated modeling methods, but budgeting could be improved through improved data collection and new analyses that relate resource levels and uses to results.
Federal departments and agencies develop budgets by first estimating what level and mix of resources they will need to execute authorized or proposed activities, consistent with legal mandates and policy objectives. Resource estimates therefore reflect both cost information and policy choices about program objectives and means. As described in Chapter 5, the budget process requires that agencies develop estimates well in advance—typically 18 to 24 months—of the period for which funding is sought, adding to the challenge. Because the budget process is lengthy and spending demands are characterized by uncertainty, agencies find it challenging to accurately estimate their resource needs when budgets are drafted.
Agencies may take one of two broad approaches to developing budget estimates. The first and most straightforward is a high-level incremental approach.1 Starting from the recent pattern of budget requests and variances (e.g., supplemental requests, reprogramming, and rescissions), budget planners account for overall spending trends and adjust for any new information expected to affect future resource requirements. To caricature: “If in the past you believe evidence suggests the U.S. Marshals Service was underbudgeted by 2 percent, then in the future bump up the budget request for the U.S. Marshals Service by 2 percent, all else equal,” or “If in the past, the U.S. Marshals Service has made do with a flat budget, then in the future provide the U.S. Marshals Service with a flat budget unless and until new information justifies an increase or decrease.”
For policy and program areas in which the processes underlying budget demands are understandable and relatively stable, incremental methods often suffice to produce reasonably accurate estimates of future resource requirements. For many annually appropriated programs, this approach works quite well. For low-income housing subsidies, for example, the funding needed to sustain a given level of service is readily
1Incremental budgeting is the oldest and simplest approach to developing budget estimates for public programs (see, e.g., Schick, 2007, Chapter 1). The distinction made here between incremental and other technical approaches is highly stylized, and it does not address the institutional and political determinants of budget and appropriations decisions that often modify or supersede technical judgments. Moreover, empirical research on budgeting decision making has thrown doubt on whether incrementalism or any other decisionmaking model can explain trends in funding for agencies, programs, or budget accounts; for a convenient summary of this research, see Meyers (1994), pp. 1-18.