One way spending agencies try to preserve or enlarge their claims on the central budget is to submit a funding request for only the first phase of a large project without specifying what later phases will cost. Because projects are difficult to cancel once begun, the best way to avoid ballooning costs is to require that no project begin without a full picture of projected future costs. "Costs" include not merely capital costs but all recurrent resources needed to complete and operate the project. The recurrent cost implications of investment decisions are often understated or overlooked. This may be because recurrent and development budgets are drawn up by separate processes, even by different groups of people, with little or no account taken of their complementarities. The problem can also be one of inadequate information. Country-specific norms, established through empirical investigation of ongoing projects, can be useful as a rough guide in forecasting recurrent costs.
Botswana's planning and budgeting system is an example of attention to the recurrent cost implications of investment spending. In preparing the the sixth National Development Plan (covering fiscal 1985-86 to 1990-91), each sector ministry was asked to list the programmes it needed to carry out its sector policies, presenting them in summary form with a brief description of each project, its purpose and its cost in both capital and recurrent expenditure. The investment ceilings were then determined from the overall targeted growth of recurrent spending, itself reflecting ceilings for use of skilled employees. The historical relation between recurrent and capital spending was an additional guide. Investment in excess of ceilings was allowed only if a ministry could demonstrate that such investments would require no further allocations from the recurrent budget.