Forward Estimates: Rocking or Rolling?

Posted by Kris Kauffmann

In my view, it is now well established that using rolling forward estimates is a central element to the successful implementation of any Medium Term Expenditure Framework approach. By establishing the future costs of existing policies, for say 2-3 years after the budget, this provides a foundation for future budget planning processes, where the focus shifts to the costs verses benefits of changing those existing policies and their established costs. Multi-year estimates also support longer term macro fiscal as well as managerial planning. In recalling Australia’s MTEF reforms and describing the impact of rolling forward estimates, a senior official is quoted as telling the World Bank that “If you had to pick out the one thing that we have done above all others, this reform would be the most dramatic change.”

Yet we often see in developing and emerging market countries that while government may have adopted what appears to be forward estimates in their budget processes, they have not been able to achieve the dramatic changes in budgeting described above. It is a characteristic of these countries that the forward estimates, while published in the previous budget, are not accepted as being highly relevant for future budget processes.

There are features of Australia’s reform experience through the mid-1980s onward that provide some clues as to how using rolling forward estimates can become the centerpiece of an effective MTEF approach.

Firstly, and most obviously, the forward estimates need to reflect (fit within) the overarching top-down medium term macro-fiscal targets adopted by the government.  It is no good having an aggregate expenditure target of X for a future year if the forward estimates for that year add to X+Y as this would automatically imply future unknown policy changes - that undermines the relevance of the forward estimates.

Further, the forward estimates need to become the centerpiece of the mechanics of the budget process. That is, the budget process needs to start with a clear restatement that the forward estimates of last year are the baseline for all negotiations during the budget process – be they top down or bottom up.  Further weight is given to the status of the forward estimates if they are published and more so if there is a clear reconciliation table showing the changes to the original forward estimate baseline that occurred in developing the final budget and the new forward estimates. 

In a high IT capacity environment like Australia, it is highly beneficial to achieve the mechanical changes to the budget process through a well-designed budget development information management system (the budget module in the IFMIS). Such a system is designed to accommodate multiple years of estimates, where the forward estimates transparently reside within the system, the next year’s forward estimates roll forward to become the starting point for the new budget, budget submissions are formulated electronically in terms of proposed changes to the multi-year baseline, and where those proposed changes sit and wait pending analysis and various decision points.

Having the forward estimates embedded in an integrated IFMIS is also potentially useful for budget execution in that it provides a point against which multi-year expenditure commitments can be controlled in a cash accounting environment with single year appropriations.  Their meaningful use in the IFMIS for multi-year planning and control gives further weight to the prospect of improved ownership of the forward estimates by all stakeholders.

For this approach to work effectively there needs to be ownership of the forward estimates at the political level – something often missing in developing countries. It means that enthusiastic line ministers are not able to change their forward estimates without rigorous analysis in comparison to other new policy proposals and approval by Cabinet.  In Australia, a committee of the most influential ministers and the Prime Minister formed a budget committee of the Cabinet called the Expenditure Review Committee which took on the role of protecting the forward estimates and forcing all expenditure decisions through the budget process. 

Finally, it must be recognized that using rolling forward estimates often imply that a large portion of spending, the rump of expenditures already embedded and committed for funding in the forward estimates, do not directly and explicitly come up for policy review each budget cycle. MTEF can actually reinforce incrementalism! As such, to ensure the forward estimates remain relevant to the government policy objectives, the rolling forward estimates/MTEF approach should be complimented by reforms directed at ensuring periodic evaluation of programs. One outcome of evaluations is to direct decision making (top down and bottom up) to desirable changes to the forward estimates that can be initiated in the coming budget cycle. Originally in Australia this involved a system of periodic program evaluation but later evolved into a comprehensive performance management approach (for very useful discussion of evaluation in Australia, see publications by Keith Mackay for the World Bank).  Through these evaluation mechanisms, those expenditures embedded in the forward estimates have their relevance reaffirmed.

In summary, making rolling forward estimates the centerpiece of an effective MTEF approach requires significant effort, careful planning, and adequate IT capacity. What the Australian experience shows us is that simply publishing multi-year projections in budget documents is not sufficient.


Note: The posts on the IMF PFM Blog should not be reported as representing the views of the IMF. The views expressed are those of the authors and do not necessarily represent those of the IMF or IMF policy.