Review: IMF Publishes Budget Classification Technical Note
The International Monetary Fund has published a new Technical Note about Budget Classifications. The IMF has been advancing the sharing of PFM knowledge with a number of recent notes that we’ve commented on including:
- Fiscal Rules-Anchoring Expectations for Sustainable Public Finances
- Modernizing Cash Management
- Commitment Controls
- A Basic Model of Performance-Based Budgeting
- Transition to Accrual Accounting
The Budget Classification Technical Note was authored by Davina Jacobs, Jean-Luc Hélis, and Dominique Bouley of the IMF Fiscal Affairs Department. This technical note provides a good case for the importance of budget classifications. It provides some good high-level technical advice in designing budget classifications. Some of the assertions, though, do not appear to follow standard practice. Sequencing PFM reform is considered a good practice. This technical note generally supports this approach. But, the implied sequence does not appear to follow the typical pattern. We agree that the budget classification is ”one of the fundamental building blocks of a sound budget management system.” Perhaps even the most important building block – we often say that all government financial management begins with the Chart of Accounts. (In our world, the COA includes budget classifications.) The budget classification enables controlling government finances, following the budget law.
Issues to consider:
- “Separate presentation in the accounting classification to ensure the proper recording of budgetary transactions” does not seem to follow typical practices. The (COA) Chart of Accounts should include all budget and accounting classifications. That does not mean that accounting data entry is more complex in automated systems. And, budget preparation and execution can be managed by filtering out any unnecessary accounting details.
- “Once established on a sound basis, a classification scheme should not be substantially changed unless there are strong reasons; a stable classification facilitates both the analysis of trends in fiscal policy over time and intercountry comparisons.” This is an interesting ideal. In practice, governments adapt the classification scheme to improve outcomes. And, as indicated in Section VII, governments are constantly reforming. So, an ideal budget classification for any government, should it exist, often requires time to implement because of legal reform. And, there is no reason why changing details in the classifications can inhibit intercountry comparisons. Like with GFS, the intercountry comparisons can roll-up from the detailed COA. And, financial systems with multi-year Charts of Accounts enable comparisons such as viewing the 2007 fiscal year based on the 2010 fiscal year classifications.
- The technical note recommends that 3 mutually exclusive classifications, administrative, economic, and functional should be used. These would represent segments within the COA. (Of course, the standard accounting or object segment is needed to support accounting functions.) The note suggests that financing source should be considered as an additional classification. This does not seem to be wise for emerging countries that have a high percentage of Official Development Assistance. The use of a financing source segment enables governments to track donor conditions. This makes it more likely that governments, rather than 3rd parties, will be entrusted in executing donor funds. This reduces transaction costs.
- The addition of a program segment is typically sequenced later. The example provided shows how the addition of the program concept can extend the administrative segment. This approach is effective only when administrative units own programs. This is sometimes not the case. Expenditures can be controlled when there is a separate segment for program and administration. Automated financial systems support valid code combinations so that the 1 to 1 or 1 to many association between program and administration can be modeled and controlled. We have found a higher likelihood that functional and program classifications share the same segment.