Priority-Based Budgeting: What It Takes to Make It Stick

Priority-based budgeting asks a different question than incremental budgeting. Instead of "what did this department get last year and what changed", it asks "what programmes does this entity operate, what results do they produce, and how do they rank against each other". The logic is sound and the failure rate is high. The failures are consistent enough to be predicted.

Building the programme inventory

The inventory is a list of every discrete service the entity delivers, with its full cost — direct personnel, direct non-personnel, and an allocation of support functions. For a mid-sized city this is typically between two hundred and six hundred programmes.

Two decisions determine whether the inventory is usable. First, granularity: a programme should be something that could plausibly be expanded, reduced or eliminated as a unit. "Public Works" is too coarse; "street sweeping — residential" is about right. Second, cost allocation method: whatever is chosen, it must be documented and stable, because the entire exercise collapses if departments can argue about the denominator.

Scoring

Programmes are scored against results the governing body has adopted, plus a set of neutral attributes: mandate status, cost recovery, demand trend, and reliance on the entity as sole provider. The scoring is the part that attracts the most process design attention and the part that matters least. Any defensible scoring scheme produces roughly the same top and bottom quartiles. The middle two quartiles are where the scheme's arbitrariness shows, and the middle two quartiles are rarely where decisions get made.

What does matter is who scores. Self-scoring by the delivering department produces a distribution that is unusable. Cross-departmental peer panels produce something defensible and take longer. A hybrid — department self-assessment, peer panel review, published disagreements — is the arrangement that most often survives.

The failure mode

Here is the pattern. An entity completes the inventory, scores it, publishes a quartile chart, and identifies fifteen programmes in the lowest quartile. In the budget that follows, none of the fifteen are eliminated. Three are reduced. Two are moved to a different department. The remainder are retained with a note that the scoring did not capture some important dimension.

The second year, departments have learned what scores well and the distribution compresses. By the third year the exercise is a documentation requirement.

The cause is not bad faith. It is that no mechanism existed to convert a low score into a decision. Priority-based budgeting is a ranking method, not a decision rule, and it is often adopted as though the ranking would decide by itself.

What makes it stick

• A pre-committed allocation rule. Before scores are known, the governing body commits to something specific: for example, that the lowest-scoring decile is presented with an explicit reduction or elimination option, and that retaining it requires an affirmative vote with a stated reason. The reason need not be a good one — but it must be on the record.
• Reinvestment. Savings from eliminated programmes should visibly fund higher-ranked programmes rather than disappearing into the general fund. Departments cooperate with a process that can benefit them and resist one that only takes.
• A real transition path. Eliminating a programme has costs — severance, contract termination, communication with residents who use it. Budgeting zero for the transition guarantees the elimination gets deferred.
• Scoring stability. If the results framework changes every cycle, comparisons across years are impossible and the process loses its main analytical advantage.

Where it works best

Priority-based budgeting performs well when an entity faces a sustained gap it intends to close through service change, and when the governing body is prepared to defend eliminations publicly. It performs poorly as a general management tool in a stable revenue environment, where the effort greatly exceeds the decisions it informs. Entities in the second situation often get most of the value from a lighter exercise: a full-cost inventory of the twenty largest and twenty smallest programmes, refreshed every three years.