NS Toor’s initiative to facilitate financial literacy ·

Banking India Update

— Independent · Daily —

Two-Choice Budget Prompts Halve Trainee Overruns by Case 4

A two-choice budget prompt cut trainee overruns by case four in a Mumbai credit-analyst simulation, showing how constraint framing sharpens allocation

Two-Choice Budget Prompts Halve Trainee Overruns by Case 4
Two-Choice Budget Prompts Halve Trainee Overruns by Case 4

In a Mumbai classroom of twenty-six credit analysts, each participant receives a simulated capital-allocation budget and a spreadsheet of live-looking exposures. The exercise is straightforward: allocate the budget across competing proposals and justify the trade-offs. What is not straightforward is the moment the facilitator replaces the open-ended instruction — "allocate the budget as you see fit" — with a two-choice prompt: "Option A: allocate ₹100 crore and stop. Option B: allocate ₹140 crore, but only if you can name the specific exposure you are dropping to fund the extra ₹40 crore." In a pilot across four cohorts of a commercial-banking training programme, the share of trainees who exceeded their sanctioned budget fell by roughly half by the fourth case in the sequence. The mechanism, I think, is not discipline. It is the structure of the choice itself.

The architecture of a two-choice prompt

The two-choice prompt is not a nudge in the soft sense. It is a constraint on the decision space. Instead of asking a trainee to decide how much to allocate, it asks them to choose between two pre-specified allocations, each with a named cost. The first option is bounded and clean. The second is larger but carries an explicit funding obligation — the trainee must identify what they are giving up.

This matters because the open-ended budget instruction is, behaviourally, a blank cheque. It invites the trainee to treat the budget as a target rather than a ceiling, and it offers no friction at the point of overrun. The two-choice version inserts friction exactly where overruns occur: at the moment of expansion. To take the larger allocation, the trainee must articulate a trade-off. Many do not, and the overrun disappears not because they were told no, but because the cost became legible.

There is a familiar behavioural concept here. Kahneman and Tversky's work on loss aversion established that losses loom larger than equivalent gains; a trade-off framed as what you give up is felt more sharply than the same trade-off framed as what you could gain. The second option in the prompt is deliberately loss-framed. It does not say "you may allocate more." It says "you may allocate more only by naming what you drop." The framing does the work that a rule would otherwise have to do.

Why case 4, and not case 1

The halving was not uniform across the four cases. In the first case, overruns were only modestly lower under the two-choice prompt than under the open instruction. By the fourth case, the gap had widened substantially. This pattern is consistent with what we know about learning under repeated structured decisions: the prompt is not a one-off intervention but a rehearsal. Each case gives the trainee another opportunity to practise the trade-off articulation, and the articulation becomes a habit rather than a compliance behaviour.

There is a second, less obvious reason. Case 4 in the sequence introduced a competing proposal that was superficially attractive but required cannibalising an existing exposure. Under the open instruction, trainees tended to fund the new proposal and retain the existing exposure, producing an overrun. Under the two-choice prompt, the cannibalisation requirement was explicit, and most trainees either declined the new proposal or dropped the existing exposure. The prompt did not make them more conservative; it made the structure of the decision visible.

This is the variable-ratio reinforcement insight applied to training design, though in an inverted form. Variable-ratio schedules — where rewards arrive unpredictably — are known to produce persistent, sometimes excessive behaviour, because the uncertainty itself is reinforcing. An open-ended budget instruction is, in a sense, a variable-ratio environment: the trainee does not know in advance whether a given allocation will be "rewarded" by the case's scoring rubric, so they keep expanding, hoping the next increment pays off. The two-choice prompt removes that uncertainty at the point of decision. It converts a variable-ratio environment into a fixed, legible one, and the persistent over-allocation behaviour drops away.

The competitive dimension

Training programmes in banking are, whether we admit it or not, competitive environments. Trainees compare allocations, defend proposals in front of peers, and are scored on outcomes. Competition amplifies risk-taking: the trainee who allocates more looks bolder, and boldness is read as competence in a room of peers. The open-ended instruction feeds this dynamic because there is no visible cost to being the boldest allocator.

The two-choice prompt changes the social calculus. The trainee who chooses the larger allocation must, in front of peers, name what they are dropping. That is a harder performance than simply allocating more. In the pilot, facilitators noted that the second option was chosen less often in group settings than in individual settings, and that the quality of the named trade-off — not the size of the allocation — became the thing peers evaluated. The competitive play shifted from who allocates most to who can defend a trade-off most precisely. That is a better competition for a finance training programme to host.

A concrete reference point

The closest published analogue is the work on implementation intentions — Gollwitzer's finding that specifying the how and when of a behaviour, rather than merely intending it, substantially increases follow-through. The two-choice prompt is an implementation intention for budgeting: it forces the trainee to specify the funding source before the allocation is made. The pilot's results are consistent with that literature, though the sample is small and the effect size should be treated as indicative rather than established.

A second reference point is the choice architecture literature associated with Thaler and Sunstein, which distinguishes between prompts that preserve all options and prompts that restructure the option set. The two-choice prompt is the latter. It does not forbid overruns; it makes them costly to articulate. That distinction matters for anyone designing training assessments, because it means the intervention is defensible on pedagogical grounds — it teaches trade-off reasoning — and not merely as a compliance trick.

What to build next

The obvious next step is to test whether the effect persists when the prompt is removed. If trainees internalise the trade-off articulation, overruns should stay low even under the open instruction by case six or seven. If they do not, the prompt is a crutch rather than a lesson, and the programme should say so.

A second step is to vary the framing of the second option. Loss-framed prompts ("name what you drop") and gain-framed prompts ("name what you add") should be compared directly, because the loss-aversion literature predicts a meaningful difference and training designers should know which one they are deploying.

A third step is to extend the prompt beyond budgeting. Credit approval, provisioning, and limit-setting all involve the same structure: an expansion decision with a hidden funding cost. If the two-choice prompt works for budgets, it is worth testing where else in the curriculum the same friction can be inserted — not as a rule, but as a question the trainee has to answer before the number gets bigger.