Skill Ratings Beat Deposit History on 6-Month Repayment

For an Indian lender staring at a loan application, the two most legible signals have traditionally been the applicant's repayment record and their deposit behaviour — the running balance in a savings account, the regularity of inflows, the average monthly credit. Both are backward-looking. Both describe what the borrower has already done with money someone else was watching. The more interesting question is whether a forward-looking measure — a demonstrated, verifiable skill rating — predicts six-month repayment behaviour better than deposit history does, and if so, what that implies for how training programmes in finance and banking should be designed and assessed.

Why Deposit History Is a Weak Predictor

Deposit history is a proxy for stability, not for competence. It tells a lender that money arrived, not that the borrower knows what to do with it. A salaried employee with a steady credit and a zero balance at month-end looks identical, in the data, to a small trader with the same pattern — yet their financial decision-making under stress may be entirely different.

This is where behavioural psychology becomes useful rather than decorative. Kahneman and Tversky's work on loss aversion established that people weigh losses roughly twice as heavily as equivalent gains, and that this asymmetry distorts decisions in predictable ways. A deposit record captures the outcome of those decisions but not the decision process. Two borrowers can produce identical bank statements while one is running a disciplined cash-flow plan and the other is simply lucky with timing.

There is also the problem of variable-ratio reinforcement. B.F. Skinner's schedules of reinforcement showed that behaviour maintained on unpredictable reward is far more resistant to extinction than behaviour maintained on consistent reward. A borrower whose income arrives irregularly — common among gig workers, commission agents, and small vendors across Indian metros — may develop erratic deposit patterns not because of poor discipline but because the reinforcement schedule itself is erratic. Penalising that pattern in a credit model is a category error.

What a Skill Rating Actually Measures

A skill rating in finance and banking, properly constructed, is not a test score. It is an observed, repeated demonstration of specific competencies: reading a balance sheet, computing effective interest on a reducing-balance loan, distinguishing nominal from real returns, building a household budget that survives a 20% income shock, or correctly identifying the true cost of a gold loan versus a personal loan.

The distinction matters because these competencies are measurable under conditions that resemble the borrower's actual environment. A training programme that ends with a proctored exam measures recall. A training programme that ends with a simulated cash-flow crisis, where the participant must allocate a fixed pool of rupees across competing obligations over six simulated months, measures something closer to judgment.

Research on decision-making under uncertainty — particularly the work following Gigerenzer on heuristics and Kahneman and Tversky on framing — suggests that what separates competent from incompetent financial actors is not raw intelligence but the possession of well-calibrated mental models. People who understand that a 24% annual rate on a reducing balance is not the same as 24% flat, or that a 3% processing fee on a ₹50,000 loan is a meaningful addition to its cost, make systematically different choices. Those choices show up in repayment.

The Six-Month Window as a Test Case

Six months is a useful horizon because it is long enough to include at least one income disruption — a medical expense, a festival-season cash crunch, a delayed payment from a client — and short enough that the borrower's underlying skill set has not had time to change dramatically. If a skill rating predicts behaviour over that window better than deposit history does, the implication is that the rating is capturing something durable about the borrower.

Consider a concrete illustration from the training sector. Several Indian skilling programmes in financial literacy have experimented with pre- and post-training assessments that include scenario-based items rather than factual recall. A participant might be shown a household with ₹18,000 monthly income, ₹6,000 in existing EMIs, and a ₹40,000 emergency, and asked to decide whether to take a new loan. Scoring is based on the reasoning shown, not just the final answer. When such scores are tracked against subsequent repayment behaviour — even informally, through programme alumni surveys — the correlation tends to be stronger than the correlation between deposit regularity and repayment. This is not a published finding in a peer-reviewed journal; it is an observation repeated across enough programmes that it deserves formal study.

The Overlap With Competitive Play and Risk-Taking

There is a second, less obvious source of skill signal: competitive environments where participants make repeated decisions under uncertainty with real consequences. Chess, debate, mock trading competitions, business plan contests, and structured case competitions all produce observable ratings. What makes these interesting is not the competition itself but the decision architecture.

In a well-designed competitive format, participants face loss aversion directly. They must decide whether to protect a lead or press an advantage. They experience variable-ratio reinforcement when outcomes are noisy. They learn, or fail to learn, that risk-taking without an edge is not risk-taking but expense.

This is precisely the cognitive training that finance and banking programmes claim to deliver but often do not, because they teach content rather than decisions. A candidate who has competed in a structured mock-portfolio competition for six months has demonstrated something a deposit history cannot show: that they can hold a position, revise it when evidence changes, and absorb a loss without abandoning the plan entirely.

The connection to repayment is direct. A borrower who panics at the first missed payment and stops communicating is a worse credit risk than one who misses a payment, recognises it as a temporary setback, and renegotiates. That capacity for calibrated response under stress is a skill, and it is trainable.

Designing Ratings That Predict Repayment

If skill ratings are to displace deposit history as a primary signal, three design choices matter.

First, the rating must be behavioural, not declarative. Asking a participant whether they understand compound interest is worthless. Asking them to compute the total cost of two loan offers and choose is not.

Second, it must be repeated. A single assessment is a snapshot; a rating built over four or six assessment cycles with varying difficulty is a trajectory. Trajectories are more predictive than levels, because they reveal whether the participant is learning.

Third, it must include a stress component. The most informative moment in any training programme is when the participant's plan breaks — when the emergency expense exceeds the buffer, when the client defaults, when the market moves against the position. How they respond in that moment, and whether they respond differently the second time, is the signal.

For Indian lenders and training institutions, the practical implication is that assessment design is credit design. A programme that produces a credible, standardised skill rating is not just an educational intervention; it is a data-generation mechanism for underwriting populations that deposit history has always misjudged. The next step is not more content. It is better instruments.