A ₹2 lakh loan costs almost as much to make as a ₹20 lakh loan. That is the whole story.

Yesterday I said lenders left the small ticket for arithmetic reasons, not appetite. Here is the arithmetic.

Almost every rupee it costs to make a loan is fixed with respect to the size of the loan.

A lead costs what a lead costs. The field visit is one officer, one trip, one shop, two hours, whether the file closes at ₹2 lakh or ₹20 lakh. The credit memo takes two to five days of human attention; a larger file takes longer, but not ten times longer. Documentation, e-sign, mandate and disbursal are an identical process. Servicing runs twenty-four months either way, and the smaller borrower generates more touchpoints, not fewer, because the cash flow is thinner.

Call that fixed cost to originate and service F. In a branch-and-RM model it lands near ₹18,000 a file. As a fraction of the loan:

F/P at ₹2 lakh = 9.0%
F/P at ₹5 lakh = 3.6%
F/P at ₹20 lakh = 0.9%

Same rupees. The drag is a hyperbola in ticket size — the smaller the loan, the larger the share of it eaten before you have taken a single rupee of credit loss.

Operating cost for this kind of lending runs at 11–15% of assets industry-wide. On a large ticket that is affordable. On a ₹2 lakh loan over 24 months it is most of the margin.

So every lender has been walking rightward along that curve, looking for where F/P stops hurting. The market's revealed answer is around ₹8 lakh — where the average ticket has landed.

Nobody chose that number. It is the solution to an equation.

Which is why "we'll go down-market next year" has been said every year since 2019 and has not happened. You cannot get there by trying harder, because F is physical. It is people, trips, hours and paper. Trying harder raises F.

There are two moves, and only two. Drive F towards zero, or stop it being flat per file, so the tenth loan costs a fraction of the first and the ten-thousandth costs almost nothing.

That is an architecture question, not an effort question.

Tomorrow: why you cannot solve it by hiring, which is the first thing everybody tries.