On 17 September the government will relaunch PM Vishwakarma. The scheme turns three that day, and the briefings circulating since Independence Day point in one direction: a bigger credit line for the artisan, somewhere north of the present ceiling of three lakh rupees.
I have watched this move made before, in other schemes, from inside the Ministry. It is the easiest announcement in the building to draft and the hardest one to defend two years later. Because the artisan under this scheme is not being turned away at three lakh. He is being turned away at one.
What three years of numbers actually say
Take the official position first, because it is genuinely impressive in parts. By March 2026 some 30 lakh artisans had registered, 26.7 lakh had cleared skill verification and 23.7 lakh had completed basic training. Nearly 26 lakh toolkit e-vouchers had gone out. On the training side, this is one of the better-executed skilling operations the Ministry has run — roughly four out of five registrants moved all the way through a three-stage verification and into a classroom.
Then the same press release reports the credit outcome: 5.9 lakh loans approved, worth ₹5,050 crore.
Sit with those two figures together. Twenty-three lakh trained artisans, six lakh loans. And divide the second pair: ₹5,050 crore across 5.9 lakh accounts is an average ticket of about ₹85,600. The first tranche under this scheme is capped at one lakh. The second, available after repayment, at two.
An average below the first-tranche ceiling means the second tranche is, for practical purposes, not happening. The reporting from mid-August 2026 confirms the shape: ₹5,259 crore across a little over six lakh loans, an average of roughly ₹87,650. Five months of movement produced about ₹200 crore of additional disbursement on a scheme with a ₹13,000 crore outlay. The funnel has not slowed. It has closed.
A 28 per cent sanction rate is not an awareness problem
The number that explains all of this has been in the public domain since early 2025, and it has not improved much since. Of about 13.93 lakh loan applications that had reached banks by then, 3.9 lakh were sanctioned. Seven lakh seventy-seven thousand were rejected outright. A further 1.57 lakh were non-compliant and 1.58 lakh artisans withdrew rather than take the loan on offer.
Twenty-eight per cent.
By November 2025 the public-sector banks were doing somewhat better on paper — around 36 per cent of processed cases sanctioned — while the regional rural banks, which sit closest to the artisan the scheme was designed for, were running at about 27 per cent on 1.92 lakh applications. State Bank of India alone had taken in over 5.3 lakh applications and sanctioned roughly 1.64 lakh of them.
Bankers, asked to explain, cite documentation, eligibility checks and credit appraisal norms. They are not being evasive. They are describing their job. A branch manager assessing a first-time borrower with a prior default, no collateral, no formal books and an unsecured exposure of one lakh rupees is making an entirely rational decision when he declines. The scheme's answer to him is an eight per cent interest subvention. Subvention lowers the borrower's cost. It does nothing whatever about the lender's risk.
Which produces the contradiction at the centre of the design. The screening filters hardest against exactly the artisan the scheme exists to reach. An applicant with a clean credit record, documented income and a working relationship with a bank branch will clear it — and that applicant, more often than not, did not need a subvented government loan in the first place.
The budget line does not agree with the announcement
Here is the detail I would want on the file before signing off on a relaunch. PM Vishwakarma was allocated ₹5,100 crore in the FY26 budget estimate, revised down to ₹4,400 crore during the year. The FY27 budget estimate is ₹3,861 crore — twelve per cent below the revised estimate and about a quarter below where the scheme started the previous year.
So the announcement scheduled for 17 September expands credit availability, and the allocation funding it has contracted for two years running.
There is a defensible reading. The sector's own advocates have argued that the smaller line reflects phased implementation and a deliberate shift of credit support onto the banking system rather than the budget. Fair enough as a design philosophy. But the banking system is precisely where the 28 per cent sits. Moving the weight of the scheme onto bank balance sheets while the sanction rate stays near a quarter is not a shift in delivery mechanism. It is a transfer of the bottleneck from a place the Ministry controls to a place it does not.
The Ministry already runs a scheme that banks the same person
What makes this avoidable is that the answer exists two corridors away.
Between FY22 and FY26 the Prime Minister's Employment Generation Programme supported 5.8 lakh projects with about ₹60,000 crore of bank lending — an average project size above ₹10 lakh, roughly twelve times what Vishwakarma manages per account. Same ministry. Same banks. Comparable borrowers, often the same villages.
The difference is not the artisan's capability. It is ₹13,450 crore of margin money subsidy disbursed over that period. PMEGP puts government money into the capital structure of the loan, where it absorbs first loss and changes the banker's arithmetic before the appraisal begins. Vishwakarma puts government money into the interest rate, where it changes nothing the banker is actually worried about.
The same logic holds at the other end of the market. The credit guarantee ceiling for micro and small enterprises was raised from ₹5 crore to ₹10 crore in April 2025, and guarantee approvals ran to ₹3.77 lakh crore in eleven months of that year. India has built a genuinely sophisticated guarantee architecture. It simply does not reach down to a ₹1 lakh artisan account, where per-account administrative cost swamps the margin and no bank will chase the cover.
Tamil Nadu declined to implement PM Vishwakarma in February 2025 and proposed its own artisan scheme instead. That was read at the time as a political gesture. Three years of sanction data suggest at least part of the objection was technical.
What 17 September should carry
A relaunch is a rare thing — a moment when a scheme's parameters are genuinely open. Four changes would be worth more than any increase in the headline ceiling.
Put a first-loss guarantee on artisan loans below one lakh rupees, funded from the existing outlay. If the subvention has to be trimmed to pay for it, trim it. A guarantee that moves a sanction decision is worth more to the artisan than three percentage points he never gets the chance to pay.
Pay the bank for the account. A modest per-account origination fee for a Vishwakarma loan acknowledges what everyone in the branch already knows — that a ₹1 lakh unsecured account with a first-time borrower costs more to originate than it will ever earn. Rural bank branches are not refusing this business out of indifference. They are refusing it because it does not pay for itself.
Release the conversion data by district. The dashboard already holds registrations, verifications, trainings and sanctions. Publish the ratio of first-tranche loans to trained artisans, district by district, every quarter. Nothing disciplines a delivery chain like a public denominator, and the districts doing this well would become visible overnight.
Reconsider the second tranche entirely. It is conditioned on repayment of the first and on adoption of digital transactions, and almost nobody is reaching it. Either the sequencing is wrong for a trade cycle that turns on festival demand, or eighteen months is too short a first tenure for a potter or a cobbler to service and re-borrow. Both are fixable. Neither is fixed by raising the number at the top.
The measurement that would settle the argument
There is a line I would put in every review note on this scheme: registrations are an input, and we have been reporting them as an outcome.
India has largely solved the formalisation problem. By March 2026 some 7.9 crore enterprises sat on Udyam and Udyam Assist between them — 4.72 crore and 3.21 crore respectively. Getting an informal enterprise onto a government list is no longer difficult. Getting capital into its hands still is, and every scheme in this space now fails at the same joint. It is the same structural point that runs through the MSME Development (Amendment) Bill that cleared both Houses this month — the statute was never the weak part; the machinery downstream of it was.
So when the relaunch is announced on Vishwakarma Jayanti, the figure to look for is not the new ceiling. It is loans sanctioned per trained artisan. Today that ratio is roughly one in four, and it has barely moved in a year.
Move it to one in two, and this becomes one of the most consequential credit interventions India has attempted for its informal workforce. Leave it where it is, and in September 2028 we will be reading another anniversary press release counting registrations.
