The first contribution cycle is closed, and a bad assumption closed with it
The Ministry of Labour and Employment notified the Code on Social Security (Central) Rules, 2026 on 8 May 2026, operationalising the Code's provisions on provident fund, employee state insurance, gratuity, maternity benefit, employee compensation, and gig and platform workers. For quick-commerce, ride-hailing and delivery platforms, the part with a hard deadline was the contribution machinery. The cycle runs on two dates: aggregators assess and deposit a provisional social-security contribution in Form-XX by 30 June each year, then file the final return in Form-XXI, with any balance contribution, by 31 July.
Both dates for 2026 have now passed. Finance teams have computed the levy, deposited it, filed the return, and booked a new statutory cost line somewhere between one and two percent of annual turnover. And in more than one boardroom, the conclusion being drawn is that worker protection is now handled, so the group personal accident and liability programme is a candidate for cuts at the next renewal.
That conclusion is wrong, and expensively so. The levy is a contribution to a benefit fund. It buys registered workers access to notified schemes. It does not indemnify the platform against anything, and it does not pay the platform's contractual promises to its riders. The rest of this post separates what the levy actually does from what the commercial programme still has to do, and identifies the one place where genuine overlap justifies renegotiating limits rather than dropping cover.
What the levy buys: fund mechanics and who qualifies
Under the 2026 Rules, aggregators must contribute not less than 1% and up to 2% of annual turnover to a Social Security Fund for gig and platform workers, capped at 5% of the amount payable to such workers. The money goes into a fund administered under the Code, out of which notified welfare schemes pay benefits to registered workers.
Who gets those benefits is narrower than most platform CFOs assume. The qualification threshold is a tenure test: a gig or platform worker must have been engaged at least 90 days with one aggregator, or 120 days cumulatively across aggregators, in the previous financial year. A rider onboarded in April who is injured in June has not crossed the threshold. High-churn fleets, which describe most quick-commerce dark-store operations, will always carry a meaningful cohort of workers the fund does not yet cover.
For workers who do qualify and register, the benefit stack is real. Registered gig and platform workers are covered for health insurance up to Rs 5 lakh per family under PM-JAY, plus accident, disability and life cover under central and state welfare schemes. That is a floor worth having. But every element of it is a benefit owed by a fund or scheme to the worker, on the scheme's terms and timelines. Nothing in it runs to the platform.
Three claims the fund will never pay
Test the levy against the three loss scenarios that actually keep platform risk managers awake.
First, the third-party claim. A delivery rider runs a red light and puts a pedestrian in hospital. The pedestrian's lawyer sues the rider and the platform, arguing the platform controlled the rider's routing, incentives and delivery windows. This is a third-party liability claim against the platform itself. The Social Security Fund owes the pedestrian nothing and owes the platform nothing. Only motor third-party cover on the vehicle and the platform's own public liability programme respond, exactly as we set out in our earlier post on aggregator liability exposure.
Second, the disablement claim. A rider loses a hand in an accident. The welfare schemes may eventually pay a notified disability benefit, subject to the rider having crossed the 90-day or 120-day threshold, having registered, and the scheme processing the claim. A group personal accident policy pays a contractual permanent-partial-disablement percentage of a capital sum the platform chose, on a settlement timeline the broker negotiated. The rider's family will notice the difference between the two, and so will the journalists who cover gig-worker deaths.
Third, the platform's own promises. Most rider agreements and city launch announcements commit to accident cover of a stated amount. That commitment is contractual. If the platform points an injured rider at PM-JAY and a state scheme instead, it invites a breach claim and a news cycle.
The four commercial layers that survive Form-XX intact
For an aggregator running riders and driver-partners, the programme after the 2026 Rules still has four layers, and the levy sits beside them as a fifth statutory line.
- Group personal accident (GPA). Death and disablement cover for the rider fleet, with a defined capital sum, a disablement scale and, where negotiated, a weekly benefit for temporary disablement. This is the layer most often confused with the fund, because both respond to accidents. The difference is who owes what: the GPA insurer owes the platform's chosen sum on a contractual trigger; the fund owes a scheme amount on scheme terms.
- Group mediclaim (GMC). Hospitalisation cover with a cashless network the platform controls. PM-JAY access for registered workers changes the shape of this layer (more below) without eliminating it.
- Motor third-party and own-damage. Statutory third-party cover on every vehicle regardless of the levy, plus own-damage where the platform owns or leases the fleet.
- Platform liability. Public liability, and professional indemnity where the platform handles money or gives advice, responding when the claimant sues the platform rather than the rider.
A useful discipline at renewal is to write each layer's answer to one question: who receives the money? The fund pays workers. GPA and GMC pay workers, faster and in amounts the platform set. Motor third-party and platform liability pay outsiders and defend the platform. No line item on that list pays the same person twice for the same loss, which is why none of them collapses into the levy.
Where the overlap is real enough to renegotiate
None of this means the programme should renew unchanged. There is one genuine overlap, and brokers should price it rather than ignore it.
For the registered, threshold-crossing cohort, PM-JAY provides hospitalisation cover up to Rs 5 lakh per family, and the notified schemes add accident, disability and life benefits. That is a real base layer for that cohort. Two structural moves follow.
Restructure GMC around the PM-JAY floor
Where a meaningful share of the fleet is registered, the GMC conversation shifts from first-rupee cover to what sits above and around the scheme: cashless access at the hospitals riders actually reach, coverage for the unregistered cohort, and speed. A platform with strong registration rates can take that data to the insurer and argue for restructured terms on the registered segment rather than paying twice for the same first Rs 5 lakh.
Re-base GPA sums with the scheme benefit in view, for part of the fleet only
The scheme's accident and disability benefits give registered workers a statutory recovery that did not exist two years ago. That can support a conversation about GPA structure for the registered cohort. It supports nothing for the rest: workers inside their first 90 days, workers who never registered, and workers whose cumulative 120-day count spans aggregators that have not reported cleanly. For that cohort, GPA is the only accident cover in existence, and its sum insured should hold.
One more caution on geography. The central fund does not absorb state regimes. Karnataka's transaction-level welfare fee, which we covered when the February 2026 notification landed, runs in parallel with the central levy. A multi-state platform is now paying into more than one statutory pool while carrying one national insurance programme across all of them.
Your compliance filings are now your underwriting submission
The contribution machinery has a side effect worth money at renewal. To compute the levy and its cap, the platform had to produce audited annual turnover and the total amount payable to gig and platform workers. To establish who qualifies, it had to build engagement records against the 90-day and 120-day thresholds. Those are exactly the exposure documents GPA and motor underwriters have been asking gig platforms for since these fleets first came to market.
Tenure data is the sharpest of these. Accident frequency in delivery fleets concentrates in the early weeks of a rider's engagement, and until now most platforms could not evidence their tenure mix. The threshold-tracking the Rules force on aggregators produces that evidence as a by-product: what share of the fleet crossed 90 days, what churned out before it, how the mix moves quarter to quarter. A platform whose records show a maturing, longer-tenured fleet should put those records in front of the underwriter and ask for the rate to reflect them.
The amounts-payable figure does similar work on the liability side. It is a clean proxy for activity volume, and it lets the underwriter scale exposure by what the platform actually paid out to workers rather than by a headcount that includes dormant accounts. Insurers will also read the filings themselves as a signal. An aggregator that deposited Form-XX on time and reconciled Form-XXI without a scramble presents as a better-run risk than one that cannot say which of its workers are registered, and policy wording negotiations tend to go better for the former.
What to do before the next cycle and the next renewal
The 2027 cycle will arrive on the same calendar: provisional deposit in Form-XX by 30 June, final return in Form-XXI by 31 July. Between now and then, five actions put the insurance programme and the statutory position on the same map.
- Build the one-page exposure map: the central levy and any state fees as statutory lines, then GPA, GMC, motor and platform liability as commercial layers, each annotated with who receives the money and on what trigger.
- Split the fleet into registered and unregistered cohorts using the threshold data, and quantify how much of the workforce actually has fund and PM-JAY access today.
- Reopen GMC and GPA structure for the registered cohort only, with a written collateral-source position from the insurer before anything binds.
- Check the insured-person definitions in every wording against the Code's gig-worker and platform-worker definitions, a mismatch we flagged when the Code first came into force and which the Rules make more pressing, not less.
- Package the Form-XX and Form-XXI figures, plus tenure records, as the renewal submission.
The closing message for any platform CFO tempted to treat the July filing as the end of the matter: the levy is what the state charges you for engaging gig workers. Insurance is what stands between the platform and the claims those workers, and the people around them, will bring. Paying the first does not shrink the second by a rupee.