Designing the Trigger, Owning the Basis Risk: A Parametric Catastrophe Underwriting Manual for Indian Commercial Buyers

Through the 2026 heat season, parametric triggers spread faster across India than most brokers expected. The most visible examples sat at the informal-sector end. In Delhi and Faridabad, insurer-backed schemes placed through broker and non-profit partners paid roughly Rs 3,000 to several thousand women informal workers once defined temperature thresholds were crossed, with a top-up reported where the heat persisted for around ten days. Other north Indian cities saw similar wage-loss covers trigger. These pilots matter because they made a previously academic product tangible, and because reporters and regulators began openly debating whether temperature-only triggers actually track the loss they are meant to replace.

That debate is exactly the corporate conversation, one tier up. Through 2026, solar developers in Rajasthan bought irradiance-based covers their lenders increasingly demand, hydropower sponsors looked at low-river-flow triggers, and food-processing and warehousing clients started asking brokers for rainfall and wind-speed layers to protect cash flow rather than physical damage. Renewables and agriculture together drive the bulk of Indian parametric uptake today, but the inquiries now reaching corporate desks are about commercial property and business interruption.

The underwriting question has shifted with that move. For a heat micro-policy the design challenge is keeping payouts, in the words used around these schemes, practical, sustainable and aligned. For a corporate account placing a parametric layer worth crores, the challenge is trigger engineering: which index, measured where, against which payout curve, and how much basis risk the buyer is knowingly accepting. Most brokers can sell the concept. Few can interrogate the trigger. This manual is about closing that gap, because trigger literacy is what wins and defends the cat-exposed corporate relationship.

Basis risk is the gap between the parametric payout and the buyer's real economic loss. In an indemnity policy the two are designed to match, subject to deductibles and limits, because the surveyor measures actual damage. In a parametric policy they are deliberately decoupled. You agree a payout against an index value, then you live with whatever difference emerges between that index and your books.

Basis risk runs in two directions, and brokers must name both to the client. Negative basis risk is the painful one: the asset suffers a real loss but the index does not breach the trigger, so nothing pays. A cyclone tracks twenty kilometres from the reference station, recorded wind speed falls just short of the threshold, and the warehouse roof is still gone. Positive basis risk is the windfall: the index triggers and pays even though the asset escaped material harm. Buyers rarely complain about windfalls, but regulators and reinsurers watch them, because consistent over-payment signals a mispriced or loosely designed trigger.

Three drivers create most basis risk in Indian placements. First, spatial mismatch: the reference weather station or grid cell sits too far from the asset, so it measures a different storm. Second, index mismatch: the chosen parameter, say peak gust, correlates only loosely with the damage mechanism, which might be sustained wind plus rainfall. Third, temporal mismatch: the measurement window misses the event, for example a trigger averaged over a calendar month when the damaging rainfall fell in six hours. Every line of trigger design that follows is really an exercise in shrinking these three gaps to a level the buyer can accept and the reinsurer will price.

The first discipline is refusing to start from the data that happens to be available. Start from the loss you are trying to fund, then work back to the parameter that drives it. A coastal warehouse fears roof loss and stock soaking, so the damage mechanism is wind plus water ingress, which points to a combined wind-speed and rainfall trigger rather than gust alone. A solar farm fears revenue shortfall from cloud cover, so irradiance below a threshold over a defined window is the honest index. A hydropower sponsor fears generation loss, so river flow or reservoir inflow tracks the economics better than rainfall upstream.

Indian corporate triggers in 2026 cluster around a recognisable set of parameters:

• Excess or deficit rainfall, measured in millimetres over a station or gridded dataset, for monsoon flood and drought exposures.
• Cyclone wind speed, often as a cat-in-a-box or track-and-intensity structure for east-coast and Gujarat-coast assets.
• Earthquake magnitude and depth at defined epicentral distances, for Himalayan-belt and Kutch assets.
• Solar irradiance and wind velocity, for renewable generation revenue.
• Temperature and AQI thresholds, mainly for workforce and operations, now extending to outdoor logistics.

Whichever parameter you choose, insist on a third-party, auditable data source named in the wording: an IMD station, a recognised global gridded reanalysis dataset, or a satellite product with a documented methodology. A trigger referencing a private, unverifiable feed is a dispute waiting to happen.

The index choice also fixes the correlation ceiling for the whole cover. A perfectly priced, generously limited cyclone policy built on the wrong parameter will still disappoint, because no payout curve can rescue an index that does not move with the client's loss. Spend the analytical effort here, before pricing, because this is the decision that determines how much basis risk is structurally baked in versus tunable later.

Once the index is fixed, the payout structure converts index values into rupees. This is where a broker earns the fee, because the curve shape changes both the basis risk profile and the premium.

The simplest form is a single binary trigger: cross the threshold, receive the full agreed sum, breach nothing and receive zero. It is clean, cheap to price and fast to pay, which is why the heat schemes use it. Its weakness is the cliff edge. An asset that experiences an index value just below the threshold gets nothing, even though its loss may be substantial, which maximises negative basis risk right at the boundary.

A linear or proportional curve softens the cliff. Payout scales between an attachment point, where it starts paying, and an exhaustion point, where it reaches the full limit. A rainfall cover might begin paying at 250 mm over the window and reach full limit at 450 mm, with proportional payouts in between. This tracks graduated losses far better and is usually worth the modest premium uplift for a corporate buyer.

A tiered or stepped structure sets several thresholds, each releasing a fixed slice of the limit. It sits between the two, offering more granularity than a binary trigger while keeping settlement arithmetic simple, useful when the loss relationship is roughly stepwise, for example operational shutdowns triggered at successive cyclone categories.

Two design parameters deserve explicit broker attention. The exit or franchise level defines the minimum index value below which nothing pays, and setting it too high reintroduces the cliff you tried to remove. The payout cap must be reconciled against the buyer's modelled loss at the same return period, otherwise you sell a limit that looks generous but exhausts well below a realistic worst case. Document the chosen curve, the attachment, the exhaustion and the cap in the wording, and walk the client through the rupee payout at three or four illustrative index values so there are no surprises at claim time.

Parametric is not a replacement for the indemnity programme. On a corporate account it is a layer with a job, and the broker's task is to define that job so the two structures complement rather than overlap or collide.

The cleanest positioning uses parametric for what indemnity handles badly. Traditional fire and property covers pay measured physical damage but settle slowly and exclude or sub-limit a great deal: business interruption with long indemnity periods, denial of access, contingent business interruption from supplier sites, and pure non-damage operational loss. Parametric pays fast, in many Indian placements within days of the index confirming, and pays regardless of whether physical damage occurred. So the natural role is funding cash-flow gaps and non-damage losses while the indemnity claim is being adjusted.

One structural point underpins all of this. Parametric proceeds are typically paid against the index breach, not against proof of loss, which raises the question of whether the buyer holds an insurable interest sufficient to avoid a wagering characterisation. Structure the cover so the limit is anchored to a genuine, demonstrable exposure, and keep the documentation that shows it.

Three placement frictions need managing. First, double recovery: if a parametric layer and an indemnity policy both respond to the same business interruption, the programme may over-indemnify, so coordinate the indemnity period, the parametric window, and any other-insurance language. Second, deductible interaction: a parametric layer can sensibly sit inside or alongside a large property deductible, effectively buying back the retained layer, which is often the most compelling corporate use case. Third, reinsurance and capacity: much parametric capacity for Indian risks is reinsurance-led, frequently routed through GIFT City vehicles or offshore markets, so confirm the paper, the security and the settlement currency before you present the structure as firm. Get these three right and the parametric layer makes the whole programme more responsive rather than merely more expensive.

Parametric pricing is fundamentally a modelling exercise, not a loss-ratio negotiation, because there is no claims history of adjusted losses to lean on, only index history. The price is built from the modelled probability of the index breaching each threshold, multiplied by the payout at that level, loaded for expenses, capital and margin.

That makes the data window a live underwriting variable. A trigger calibrated on twenty years of station data will price very differently from one calibrated on forty years, and Indian climate trends mean the recent decade often carries a higher breach frequency than the long-run mean. Brokers should ask the underwriter, in writing, which historical period anchors the burn analysis and whether any climate-conditioning adjustment has been applied, because that single choice can move the rate materially.

The modelling conversation also exposes basis risk quantitatively. A credible underwriter will share the historical correlation between the proposed index and a proxy for the client's loss, often modelled damage at the asset. If that correlation is weak, the right answer is to redesign the trigger, not to widen the limit. Where data is thin, expect conservatism: higher attachment points, lower caps and a basis-risk loading in the rate.

For the buyer, the value test is straightforward. Compare the parametric premium against the cost of the alternative, which is usually carrying the retained layer on the balance sheet or buying back a property deductible at indemnity rates. Parametric wins where speed of payment, certainty of trigger and freedom from loss adjustment carry real economic value, for example where a generation or processing asset has covenanted cash flows. It loses where the buyer simply wants damage indemnified and is comfortable waiting for a survey. Frame the recommendation in those terms, with the modelled annual expected payout and the worst-case net position shown side by side, and the corporate risk manager can defend the spend internally.

The promise of parametric is a settlement with almost no argument, but that promise lives or dies in the wording. The single most important clause is the calculation agent and data source provision. It must name who reads the index, from which dataset, at what resolution, and what happens if that source fails or is revised. Reanalysis datasets and even IMD records can be restated after the fact, so the wording should fix whether the cover responds to the first-reported value or a final corrected value, and within what window.

A second clause that decides disputes is the measurement geometry: the exact coordinates of the reference point, the radius of any cat-in-a-box, and the averaging window. Vague geometry is the most common source of basis-risk surprise, because the client assumed the trigger watched their site and it actually watched a grid cell several kilometres away.

Practical claim discipline for the broker:

1. At placement, obtain and file the index methodology document, not just the schedule, so you can audit a triggered or non-triggered event later.
2. Confirm the settlement timeline in the wording, the few-day turnaround that justifies the product, and the named bank account, so funds move without a fresh negotiation.
3. Pre-agree the dispute path, ideally expert determination on the index reading rather than full arbitration, because the only real question is what the data said.
4. Brief the client honestly before binding that a near-miss non-payment is a feature of the structure, not a denial, so a marginal event does not become a relationship rupture.

Handled this way, the parametric layer does what it promises: it pays quickly, objectively and without a surveyor, and the broker who engineered the trigger keeps the account.