In a flood, pay first and audit afterwards

Every few years an Indian city floods, and the same story follows. Thousands of vehicles stand in water for days. Shops lose their stock and small factories their machines. Insurers open help desks, and the regulator and the public expect claims to be handled quickly. Then, for weeks, many policyholders wait, not because anyone doubts that their car was under water, but because each claim has joined a queue for a surveyor's visit.

I have managed claims through five such events, from Cyclone Hudhud in Visakhapatnam in 2014 to the Chennai floods of December 2023, and my position has hardened with each one. For routine catastrophe losses, an insurer should pay on remote assessment and control the risk of overpayment by auditing after payment, not by inspecting every claim before it. The physical survey still matters. It belongs where the loss is large, the facts are unclear or the claim looks wrong, not as a gate through which every family must queue.

The arithmetic of a surveyor queue

The case starts with numbers nobody disputes. In the first ten days of December 2023, Nilakhya received 11,900 motor claims from Chennai, more than nine months of the city's ordinary volume. Our surveyors there could visit about 250 vehicles a day, and many of the garages they would have visited were flooded themselves. A queue of 11,900 at 250 a day is seven weeks long before a single repair begins.

Seven weeks is not an administrative inconvenience. For a family that uses its car to reach work, or a driver whose taxi is his income, it is the difference between a bad month and a lost one. Garages begin to charge for storage after a period, and vehicles left standing deteriorate while they wait, which raises the eventual claim. The delay also falls unevenly. Policyholders who can telephone a manager, or who bought through a large dealer, tend to move up the queue. Those with neither wait longest.

The traditional answer is to bring in surveyors from other regions. We did that, engaging 48 from other states. It helps at the margin. It does not change the arithmetic, because the constraint is not only the number of surveyors but the number of places they can reach in a flooded city each day.

Sampling is a control, not a concession

What changes the arithmetic is separating assessment from inspection. A licensed surveyor can judge a great deal of flood damage from video and photographs taken by the owner or the garage: the height of the waterline, the condition of the interior, whether the engine was started after water entered it. For cars submerged above the dashboard, the question is usually whether repair would cost more than three-quarters of the insured declared value, at which point the policy treats the vehicle as a constructive total loss. That too can often be judged remotely, and early.

In Chennai we assessed vehicles with an estimated loss below ₹3 lakh in this way, paid half the expected settlement on account for likely total losses, and required a geotagged photograph of standing water with every claim to support our reinsurance documentation. We then audited one paid claim in eight. Of 1,490 claims audited, 2.6 per cent had been overpaid, about ₹3.4 crore across the event. Forty-one claims were referred for investigation. The median motor settlement was nine days, and 92 per cent of claims were paid within thirty.

Our head of fraud control opposed the approach before the event, and his objection deserves a fair statement. Remote assessment, he argued, makes it easier to add old damage to a flood claim, to stage photographs or to claim for a car that was never in the water. All of that is true. The question is whether a physical survey of every claim is the best defence against it. In my experience it is not. A surveyor visiting the hundredth flooded car of the week is not conducting a forensic examination. An audit after payment, designed by people who know where fraud concentrates, can look harder at fewer claims: those from garages with unusual patterns, those just below the remote-assessment threshold, those on policies bought days before the rain.

Sampling is therefore not a concession to speed. It is a different control, and in a catastrophe a better one. It accepts a small, measured rate of overpayment in exchange for weeks of policyholders' time, and it puts an insurer's scarcest people where the risk actually lies. It also produces something a queue never does: an estimate, three months later, of exactly what the speed cost.

What has to be bought before the rain

The approach has costs that are easy to understate. The 160 staff we redeployed to Chennai came from Pune and Lucknow, and ordinary claims there slowed; in Pune the median motor settlement rose from eleven days to sixteen for December. We did not tell those customers why, and we should have. We also lost the first three days of the event because we had not contracted salvage yards and additional surveyors in advance, and had to negotiate both while the water was still rising.

That is the real lesson for boards. Catastrophe operations are decided before the catastrophe. An insurer that wants to pay quickly after a flood needs agreements with surveyor pools and salvage yards signed before the monsoon, a rehearsed plan for redeploying staff, remote-assessment rules that its reinsurers have seen in advance, and audit criteria written when nobody is under pressure. None of that is expensive compared with the claims involved. All of it is easy to defer in a quiet year, and every year without a cyclone feels like a quiet one.

There is a fairness argument too. Physical surveys before payment feel rigorous, but their burden falls on the people least able to carry it: the taxi driver, the shopkeeper whose stock was the business, the family whose only vehicle is in a garage that is itself under water. An audit after payment moves the burden from the many honest claimants to the few doubtful ones, which is where it belongs.

I would ask any insurer's board three questions before the next cyclone season. What would our settlement time be if we received nine months of claims from one city in ten days? Which decisions would we make differently from the ordinary rules, and who has the authority to make them? And how would we know, three months later, what the speed had cost us? An insurer that can answer all three has a catastrophe plan. One that cannot has a press release about its help desks waiting to be issued.