Catastrophe Modelling Maturity in Indian Insurance

Most Indian insurers writing property, engineering, or motor business now use some form of catastrophe model. Some run global vendor models (RMS, Verisk-AIR, KCC), some run partial in-house implementations, some rely on reinsurer-supplied outputs. The variation in how the models are used is wider than the variation in whether they are used.

A maturity lens captures this variation usefully. Across global markets, catastrophe-modelling maturity tends to develop along five dimensions: exposure data quality, model coverage and validation, integration into pricing and capital, governance and challenge, and use in reinsurance and accumulation management. Indian insurers cluster across the lower-mid range on most dimensions, with significant variance between large mature insurers and smaller players still building basic capability.

Catastrophe modelling output is bounded by the quality of exposure data: where the insured assets are, what they are worth, how they are constructed, and what they contain.

Indian exposure data is heterogeneous. Large commercial fire policies on listed corporates typically carry good geocoding, construction-class identification, and BI-value detail. Mid-market commercial fire policies often carry city-level rather than pincode-level location, broad construction classes, and BI sums insured derived from book values rather than reinstatement. Retail fire and contents policies frequently carry only the postal address with no validated geocoding.

A maturity assessment on exposure data should look at:

• the share of insured value with geocoded locations at building or street level
• the share with validated construction class beyond simple kucha/pucca distinctions
• the share with occupancy and hazard tags beyond generic descriptions
• the periodicity of exposure refresh (annual at renewal versus continuous)
• the reconciliation of policy admin exposure against accounting and underwriting records

Indian insurers running catastrophe models on exposure data that is 60 to 80% geocoded at city level are producing modelled losses with implicit uncertainty that the model output does not reflect. Improving exposure data quality is often a higher-impact intervention than refining the model itself.

Indian commercial catastrophe modelling primarily covers four perils: cyclone, flood, earthquake, and terrorism. Coverage varies in depth.

Cyclone: well-modelled by global vendors with India-specific module development. Indian east-coast and west-coast wind models incorporate IMD historical track data and increasingly include storm-surge sub-models. Reasonable maturity at the major insurers.

Flood: less mature. India lacks the high-resolution river-network and digital-elevation-model data that vendors use in developed markets. River flood, urban flood (Bengaluru, Mumbai, Chennai), and coastal-surge flood remain separately modelled with material differences in vendor approach.

Earthquake: well-modelled for major fault systems (Himalayan, Western Ghats, Bhuj region), but population-level vulnerability functions are coarser than US or Japanese equivalents.

Terrorism: stylised models exist but Indian-specific calibration is thin. Most insurers treat terrorism through a separate pool mechanism rather than fully modelled.

Validation, where the insurer compares modelled losses to actuals, is the second-line discipline that distinguishes mature from less mature programmes. Indian insurers that have not yet built validation programmes are running models whose outputs they cannot independently defend in pricing, capital, or reinsurance conversations.

Catastrophe model output is most valuable when it actually drives decisions. Mature programmes integrate the output into three areas.

Pricing: cat-adjusted technical pricing for fire and engineering business, with location-specific loadings that reflect modelled loss costs. Some Indian insurers run uniform tariffs across cat-exposed and non-exposed locations, leaving substantial cross-subsidy in the book. Others differentiate clearly but only at the catastrophe-zone level (Zone I/II/III/IV), missing within-zone variation that high-resolution models can capture.

Capital adequacy: the IRDAI's risk-based capital framework, in development, will require explicit catastrophe-risk capital charges. Mature insurers are already running internal models that produce probable maximum loss (PML), value-at-risk, and tail-value-at-risk metrics aligned to capital methodologies.

Reinsurance placement: cat-model output drives the choice of attachment points, layer sizing, and reinsurer selection. Indian insurers with sophisticated cat-modelling capability negotiate treaty terms from a stronger evidentiary base than those who rely on reinsurer-supplied modelling. The information asymmetry has historically favoured reinsurers in the Indian market; closing the gap is a strategic priority for primary insurers building modelling capability.

Cat models are technical artefacts that require governance. Mature programmes treat the cat model on the same footing as other major actuarial models, with structured governance covering:

• board-approved model risk policy explicitly covering catastrophe modelling
• second-line validation by an independent function reviewing model assumptions, data, and output
• periodic vendor review: are we using the right vendor model, are we using it correctly, when did we last test against alternatives
• annual model review documenting changes in calibration, data, or scope
• disclosure to reinsurers, auditors, and the regulator of model basis and limitations

Catastrophe models are most operationally consequential in accumulation management. Indian insurers writing fire, engineering, and motor business carry concentrated exposure to specific events: a Mumbai monsoon, a Gujarat earthquake, a Chennai flood, an Andhra cyclone. Effective accumulation management requires:

• portfolio-level PML by peril, peril-region, and return period
• scenario stress testing against named historical events (1999 Odisha cyclone, 2001 Bhuj earthquake, 2005 Mumbai flood, 2018 Kerala flood) and against severe-but-plausible counterfactuals
• real-time exposure tracking so a new policy bound today is immediately visible in the accumulation view
• reinsurance recovery modelling ensuring the cat treaty covers the modelled events with appropriate certainty
• portfolio shaping decisions, where new business is sought or declined based on accumulation impact

Mature insurers run quarterly accumulation reviews at the board risk committee, with the chief underwriting officer and chief actuary jointly accountable for the position. Indian insurers without this discipline find themselves over-concentrated by accident: each individual policy looked reasonable at underwriting, but the cumulative book is exposed to a single peril event well beyond risk appetite.

Across the five dimensions, the Indian market shows a wide spread. Large general insurers (New India, ICICI Lombard, Bajaj Allianz, HDFC ERGO, Tata AIG) have built materially mature programmes, with the largest running internal-model capability that complements vendor outputs. Mid-tier insurers are at varying stages of integration. Smaller and newer entrants often rely on reinsurer-supplied outputs without independent assessment.

Three developments will pull the market toward higher maturity over the next 24 months.

First, the IRDAI Risk-Based Capital framework will require explicit catastrophe-risk capital methodology, forcing insurers without internal capability to either build it or rely on supervisor-validated standard formulas with less favourable charges.

Second, climate-change adjustments to historical loss data are becoming standard in global cat models. Indian insurers that update their modelling to reflect these will price more accurately; those that continue using uncalibrated historical data will under-price in the worst-affected segments.

Third, reinsurer expectations have tightened. Treaty renewal conversations now routinely include reinsurer review of the cedant's cat-modelling sophistication, with terms reflecting the reinsurer's confidence in the insurer's exposure view. Less mature insurers face worse terms; the gap is no longer subtle.

Indian insurance is at the stage where building catastrophe-modelling maturity has shifted from optional to strategic. The insurers that invest now will price better, retain risk more profitably, and access better reinsurance support than those that wait for regulatory or commercial pressure to force the issue.