The monsoon returns to India every year. This year, early in July 2006, Mumbai's financial centre was disrupted for three days, and schools were closed as traffic came to a standstill and low lying areas were flooded. August 2005 saw heavy rains throughout India, claiming over 1,000 lives and costing over $3 billion in economic losses, $700 million of which were paid by the insurance industry. In Mumbai, up to 944 mm of rain fell in a single day.
The flooding in early July 2006 had a fraction of the impact of the 2005 event, but many Mumbaikars feared the worst and left their jobs, bringing public services to a halt. Are the insurers and reinsurers equally fearful of the repeat of the 2005 claims?
While some argue over the exceptional nature of the offshore vortex phenomenon (see box) that caused unprecedented rainfall in the city, others point to the human contribution in exacerbating the severity of the flooding. Mumbai has overgrown its early 20th century drainage system: a low capacity network with few floodgates and frequent blockages. The expansion of the city on reclaimed land by building over mangroves along the Mithi River and swamps in the Mahim Creek has reduced the capacity of the Methi River and the extent of natural flood plains.
Cat losses and insurance
Stretching from the Himalayan Arc to the Gulf of Mannar, India is not unused to devastating natural catastrophes; the country was second only to China in terms of pure numbers of catastrophes in 2005. After monsoon floods affected states throughout India, the disputed region of Jammu and Kashmir suffered damage from the magnitude 7.6 earthquake centred across the border in Pakistan. In 1999, a super cyclone swept through the state of Orissa inflicting $200 million worth of damage on the industrial sector alone.
However, the state most affected in the recent past is the wealthy and developed state of Gujarat. In 1998, a cyclone caused $500 million in economic losses and brought Kandla Port to a standstill. The 2001 Bhuj earthquake resulted in losses of $2.6 billion. Over 19,200 people were killed and nearly 10,000 industrial units went out of production.
So far, the insurance industry has borne only a small proportion of these losses, but with the current trend of increasing insurance penetration, events such as these will shortly be felt internationally as the reinsurance market expands into the country.
Increasing density
India's history of natural disasters does not appear to be stopping the country's phenomenal economic growth, which is accompanied by an impressive increase in the penetration of insurance in all lines of business. Between 2001 and 2005, total premiums grew on average by 13% per annum with non-life premiums approaching $4 billion. The growth of penetration is greatest in the state of Maharashtra, which includes the country's economic capital Mumbai. It accounts for 27% of the total premiums with only 9% of the population. Insurance density, however, is still one of the lowest in the world at 0.5% of GDP for non-life insurance.
The growing exposures are providing the reinsurance market with an attractive non-correlated risk. "Insurance companies in India are getting larger. This is increasing the appetite of the reinsurance companies to write catastrophe programmes in the region" says Ian Kerton of HSBC brokers.
The growth of the insurance industry, the history of losses and the forthcoming loosening of the tariffs which will take place in January 2007 suggest an important role for catastrophe modelling. The results of catastrophe models can provide technical rates to inform reinsurance renewal negotiations. This is particularly relevant given the increase in rates following the 2005 season. However, it will be in the primary market that detailed loss models will most benefit the more experienced underwriters. At primary level, cover against natural hazards is provided by two extensions of the fire policy: one for storm, tempest, flood and inundation (STFI) and the other for earthquake.
According to Ian Kerton, "Detariffing will create a very competitive market. Smaller, emerging companies and joint venture companies will need greater sophistication to refine their pricing strategy to consolidate and grow. The latest generation of catastrophe models will be key to this process. "
Models
The Himalayan Arc produces some of the largest continental earthquakes globally, and the seismic hazard from this boundary extends to include the capital New Delhi. Large intra-plate earthquakes also occur further south within the Indian peninsula. The magnitude 6.2 Latur earthquake of 29 September 1993 occurred in Maharashtra. At least 52 villages were completely destroyed by the earthquake and around 8,000 people were killed. On 11 December 1967, a magnitude 6.5 earthquake occurred near the Koyna Reservoir in Maharashtra province. This earthquake killed over 125 people and injured at least a further 1,400. The significance of these events should not be under-estimated. Both occurred in the developed state Maharashtra, and highlight the risk that intra-plate events pose to areas of urban development in the central regions of the country.
Jointly with our sister company RMSI, Risk Management Solutions (RMS) has been building and consolidating knowledge of India's natural disasters and the vulnerability of the country's economic exposure. In the autumn of 2006, we will release an earthquake aggregate loss model, developed with the support of brokers Aon. Plans are also underway to release the detailed earthquake model, which underlies the aggregate model, and in the short term this will be available to the market. The detailed earthquake model uses a stochastic set of over 40,000 events, generated from 96 seismic sources to model the ground motion as peak ground acceleration (PGA). The vulnerability curves developed for the Indian building stock are calibrated on the most recent events and use intensity to calculate the mean damage ratio.
The aggregate loss model has been developed at Cresta level country wide, but in the states of Maharashtra and Gujarat the resolution is increased to district level. There are 19 Cresta zones in India and 60 districts in Maharashtra and Gujarat.
Understanding earthquake risk is only a part of the catastrophe landscape in India; focus on climate hazards is increasing. A project RMSI recently completed for the World Bank has allowed comparative analysis between the perils on a probabilistic basis. The risk assessment considered housing and public infrastructures in the states of Andhra Pradesh, Gujarat, Orissa and Maharashtra. As well as earthquake risk, losses from flood and cyclone formed part of the study.
The key findings of the study indicate that flooding remains the largest cause of loss throughout all states with the exception of Orissa, where the loss cost from cyclone is greatest.
Detailed flood modelling poses significant challenges in any country, but even more so in India. Critical to the understanding of urban floods, such as those that affect Mumbai among other cities, is the availability of digital terrain model data and a up-to-date information on the network and condition of the drainage system. Government restrictions often impede modellers' access to this information, thus complicating the modelling process.
The full use of the sophisticated model features presents a challenge in India, given the paucity of data on the individual risks. The release of models on the market may be a catalyst to improve in this area, as George Walker of Aon states, "Data capture in India is still very limited. Availability of catastrophe models in the market will be a strong incentive across all insurance companies to use the full potential of the newly implemented IT systems and make the exposure data available to their risk managers and the intermediaries."
With on average 8,600 casualties and nearly 100 million people affected each year by natural disasters between 2000 and 2006, there is also a strong role for models and their users to use the capabilities of these tools to quantify and reward reduction in vulnerability, in particular, the ability to capture the long term economic benefits of improved construction design and quality, both of housing and infrastructure.
The scale of natural disasters, the size of the country and scarce information about the exposure make the complexity of catastrophe modelling in India a great challenge.
1. CRED www.em-dat.net/index.htm
2. Disaster Risk Management and the Role of the Corporate Sector www.ndmindia.nic.in
- Domenico del Re works in model management, Asia Pacific Region for RMS, London. Email: Domenico.delRe@rms.com Website: www.rms.com Offshore vortex phenomenon
High winds blow in from the Arabian Sea and turn northwards. They create troughs and vortexes. Pressure drops and strong winds shoot upwards, resulting in heavy rain.