Sovereign catastrophe risk transfer in Asia faces barriers

World Bank has successfully run out programmes in the Caribbean and Mexico but so far not in peril-prone South-east Asia

Take away batik and ketchup and Indonesia's contribution to the English language is pretty scant – but one word that every English speaker knows is Krakatoa. When that volcanic island, tucked between Java and Sumatra, exploded in 1883 the sound was heard nearly 5,000 kilometres away and 36,000 people are estimated to have died in the resulting series of tsunamis.

The Krakatoa eruption is famous because of its size and rarity – but the region it inhabits continues to be plagued by natural disasters, or perils to use insurance speak. While last year was notable for the low number of fatalities globally from natural catastrophes – total 2014 casualties stood at 7,700, well below the 30-year annual average of 56,000 according to figures by Munich Re (see table below) – Asia still led the way both in terms of total insured damage caused and numbers killed.

And this is no exception: a combination of seismic activity, volcanoes, typhoons and floods means that Asia – particularly in its southeast region – is acutely exposed to a number of natural catastrophe types. The last decade has seen Typhoon Haiyan – known as Yolanda locally – inflict more than 6,000 deaths in 2013. In 2011 flooding in Thailand crippled the country's car industry and landed the government with a $46.5 billion bill, equal to 5% of its annual revenues, to clear up the mess. While Boxing Day 2004 saw the Asian tsunami, the worst natural disaster of modern times leaving 180,000 dead and nearly two million homeless, and whose legacy is still felt today.

Governments cannot control the weather but they can attempt to limit the impact of the damage it causes by improving infrastructure to better withstand extremes and to recycle risk via the capital markets. One entity active in this area is the Caribbean Catastrophe Risk Facility, a public-private partnership based in the Cayman Islands. Set up as a non-profit mutual insurer, CCRIF uses a variety of structures to pass on the catastrophe risk faced by countries in the region.

Its website describes CCRIF as the first multi-country risk pool in the world, and "also the first insurance instrument to successfully develop parametric policies backed by both traditional and capital markets."

CCRIF is in essence a regional catastrophe fund for local governments which is intended to limit the impact of natural catastrophes by disbursing funds to affected countries quickly. Likewise Mexico – afflicted by hurricanes, earthquakes and lacking the financial resources to deal effectively with a large-scale natural disaster – has since 2009 been transferring risk to the capital markets via the World Bank's MultiCat programme (see box below – 'How does natural catastrophe risk transfer work?')

The MultiCat initiative enables Mexico to issue catastrophe-related structures through a common documentation platform. It has issued two bonds under the programme: a $290 million three-year deal in 2009, and a $315 million three-year deal in 2012.

In June 2014 the World Bank issued its first transaction for CCRIF under its capital-at-risk programme, a variant of the MultiCat scheme.

Transferable skills

So why not expand the programme to include countries in South-east Asia which are acutely exposed to these issues? The Philippines finance secretary Cesar V Purisima has asked himself this question. He first mooted a regional natural catastrophe risk transfer facility, under the auspices of the Asian Development Bank and the World Bank in 2012. Three years, and Typhoon Haiyan later, why hasn't it happened yet?

Michael Bennett (pictured), head of derivatives and structured finance at the World Bank, based out of Washington, DC, sits back in his seat and says "in principle" there is no reason why the Philippines, Thailand and Indonesia – or indeed other countries in the region – couldn't join the latest capital-at-risk notes programme. Indeed the addition of some South-east Asian risk to a natural catastrophe market that is highly concentrated in the "peak perils" of Florida hurricane and California earthquake risk, would be a welcome diversifier with a positive pricing impact.

"Our first risk transfer transactions for clients were all in swap form – for example, for seven years we helped CCRIF to pass risk to the capital markets through cat swaps. We then created the MultiCat programme, which has been used twice by Mexico. MultiCat is a classic catastrophe bond programme where a special-purpose vehicle (SPV) is set up for each issuance, and the World Bank acts as the arranger and adviser to the sponsors. And then in 2014 we set up the capital-at-risk programme which allows us to now issue cat bonds off our own balance sheet."

The head of derivatives says that replacing the SPV with the bank's balance sheet simplifies the process because the World Bank can now go to countries and ask them to enter a transaction directly.

"We can structure the contract with the client in a bespoke way and that risk can then be passed on to the capital markets in bond form. There is no need for any collateral arrangement because we have a real balance sheet instead of an SPV. The client just sees us as its only counterparty."

Market sources indicate that the World Bank has been in discussions with a number of governments in Asia over extending the MultiCat programme to countries in the region but Bennett declined to comment on this. He does say that any further transactions could be some way off due to the practical and political risks involved in setting up this type of programme.

"There are countries looking at this and they have access to the necessary data, but at this stage we don't have a pipeline of deals. Transactions like these take a very long time, the data is very complicated and you usually need a champion within the government to push the idea.

"Also, it's not typical for most governments to seek risk transfer to the market. Some may also be concerned with political issues like the risk of paying premiums if the insurance is never used, because opposition parties could then say it was a waste of government resources."

Trigger shy

Political issues aside there are also significant practical obstacles to getting deals into the capital markets – mainly gathering enough data and then using those numbers to construct a deal that is attractive to both governments and investors. The issue here is that governments typically prefer a parametric trigger, for example an earthquake greater than eight on the Richter scale or a typhoon with wind speeds over 100mph, versus capital market participants' preference for the more orthodox indemnity approach which pays out based on the size of loss.

"So we are usually talking about parametric triggers for sovereign transactions where the capital markets price something on a probabilistic basis – what is the probability of those parameters being hit? It may be hard for investors to deal with an indemnity trigger for a sovereign deal since we are often talking about fiscal losses and in most cases the government would be in the sole position to determine those losses," Bennett says.

Calculating these types of probabilities requires large amounts of data, something that typically emerging market economies that are in most need of this type of risk transfer don't have. While there may be pockets of data – insurers covering the Japan car industry in Thailand will have good understanding of the flood risks in that country for example – it won't be sufficient for catastrophe bond purposes and is unlikely to be in a form accepted by the industry.

"The first step is often to bring a third-party modelling firm that the markets trust in order to model the probability of certain outcomes."

With seven iterations of the World Bank's Caribbean cat risk transfer programme and two versions of the Mexico MultiCat deal already complete, there is an existing pool of highly detailed data. The problem is these programmes are totally uncorrelated with the perils faced in South-east Asia and, outside of what Bennett describes as a "Mayan end of days scenario", there is no conceivable weather event that would impact both Mexico and the Philippines at the same time.

Damage control

Likewise the idiosyncrasies of each country's topography and infrastructure setup mean it is not possible to use, say, Mexico hurricane data to estimate the potential impact of a typhoon in the Philippines because the storm patterns and the damage caused in each location are different.

Hurricanes in Mexico usually have one point of contact before the storm goes inland and dissipates its strength, whereas typhoon tracks in the Philippines hit multiple islands at different times, sometimes changing direction in the process. As a result, modelling the risk exposure in both countries is different, says London-based Milan Simic, AIR Worldwide's head of international operations.

The 2011 Tohoku earthquake and tsunami was the second-costliest natural disaster in terms of insured losses since 1950, at $40 billion, behind only 2005's Hurricane Katrina at $62.2 billion (source: www.iii.org)

 

AIR is a modelling agent active in private sector nat cat risk transfer as well as being involved with all the sovereign transactions so far. Simic says the Philippines, Indonesia and Thailand are all capable of being modelled but even though they are a geographically compact group of countries, the perils they face are markedly different.

"There is very little you can do in terms of proxy data. Although the physical appearance of catastrophes is similar – there is very little practical difference between a hurricane and a typhoon – the impact of those events will vary depending on the country they hit.

"Additionally all three of those countries suffer from different perils – Philippines has a very strong typhoon activity, plus it has relatively high risk of earthquake activity, and also volcanoes. It's a country with a significant exposure to natural hazards. Whereas Indonesia's relatively southern position means that while it has less weather risks than the Philippines it has a much greater exposure to earthquakes and tsunamis," says Simic.

The lack of correlation between the risks underlying the existing nat cat risk transfer programmes and potential new recruits such as the Philippines and Thailand may give headaches to the quants assigning probabilities for natural disasters but it offers diversification benefits to investors. Until now, no cat bond has been issued which is linked to South-east Asian underlying; the overwhelming majority of exposure is found in North America.

"Diversifying risk is generally attractive to the market and these then trade at lower multiples And since those multiples have been going down, you could say now is a good opportunity to bring those risks to market. Broadening the market will happen - it will just take time," says Bennett.

 

How does natural catastrophe risk transfer work?

The World Bank's MultiCat programme uses a special-purpose vehicle (SPV) to write insurance contracts and then issue a catastrophe bond which the World Bank places with institutional investors. The SPV invests the funds in triple-A rated assets, which form the source of payouts if cover is triggered.

If no event occurs during the life of the bond – typically three years – the SPV returns the principal to the investors. "Investors in the bond take the risk that all or part of their principal may be lost if a covered event occurs during the life of the bond and receive a coupon that reflects this risk," the World Bank says in the issuing documentation.

The World Bank intends the MultiCat programme to be capable of supporting a number of structures across various perils in different regions, with each bond issued under a common legal framework.

The capital-at-risk notes programme which the development bank launched last year works in the same way but with the World Bank's balance sheet replacing the SPV.

 

Pandemic bonds

Pandemics and natural catastrophes share many characteristics – their random nature, the extreme tail risk they contain and the strain they can put on emerging economics that lack the finances and infrastructure to cope with them. Crucially pandemics and natural catastrophes are also uncorrelated with the risks driving the wider financial markets. So in principle the potential risk of an outbreak of avian flu, Ebola or a new undiscovered disease, could be transferred to the same types of investors who put cash into catastrophe bonds.

According to Michael Bennett, head of derivatives and structured finance at the World Bank, concern over the slow international response to the Ebola outbreak in parts of West Africa has led his team to look into issuing pandemic bonds using the nat cat risk transfer technology.

"With a pandemic if you can spend the money earlier and stop transmission promptly, the total costs are significantly lower. So now the World Bank is looking at whether we can do something using the cat bond and swap infrastructure we have developed where we could essentially provide insurance against pandemics. "

Interestingly Bennett says that epidemiologists he has spoken to say that – by the standards of pandemics – Ebola is relatively simple to deal with: its symptoms are unique, and it only becomes transmissible at this point meaning it is easy to isolate those who are suffering. By comparison HIV sufferers can transmit the disease to large numbers of people without realising they have been infected.

"For those in the field those types of diseases are the really scary ones."