Some Things Are to Hot to Touch

WilliamsCollege

April 2008

Some things are to hot to touch

The human mind can only take so much

You can’t win with a losing hand

Bob Dylan – Things Have Changed

Climate change, Development and the Insurance Sector

Rodney Lester, Senior Advisor, the world bank

There is growing evidence that the world’s climate is changing in a way that will increase the frequency of severe weather events. 2005 was the hottest year for a millennium and the 1990s the hottest decade. The Iberian Peninsular and Brazil have recently seen their first proto hurricanes in living memory. There is also growing evidence that this process is being driven by the manmade carbon dioxide emissions that are an intrinsic output of modern carbon based economies. These developments are likely to fundamentally change the way in which the financial sector designs and prices its products. In addition the insurance sector has been formally identified as part of a suite of adaptation tools mentioned in both the UN Framework Agreement and the Kyoto Protocol.

The insurance industry has already been very active in raising awareness of the consequences of climate change, particularly in Europe where the precautionary principle is perhaps better accepted than in North America. However underwriting practices in North America indirectly carry the message. It is well known for example that the leading private insurers are withdrawing from certain parts of the United States that are exposed to hurricane and storm surge risk, price controls and populist judiciaries. State governments are in consequence increasingly becoming insurers of last resort. In developing countries catastrophe sourced contingent fiscal and developmental risks are also present. All indicators point to climate change as having a disproportionately severeimpact on the populations of such countries.

However it needs to be understood by those making public policy that catastrophe insurance typically costs significantly more thanthe expected loss and insurance capacity is not infinite. On the demand side cognitive research on low probability events and experience on the ground indicate that individuals underestimate the probability of infrequent severe events and their risk aversion is consequently not sufficient to justify paying much more than the expected loss. Thus risk transfer markets can have trouble forming. If insurance is to fulfill an adaptation role it will increasinglyhave to be in the context of public private partnerships, possibly involving well targeted subsidies of transparent market prices and intensive public education efforts.

Introduction

To fully understand the impact of climate change at the country, corporate and household risk management levels one needs a little historical perspective.

Carbon dioxide concentrations most recently began to rise around the same time the industrial revolution began to raise standards of living from the relatively static level of the previous few millenia. Since 1800 atmospheric carbon dioxide has risen from 280 ppm to 383 ppm and concentrations continue to rise. Standards of living and world population have risen in tandem. The industrial revolution is carbon based. Sweet oil in particular is an unbelievably efficient source of energy – one barrel is equivalent to 6.1 x 109 joules. Put another way one gallon of gasoline is worth about 500 hours of human physical toil. Some estimates indicate that the carrying power of the world would be one or two billion people rather than the current 6 billion (and rising) if we did not have access to a source of energy that is effectively free in terms of market price per joule.

However we now see evidence that, with an ocean sink induced time lag, increasing concentrations of carbon dioxide, both directly and through positive feedback loops, are raising surface air and water temperatures.The significance of this for insurers is that a relatively small increase in average temperatures means a much greater incidence of extreme temperatures because the temperature distribution curve is moved to the right. Mills (see key readings) quotes a UKstudy that indicates that a 1.6 degree increase in average temperatures increases the likelihood of extreme temperatures by 25 times.

Chart 1: Trends in high temperature months in the UK

Source: Dlugolecki

The picture remains clouded becauseextracting the recent impact of greenhouse gases from normal cycles is complicated by the fact that the North Atlantic is entering a period of cyclical warming, the last such period corresponding with the heavy hurricane seasons of the 1950s. However ananthropogenic influenceis increasingly evident andif this interpretation of the data is correct it will be the first time in the last 650,000 years that CO2 has become an initiating forcerof a warming trend.

On the positive side it is likely that such trends, if contained at relatively modest levels, will increase growing seasons and extend the range of agriculture in somecolder parts of the world, although others regions nearer the equator are likely to see reduced and more concentrated rainfall. South Eastern Europe for example would have to give up its traditional agricultural practices and way of life.

Thus until recently it could be argued that we live in a world that owes its wealth to carbon based fuels and that with a bit of adaptation all would be well. This attitude is now changing, driven by 2 realities:

1. There is widely disseminated evidence that theGreenland and Arctic ice sheets could slip into theircontiguous oceans, and much sooner than originally modeled. Even one of the 2 lesser events, Greenland or the Western Arctic ice sheet, would see a good part of the US East Coast submerged. In addition glacial retreat is now clearly threatening fresh water supplies in a number of industrial, transition and developing countries.
2. The financial system is beginning to understand the potential risks inherent in climate change and this in turn is likely to materially affect peoples’ lives. In fact it may already be doing so through insurance and weather risk derivatives markets. To date the financial sector has largely relied on past experience to price its products and in setting risk management parameters such as economic risk capital. However there is an increasing trend to use prospective measures and this may have significant implications for the availability and price of insurance and other risk management instruments and for regulatory policy.

The first of these two developments is leading some people to argue that strong and urgent mitigation measures are required, and in particular steps to effect a relatively rapid reduction in the concentrations of greenhouse gases such as carbon dioxide, methane and nitrous oxide in the atmosphere. The political problem of course is that this would disrupt the superficially free carbon based lunch that the world has been having for the last 2 centuries, and could in fact lead to global conflict and chaos. Current responses have been largely restricted to free market mechanisms such as the trade in carbon credits (which are mandated in Europe). However there is little evidence that this approach is having any significant impact on the global carbon footprint. Other responses have involved the use of alternative energy, of which solar energy looks to be the most promising long term solution, although the technology is still not particularly cost effective. Finally there is a clear need to use energy more efficiently and many initiatives are under currently way at the civil society, commercial and household levels. National governments tend to lag their populations in this regard: fossil fuel subsidies still greatly outweigh those for alternative energy in the major economic blocs and carbon is trading at well below its long term shadow price as estimated by the EC.

The financial sector is likely to produceuseful if uncomfortable forms ofnegative feedback and hence strong signals for investment in mitigation. However insurance in particular is currently seen by the policy makers as being more of an adaptation mechanism (see for example Acosta’s presentations on the UNFCCC web site – The reality is that key sectors of the insurance industry have been concerned about climate change for many decades and in a number of areas are at the cutting edge of mitigation knowledge generation and dissemination, including the shaping of more resilient infrastructure investments.

Climate change and risk transfer

Risk is an intangible – it represents the possibility of loss. The measurement of risk depends on one’s perspective. At its most fundamental it can be represented by expected loss. This is the way many macroeconomists see it as individual variationsin loss experience tend to wash out at the macro level. Thus large well diversified economies can theoretically set aside expected losses through funding and insurance mechanisms each year and these mechanisms will broadly meet ongoing losses. This is one reason large wealthy countries retain the risk on government owned properties and infrastructure. Similar considerations apply to very large multinational enterprises which effectively have more stochastically efficient portfolios than most insurers.

However when it comes to smaller countries and enterprises and households, which have more concentrated and vulnerable productive capacity and essential social assets, the picture is very different. In fact even the larger countries are now finding that risks are becoming significant relative to available resources. There are at least two risks in the US that have been modeled as producing losses of in excess of $100 billion with a probability high enough to warrant public policy attention. In addition a number of US states have sponsored catastrophe insurance pools and the national flood insurance facility is administered by FEMA, a Federal agency.

In the case of seismic and other non hydro meteorological hazards increasing risk levelslargely reflect growing concentrations of population and property/ infrastructure in risk prone areas. The doubling of the population of Instanbul in a single decade in a highly seismic region is one well known example outside the USA.

Chart 2: Hydro meteorological versus all natural catstrophes

However in the case of hydro meteorological threats we see both increasing exposures of people and property, and the likelihood that the frequency of severe weather events is increasing, even if overall hazard frequencies are remaining relatively static.

A fundamental rule of nature is that the less frequent is an earthquake, tropical cyclone, lahar, wild fire etc. the more intense it is. Thus the distributions of losses arising from catastrophes (defined as events affecting a lot of people and assets at the same time)havevery fat tails and any pricing based on the bell curve is likely to be woefully inadequate. One may draw a valid analogy with structured financial instruments, which clearly show cooperative behavior when market volatility gets beyond a certain point. In fact the processes are remarkably similar and it is surprising that the financial markets have not been using more actuarial input in pricing risk and setting capital requirements for those institutions holding direct and residual exposures to leveraged instruments .

When applied to catastrophe risk the 2nd and 3rd moments of the loss distributions do in fact manifest themselves through capital requirements. However the role of capital in the markets for insurance risk does not appear to be well understood by many influentialeconomists or public policy types, although there are honourable exceptions including a small cadre of academics who are now actively debating the issues and engaging with the insurance sector and development organizations.

To be precise the premium charged by an insurer or reinsurer to accept the transfer of a risk has three components:

1. The expected loss
2. Expenses
3. The cost of capital

The expected loss under an indemnity contract for a severe hazards with say a 50 year return period (i.e. a probability of 2% of occurrence in any given year) , such as Hurricane in the Caribbean, might be of the order of 1% of sum insured. It is less than 2% because a total loss is not the norm under an indemnity arrangement. Expenses for catastrophe reinsurance are usually small, around 4% of the premium.

Capital allocated to catastrophe risk is increasingly being driven by probability of ruin (VAR in bank-speak)considerations although portfolio diversity is likely to be taken into account. The large European reinsurers have already published their capital requirements and shown theirdiversification offsets. This is probably the key issue for insurers as compounded non linear processes are involved. As already notedrelatively modest upwards shifts in the temperature distributioncan have amuch greater impact on the incidence of extreme weather, and hence loss distributions,than a simple prorating of temperature averages would indicate. For example property damage increases as the cube of wind speed, which is broadly related to water surface temperature in the case of tropical cyclones. Other manifestations of severe weather show even more extreme non linearity when it comes to damage curves.

Reinsurers’ costs of capital arenow also being determined using prospective measures such as the term sensitive market derived capital price (MCPM) rather than the backward looking and highly unstable CAPM. Thus, to the extent that markets factor climate change into equity options on insurers and reinsurers both the risk capital and cost of that capital will reflect future weather patterns. This is likely to increase the long term cost of catastrophe reinsurance and hence direct insurance prices. The reason for this is that we are dealing with annual contracts (i.e. short term MCPM cost of capital) and the events of concern are infrequent, so data is limited and thanks to climate change there is some evidence of non stationarity and increased scope for parameter or even model error in loss distributions.

In the US this will raise thorny regulatory issues as a free insurance market does not always operate in this country. In a number of states retail insurance pricing is subject to supervisory overview and in many of these states the supervisor is elected. It will be interesting to see if stochastically based prospective pricing will be permitted – the National Association of Insurance Commissioners has it on its agenda but is yet to opine. The growing pressures on insurers to make full risk disclosures under such measures as Sarbanes Oxley and Solvency II could lead to some interesting regulatory conflicts.

In practice catastrophe risks are broken up into layers. The lower, or more frequently hit layerswhere implicit premiums are not much greater than expected loss (commonly called working layers), are usually retained by the direct insurer, depending on its size, capital resources and portfolio diversity. The higher layers are laid off to reinsurers or the capital markets (the latter most often through catastrophe bonds). As the higher layers represent the least frequent but most severe events, about which there is little data(and where climate change is likely to have the most effect), there is a much greater amount of capital required per unit of sum insured and the premium is a greater multiple of expected loss. Ken Froot of Harvard brought this to light in the much quoted introduction to his 1999 book (see key readings), with the multiple ranging between 1 and 2 at the lower layers and up to 8 for the top layer.

What was perhaps of most concern at the time the book was written was that direct insurers were not buying sufficient coverage at the upper layers, presumably because of a combination of price and career horizons that are short compared to the return periods of the catastrophes concerned. However there is some evidence that this is changing, possibly driven by the increased frequency of natural disasters in the last decade and a half. From an insurer’s point of view it is of note that this much increased frequency and cost of cat. events is entirely due to hydro meteorological events and may reflect the lagged impact of anthropomorphic CO2finally emerging.

Chart3: Hydro meteorological insurance losses versus total non life insurance premium - 1980 provides the base ratio normalized to 100

Source: Mills, Science 309, 2005

Aside from balance sheet realities there are other imperatives for direct insurers to offload catastrophe exposures. In particular it is very uncommon for insurers to be allowed to build reserves for losses that may occur after the current operating cycle. The insurance elements of the new international financial reporting standard (IFRS) do not allow for such reserving, and any catastrophe buffers will need to be held as equity capital. In addition most tax regimes do not allow catastrophe reserves to be set aside from pre tax income. And of course no executive or board with any sense of self preservation is likely to allow large amounts of apparently unutilized capital to build up in the normal course of events.

This incentive to hold minimal capital will to some extent be offset if current thinking about insurer regulatory capital becomes the norm. This thinking is best captured by the proposed Solvency II regime in Europe, which follows the Basel II formulation for banks. Under Solvency II insurers will be required to establish two capital levels, a solvency capital requirement (SCR) which represents the economic capital required given the risk characteristics of the insurer and a minimum capital requirement (MCR) below which the insurer will lose its right to trade. If an insurer falls below its solvency capital level the financial supervisor will be required to intervene in its affairs. The solvency capital will reflect a range of risks including market, capital, reserving and underwriting risk. All four of these will be affected by climate change, and the capital allocation for reserving and underwriting risk will include an explicit allowance for retained catastrophe risk. In addition there is a trend for insurance supervisors to look at quality of reinsurance when determining insurer solvency.