(Reprinted from HKCER Letters, Vol. 82 Mar-Apr 2005)
Information and the pricing of risk
Risk is a wide-ranging subject, which conjures up different images to different people: the risk of a hardware failure or a software virus on your computer; the risk of being burgled; the risk of being flooded; the risk of collapsing from a heart attack; or, more specifically to those working in the financial world, the risk of a price moving more than two standard deviations from its historic mean, the risk of a dealer trading beyond his authorised limits, the risk of underwriters being hit by a colossal insurance claim, the risk of a massive fraud, or simply the risk of a bank customer failing to repay a loan.
This paper addresses certain types of risk from the perspective of financial business, examining how the way in which we cope with risk is being affected by the increased availability of information, as well as the increased sophistication of some of that information.
Consider first the provision of bank credit. It has always been the complaint of small and medium sized enterprises that not enough credit is available, or that it is too expensive. Banks, for their part, defend themselves by referring to the uncertain credit-standing of SMEs, and the burden of the banks' own internal costs.
There is a corner of economics where it is argued that banks do not lend at the highest rates that they might be able to obtain, for fear of adverse selection. The reasoning runs as follows. The most dependable bank customers are probably those which earn a modest but steady profit in their business, and which cannot, because of that modestness, afford too high an interest rate. On the other hand, any customers who would be willing to contract to pay significantly higher rates are likely to be those with a vision of high profitability but probably also a much greater risk of failure. The banks do not know enough about those risks, so generally prefer not to take them on at all. Thus, rather than fill its loan book to capacity at the highest rates obtainable, a bank settles for somewhat lower rates, selecting customers with the patently lower risks, and uses rationing to deal with any excess loan demand that may arise at those rates - whether in excess of particular customer limits or in excess of the bank's balance sheet capacity.
One might wonder why banks in such situations do not spend a little more time looking at customer credentials and thus sorting out the good from the bad among the higher risk group, thereby pinpointing some more lucrative loans. One simple reason, which is often overlooked by the customers, is that credit assessment takes time, and management time is expensive. In the case of an SME for which only a relatively modest (from the banker's point of view) credit would be appropriate, the management cost element could quickly turn prospective profit for the bank into loss. The alternative of raising the interest charge to fully cover such costs would result in seemingly outrageous and probably unaffordable interest rates, which might attract only the most reckless clients. Thus the bank will tend to limit the time spent on each application, and simply decline to be seduced into exploring the more risky areas.
Nevertheless, there are forces at work towards more efficient and cost-effective credit assessment. The information revolution produces increasingly sophisticated tools, based on credit scoring and so on, which aim to enhance assessment without adding to that crucial cost element. These methods may often be impersonal. This gives rise to complaints from bank customers that they never seem to be dealing with a real person; the cosy chat with the bank manager becomes, for many, a mere fiction from the black-and-white movie era. But the fact that procedures are impersonal does not mean they are inefficient. It is even arguable that they are superior, because they remove any risk arising from errors of human judgement.
The latest addition to this credit assessment armoury in Hong Kong has been the commercial credit reference agency, which was launched a few months ago. By enabling banks to share credit information on customers and potential customers, the banks should be able to reach more assured decisions. This should help them to discriminate better among SMEs, and to allocate and price their credits more efficiently. Many SMEs may benefit. But the most risky ventures may be screened out even more decisively than previously. This would appear to make sense for banks, which, by their very nature as takers of deposits from the public, as a result of which they are also subject to quite tight regulation, are likely to remain at the cautious end of the funding spectrum. There are other channels more suitable for the provision of risk capital, such as the equity market or proprietors' own pockets.
Let us now address the same type of question in the context of equity capital. If, as investors, we are nowadays all better informed about risks, are we prepared to accept more risk? The conventional view is that the majority of people are rather averse to risk - although contrary evidence may be found in the popularity of gambling on horses or in casinos. In the investment field, we have come to expect very full information, and if we then judge the risk to be high, even the few of us who may be attracted to invest may need the enticement of a disproportionately handsome reward. If the sponsor cannot convince us, the venture will be a non-starter.
Historically things may have been somewhat different. The south sea bubble in the early 1700s, involving numerous scams, would never have developed so fully as it did if investors had insisted then on the sort of transparency we expect today. There were also some more respectable ventures which had a similar fate. What about some of those railways constructed across the world in the nineteenth and early twentieth centuries which ended in financial failure? Would they ever have been built if the investors had possessed an even half-complete picture of the risks? Perhaps not, but arguably those investors did us a favour by putting their money to work for a positive social return, even if at a personal loss.
So, as we all become better informed, and as many, perhaps most of us consequently shy away from great entrepreneurial ventures, of the type which in past centuries may have preyed on ignorance, sucked money from the uninitiate, and bankrupted many of them, is it safe to conclude that capital is necessarily going to be allocated more efficiently? Yes and no.
The flaw in that argument lies in the assumption that the ex ante information on which decisions are based is always reliable. In matters of investment appraisal that can never be entirely correct. If all ex ante high risks are screened out, then that one-in-a-hundred venture which may actually work has much less of a chance of getting off the ground. We shall perhaps be safer on average and more profitable on average, but just occasionally society may miss out on something exciting.
On the other hand, at the same time as those who seek to raise capital are being screened more discriminatingly, so too the providers of capital are perhaps becoming more stratified. As the world becomes more prosperous and the number of millionaires more numerous, we see an increasing number of boutique financial institutions, venture capital funds and the like, catering for investors who have, in effect, realised that the "average" market is too risk averse for their liking, and who have set out to deliberately court the higher risk business. The supply of risk-seeking capital of this sort is in part dependent upon the tax system being friendly enough towards profits, if or when they eventually materialise.
So perhaps, at the end of the day, both low and high risks are being better catered for, in which case we are all better off, and the textbook teaching that information promotes allocative efficiency is therefore proved correct.
The risks referred to thus far have been those specific to individual businesses or customers, where customer data and profiling can therefore assist in decisions on the allocation of funds. There is another, more general class of risk to which businesses may also be exposed, which may be termed macro-risks. These include such things as terrorism, disease or tsunamis. Banks or even stock market investors may think it unusually bad luck if they lose money to such events. For insurance companies, on the other hand, these may be regarded as part of their core business. Let us consider this in a bit more detail.
The recent tsunami disaster is a useful starting point. Property and lives were swept away. Few of those who suffered loss will have been insured. One can hardly blame poor people in Aceh or Sri Lanka for living alongside their livelihood, on the waterfront. The same could be said, for example, of the thousands in Bangladesh who are so frequently victims of flooding rivers. There are many other examples around the world. Could any of the vulnerable get insurance cover if they tried?
If the insurers reckon that such natural disasters are sufficiently rare, they may be able to offer cover at rates that are affordable, at least to some. Thus, insurers might price a fatal tsunami in the "once in a century" category (although the second quake in Indonesia, in late March, may have cast doubt even on that), but not so Bangladesh's frequent floods. Anyway, for most inhabitants of those regions the question may be largely academic, since there are no possessions to insure. For businesses, however, things are different. They may, if they can get an insurance quote, decide to pay the premium, or else consciously carry the risk themselves. One might expect them to retreat a bit from the shoreline or to slightly higher ground. But, judging from what has been happening in Thailand, for example, the businesses tend to want to stay right where they were; the lure of the tourist dollar is strong, and if you're not on the edge of the beach you're a loser. That may either be very foolish, or a highly rational actuarial decision.
One advantage of commercial insurance in such circumstances is that it relieves government of some of the burden of aid and reconstruction. However, government or aid agencies are always likely to be there as a last resort, and that expectation may discourage people from obtaining insurance cover, and make people less cautious than they might otherwise be - an example of moral hazard.
We can find examples in the developed world with many similarities. But the main difference is that people there may be deemed prosperous enough to have the choice not to live in places prone to natural disaster. Nevertheless, by way of example, despite the severe earthquake of 1906 and the lesser but still serious one of 1989, people still like to live in San Francisco. According to the official estimate from the US Geological Survey, there is a 70% probability of at least one quake of Richter magnitude 6.7 or higher, capable of causing widespread damage in the Bay region, before 2030. San Franciscans can probably, at a price, obtain earthquake insurance cover. It is their choice, both whether to insure and where to live, much more so than there was ever any real choice for the inhabitants of Aceh.
Let us take another example. In wealthy societies many people voluntarily live on river banks - indeed, enthusiastically so. It's very nice when the sun is shining and the ducks are quacking, but once in a while they get flooded. People may adopt one of three strategies. The first is to buy insurance cover, which should be available for all but the most ridiculous locations, but which may be costly. If one considers the premium too exorbitant, the second option is, in effect, self-insurance. This might be a cheaper option in terms of statistically expected values, because none of the costs of intermediation by an insurance company arise, but it will only be a feasible strategy for maintaining one's standard of living if one can access the necessary capital or liquidity to settle one's own claim when disaster strikes. The third strategy is simply to throw caution to the wind and presume upon the state to provide at least the basic necessities of life if disaster strikes.
Insurance nowadays is not, if indeed it ever was, a simple textbook business of pooling, indiscriminately, risks across a vast population, in the sense of the following example. Suppose the insurance company knows that the historical incidence of some event was 1 in 100, and has no reason to believe the future probability to be any different; and suppose all people are deemed equally likely to suffer the event; then the insurer can pool risks at a uniform premium of 1% plus a management margin.
In reality, however, life is not so simple. Any insuree who suspects that he faces a less than average risk may decide either not to insure, or to seek a discount; at the same time insurance companies may compete in trying to cherry pick the lesser risks. Meanwhile, those who suspect they are an above average risk will pile in to buy cover at the 1% premium - the phenomenon of adverse selection. The insurance companies try to counter this by making more sophisticated assessments of the risk relevant to each client, and setting premiums accordingly. Individual risk profiling takes over from mere risk pooling. This story occurs in all sorts of insurance. The nice simple concept of insurance as a homogeneous pool unravels.
One area where this is especially evident is medical insurance, where someone with a poor medical history or heredity faces a very heavily weighted premium, and may only be able to obtain cover within a group scheme run by his employer, where the employer and insurance company collude, effectively to enforce some cross-subsidisation from healthy employees to less healthy ones.
The more that we have sophisticated techniques for calculating risks, be it through the likes of DNA testing for medical or life insurance, seismic or weather forecasting for building cover, or behavioural profiling for car drivers, the less cross-subsidisation will there be between low and high risk categories, and the more expensive, perhaps prohibitively so, will insurance become for the top tail of the risk distribution.
Then, in the case of the car, a driver assessed as very high risk may have to give up driving altogether because he cannot afford the premium. In the case of the house in a high risk location, you can move to a safer place, or, if you cannot afford to insure, you might just sit it out and depend on the mercy of the state if the worst happens. In the case of health, there are limits to making yourself healthier - you could slim or give up smoking, but not alter your DNA. Ultimately you may be denied insurance, or have to opt out because you cannot afford it; thenceforth the state (or possibly some charitable organisation or a benevolent employer) is your fallback.
In some respects the state may be able to limit its potential liability. In the case of housing, it could refuse planning permission to build houses on river banks in the first place. But, despite all the encouragement and exhortation about the benefits of insurance and the need for individuals to take their own steps to obtain cover and not rely on the state, there will always be a substantial residual liability on the state, and in some fields of insurance that liability may grow as society's ideas of social responsibility and decency of life grow, and as the sophistication of the insurance industry's screening procedures exacerbates rather than resolves some of the problems.
Ultimately, insurance becomes so refined that, from the customers' standpoint, it is not a risk pooling within a large population, but something more akin to an individual profiling where, in the theoretical extreme, the present value of the premiums you pay equals the present value of the statistical expectation of the claims which you yourself will make, plus a fee component. It serves as a savings medium or liquidity provider. The advantage of this development is that it focuses the minds of high risk people to try to improve their profile. The disadvantage is that, regardless of the extent to which the government may have successfully persuaded Mr Average to insure through the private sector rather than depend on the state, it leaves the tail to the state (possibly to a greater degree than before), and relegates those who are incapable of raising their profile - for example because of heredity - to a degree of destitution, the exact extent depending on the generosity of the state.
But governments are becoming increasingly nervous about assuming open-ended obligations, in the role of a sort of residual insurer. They worry about the likely burden on the budget from such things as rising longevity or burgeoning medical costs. Hence one witnesses the introduction of mandatory savings for old age, such as the MPF introduced in Hong Kong a few years ago; and there is talk now about medical cover along similar lines. And George W Bush has indicated his intention, or perhaps better described as his wish, to in effect privatise US basic pension provision so as to alleviate the potential long-term burden on the federal budget.
But private insurance or saving schemes are unlikely to provide satisfactorily for those at the tail end of the distributions - the very sick or those who have lived so long as to exhaust their savings. Private insurance may never be able to provide enough for them; the insurance companies would be reluctant to take on such clients if they could accurately foresee the circumstances, and scientific advances are increasingly enabling them to do just that. So it is fanciful to think that the state can disengage completely. The more astute that private providers become, the more the high risks are likely to end up on the government's plate.
Fairness. morality, etc
This paper has been written on the assumption that investors or insurers are entitled to obtain all relevant information about the persons or entities which they plan to fund or insure, and to use that information to discriminate between those parties and so set differential prices. Particularly in the field of insurance, questions have been raised over the morality of discriminating between genders, or between able-bodied and disabled people, and so on. And there have been associated calls for legislation to outlaw certain types of discrimination or compel insurance companies to provide cover. However, those issues are not addressed in this paper.
The availability of increasingly sophisticated information and methodologies enhances predictive capabilities, and so allows finer identification and pricing of risk.
In the context of the provision of capital, this might be seen as likely to reduce the funding available, from a predominantly risk-averse population, to ventures bearing the highest risk. But it seems likely that, amid rising affluence, so long as people are allowed to become rich and, importantly, to retain a reasonably high share of any rewards from risk-taking (rather than having to surrender them as tax), there should be a compensating increase in the supply of high risk takers, albeit a minority of the overall investing population. The end-result may be that capital as a whole is allocated at least as efficiently, and perhaps more so, than formerly.
On the insurance front, risk profiling has also become much more refined. In circumstances where the risks cannot be averted, such as in matters of basic livelihood, the state may find no escape from having to shoulder the responsibility for the high risk categories, even if the average and lower risks may increasingly be covered through private sector provision.
Adapted from an address to the Informatics Symposium on Risk Management, Hong Kong, 1 April 2005.
Prof. Tony Latter is Visiting Professor at the School of Economics and Finance of the University of Hong Kong.