Monday, 22 September 2008

New ways of managing risk


We need certainty about the future in order to do anything. It’s a basic requirement of the action-feedback cycle that intelligent beings use to achieve their goals.

On the simplest level, if I want to pick up an apple to eat, I need to know that when I tense my arm muscles, my hand will move in such a way, and when I grip it and pick it up, it will weigh (more or less) how much I expect, so that I need to apply just this much pressure to bring it to my mouth. I need to know that it will be this hard to bite and that it will give me some energy and assuage my hunger. If I am missing any of this information it’s highly unlikely I’ll be able, or even want to try, to pick up and eat the apple.

In a more sophisticated decision I need the same predictability. I have to know that when I hire this person, they’ll show up at work, and when I sign this contract, the customer will pay the money on time, and when I build this car, someone will buy it. If we have certainty, it lets us see clearly how to make our lives better – because we can simply look at the future consequences of each action and determine which one is the best.

However, I can’t get absolute certainty on any of these decisions. The new employee may stop turning up, the customer may go bust or nobody may want the car. I can estimate the probabilities but I can’t make the uncertainty go away. So how do I take the action if I don’t know what the result will be?

To some extent, of course, I can act without complete certainty. If I’m not sure how much the apple will weigh, I can apply extra effort just in case. To mitigate the consequences of my employee not showing up, I can make sure not to book any meetings during her first few days, so I’m available to fill the gap if she isn’t here. These solutions clearly reduce efficiency and waste time or resources. The greater the certainty I can achieve, the more efficiently I can work. Tools for certainty are one of the critical contributors to a successful economic system.

This is one of the three reasons that credit exists (the others are to enable pooling and specialisation, and to transfer resources through time; to be dealt with later). Credit allows me to take actions, without full certainty about the outcomes, because I know that I can at least feed myself in the meantime. It lets the risk of default be pooled in return for an insurance component to the interest rate; it allows us to act as if we have full certainty without incurring the high cost of achieving it. It gives me the breathing space to know that even if something does not go as planned, I’ll have the chance to fix it.

Indeed the availability of credit actually increases the net amount of certainty in the world. Here’s a short example illustrating why. If my risk of default is 10% and so is yours, the risk of either of us defaulting is 19% but the risk of both of us is 1%. When you have 100 people involved, the chance of at least one person defaulting is 99.997%, the chance of at least 10 people defaulting is 57%, but the chance of 100 people defaulting is only 1 in 10100. You can make a highly accurate and certain prediction of your loss, and therefore turn the situation into one of certainty for all 100 participants. (This assumes that the chances of default on each loan are mutually independent – Nicholas Taleb has explained in depth why that’s not always true. One of the mistakes was to package up debts – such as 100 separate subprime mortgages – with a highly correlated risk of default. But the principle works if applied sensibly).

John Gapper writes in this week’s Financial Times that credit derivatives are a zero-sum game, because losses equal profits. But that’s not quite correct. Even though the money that’s transferred under the derivative contract nets to zero, the certainty that the tools provide (if designed correctly) creates non-zero-sum outcomes.

So the growth of credit in the last ten years has enabled people to act as if they have a lot of certainty. Companies have invested hundreds of billions on the basis of future expected returns, because they’ve been able to borrow to finance it. Millions of new services and products have been created out of this investment. Consumers have increased consumption because they ‘know’ that they can pay for it later, and it will bring them more joy to own something now than in the future. Our lives have genuinely become better because of credit.

If there is no more credit in the economy this is going to stop. Of course, there won’t be no credit but there will probably be much less. This takes away one of our major tools to deal with uncertainty. This will reduce the amount of economic activity because people cannot invest or spend without some certainty that they are able to pay for it, and that they will get something back in return.

So what will we do instead to reduce uncertainty and enable people to spend again? There are some alternatives:


  1. Better data and analysis. Storing and using information better, and interpreting it correctly is an excellent way to reduce uncertainty. We have much better IT now than we ever had (and no reason this improvement should stop) so the likelihood is that we will gain certainty in this way.

  2. More transparency. One big reason for lack of certainty is that other people have information that we don’t know. The more open we are about our businesses and our lives, the more accurately and confidently others can make decisions on that basis.

  3. Participation in commitment structures. If we commit ourselves to act in a predictable way and make it credible that we’ll do so, others can make accurate decisions on the basis of our future actions. This is what contracts and laws are for – but we can go beyond the existing structures and make ourselves more reliable still.

  4. Tests and prototypes. By providing ourselves with laboratories – spaces to try things out quickly and test the consequences – we can gain greater certainty about how things will develop.

  5. Raising the status of knowledge and trust in experts. Knowledge is what provides certainty, and a time investment is needed to create knowledge. That time investment will get the best return if it is not duplicated – if we trust in worldwide centres of excellence which have the opportunity to explore aspects of the world (and especially the social sciences) and publish their results so we can all use them.

All this isn't quite a coherent economic strategy. Yet. Give me a little longer...

1 comment:

Ilya said...

I know this is an old post but might as well leave a comment.

I take issue with item 5. It has been shown in the past that experts are not much and in many cases may be inferior to either the average pundit or, as is sometimes argued, the wisdom of crowds. See Tetlock's book w/r/t expert judgement as well as Jim Surowiecki's book.

The other problem is that in some respects this may prove to be counterproductive. Expert knowledge implies concenrated rather than diffused knowledge raising nodal correlation risks. A better way may be increasing everyone's level of knowledge rather than relying on a select few, thereby dispersing the risk of a bad judgement by an individual expert.