They really aren't out to get you, George

There's an interesting article in the London review of books this week by Slavoj Zizek. It quotes the Commandments of Liberal Communism (original from Technikart magazine):

1. You shall give everything away free (free access, no copyright); just charge for the additional services, which will make you rich.

2. You shall change the world, not just sell things.

3. You shall be sharing, aware of social responsibility.

4. You shall be creative: focus on design, new technologies and science.

5. You shall tell all: have no secrets, endorse and practise the cult of transparency and the free flow of information; all humanity should collaborate and interact.

6. You shall not work: have no fixed 9 to 5 job, but engage in smart, dynamic, flexible communication.

7. You shall return to school: engage in permanent education.

8. You shall act as an enzyme: work not only for the market, but trigger new forms of social collaboration.

9. You shall die poor: return your wealth to those who need it, since you have more than you can ever spend.

10. You shall be the state: companies should be in partnership with the state.

The author then goes on to say that people who adhere to these commandments (in whose ranks he counts Bill Gates and George Soros) are the real enemies of progress.

Now, one never likes to punctuate a well-crafted delusion, but sometimes they really are not out to get you, Slavoj. A major proponent of the capitalism is an evil conspiracy theory is George Monbiot, hence the title of the article: he makes the same error as the LRB writer, the mistake of anthropomorphising the system. So, George and Slavoj, there is no single entity which is the capitalist system. There is no smoke filled room full of oligarchs plotting the destruction of liberties and the enslavement of the world's workers. There are just (not very) rational economic actors, working within an economic system. The system can sometimes produce consequences most of us would agree are desireable, like increasing wealth; and sometimes ones that are less so, like the farrago that is most of the developed world's protection of its farmers at the expense the less developed countries. So let me add an eleventh commandment.

11. You must understand the system and its rules, for that determines everything that is possible. If you do not like what is possible, then you must strive to change the rules.

Throwing out the rubbish is the hardest thing

Time and time again it strikes me that much software would be improved by asking the question at the design state `when do we expect to throw this way' and budgeting for that. How often do you have to deal with a plain out-of-date system that is absorbing massive amounts of its creators effort simply to keep running? Fairly often if you have anything to do with UK public sector systems. My favourite example this week: government gateway logins for the inland revenue are incompatible with ones for VAT - what?

Anyway. Something to add to your list of design principals when you are designing software: write down the criteria for throwing the system away before you get locked in. You'll be glad you did.

The Local Meme, Part II

Plato was an aristocrat: the idea that you could know everything, that you sat, imperial, having information delivered to you, was completely natural. Perhaps if epistemology had started on a different track, it would not now be so challenged by finding that this model isn't the way nature seems to work. For now at least, it seems that the Platonic observer is far too distant from the world. Instead, we have to leave our chairs to find out about reality, and whatever we do changes things. Is that really so unnatural, so dissatisfying?

In mathematics, isn't the idea that you select a meta-mathematics for the task at hand completely reasonable? You don't try to do the plumbing with just a hammer and nails, after all. And the axiom of choice is no more or less than a tool to be chosen or ignored, as the task at hand requires. It is no more `true' than a hammer is: asking for the global truth of mathematical foundations increasingly seems to be a category error.

So, gentle reader, celebrate the triumph of local truth, of good enough for the here and now. Global truth is not just a dangerous ontological abstraction, it doesn't even make sense.

The Local Meme, Part I

A couple of years ago I was chatting to an acquaintance and, like many people, he was fascinated by quantum mechanics. I resisted the temptation to suggest he learn about Hilbert spaces before trying understand the theory of observation in QM (so, right, you take this operator wotsit, and you apply it to the state vector, right, and ...) He did make me realise one thing, though: several of the great intellectual achievements of the last century have a common thread - the importance of the local. So at the expense of 99.9...% of the important stuff, here goes:

Special relativity overturns the primacy of the Newtonian observer and focusses attention on the relationship between different observers. What you can see depends on where you are and how fast you are going;

Quantum mechanics says that there is no information without observation. You interact with the system to measure it, and what you do changes the system. Again, there is no gros picture, no privileged observer (at least leaving aside the vexed question of the wavefunction of the whole universe).

The various incompleteness theorems of Gödel and others again assert limits to global knowledge: you can have this property of a proof system, but not that too. If you want all of arithmetic, you have inconsistency. (And interestingly these proof theoretic results now seem to be important for understanding the P = NP problem.)

And then there is constructive mathematics where to prove that the exists an x such that some property holds, we have to actually hold up that x and show it has the property - another local idea.

In Part II, a (probably jejune) way of thinking about how this fits together.

The Price of Obscurity

What price knowledge? Recent graduates may have difficulty attributing any positive value to this commodity, but I want to be more specific - what is it worth to conceal your position?

A good example of this question comes up in the liquidity market. Suppose you are a bank and you need to raise 10B euros of short term funding, and this requirement is both persistent and not particularly volatile. You could go to the cheapest source, perhaps the interbank market. But after a while, given the limited number of participants, the market would know your funding requirement.

There are other sources of funding, of course: CP, CDs, lines, and so on. Spreading your funding around different markets is viewed as good practice, but it is obviously more expensive than just using the cheapest. (I'm not talking about duration here, just source of funds.) But if you move your funding around, a billion here, a couple of billion there, it is much harder for other participants to figure out what your requirement is and how volatile it is. Then if you need to draw on a line, it won't be a surprise, because you have done it before: there will be no information attached to you making a draw down.

Clearly a strategy that means you do not signal your intentions and situation to the market is helpful. But what is it worth? If extra line costs 10bps, when is that worth having? If you have to pay another big blind to be allowed to wear shades when playing poker, what would you do? An interesting problem...

The Game of Monetary Policy

Given the budget is close, let's talk Economics. Here's a model of how to win a Nobel prize in that esteemed discipline. Come up with a macro economic theory, publish it, have it suffiently well accepted that governments use it to set monetary policy.

This is difficult, of course, but at least the programme is clear. It is a model from science: discover something, publish it, help explain the world, add to knowledge.

But what if that was not possible? What if Economics is fundamentally different from Physics, in that the very act of discovering something and using it makes it less likely to correctly predict future behaviour? It could be that the economy does not just change fast, the dynamics change too. The way to understand it next year won't be the way to understand last year, not just because we will know more, but because last year's theories won't work any more.

Bigger and better?

I have just got a new display for my computer, a deliciously big, high res LCD panel. And there's just one problem, a problem I didn't anticipate. My pictures don't look nearly as good on it. It's the resolution. At 1024x768, my old resolution, it's easy to find a nice looking desktop, for instance. But there isn't much available at 1920x1200, and the old stuff just doesn't look very good blown up to fill the new screen. So, gentle reader, be warned, your new display might very well make your old pictures look terrible, which might not be quite why you bought it.

Good going on bad

Why are expensive hotels so dissatisfied? Thanks mostly to my job, I have had the opportunity to stay in a variety of not-cheap hotels around the world, and overall, the experience is much worse than staying in budget places. The problem is they promise so much: in a cheap hotel, you are grateful for a comfortable bed, a quiet room, hot water in the morning and neither boiling nor freezing during the night. It is not difficult to meet these rather basic needs.

An expensive hotel, on the other hand, promises fine dining, attentive service in every detail, a delightful experience. But these hotels are staffed by people, and people are often not very good at considering others. In the Ritz, a snooty waiter ruins the experience. At the Georges V, a small room leaves you vowing never to return. The pervasive smell of boiled cabbage in the Ritz Carlton's meeting room is intolerable. A maid who cannot read the 'Do Not Disturb' sign at the Park Hyatt infuriates. Hotels at this level have set up the game so that they are likely to fail: anything less than fantastic is unacceptable, because you have been promised fantastic.

One motto for anyone delivering a good or a service should be `meet customer expectations'. If you can't, change their expectations. Sadly, most luxury hotels are competing to raise expectations, not to surpass them. Perhaps a new business model is needed?

Market success, market failure

Liquidity is a coming theme in the financial markets. It is not the same thing as volume or turnover: having liquidity means you can sell something when you want to, not just in ordinary market conditions. There are signs that, despite rising volumes, liquidity (in this sense at least) is decreasing: Avinash Persaud uses the term `liquidity black holes' for those times when the market dries up and prices change dramatically without trading being possible. I think there is a good deal to be said for his thesis that these black holes are getting more frequent and more intense, but that will have to wait for another day: today I want to talk about the importance of liquidity in establishing a market.

In recent years we have seen several whole new markets appear: credit derivatives have transformed the trading of credit; weather derivatives have dramatically extended the trading of environmental risks; electricity derivatives are broadening the trading of power-related underlyings from the physical to the purely transferrable. Why do these new markets sometimes take off (credit derivatives being a good example) and sometimes fail to gather momentum (cat futures, despite the moderate successes of cat bonds, have never drawn much trading interest)?

One key is obviously liquidity, or at least the ability to hedge. The (single name) credit derivatives market started as an adjunct to the corporate credit market, and you could always hedge there. Moreover, most potential participants had risks that could roughly speaking be covered by buying a default swap. Weather trading has been slower because there is a lot of basis risk: you want to cover the risk of a low rainfall in Yorkshire because you are a water company; I am a Lancashire ice cream company who is willing to give up some of my excess profits if the summer in Manchester is hot. There's nearly a trade between us, but not quite: hot summers in Manchester do not necessarily mean low rainfall over Leeds and vice versa. Hence we both end up paying an intermediary, an investment bank perhaps, to run that basis risk. The bank charges both of us and the market has less liquidity.

The thing that brought this theme to mind was the nascent market in property derivatives in the UK. This has all the ingredients for success: a good underlying (the investment property database, which covers over three quarters of commerical property); an attractive asset class; high costs for trading directly in the underlying (roughly a 7% bid/offer spread); and a reasonable regulatory framework (now property derivatives are admissable assets for insurers). It is not perhaps quite the Kaiser Chiefs (I predict a riot), but certainly considerable growth can be expected.

Just one word

I've just installed Office 2003 and, to be fair, it was incredibly straightforward. Insert disc; enter serial; wait; done. For what is one of the most complex pieces of software commonly available, that is pretty impressive.

Unfortunately, that's where the problems start. I had forgotten how much `I know better than you' there is about Word and its friends. The full horror only returns when you have to remember how to turn all that nonsense off, again. It's nanny software: here, let me help you with that, did you mean this?, can I make that ugly for you?

Microsoft have, understandably, produced a lot of `features' that people may find convenient. But because they have constructed something over years, in archeological layers, there's no single place, no design model. Some features are defined by registry keys, some are in normal.dot, others live in yet more obscure files shared across office or program specific. It's a mess and it's hideous to mess with.

So today's law of software is `when you improve it, don't just add features, add the management of those features consistently'. Think how much time the microsofties must spend on maintaining that piece of baroque complexity. Modularity and consistent interfaces are hardly new ideas, after all. Maybe someone should send Bill a copy of Learning to Program in Pascal?

Taxing problems

One last example from Finance, and I promise I'll talk about something else next. But an article in today's paper made me think again about the tax system and how silly the rules are. It's not that I think taxes are too high, or too low for that matter - it is just that the system is so complicated. Think of what that complication supports: tax lawyers and accountants, offshore SPVs for companies and Guernsey Trusts for individuals, the whole panoply of tax planning and tax optimisation. What does this do for the real economy? Nothing. Indeed, less than nothing, as it deprives the chancellor of a goodly number of hospitals and schools and all the growth they would generate.

This is why a flat tax system is a good idea. Of course there would be winners and losers, and if absolutely necessary I'm sure the Treasury could find some system of grants to address any perceived gross injustices. But think what we would gain from this simplicity. A whole useless segment of the economy disappears, and some of those people might do something that makes an economic contribution to the country instead. The brutal burden of completing all those tax returns would be lifted, freeing up millions of man hours. And Gordon could manage the Revenue with a fraction of the staff it has now. How's that for an efficient rule change?

Getting paid for Gaming

Suppose you are a trader in an investment bank. Suppose the world is fair, so you get paid for every pound you make over your budget. In options terms, that means you have a call on the P/L struck at the budget. How do you increase the value of a call? One way is to make a lot of money: raise the forward. But another is to raise volatility of the underlying: make the P/L more volatile. Taking bigger risks is good for this. Now, was this the kind of behaviour we wanted to encourage when we designed the bonus programme?

It gets better. Suppose you manage traders. You get paid for every pound your group of traders makes over budget. So you have a call on a basket of P/Ls. How do you increase the value of a call on a basket? Well, the two methods above work, but so does increasing correlation - get everyone who works for you to take the same risks. Hmmm. Not quite what we wanted there either...

Why two strategies are often better than one

Suppose you're playing poker. You look at the cards on the table, and those in your hand. You can't be beaten. So you raise.

Corporate strategy discussions use lots of analogies like that, and they can be helpful. But what they hide is the fact that in the real world, we very rarely have complete information. We don't know what our competitors are doing; we don't know what factors in the world might change. So even if we could come up with an optimal strategy based on what we know, it might be invalidated by facts unknown to us.

The risk mitigant to uncertainty is often diversification. We don't know which stock will go up, so we buy them all, or a lot of them. But in strategic discussions, how often do companies consider having two strategies or three or a dozen?

Clearly it is not efficient to try to do three things when you could do one. But it is not efficient to buy the top three performing stocks: you could have put all your cash in the best one if you knew which one that was going to be.

There is a lovely piece of research about how this relates to executive behaviour. Suppose I give a hundred pounds to ten people to enter a large poker tournament, with a prize for the one who wins the most. It is likely that the winner will be the person who made the biggest bets and was lucky, not the prudent gambler. So if we reward executives solely by the size of their financial successes, we discourage diversification, so when they do get it wrong, they get it very wrong. There have been so many examples of this recently it would be jejune (as well as potentially actionable) to name names.

So by all means have a strategy. But part of that strategy should be what you do when the strategy goes wrong and how you spot that that has happened early enough. One solution might be to put, say, 50% of your resources on your best guess as to the right thing to do next, 30% on another likely possibility and 20% on a rank outsider. At least you'll have the other projects to fall back on it you turn out to be wrong about the best strategy. Think of it as an ecological approach to corporate strategy.

Finally, Maynard Keynes is important here:

When somebody persuades me I am wrong, I change my mind. What do you do?

Strategies have to be monitored regularly and resources reassigned in the light of new evidence. Keeping an open mind about what might work in changing conditions is vital.

Actuarial Advice, Part II

Now my point in Part I was not to be the latest in a long line of people to point out how foolish the actuarial assumption of equity returns being 10% forever are.

Rather, it is that the rules of the system have produced the behaviour.

If you set things up so that you are given advice where the 'best' thing to do today is purely defined by what happened in the past, you run a lot of risk that the future will be different to the past and hence that the advice will be misleading. One of the pieces of information missing in the old actuarial advice was some measure of the probability of things going sufficiently differently that pension obligations could not be met: perhaps if trustees had had that, they might have invested differently?

In the absence of any theory which we have reason to believe governs the behaviour of a variable, didn't David Hume point out the error of thinking the past behaves like the future? For the financial markets this is even more clearly silly than elsewhere: the global economy is obviously very different now from the one we had in the 80s, let alone the 50s. So why should stock prices follow paths characterised by statistics from long ago? That is not to say that we should not pay any attention to the statistics: just that we should be aware that there is model risk in how we use the past to predict the future, and for the sake of the next generation of pensioners, perhaps that risk should be considered along with all the others involved in running a long term investment portfolio.

When we use mathematics to model the world, as in fitting a return distribution of some financial asset, there is the danger that we use the maths that is convenient rather than the maths that captures the essential features of the problem. In finance, for instance, we are so obsessed with normal distributions that we use them whereever possible. Part of the reason for this is that so much is known about them -- we have a lot of tools to hand. Also, the errors made by using a normal distribution are often small for typical financial applications (especially once we hack in the implied volatility smile). That doesn't mean that the assumption that (log) returns are normally distributed is always good, though.

Phillipe Jorion has a insight into the dangers here in his paper on the fall of LTCM ( then search for Long Term Capital Management): he shows how making an modelling assumption, that correlation is stable and the return distribution is normal, leads to a dramatic understatement of risk. Sometimes, which tool you pick makes a lot of difference, and familiar tools can be the riskiest ones, not least because everyone else is using them too.

This is an education issue: the next generation of mathematical modellers needs to be taught how to model, but also about the dangers of modelling, about the need to look at a problem through the prisms of different models.

Turning back to actuarial advice, we have people trying to model the future using the past but without a theory that explains the dynamics, a system that encourages them to give their best guess with quantifying how wrong that guess might be, and a predilection for using tools that have nice mathematical properties but fail to capture significant features of the real world. Is it any wonder we have a pensions crisis?

There, I managed to talk about actuarial advice without mentioning the Ljung Box statistic once...

Actuarial Advice & How It Serves Us, Part I

Pensions are hot, possibly for the first time ever. There is a lot of press comment about underfunded pension funds and many people are becoming aware that their retirement may not be as comfortable as they had hoped. So it is not a bad time to look at one of the systems which underlies pensions, that of actuarial advice, and what it does.

Now what pensions actuaries do is rather complicated and any attempt to summarise it in blog length is at best vainglorious. But here goes. Firstly the problem. A pension fund takes money from people over some fraction of their working lives. The cash is invested. The fund has the liability of providing a pension, either based on pensioner's final salary (defined benefit, or DB) or on how the investments have done (defined contribution, or DC). In the absence of free money (aka a government guarantee) the amount a fund can pay in a pension depends on how much has been contributed and what the investment performance has been.

In a DB scheme, a individual's pension is determined by how the whole scheme has done; more recently for most DC policies, individual pensioner's assets are ring fenced.

So where do the actuaries come in? For a DB scheme, they advise on the investment strategy and the contributions needed to meet the liabilities. I'll concentrate on this kind of scheme, as they are the most interesting and the most problematic.

An aside on the individual versus the collective: In a DC scheme the typical member chooses how to make his contributions. If they do well, the member has a great retirement income. If they do badly, forget about that villa on the Riviera. The problem is that many people, myself included, find managing investments fundamentally boring. Most also have little or no education in it. So the growth of DC schemes combined with low education in investment fundamentals is likely to result in a significant number of pensions which will not provide their beneficiaries with a good standard of retirement income. This is hardly good for society. In a DB scheme, in contrast, short term mistakes in investment performance can be corrected by higher contributions by everyone: the scheme is a pool with money always coming in from current contributions and (after an initial delay) always going out to pensioners. In this setting, if more goes out than comes in, everyone suffers. So we have a classic prisoner's dilemma: to what extent should the individual subsidise the collective, and to what extent should he or she be able to rely on them for support in the event that things go badly?

Next, actuarial advice. A range of assets - different equities, bonds and so on - are available for the pension fund to buy. Which ones should they pick? In the very long term, it seems so far, investing in equities has resulted in higher capital appreciation than investing in safer assets like good quality bonds. On the other hand, in the very long term we are all dead, and in the shorter term there have been several extended periods where equities have underperformed bonds. DB pension fund trustees have to invest members contributions in order to have enough assets to meet the required pension liabilities. The advice they receive from actuaries in the past has often highlighted the historical outperformance of equities and hence influenced trustees to pick higher risk investments (the mean of the return distributions). What it sometimes did not highlight was the risk that equity markets might not outperform (some measure of the sample variance*).

What some actuaries did, in other words, was to build a model based on the past to predict the future. This is not a model in the sense of Newtonian mechanics or Relativity: extrapolation might be more accurate than model as there is very little theory underlying the idea that if something has grown at 10% for the last ten years it will carry on growing at the same rate for the next ten. The risk of doing this is obvious: if the world does not behave as your prediction suggests it should, decisions taken on the basis of the prediction can seem to be rather bad ones. And of course this is what happened with DB pension funds: investment decisions were made on the basis of the outperformance of equities, then when equity markets fell in 2001-2002, many of those funds did not have enough cash to meet the promises they had to keep. Hence the pensions crisis.

In part II, why the way actuarial advice was framed made this more or less inevitable.

* Assuming there is a process rather than just a sample variance is another modelling choice which might or might not be sound.

Wake up and smell the coffee?

Firstly an apology for the dreadful pun.

This blog is going to be about systems thinking and games in the large: rule-based systems and their behaviour. Part of the motivation for the name came from thinking about one of my favourite cafés, Dal Professore in Naples (try the Nocciolato, it's great). In most large Italian cities, coffee houses of this quality are routine and the coffee is cheap: perhaps one euro for an espresso. Yet outside Italy, 'coffee houses' which are both much more expensive and serve considerably worse coffee dominate. Why? The outcome in Italy seems, to me, better than the outcome elsewhere, and it is certainly different. What is it about conditions there that has allowed places like Dal Professore and thousands of others to thrive?

The answer is clearly complex: the Italian habit of taking a quick coffee standing up obviously allows for faster throughput; a focus on the flavour of the coffee (rather than the size of the cups or the number of different varieties of muffin); a culture that encourages opinions about taste; an economy that supports small businesses versus large chains; and so on. I'm interested in how the rules determine the behaviour. Usually, I think there is no deus ex machina dictating the outcome, no system to be praised or condemned. There are just complicated interacting constraints, contexts and capabilities which determine what happens. Hopefully this will be
a space to discuss this kind of systems thinking in a number of domains. Welcome.