Monday, March 3, 2008

What took you so long? Mismatched incentives and price-less risk

From the FT this morning:

“It seems that equities are finally becoming aware that all other asset classes are in risk aversion mode,” analysts at BNP Paribas said. “They are also recognizing that the prospects for profit growth in the near term could be constrained, given the procession of economic data highlighting the fragile state of the US economy.”

Analysts get paid to stay informed, process the information efficiently, and act accordingly. So it's amazing, ain't it, what they come to find only at the start of March? At least they didn't wait for the ides.

One can no longer be surprised by the capacity of analysts/traders/fund managers to cover the totally obvious through a mystical invocation of a new set of laws for the Universe of Vanishing Risk. Housing bubble, dot-com boom, this time, at least, we're being spared the justifications: US stocks had simply been doing quite well since January relative to other domestic and foreign assets in the face of worsening domestic econ news (even housing stocks have had their rally). If anything, I would venture it's the conviction that bad news are merely triggers for the Fed to shoot down rates further so the party can go on.

The magic spell of the moment is really irrelevant: it will happen again, we will recover and then, a few years from now, the experts will claim they've again discovered a new universe beyond the laws of economics ("this time it's for real!"), will forget to account for risk, and will try to pull us along with them. The irony is that we know it will happen precisely because it's all laid out there by those very same laws of microeconomics, most basic of all: people react to incentives.

What we ought to do is look at what the incentives facing analysts, traders, and fund managers (specially hedge fund managers) are. And the constant has been, as far back as we care to go to understand today's financial architecture, that these incentives are skewed towards short-term horizons and divorced from the downside-risk the agents' actions generate. The results are rewards virtually unlimited on the upside and limited on the downside to, in the worst case scenario, temporary unemployment (has anyone tracked the career paths of hedge fund managers whose portfolios tanked?).

To go back to my leitmotif: a lot of the risk gets priced close to zero because the agents making the calls are essentially not facing it.

With that in mind, there's no mystery left to explain: it's not stupidity and there certainly is nothing irrational (nor evil) at play; it's plain self-interest on the parts of both these market-movers as well as the institutions who hire them, at the expense of clients with (rationally?) limited information (who are also acting on their info-constrained best interest).

(In a darker angle: Kerviel's defense in the SocGen scandal boils down to "SocGen knew what I was doing and didn't care as long as I was making money for them.")

So if it's all about self interest, should we care? Isn't it what markets are all about? My pet peeve here is the considerable negative externality (and the perverted incentives) exerted on others by these actions when they find fertile ground in enough actors on the other side of the fence to tilt the markets. Those who do get informed and "do the right thing" (in this case, those on both sides of the market who follow the path given by fundamentals up to the existence of a bubble) not only miss the unsustainable gains on the upswing, but even worse, when things inevitably come crashing down, end up suffering the adjustment just like those who played hard'n'reckless, and worst of all, have to bail out those who bet and lost.

But no, the point here is not that the government should step in and regulate the life out of markets: please, I'm an economist! Indeed, I think federal and state governments have contributed to the mess in the housing market through their tax incentives in favor of house ownership (or, rephrasing this, against renting). And I really don't see how the government could improve on the reasons causing the market to fail to self-regulate.

On the contrary, it is conceivable that the mismatches in the planning horizons and allocation of risk could be eliminated so that private contracts and compensation schemes would suffice to produce better results.

For this to happen, stakeholders in the financial markets need to find a way to overcome their collective action issues and design better incentives for themselves, by which I mean more closely aligning their corporate interests and that of its market-informing/moving employees with those of the clients they inform and for whom they act as agents. In particular, this means incorporating into compensation schemes a sense of the risk, both on the upside and the downside, that their employees' decisions generate on clients' returns, and to extend the horizon of this effect on compensation over a few years.

In fact, both things work together nicely: smoothing earnings over a longer period provides incentives to think in terms of risk over the longer-term while allowing compensation to be based on realized risk, including a bigger downside. On average, agents' compensation would better track their record.

Easier said than done: the coordination problem starts when investment banks, hedge funds, etc. grow afraid they will chase talent away if they move away from immediate, limited-liability returns in favor of time-smoothed contracts that load more risk. It grows deeper when imperfectly informed clients believe the illusion of risk-less returns coming out from the "experts" and demand such returns. More generally, there are too many stakeholders (on both sides, but so many more on the clients' side) for transaction costs to allow spontaneous Coasian bargaining to get to the efficient result.

So yes, perhaps some (public) institution might be needed to provide a forum that allows this coordination to take place in. Not to blurt out even more complex regulations that will inevitably be ran loops around, but to help the parties involved overcome the costs of re-starting the system from a more efficient equilibrium. Unfortunately, the financial markets have consistently proven their inability to stop these "misalignments" (I won't call them all bubbles) from happening over and over again on their own.

Otherwise, the rational thing will be for all of us to play dumb and go for broke every few years when markets spin out of control because of some screwed up incentives scheme. Not a good equilibrium.


Luis said...

Ok, so this may be the first comment on your blog. I'm popping your blog cherry. Anyway, metaphors aside, some business first. To check out the FT article w/o registering (how annoying) head over to They can provide you with all the phony e-mail and password combinations you need to avoid spam.
n to the post:

You said: " The irony is that we know it will happen precisely because it's all laid out there by those very same laws of microeconomics, most basic of all: people react to incentives.
What we ought to do is look at what the incentives facing analysts, traders, and fund managers (specially hedge fund managers) are."

While I agree in principle, the problem with this approach is that very well paid people are already engaged in this very activity. We DO look at the incentives facing analysts.

There are two problems hindering actions, as I see it, to change the current system. The first I think, is a basic problem with our brain's wiring. As you stated, we tend to price down to zero risks that are outside our immediate timeframe. I don't believe we do it out of malice, but rather out of human nature. Our reptilian brains react to immediate dangers. The same thing happens when we look at incentives. Unintended consequences love to creep on in. Part of the reason is is that often we do not behave in a rational manner. What?! That's not true! Economist cannot tolerate this! Well, I'm sorry to inform you, its true. As an example Dan Ariley so artfully points out in "Predicatbly Irrational", we do irrational things such as get suckered for "free" things. (we will even give up a better option for the lure of free!)

The second reason I think the incentive system has not changes is because (for the most part) it "works". It works for the way we are wired and it works in that we achieve (while maybe a "sub-optimal;" in your view) growth. Yes, occasional disastrous blunders are made and a trader loses millions or we headily invest in web-based-grocery-delivery schemes. BUT, and here is the but: what is the incentive not to make fist-fulls of cash in the here and now (both the trader and the company) when it is only a remote possibility that a disaster may occur? What's more - as you said - a time-smoothed risk curve offers consumers (investors) a negative incentive given we cannot price future risk correctly. We cannot help ourselves. The question is: which is more optimal given our human nature - a boom bust cycle or smooth sailing? You have to admit, the rush of boom-bust has its appeal.

ram said...

Luis, thanks for your posting!

As I see it, there are three parts to it. I agree with two, but with caveats.

"What we ought to do is look at what the incentives facing [them are]" was meant as a rhetorical device to point at where I thought the problem was (not evil, not stupidity, but poorly designed incentives), not as a suggestion that no one had paid or was paying attention to compensation mechanisms. And I think the evidence of one crisis after another suggests that whoever designed these botched things before and might keep doing it as long as the risk passed onto clients by the decision makers is so removed from the (much smaller and asymmetric) personal risk assumed by the decision makers.

Let me put it another way: after a lot of thought, someone designed stock options schemes too, thinking it'd make things better aligning the interest of managers and owners, but then evidence came back to show that we'd missed a crucial aspect or two. Same thing here: once the evidence shows we screwed up our best laid plans, it's time to change those plans.

On your second point: I agree that the rational agent model so beloved by economists is controversial and can be used incorrectly. But I think that those who like to point out that humans are not (as) rational as the simplest econ models claim are not only missing the point, but getting it wrong through exaggeration in their desire to pull an "AHA! gotcha!".

First, I agree: reliance on the rational agent model is incorrect when taken as an article of faith, as "the truth" (and I make that mistake a lot). However, it is a fantastic device when it serves as the starting point that gives structure to a research agenda; moreover, a lot of microeconomics in the last couple of decades has precisely been an effort to show what are the effects of lifting some of the constraints imposed by such an abstract model. In fact, our well loved, but not-so-new (though only now available for mass consumption) field of behavioral economics (and neuronomics coming up soon) works in exactly that way: you start from the rational agent model, you observe something that doesn't match, you see what you need to lift in the model to get there, you then ask what other of the old conclusions that one modification forces you to change (that's when the policy advice is generated). Great structure, which is what is unfortunately missing in some of the unorthodox research agendas in economics and in a lot of the other social sciences.

So we now land this in your topic about the reptilian brain and how bad we are at measuring risk: of course I agree... in a bounded way. There is a continuum: the more knee-jerk our reaction and the more complex a context is relative to the tools we use to tackle it, then the less rational our response will probably be (and also, the more outright good ol' mistakes we will make, even if we got the algorithm right); but when we can sit down and think about it and if our tools are better relative to the complexity, the more we will behave like a rational agent. All evidence I'm aware of shows we tend to behave more like one (emphasis on "tend" and "like"), on average, than like a... whatever the alternative is. We are not fighting or fleeing all the time.

Wrapping my argument so far another way and since you cite Ariely's book: the "Free!" experiment his paper gives fantastic insight into an aspect of the rational agent model that needs to be modified some way (for some contexts being analyzed, but perhaps not for others). But what is not included there is the next iteration: we learn. When we play something in a suboptimal way AND it is pointed to us, we tend not to make (exactly) the same mistake again. Once I make the "Free!" mistake once and it is pointed out at me, I am more likely to pause, think about it, and make the correct decision next time I'm faced with a situation I can recognize as similar. Same thing happens with repeated prisoner dilemma's experiments, ultimatum games, centipede games, etc.

So we might not be rational all the time, but we learn from mistakes (which happen when we deviate from rational behavior) and act more like a rational agent next time.

The agents I'm talking about in the posting are a smart crowd with one job only: making these calls. They have the best tools available, the best info, computers, etc. But most importantly: they have been trained to conceptualize risk, measure it, and incorporate it in their decision. The concept we use of risk might be wrong, but it's the one all of us who've studied the topic share, so we cannot be arriving at such opposite conclusions just because of some limitation inherent to them (and I say "them" because of the evidence given by one bubble burst after another and some pretty out-there market reactions and comments). So I instead suggest that it's because the incentives they face lead them to take the right amount of risk for themselves... which is too much risk for their clients.

Your final point is where I totally disagree: when you say that, perhaps, the system works for the most part. If my assessment is right, the way the incentives work for these agents (who make decisions for or inform the rest of the market and so play a pivotal role) is adding unnecessary instability to the system by making the amplitude of the business cycles larger than it would otherwise be. Now, one could argue that this would not be a problem if: (a) the long term growth trend would be the same (only with larger cycles around it) and (b) this does not create even more perverse incentives when the troughs of the cycle hit us.

Neither (a) nor (b) hold. A more unstable system will have lower long term growth because, at the very list, the credit markets become more expensive. And worse, if the gains and losses of the "gamblers" would be limited to them, whatever; but while they get to keep the gains, their losses, when large enough, can create such havoc that they affect the rest of us and interact with politics to introduce policies that, in effect, redistribute wealth from those who did not gamble to those who did so irresponsibly.

The result are perverse incentives for all to be "irresponsible gamblers" next time around.

An example: Family A saw through those bubble-fueling mortgages which could only work if house prices went up forever and didn't take one. The "need" for such a loan arises out of a vicious spiral from bubble-fueling speculation which has pushed prices up so much that Family A cannot buy a house of the quality they could have afforded if the market had been behaving according to fundamentals. So even during the inflation of the speculative bubble, Family A must accept a smaller house than otherwise or, worse, must pack and go live somewhere else and commute two or three hours to work every day. Now the bubble bursts. Everyone is hurt by it, both Family A, who has already lost, and those who created the bubble. But now foreclosures are threatening the financial system for all and the value of homes in neighborhoods were too many are happening, perhaps the very same neighborhood where Family A moved into. So the Fed, the White House, and Congress come rushing in with some pretty crazy stuff to try to "rescue" everyone... But government spending and inflation are, in the end, paid by everyone through higher taxes and future slowdowns/recessions.

Ah! This has been another posting in disguise!