Monday, December 14, 2009

Why the Federal Reserve will Not Stop Bubbles

Professor Jim Hamilton has an interesting post up about whether the Federal Reserve should try to put a stop to bubbles:

Before we can discuss this issue, we'd need to agree on what we mean by a "bubble". Here's one definition that a lot of people may have in mind: a bubble describes a condition where the price of a particular asset is higher than it should be based on fundamentals and will eventually come crashing back down.

If that's what you believe, then there's a potential profit opportunity from selling the asset short whenever you're sure there's a bubble. And if that's the case, my question for you would be, why don't you do put your money where your mouth is instead of telling the Fed to do it for you? Your answer might be that it could take years for the bubble to pop, and you're not willing to absorb the risk in the interim. Or maybe you don't have the capital to cover the necessary margin requirements while you're shorting the bubble on the way up.

Even so, posing the statement in this way should bring a dash of humility to those currently claiming to see a plethora of bubbles that the Fed supposedly needs to fight. What exactly persuades them that they are right and all the other players in the market are wrong? How much of their personal wealth are they staking on the strength of their convictions? And even if you're absolutely sure you know how to identify bubbles, raising interest rates as a response is, as Tim Duy observes, "a rather blunt weapon that kills indiscriminately".

Regardless of whether the Fed should pop bubbles, I claim that it never will. And the reason is that it doesn't make any sense in terms of the social dynamics of a bubble for it to do so.

Let me explain.

Bubbles happen because for certain dimensions of the price space, there is significant uncertainty about the underlying fundamentals and no fully reliable algorithm for resolving that uncertainty that everyone can agree on. Once you have that situation, the door is open to humans, who are profoundly emotional and social creatures, and constantly look to our networks for clues about what's going on, to be subject to waves of emotion (greed/fear) that color prices.

Let me contrast two things: transitivity of exchange rates, and the price of Internet stocks in the late 1990s. Any fool can see that if you change yen into dollars and then into euros and then from there back into yen, very quickly, you ought to get about the same number of yen you started with, modulo a little transaction cost. If it were not so - if you ended up with either more yen or less yen than you started with, then there's a profit opportunity that can be exploited very rapidly. That places an algebraic constraint on the dollar-euro, euro-yen, and yen-dollar exchange rates. So I would expect that, regardless of how much exchange rates fluctuate, the overall vector of currency prices will be tightly constrained to the subspace in which this relationship is satisfied for all currency triplets. No bubble is ever going to occur in which people lose sight of this.

However, if you were deciding whether to buy Internet stocks in, say, 1997-1998, you faced a genuine conundrum. The Internet was clearly a phenomenon of enormous importance that would influence the future greatly, and clearly there were a lot of profit opportunities. On the other hand, others had been bidding the prices of those stocks up already, were they now too high? Clearly, no algorithm can settle this - it's a judgement call. Reasonable people could, and did, differ. And if you differed, there was no quick or certain way to profit, in the way there is with the currency rate example.

We have a lot of historical experience that tells us this kind of situation sets human decision-making heuristics up to fail. Early on, as the genuine possibilities of the future seep in, excitement builds amongst the bulls. This creates a self-reinforcing paradigm - I use the word deliberately in the Kuhnian sense. The more people come to believe in a very rosy assessment of future prospects, and the more they are willing to risk based on that view, the higher prices get bid up, further reinforcing the bulls in their views. The whole thing continues until it becomes impossible to find new recruits to the increasingly manic bull view and the crash sets in.

Now, let's think about the role of the economics profession, and economic science, in a situation like this. The first thing to note is that economics is not isolated from its subject matter in the way we normally expect of sciences. A chemist studying reactions in the test-tube doesn't worry about the possibility of the chemicals behaving differently because he is watching. Even an anthropologist living with a tribe in the jungle, although she certainly interacts with the tribe, doesn't have to worry about them reading her academic papers and deciding to behave differently in consequence.

But economists should worry about this! Financial market firms are full of smart money-driven people who can certainly read the economics literature and have and will absorb any useful ideas from it. To the extent an economic theory has real-world predictive power, we can be confident that financial market participants will pick it up and use it in an attempt to profit. And this of course will change the behavior of the system. In particular, it will tend to change the system in such a way as to make the theory less true.

Consider if you had a perfect way to predict recessions. Take three data series X, Y, Z, apply your brand-new super-duper algorithm and a liberal pinch of salt, and out pops an infallible prediction of whether or not there will be a recession in six months. You publish, other economists admire and begin to work out minor variants and improvements, the thing becomes a consensus. The recession-prediction problem is solved! You win the Nobel prize.

Will all this escape the attention of Goldman Sachs and Bank of America? No way! And what will they do when the algorithm says there's a recession coming in six months? Well, there's a thousand things a smart executive would think of. Short stocks! Short commodities! Interest rate swaps! And maybe we should start doing some quiet layoffs in preparation.

And what about our smart executive's country cousin running Applied Widgets down in Arkansas? Even he reads the Wall St Journal and is plugged in enough that the upcoming recession prediction will not escape his attention. Clearly the volume of widget production should be ramped back in anticipation. Better start laying off staff. Probably should cut the marketing budget, and postpone the launch of the revolutionary new SuperWidget-3.0 until the situation becomes a little clearer. Take a look at the travel budget.

In short, everyone, in anticipation of the recession, will take actions that are contractionary in nature. In the limit, once the recession prediction algorithm is fully absorbed by all economic decision-makers, the effect is to cause a recession immediately, as soon as the algorithm predicts one. But this means it has no predictive power any more! It only tells you there's a recession once you are already in one.

I can't think of any other science quite like this, where the very truth of an insight makes it self-falsifying.

Does this mean no economist can ever predict a recession? No! There could be, and indeed I think are, economists and procedures that predict recessions quite well. But the point is that any such persons and procedures must occupy a slightly maverick, liminal status outside the mainstream consensus of the profession. They will be Cassandra figures - doomed to successfully predict the future, but also to have their predictions ignored. If they became too generally accepted, their success would inevitably go away.

In a very similar way, it's clearly impossible for the economics profession as a whole to successfully diagnose a bubble. If economists were in some reasonable agreement that "prices are overheated and there will be a crash next year", then of course it's very likely that market participants would be trading based on the same theory the economists were using, and the price fall would be here already. The bubble can only last as long as most people believe it will continue, and economists are centrally included in that.

And of course the Federal Reserve is the commanding heights of the economic profession. It is staffed by professional economists, overseen by experienced economic thinkers and financial market participants, and is a pinnacle of respect and power for them. There is naturally an expectation that very good people with excellent judgement will be hired there. But this is "good" and "excellent" as judged by the mainstream of economic thought. How likely is it that the Federal Reserve will give major decision-making authority to marginal figures with off-kilter views? Not very likely - that's not how human institutions work. But if it won't do that, then it's inevitably subject to the curse of self-falsification outlined above.

And so we get what we have - a Federal Reserve that utterly failed to see the housing bubble in advance, staffed by prestigious representatives of an economics profession that also, by-and-large, denied the existence of the bubble as it was developing. Including, I might add, the otherwise excellent Professor Hamilton.

It couldn't have been any other way.


Burk Braun said...

Excellent post, and thank you for your very rich posts. But the Fed could detect and address over-leverage, if it doesn't allow rogue entities to escape its reserve banking system. The crisis was not only an asset bubble, but an economy-wide overleverage situation brought about, not by regular banking, but exotic instruments and institutions.

I thought that the whole point of the reserve banking system was to allow money creation and lending on a relatively free basis, but with ultimate monetary control by the Fed. I'd argue that control was lost in the recent period.

Stuart Staniford said...

Hi Burk, and welcome. I agree with you the Fed, could, in theory, prevent overleverage. But if you look at what actually at happened, is they were instrumental in dismantling traditional protections against too much leverage (eg effectively dismantling the reserve requirements for banks, and doing nothing to prevent all kinds of off balance sheet entities). I argue this is what one would expect - they were at the center of the social consensus amongst economists that overleverage was no longer a problem.