Wednesday 22 April 2009

Rules Versus Discretion, Once More

(About the emotional decision making of the Global Economy)

One of the earliest posts of this blog was about policy rules and policy discretion. The original economics model is generally used to provide evidence for how short term optimisation of policy can screw up your long term rules, and thus lead to a sub-optimal outcome. It is an economy level procrastination problem, which - on the individual level - is also well described, by the behavioural economics lot.

A ‘rational expectations’ approach to the rules versus discretion problem, however, changes it a bit. In this form, the economic agents come up with new assumptions about policy rules when they see the policy maker’s short termist behaviour. And thus not only do the originally announced or implied policy rules go out of the window with the discretionary measures, but new rules also emerge. This is especially the case when the policy maker herself does not have any idea what is going on. Which is, incidentally, the situation now…

Given the above, it is striking how little is discussed about the long term consequences of all these semi-random and fully-random and even-more-random policy ‘innovations’ that governments around the world keep coming up with.

Like, what is going to be the long term consequence of a government backed mortgage payment holiday during crisis? How on Earth will anybody be able to come up with any robust probability function for mortgage defaults? Like: if the debtor gets into trouble, she defaults, unless, there are a lot of others, who are also in trouble, and all of these happen to be the constituencies of the ruling political party… Now try to squeeze that into an algorithm…

Or what is going to be the basis for assessing financial risk? Where will bank valuations come from? Will you have to categorise financial institutions into sizes, the big ones being saved for sure? Is it going to be a continuous distribution? Is there going to be a threshold for triggering government action? How many peaks will this distribution function have? E.g., you don’t save the first big one, then save all further big ones, until too many need saving. Or maybe just one or two more… I would shake the hands of the bank equity analyst who can deal with this one in any credible way.

Or countries. What will be the default probability assigned to any country? Now are these lower or higher than before? Lower, for the behaviour of the global governing institutions the past months suggests that no country is being allowed to run into real bankruptcy. But higher, for if these institutions run out of money, or more importantly political capital, then a lot of countries will default at the same time, pulling each other down.

The trouble is that although all of the above can be modelled, but not in any robust way.

Hence the importance of short term discretionary policies re-defining long term rules. Seeing how these novel ‘rules’ come to life is really disheartening. As if we all, as individuals would make our most important, long lasting decisions based on ad hoc emotional impulses.

Which, incidentally, we all do, of course… Life is a mess for everyone, even if you happen to be the Global Economy.

12 comments:

  1. Have a look at Olivier Blanchard's new paper on the crisis.

    ReplyDelete
  2. You keep insisting that economics died, a new theory is needed. This cannot be further from reality. Look at all the intensified research WITHIN mainstream economics. All that is happening is one more cyclical downturn. We will plug the new data into our models and that's it. You read far too much into it.

    ReplyDelete
  3. surely, there is a reason for "intensified research" (what is that anyway?). perhaps since the state of the discipline is truly dismal?

    more and more people realise that the old thinking is out. or at least the old thinking should be out. or something along those lines

    and in any case, tamas's post, at least this one, actually stays within mainstream macro. his point that the long term rule formation based on the current ad hoc actions is rather interesting. and not very much discussed.

    the markets are completely focused on the superficial interpretations of the temporary price movements of the ignorant market itself. leading light, when will you come?

    ReplyDelete
  4. Bee,

    Your point is 100% BS. It has been shown by a million studies that the Market is the most efficient mechanism to synthesise information. If there is volatility, all that shows is that there is volatility implied by the data itself. I does not have to do anything with the "ignorance" or "superficiality" of anyone.

    ReplyDelete
  5. Very interesting post, Tamas. Thanks

    ReplyDelete
  6. dragon_tamer: come on. the markets clearly have no idea. have you talked to anyone recently about where their forecasts are coming from? the stories change by the hour

    ReplyDelete
  7. Bee is right. The stories that analysts offer vary to a ridiculous extent. There is no market consensus whatsoever, and even individual analysts regularly reinvent their entire argument from scratch.

    There is a Brave New World out there, and nobody has any idea what it really looks like.

    ReplyDelete
  8. Bee,
    Where are YOUR arguments? The "shifting stories" story of yours, even if true, does not go against the efficient market assumption.

    ReplyDelete
  9. dragon_tamer:where do you think the valuations that the market aggregates so efficiently come from? they are the stories. we just call models differently: it used to sound more sexy.

    when the stories change radically, the information set that the market digests is not stable at all. even if the aggregation exercise turned out to be right, it would be a pure chance event.

    ReplyDelete
  10. It is not only the market that changes the stories. Look at the IMF updates. What kind of models can be that wrong?

    ReplyDelete
  11. The long term rules growing out of a series of discretionary measures is the point. We had already known that the probability distribution tails are much thicker than assumed under Gauss. What we learn these days is that it will be literally impossible to estimate the multidimensional probability structure. Not only that the different dimensions are cross-cutting (Engel, T-copula), but also that they probably have very odd shapes.

    Re the above debate: the 'price myth'. People in emerging market research have long known that capital market prices could stay away from the underlying 'fundamental' value for a very-very long time. Long term implied valuation, long term changes to the valuation, and short term price movements often have little to do with the real dynamics. You can blame liquidity, or institutions, or sentiment bubbles. However, often these are mere symptoms of a deeper problem of not having adequate understanding of what is going on. Not that some people have usable models, and just our lot had not figured it out, rather nobody has, or could have, a proper model. For the data is bad quality, and the structure changes too much, too fast. The global economy increasingly resembles an emerging market in this respect (see previous post ).

    ReplyDelete
  12. The recession is already in the last phase. The policy mix, tough as it was, worked. There is no point discussing discretionary measures any more. The nightmare of global government is gone.

    ReplyDelete