Wednesday, 11 March 2009

Multi-level rationality

Robert Peston is a bit torn today as to whether the markets are being rational or irrational - are they taking into account a future potential fall in gilt prices in this week's trading, or are they focusing only on a short-term rise in demand and ignoring the future?

My take is that they are following a plausible model of bounded rationality: being rational within the reasonable scope of the foreseeable consequences of their actions, but not beyond.

Traders can see a certain distance into the future with reasonable certainty.

They can see the immediate consequences of the Bank of England's policy: that a fair proportion of available bonds are going to be bought up, and thus the price should increase in the short term. There are basic reasons that people need to buy government bonds (e.g. pension funds selling annuities) and so regardless of long-term returns, a temporary shortage can push up prices.

They can also foresee the likely short-term effects for the economy of additional money supply: increased economic growth (or the recession moderated) without a short-term increase in interest rates, because we are essentially at the zero interest rate bound already. They aren't expecting much increase in inflation - the relative price of index-linked bonds compared to non-index-linked has not jumped.

But the transmission mechanism from current QE to long-term inflation or economic growth, and the reflexive effects of QE on medium-term gilt demand, the future costs of money-raising for the government, and the resulting levels of issuance of further gilts are just a bit too far away and too self-interacting to realistically calculate the effects. Therefore, due to simple constraints on the ability of traders to think through these effects, they will probably be discarded rather than factored in.

Stephanie Flanders has a more sanguine take on the whole thing - pointing out that if the Bank of England does make losses in future, the Treasury will have made compensating savings in the meantime.

In essence the government should hardly ever be in the business of worrying about the secondary costs to itself (such as the Bank of England's losses on bonds, or indeed its profits from overnight lending) of boosting economic growth. Additional growth makes such a positive impact on state finances that it outweighs nearly any second-order cost of achieving it.

Peston also asks whether this trade of bonds for money should be subject to Ricardian equivalence, so that it would leave the markets totally unaffected. Ricardian equivalence assumes perfect rationality and perfect markets - but as Nick Rowe has pointed out, if we had perfect rationality we wouldn't have recessions anyway. In reality, people are either irrational decision makers, or they have insufficient information about the future and thus can't make perfect decisions. Part of the reason they don't have information about the future is the assumption that other people are going to be irrational - which is essentially self-fulfilling.

So it can't be expected that swapping bonds for money will leave the markets unaffected. The immediate effect, as you'd expect, is to increase the price of bonds. We hope that the longer-term effect will be to improve the performance of the economy and therefore reduce the price of bonds again.

I would hypothesise that markets will actually approach the ideal of a correct reflection of weighted expectations, but that they require a certain amount of processing time to get there. During this time, analysts will opine on the likely path of the economy, traders will evaluate their analysis, and everyone will wait for evidence on whether the easing has been effective in boosting activity. Traders will gradually feel their way towards a consensus - with nobody wanting to take a big risk by moving immediately to their prediction of the final price - and gradually, prices will reach a figure that reflects the best possible estimate of the future course of the economy.

In the meantime we may see the phenomenon of 'herding' - traders interpreting the actions of other traders as if they reveal the underlying truth, and thus acting to confirm them. This in turn is short-term rational but results in an inaccurate price being placed on long-term returns. The herding takes time to take effect, and it may be too early for it to have happened yet. Or, herding may have occurred already and the slow long-term reversion to rational prices may only now be beginning.

Eventually, of course, rational pricing must return, because the bonds will be redeemed and the as the final payments take place they will be highly predictable. Indeed after redemption the rational price of zero is guaranteed to go into effect. In the meantime, we need to mix together short-term rationality, mid-term herding and slow, long-term rationality in order to come to an understanding of trader behaviour.

A final point: bonds are priced relative to money. Thus if bond prices are going up, it might also imply that the price of money is going down. Of course the returns on bonds, which justify their prices, are also in money, so this money price actually reflects the price of liquidity. If we are less hungry for liquidity, this could explain a drop in the money price relative to less-liquid bonds.

No comments: