Behind the forecasts

Like Stephen Pollard, I don’t like the new year: I see no reason to celebrate the anniversary of the death of the greatest man who ever lived.

One thing I especially hate about this time of year is that it focuses attention on economic forecasts.

Sensible economists make forecasts for one reason only – people are stupid enough to pay them to do so.

Economic forecasts in themselves are boring, and bring only discredit to the economics profession. What are interesting are the ideas underneath the forecasts. Here are six of them:

1. The Meese-Rogoff puzzle. Conventional textbook models of exchange rates are consistently hopeless at predicting actual exchange rate moves, as Menzie Chinn has pointed out (pdf). Why is this? And does it mean economists have nothing useful to say about exchange rates? Not necessarily, because of….

2. Uncovered equity return parity (pdf). This is one of the most interesting recent theories about exchange rates. It’s a variation of uncovered interest rate parity (another textbook theory that doesn’t work). The idea is that expected equity returns must be equal across countries. So, if one country offers low expected equity returns, its exchange rate should be expected to rise.

I reckon this can help explain the rise in the dollar this year; dividend yields on euro zone and Japanese stocks have fallen –implying expected equity returns there have fallen – so their exchange rates have had to fall to levels from which a rise is considered more likely than a year ago.

So, have we finally found a useful theory of exchange rates? How consistent is this with rational asset pricing?

3. Relative productivity and the dollar. Since the late 90s, productivity has risen in the US relative to the euro zone. This should have strengthened the dollar, due to the Balassa-Samuelson effect. To what extent does this offset the negative impact on the dollar of the current account deficit?

4. Short-term interest rates should be unpredictable. If a central bank were pursuing rigorous inflation targeting, official rates should follow a random walk. This is because rates are set according to where inflation is forecast to be. An optimal forecast embodies all available information and so should only change because of surprises. It follows than interest rate changes should be random.

But they don’t seem to be. Is this a good thing? Do the welfare gains from interest rate smoothing outweigh the losses from deviating from optimal targeting? Do they do so when interest rates are at levels low enough to threaten to raise inflation, when we know that inflation is welfare-destroying?

5. The predictability of share prices. In my day job, I show that annual returns on the All-share index are partly predictable by: the dividend yield; ratios of money stocks to market capitalization in the UK and world; consumption-wealth ratios; foreign buying of US equities; retail buying of unit trusts; and mortgage approvals. Except for consumption-wealth ratios, these are all measures of risk aversion – and high risk aversion should lead to high returns.

The trouble is, these point in different directions. Money-capitalization ratios point to decent rises, but foreign buying of US stocks and consumption-wealth ratios point to falls.

6. UK house prices. The recent rise in mortgage approvals points to pricing rising. But there’s also a strong tendency for price inflation to feed on itself, and for high price-income ratios to lead to falling prices.

The lesson of 5 and 6, I guess, is that in-sample predictability is not sufficient to generate useful forecasts. Which raises the question: on what are forecasts based?

My hunch is: bluff and self-deception.

Comments

[Angry Economist]

Who is the bigger fool!? I would say that the bigger fool is the person who places forecasts as the most influential factor in their decision making. The lesser fool is the person who provides the forecasts and pretends that they can either accurately predict the future or can predict it more accurately than anyone else.

I think that forecasts generally are alluring because they offer some kind of pophetic nonsense for people to follow.

In my experience we don't yet fully understand why things happened the way they did, but some folks are quite happy to predict them anyway. Its fools gold if you ask me.

A wee bit on using forecasts: http://accidentaleconomist.blogspot.com/

[Jeff Ross]

"The recent rise in mortgage approvals..."

Are the figures for mortgage approvals, published monthly, just the new approvals issued that month, or a cummulative total of all outstanding approvals that have yet to be taken up? If the latter, a rise in mortgage approvals could be due to lengthening chains. How long does it take on average for mortgage approvals to be taken up? What percentage of approvals lapse? What percentage are for remortgages and MEW? How have these figures changed?