Monday, February 11, 2013

Reserve Bank burst bubble of certainty about future

There's never any shortage of people convinced they could do a much better job of managing the macro-economy than the outfit that does manage it, the Reserve Bank. And sometimes I suspect there's a geographic dimension to their criticism.

Economists and others who live in Canberra seem terribly confident they know better than the Reserve - much more confident than those living in Sydney, the same town as the Reserve. Indeed, the self-proclaimed superior understanding of the Canberrans is exceeded only by that of economists from Melbourne.

The Reserve is, of course, far from omniscient. Its forecasts are often astray. And these days, forecasting is more important than ever. In the old days, governments waited until they had hard statistical evidence inflation was getting out of hand before they took corrective action by raising interest rates.

Which meant they were almost always acting too late - sometimes so late they ended up making matters worse rather than better. That's because changes in rates have their effect on demand and then prices only after a "long and variable lag".

Since the Reserve attained its independence from the elected government, it has sought to correct for monetary policy's long "response lag" by conducting policy on a forward-looking basis, or "pre-emptively".

That is, policy decisions are based on forecasts for growth and, more particularly, inflation over the coming 18 months to two years. The arrival of actual figures is used just to adjust the forecasts.

And, as I say, the Reserve's forecasts are often astray. But this just reflects the limitations of the economics profession's art. The question is whether any of the Reserve's many second-guessers are any better at forecasting than it is. I remain to be convinced any are.

Although the Reserve's present course of action is always being criticised by someone - and not only the business lobby groups that make their living by always arguing rates should lower - I see little reason to believe they could do any better.

Indeed, they could easily do a lot worse. The Reserve makes a lot of small errors, but it's yet to make any really serious ones - the reason its critics have failed to gain much credibility.

One reason the Reserve never gets too far off beam is that it revises its forecasts every quarter and generally moves in tiny steps of 25 basis points (0.25 percentage points). And it's never too proud to change direction if it becomes obvious it should.

The other reason the Reserve has yet to get things badly wrong is that no one understands better than it how fallible its forecasts are - all forecasts, for that matter. And it's never afraid to admit its fallibility to the world.

Just as newspapers that regularly correct their errors are more trustworthy than those that rarely do, so those official forecasters who freely acknowledge their failings engender more confidence in their competence rather than less.

The Reserve revised its forecasts in the statement on monetary policy it issued on Friday. And for the first time it provided "confidence intervals" for its latest forecasts for growth and underlying inflation. These intervals were based on the range of the Reserve's actual forecast errors between 1993 and 2011.

It advised that a 70 per cent confidence interval for the forecast of underlying inflation over the year to the December quarter of 2014 extends from 1.6 per cent to 3.2 per cent. That is, if the Reserve makes similar-sized forecast errors to those made in the past, there is a 70 per cent probability that underlying inflation will lie between 1.6 per cent and 3.2 per cent.

Similarly, there's a 70 per cent probability that growth in real gross domestic product (GDP) over the year to the December quarter of 2014 will lie between 1.5 per cent and 4.4 per cent.

Hardly particularly informative? At least it avoids the illusion of certainty about what the future holds. But if your own fallibility makes you prefer a central, single-number forecast (a "point estimate"), you can use the fact that the confidence intervals are assumed to be symmetrical to work out what it is.

Add 1.6 to 3.2 and divide by two and the central forecast for underlying inflation is 2.4 per cent. Similarly, halving 1.5 plus 4.4 tells you the central forecast for growth is a fraction less that 3 per cent.

Happy now? If you're really keen you can apply a ruler to the confidence interval graphs in the statement and work out the Reserve's central forecast quarter by quarter - something it has never previously (sort of) made public. Whether it continues doing so has yet to be decided.

The width of the confidence interval (plus or minus 0.8 percentage points in the case of underlying inflation; plus or minus 1.5 points in the case of growth) indicates there is always substantial uncertainty about the economic outlook. (Though less about the more inertia-driven inflation than about growth.)

The Reserve says such high levels of uncertainty are also found in other countries and for both official and private forecasts. Similarly, it's typical (and hardly surprising) for the degree of uncertainty to increase the further into the future you're forecasting.

But if economic forecasts are so universally inaccurate, how come we hear so little about confidence intervals? It's partly because economists don't like advertising the considerable limitations of their art. They don't even like reminding themselves of their own fallibility.

But it's also because economists are selling their services and are very conscious of how much their customers value the illusion of certainty, which allows the customers to delude themselves they have more ability to control the future than they actually do.