Monday, November 28, 2011

Econocrats get smarter on dodgy forecasts

You've heard the joke that economic forecasters are there to make weather forecasters look good. What you haven't heard is that the nation's top economic forecaster, Glenn Stevens, the governor of the Reserve Bank, thinks the joke "has something going for it".

There's an even older joke: everybody complains about the weather, but no one does anything about it. Actually, nothing we could do would change the weather. But, as Stevens remarked in a speech to the Australian Business Economists last week, some decisions based on economic forecasts can alter what happens (thus making forecasting the economy even harder than forecasting the weather).

That's true of the decisions made by central banks and governments, but it's also true of decisions made by businesses and households - even when their "forecast" is no more sophisticated than a bad feeling or a good feeling about how the economy's travelling and what lies ahead.

Actually, you can't not make a forecast. Even if you refuse to think about the future, you're implicitly making a forecast that things will stay as they are.

The Reserve Bank has no choice but to make the best forecasts it can because it can take two or three years for a change in the official interest rate to have its full effect on the economy. So the Reserve has to act before things get off the rails. Were it to wait until problems actually happened, it would always be acting too late.

But something I've always admired about the Reserve, and Stevens in particular, is their humility and realism on the subject of forecasting.

"It is only natural to desire certainty," he says. "Everyone wants to know what will happen. We all want to believe that someone, somewhere, does know and can tell us what to expect. But the truth is that the best we can do when talking about the future is to speak about likelihoods and possible alternative outcomes."

Like almost everyone else, the Reserve has expressed its forecasts as a "point estimate" - one number. But this gives forecasts an air of precision they don't possess. They're actually a "central forecast" within a range of possibilities.

People (and journalists) who don't understand this can attach too much significance to small changes in forecasts, or to small differences between the Reserve's forecasts and Treasury's. (Tip: they're never going to be very different because they're produced in the same factory, the Joint Economic Forecasting Group.)

Stevens says that "when consideration is given to the real margin for error around central forecasts, such differences are often, for practical purposes, insignificant". "When comparing forecasts, if we are not talking about differences of at least 0.5 percentage points, the argument is not worth having."

Consider this. In the case of a year-ended forecast for the growth of real gross domestic product four quarters ahead, the record over the past couple of decades says the probability of a point forecast being accurate to within 0.5 percentage points is about 20 per cent.

Experience since the start of inflation targeting in 1993 says the probability of underlying inflation over the next year or the next two years being within 0.5 percentage points of the central forecast is about 67 per cent. That is, if the forecast was 2.5 per cent, the chances of the outcome being between 2 and 3 per cent would be about 67 per cent.

"So any point forecast will very likely not be right," Stevens says.

According to Stevens, it would be vastly preferable for discussions of forecasts to be couched in more "probabilistic" language than tends to be the case, and for there to be more explicit recognition that the particular numbers quoted are conditional on various assumptions. To this end, the revised forecasts the Reserve published earlier this month, particularly those for the year to December 2013 (that is, more than two years away), were expressed as a 0.5 percentage point or even 1 percentage point range. Now you know why.

And, Stevens adds, the forecasts include "more extensive discussion these days of the ways in which things could turn out differently from the central forecast". "This goes at least some way to recognising the inherent uncertainties in the forecasting process, and is also important in relating the forecast to the policy decision."

But if forecasts are so dodgy, why bother? Why not merely assume things don't change, since that at least would be quicker and cheaper? Stevens insists economists can shed useful light on the future.

"We know something about average rates of growth through time," he says. That is, forecasts that the economy will return to its trend rate of growth are likely to be closer to the truth than forecasts that it will stay at the rate it is now.

Stevens says economists also know something about the long-run forces that produce economic growth: productivity and population growth. "We know that there have been, and will be again, periods of recession and recovery, though our ability to forecast the timing of those episodes is limited," he says.

"We know from experience some things about the nature of inflation, including its characteristic persistence, and the things that can push it up or down." But above all, Stevens says, we know some of the "big forces" working on the global and local economies at any time. The two big (and conflicting) forces at present are the resources boom and the troubles of the euro.

A lot of the work of forecasting boils down to weighing up the net effect of the conflicting big forces at the time. We'd be better off debating and understanding the effects of those forces than arguing about point estimates.