RPT-Watchdogs struggle to explain "optimism bias" in rosy GDP forecasts

(Repeats story, text unchanged)

* Of 30 IMF and World Bank GDP forecasts, 28 too optimistic

* 10 out of 10 OECD forecasts too optimistic

* G8 economic growth since 2008:

By Jamie McGeever

LONDON, Nov 4 (Reuters) - Chances are, the world economy has been in worse shape than you thought for years. Especially if you were reading the headline forecasts from the IMF, World Bank and OECD.

Having repeatedly overestimated the world economic rebound since the 2007-08 crisis, they say they are gradually gaining a better understanding of why their forecasts were so wrong, and are working to fix their models.

Forecasting gross domestic product precisely can be a lottery for even the best of economists.

But the International Monetary Fund, World Bank and Organization for Economic Cooperation and Development have not just been wrong; for years they have all been wrong in the same direction, persistently forced to revise down predictions that proved too rosy.

“There’s an inbuilt ‘optimism bias’,” said Stephen King, senior advisor to HSBC and a former economic adviser at the British treasury. “But facts have to dominate a forecast eventually.”

Twenty-eight out of the 30 initial calendar year forecasts by the IMF and World Bank for global, developed market and emerging market GDP growth for 2011-2015 proved too optimistic, in many cases wildly so.

It’s a similar picture at the OECD, which mostly tracks the world’s industrialised nations. Of its 10 initial growth forecasts for the U.S. and global economies over the same period, all have been too optimistic. In some cases, growth turned out to be half the pace originally estimated.

There’s no single answer for why, but economists say explanations include an under-appreciation of the damage done by the 2007-2008 financial crisis both to consumer demand and to banks’ willingness and ability to lend; the poorly-understood effects of increased global debt levels and a failure to predict surprisingly weak productivity.

Olivier Blanchard, chief economist at the IMF from 2008 to October of this year, recognised that forecasts have been consistently revised down over the past five years but denied there was any structural bias to them. Forecasters just made the same mistakes.

“We and others misinterpreted permanent adverse shocks for temporary ones,” Blanchard said.


The IMF, World Bank and OECD told Reuters that post-crisis forecasting has become particularly complex and that they have each carried out research into why this is.

Researchers concluded that models were underestimating the damage wrought by the 2008 crash and overestimating the durability of the subsequent recovery.

The IMF points to the “poorly understood” slowdown in productivity growth across developed economies, and a “protracted slowdown” in emerging economies, among other reasons for the persistent overestimates.

All say their models are constantly updated, which means that just because they have repeatedly overstated growth in the past, investors should not expect them to continue to do so forever.

“For the post-crisis period, say from 2012 onwards, events are still unfolding, but we are continuing to make adjustments and reassess all the time the models we use,” said Sebastian Barnes, senior economist at the OECD in Paris.

Economists have known for some time that recessions following banking collapses are longer and deeper than other recessions, and recoveries are slower and shallower.

But they still failed to predict the lasting impact of the 2007-2008 crisis on demand for goods and credit, on businesses’ ability and desire to borrow to invest, and on banks’ willingness to lend.

Despite trillions of dollars of central bank stimulus, interest rates slashed to zero and an early and heavy bout of classic government deficit spending immediately after the crash, demand, lending and investment have still to fully recover.

The policy response to the crisis continues to be unprecedented. Eight years after the first tremors were felt, major central banks still have interest rates at or near zero and are buying bonds. Global debt is far higher now, by some $57 trillion, according to McKinsey & Co, than it was in 2007.

The effect of that debt build up on growth has yet to be fully understood. These are uncharted waters. But if so, “forecasts should be all over the place, not persistently one-sided,” said Willem Buiter, chief global economist at Citi and a former Bank of England policymaker.

Global institutions like the IMF and World Bank rely on data from national authorities. If those figures are prone to upward bias - few observers, for example, believe China’s official growth figures - then the overall picture will be skewed.

Emerging markets also have less reliable historical data on which to base forecasts than more developed economies. Given that the world economy is largely driven by emerging market growth, this makes the global picture more susceptible to wayward forecasts. While weak data can theoretically produce errors in either direction, most recently it resulted in a failure to spot an emerging market slowdown.

The IMF and others also failed to spot the degree to which companies have become more reluctant than in the past to invest, not only as a result of the 2008 financial crisis but also from the dotcom bubble that burst eight years earlier.

Companies have opted for short-term measures to boost returns, rather than long-term investment. Stock buybacks from U.S. S&P 500 firms since 2010 have reached $2.6 trillion, according to Barclays, returning money to shareholders that in past recoveries might have been spent building factories and hiring workers.

Some economists blame professional habits of their field for persistently faulty models that overestimated growth.

Buiter at Citi notes that economists are prone to “confirmation bias”, a tendency to emphasise evidence that supports an earlier view and play down evidence that contradicts it, which makes it harder to see when models are wrong.

Another tendency among economists is to expect “mean reversion” - that statistics will move back towards their historical averages and previous cycles will be repeated.

“Everybody assumes that there’s mean reversion, which is what happened pre-2008 happens now,” said David Blanchflower, professor at Dartmouth College in New Hampshire and another former policymaker at the Bank of England.

“They don’t get that it’s a different world now.” (Reporting by Jamie McGeever; Additional reporting by Marc Jones; Editing by Peter Graff)