IMF shows poor track record at forecasting recessions

Economic predictions presented as precise numbers are far from that in reality

Valentina Romei and Keith Fray


Blowing hot and cold: The difficulty in getting forecasts right is not unique to the IMF


The IMF is set to release its twice yearly global outlook on Monday, providing economic growth forecasts for almost every country in the world.

However, a closer look at its record on reading the future suggests it is unlikely they will accurately predict even how many economies will expand or contract this year.

The FT looked at the number of countries that the IMF expected to be in recession for every year since 1991 and compared it with the number of economies that turned out to have actually contracted.

Over the last 27 years, the IMF has predicted every October that an average of five economies will contract the following year. In practice, an average of 26 have contracted. This suggests that the six countries that the IMF predict will be in recession for 2018 could rise to as many as 31.

The difficulty in getting forecasts right is not unique to the IMF. “All macroeconomic forecasters are poor at predicting downturns,” David Turner, head of the economics department at the OECD told the FT.

Countries in recession

“Recessions are not rare,” echoed Prakash Loungani, a macro-economist at the IMF. “What is rare is a recession that is forecast in advance.” Despite an increased amount of economic data being available, “the ability to predict downturns remains dismal”, he told the FT.

“Extreme volatility during the global financial crisis complicated economic forecasting” said the OECD. In October 2008, after the collapse of Lehman Brothers, the IMF forecast that seven countries would be in recession in 2009 and it predicted global output would expand by 3 per cent. In reality, world GDP contracted by 0.1 per cent and 91 countries went into recession.

In that October, the IMF predicted an economic expansion for 2009 for both the US and Japan. Instead, the world’s largest economy contracted by nearly three percentage points and Japan’s economy shrank by 5.4 per cent, its worst annual performance since the second world war.Also, the IMF sometimes gets the countries wrong.

Change in World GDP

Greece, for example, is one of the eight countries that the IMF expected to be in recession in 2016, but it is not among the 25 countries that the IMF now considers to have been in recession in that year.

The fact that forecasts are “typically over-optimistic for horizons beyond the current year” is not necessarily the result of economist optimism. They “fail to forecast strong booms, just as they fail to predict recessions,” said Mr Loungani, suggesting that economic forecasts “are too rooted in thinking that things stay close to normal or will revert to normal soon”. Economic models are “not precise and they are based on lots of assumptions that may not turn out to be true, ” said Paul Donovan, the chief economist at UBS Wealth Management.



The problem is not only with models and assumptions, but also with hard data. Economic data are continuously revised, sometimes in a significant way, even in countries with usually reliable data.

In the last 24 years of UK quarterly GDP data, almost every final figure has been different from the initial release, sometimes by one full percentage point.

The US economy expanded by an annualised rate of 3.2 per cent or contracted by 0.7 per cent in the first quarter of 2015 according to different revisions of the same data. Similarly, in Q1 2011 the US economy was revised from registering a strong expansion to a strong contraction.




The IMF’s April outlook is often more accurate. This is because it is easier to get a forecast right for the current year than the following year. The April report is better able to signal a recession for the current year than the October publication, “but one that is much milder than what transpires”, says Mr Loungani, author of several studies.

Economic forecasts should provide “broad trends, not specific numbers” said Mr Donovan, they should assess whether an economy is growing about trend, below or above trend. “Specific forecasts create an illusion of precision,” he added. Having decimal points in forecasts is “purely to prove that economists have a sense of humour” said Mr Donovan.

0 comentarios:

Publicar un comentario