OECD Forecasts During & After the Financial Crisis: A Post-Mortem

I attended the titled event at Bloomberg's futuristic London office on 11 February 2014.  Pier Carlo Padoan, Chief Economist at OECD, gave us a keynote on one of the policy notes produced by OECD's economics department. The note was to dissect (to echo the name of the note) the forecasting errors over 2007-2012 with the hope to draw lessons and improve both the data and the methodology of future forecasting.

One of the main findings is that the GDP growth was overestimated on average across 2007-2012, in other word, none of the forecasting models from OECD, IMF or other major international organizations predicted the 2008/9 financial crisis and the 2010 Euro crisis. This is not new news.  The good news is OECD have beat themselves really hard by reviewing where went wrong and have come up with a list of developments so the audience started to feel a little better.  

There are two questions lingering in my mind: First, are we fooling ourselves by believing we can accurately forecast future crises if we had considered all relevant factors and used all the data in the world? Mr Padoan mentioned that their model failed to forecast the Oil Crisis in the 1970's and they have drawn lessons and improved their forecasting technique since. Unfortunately, 30 odd years later, their forecasts failed again. The available data and indicators have grown in both quality and frequency in the last few decades, and forecasters are now even talking about "big data" (suppose they know what it is). But are data really the solution? This lead to my next question, are we using the right model or is there a right model? One of the panelists, Prof. Paul de Grauwe, from LSE argued that the model used by mainstream economists is wrong. This was due to the failure to diagnose what caused this crisis. In his view, the prolonged crisis was caused by demand side problems whilst fiscal policy makers are looking for answer from the supply side.  I am not in a position to agree or disagree with this view but this could partly explain what went wrong in previous forecasts.  

I believe that forecasting is indeed difficult. Mr Padoan pointed out that major international organizations tend to use similar methods, which means they tend to be wrong at the same time.  This is probably why the creditability of these forecasts were called into questions, a bit like the ratings from credit rating agencies. This is also why I personally spend no more than 5 second on a piece of news with "GDP forecast" in its title. Having said all that, I am hopeful that by thinking independently and creatively, these orgnisations could provide invaluable market intelligence, which could assist investors, business owners, and governments in making their decisions.

My mind was tangled with these questions (too high IQ to cope) after the event until my husband asked me an excellent question: what does OECD do.  According their website:

The mission of the Organisation for Economic Co-operation and Development (OECD) is to promote policies that will improve the economic and social well-being of people around the world.

Can a wildly inaccurate GDP forecast contribute to this mission?  To me, what I really hope to get from them is not a GDP number. I hope they could communicate these numbers better by breaking them down and explain what these means. For example, a 2% GDP growth in country A is different to a 2% GDP growth in country B because country A has 2% more people working with 0 productivity growth whilst country B has 2% productivity growth without adding one man in the workforce.  Interpreting numbers is far more important if business and investment decisions are to be made based on them.

Perhaps OECD can educate users on the extend (probability) users should apply to the numbers' accuracy, equip users with data and tool to do their own forecasts.  In a perfect world where individuals have their own forecasts, you may get 50% of the people right and another half wrong. For example, one out of the two exporting business owners decided to increase production in China, which matched the demand exactly.  Whereas, if we have international authorities doing synchronized forecasts, with a poor historic error rate, and users use them without pinch of salt, you could potentially end up having 100% of the people wrong. This means two out of the two exporting business owners decided to increase production in China which exceed the demand, causing overcapacity, financial loss, and etc etc.

The event is nonetheless thought provoking and not a single second has been wasted both during and after the event.