The reply came back: “The commanding general is well aware that the forecasts are no good. However, he needs them for planning purposes.”Arrow may still be the youngest person ever to have won the Nobel Prize for Economics, having picked up the award in 1972, but clearly his general has proved the greater influence on many of today’s business leaders and policymakers.
Take, for example, forecasts relating to energy consumption and prices. The National Audit Office and the Department of Energy & Climate Change could do worse than consider following the lead of the Energy Information Administration (EIA) in the US, which has a policy of continually reviewing its previous forecasts.
To take one example, the organisation makes regular projections of the US economy’s ‘energy intensity’ – how much energy it will use per dollar of GDP. Back in 1995, the EIA forecast the US’s 2012 energy intensity would be 11 billion BTUs (British thermal units) while, by 2011, it had reduced its forecast to 9 billion BTUs. In the end, the actual figure turned out to be 7 billion BTUs.
A similar pattern emerges with the rest of the EIA’s forecasts – anything predicted a decade or two ahead has tended to be wrong by double-digit percentages. The table below shows the long-term accuracy of the EIA’s most important forecast that of the oil price, during the 20 years to 2005.
Two- and five-year projections of oil price 1985 to 2005
Source: Resources for the Future 2008
This retrospective analysis is something of a double-edged sword, of course, because while The Value Perspective believes this to be very good practice, it serves to highlight just how poor the EIA – in common with so many others – is at forecasting.
Long-term forecasting is even more fraught – anything predicted a decade or two ahead has tended to be wrong by double-digit percentages. An oil price in the $30s seems almost quaint but, as you can see in the table below, this was the forecast for 2012 as recently as 2005.
Forecast of 2012 oil projection
Source: Resources for the Future 2008
Nor do forecasts have to be so very long term in order to be so very wrong. To pick on Centrica, back in the UK, the energy supplier makes predictions as to what power prices will be in two years’ time and these include an estimate as well as a high and a low range.
It turns out Centrica’s estimates for both 2010 and 2012 were outside that range about one-third of the time.
So it seems even if you are trying to forecast only one thing just a few years ahead of time – and that one thing is your specialist subject, no less – you might as well, as Kenneth Arrow suggested to his superior officers, be picking numbers at random.
The good news is that the best forecasters in the world regularly review their predictions – the bad news is this generally flags up how badly they have done.
So should you simply not bother to try? Well, bearing in mind our consistently low opinion of experts’ ability to predict the future with any accuracy, it may come as a surprise to learn that, whenever we buy a business, we do include forecasts in our investment thesis.
This is not with a view to broking the company to anyone else, however, but rather to lay down a marker in time.
Each forecast sets out what we think about the future and the model we have built on the business’s financial statements and so acts as a reference point to which we can return should our investment not work out quite as well as we believed it would.
Such forecasts serve the purpose of helping us to see where we may have gone wrong and thereby, over time, enabling us to improve as investors.
Ian Kelly (pictured on page 1) is co-manager of the $2.3bn Schroder ISF Global Dividend Maximiser fund and writes on The Value Perspective. The views expressed above are his own.
