The unbearable lightness of forecasting with bad data

It takes no genius in order to paint the broad strokes of sound economic policy advice. That such advice may involve the M-Word is also not a surprise. This was the case with the annual report of the German Council of Economic Experts (whose members are respected economists) published earlier this month:

“Economic policy should place greater confidence in market processes instead of increasingly seeking to fix market outcomes in order to achieve distributional objectives”

states the Council’s press release in its very first paragraph. That market processes are part of the equation in getting things right is as self-evident as saying that training and competition among teammates will make a better soccer team. That you still need a coach of course to regulate, discipline and stir as Peter Bofinger stated in dissent is also self evident. It really is not “rocket science”.

The problem, as is often the case with self-evident truths, is that when it comes down to sizing the portion of each ingredient in an economic policy, polarization ensues and leads most people to come down on one side of the argument or the other. The quality of the debate suffers as a result. One of the reasons this happens, besides the fact that some people first choose their truth and then seek the “evidence” to support it, is that both politicians and economists are missing the right data to work with. The exact portions of “market” and “state” are a result of one’s own bias but adjusting for that it is a result of current dynamics in the economy. You need to measure the right variables, timely and reliably, and act accordingly, swiftly, preventing the economy from departing for detrimental equilibria.  If you are asking whose fault it is that the data we need is not available then I will say:

politicians don’t listen when scientists say they need more data and economists don’t complain enough when they are called to supply economic advice based on bad data

so yes it is a bizarre sharing of guilt and by all accounts a bad equilibrium. Knowing that “garbage in” will cause “garbage out” it is unfortunate to see that despite the lack of reliable data there is a lightness with which hard lines are drawn in the sand on both sides of arguments. Extensive costs for society are the result.

In their annual report the Economic Experts reiterated a revision of their GDP growth forecast for 2014 from an optimistic +1.6% (made in November 2013) to a less so +1.2%. This revision was first made known a couple of months ago and was consumed by the country’s economic press less as a prediction error of -25% on the part of the Council and more as a sign that the economy is underperforming at the rate of -25%. Reporting is gloomy ever since, talking the economy down as pundits of varying persuasion use this “economic under-performance” to promote their own agenda. But how sound are these numbers and how much faith should we put in them? I went ahead and took the Council’s reports from the last years and plotted them here:

Beliefs of the Council of Economic ExpertsEvery year there are revisions of the past numbers (years and the predictions for past, present and future are color coded) and new forecasts and revisions may go several years back (which goes to show that we have bad unreliable data). Let us look at 2013. In November 2012 the Council believed that the growth rate in 2013 will be +.8%, that was revised down to +.4% in November of 2013 and again down to +.1% 10 days ago a total prediction change of -87.5%! To make things worse the revision of the 2012 growth rate went down from .8% to .4% i.e. a prediction change of -50%. So the total forecasting error committed in November 2012 for 2013 is much worse because the revised .1% of 2013 is applied to the also revised .4% of 2012 and not the original .8%. The difference of .8% (i.e about 1%) and .1% (i.e. about 0%) is the difference between growth and near contraction. That such errors occur is by no means a surprise and they are so severe implies that we should at least take them with a grain of salt. The problem becomes acute when public opinion is bombarded with hysterical reports when an erroneous 1.6% drops to a (probably still erroneous) 1.2% (I will revisit this next year to see where it will be, stay tuned).

Without better data I think these numbers are not worth the ink to print them with so we might as well stop computing them. Flying in hallucination is at least as dangerous as being in a blind flight.