This week, investors, journalists, and economists expect that “Thursday’s release of July retail sales data by the Commerce Department could merit a market reaction, particularly if U.S. sales rebounded last month from a weak showing in June. Economists are expecting a 0.5% bump thanks to increased motor vehicle sales.” [Fortune, Aug 9th 2015]. Every month, economists and other pundits try to “predict” how much American consumers bought at the till, and they are always wrong. This is not even forecasting, because the estimate, like most economic data, is 2-6 weeks old. Should be easy, right?
Well, here is the series, and we can see right away that it is probably pointless to guess at a given datapoint, because the signal is swamped by the noise.
So what should we be paying attention to, and trying to predict? Rather than intensely scrutinizing the random-number-generator that produces the wiggles of the blue line (monthly data) around the red line (the underlying trend), what matters most is the big changes in that trend, whether from high levels to low levels (as in 2002-3) or sudden shifts to a recession (2009) or the V-shaped recovery that followed, and is now threatening to peter out. A forecaster should look ahead, rather than be a man who “puts his head in the past and backs his ass into the future” to quote Ross Johnson of RJR Nabisco infamy.
In other words we should be looking at the series shown below, and the prediction that counts is if 2015’s total retail sales will total $5.2 trillion as in 2014, rise just 2% as the current trend suggests, or resume their earlier 4-5% rises after the effects of the polar vortex and commodity declines fade away.