Ripple Me This

All models are wrong, but some are useful – George E. P. Box

Albert Einstein might be the most famous scientist yet. One part of his work created models to help understand the nature of reality. Using mathematics, he predicted something called gravitational waves or ripples in space itself. Think of the wave you can create with a sheet while making your bed. Einstein postulated that when two significantly massive bodies spin around each other they would create ripples that would move through space at the speed of light. Just over two years ago, physicists in America claimed for the first time to have detected such a ripple. Since then (including the most recent on August 17th of this year which was sensed by two apparatuses in separate countries) three other undulations have been discovered, further confirming the accuracy of Einstein’s model. Here’s an interesting article on it.

Of course, Albert Einstein is not the only intelligent person to use math as a way to predict or explain reality. George E. P. Box was a statistician from Britain known for various breakthrough discoveries in time-series analysis. A time-series consists of a collection data points listed in sequential order. Temperatures over the last week, several years of quarterly gross domestic product figures, or previous closing values of a stock market index are examples of times-series. His methods are implemented by forecasters trying to predict future outcomes in all sorts of disciplines, including natural and social sciences. While considered a legend in the field of statistics, he remained cognizant of the limits of models; see his quote above.

Professionals in the private, government, and educational sectors are constantly tweaking their models of markets and economies. Yet even after all of the help from George E. P. Box and the number crunching capability of modern technology, predictive accuracy (and reliability) of a time-series model of America’s gross domestic product still eludes mankind. Atlas submits the Federal Reserve’s inaccurate predictions for both output growth and inflation during the current recovery as evidence. Before you think we’re being too hard on the hundreds of Ph.D. economists staffed by the central bank, Atlas acknowledges the inherent flaw in expecting better results.

These folks are dealing with subjective variables (people) and not the steady-state nature of science. George E. P. Box was right; consistently modeling the behavior of individuals making multiple economic choices each day is impossible, so the models will always be wrong. Atlas just hopes there are models that are actually useful and that the central bank knows about them. Of course, we could always let them use ours.