The Trouble With Economic Statistics

"If central bankers threw out all the data that was poorly measured, there would be very little information left on which to base their decisions." - A former research director at the Fed, quoted by Bloomberg's Caroline Baum

It all started in Germany and was called the Methodenstreit (the clash over methods), an intellectual war over how the study of economics would be conducted - either by the logical deductive reasoning embodied by the "Austrian" school or by the mathematical methods embodied by those leading lights of the positivist German Historical School, such as Professor Gustav von Schmoller. The war ended in a complete victory for the latter and now, a century and a half on, the mathematical methods championed by Schmoller have swept the globe.

In our times the Econometric Society's old slogan "Science is measurement" holds complete sway; math has become so imbedded into the study of economics that, to the untrained eye, your average economic tome can be easily mistaken for a calculus textbook. Yet, the accuracy of the flood of economic statistics we eagerly follow is rarely questioned. That is a question, though, of utmost importance given that we live in a country with a strong predilection for "economic planning", and the hordes of technocrats tasked with carrying out this function use those economic statistics to guide the decisions they make for the rest of us. So the question stands - how accurate are these numbers?

One of the more interesting (if forgotten) looks at this question was a book published in 1950 by the mathematician and economist Oskar Morgenstern. Called On the Accuracy of Economic Observations it held that economic data could be, and often was, compromised by a variety of factors, among them a lack of designed experiments, the possibility of lying, inadequate training of those tasked to provide or collect the data, and a lack of clear definitions or classifications. (Quick, what industry code should GE fall under?)

While in the "hard sciences" that study the physical world statistics are always supplied with an accompanying warning of their probable margin of error, nothing of the sort is hinted at when, for instance, GDP or employment numbers are released. There are a number of economists who have brought up this lack of transparency (among them Ben Bernanke) but for the most part the investing world seems to take economic statistics as gospel. Yet, judging by history, it behooves us to take them with a grain of salt.

One recent, glaring example of the trouble with numbers has been provided by the admission of Greece into the Euro, an admission allowed only due to what the world is now aware of - the Greeks cooked the books. Lest our Greek friends feel put upon, allowance must be made that they are not unique in this fashion; Japanese and American officials negotiated what the reported size of Japan's National Income would be during the post-war occupation period, as that number was used to determine the amount of financial aid that would be given to the Japanese.

Even in the absence of political meddling, economic statistics can be slippery things. In a November 2006 speech, Dallas Fed president Mr. Fisher stated that the Fed miscalculated its favored "core PCE" inflation gauge by 0.5% in early 2003, causing them to create more money and credit then it otherwise would have. Our current central bank head Ben Bernanke remarked in his Essays on the Great Depression that some data from that time is doubtless compromised because "actual wages paid fell relative to reported or official wage rates". In other words, when asked to provide statistical information some of the people involved simply lied.

Besides the imposition of politics and unavoidable human error, it has been pondered by minds far greater than mine as to whether it is even possible to mathematically measure certain economic activities. In his opus General Theory the great John Maynard Keynes writes of "the well known, but unavoidable, element of vagueness which admittedly attends the concept of the general price level" and therefore any attempt to compare price levels from one period to the next is "unsuitable for differential calculus." Nonetheless, from sea to shining sea highly trained econometricians collect reams of data, laboring under the concept that there is a "general price level" and, furthermore, that it can be measured.

I am in no way implying that rank dishonesty and slovenliness is endemic throughout our statistical bureaus. I am merely thinking out loud as to why the results of their honest efforts are never looked upon with a more critical eye. After all history - and logic - tells us to be careful, to question not only the accuracy but even (in certain cases) the usefulness of all these numbers. Sir John Cowperthwaite (the British Empire's financial secretary in Hong Kong from 1961 to 1971) flat out refused to allow the collection of economic data; regardless, the economy of Hong Kong, unmeasured and therefore unplanned, raced to the top of world rankings.

I will end by playing devil's advocate, all too happy to mimic Sir John and wonder: how many of our econometric castles, despite all the time and effort expended in creating them, are built on statistical sand?

The cynic in me replies - all of them.