The Second Annual Howard Littlefield Award for Putting the Con in Economics Awarded to Boozer Allan Hamilton - 3

Awarded by Pig Bodine, M.Sc., Ph.D., BM2, BEM, MAD, MDMA

Science, testing, smoothing-filtering-predicting, economic indicators

What distinguishes science from other disciplines is that science develops what are called theories, that is to say, explanations of how things work with a distinguishing feature that these explanations allow prediction of events that are precise and repeatable. These precise predictions become the basis for experiments. Without precise predictions, theory is not science but another form of literary fiction. This missing feature is what sets science aside from both economics and astrology, two peas in a pod.

That is, except for the work of Boozer Allan Hamilton. His explanation of how things work stands apart from the remainder of the feckless blather: it provides precise, repeatable predictions as well as control mechanisms. This is in sharp contrast with the entire spectrum of what has come to be called "social science," a type of ceremonial magic in which it is presumed the power of scientific prediction will be transferred by the use of the word "science." In similar fashion to cargo cults marching around mock airstrips while carrying twigs and wearing the words US painted on them in hope of coaxing the gods to drop manufactured goods, social scientists develop grandiose schemes aping theory and then devise "experiments" without predictive power, as it is dangerous to their prestige when their experiments fail. Their mock behavior in hindsight provides no insight, only an occasional success that is seldom repeatable except in the most trivial of cases and none of the powerful "god-given goods" that the theory of Newton provides even now, centuries after his work. In fact, it was in aping Newton that Smith devised his absurd automatic control scheme in which fealty and prayer to the God of the Free Market would cause Him to guide human affairs with His invisible hand. There is no doubt that, assuming humanity survives, future versions of what are now termed anthropologists and historians will marvel at the primitive cargo cults headed up by central banks.

Economics cannot be called science; nor is it engineering, though people sometimes use the term soft science. Scratch the term soft and it turns out that everything not intended to be fiction or poetry is science. (Ask an economist to explain how the "control mechanism" of overnight interest rates between US banks can affect overall US interest rates which are determined at international auction, or why there is no given temporal determinant regarding how long such a lowering will take to "correct" an economy, an idea at odds with the mythology of Smithism at whose altar the hierophants of the central banks pray.)

Boozer Allan Hamilton stands alone in having formulated a theory with the requisite predictive power for scientific work. His predictions can be and have been repeated by others. Were it not for Boozer's work, economics would have no credible basis for calling itself anything except ceremonial magic or fiction.

The most serious problem with Boozer's work is not that it doesn't predict well, which it does with the most uncanny reliability, but that his interpretations are not well received, a problem similar to that found within quantum mechanics (a discipline to which he has in fact made serious contributions). They are not only beyond the mathematical training of even the most mathematical of econometricians, but are indeed beyond many mathematicians and most engineers, crossing as they do a wild and unexplored borderland between analysis, differential geometry, Lie algebras and probability in infinite dimensions, with ties to quantum field theory and deep connections to recent developments in Cantorian set theory associated with names like Shelah, Martin, Solovay, Woodin, Jech and others who have built on the work of Cohen and Godel. Some of his results are worse than startling; they are downright unacceptable to economists refusing to believe such issues as the existence of inaccessible cardinals or the generalized continuum hypothesis or problems with even more abstruse origins can be the lynchpin of economics as a discipline. But this is a reality. Whether or not economic conditions are measureable, let along controllable, depends on questions that are independent of the standard axioms of set theory. Unthinkable. If this were the case in physics it would be a crisis, though it may be in the long run.

As Boozer explains it, the issue is the number of degrees of freedom the problems demand, which must needs be a Mahlo cardinal even to be able to define simple economic indicators meaningfully. To control requires more formidable machinery and wilder existence results.

I had intended to provide a survey of high level, a necessity as I will be the first to admit I not fully cognizant of the nuances of the Boozer Allan Hamilton model, commonly denoted BAH. Spatial and temporal constraints intervene.

The only predictions discussed herein regard national events for the US, but be aware that Boozer has been able to zoom the model to the corporate level. Unfortunately, these is a granularity that prevents sharp resolution on corporations operating within highly competitive areas, which is to be expected in any case (though his model does forecast where there will be competition and when it will disappear, with a quantitative definition of the "amount" of competition). BAH has also predicted new areas of technology sales, often well ahead of their appearance. Sometimes these are so surprising he has been laughed at, but that laughter dies. His predictions regarding drug companies, auto manufacturers, utility companies (using an early prototype of BAH, he predicted the entire Enron fiasco a full decade before it showed a blip on the radar screen), defense companies and software firms, among others, have been on the money. He has had some success with the computer hardware industry, but more with chip manufacturers than with individual box makers, though he forecasted the rise of Dell before there was a Dell and the AMD challenge to Intel. What we mean is that his model predicts revenues, dates of expansion or contraction, and many other factors including stock prices within finite intervals as a percentage of total sector valuation (the error is nonlinearly inversely proportional to the size of the interval).

Some of his macroeconomic predictions of such quantities as unemployment, gross national product, and similar magic numbers used by the ceremonial soothsayers are of interest because they come out of BAH much different than as officially crunched. Puzzled at first, he realized that by massaging data he could match the government's official results closely. For example, using the small business hidden job creation fudge factor and other fudge factors, he aped the labor department's figures for jobs created. Similarly he was able to bring down unemployment rates his model predicted by counting those barely employed or marginally employed (that is, people working only one hour per week) or those who had given up the search. His model fell from ten percent to the seven of the Boston Federal Reserve Bank to the official numbers of nearly five percent with the appropriate paring of categories. There were similar results with the various measures of gross domestic product.

Because of this, he considers these numbers meaningless and concentrates on other items, such as money supply. This can be a difficult number for the government, depending on the actual money type, that is M1 or M2 or M3. But he has also been able to get close to the official government figures for all three of these values. He has been able to provide an estimate of money created by credit transactions, a number the government seems to have difficulty with since it involves future flows, and impending credit crunches, defaults, and deflation due to default. This allows estimating inflation and deflation, the latter occurring more often than the government believes.

Boozer has developed quantitative and qualitative indicators across the entire social-economic spectrum. Some of them he calls economic indicators, but he cautions that the idea of an economy is nonsense. Such a entity for a nation or a household does not exist in isolation from the society at large. To attempt to study it in such isolation is the most pernicious of flapdoodles. His model includes the society as a whole and the globe as a whole. It is theoretically possible for its resolution to go to the individual household, but he points out that such computations are mathematically infeasible. He has proven they belong to the class NP.

It is my opinion that his most important predictions are political in nature, that is to say, electoral. He predicts the sorts of candidates to emerge and be chosen decades before they are elected. Of course, this sort of prediction does not lead to names, only parties and characteristics. The higher the level of office, the more interesting the prediction since as he points out, on a local level the local culture predominates. We concentrate on this aspect of his work in the next section.

He considers qualitative results to be most important. These are determined by looking at projections of flows from infinite-dimensional spaces to finite-dimensional spaces, leading to numerical results that are statistical predictors of events with precise values. The flows generally are deterministic predictors of qualitative events. The output of the predictor in the qualitative case is information regarding obstructions to flows, sources without flow (corresponding to beliefs out of synch with events), and some other arcane factors. It is at this point that I lose my footing and which the reader will need to glean from Boozer's write-ups to follow.

Boozer allows no discussion prior to the publication of predictions, as according to his methodology they cannot be made public until after their forecast dates. This is to ensure against the well known and little understood self-fulfillment syndrome in the area of social prognostication. However, the forecasts are made well prior to the dates, with model outputs to support the predictions, and then vaulted with an escrow service. His paper with predictions for the US 2008 elections is already in a vault. His basic predictions regarding the political shape of the 2006 elections were made prior to the 2000 election, but without specific candidates. The most recent forecast includes some candidates by name and outcome modulo mitigating factors such as accidental death (assassination attempts are not considered accidental and are part of forecast, though the success or failure is of course not something the BAH can predict).

I had hoped to sketch below an outline of the theory based on the science of information, the backbone of the model, but this has not materialized. Perhaps after Boozer's own write-ups they might be more appropriate. Instead I present an overview of his predictions within a long time horizon view of the US. The first of these predictions were made in the 1960s, with a prototype of BAH that evolved into the model today, which passes its tests so well it is no longer tinkered with.

Note that there have been backward predictions as well, that is to say, inputting historical data up to a certain period and then predicting. Those predictions have been accurate as well, depending on the availability of supporting data. They are now studied as an aid in historical research. Though the real purpose of BAH is filtering and prediction, the model can smooth as well (smoothing is to determine past events given future knowledge, which is useful in such tasks as orbit or missile or rocket trajectory determination, a sort of improved hindsight, while filtering is simply to determine what is happening at the time of the final data input, the "now" of real time). In the case of economics, this smoothing hindsight examined by historians to determine well defined mistakes ended a number of significant disputes. That is to say, hindsight being twenty-twenty, decisions on different levels from national to corporate can be seen to have been mistakes when smoothed, since then other decisions are substituted. A notable example is the Vietnam War, which the model smoothes away. Another is the invasion and current occupation of Iraq.

• << PAM Home
• 1
• 2
• Page 3
• 4
• 5
• Next >>