Economists Make Up a Worsening Story As They Go Along

X
Story Stream
recent articles

As I was driving in to our Buffalo office just before New Year's, the area was in the midst of its first real winter storm in two years. Despite Buffalo's wintry reputation, heavy snow fall had been absent since the winter of 2010, causing some perhaps unrealistic expectations. As the accumulation on the roads became troublesome, in my head I ran through some of the factors that would be involved in my safe arrival despite worsening conditions. There was the chosen route, whether that route had likely been plowed and whether that plowing had been sufficient to make up for drivers that would inevitably fail to exercise due care, etc. I also thought that it was likely that my wife prayed for my safe arrival.

Prayer might not be in the list of top factors determining safe travel, but after repeated observations over the course of the nearly ten years of wonderful marriage, I would be able to statistically "prove" that prayer was not only relevant, but a better predictor of safe travel than randomness. There are certainly enough observations over that period, run through a simple regression, that would no doubt yield results that claimed my wife praying for my safe arrival led to that very result.

The correlation of these two events, particularly in chronological sequence, would have statistical significance but would also be equally dubious. Post hoc ergo propter hoc is perhaps the most easily committed logical fallacy since we intuitively look for causation in sequencing. Just because two events occur in sequence does not mean that the first causes the second.

Rather, it is far more likely, from a non-mathematical, common sense perspective, that those other factors cited in the opening paragraph were more determinative toward cause and effect. In any statistical model that I create where prayer is the independent variable, I have subjectively chosen to ignore other factors as the whisps of randomness.

The problem is not just one of mathematics or even common sense. For much of our existence, cause and effect elude us. We cannot directly observe gravity, yet we infer its existence solely on the basis of repeated observation - to the point of predictability. Those observations are nothing more than correlations that seem to hold under all circumstances.

Philosopher David Hume postulated in the 1700's that since we cannot directly observe much of our world, only the correlations of otherwise seeming random events, our minds infer causation as nothing more than a fiction to make sense of it all. If two events occur, particularly in sequence, then our brains will come up with a story, causation, to connect those two events, providing context and the self-delusion of order out of chaos.

This lack of direct observation and dependence on the creation of causation from disorder afflicts more than just individual experience and logic. It is a feature of any complex system (in the technical sense) where there often exist far too many variables and interactions to infer much of anything out of isolation.

The human body is certainly a complex system, and despite modern advances in medicine and the science of biotechnology, there is much that is not understood about the actual physiological interactions amongst all the compounds and chemicals contained in our corpus. Because of that complexity, we do not really know for sure how medicines and drugs cause interactions and therefore lead to desired results. We can only observe correlations of when the introduction of an independent variable happens to coincide with desired outcomes.

Rigorous and complex statistical analysis is required to "prove" causation (for the FDA, anyway). But it does nothing of the sort; there is no proof, only confidence in inference. Yesterday morning, for example, Biogen halted a Phase III study of a new ALS medicine. According to Bloomberg, dexpramipexole did not show any significant efficacy above a placebo in improving the trajectory of the disease. This was after a Phase II trial that had shown 35% improvement.

What this likely means is that the 35% improvement was simply inferred from the introduction of the drug in those patients. And since some improvement was observed in those patients, such correlation led to a statistical analysis of inferred causation. But, according to the Phase III results, it was nothing more than post hoc ergo propter hoc simply because the complex interactions in the human body cannot be observed directly. Some other unseen and misunderstood factor was determined to be at work, forcing Biogen back into correlation mining among seemingly promising compounds.

The same holds true for the economic system. The billions or even trillions of daily interactions and facets that are featured in the economy are far too complex for economists to observe directly. The entire schematic of economics and econometrics is the inference of causation among disparate variables. The most obvious, and the one dangerously unquestioned, is that low interest rates stimulate activity.

Those two variables are intertwined in the canon of modern economics because of coincidence. Economists only suppose that the coincidence of low interest rates and increased economic activity are causally linked, and even have high statistical confidence that is the case. But, as David Hume noted, it might just be that economists are making up a story to fit two otherwise unrelated events. Or, as in the case of low interest rates, that coincidence is not necessarily predictive of future outcomes because a further inference is made that the future will look like the past.

Repeated observation leads to confident inferences, but it offers nothing in the way of dispositive proof. That is the essence of Nassim Taleb's The Black Swan. If Hume is right about inference of causation, then there is more than a bit of circular logic embedded within the foundation of all of modern economics. If inference is based on experience (observation), and that the future will look like the past, since the past is based on experience there is no way to conclusively derive anything about the future since we cannot, by definition, experience the future. It is nothing more than a leap of faith to assume that the future will be like the past. Repeated observation is not enough, particularly in a complex system, to assume such. We can safely infer gravity because of the standard of predictability; we cannot do the same with ZIRP.

To model the future in social systems exclusively on inferences of the past is a dangerous game. To do so requires the elimination of the dynamic nature of humanity. It requires that fit of circular logic, that we can infer the future out of experience when none of us has ever experienced the infinite possibilities of what the future can possibly hold.

In terms of economics, that is the story of the last few decades. The elimination of human dynamic variables from the economic equation has led to a false comfort of repeated observation. Low interest rates did coincide with stimulated economic growth, but it was not a simple cause and effect relationship. Repeated accumulations of debt led to an all-too-real upper limit in debt saturation - a limit not postulated by past experience, but one that existed in the praxeology of common sense. It was a factor cast aside in the vanity of presumed "knowledge" of the governing dynamics of the economic system.

The repeated nature of asset bubbles was also not inferred by past experience because of the unique nature of interactions in the present tense. How and why money or goods circulate in 2013 is unquestionably different than in 1993 or 1933. The financial system has undergone a radical shift in the decades since econometrics grew up out of the Great Inflation. The economics profession intentionally chooses to narrow the definition of inflation whether or not such a narrow definition fits the current circumstances. Because it supposedly "worked" in creating "meaningful" models before, it is simply assumed those correlations will prove durable.

What is missing out of those models is a more complete understanding of the complex pathways of the real economy and financial system. Practitioners of econometrics (and that includes every major central bank) are the equivalent of dexpramipexole stuck in Phase II. The CBO forecasted in 2010 that 2012 - 2014 would see 4.4% real growth. In June 2011, the FOMC's forecast for growth in 2012 was 3.3% to 3.7%.

In Europe, the economists at the ECB predicted 1.6% growth for 2012 as late as Q3 2011. Current expectations (Q4 2012 numbers have not been fully estimated yet) are for a continent-wide contraction of at least 0.5%. Not only were these simulations off by a matter of degree, they also got the sign wrong. Even in Germany, the vaunted Bundesbank was in December 2011 expecting 3% growth for 2012. It is likely that German economic growth was 0.4%.

All of these projections and predictions are based on post hoc ergo propter hoc. All of these projections fail because economists refuse to learn what they don't know. They have assumed a priori knowledge of a narrow slice of events and that statistical significance of correlation among those events are enough to map out the future. What is left out of their equations is assumed to be random noise - unimportant to the eventual outcome. But like the factors I left out of my model of arriving at work, those ignored facets are something other than random. We may not be able to observe or model these hidden factors, but ignoring them does not make them go away. Instead, it might be better to explore the contours of economic ignorance, and the "evolution" of economic thought that began to decry and debase common sense.

Rather than realizing the limitations of inference and scaling back or undergoing a meaningful retooling, correlation inference has gone into overdrive, spreading everywhere. My colleague Joe Calhoun pointed out a "study" reported in Psychology Today about how too much knowledge may actually reduce or disable optimum decision-making abilities. Joe's point in referencing the study was that it seemed to reinforce an old idiom traced to Albert Einstein to "make things as simple as possible, but no simpler".

To me, as some people expect of me, it was night and day different; my reaction was much more overstated. Einstein's cliché is a helpful guide, the brilliance of a brilliant man distilled into a small piece of advice to be left in the recesses of your mind and considered individually on individual terms. The study by Princeton and Stanford psychologists is purportedly evidence of a scientific conclusion. Despite the absurdity of the study's construction for extracting such a sweeping generalization as it does, the veneer and gloss of science is added to post hoc ergo propter hoc. In that it is given weight and treated far differently than Einstein's far more helpful advice.

The emerging "science" of behavioral economics is changing the very nature of not just economics, but as far reaching as government and law. Like straight economic theory, it is posited that these "scientific practitioners" can find optimal levels of certain variables that lead to optimal outcomes. Then, like computer programmers, they can descend into the real world of humanity and change the course of history through "nudging" people into those optimal routines whether they want to or not. That is one of the primary pillars of the Federal Reserve's ZIRP program - to make savings expensive so people will consume and spend regardless of personal and individual preference. Because it is "science" produced through rigorous observation and run through all manner of accepted statistical constructs, it cannot be argued against except by those who wish to wage a "War on Science".

In 2008, the Sacramento Municipal Utility District began to give some of its customers "grades" on their utility usage. They issued them to a randomly selected group of 35,000 households by comparing their electricity usage to comparable homes of their unnamed neighbors. This was largely based on a 2004 experiment conducted by a social psychologist at Arizona State University.

Dr. Robert Cialdini left three different messages on doorknobs in a "middle class" neighborhood north of San Diego. One message counseled energy consumers to save the earth for future generations, while a second emphasized financial savings. Neither of those messages, according to the study, produced significant results. It was the third message, that other neighbors had already started to conserve energy, that reported the most significant changes in behavior. Peer pressure had been given the scientific veneer for utilities to try to shame people into lowering energy consumption (of between 1% and 2%). Human or individual preference is not optimal, so people have to be nudged out of the tails and into the fat part of the bell curve.

In essence, that is what all this statistical penetration is really about: eliminating tails and curtailing kurtosis by crafting policies and incentives that prevent them from happening. It sounds like a wonderful idea - smoothing out the business cycle, defeating bank panics, making humans more robotic, etc., but it misses the essence of dynamism. Human systems are truly dynamic in that imagination is boundless. While it may seem a worthwhile endeavor to try to banish left tails (the "bad" outliers) in such a truly dynamic system, doing so may actually destroy the good outliers as well. It is a lack of recognition that progress itself is usually not optimal toward the status quo.

We do not know, because there is no way to observe, how human progress and innovation actually occurs. There is much common sense (and even observations) that innovation is actually a creature of failure. We can infer that eliminating failure and creating an "optimal" society would actually be counterproductive because the "science" of statistics is willfully blind to factors beyond its grasp, those same factors that give rise to meaningful advancement. The very upswing of human society, not the least of which was driven by economic progress, is dependent on "tail risks" actually coming true. Real imagination and innovation is, by definition, a tail event itself. Nudging all of society into the narrowness of the Bell Curve is not optimal for anything but debilitating sclerosis.

What's worse is that all of this is being done in the name of science. The scientific veneer is used as cover for political control in the inarguable trajectory of centralization. Only the priests of statistical knowledge are allowed to define what is "optimal". Individual preferences are now scientifically proven to be against the Greater Good, and thus the central agency self-delegated to provide an optimal utopia must "nudge" the tail outliers into compliance through all-too-real penalties (that will only increase over time).

There is no science here. The Federal Reserve can no more provide an optimal economy than I can provide an optimal length article. The human economy, indeed all of humanity, does not fit in the Bell Curve. If it did, there would be no progress or growth because such factors are driven by differences. Trying to remove those differences is the height of ignorance and arrogance since it betrays a stunning misunderstanding of what actually constitutes science and what it means to be an individual. This vain pursuit of objective standards to divine an optimal society is the recipe for stagnation and disaster. We are already living in the economics end of that vanity since any real recovery would be an outlier or tail event to the financial repression of this very type. Perhaps praying for a recovery would be statistically more likely to convey that outcome than relying on the pseudo-science that has arisen out of post hoc ergo propter hoc.

Jeffrey Snider is the Chief Investment Strategist of Alhambra Investment Partners, a registered investment advisor. 

Comment
Show commentsHide Comments

Related Articles