Forecasting Futures

by R. Passov

“In … economics we are faced with … a need for accurate forecasts, yet our ability to predict the future has been found wanting”

—Systems Economics: D. Orrell and P. McSharry, International Journal of Forecasting, Vol 25 (2009)

*          *          *          *

The Stanford Encyclopedia of the Philosophy of Economics (2018) stabs at a definition of the science:

… At first glance, the difficulties in defining economics may not appear serious. Economics is, after all, concerned with aspects of the production, exchange, distribution, and consumption of commodities and services. But this claim and the terms it contains are vague…

Stanford [] portrays economics as a new science only coming into its own under Adam Smith, whose work “… offers a systematic Inquiry into the Nature and Causes of the Wealth of Nations.

In Smith’s economics, an actor led by an “…Invisible Hand…intending only his own gain … gives rise to regularities …”

These “…regularities…” – the unintended consequences of individual choices – “…give rise to an object of scientific investigation.”

The individual choices, it can be argued, are the domain of contemporary Microeconomics while the regularities to which they give rise, might in some sense be our Macroeconomics.

*          *          *          *

Smith, a jocular, bulbous-nosed Scotsman, after graduating from Oxford in 1748 parlayed a penchant for soap-box speeches into a professor-ship at the University of Glasgow. There he rose to Chair of Philosophy. Economics would wait until 1903 when, finally, Cambridge set it apart from the moral sciences.

In 1759 Smith produced a work entitled “A Theory of Moral Sentiments” in which he mused on “… how a man who is interested chiefly in himself [can] make moral judgements that satisfy other people.”

His answer: “When people confront moral choices they imagine an Impartial Spectator who … advises them…Instead of following their self-interest, they take the imaginary observer’s advice,” and in so doing, “…decide on the basis of sympathy, not selfishness.”

After publishing Moral Sentiments, Smith followed the money. For two years, he wandered through France tutoring the son of a gentlemen who, as the Chancellor of the Exchequer, devised the tax policies that sparked the Boston Tea Party.

During his wanderings Smith sought, among others, Hume, Voltaire and Benjamin Franklin. After exhausting his stipend, he spent a decade socializing at the Literary Club of London, turning his notebooks into The Wealth of Nations. The Impartial Spectator morphed into the Invisible Hand. Empathy turned into self-interest.

*          *          *          *

According to Stanford, economists have struggled in trying to move beyond Smith’s views:

… flirt[ing] with a less substantive characterization of individual motivation and with a more expansive view of the domain of economics. In his influential monograph, An Essay on the Nature and Significance of Economic Science, Lionel Robbins defined economics as ‘the science which studies human behavior as a relationship between ends and scarce means which have alternative uses’…

But, in the manner of an economist, Robbins has more to say:

The efforts of economists during the last hundred and fifty years have resulted in the establishment of a body of generalizations whose subsequent accuracy and importance are open to question only by the ignorant or the perverse. But they have achieved no unanimity concerning the ultimate nature of the common subject-matter of these generalizations. …

We all talk about the same things, but we have not yet agreed what it is we are talking about…

Robbins then tries to solve his conundrum by quoting John Stuart Mill:

Like the wall of a city it has usually been erected, not to be a receptacle for such edifices as might afterwards ring up, but to circumscribe and aggregate [what is] already in existence.

*          *          *          *

Economics, according to Robbins, could not be defined in the absence of the development of the activities which are its subject matter. And so commerce, markets, trade, borrowing and lending, specialization and so on and so on, after coming into existence without the aid of economists, deserve scientific inquiry.

Contrast this with physics which, it can be argued, is discovering natural laws in existence for a really long time. But though physics might one day be responsible for the end of the world as we know it – through some form of nuclear catastrophe, say – physicists are not held to a prediction of when this might happen.

Economists, on the other hand, notwithstanding that they may define their science as backward-looking, are held to account for their ‘dismal’ record at forecasting. According to a growing chorus, a critical fault of the profession is failing to anticipate the great financial crises of the early 21st century. This failure, it’s argued, is one of two that now confront the science.

The other failure, perhaps the systemic reason for the first, lurks at the core of modern, Neo Classical, economics: The assumption that Smith’s selfish agents are rational actors.

*          *          *          *

One of the louder critics, Richard Bookstabber, writes in The End of Theory that humans are not rational; our processes are not ergodic and so cannot be contained within statistics. Consequently, the toy models and flawed theories of economics will never map the real world, i.e. never predict the great recessions.

But suppose the science of economics could predict Great Recessions. What would this imply?

Robert Lucas won the 1995 Nobel in Economics for “…having developed and applied the hypothesis of rational expectations, and thereby having transformed macroeconomic analysis…” [Wikipedia]

Lucas credits his insights to John Muth, known as:

…the father of the rational expectations revolution in economics”, primarily due to his article “Rational Expectations and the Theory of Price Movements” from 1961. [Wikipedia]

Muth argues that if a model is predictive, then expectations of actors in the modeled economy will align with the objective, probabilistic forecasts of the model. For if the expectations of actors differed from those implicit in an accurate model of their behavior, then those in control of the model could trade against this discrepancy.

This idea that an accurate model of the economy, by logic, forces actors within the model to expect a future as predicted by the model is what Muth deemed ‘rational’:

…expectations, since they are informed predictions of future events, are essentially the same as the predictions of the relevant economic theory. At the risk of confusing this purely descriptive hypothesis with a pronouncement as to what firms ought to do, we call such expectations “rational.”

As Muth points out, this is not the same as saying that economic agents are rational or that economists are accurate forecasters. This simply says that if you have a model that is an accurate predictor of the future then the agents that are the subject of such a model must by logic have expectations that align with the model.

(Try this thought experiment: Suppose there is an economy with only one economist. Suppose that one economist publishes a forecast calling for a recession six months forward.

Now suppose that sole economist, having predicted a recession, witnesses those in the economy acting to protect themselves from her forecast by increasing their savings, with the result being the recession is brought forward six months.

What is this all-knowing forecaster to do when her model predicts a second recession? Should she let out the forecast in advance, knowing that it would accelerate the timing of the recession? Or, should she offer a false prediction that calls for additional growth with the hope that, in acting as a result of her false forecast, economic agents increase spending and investment, thereby averting a recession?)

*          *          *          *

What does Bookstabber offer in place of the faulty theories of modern economics? Less theory and more computation. Actors aren’t smoothly rational. Instead we are individual agents whose behaviors exhibit irreducible computational challenges.

The way forward, argues Bookstabber, is to let agents run loose in a vast simulation, consuming reams and reams of computation. But is this new way helpful? It’s possible because computation is virtually costless. But upon close inspection, costless computation simply pushes assumptions down to lower levels.

Humans in fact are not acting within the machine; those many agents are themselves models whose starting points are assumptions and whose decisions are sampled from a distribution.

Properties of systems emerge in ways that illuminate. But exactly what will emerge, by the very definition of the process, cannot be determined in advance. And furthermore, it’s also evident that whatever does emerge is indeterminately sensitive to initial conditions.

So while these simulations are illuminating, they are no more predicative as to a specific future than the models and theories they strive to replace.

*          *          *          *

George Akerloff, also a Nobel laureate, also sees something wrong in his profession. But he defines the problem as one of “…sins of omission…” – sins that are the result of the apparatus grown to support economics as a hard science.

Akerloff observes that economics is as much about how as it is about what. The consequence of reducing the definition of a science to its methodology is, he argues, something Thomas Kuhn (The Structure of Scientific Revolutions) failed to anticipate – a science that, by filtering everything through the lens of mathematics, strangles its own progress.

Perhaps Akerloff misses the more interesting consequence of his observation.

*          *          *          *

I suspect that neither Smith nor any famous economists that followed will admit to having met a truly rational actor. But from Smith’s time forward, the argument is that actors need not know how they choose. The Invisible Hand gives rise to regularities, or as Milton Friedman put it: it doesn’t matter what the decision maker knows, only that it appears that he decides as if.

This convenient excuse lasted until the costs of getting more granular reduced to the point where it was no longer possible to rest on simple assumptions.

Mathematics invaded economics as part of the renaissance in rational thinking – the Enlightenment. But economists have been greedy collaborators. Freidman, in his seminal essay The Methodology of Positive Economics, begs mathematics to drive further into his science. Mathematical rigor promised Freidman a sanctuary from those whose contributions seemed no more than cocktail-party opinions.

But as he writes in that essay, Freidman coveted more. Grounded in the rigors of mathematics, economists would continue to enjoy their vaunted domain, high above the other social sciences. Today, there is a Nobel Prize waiting for economists but not for their fellow social scientists. And perhaps more importantly, there is a Council of Economic Advisors with whom US Presidents consult but no such council for, say, psychologists or sociologists.

*          *          *          *

Bookstabber is wrong in believing that economists did not anticipate the Great Recession. Many did but with few exceptions they were running hedge funds (See R. Rajan, 2005). He’s also wrong in suggesting that economists cannot predict the future.

There are many futures. The more difficult to predict are those subject to the capriciousness of political actors, along with the reflexivity of economic agents. But anyone who reads Brian Arthur from 1989 will find an accurate forecast of today’s network economics.

Over the course of its history, the science of economics has built upon the assumptions necessary to map a discipline into mathematics. The result of building a science in this way is a domain subject to the limits of computation. Assumptions once necessary to mathematize economics, appear wrong. The truth is simpler; in a world of declining computational cost, the half-life of assumptions grows ever shorter.

Economics, since it admits to a science that sees itself through a rear-view mirror, evolving as circumstances evolve, should acknowledge this aspect about itself; that it is a science driven by the cost of excavation and its major tool for exploring is computation.

Some practitioners hold to views of rationality, stretching heuristics to the breaking point. This is a resistance to the hard work necessary to revolutionize the field. If this resistance remains, it’s likely that sociology, once thought less hard than economics and so lower down in the hierarchy of sciences, will, relishing data and agnostic to its lack of axioms, rise above.



System Economics:  Overcoming the pitfalls of Forecasting Models, D. Orrell, P. McSharry, International Journal of Forecasting 25 (2009)

New Ideas From Dead Economists: T. Buchholz, Plume (2007)

An Essay on The Nature and Significance of Economic Science: L. Robbins, MacMillan & Co., LTD, St. Martin St. London, (1932)

The End of Theory. Financial Crises, the Failure of Economics, and the Sweep of Human Interaction: R. Bookstabber, Princeton University Press (2017)

Sins of Omission and the Practice of Economics: G. Akerloff, Journal of Economic Literature, Vol. 58 (2020)

The Methodology of Positive Economics: M. Friedman, University of Chicago Press (1966)

Has Financial Development Made the World Riskier? R. Rajan, NBER (2005)

Competing Technologies, Increasing Returns, and Lock-In: Brian Arthur, The Economic Journal (1989)