O, for a one-armed economist! Which politician was it who decried the on the other hand rhetoric of his economic advisors? Doesn't matter. He just wanted an answer, and the economists he consulted kept on adding complexities.

It's no wonder. The world's a complex place. And economics has to accommodate and explain that complexity.

Milton Friedman insisted that prediction was the hallmark of science, and therefore economics. But Oscar Morgenstern noted the problem with this, long before Friedman: Any prediction of social events which is a causal factor for the predicted effects must necessarily go wrong. Willi Meyer, in Beyond Choice (Subjectivism, Intelligibility, and Economic Understanding, Israel Kirzner, ed.), correctly denies this Strong Impossibility Theorem a logical certainty, but gives it its due as a synthetic statement. It is, well, almost common sense.

Say an economist studies the development of an industry. He comes to the conclusion that investors have overvalued stock in several major corporations in said industry. He predicts that, down the road, this will lead to a major bubble burst.

The article gets published, and is picked up by financial journalists. And the stock goes into an immediate dive.

The prediction has influenced the outcome, speeding up the process of stock devaluation. And, perhaps, by doing so, has prevented a major bubble by pricking the bubble early on.

That shows an influence that prediction has, and that must be modified in the predictions.

Trouble is, it's hard to predict what will be made of any one prediction: nothing? what the predictor thinks savvy? or too much?

This last leads to a factor far more troublesome than the Strong Impossibility Theorem. It is the possibility of the self-fulfilling prophecy. One makes a prediction in order to influence events, perhaps to increase one's own reputation as a predictor.

This make economic prediction a form of wizardry. An art, perhaps, not unlike that of a confindence game. (And, after listening to the former head of the Federal Reserve, one might very well come to believe this was economics' very method.)

This shows the inherent limits that social sciences have at the predictive level, the level of exact predictions.

Hayek suggested that economics should be, instead, about the making of pattern predictions, and that gets it closer, but still, the problems of interference with the result by the very act of prediction remain a big problem.

This is the result of the key feature of human action: human ends change to adapt to information about possible goals and means, and predictions affect the relationship between various goals and means; so people choose differently based on predictions.

And this can lead to an amazing degree of complexity, far beyond the complexity of the statics that economists have traditionally theorized about. Predictions can be made that

• intentionally affect the outcome to prove the prediction (self-fulfilling prophecy as wizardry)
• intentionally affect the outcome contrary to the apparent content of the prediction (sly manipulation)
• unintentionally affect the outcome in a way that favors the prediction (self-fulfilling prophecy by means of Thomas Theorem factors)
• unintentionally affect the outcome to muddy or disprove the prediction (impossibility theorem)

These factors turn the nature of economics as a social science away from any simple predictive science. It turns the science into a critique of simple-minded scientism, doesn't it? A more dialectical enterprise, perhaps. I don't see how the science can be looked upon as a science in quite the same way physics is. The very nature of the science as influencing its observation may seem like just a macro instance of the Heisenberg Uncertainty Principle in quantum mechanics. But the problem is much larger, and on every level of economics.

Perhaps this makes economics more like evolutionary biology. Or ecology. There's a place for prediction, but a limited place.

Even meteorology and climatology have to incorporate human influences. But they can do so in pretty simple modelling. Trouble is, in economics, human influence is at the discipline's heart, and susceptibility to influence from predictions and the science itself can change outcomes in unforeseeable ways, due not only to complexity but to the very structure of the central events: the human choices and actions.

Funny, though, I've never heard anyone suggest that the proper way to conduct economic research would be in guild secrecy. One could only test predictions that are never allowed to reach the populace!

Perhaps one reason I've never heard of this is not merely the trouble with keeping secrets. It's the problem that any observation and prediction that an economist can make can be made by actors in society itself. After all, some investors are extremely sophisticated. They have every reason to be.

These thoughts come to mind as I break open a book that just arrived in the mail, Fooled by Randomness: The Hidden Life of Chance in Life and in the Markets, by Nassim Nicholas Taleb. I'll see what new levels of complexity become apparent by reading this much-praised work.

After writing the above, I came across this, by Nicolai Foss:

While we can all agree that ideas matter, how much do they matter? Is much, and perhaps most, of social reality essentially bootstrap phenomena in which the Thomas Theorem (i.e., the the situations that men define as true, become true for them) holds true? Do the social sciences bootstrap much of social reality in the sense that social science decisively affects the agents that social scientists study? Or, are there constants, stable mechanisms, etc. that exist and work regardless of what social scientists believe about them (as, I suppose, many economists would hold)?

The idea that theorizing affects the objects of theorizing — that is, the notion of “reflexivity” — has been an important one in sociology for a long time (Thomas wrote about it in the 1920s; Merton in the 1940s). It has become a Leitmotiv in the sociology of knowledge. It has also been a recurring theme in economics (in connection with predictions and the modeling of expectations; e.g., the debate surrounding the Lucas critique), and it has been treated by philosophers as well (e.g. Popper). However, this literature has not discussed the extent to which reflexivity (in the above sense) obtains.

The issue is obviously very difficult to get a hold on. However, I also believe it is a crucial one, particularly in management where we have recently witnessed people essentially arguing that economics-as-applied to management is (nothing but?) a self-fulfilling prophecy (i.e., this paper). So, do any of our readers know of literature that can help to frame and answer the questions with which this bleg began?

Ah, another call for help. I just asked for such citations on two email discussion groups.

But, as I try to indicate in my post above, the Thomas Theorem is not the whole of the problem. It's not even the half of it. It's about the quarter of it. At most. The Thomas Theorem does not really take in the possibility of a intentional influence of a prediction, nor the unintended process of a prediction unintentionally influencing reality so as to seemingly falsify the prediction.

But then, Foss was mainly talking about reflexivity, reciprocal influence and all that, and used the Thomas Theorem as if in synecdochic relationship to reflexivity.

The blog I quote from, by the way, looks fantastic. Organizations and Markets: great title for a very large endeavor with a wide perspective. I had not seen it before. I've now bookmarked it as a savvy source for new and old ideas.