Sunday, April 20, 2014

There is no information from the future: the decision irrelevance of background uncertainty

An expansion of comments I made here.

There is a certain sort of regress argument that David Glasner, following Earl Thompson, is fond of using. He uses it about the value of bitcoins here:
The problem I have is that bitcoins can’t be used for anything except as a means of payment for something else. Bitcoins provide no real service distinct from being a means of payment. Think about it; if a bitcoin can’t be used for anything except to be given to someone else in exchange, that means that someday, someone is going to be stuck holding a bitcoin with no one left to give it to in exchange. When that happens, that stinky bitcoin won’t be worth a plum (or plugged) nickel, or a red cent. It will be as worthless as a three-dollar bill.
Now I grant you that that final moment of clarity might not happen for a long time – maybe not even for a very long time. But if anything is certain, it is certain that, sooner or later, such a moment must certainly come. But if it is certain that ultimately no one will accept a bitcoin in exchange, then it follows that no one forseeing that inevitable outcome would accept a bitcoin in exchange prior to that moment unless he or she is confident that there is some sucker out there who will accept in the interim. But since when does a theory of asset valuation premised on the existence of an unlimited supply of suckers count as an acceptable theory? Under the normal rationality assumptions that economists like to use, it is not possible to rationalize a positive price for a bitcoin at any point in its history.
Against which, I invoke economist Frank Knight's distinction between risk and uncertainty and the fundamental principle of modern physics that there is no information from the future. Knight distinguished between risk, which was quantifiable (in some broad sense at least) and uncertainty, which was not:
Uncertainty must be taken in a sense radically distinct from the familiar notion of Risk, from which it has never been properly separated.... The essential fact is that 'risk' means in some cases a quantity susceptible of measurement, while at other times it is something distinctly not of this character; and there are far-reaching and crucial differences in the bearings of the phenomena depending on which of the two is really present and operating.... It will appear that a measurable uncertainty, or 'risk' proper, as we shall use the term, is so far different from an unmeasurable one that it is not in effect an uncertainty at all.
We live with lots of background uncertainty all the time. It is certain that homo sapiens will become extinct, that the polity I belong (and pay taxes) to will cease to exist, that the Earth will cease to exist. But we have no information about specifically when these things will happen. So, they are certain to happen, but when they will happen is completely uncertain. That is, they are part of the normal background uncertainty about the certain-to-happen.

If I have no information about when the certain-to-happen-sometime event will occur (not merely if, but when) then it is not a factor in current exchanges because it is not part of current expectations because there is no specific information for it to become part of current expectations. It is not a risk, because there is no information to quantify it. It is, at best, part of the aforementioned background uncertainty about irrelevant-for-current-decision-making certainties.

An example is the dollars of the Confederate States of America (C$).  As the probability of the Confederacy ceasing to be due to the increasing likelihood of the final victory of the United States of America increased, the (at least roughly quantifiable) risk of there being no one to trade with in C$ increased dramatically, hence the value of the C$ plummeted, the time held between transactions [-] fell towards 0 [and so the velocity, and thus] the price level towards infinity.  But there was information to base specific expectations on and the risk could be quantified (and clearly was).

To put it another way, the certainty of occurrence at some no-specific-information-about time does not, of itself, constitute a likelihood of occurrence within the frame in which current expectations are formed and decision calculations are made.

Which applies to all such "regress from certain to happen but-no-specific-information-about-when" arguments, not merely bitcoins. 


  1. Indeed, the same point could be made about any stable currency, such as the US dollar or the British pound. It is certain that one day the United States of America and the United Kingdom will cease to exist, and hence there will be no nation backing these currencies. However, there is no reason to assume that this day will come soon -- or indeed, within either of our lifetimes!

  2. Lorenzo, I don't see how Knight's distinction between risk and uncertainty is connected to the physical principle that there is no information from the future. I am also dubious about the relevance of the physical principle to economics. Seems kind of reductionist to me.

    As to the collapse of the value of Confederate money at the end of the Civil War, I am inclined to view it as support for my position rather than for yours. After all, the Confederate dollar had already established itself as a generally acceptable medium of exchange in the Confederacy when the Confederacy still seemed to be militarily viable. Why should doubts about its acceptability in discharging future tax liabilities have affected the demand for it as a medium of exchange? Your example shows that acceptability as a medium of exchange requires more than current acceptability as a medium of exchange. If a non-monetary source of value is not necessary for a bitcoin to continue to serve as a medium of exchange once it has been widely accepted as a medium of exchange, why did the Confederate dollar require a non-monetary source of value and collapse once the non-monetary source of value became seriously suspect?

    Jordan179, The same point could not be made about any stable currency, as Lorenzo's example actually demonstrates. Any stable currency is acceptable in payment of taxes, providing a non-monetary source of value that a bitcoin doesn't have. When the expectation that the Confederate dollar would no longer be acceptable to discharge tax liabilities took hold, it rapidly lost its value.

    1. I am not arguing against the notion that expectations about future use affect the value/acceptability of a medium of exchange. If something is being used in transactions for its transaction utility, positive expectations about future use are required.

      I am arguing against the use of the "certain to end" regress without any further information. The existence of homo sapiens is certain to end. That certainty does not affect the value of any existing assets because we have no information about when that eventuality is likely to occur. Using the regress entails a presumption of information from the future that we do not have. And if physics says we cannot have information from the future, I fail to see how economics can conjure it up.

      What we have instead--in fact, all we have--is expectations. Expectations formed on the basis of past and present information. When such information allows us to, at least roughly, calculate when something is likely to occur, or how likely it is to occur within a specific time frame, then it enters into our expectations and our decision-making and not until then.

      So, in the absence of such information, such terminal certainties are, in the sense of when they will happen, just background uncertainties that do not affect our decision making about specific assets.