Betting your views, follow-up

My claims that you are not required to publicly bet your views have received enough denunciations (mostly commentators on other blogs) that I feel it is time to rub some salt into the open wounds. 

First, I have nothing against betting one's views and indeed I've done it with Bryan Caplan, though mostly for fun.  No one bets his or her views consistently, or that frequently relative to the number of views, so I am simply pointing out that a default of "no betting at all" is an OK point along that spectrum.  In fact it is a homage to the idea of division of labor.  Nor have I seen the critics publicly revealing their equity and other asset portfolios, the real bets which matter in financial terms.

More generally, we may wish that researchers express "real commitment" to their views.  I don't see betting as an essential part of such a commitment portfolio.  "Simply being a certain way" when it comes to inquiry is #1 on my list.  Having a good and deserved personal reputation for truth-seeking is another.  An emphasis on betting, in my view, represents an odd economistic view that commitments should be viewed essentially or primarily in monetary terms.  From a variety of other settings (try giving your wife "cash" for Valentine's Day) we know that signaling commitments through money can backfire.  Might that be the case here as well?

Doesn't the very offer to bet signal that your view is possibly based on private, not easily verified-in-public information?  Doesn't it signal a lack of confidence in the publicly available information itself?

If I think of the scientists who have influenced me most, or done the world the most good, very few of them were practitioners of public betting.  Furthermore that correlation is no accident (I would bet that most of them regarded, or would have regarded, the idea skeptically, or as a kind of public relations stunt).  You might think that is a market failure of some kind and maybe it is.  But in the meantime, if you do have a truly good idea, maybe the best course of action is to mimic the other holders of truly good ideas and that means not betting publicly on your idea.

I'm not suggesting that such a mimicry strategy is best with p = 1.0, only that it gives you one of a number of rationales for not betting at all.

Comments

But how do you acquire "a good and deserved personal reputation for truth-seeking" or being a certain "good way" "when it comes to inquiry", and why do you believe prestigious scientists who most influenced you were that way or had such a deserved reputation? Yes, the most prestigious people usually don't bet, as bets would signal a lower status; few are publicly willing to doubt the "good way" of the highest in status. But this doesn't imply they were actually truth-seeking.

Tyler,

I think you're correct that scientists often have many non-monetary ways to "bet" on their theories (performing an experiment in a natural science comes to mind here). But they often do not, or these ways also happen to correlate with other rewarding behavior disconnected from the truth. In addition, scientists often seem to actively avoid non-monetary (status?) bets. Robin's observation that nearly no one keeps or cares about track records of scientific or pundit predictions seems particularly relevant here.

Perhaps you have difficulty distinguishing between beliefs you believe to be true and beliefs you hold for signaling reasons? I know I do. I have a simple litmus test with regards to beliefs: if I'm not willing to bet on them, they aren't real beliefs.

In the spirit of non-monetary betting, I'll put my (non-existent, admittedly) status on the line: I predict that betting (if restricted to scientists) produces more accurate predictions from each individual bettor than simply polling those scientists. I believe a number of these experiments have already been performed.

The biggest problem with non-betting seems to be for a layman deciding which academic to believe. Without keeping track records or bets, how are we to know what ideas or thinkers are best? You mention that most of the great scientists didn't bet on anything, but who regards them as great? A liberal, conservative and a Marxist might each have very different ideas on which scientists were great. Betting and track records offers ways to objectively determine the worth of scientists.

So what is your alternative? How else do you suggest we measure the sincerity of academics, and get more accurate predictions? Obviously non-academics can easily bet on academic subjects, but they'd be able to do so in a more informed manner if they knew the sincerity of the academics.

TC's remark about his bets with Bryan Caplan point to a problem with betting on academic views: it can reasonably made to work with not too large amounts and between friendly, or at least mutually respecting people.

But in issues directly relevant to the larger public, and therefore with a political angle to them, we need substantial amounts, and probably parties that do not like or really trust each other, like Krugman and Mankiw.

In such cases there will either be an enormous amount of haggling over terms, and perhaps legal involvement, or you have a bluffer's world, where people are forced to accept unfavourable terms in order not to look evasive. Neither seems really attractive as a way to generate debate on important issues.

Tyler,
I think what bothers me is that when economists such as Paul Krugman recommend a very large macro policy intervention, they are asking the tax payers (me) make a bet. Because having an even bigger stimulus package would require the average tax payer to have a lot of skin in the game, it seems only fair that Mr. Krugman be equally willing to put his money on the line. Not every belief needs to be signaled with a bet, but when economists are asking others to make a large gamble based on their advice it seems fair to ask them to ante up.

One thing that ought to be pointed out in this context is that some knowledgeable people (not scientists) are effectively prohibited from “betting their views† by professional, ethical, legal and policy restrictions. Here I am specifically speaking of attorneys and accountants who, because of their professional duties, access to inside information and need for independence, are extremely limited in their ability to participate in markets. I know people who would dearly have loved to bet their views over the past decade but have been locked out of a very wide range of investments, or have been so restricted in their ability to trade that it is impractical to do so.

Every blog-post title should end with a p! This one included.

Here is the ironic part about Paul Krugman not betting his view:

His view is that we need a HUGE stimulus paid for by taxpayers, but that government debt is unlikely to be a problem because he believes the Obama administration's forecast for the recovery. So in essence, he is unwilling to bet his view directly, yet he is asking (demanding?) taxpayers to bet massively on it.

I think that if someone is spending time and resources formulating a theory or making up an opinion in his mind, he is already betting. I mean he/she could be doing something else with that time right? So, if they are proved wrong, they lose anyway. And if they are right, they will get their rewards. It's just not a monetary bet, but a bet nonetheless.

The three main advantages that I see to betting are 1) it can make you think more clearly and precisely about your predictions & analyses, 2) it draws attention to an issue, and 3) the result can be a test of someone's credibility.

But any of these can backfire (I'll illustrate with the Krugman-Mankiw example). 1) Sometimes a bet challenge becomes a distraction and makes people combative instead of analytic (has anyone involved in this dispute even given their own best estimate for 2013 GDP?). 2) Sometimes the issue isn't important enough for the attention (what's at stake here? A budget projection? The unit root hypothesis?) 3) Some outcomes (like 5-year GDP growth) are inherently unpredictable (they have wide confidence intervals), so even someone who makes the best estimate possible could have a 30 or 40% chance of losing the bet (Mankiw's proposed terms give Krugman nearly a 50% chance of losing even if Obama's projections are the best possible estimates).

A system that involved frequent betting between economists might work (although its advantages over a system that involved explicit, easily scoreable predictions instead of bets isn't obvious). But without that system, occasional predictions don't seem very useful unless two people have a stark disagreement on an issue that they both find very important (as with the Simon-Ehrlich wager).

I think what bothers me is that when economists such as Paul Krugman recommend a very large macro policy intervention, they are asking the tax payers (me) make a bet. Because having an even bigger stimulus package would require the average tax payer to have a lot of skin in the game, it seems only fair that Mr. Krugman be equally willing to put his money on the line.

[Krugman]'s view is that we need a HUGE stimulus paid for by taxpayers, but that government debt is unlikely to be a problem because he believes the Obama administration's forecast for the recovery. So in essence, he is unwilling to bet his view directly, yet he is asking (demanding?) taxpayers to bet massively on it.

Excuse me. Is Krugman exempt from paying taxes? I'm guessing his income tax bill is well above average, so how is he asking others to make a bet he is unwilling to make?

The notion, that a bet on a given theory should indicate the strength of one's belief in that theory is, I think, misleading. Rather, bets should be seen as a way to manage risk. If an academic economist has a theory predicting X, then it would be fair to say, that economist gains reputation in the event that X occurs. However, if instead of X, Y occurs, the economist might lose some reputation. Assuming, that there is diminishing marginal utility to "reputation" and that reputation can be easily quantified in terms of money, the economist will find it optimal to bet against X occurring. If X occurs, reputation but loses some money. If X does not occur, then some reputation is lost, while some money is won.

Concluding from the above, economists who bet their prediction will be fulfilled are simply classified as risk-loving, rather than being confident in their own theories. Furthermore, the size of the bet would then depend on how much economists value reputation.

I'd be pretty shocked if the Dalai Lama bet his views, but I always listen to him for ideas, just the same.

Comments for this post are closed