Individual truths, democratic truths, expert truths and market truths
By Scott Sumner
Most people make a distinction between what people believe to be true and what is actually true. I accept Richard Rorty‘s claim that this distinction is not useful. Because our view is in the minority, I’d like to reframe this discussion of truth in a way that is broadly accepted.
I’d like to consider propositions that are currently highly uncertain, but where scientific progress is likely to lead to a broad consensus in the future. Now Rortians like me and normal people can agree on a useful distinction. In this case it’s the distinction between things currently regarded as likely to be true with varying degrees of probability, and things that will be broadly accepted as true (or false) at some specified date in the future, as new information comes in.
For instance, let’s consider a question that was once uncertain but is now widely accepted. Consider what would have happened in 2008 if people were asked, “Will the Large Hadron Collider staff announce the discovery of a Higgs boson within 10 years. Please give us a probability value for the likelihood of this announcement.” (In fact, the Higgs boson was discovered in 2012; or at least it was announced, for you conspiracy buffs.)
I’d like you to consider how much faith we should have in answers provided by each of the following 4 groups, back in 2008.
1. Average individuals. (Individual truth)
2. The average probability provided by all individuals. (Democratic truth.)
3. The average probability provided by all PhD physicists (Expert truth)
4. The price of a Higgs boson futures contract that pays $1 if a Higgs boson is discovered within 10 years, and $0 otherwise. (Market truth)
I view these 4 types of truth as a reasonable summary of the options. In our society, various choices are delegated to various groups. Choices about what shoes to buy are made by individuals, the choice of President is made democratically, monetary policy is made by experts, whereas asset price values are determined by markets.
I suspect that most economists believe that Expert Truth is best when you are trying to figure out the probability of a Higgs boson discovery. I do not agree. I believe that market truth is almost always the most reliable method of identifying the truth.
Of course in any single case there’s lot of luck involved. If physicists said there was a 53% chance and markets implied a 48% chance, then the outcome of that single Higgs boson prediction experiment would tell us very little. But you can imagine hundreds of such experiments. Earlier questions might have included: “How many moons will Voyager discover around Saturn?” Or, “Will a mammal be cloned within 10 years?” Today we could ask, “What is the probability that sea levels will rise by at least 3 feet by 2100?” Or, “What is the probability of a 8.0 or greater earthquake in California within 20 years?” Etc., etc.
With hundreds of such questions, a “horse race” between markets and expert opinion would be quite interesting. I believe that markets would come out slightly ahead. Not because traders are smarter that scientists, rather because markets would incorporate all useful information, including the appropriate weight to put on each experts opinion, as well as any bias that might occur on politically charged questions.
Most economists do not agree with me. Although they teach the efficient markets hypothesis in class, they prefer that monetary policy be set by the FOMC, not an NGDP futures market. They also accept the criticism (from Queen Elizabeth) that our inability to predict the 2008 financial crisis was a black eye for the economics profession, whereas I noticed that the markets also failed to predict the crisis, and hence don’t believe that economists should accept any blame for not seeing the crisis coming.
Richard Rorty famously said, “Truth is what your contemporaries let you get away with saying.” Which I gather means that claims are regarded as true when they are broadly accepted by the public. But which public? Everyone? College professors who are Rorty’s peers?
For me, the most useful types of truth are the beliefs revealed in market prices. Markets are giant engines for generating true statements about the world—the most efficient truth engines ever discovered. Markets are especially useful in economics. It’s possible that nature will never reveal to philosophers the answer to basic questions about God, free will, consciousness, etc., but we economists are bombarded with interesting new data points almost every day. While the data measure the underlying reality with some imprecision, they are obviously useful enough to move asset prices. We’d really like to know the level of NGDP in the first quarter of 2021 as estimated by the BEA (with all its flaws), and an NGDP futures market is the best way to estimate that future BEA announcement.
What are we to make of the fact that there are currently no subsidized prediction markets for 2021 NGDP, or the 2050 sea level rise? I recall that Robin Hanson is as cynical as Jack Nicholson on that question:
I also believe that markets are the best way to answer moral questions. Thus the best way to figure out whether eating meat is immoral is not to read lots of books on moral philosophy (although that might have some value). Rather we need a prediction market on the probability that meat consumption in the US (taken from live animals) will be banned by 2100. Ditto for abortion. The same logic applies to aesthetics. Which films will be on the top 100 lists of film critics in 2100?
For paintings, the market already provides a sort of prediction, as explained by Tyler Cowen. Yes, one could argue that Tyler slightly overstates his case, as the “question” is not precisely specified. Paintings can have historical value that is separate from aesthetics—consider the recent Da Vinci that sold for $450 million. But I believe that Tyler is broadly correct.
Some people distinguish between scientific truth and moral “opinions”. I don’t. We can be more confident that murder will still be regarded as wrong in the year 2119 than we can be that current models of quantum mechanics and relativity will still be true 100 years from now.
PS. Each year the blogger Scott Alexander gives probabilities on a wide range of outcomes, and then a year later evaluates the results. As far as I recall, the outcomes do positively correlate with the probabilities that he assigns. I believe that the 538 website does something similar.
PPS. I would not rule out that some individual forecasters would usually beat the democratic forecast. For instance, Bryan Caplan almost always wins his bets. But the average individual forecast is probably inferior to the average forecast of all individuals.
PPPS. For those who never saw the movie, Jack says, “You can’t handle the truth!” But I knew that even though I never saw the film. I hope your knowledge of pop culture is not even worse than mine.