tag:blogger.com,1999:blog-17232051.post7356318220211890765..comments2024-03-18T22:32:52.802-04:00Comments on Noahpinion: Risk premia or behavioral craziness?Noah Smithhttp://www.blogger.com/profile/09093917601641588575noreply@blogger.comBlogger54125tag:blogger.com,1999:blog-17232051.post-10124176416650257122013-12-25T13:52:59.957-05:002013-12-25T13:52:59.957-05:00"Decisive: I was told in my first econometric..."Decisive: I was told in my first econometrics class that we can reject but we never accept. Science is never done.<br /><br />That has always been my understanding as well."<br /><br />You're both bloody crazy. In the first place, I can accept anything I want to. The question is when SHOULD I accept. That's basically trivial. <br /><br />But in the real world, as opposed to the fantasyland you two evidently inhabit, actual decisions need to be made. You can't just say "we're never done;" you have to actually do something. The something might be to determine what line of research to pursue, or whether to buy insurance personally, or whether society should collectively buy insurance against Global Warming. <br /><br />So, hypotheses need to be accepted, even granting that we might need to change them in the future.<br /><br />All of which was pointed out in 1954 by Richard Rudner in a paper published in Philosophy of Science titled "The Scientist Qua Scientist Makes Value Judgements."<br /><br />And while we're on the subject of Richards, I can't resist pointing out the Feynman's understanding of the real Cargo Cultists, as opposed to his straw men, was essentially nonexistent. Anthropology was not his strong suit. Nor was philosophy, albeit the advice to bend over backwards to be completely honest is excellent. His undoubted genius didn't make him infallible (not even in physics; he made some famous blunders.)<br /><br /><br />"I think we would find (Lake Wobegon fashion) that the average person thinks they are well above average."<br /><br />You THINK!? You're talking about this in ignorance of the decades of experimental results on the subject? The results of which were that yes people do think that, just in case you want to know.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-38692669828127202802013-12-25T13:00:23.606-05:002013-12-25T13:00:23.606-05:00"look up Phillip Tetlock if not familar), the..."look up Phillip Tetlock if not familar), they both sound like big hedgehogs rather than foxes"<br /><br />The terminology was originated by Isaiah Berlin, as reading Tetlock (of whom I think very highly) will tell you.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-21460401560319697172013-12-25T12:53:17.791-05:002013-12-25T12:53:17.791-05:00I'm not so sure about that. For example, Dean ...I'm not so sure about that. For example, Dean Baker detected the housing bubble in real time. He pointed out e.g. the huge deviation from historical norms in the ratio of housing and rental prices. He was certainly right, and I don't believe he had any general definition of bubble from which he was working. He just said something like "whatever a bubble is, this is a bubble."<br /><br />If I'm wrong please correct me.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-19445963468169376012013-12-23T15:33:44.956-05:002013-12-23T15:33:44.956-05:00In any case your actions are marked to market, ver...In any case your actions are marked to market, very quickly.<br /><br />(Quite different from writing posts in economics, where you can get away predicting high inflation and a run on UST, and then very quietly coming up with models where the tail wags the dog!)Som Dasguptahttps://www.blogger.com/profile/11848089230329819807noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-56915011812356462572013-12-23T15:29:20.143-05:002013-12-23T15:29:20.143-05:00What happens in the (financial) asset markets, is ...What happens in the (financial) asset markets, is roughly something like this: an influential investor (think of a fund leveraging $15b+, not just one or two) spots the potential to start a price run. He goes on TeeVee. at that point, as a trader you either decide to go along with him or stay out. Other large fund managers decide to back the momentum and we have the beginnings of a "bubble". If you can exit with the big group (not necessarily telegraphed in real time), you win. <br /><br />For the most part, many are engaging in behavioural craziness.<br /><br />Som Dasguptahttps://www.blogger.com/profile/11848089230329819807noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-6553249473376876282013-12-22T18:55:35.399-05:002013-12-22T18:55:35.399-05:00As someone who has been watching cryptocoin market...As someone who has been watching cryptocoin markets a lot lately, I'm pretty sure it's behavioural craziness.Anonymoushttps://www.blogger.com/profile/17797928102421103389noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-14245956583788068492013-12-21T22:37:01.239-05:002013-12-21T22:37:01.239-05:00Actually there are quite a few studies that get di...Actually there are quite a few studies that get directly at whether or not risk premia vary over time. The best are panel studies following groups of people to see what they do, especially with portfolio allocation. One such from 2006 is at www.columbia.edu/~pc2167/ChiapporiPaiellaOct2006.pdf . They find no evidence of any noticeable change in relative risk aversion for people in their sample panel, and this is consistent with pretty much every other study that has attempted to study how peoples' attitudes to risk change over time. They don't very much, certainly not remotely enough to support the Fama/Cochrane fantasy, and this has been in the lit for decades now, only getting more strongly supported and refined as we go along. One does not need to wave hands about people completely rewiring their brains. The data is there and pretty darned robust. This ballgame is basically over for anybody willing to look at the careful studies, not all of them just surveys but ones based on behavior as well.<br /><br />Barkley Rosserrosserjb@jmu.eduhttps://www.blogger.com/profile/09300046915843554101noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-45534543827742442602013-12-21T17:58:19.180-05:002013-12-21T17:58:19.180-05:00Stephen,
I think you are on the money with the La...Stephen,<br /><br />I think you are on the money with the Lake Wobegon effect argument. People start bubbling because they think they are smarter than the others and can make money by selling high and getting out before the others. Something like that. When it is repeated a bunch of times with the same group, they figure out that they cannot do so, at least not systematically.rosserjb@jmu.eduhttps://www.blogger.com/profile/09300046915843554101noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-7409195915492424462013-12-21T15:42:24.015-05:002013-12-21T15:42:24.015-05:00You may like my next experiment... ;-)You may like my next experiment... ;-)Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-53363930309630554792013-12-21T12:19:28.470-05:002013-12-21T12:19:28.470-05:00Thought I was done, but not yet. Here's an ide...Thought I was done, but not yet. Here's an idea. It's not a craziness theory, but along the lines you're thinking about. It seems we have evolved to be optimistic and to think well of ourselves. If we took a poll, I think we would find (Lake Wobegon fashion) that the average person thinks they are well above average. That has served us well as a species - we survive, we prosper, we innovate. So, if we put these average people in a Vernon Smith experiment, they tend to be willing to accept an asset in exchange at a price higher than the fundamental, because they think there is some other dummy in the crowd who will take it at a higher price. I don't know what has already been done on this, but the theory project would be to put all of that in a model, and see what you need to make it fly. You want some initial evolution stage where the survivors tend to be the ones who think well of themselves, and then a stage where the survivors trade assets. Stephen Williamsonhttps://www.blogger.com/profile/01434465858419028592noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-8134803471374866372013-12-21T11:38:53.100-05:002013-12-21T11:38:53.100-05:00I guess the bottom line for me is that I haven'...I guess the bottom line for me is that I haven't seen experimental work or survey evidence that has changed my views about how the world works, or about how to do research. Experimental work is fraught with problems, and I already know that the actual behavior of any individual economic actor - me, for example - will look nothing like what I write down in models. I think learning on the part of market participants is probably important, but I also think that the learning models that people write down are pretty crappy. Solving learning models in a sensible way seems beyond what we know how to do technically. So, when I'm choosing how to do research, I tend to focus on the things that are feasible, and try to build on what we know well. Stephen Williamsonhttps://www.blogger.com/profile/01434465858419028592noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-17131884350931185242013-12-21T11:28:21.970-05:002013-12-21T11:28:21.970-05:00I'll check out her work! Thanks!
Yes, that...I'll check out her work! Thanks!<br /><br />Yes, that's also how I've simulated an infinite horizon. When you get only a 0.001% chance of the experiment going past 5:00, I think you probably get pretty close to true infinite-horizon behavior, in terms of infinite # of periods. But of course, infinitely many periods and infinitely long time aren't the same thing, and you can't do an experiment where the results will affect a subject's entire future lifetime, so life-cycle effects are one thing that experiments will probably never have much to say about.Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-1023309361198966682013-12-21T11:17:56.285-05:002013-12-21T11:17:56.285-05:00"but when things get dynamic+stochastic, we j..."but when things get dynamic+stochastic, we just don't understand much of what we see in the lab."<br /><br />Some people have done monetary experiments, for example Gabriele Camera works on this. Those experiments are particularly difficult. In monetary models, there's typically in infinite horizon, and valued money is supported by the belief that people will accept the money in the future, which is supported by the belief that people in the future will accept money in the further future, etc. But how do you set that up in the lab? Some experimental designs have the experiment proceed in rounds, and we determine whether we proceed to the next round by rolling a die - so play stops with probability 1/6. So that might seem OK, as it seems there is always positive probability of going to the next round. But the players know that, when 5 pm comes, they are going to go home no matter what. When the subjects play the game, they typically trade the money - the stuff has value even though its "fundamental value" is zero. But note that, the way the experiment was set up, they're already violating the theory, as it's a finite game they're playing. I think they're valuing the money because they come into the experiment knowing what money is from their daily lives. They see some objects in the experiment and think, "this is money," and then they trade the stuff.Stephen Williamsonhttps://www.blogger.com/profile/01434465858419028592noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-14299462205484172942013-12-21T10:21:26.446-05:002013-12-21T10:21:26.446-05:00Decisive: I was told in my first econometrics clas...<i>Decisive: I was told in my first econometrics class that we can reject but we never accept. Science is never done.</i><br /><br />That has always been my understanding as well. <br /><br /><i>You seem to want to take these lab experiments very seriously. In the experiments I have seen written up, I'm never much surprised by the results. Basically a group of undergraduates trying to figure something out.</i><br /><br />Actually, in some recent experiments (including one of mine), people are finding that even when the subjects provably understand how the thing works, they act very weird unless we find some way to make them believe that the *other* subjects understand it too.<br /><br /><i>nd you can't control what baggage the subjects bring into the experiment. A lot of people do this stuff now, but I think you have a lot of convincing to do with the rest of the economics profession.</i><br /><br />That is certainly true! I've been thinking of ways to do that.<br /><br />Have you read David Levine's survey of experiments? He didn't say it explicitly, but I think the upshot is that when things are either static or non-stochastic, standard micro theories tend to do a good job of predicting lab results, but when things get dynamic+stochastic, we just don't understand much of what we see in the lab.<br /><br /><i>Then there's the problem of where all this leads. Some policymaker gets it in his or her head that people are acting crazy, and then starts trying to make them behave.</i><br /><br />Well, maybe. But given the way policymakers behave in real life, I'd say heeding economics experiments is the least of their sins...<br /><br />Anyway, I think the main implication of "inefficient markets" is in corporate finance. I strongly suspect that compensating executives with options and stock is a much worse idea than we think.Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-6313194027675585442013-12-21T08:58:27.573-05:002013-12-21T08:58:27.573-05:00Then there's the problem of where all this lea...Then there's the problem of where all this leads. Some policymaker gets it in his or her head that people are acting crazy, and then starts trying to make them behave.Stephen Williamsonhttps://www.blogger.com/profile/01434465858419028592noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-51562064163758971262013-12-21T08:52:04.197-05:002013-12-21T08:52:04.197-05:00Decisive: I was told in my first econometrics clas...Decisive: I was told in my first econometrics class that we can reject but we never accept. Science is never done.<br /><br />You seem to want to take these lab experiments very seriously. In the experiments I have seen written up, I'm never much surprised by the results. Basically a group of undergraduates trying to figure something out. And you can't control what baggage the subjects bring into the experiment. A lot of people do this stuff now, but I think you have a lot of convincing to do with the rest of the economics profession.Stephen Williamsonhttps://www.blogger.com/profile/01434465858419028592noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-73788505097619724942013-12-20T20:26:00.032-05:002013-12-20T20:26:00.032-05:00Also, I don't know how you're defining P a...Also, I don't know how you're defining P and FV, but as I remember it, If P>FV for an asset, then you could not buy it and get different assets instead that would give you the same risk-adjusted expected return at a lower price. And, if that's the case, "hyper-rational people would not pay that P, and it would fall if everyone was hyperrational. So I don't know how you're defining fundamental value.Richard H. Serlinhttps://www.blogger.com/profile/09824966626830758801noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-22463268166365732662013-12-20T20:09:52.336-05:002013-12-20T20:09:52.336-05:00"I saw a dog acting funny yesterday, which su..."I saw a dog acting funny yesterday, which suggested to me that all humans are crazy."<br /><br />Oy, com'on Stephen. Are you actually that detached from the real world that you think this kind of behavior and ginormous lack of expertise and knowledge is some rare crazy dog phenomenon. Well, here's some empirical data that's not crazy dog, it's percentages like 2/3rds:<br /><br />Consider these microfoundations from Chicago economist Richard Thayler:<br /><br />• Suppose you had $100 in a savings account and the interest rate was 2 percent a year. After five years, how much do you think you would have if you left the money to grow? More than $102, exactly $102 or less than $102?<br /><br />• Imagine that the interest rate on your savings account was 1 percent a year and that inflation was 2 percent. After one year, would you be able to buy more than, the same as or less than you could today with the money?<br /><br />• Do you think this statement is true or false: “Buying a single company stock usually provides a safer return than a stock mutual fund”?<br /><br />Anyone with even a basic understanding of compound interest, inflation and diversification should know that the answers to these questions are “more than,” “less than” and “false.” Yet in a survey of Americans over age 50 conducted by the economists Annamaria Lusardi of George Washington University and Olivia S. Mitchell of the Wharton School of the University of Pennsylvania, only a third could answer all three questions correctly.<br /><br />At:<br /><br />http://www.nytimes.com/2013/10/06/business/financial-literacy-beyond-the-classroom.html?pagewanted=all <br /><br />And these:<br /><br />From a Center for Economic and Financial Literacy survey:<br /><br />"65% of people answered incorrectly when asked how many reindeer would remain if Santa had to lay off 25% of his eight reindeer."<br /><br />"1 in 3 people didn't know how much money a person would be spending on gifts if they spent 1% of their $50,000/year salary."<br /><br />– Personal Finance for Dummies, 7th edition, 2012, page 9Richard H. Serlinhttps://www.blogger.com/profile/09824966626830758801noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-75556948488220633512013-12-20T20:03:06.616-05:002013-12-20T20:03:06.616-05:00This comment has been removed by the author.Richard H. Serlinhttps://www.blogger.com/profile/09824966626830758801noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-59776891649414264592013-12-20T18:52:26.118-05:002013-12-20T18:52:26.118-05:00Now we're getting to it. That's suboptimal...<i>Now we're getting to it. That's suboptimal given some objective function and some constraints. Maybe you just have the objective function or the constraints wrong.</i><br /><br />That's always true. Objective functions will always be a free parameter. You can only restrict further and further the set of objective functions consistent with observed actions.<br /><br /><i>By the way, what do you mean by decisive?</i><br /><br />I mean, something that would convince Noah Smith that real-world bubbles (of the "price goes up and then down") variety are often the result of investor mistakes rather than changes in discount rates.<br /><br /><i>http://www.lhup.edu/~DSIMANEK/cargocul.htm</i><br /><br />I feel like I've made this joke before, only not about experimental economics... ;-)<br /><br /><i>So, the experimental subjects behave in predictable ways. They're not crazy at all.</i><br /><br />Predictable in a lab doesn't mean predictable in the real world, even if the behaviors are one and the same in both domains. For example, suppose I find that in the lab, when I induce bad weather conditions, people predictably sell stocks. That doesn't mean that bad weather is predictable in the real world. See?<br /><br />This also addresses Point #2.<br /><br /><i>I (and other people) have written down models which determine asset prices. If Shiller lived in those models, and measured P and FV, he would find P > FV. And the other people who live in that model are all hyper-rational.</i><br /><br />Sure, investor irrationality is only one possible reason for P>FV in the real world. In the lab, the causes of P>FV can be (to some degree, hopefully to a large degree) controlled and separated. Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-25749343150930624672013-12-20T18:33:30.233-05:002013-12-20T18:33:30.233-05:00Back again.
1. I think what you're saying is ...Back again.<br /><br />1. I think what you're saying is that the Vernon Smith bubble experiment has been repeated many times with different subjects, and the experimenters get roughly the same results. So, the experimental subjects behave in predictable ways. They're not crazy at all.<br /><br />2. Suppose that we could rerun the last 15 years with the same time series of "fundamentals" (you get to choose what those are). Does craziness theory tell us that we should observe the same time series of asset prices the second time around, or something different?<br /><br />3. I assume that FV is financespeak for "fundamental value." I (and other people) have written down models which determine asset prices. If Shiller lived in those models, and measured P and FV, he would find P > FV. And the other people who live in that model are all hyper-rational.Stephen Williamsonhttps://www.blogger.com/profile/01434465858419028592noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-50145922072327553162013-12-20T17:24:51.581-05:002013-12-20T17:24:51.581-05:00""I always see dogs acting funny"
...""I always see dogs acting funny"<br /><br />Now you're becoming overconfident. You claim to be a physicist, so you're probably aware of Feynman's commencement address:<br /><br />http://www.lhup.edu/~DSIMANEK/cargocul.htmStephen Williamsonhttps://www.blogger.com/profile/01434465858419028592noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-39834987203791620282013-12-20T17:21:33.481-05:002013-12-20T17:21:33.481-05:00"suboptimal behavior coordinated across indiv..."suboptimal behavior coordinated across individuals"<br /><br />Now we're getting to it. That's suboptimal given some objective function and some constraints. Maybe you just have the objective function or the constraints wrong.<br /><br />By the way, what do you mean by decisive?Stephen Williamsonhttps://www.blogger.com/profile/01434465858419028592noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-78830699242595987262013-12-20T16:43:01.419-05:002013-12-20T16:43:01.419-05:00I saw a dog acting funny yesterday, which suggeste...<i>I saw a dog acting funny yesterday, which suggested to me that all humans are crazy.</i><br /><br />You should know that the original Vernon Smith bubble result has been replicated with professional trader participants and with much larger numbers of participants. Still suggestive rather than decisive, but more suggestive than just the original 6-undergrads-in-a-lab thing. Also, the original Smith experiment has been replicated hundreds of times (I have replicated it four times myself), so you should change your straw man from "I saw a dog acting funny" to "I always see dogs acting funny".<br /><br /><i>What exactly is this craziness theory?</i><br /><br />P>FV<br /><br /><i>What does it imply about what we should see in the data?</i><br /><br />P>E[FV|observables]<br /><br /><i>Actually, what I'm thinking is that there are no implications at all.</i><br /><br />That's a perfectly logical way to think, given the paradox of induction.<br /><br /><i>In human behavior, craziness is another word for mental illness. But mentally ill people actually behave in very predictable ways, once your veil of ignorance is lifted.</i><br /><br />This is a good point, and the reason I used the word "craziness" is because I had initially used the word "spazzing", but found out that this is a fairly serious slur in the UK. So I changed it. What I really mean is "suboptimal behavior coordinated across individuals". But that doesn't sound as cute.Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-57876260114995963752013-12-20T15:51:47.376-05:002013-12-20T15:51:47.376-05:00Interesting, I hadn't realised. What do people...Interesting, I hadn't realised. What do people with cerebral palsy think about it in the US? As an englishman the whole post was basically unreadable before.<br /><br />I wouldn't use it, and I definitely don't think Noah would have meant to offend anyone, but intent isn't really the point is it, it's perception. Funny how that's different. Left Outsidehttp://leftoutside.wordpress.comnoreply@blogger.com