tag:blogger.com,1999:blog-17232051.post1103474074005602830..comments2024-03-18T22:32:52.802-04:00Comments on Noahpinion: The power and the terror of Irrational ExpectationsNoah Smithhttp://www.blogger.com/profile/09093917601641588575noreply@blogger.comBlogger81125tag:blogger.com,1999:blog-17232051.post-52290687832442878632013-02-01T07:44:10.886-05:002013-02-01T07:44:10.886-05:00Probably, but the Harper government is far to the ...Probably, but the Harper government is far to the right of most Canadian citizens.<br /><br />I have long suspected that we are far to the left of Americans because we are so closely involved with them. We see their news, we have many relatives cross border, we visit and travel in the US, and we share enough culture that we can easily watch -- and judge-- their culture and politics.<br /><br />Unfortunately, they don't seem to learn from our example of cheaper health care, much lower violent crime levels, much smaller prison system, etc. also unfortunately, their political strategies have been percolating across the border. Harper had Republican advisors during the last election, and has taken a lot of their strategies and approaches. Noni Mausahttps://www.blogger.com/profile/11198589990083966806noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-47574187781168863802013-02-01T00:04:09.852-05:002013-02-01T00:04:09.852-05:00Essentially all models are internally consistent, ...Essentially all models are internally consistent, mathematically. Rational Expectations is not required to make a model internally consistent. <br /><br />A model can be perfectly internally consistent and have the most irrational, crazy expectation formation process imaginable.Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-85507669744835082262013-01-31T22:49:27.361-05:002013-01-31T22:49:27.361-05:00"That's right. But why should we want an ..."That's right. But why should we want an "obvious way to discipline our modeling choices"? We should seek models that match observed reality, even if the methods used in those models are non-obvious"<br /><br />As long as you impose internal consistency to your model, be my guest.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-40556241870383155662013-01-31T13:27:44.927-05:002013-01-31T13:27:44.927-05:00Aren't there many real time surveys such as TI...Aren't there many real time surveys such as TIPS and Treasuries which implicitly embed expectations in prices?<br />Wouldn't it be very difficult to design a survey that adds any information that a spot price for a physical would not have.<br /><br />I am also bothered by your closing few paragraphs, they seem to invert the map and the territory. RE is an imperfect map, but that doesn't have any effect at all on the territory. Expectations are what they are. Whether or not some model of expectations turns out to be wrong has no scary implications whatsoever on expectations or what expectations may be. The only possible implication of showing an existing effort at modeling expectations to be incorrect is positive in as much as it would reflect an advancement of our collective knowledge of what expectations are.Danhttps://www.blogger.com/profile/15136541075745913165noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-48806160065916778812013-01-31T12:09:49.382-05:002013-01-31T12:09:49.382-05:00Incestuous amplification is just another name for ...Incestuous amplification is just another name for resonance. When the wavelength of the bullshit is an integer divisor of the size of the echo chamber, the bullshit reinforces itself until everything explodes. PHYSICS!Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-16251110319226899782013-01-31T11:51:47.829-05:002013-01-31T11:51:47.829-05:00I have been hearing my Canadian federal government...<i>I have been hearing my Canadian federal government (Conservatives, I'm afraid)</i><br /><br />The Canadian government is probably to the left of the Obama administration.Absalonhttps://www.blogger.com/profile/09131268683451462949noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-47768953682762028222013-01-31T10:45:52.545-05:002013-01-31T10:45:52.545-05:00Incestuous amplification, you mean, like you comin...Incestuous amplification, you mean, like you coming here to repeat Krugman's theories?CAnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-30436889332408026952013-01-31T10:12:44.542-05:002013-01-31T10:12:44.542-05:00@To
"parametrize and (fit to/validate with) ...@To<br /><br />"parametrize and (fit to/validate with) survey data."<br /><br />Why survey data? There's a lot of different ways of finding estimates for parameters.<br /><br />"Any noise in the surveys will end up in the model anyway."<br /><br />No, not necessarily "noise" if there's bias in the survey that leads people to consistently pick one way. The classic example was:<br /><br />Are you happy? Do you have a significant other?<br /><br />And<br /><br />Do you have a significant other? Are you happy?<br /><br />That's not noise since it will systematically make the first group happier, and the second less happy.<br /><br />Unless you mean multiple different surveys and their effects average out. I don't think that's the case, but even if it is, I'd say that still doesn't eliminate other ways of estimating parameters.<br /><br />Good points, though. What do you mean by "make one up"? As in a representative agent?<br /><br />@Anon <br /><br />"Whether you consider this a bug or a feature of the model depends on your preferences."<br /><br />Very interesting, but I guess it depends on the scenario. Perhaps it could diverge to a distribution of beliefs? Although that brings in noise, it won't account for the more tangible effects of why people have heterogeneous expectations.<br /><br />All good points, and thanks for the link.InjuredEhttps://www.blogger.com/profile/00685282771952562871noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-50137765363241745412013-01-31T08:11:00.620-05:002013-01-31T08:11:00.620-05:00@Noah Smith: The paper uses a survey but AFAICT bu...@Noah Smith: The paper uses a survey but AFAICT building a model is a stated goal. My point was: is it possible to skip the expectation-modelling stage altogether and use survey data directly in macro models ?<br /><br />@Injured: To produce an expectation model you have to either (a) simulate the brains of a few million humans and their environment (b) make one up or (c) parametrize and (fit to/validate with) survey data. To this non-economist (c) looks like the only viable (i.e. scientific) option. Any noise in the surveys will end up in the model anyway.<br /><br />@Anon: Interesting, thanks.Tonoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-80633464390247567652013-01-31T06:59:28.832-05:002013-01-31T06:59:28.832-05:00In addition to the general problems of surveys, th...In addition to the general problems of surveys, there are some additional problems which prevents researchers from using expectations in their models.<br /><br />First, there does not exist an aggregation theorem for beliefs, i.e. something similar to the aggregation of preferences. So researchers have to rely on static models that can be solved or keep the numbers of agents small in a computational model.<br /><br />Second, to get interesting dynamics the beliefs of the agents in a model have to change, i.e. in an asset pricing model with fixed heterogeneous beliefs agents will only trade because of endowment/dividend shocks. Hence, we will get the same aggregate dynamics as in the standard model.<br /><br />So, in order to get some interesting dynamics in the model we will need the beliefs to change, i.e. we need some theory for learning. However, many learning theories have the property that the beliefs of everyone will converge. Whether you consider this a bug or a feature of the model depends on your preferences.<br /><br />The third problem is the welfare evaluation in these models, i.e. ex-ante at most one belief can be correct but the question is which one.<br /><br />To sum it up: Deviating from the rational expectations hypothesis and allowing for heterogeneous beliefs is not as trivial as it sounds. While some progress has been made, there is still a lot to do.<br /><br />There is also a nice survey from Woodford about the different approaches in the literature:<br /><br />http://www.columbia.edu/~mw2230/AREcon.pdfAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-70036402420080371762013-01-31T03:14:39.459-05:002013-01-31T03:14:39.459-05:00They are trying to scare Canadians into laying off...They are trying to scare Canadians into laying off from a debt fueled consumption bubble that is rivaling the American pigs at their high point of greed. Real estate in Vancouver and Toronto would be unaffordable to an average Canadian and stagnant productivity means most of the recent consumption has been enabled by ultra cheap money. And while its true that if you can afford it now is the time to lock down cheap credit, the problem is that most credit in Canada is on a 10 year re-financing schedule and if you go out and buy a Vancouver condo for 400,000 expecting to pay 1000/1500 a month and in 5 years when you try to refinance it jumps to 3000 a month youll be out of a home. And your bank will have a bunch of holes in its balance sheet where its customers used to be. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-79942835533728010212013-01-30T21:12:22.327-05:002013-01-30T21:12:22.327-05:00I have been hearing my Canadian federal government...I have been hearing my Canadian federal government (Conservatives, I'm afraid) for the past couple of years warning citizens to beware of the inevitable rising interest rates on the horizon. None of the Harper team, as far as I have heard, have explained what they think will trigger this rise.<br /><br />As I understand it, prices rise when demand increases or supply falls. Is our prime minister implying that the money supply is going to tighten up? Since we have sovereign currency, I can't see how that would happen. (Though he has banished the Canadian penny, and is talking about scrapping the nickel, too.)<br /><br />So, can we look forward to higher demand then? I'm not sure where that demand would come from. Although Canadian citizens are not so bad off as the Americans, we're also not springing forward with higher wages, more secure jobs, and in general better prospects for our households. We're going the opposite direction, if anything.<br /><br />So, do our conservatives expect to see businesses or foreign investors driving up interest rates? In which case, I am thinking that the time for ordinary Canadians to borrow is now, while rates are laughably low, before the big financial hippos jump in the watering hole and get it all churned up.<br /><br />Or lastly, do they not actually expect rates to rise, but are saying so at every opportunity for some arcane reason of their own?<br /><br />Anyone who can offer some illumination on this topic would be my best buddy. This week, anyway.<br /><br /><br /><br />Noni Mausahttps://www.blogger.com/profile/11198589990083966806noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-39579508292256773932013-01-30T18:48:21.300-05:002013-01-30T18:48:21.300-05:00Surveys are plagued by multiple problems which mak...Surveys are plagued by multiple problems which make them relatively less trustworthy. I haven't looked into the Malmendier paper, so I can't say anything about that. However, in what order you ask, how you ask, etc., all have tangible effects on the outcomes. What's the most objective way of wording questions, and what's the most reasonable order, etc.? Who knows.<br /><br />Hence why you think they circumvent. It is a good point, though. I imagine it wouldn't be too difficult to do. I'd just take it with a grain of salt.InjuredEhttps://www.blogger.com/profile/00685282771952562871noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-45560197405900555552013-01-30T16:53:43.774-05:002013-01-30T16:53:43.774-05:00Not unless we can observe the financial portfolios...Not unless we can observe the financial portfolios of various age cohorts!Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-91644073595883839132013-01-30T16:49:46.480-05:002013-01-30T16:49:46.480-05:00Yep, everyone should read Weitzman.Yep, everyone should read Weitzman.Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-78687805631905002882013-01-30T16:47:58.679-05:002013-01-30T16:47:58.679-05:00The Malmendier paper is a survey.The Malmendier paper is a survey.Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-81006569996273635882013-01-30T16:27:02.901-05:002013-01-30T16:27:02.901-05:00I wonder: wouldn't the simplest way to get (re...I wonder: wouldn't the simplest way to get (reliable) expectations to feed into models be simply to, you know, ask people ? A good old-fashioned survey. I can't fathom why economists insist on stumbling on what is perhaps the toughest problem in all of the sciences, instead of circumventing it.Tonoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-85345302550688594832013-01-30T12:41:51.455-05:002013-01-30T12:41:51.455-05:00I have more money than usual in term deposits. I ...I have more money than usual in term deposits. I have from time to time shorted stocks and made money at it overall but you run up against the problem that "the markets can stay irrational longer than you can stay solvent." Right now you also have the problem that government, including the Fed, is working very hard to manipulate the markets higher. Absalonhttps://www.blogger.com/profile/09131268683451462949noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-87391483910425330632013-01-30T11:46:09.783-05:002013-01-30T11:46:09.783-05:00The paper you posted before is an example, of cour...<i>The paper you posted before is an example, of course (maybe that's why it would be better just to call rational expectations an internal consistency requirement). The problem is that once you decide that your model doesn't need to be internally consistent, then anything goes.</i><br /><br />No, I disagree. Labeling RE "internal consistency" is just the same as inventing a new definition for "internal consistency". Non-RE models can easily be mathematically consistent. RE requires that expectations be consistent with outcomes. That is the definition of RE. Thus, saying "RE = internal consistency" is just restating the RE hypothesis. But it still could easily be wrong.<br /><br /><i>The problem is that after you take away that requirement, there's not an obvious way to discipline your modelling choices either. Period...</i><br /><br />That's right. But why should we want an "obvious way to discipline our modeling choices"? We should seek models that match observed reality, even if the methods used in those models are non-obvious.Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-39135061884748867322013-01-30T11:38:03.217-05:002013-01-30T11:38:03.217-05:00Great idea, and I think there are lots of examples...Great idea, and I think there are lots of examples of this. Larry Summer just did an interview where he quoted his adviser as having said that things always take longer than you think they should and then go faster than you think they will. It takes a long time to become a flash-in-the-pan!<br /><br />There are oodles of examples of discontinuous events. The question to start with would be why WOULDN'T expectations exhibit discontinuity?<br /><br />Lucas himself will of course be the cause of just such an event in Economics. He's clearly wrong, a tremendous amount of work is being expended on making this case. There will be a moment where the weight of the evidence outweighs the weight of his influence, (when the 100th monkey gets it so to say) and viola! Macro will be free to advance.<br /><br />[as an aside, the 'weight of Lucas' influence is another phenomena, you likely saw the recent post about the military term for how highly dubious ideas become not just accepted, but viewed as certainties. “Incestuous amplification” happen when a closed group of people repeat the same things to each other – and when accepting the group’s preconceptions itself becomes a necessary ticket to being in the in-group. A fundamentally flawed notion becomes part of what everyone knows, where “everyone” means by definition only people who accept the flawed notion. IE creates discontinuity when the evidence overwhelms beliefs.]<br /><br /><br /><br />Danhttps://www.blogger.com/profile/15136541075745913165noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-77419479717526727052013-01-30T11:26:50.083-05:002013-01-30T11:26:50.083-05:00The more complicated things get, the more opportun...<i>The more complicated things get, the more opportunity there is to draw conclusions without really knowing what's going on.</i><br /><br />Possibly the most perspicacious sentence I have ever read...Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-5712569986080491402013-01-30T11:11:43.326-05:002013-01-30T11:11:43.326-05:00"Also, remember that forecasting doesn't ..."<i>Also, remember that forecasting doesn't care about the Lucas Critique.</i>"<br /><br />Oh, yes, definitely. Private sector forecasters who use RE are being foolish IMO (and I doubt there are a whole lot of them, because they will have learned that it was foolish if they hadn't already figured it out). <br /><br />However, forecasting by policymakers is a bit more complicated issue. If the forecast is just meant to be an honest best guess of the outcome, for its own sake, then there's no point in using RE. On the other hand, if the forecast is going to guide policy, then the Lucas critique applies, so there's a case for using some Lucas-robust model of expectations to generate a forecast, or at least to test the robustness of the forecast. As you note, Lucas-robust doesn't imply RE, but I'm a bit skeptical of introducing more complexity by using more realistic but still Lucas-robust models of expectations. The more complicated things get, the more opportunity there is to draw conclusions without really knowing what's going on.Andy Harlesshttps://www.blogger.com/profile/17582263872850949568noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-39517175072464140612013-01-30T10:13:03.677-05:002013-01-30T10:13:03.677-05:00Noah, the difficulty of updating expectations also...Noah, the difficulty of updating expectations also jibes with the idea that for right-wing economists, it's always the 1970's again.Barry DeCiccohttps://www.blogger.com/profile/04735814736387033844noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-52201271418406741152013-01-30T09:37:26.518-05:002013-01-30T09:37:26.518-05:00I was addressing Noah.
To answer to your comment,...I was addressing Noah. <br />To answer to your comment, "Rational expectations" has never pretended to be the only perfectly rational behavior (?). The paper you posted before is an example, of course (maybe that's why it would be better just to call rational expectations an internal consistency requirement). The problem is that once you decide that your model doesn't need to be internally consistent, then anything goes. The equity premium puzzle is a puzzle for models with internal consistency and the other usual stuff we know; once you take that away, there's no puzzle anymore and that should be obvious. The problem is that after you take away that requirement, there's not an obvious way to discipline your modelling choices either. Period...Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-38629569404498899602013-01-30T08:41:49.227-05:002013-01-30T08:41:49.227-05:00"the central bank might be unsuccessful in be..."the central bank might be unsuccessful in beating deflation, right up until the moment when hyperinflation runs wild."<br /><br />FWIW, I once tried to think about such discontinuity in terms of "phase transition" and "latent heat" <a href="http://himaginary.blogspot.jp/2011/06/phase-transition-analogy-of-inflation.html" rel="nofollow">here</a>.himaginaryhttp://himaginary.blogspot.com/noreply@blogger.com