Tuesday, May 13, 2014

Today's econ ain't your pappy's econ




I have a new article in Quartz, about how the economics discipline has changed in the last 40 years. Obviously these aren't the only things that have changed, but I thought that these are the three things that will have the biggest impact, on the world and on people going into the field. Excerpts:
Do a YouTube search for “Milton Friedman.” Most of the hits will be speeches mixing economic theory with political philosophy. You’ll see Friedman talking about the value of greed, for example, or holding forth on socialism versus capitalism. Most entertaining is the series of videos titled “Milton Friedman schools young idealist,” in which young, hippie-looking kids stand up and challenge the old man’s capitalist values, only to be hit over the head with Econ 101. 
As an econ blogger, I get the sense that this is exactly how many Americans still think of economists—as self-appointed defenders of the free market, spinning theories to show that greed is good. Watching those old Milton Friedman videos, I wonder if that picture might have been accurate in the 1960s and 1970s. But some big things have changed in the field of economics, and America should know about them. Three big changes stand out in particular: Econ today is more data-driven, far less politically conservative, and in general much more like engineering than it used to be.
Read the whole thing here! The title of the article was not the wording I would have used - it says economists are now "just" engineers, but I actually think that successful engineering is a much tougher, more rewarding, and more admirable activity than political-philosophizing-via-theory...


Update

Actually, one of the "trends" I mentioned might not be a recent change. It looks like economists' policy views in 1979 were just about as "liberal" as today, on average (though the two data sets are not directly comparable). So the change in terms of public intellectuals (Friedman then, Krugman now) might not reflect a change in policy views in academia. But the important fact remains that economists lean to the left now, which is not in keeping with the popular stereotype...

David Henderson - who was alive to see all this happen, unlike me - points out that the "old econ culture" I attributed to the 60s and 70s was really not so prevalent in the 60s. If so, I got my decades slightly wrong...

57 comments:

  1. All Star11:07 AM

    What economics does is it dehumanizes people by taking away their intuition. Economics is rife with assertions contrary to what seems intuitively right or correct. Once the trust of the people is gained by the economist, the people no longer can think for themselves and the economist is free to manipulate consensus any which way.

    ReplyDelete
    Replies
    1. Anonymous2:10 PM

      "assertions contrary to what seems intuitively right or correct"

      You mean, like the entirety of your comment?

      Delete
    2. Anonymous3:38 PM

      Cheap shot Anonymous. Actually the point was a good one about political apathy and unquestioned allegiance to orthodoxy. People should have been questioning what economists including those in the IMF were saying during the time of the Great Moderation and Washington Consensus. No doubt you having something interesting or contribute? Or rather, **.

      Delete
    3. Ferdinand9:14 AM

      It seemed intuitively right that the Sun revolved around the Earth.

      It seemed intuitively right that the Earth is flat.

      It didn't seem intuitive that humans evolved from apes.

      It didn't seem intuitive that

      Intuition is a way to handle basic daily challenges and decisions, not a tool to understand the world in all its complexity.

      The whole point of science is to do away with intuition because it's a flawed tool. It is based on limited, biased observations (what you see in your life and what informs your intuition is likely not a random sample of what the whole world looks like).

      It is rarely consistent (psychologists have long know that most people hold strongly conflicting views on all kinds of things- from politics to religion).

      And let's not pretend that a couple of dinner conversations with Uncle are enough to develop a solid grasp on multi-trillion dollar economy consisting of hundreds of millions of workers, millions of firms, and thousands of connected markets.

      So I am not sure how we can claim that intuition is somehow the right approach to issues like taxation, business regulation, trade, etc.

      Delete
    4. Ferdinand9:27 AM

      Fourth line should have read:

      "It didn't seem intuitive that time keeping and motion are relative."

      Delete
    5. Shooby7:43 PM

      The whole point of science is not to do away with intuition anymore than the point of meat is to do away with vegetables or the point of Ford is to do away with Toyota or the point of no return is to do away with the the point of it all.

      The point of science is to explain natural phenomena.

      Delete
    6. That was pointless. Economics isn't science.

      It seems unintuitive that snake oil cures warts.

      Delete
  2. @Noah: Check your image, brother.


    [One failure after a three-year run of consistently hot-diggity-awesome image-to-content matching is nothing to be ashamed of.]

    ReplyDelete
    Replies
    1. You mean the image isn't working, or makes no sense?

      Delete
    2. It initially displayed as a "this image is stolen," or some sort. Not anymore, however.

      Delete
    3. Your RSS feed is showing this in Feedly.
      http://www.foxtaleedit.com/images/nohotlink.gif

      But I get the correct image when I click through to the article.

      Delete
    4. By the way, your captchas are REALLY hard..

      Delete
  3. I don't think economics was ever the caricature of "conservative science" even if at times the public sees it as that way. The dominant textbook during the Friedman era was Samuelson's.

    Even the rise of Chicago economics (which was an outlier after all) was data driven not theory driven. The rise of monetarism was due to Friedman and Schwartz's monetary history not the theoretical work.

    ReplyDelete
    Replies
    1. Anonymous5:01 AM

      There were two Chicago Schools here. One was to some extent history and data driven - Friedman.

      The other (Lucas/Sargent/Prescott) is more engineering and theoretically driven. It is much more extreme. Largely groping around for mathematics there is around and calling it to "economics".

      Delete
    2. Well, neither Sargent or Prescott are Chicago economists. Sargent has spent no time there, while Prescott has held a visiting position for less than 2.5 years of his career. They weren't part of the Chicago school.

      Delete
    3. Anonymous3:49 AM

      True. I think the Chicago School, perhaps wrongly, refers to Monetarism I and II (you may not have been around when those terms were used). Essentially I guess it was a way of distinguishing a group of economists from the Neo-Classical synthesis. They dominated the orthodoxy from the 1980s, taking over from Friedman in the 1970s. They liked free markets - but with tough rules to protect those markets (so they actually were often in favor of government regulation even if they did not like big government), rules for money supply growth etc. Monetarism II is more dogmatic - rational expectations, even more doctrinaire on government commitment and credibility, insistence on optimization constraints and the linking of macro-economics with the theory of the the firm and households. By doing this they are delinking macro-economics from a different view - it is fundamentally not about markets, but society and the role of the state.

      Delete
    4. Anonymous4:03 AM

      To sum it up the Lucas/Sargent project was about removing the case for macro-economic stabilisation through active policy and for income distribution.

      The latter point is why they are big on micro-theory relating to the efficiency-equity tradeoff.

      Delete
    5. Anonymous4:05 AM

      Sorry that was removing the case for income redistribution through active government intervention

      Delete
  4. Anonymous1:10 PM

    The current change is actually the decreasing politicization of economics. Economists care less for politics and many purposefully attempt to avoid its rhetoric even in their non-academic writings. Popular economics is less laissez faire, but in the 70s most of it was not laissez faire and libertarian, just the slightly more convincing economics was laissez faire, its purpose in the US is no longer needed. The conversation has needed to become permanently nuanced as more knowledge is acquired.

    ReplyDelete
  5. Big data is the game in all fields, not just economics. The problem is arbitrary correlations to support advocacy using big data.

    ReplyDelete
    Replies
    1. Sherman4:18 PM

      No, big or small data is not unpolitical. Economy, real or surreal economy is increasing in the politicization, some economists think that they are free thinkers, but that is just an illusion.

      Delete
  6. This is hilarious. Econ is like engineering because mechanism design might work in very strictly defined and limited circumstances. Plus, now we don't need theory since we have data! Just ridiculous. Guess ever more economists enjoy the pretense of knowledge disease.

    Interestingly enough, the only way they know how to deploy their deep insights is through power. It would seem that if your powerful understanding of markets etc is so firm, you should not waste your time in academia but try it out on actual markets, by say, trading. Strangely enough, I don't see the eagerness anymore.

    yes, if you cannot convince them, call the police.

    ReplyDelete
    Replies
    1. Hot Shot5:43 PM

      I'm in agreement here. Its time for the government to step in and say to all these self proclaimed or otherwise credentialed economists "OK hotshot, let's see what you got."

      The regulators should chain these wretches to a Bloomberg station in an office tower leased by one of the big investment banks and they should be given a hedge fund account and instructed to trade. They can borrow as much money as they want as long as they can pay for the hedges that protect the bank from loss. The VAR, or value at risk cant not exceed the value of their pension because if they trade unsuccessfully, they're going to lose it.

      This is the type of babtism in fire I prefer for the top notch economists.

      Thanks for listening.

      Delete
    2. Andrew7:04 PM

      Yeah, and all physicists should all have to build a rocket and send a man to the moon! Oh wait, that just sounds silly.

      Delete
    3. Anonymous8:26 PM

      Jeesh, some people really don't have any idea what economists do. Krzys and Hot Shot would do the world a big favor if they sterilized themselves.

      Delete
    4. Anonymous5:04 AM

      The tenure and pay of economics professors should be tied to their forecasting record.

      Delete
    5. How is it any different than statistical based methods in physics or engineering? Does anyone solve Navier-Stokes equations when modeling any large system? Plus the answers given are far from definite.

      We take for granted weather forecasts out to about 5 days but beyond that are just general probabilistic models that say it will be hotter than normal or wetter than normal.

      It isn't about the science as much as it is about the models. Economists need to rigorously verify their models on a continuous basis so that they conform to reality in more than superficial moment matching. This is starting to happen but economics shouldn't be condemned because it takes longer or is more difficult to determine relationships compared to sciences that have hundreds of years of a head start.

      Delete
    6. Of course, economics (and economists) shouldn't be condemned. It deals with a very hard set of problems with complex systems and limited data sets. Consequently resolving noise is very hard or, in many cases, simply impossible.

      However, this makes it very different from Engineering. It's more like a clinical science (like say medicine). The danger is the false confidence in the quality of your information and consequently the quality of decisions you can make. More specifically, how (non)robust the optimizations you want to perform are.

      Delete
  7. "Modern Portfolio Theory has helped the finance industry move away from expensive human stock-pickers toward “passive” allocation of capital"

    Portfolio Theory however has also fooled thousands of investment managers to believe they can fully cover a part of their risk by investing in 30 or so stocks. It has also made them believe their whole risk can be quantified in those Greek letters (betas and sigmas, β,σ) and their performance can be measured in terms of a Greek letter too (alpha - α)!

    And no PT is not a data based model - it is a theory in the strictest sense. Data just don't agree with this theory - no normality, no independent returns, no constant volatility, no linearity etc

    ReplyDelete
    Replies
    1. Portfolio Theory however has also fooled thousands of investment managers to believe they can fully cover a part of their risk by investing in 30 or so stocks.

      What does it mean to "fully cover a part" of something?

      It has also made them believe their whole risk can be quantified in those Greek letters (betas and sigmas, β,σ) and their performance can be measured in terms of a Greek letter too (alpha - α)

      That's not ideal but it's a heck of an improvement over what was being done before!

      And no PT is not a data based model - it is a theory in the strictest sense. Data just don't agree with this theory - no normality, no independent returns, no constant volatility, no linearity etc

      There have been improvements on all those fronts. Come out to Stony Brook and we'll show you! ;-)

      But I agree that MPT is not the ultimate solution to investing.

      Delete
    2. This comment has been removed by the author.

      Delete
    3. "What does it mean to "fully cover a part" of something?"

      As you already know according to the MPT the total portfolio risk can split into two categories - the unsystematic and the systematic risk. Hence the my expression "fully cover a part" is identical to "fully cover the unsystematic risk". The latter is nothing more that an fictitious entity which we assume we can fully control! Sorry it seems I should have been a bit more explicit.


      "That's not ideal but it's a heck of an improvement over what was being done before!"

      I am not so sure the improvement is that big. MPT is a naive quantification of the very old proverb "don't put all your eggs in one basket". And by attempting to do so it doesn’t only make the wrong assumptions but it also gives the wrong answers to some very crucial questions (how/when to invest and how to assess and manage the associated risk!).

      What I mean is that we can’t assess the total acidity of an unknown liquid substance by measuring its volume. And that despite the fact that there is a positive correlation between the acidity of an *unknown* substance and its volume! Using the wrong measure could be worse than using no measure at all (but just some robust heuristics)!



      "There have been improvements on all those fronts.”

      Yes I know there have been many improvements (e.g. EVT) but those are to a big extend unrelated to the MPT. The MPT implies the only risk you can eliminate (and that completely!) is the so called unsystematic and that can be done just by mimicking the market portfolio. This is completely wrong. The MPT’s sole risk measure (linear correlations) in many cases is spurious, meaningless and/or can not even be defined (that depending on underlying stochastic process of the asset returns)! Furthermore timing plays a huge role as the returns have got long memory. Big gains and big losses cluster in time!


      "Come out to Stony Brook and we'll show you! ;-)”

      Sure why not. Next time I am in NY I may come and visit you ;)

      Delete
    4. Next time I am in NY I may come and visit you ;)

      Please do! You should meet the people from our QF program in the AMS department, or come to a seminar or something.

      Delete
  8. Bad economics today also seems to be data driven - like the infamous R&R or Alesina papers on austerity. I get the impression that iconoclasts use data to challenge everything we thought we knew - about keynesian multipliers, the minimum wage, progressive taxation.

    So maybe the reason why economists stopped being priests is that they stopped being so confident in what used to be dogma.

    ReplyDelete
    Replies
    1. Anonymous9:40 AM

      I think R and R were not typical. Abstraction and conformance to empirically dubiou microfoundations has been far more typical.

      As for R and R, they went for big data, sure, but what their analysis lacked was historical context - a bigger problem than theoretical underpinning.

      Delete
  9. Anonymous9:37 AM

    A lot of people seem to see the problem as big data. But surely the problem is the opposite - excessive abstraction and a desire to defend theories irrespective of whether they have any empirical foundation at all?

    ReplyDelete
  10. Big Data is the solution, more importantly social data. I've written a reply to @Noahpinion regarding the use of social data in economics http://www.mblast.com/blog/how-to-use/social-media-and-economics/

    ReplyDelete
  11. Big Data is only as useful as the models you use to evaluate the data relationships.

    ReplyDelete
    Replies
    1. Anonymous3:31 AM

      Big Data is only useful when you understand what the data means. Ie the history and context behind it.

      Delete
  12. Noah, What makes you so certain that Friedmanite approach was needed in the 70's? The debate, for instance, over exactly what happened with inflation in the 70's does not seem to have a definitive answer - and the change in direction made in particularly macro seem to have as much to do with the rising political and cultural authority of Big Money as anything else) You could be more specific (and you might be falling into someone's ideological force field without realizing it).

    ReplyDelete
  13. 1. In 1968, Milton Friedman was hardly in the mainstream of the profession. The University of Chicago housed a group of renegades, and the power in the economics profession was concentrated on the east coast - e.g. Samuleson, Solow, and Tobin were big guys in the profession, and they were (are) all liberals.

    2. Macro in 1968 was all about IS/LM, which had been popularized by Samuelson (see above) in his textbook. That was mainly a story about how to "engineer" the economy.

    3. I wouldn't say structural estimation is the "hot new technique." My impression is that the majority of empirical researchers don't like it.

    You're correct in pointing out that auction theory and matching theory have been big successes for economics. Otherwise, you're off track on the trends I think.

    ReplyDelete
    Replies
    1. In 1968, Milton Friedman was hardly in the mainstream of the profession.

      Really? When did he become such a respected guy? As long as I've been aware of the outside world (i.e. since the late 80s or early 90s), it seems like he's been considered THE economist of economists.

      Macro in 1968 was all about IS/LM

      Well, yeah. I was trying to talk less about macro specifically and more about all of econ in general. Macro might be "out of phase" with the rest. But in any case, don't you think the stuff that was dominant in the 00s (NK models) and the stuff that's hot right now (institutional finance-macro stuff) is much less "perfect-free-market" than the stuff that was big in the 80s?

      wouldn't say structural estimation is the "hot new technique." My impression is that the majority of empirical researchers don't like it.

      Seriously?? At Michigan it was all the rage, and those are pretty much all empirical people. And people like Card who I've seen give talks on "the state of empirical econ" or whatever were all very big on structural estimation. I think you might be wrong about this.

      Otherwise, you're off track on the trends I think.

      So of the other two trends - more empirics and less libertarianism - which do you think is wrong? Are you not persuaded by Hammermesh's study of publishing trends and Klein's study of political/policy leanings?

      Delete
    2. In 1967 he delivered his address as the president of AEA
      http://www.aeaweb.org/honors_awards/officerspast.php

      Speech: http://www.aeaweb.org/aer/top20/58.1.1-17.pdf

      I was under the impression that it is a pretty big honor and Tobin and Solow became presidents after him. He was also awarded the John Bates Clark medal in 1951.

      Delete
    3. Noah and Stephen, have either of you seen this?:

      http://informationtransfereconomics.blogspot.com/2014/05/blowing-anti-neo-fisherite-model-out-of.html

      Delete
    4. Dimitar: That fits more with my impression that Friedman was always a really well-respected guy.

      Delete
    5. Your "impression?" Doesn't sound very scientific.

      Delete
    6. How can you scientifically prove that Friedman was a rebel-maverick in 1968?

      He got published, awarded, honored and listened to in central banks (like in Germany). What more would he need in respect? Getting a a yearly triumph in the streets of Rome?

      Delete
    7. Your "impression?" Doesn't sound very scientific.

      That's true, but president of AEA seems fairly indicative of non-renegade status.

      Delete
    8. " I wouldn't say structural estimation is the "hot new technique." My impression is that the majority of empirical researchers don't like it."

      Even people doing CS do like it! See for example the influential work of Judea Pearl (Turing Award 2011).

      Delete
    9. Part of your narrative appears to be that Friedman is representative of mainstream views at the time. He may have been prominent, but the seat of power in those days was not in Chicago. Your pappy's representative economist was not Milton Friedman - more like Samuelson/Tobin/Solow.

      Delete
  14. I finished grad school in 1973 (finishing my dissertation took a little longer). So let's look at what macro data looked like then. At the end of 1983, we had 26 (quarterly) observations for national income accounts, and our output measure was GDP, not GNP. We had 15 years (!) of monthly data on money supply variables. We had 26 years of monthly labor force data. So trying to do anything particularly meaningful in terms of modeling was, well, difficult.

    And our ability to process data was equally limited. The first stat program I used (SPSS for main-frame) was slow and could not handle very large data sets. (The amount of data that I can handle in Excel now is vastly larger than I could have handled in SPSS in 1973.) Even if we had been able to imagine something like the data sets that Google has developed (and Hal Varian is helping to exploit), we couldn't have done anything with them.

    The developments in software and hardware proceeded in tandem--we had to have both developments; either one was useless without the other. And the incentive to develop and use large data sets arose out of--and helped stimulate--those developments.

    ReplyDelete
    Replies
    1. That's "...at the end of 1973..."...

      Delete
    2. Has the availability of data improved since then? You can sure process the heck out of it now, but is there much more data than before?

      Delete
    3. Anonymous6:03 PM

      I think Dimitar, it was always there, especially foreign country data, but before it was just much more difficult to find.

      Delete
  15. Three thoughts:

    Think of the title as a piece of subtle irony.

    Or would you have felt better if the copy editor had gone in the opposite direction?
    “Economists are Idiots!..." And as recently as the mid-1980s, he was right.

    Did the two disciplines pass each other in the night? (And here I have string theory in mind.)

    ReplyDelete
  16. This is because most of the free market economics types followed the money into the business schools...mainstream finance is built solidly on a foundation of market solutions and lack of trust for regulation, even in situations rife with asymmetric information. That left only the hippies in economics departments...

    ReplyDelete
  17. Anonymous2:45 AM

    Levitt shows a poor understanding about how social health systems work. But what I dislike more is when I see a lot of ignorance about how a system, or economy, or society actually works, and then an argument is presented in algebra or geometry, obeys optimisation conditions and somehow it is considered informed and authoritative.

    In this way perhaps Levitt is not as bad as some others?

    A lot of people get tired of economists going on about "their model".

    What we want is informed analysis.

    ReplyDelete