The Prize in Economic Sciences 2020(nobelprize.org)
nobelprize.org
The Prize in Economic Sciences 2020
https://www.nobelprize.org/prizes/economic-sciences/2020/summary/
11 comments
2017, 2018, 2019, the committee has been on a tear, picking interesting, slightly-less-mainstream but still really defensible winners. E.g. Banerjee, Duflo, and Kremer (2019) are unusually (for economists) engaged in the real world. Nordhaus and Romer - both considered on the quirky end of mainstream (though Romer's textbook of course gave him total mainstream cred, look at the rest of his career). Thaler (2017) had a huge body of writing that verged on popular economics. When I was an economist, each of these really shook up the way I thought about what was possible in the field.
Milgrom and Wilson is... expected. I mean, auction theory has been very influential, but I think Roth + Shapley (2012) just recently covered more fun intellectual predecessor and successor movements to much of Milgrom and Wilson. Classic theory-heavy work that has had practical influence as well. Arguably, their emphasis on supermodularity was a deep theoretical influence on follow-on work, so I have to count that as high impact. But it does feel a little less exciting than past awardees. Maybe just because we've lived with auction theory for so long now?
Milgrom and Wilson is... expected. I mean, auction theory has been very influential, but I think Roth + Shapley (2012) just recently covered more fun intellectual predecessor and successor movements to much of Milgrom and Wilson. Classic theory-heavy work that has had practical influence as well. Arguably, their emphasis on supermodularity was a deep theoretical influence on follow-on work, so I have to count that as high impact. But it does feel a little less exciting than past awardees. Maybe just because we've lived with auction theory for so long now?
I judge laureates by their published Nobel lectures.
There’s a footnote in Heckman’s that pretty much renders Duflo’s work invalid.
There’s a footnote in Heckman’s that pretty much renders Duflo’s work invalid.
Elaborate please?
> Romer's textbook of course gave him total mainstream cred
You might be thinking of David Romer's textbook. As far as I know, Paul Romer never wrote one.
Paul Romer had tons of mainstream cred from his two classic publications. He left the profession and really didn't have anything to do with it for many years, even becoming a startup founder. Then he showed up to coin the term mathiness while criticizing his advisor, and later wrote a critique of modern macro.
You might be thinking of David Romer's textbook. As far as I know, Paul Romer never wrote one.
Paul Romer had tons of mainstream cred from his two classic publications. He left the profession and really didn't have anything to do with it for many years, even becoming a startup founder. Then he showed up to coin the term mathiness while criticizing his advisor, and later wrote a critique of modern macro.
Starting point for HN discussion: "Applicant Auctions for Internet Top-Level Domains: Resolving Conflicts Efficiently” [1] by Professor Robert Wilson.
[1] http://www.cramton.umd.edu/aa/cramton-gall-sujarittanonta-wi...
[1] http://www.cramton.umd.edu/aa/cramton-gall-sujarittanonta-wi...
> Robert Wilson developed the theory for auctions of objects with a common value – a value which is uncertain before- hand but, in the end, is the same for everyone. Examples include the future value of radio frequencies or the volume
of minerals in a particular area. Wilson showed why rational bidders tend to place bids below their own best estimate of the common value: they are worried about the winner’s curse – that is, about paying too much and losing out
> Paul Milgrom formulated a more general theory of auctions that not only allows common values, but also private values that vary from bidder to bidder. He analysed the bidding strategies in a number of well-known auction formats, demonstrating that a format will give the seller higher expected revenue when bidders learn more about each other’s estimated values during bidding.
> Paul Milgrom formulated a more general theory of auctions that not only allows common values, but also private values that vary from bidder to bidder. He analysed the bidding strategies in a number of well-known auction formats, demonstrating that a format will give the seller higher expected revenue when bidders learn more about each other’s estimated values during bidding.
Here's an interesting story from Priconomics about spectrum auctions & Milgrom and goes into the past to discuss how Milgrom created an Excel spreadsheet on multiple floppy disks to run this first "online" auction for the FCC - https://priceonomics.com/the-spectrum-auction-how-economists...
It should be noted that the auction format Milgrom created for the first FCC auction, the SMR auction, is no longer the preferred auction format for spectrum auction. The preferred format has shifted to a general clock auction (round-by-round for generic blocks) with a supplement round where the actual blocks are assigned in a Vickrey algorithm. Additionally, as computing power has increased, the "combinatorial" format has become popular in many countries (Canada, UK) where bidders can submit different package bids which are all taken into account with the winner determined by a SAT solver.
It should be noted that the auction format Milgrom created for the first FCC auction, the SMR auction, is no longer the preferred auction format for spectrum auction. The preferred format has shifted to a general clock auction (round-by-round for generic blocks) with a supplement round where the actual blocks are assigned in a Vickrey algorithm. Additionally, as computing power has increased, the "combinatorial" format has become popular in many countries (Canada, UK) where bidders can submit different package bids which are all taken into account with the winner determined by a SAT solver.
>bidders can submit different package bids which are all taken into account with the winner determined by a SAT solver
Is anyone doing this for elections?
Is anyone doing this for elections?
It doesn't work that way - the SAT solver takes all the bids and maximizes revenue for the auctioneer. What variable would you try to maximize/minimize in an election?
Auction theory is excellent at predicting how other economists also using auction theory will behave.
Apprently Milgrom didn't pick up his phone when the commitee called him. So his neighbor who happens to be Robert Wilson woke him up.
The best thing, there is a doorbell video of that happening: https://twitter.com/Stanford/status/1315631500080148480
The best thing, there is a doorbell video of that happening: https://twitter.com/Stanford/status/1315631500080148480
Proving that you can be a Nobel laureate, and still dumb enough to have a telescreen in your home.
So how do you determine the difference between the "True Value" of a product let say a Used TESLA vs the "Market Price" is it just subjective? Im working on True value app to value a used tesla but can't figure out what formula to use.
The items placed in an auction don't generally have a market price - that's the point of the auction. So, there's a market price for gold. It wouldn't make sense to hold an auction for a 1oz block of gold because the price is known to everyone participating in the auction. On the other hand, say there's a 1oz rock taken back to Earth from a Mars mission. There's no market price and therefore no way to just sell it. The auction will find the True Value by auction - in Milgrom's case, going round by round and asking every bidder if they're willing to buy the rock at X price. In next round they ask every bidder if they're willing to buy it at 1.1X. So on and so on until there's only bidder left.
To answer your question about valuing a used Tesla - a regression formula would work best, take the variables like mileage, features, color, etc. and find 30-40 used Tesla sales to create your formula.
To answer your question about valuing a used Tesla - a regression formula would work best, take the variables like mileage, features, color, etc. and find 30-40 used Tesla sales to create your formula.
THANK YOU. So your saying I only need a sample size of say 50 used teslas to make a formula? I need to apply a score to the Autopilot software. And the other thing is that Tesla keeps reducing prices, so used values are always changing based on new information.
Don't forget to report the uncertainty on the regression estimate too!
Economic "Sciences". Economics is more of an art or a social "science" than a real science, but I guess use "science" to muster credibility if they must.
What's next. The nobel committee is going to rename the literature prize to literature "sciences" because the literature prize has lost any credibility or value?
Why not just say economics? Why try to sneakily and falsely latch onto the sciences. It may work for the short term but in the long term it really detracts from the field as it tries to pretend to be what it is not. There is value in economics, but not in the sciences.
What's next. The nobel committee is going to rename the literature prize to literature "sciences" because the literature prize has lost any credibility or value?
Why not just say economics? Why try to sneakily and falsely latch onto the sciences. It may work for the short term but in the long term it really detracts from the field as it tries to pretend to be what it is not. There is value in economics, but not in the sciences.
Economics is absolutely a science. Though the article just discusses auction theory, there are complex equations beyond most of our understanding behind the theory, just like you’d find in physics or chemistry.
The presence of complex equations does not determine science-ness.
Arguably a science should feature theories with some ability to predict behaviour observed in the world, and those theories should be falsifiable.
Some offshoots of theoretical physics that generate large parameter spaces of non-testable mathematical models packed with complex equations are arguably not science (string theories...)
Arguably a science should feature theories with some ability to predict behaviour observed in the world, and those theories should be falsifiable.
Some offshoots of theoretical physics that generate large parameter spaces of non-testable mathematical models packed with complex equations are arguably not science (string theories...)
That's a contemporary definition of science deeply rooted in a specific philosopher of science (Popper).
There was nothing falsifiable about Newton or Darwin. They were predictive (Darwin to a lesser extent) and, more importantly, they fit, like a puzzle.
Falsifiability is a good rule of thumb, like Occam's razor, to tell apart obviously pathological research (this possibly includes particle and high-energy physics); but the point of science is the unity of science. If you're a believer in science, you have to believe there's true knowledge to be had (however unobtainable); this true knowledge has to be a single thing, a single puzzle to piece together.
However inefficiently and plagued by politics, economics is doing a bit of that. Not long ago we were thinking that it should be replaced in favor of psychology, which is closer to the matter of human behavior. But just look at how psychology is doing.
There was nothing falsifiable about Newton or Darwin. They were predictive (Darwin to a lesser extent) and, more importantly, they fit, like a puzzle.
Falsifiability is a good rule of thumb, like Occam's razor, to tell apart obviously pathological research (this possibly includes particle and high-energy physics); but the point of science is the unity of science. If you're a believer in science, you have to believe there's true knowledge to be had (however unobtainable); this true knowledge has to be a single thing, a single puzzle to piece together.
However inefficiently and plagued by politics, economics is doing a bit of that. Not long ago we were thinking that it should be replaced in favor of psychology, which is closer to the matter of human behavior. But just look at how psychology is doing.
auction theory is regularly tested by people running auctions with billions of dollars at stake.
of course the assumptions of the theory are never perfectly met in the real world, and things can get quite messy. but for instance, the theory predicts that certain kinds of auctions should fail quite badly, which has happened; and it gives prescriptions for better types of auctions, which have resulted in more revenue.
of course the assumptions of the theory are never perfectly met in the real world, and things can get quite messy. but for instance, the theory predicts that certain kinds of auctions should fail quite badly, which has happened; and it gives prescriptions for better types of auctions, which have resulted in more revenue.
I'd argue that Economics is not a science because it has no models that can predict anything. Economists today are just modern day Oracles/Seers that use their equations/models in a similar manner to how a Seer would throw chicken bones to foretell the future. Because of these Economists seem to break up into factions making it more similar to religious belief IMO. You even see many Economists ignore evidence that contradicts their pet theory.
I think it's dangerous to take their ideas too seriously.
I think it's dangerous to take their ideas too seriously.
I'm not sure where you're coming from. Economics does indeed produce models that are predictive.
Tulip prices can only go up.
See also 1929, 2008, etc etc.
See also 1929, 2008, etc etc.
The link explains some practical applications of Milgrom's and Wilson's work on auction theory. Mechanism design and auction theory are definitely predictive and useful in the real world.
The current prize disproves your point.
This seems to conflate predictability with determinism.
The value of stock options hews very close to what Black-Scholes theory would assume. There's one loose bone: volatility. But reasonable estimates of volatility are close enough for any real applications of derivatives (i.e. hedging, not speculation).
The uncovered parity equation of exchange rates has one loose bone: sovereign risk premia. If Cristina Kirschner dies and her successor starts privatizing foreign-owned capital assets, well, there goes the model. And to a smaller extent this is happening all the time. But if you had to take a decision on how local interest rates affect exchange rates, this is an excellent guide.
If anything, economics is insufficiently mathematical in the treatment of what it can't determine. It's a reasonable mean field theory but should be more Bayesian about what it takes as assumption (markets will open tomorrow -- fruit and stock alike...). The assumptions are mostly fine but we should, as a general public, have a clearer picture of how fine.
Then, how well is epidemiology doing this? 99% confidence intervals from March Covid models should encompass the possibility that there's almost no fomite transmission but the virus remains alive in airborne aerosols for longer distances and time. We might all have gone the Swede route, where true agglomerations were instantly shut down, or at least prevented having the WHO and the Surgeon General telling us not to wear masks.
The value of stock options hews very close to what Black-Scholes theory would assume. There's one loose bone: volatility. But reasonable estimates of volatility are close enough for any real applications of derivatives (i.e. hedging, not speculation).
The uncovered parity equation of exchange rates has one loose bone: sovereign risk premia. If Cristina Kirschner dies and her successor starts privatizing foreign-owned capital assets, well, there goes the model. And to a smaller extent this is happening all the time. But if you had to take a decision on how local interest rates affect exchange rates, this is an excellent guide.
If anything, economics is insufficiently mathematical in the treatment of what it can't determine. It's a reasonable mean field theory but should be more Bayesian about what it takes as assumption (markets will open tomorrow -- fruit and stock alike...). The assumptions are mostly fine but we should, as a general public, have a clearer picture of how fine.
Then, how well is epidemiology doing this? 99% confidence intervals from March Covid models should encompass the possibility that there's almost no fomite transmission but the virus remains alive in airborne aerosols for longer distances and time. We might all have gone the Swede route, where true agglomerations were instantly shut down, or at least prevented having the WHO and the Surgeon General telling us not to wear masks.
If science is about having equations and maths in it, then wait until you hear about astrology.
> Economics is absolutely a science.
It absolutely is not. It's is a societal study based on artificial rules created by man. It's no more a "science" than a study of monopoly or minecraft. After all, the laws/rules of monpoly or minecraft can be changed since they are man-made and not natural.
> there are complex equations beyond most of our understanding behind the theory, just like you’d find in physics or chemistry.
Except that in physics or chemistry, the complex equations are testable in nature and hence reproducible and testable. Nothing in economics is objectively testable. The better comparison would be to astrology or christian numeralology. Where people make arbitrary rules and within those rules they use "numbers" to create associations.
It absolutely is not. It's is a societal study based on artificial rules created by man. It's no more a "science" than a study of monopoly or minecraft. After all, the laws/rules of monpoly or minecraft can be changed since they are man-made and not natural.
> there are complex equations beyond most of our understanding behind the theory, just like you’d find in physics or chemistry.
Except that in physics or chemistry, the complex equations are testable in nature and hence reproducible and testable. Nothing in economics is objectively testable. The better comparison would be to astrology or christian numeralology. Where people make arbitrary rules and within those rules they use "numbers" to create associations.
Also, this isn't the real Nobel Prize, it's just "The Nobel Memorial Prize in Economic Sciences, officially the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel"
i thought there wasn’t an economic nobel prize?
The Prize in Economics is not one of the Nobel Prizes which were endowed by Alfred Nobel in his will. However, the nomination process, selection criteria, and awards presentation of the Prize in Economic Sciences are performed in a manner similar to that of the original Nobel Prizes.
Well, there's a prize. It's in memory of Nobel, and it's awarded based on the same rules as the other prizes. But it wasn't actually established by Nobel. So, is that a Nobel prize or not? Technically, no... but everybody doesn't want to be technical all the time. And the full name of the prize does have "Nobel" in it. So people call it "the Nobel in Economics" for short. And then other people get all huffy, because technically that isn't quite accurate.
The full title is
"The Sveriges Riksbank prize in Economic Sciences in memory of Alfred Nobel"
which tells you all you need to know about it.
"The Sveriges Riksbank prize in Economic Sciences in memory of Alfred Nobel"
which tells you all you need to know about it.
Also, there is no price in math, which might add to the confusion as well.
Mandatory response to this mandatory comment to every discussion on this topic:
https://dilbert.com/strip/2015-04-02
https://dilbert.com/strip/2015-04-02
The title is designed to escape that perennial black hole, so let's steer clear.
Milgrom's Putting Auction Theory to Work [2] is an excellent introduction to this world.
PS: OneChronos (YC 16) [3] is working on bringing these mechanisms to capital markets. It's an incredibly challenging but rewarding problem that spans deeply theoretical and applied problems. We're always hiring top academic and engineering talent regardless of your background.
[1] https://www.goldengooseaward.org/01awardees/auction-design
[2] https://www.google.com/books/edition/Putting_Auction_Theory_...
[3] https://www.onechronos.com