transcript
- 1. Confessions of a Theorist: Why Reality is Overrated1 Bart
Lipman Boston University March 28, 2013 1 Very loosely based on
Dekel and Lipman, How (Not) to Do DecisionTheory (2010).
- 2. Quotes about Economic Theory:
- 3. Quotes about Economic Theory: Pure theorists sometimes . . .
prefer a theory that is so pure as to be uncontaminated with any
material content. Joan Robinson
- 4. Quotes about Economic Theory: Pure theorists sometimes . . .
prefer a theory that is so pure as to be uncontaminated with any
material content. Joan Robinson Page after page of professional
economic journals are lled with mathematical formulas leading the
reader from sets of more or less plausible but entirely arbitrary
assumptions to precisely stated but irrelevant theoretical
conclusions. Wassily Leontief
- 5. Quotes about Economic Theory: Pure theorists sometimes . . .
prefer a theory that is so pure as to be uncontaminated with any
material content. Joan Robinson Page after page of professional
economic journals are lled with mathematical formulas leading the
reader from sets of more or less plausible but entirely arbitrary
assumptions to precisely stated but irrelevant theoretical
conclusions. Wassily Leontief An economist is someone who sees
something that works in practice and wonders if it would work in
theory. Ronald Reagan
- 6. Ironically, the only one of the three who gets it right is
the onewhos not an economist.Reagans line is correct about how
theory (often) works and how itshould work.
- 7. Ironically, the only one of the three who gets it right is
the onewhos not an economist.Reagans line is correct about how
theory (often) works and how itshould work.In my view, economic
theory is not about guring out what factsare true in the world.
Thats what empirical work is for.
- 8. Ironically, the only one of the three who gets it right is
the onewhos not an economist.Reagans line is correct about how
theory (often) works and how itshould work.In my view, economic
theory is not about guring out what factsare true in the world.
Thats what empirical work is for.When empiricists nd a new fact, it
might be an interesting fact tothink about from a theoretical point
of view.
- 9. Ironically, the only one of the three who gets it right is
the onewhos not an economist.Reagans line is correct about how
theory (often) works and how itshould work.In my view, economic
theory is not about guring out what factsare true in the world.
Thats what empirical work is for.When empiricists nd a new fact, it
might be an interesting fact tothink about from a theoretical point
of view.The relationship between economic theory and reality is
tenuous.
- 10. Example: No sensible person has a utility function.
Proof:
- 11. Example: No sensible person has a utility function.
Proof:
- 12. Example: No sensible person has a utility function.
Proof:
- 13. Example: No sensible person has a utility function.
Proof:
- 14. Example: No sensible person has a utility function.
Proof:
- 15. Example: No sensible person has a utility function.
Proof:
- 16. Example: No sensible person has a utility function. Proof:
etc., until
- 17. Example: No sensible person has a utility function. Proof:
etc., until
- 18. Incorrect Responses:1. From economic theory, we know that
people have utilityfunctions. Hence such behavior cannot
occur.
- 19. Incorrect Responses:1. From economic theory, we know that
people have utilityfunctions. Hence such behavior cannot occur. In
mathematics, names are free. It is perfectly allowable to call a
selfadjoint operator an elephant and a spectral resolution a trunk.
One can then prove a theorem, whereby all elephants have trunks.
What is not allowable is to pretend that this result has anything
to do with certain large gray animals. Hector Sussman
- 20. 2. Economists believe that people have utility functions
and henceeconomists are idiots.
- 21. 2. Economists believe that people have utility functions
and henceeconomists are idiots.Correct Response: People dont have
utility functions, but if itsa useful hypothesis, go for it.
- 22. So how can false theories be useful?
- 23. So how can false theories be useful? [A] theory is not like
an airline or bus timetable. We are not interested simply in the
accuracy of its predictions. A theory also serves as a base for
thinking. It helps us to understand what is going on by enabling us
to organize our thoughts. Faced with a choice between a theory
which predicts well but gives us little insight . . . and one which
gives us this insight but predicts badly, I would choose the latter
. . . Ronald Coase
- 24. General principles are unlikely to be accurate but may be
morecomprehensible and hence more useful than detailed,
accuratespecications.
- 25. General principles are unlikely to be accurate but may be
morecomprehensible and hence more useful than detailed,
accuratespecications.All understanding is generated by deciding
which aspects of realityto ignore.The world never repeats exactly,
so if everything is relevant, wecant learn from one situation about
another.
- 26. General principles are unlikely to be accurate but may be
morecomprehensible and hence more useful than detailed,
accuratespecications.All understanding is generated by deciding
which aspects of realityto ignore.The world never repeats exactly,
so if everything is relevant, wecant learn from one situation about
another. I have a full-size map of the world. At the bottom it says
1 inch = 1 inch. I hardly ever unroll it. Steven Wright
- 27. Implication: Because models are always false, we should not
besurprised when they are refuted.
- 28. Implication: Because models are always false, we should not
besurprised when they are refuted.If they arent refuted, its only
because we dont have enough gooddata to reveal the inaccuracies we
know are there!
- 29. Implication: Because models are always false, we should not
besurprised when they are refuted.If they arent refuted, its only
because we dont have enough gooddata to reveal the inaccuracies we
know are there!Experimental/empirical observations refuting the
serious parts ofthe theory should concern us, but not those that
refute the detailswe chose to ignore. Refutation per se is not
surprising.
- 30. Implication: Because models are always false, we should not
besurprised when they are refuted.If they arent refuted, its only
because we dont have enough gooddata to reveal the inaccuracies we
know are there!Experimental/empirical observations refuting the
serious parts ofthe theory should concern us, but not those that
refute the detailswe chose to ignore. Refutation per se is not
surprising.Hence the question should not be Is the theory true?
since theanswer is always no.The question should be Is the theory
useful?
- 31. Uses of Theory:First: Creating a language.
- 32. Uses of Theory:First: Creating a language.Have to name the
relevant objects and create a basic taxonomy.An important rst step,
but only a rst step.
- 33. Uses of Theory:First: Creating a language.Have to name the
relevant objects and create a basic taxonomy.An important rst step,
but only a rst step. I admit that these terms (nal utility,
marginal production, etc.) . . . repel some readers, and ll others
with the vain imagination that they have mastered dicult economics
problems, when really they have done little more than learn the
language in which parts of those problems can be expressed. . .
Alfred Marshall
- 34. Second: Checking an intuition.
- 35. Second: Checking an intuition.Intuitive reasoning is often
wrong.
- 36. Second: Checking an intuition.Intuitive reasoning is often
wrong. What I did at the time, a very economist thing to do, was to
build myself a little model to prove the point that I believed. So
I built a little intertemporal optimizing whatever and to my shock
and this is the point, of course, of doing models it actually gave
me the opposite answer. Paul Krugman
- 37. Third: Finding unexpected results which trigger new
directions andnew insights.
- 38. Third: Finding unexpected results which trigger new
directions andnew insights.Example. Earliest work on auctions
considered independentprivate values.
- 39. Third: Finding unexpected results which trigger new
directions andnew insights.Example. Earliest work on auctions
considered independentprivate values.Revenue Equivalence Theorem: A
very wide range of auctionsgive the seller same expected
revenue.
- 40. Third: Finding unexpected results which trigger new
directions andnew insights.Example. Earliest work on auctions
considered independentprivate values.Revenue Equivalence Theorem: A
very wide range of auctionsgive the seller same expected revenue.So
why are English auctions so much more common than manyauctions
which would have same revenue?
- 41. Third: Finding unexpected results which trigger new
directions andnew insights.Example. Earliest work on auctions
considered independentprivate values.Revenue Equivalence Theorem: A
very wide range of auctionsgive the seller same expected revenue.So
why are English auctions so much more common than manyauctions
which would have same revenue?This led to research on common value
auctions which havesurprisingly dierent properties.
- 42. Third: Finding unexpected results which trigger new
directions andnew insights.Example. Earliest work on auctions
considered independentprivate values.Revenue Equivalence Theorem: A
very wide range of auctionsgive the seller same expected revenue.So
why are English auctions so much more common than manyauctions
which would have same revenue?This led to research on common value
auctions which havesurprisingly dierent properties.Doesnt prove
facts about reality, but suggests interestingconnections between
nature of objects and auctions for them.
- 43. Fourth: Developing an intuition for how things might
work.
- 44. Fourth: Developing an intuition for how things might
work.Example: Knowles, Persico, and Todd, Racial Bias in
MotorVehicle Searches: Theory and Evidence, JPE, 2001.
- 45. Fourth: Developing an intuition for how things might
work.Example: Knowles, Persico, and Todd, Racial Bias in
MotorVehicle Searches: Theory and Evidence, JPE, 2001.Empirical
observation: Comparing white and black male drivers,blacks are
stopped more often but the rate at which they are foundto be
carrying drugs is the same as that of whites.
- 46. Fourth: Developing an intuition for how things might
work.Example: Knowles, Persico, and Todd, Racial Bias in
MotorVehicle Searches: Theory and Evidence, JPE, 2001.Empirical
observation: Comparing white and black male drivers,blacks are
stopped more often but the rate at which they are foundto be
carrying drugs is the same as that of whites. Stop Not drugs 0, 10
10, 0 noncriminal work wt 5, 4 wt , 5where 5 < wt < 10.
- 47. Implication: If police enjoy stopping a particular type,
that typeshould carry drugs less often and be stopped equally often
as atype with same wages.
- 48. Implication: If police enjoy stopping a particular type,
that typeshould carry drugs less often and be stopped equally often
as atype with same wages.Group that satises this property: white
females.
- 49. Implication: If police enjoy stopping a particular type,
that typeshould carry drugs less often and be stopped equally often
as atype with same wages.Group that satises this property: white
females.This doesnt prove police arent prejudiced against black
males orthat they enjoy stopping white females.This just gives an
intriguing interpretation of the data.
- 50. Fifth: Developing tools.
- 51. Fifth: Developing tools.Example: Spence, Job Market
Signaling, Quarterly Journal ofEconomics, 1974.
- 52. Fifth: Developing tools.Example: Spence, Job Market
Signaling, Quarterly Journal ofEconomics, 1974.Spence considers how
agents may signal their private information,focusing on potential
workers signaling abilities to potentialemployers.
- 53. Fifth: Developing tools.Example: Spence, Job Market
Signaling, Quarterly Journal ofEconomics, 1974.Spence considers how
agents may signal their private information,focusing on potential
workers signaling abilities to potentialemployers.Basic idea is
that higher ability students nd school easier andhence go longer;
so more schooling signals higher ability.
- 54. Fifth: Developing tools.Example: Spence, Job Market
Signaling, Quarterly Journal ofEconomics, 1974.Spence considers how
agents may signal their private information,focusing on potential
workers signaling abilities to potentialemployers.Basic idea is
that higher ability students nd school easier andhence go longer;
so more schooling signals higher ability.Spence assumes: education
has no eect on ability,
- 55. Fifth: Developing tools.Example: Spence, Job Market
Signaling, Quarterly Journal ofEconomics, 1974.Spence considers how
agents may signal their private information,focusing on potential
workers signaling abilities to potentialemployers.Basic idea is
that higher ability students nd school easier andhence go longer;
so more schooling signals higher ability.Spence assumes: education
has no eect on ability, there are nogrades,
- 56. Fifth: Developing tools.Example: Spence, Job Market
Signaling, Quarterly Journal ofEconomics, 1974.Spence considers how
agents may signal their private information,focusing on potential
workers signaling abilities to potentialemployers.Basic idea is
that higher ability students nd school easier andhence go longer;
so more schooling signals higher ability.Spence assumes: education
has no eect on ability, there are nogrades, all else equal, every
student prefers less education, etc.
- 57. Fifth: Developing tools.Example: Spence, Job Market
Signaling, Quarterly Journal ofEconomics, 1974.Spence considers how
agents may signal their private information,focusing on potential
workers signaling abilities to potentialemployers.Basic idea is
that higher ability students nd school easier andhence go longer;
so more schooling signals higher ability.Spence assumes: education
has no eect on ability, there are nogrades, all else equal, every
student prefers less education, etc.None of these things are true,
but they make it easier to see theconcept hes exploring. This helps
us understand signaling in ageneral way so that we can see how it
could work in many verydierent settings.
- 58. Example: Gul and Pesendorfer, Temptation and
SelfControl,Econometrica, 2001.
- 59. Example: Gul and Pesendorfer, Temptation and
SelfControl,Econometrica, 2001.What behavior would indicate that
choices are due to temptation?
- 60. Example: Gul and Pesendorfer, Temptation and
SelfControl,Econometrica, 2001.What behavior would indicate that
choices are due to temptation?Not enough to observe unhealthy
choices. Maybe the person isjust a regular utility maximizer who
likes fattening foods.
- 61. Example: Gul and Pesendorfer, Temptation and
SelfControl,Econometrica, 2001.What behavior would indicate that
choices are due to temptation?Not enough to observe unhealthy
choices. Maybe the person isjust a regular utility maximizer who
likes fattening foods.Even harder question: What behavior would
indicate that thechooser is subject to temptation but exerted
selfcontrol?
- 62. Gul and Pesendorfers answer: Consider the choice of
menus.
- 63. Gul and Pesendorfers answer: Consider the choice of
menus.Suppose someone with a choice between a restaurant with
onlyhealthy dishes and another with all those same dishes plus
someunhealthy dishes chooses the former.Natural interpretation: He
knew hed be tempted by the unhealthydishes.
- 64. Gul and Pesendorfers answer: Consider the choice of
menus.Suppose someone with a choice between a restaurant with
onlyhealthy dishes and another with all those same dishes plus
someunhealthy dishes chooses the former.Natural interpretation: He
knew hed be tempted by the unhealthydishes.Suppose it turns out
that the rst restaurant is closed, so he has togo to the second one
and he chooses something healthy.
- 65. Gul and Pesendorfers answer: Consider the choice of
menus.Suppose someone with a choice between a restaurant with
onlyhealthy dishes and another with all those same dishes plus
someunhealthy dishes chooses the former.Natural interpretation: He
knew hed be tempted by the unhealthydishes.Suppose it turns out
that the rst restaurant is closed, so he has togo to the second one
and he chooses something healthy.Natural interpretation: He exerted
(costly) selfcontrol.
- 66. Gul and Pesendorfers answer: Consider the choice of
menus.Suppose someone with a choice between a restaurant with
onlyhealthy dishes and another with all those same dishes plus
someunhealthy dishes chooses the former.Natural interpretation: He
knew hed be tempted by the unhealthydishes.Suppose it turns out
that the rst restaurant is closed, so he has togo to the second one
and he chooses something healthy.Natural interpretation: He exerted
(costly) selfcontrol.Cant know that this is what causes such
behavior, but gives anintriguing interpretation.
- 67. Last point: We develop these tools in the hopes that well
be ableto use them to do better in the future. We do understand
somethings better now than economists 50 years ago did.
- 68. Last point: We develop these tools in the hopes that well
be ableto use them to do better in the future. We do understand
somethings better now than economists 50 years ago did.I expect
that 50 years from now that what we do now will lookpretty
primitive.