Freakonomics by Stephen Levitt and Stephen Dubner

1
Freakonomicsby Stephen Levitt and Stephen Dubner

• Whats it all about, then?
• Richard Disney, University of Nottingham
• for L11200 Macroeconomics

2
The aim of this book is to explore the hidden
sideof everything (p.12)

• but from the point of view of a Harvard-trained
  Chicago economist.
• Incentives are the cornerstone of economic
  life
• The conventional wisdom is often wrong
• Dramatic effects often have distant, even
  subtle, causes
• Experts use their informational advantage to
  serve their own agenda..
• Knowing what to measure and how to measure it
  makes a complicated world much less so.

3
Overall

• Freakonomics is well-written,
  thought-provoking, sometimes funny.
• Its also glib and self-indulgent in parts
• Although promising to examine everyday life
• It only treats a sub-set of economic problems
  namely how individual agents respond to
  incentives in rather idiosyncratic settings
• Some examples..

4
Teenage gun crime in US

• Went up sharply in 1980s and early 1990s in US
  experts predicted even faster growth in 2000s.
• But crime rate actually went down in 2000s. Why?
• A Fewer teenagers around prone to taking part in
  gun crime. Why?
• A Roe v Wade case in 1970 legalised abortion in
  US
• Gun-carrying teenagers generally have teenager
  mothers, live with single parents etc. These
  women could now have abortions.
• 20 years on, fewer of these teenagers exist.

5
Cheating

• What have sumo wrestlers in Japan and Chicago
  high school teachers have in common?
• A They both have an incentive to cheat, and do
  so.
• Sumo wrestlers throw matches how?
• When result of a bout matters to 1 sumo but not
  the other, the former has an increased
  probability of winning irrespective of skill
• High school teachers cheat why?
• Getting better grades improves school performance
  and salary
• An algorithm was developed to spot teachers
  changing outcomes of multiple choice tests.
• California abandoned bonus to successful teachers
  (measured by students test scores) as there was
  so much cheating.
• Question (LD) not why is there so much crime
  in modern society but why isnt there a lot more
  crime? (p.17)

6
Does how you bring up your children matter for
their school performance?

• No!
• The following factors DONT matter
• Wife giving up work to look after children,
  parents being married, reading to children,
  taking them to museums, stopping them watching
  television, forbidding corporal punishment.
• The following factors DO matter
• Age, education and socio-economic status of
  parents, English spoken in home (good), low
  birthweight (bad), adoption (bad influence of
  natural parents outweighs that of adopting
  parents?)
• For parents, its who you are not what you do.
• And the main influence on school performance is
  effect of bullying, loitering, absence of PTA,
  NOT number of computers, pupil-teacher ratios
  etc. (peer group effects are crucial)

7
Childrens names I

• Afro-Americans in US give their children
  (especially girls) different names from rest of
  population.
• Top ten girls names among whites in California
  in 1990s
• Molly, Amy, Claire, Emily, Katie, Madeline,
  Katelyn, Emma, Abigail, Carly
• And among blacks
• Imani, Ebony, Shanice, Aaliyah, Precious, Nia,
  Deja, Diamond, Asia, Aliyah
• So this allows easy discrimination e.g in CVs for
  job applications (where, obviously, race not
  recorded)

8
Childrens names II

• (This is work joint work with economist Roland
  Fryer)
• The parent who named his children Winner and
  Loser but Loser ended up a Sgt in Police while
  Winner ended in gaol.
• Suppose Afro-Americans gave their children
  white names, would it make much difference to
  their economic outcomes?
• Not really, once you control for background
  characteristics (see what makes a good parent)
  
• Name fashions change over time and migrate down
  the socio-economic ladder.
• E.g Britney/Brittany has gone from middle to
  lower class white name.

9
Comments

• Lets face it, economics has not earned the title
  the dismal science for nothing!
• Most economics research is pretty boring.
• Levitt uses large scale data sets from
  households, administrative records etc, to
  examine interesting subjects.
• Some subjects e.g. family economics have
  developed as a result of his work (and influenced
  by his mentor at Chicago Gary Becker).

10
Multiple or single explanations?

• Most economic explanations have multiple causes
• So econometricians use multiple regression which
  tries to identify the marginal effect of
  different explanatory variables (as does Levitt)
• Levitt is not always consistent in how he looks
  at multiple causes
• With childrens names, he wants to argue there
  are multiple reasons for the association between
  names and economic outcomes
• With teenage crime, he wants to emphasise one
  association in the data (abortion law change)
  when there are clearly many causes.

11
Statistical correlations

• Everyone knows (including Levitt!) that
  correlation need not mean causation.
• In any large scale data set, there will be
  correlations a (high?) proportion will be
  associations with no direct intuition.
• Some of these correlations are just spurious and
  not derived from any underlying structure
• When many correlations, and also two way
  causation, econometricians look for instruments
• Something that correlates with the regressor but
  is not subject to 2-way causation (e.g. example
  of abortion law reform 20 years earlier)

12
A potential risk with this approach is.

• To get your paper published, or cited (e.g. by
  Levitt)
• You look for weird and unusual instruments
  variables that correlate with an X regressor but
  can be treated as exogenous (to Y)
• But if the instruments dont have any intuitive
  sense (from the point of view of inductive
  economics), what have we gained?
• We are no nearer understanding economic
  structure or more difficult structural
  economic problems. Some critics have labeled
  Levitts work cute-o-nomics
• That is find a problem thats small enough or
  sexy enough to get an answer, but it doesnt
  answer a big question.
• Much the same critique can be levelled at some
  experimental economics that looks for (alleged)
  refutations of orthodox theory by devising cute
  experiments.

13
Some references

• Steven D Levitt and Stephen J Dubner (2006)
  Freakonomics, (Revised edition) Penguin Books.
• Critiques
• (Heavy) John DiNardo (2007) Interesting
  questions in Freakonomics Journal of Economic
  Literature, 45, December.
• (Light) Noam Scheiber (2007) How Freakonomics is
  ruining the dismal science Freaks and Geeks, The
  New Republic Online (posted 21.03.07)