Behavioral%20economics%20field%20evidence%20(Della%20Vigna%20JEL%20in%20press) - PowerPoint PPT Presentation

View by Category
About This Presentation



Behavioral economics field evidence (Della Vigna JEL in press) A boom in clever field studies showing impact of psychology on economic behavior – PowerPoint PPT presentation

Number of Views:90
Avg rating:3.0/5.0
Slides: 33
Provided by: Colin262


Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Behavioral%20economics%20field%20evidence%20(Della%20Vigna%20JEL%20in%20press)

Behavioral economics field evidence (Della Vigna
JEL in press)
  • A boom in clever field studies showing impact of
    psychology on economic behavior
  • Largely fueled by experimental data ? tells you
    what to look for
  • Wide variety of topics methods
  • High impact (citations, buzz) and easier to
    publish than experiments

  • Non-standard preferences
  • Self control, reference-dependence, social
  • Non-standard beliefs
  • Overconfidence, non-Bayesian, projection bias
  • Non-standard decision making
  • Limited attention, menu effects, social pressure,

Non-standard preferences
  • Self-control
  • Exercise People overpay for annual health club
  • Deadlines
  • Imposed equally-spaced deadlines for proofreading
    improved performance (135 vs 70)
  • Endogenous choice? Majority picked deadlinesbut
    bunched too close to end of the term
  • Credit and savings. Many examples

Non-standard preferences
  • Reference-dependence
  • Organ donation Countries with opt-out of organ
    donation have much higher rates than opt-in
  • Huge effect of default into savings (SMaRT plan
    06 Automatic Pension and Savings Protection Act
  • SMaRT plan (Benartzi-Thaler 04 JPE) exploits two
    elements of human nature
  • Tendency to commit and not switch
  • Dislike nominal take-home pay falling
  • ? commit workers to put 1/3 of future pay
    increase into 401(k)
  • Raises savings strongly
  • The major practical advance from behavioral
    economics so far

Reference-dependence in labor supply
  • Basic questions
  • Does supply rise with wage w?
  • Participation (days worked) vs hours
  • A Very low supply elasticities for males
  • but most data from fixed-hours
  • Intertemporal substitution
  • Do workers work long hours during temporary wage
    increases (e.g. Alaska oil pipeline)? (Mulligan
    JPE 98?)
  • Participate on high-wage days (Oettinger JPE
    baseball stadium vendors)
  • Alternative Amateur income targeting

Cab driver income targeting (Camerer et al QJE
Cab driver instrumental variables (IV) showing
experience effect
Possibility of Poisoning Is Raised In the Death
of a Haitian Colonel ... Is Raised In the Death
of a Haitian Colonel ... November 8, 1988 -
AP (NYT) - International - News - 431 words
  • 27
  • 36
  • 43

    Critic's Notebook Tokyo, City of the 12
Movie ... , City of the 12 Movie ...View free
preview November 8, 1988 - By VINCENT CANBY,
Special to the New York Times (NYT) - Movies -
News - 1250 words
Save the Catskills Also ... Save the Catskills
Also ... November 8, 1988 - (NYT) - Editorials
and Op-Ed - Letter - 121 words
Interviews Great idea!
  • Q From passage above, do you think others did
    interviews too? Were they more or less systematic
    than Farbers? (his are admittedly not

Farber (JPE 04) hazard rate estimation Do hrs
worked or accumulated income predict quitting?
  • Note If workers are targetting, why isnt the
    income distribution more spiky?

Do they quit because of hours or ?Getting tired
is a stronger regularity than targetting
  • Note Which has more measurement error, hours or
  • Big tip experiment!

Alan Krueger 6/26/03 NYTimes column
  • Now their findings are being debated. First,
    Gerald S. Oettinger of the University of Texas at
    Austin published a paper in the Journal of
    Political Economy on the daily work decisions of
    food and beverage vendors at a major-league
    baseball stadium. The vendors were independent
    contractors, required to work until the seventh
    inning, but they could choose which games to
    work. Vendors make more when the number of fans
    is high and the number of other vendors is low.
    Professor Oettinger found that vendors were more
    likely to go to work when the expected payoff was
    higher -- for example, on days when a larger
    crowd was expected because of a pivotal game or a
    quality opponent. The decision of whether to work
    at all on a high-payoff day -- as opposed to how
    much to work -- was not considered in the
    cabdriver study. A YES IT WAS. PERHAPS KRUEGER

  • And most recently, my Princeton colleague Henry
    S. Farber revisited the question of cabdrivers,
    studying a different set of drivers. He found
    that cabdrivers quit after they work a lot of
    hours and grow weary. How much they have earned
    to that point has little or nothing to do with
    their decision. Moreover, the amount the drivers
    earn varies substantially from day to day,
    suggesting that their target income levels, if
    they have them, fluctuate wildly. He suggests
    that the earlier findings possibly resulted from
    reporting errors in the data because daily wages
    were derived by dividing total revenue by hours
    worked, any mistake in reported hours would cause
    a mistake in the opposite direction in the
    calculated wage, inducing a negative correlation
    between wages and hours worked.

Big tip experiment
  • Prediction of reference-dependent model
  • A large windfall will lead to lower labor supply
  • Example Give drivers a big surprising tip
  • predict they will quit early (or street
    musicians etc)
  • Tip must not be an indication of a shift in wages
  • Do it? Only if it would convince true-believer
    labor economists
  • Thaler asked Kevin Murphy
  • They might just go home to celebrate
  • Implication Some labor economists will not
    commit to reputational bets about whether
    theories are true

Non-standard preferences
  • Social preferences
  • Charitable giving in the field
  • Falk (04) Do people give more after small or
    large gift? (postcards)
  • Control 12.2, small 14.4, large 20.6
  • Does gift exchange for high wage wear off?
  • Yes Gneezy-List 07 (tiny n, absent in one
  • No Kube 06, offer 15 Euro/hr
  • Low group (paid 10) do 25 less, does not decline
    w/ time
  • High group (paid 20) do 5 more, does not decline
    w/ time

Field and lab experiments
  • Does lab generalize to field?
  • Gary Becker (2002 interview) Economists have a
    theory of behavior in markets, not in labs, and
    the relevant theories can be very different.
    (italics mine)
  • Labs and markets are not very different in
    principle (stereotyping error to think they are)
  • ? For every market M1, can find another market M2
    which is more different than a lab experiment L1
  • Focus should be on generalizing from empirical
    findings to a particular domain of interest, not
    from lab to field
  • Example Interested in Las Vegas slot machine
  • Lab experiment generalizes well
  • Field data from betting on cockfighting in
    Phillipines may not generalize well.
  • External validity is a misleading phrase because
    the external world varies

Behavioralist meets the market (List JPE 06 )
  • Gift exchange
  • Offer price W, supply quality q
  • Dealers W c(q) buyers v(q)-p
  • q 1 2 3 4 5
  • c(q) 4 5 8 15 50
  • v(q) 6 8 15 30 80
  • Twist
  • Use sports card dealers in lab and field
  • Field Approach, offer 20 or 65, request high
    quality (PSA 9) or top quality (PSA 10)

John List JPE
Basic lab replication is weak (t1.80, .05ltplt.10
Effect of quality on price Lab (I) to field
(III, IV-P) generalizes well
but driven by local (L) dealers
Levitt-List (JEcPers 07) articulated concerns
about stereotypical lab experiments
  • 1. Scrutiny
  • 2. Context in which decision is embedded
  • 3. Self-selection and experience
  • 1. is a very minor problem in most experiments
  • Largest in dictator games, a weak situation
  • Scrutiny is no greater in lab (web) than field (
    is a consequence of IRB informed consent)
  • 2-3 are strong arguments for lab experiments
    because of extraordinary control over subtle fx

Levitt-List conclusion re List 04
  • Problems
  • The could not have card graded condition is not
    present in the lab experiments
  • The local/nonlocal comparison is not reported
    for lab experiments
  • ? There is no clean comparison between lab and
  • In fact, Levitt-List have zero examples of an
    unconfounded lab-field failure of generalization

Non-standard beliefs
  • Overconfidence
  • Malmendier-Tate 04,05 Overconfident CEOs hold
    options to expiration
  • 55 more likely to do merger, also invest more
    heavily in response to free cash flow
  • Investors trading (Barber-Odean)
  • Trade too much (esp. men)
  • Non-Bayesian
  • Winning lottery numbers are underbet in later
    weeks (law of small numbers)
  • Hot hand biases betting on NBA teams
  • Projection bias (overforecast persistence of
  • Catalog returns for cold-weather clothes higher
    if ordering day is cold (as if buyers expect cold
    weather to persist)

Non-standard decision making
  • Limited attention
  • Attention is scarce!
  • Value of a good is vi (i is invisible)
  • Choices reveal v(1-?)i
  • ? 0 full attention ? gt 0 degree of limited
  • Hossain-Morgan 06 Ebay auctions with shipping
    costs included or excluded
  • Estimate ? .45, .18
  • Chetty 07 Do consumers include sales tax?
  • Field experiment with sales tax included or
  • Estimate ?.75
  • Ranking sensitivity
  • Hospital, college rankings are continuous (0-100)
    but often presented as ranks (e.g. ranks 3-4
    might be 93, 94)
  • Pope 07 shows sensitivity to ranks, not
    (reported) continous variable

Information vs news Attention to NYTimes story
about cancer drug moves a stock (Huberman-Regev
01 JFin)
Limited attention in finance
  • M-Th announcements of earnings surprises
    estimate ?.42 Friday announcements ?.59 (Della
    Vigna-Pollet 06 )
  • Attention to announcements lower on days when
    more announcements take place (Hirshleifer 07)
  • Stocks of supplier companies decline 1-3 mos
    later after bad earnings news from companies they
    supply (Cohen-Frazzini 06)
  • Markets forecast effects of demographics (e.g.
    baby boom ? school busses) only 5yrs ahead (Della
    Vigna-Pollet in press)
  • Investors buy 20 more often for stocks in
    highest decile on volume or price change (gets
    attention) (Barber-Odean 06)

Non-standard decision making
  • Menu effects
  • 1/n heuristic (partition-dependence)
  • Choices of retirement plans (next slide)
  • In lab exps, field exps, field data too
    (Goldman-Sachs derivative markets)
  • Choice overdose paralysis
  • Jams More sample w/ 24 jams, but fewer buy
    (Iyengar-Lepper 00)
  • Firms w/ 2 funds available, 75 participate gt40
    funds, only 65 (Iyengar 04)
  • Equilibrium implications? (How do people ever
    choose in the face of complexity?)

Horse and rabbit stew(Benartzi-Thaler 01)
  • Fund allocations ? stock/bond mixes
  • Equal-spreading heuristic can lead to too much
  • Stock-balanced fund
  • ? heavy stock weight

Non-standard decision making
  • Social pressure
  • Soccer refs add more extra-time when it
    benefits home team ( 4 mins when -1 goal, 2 mins
    when 1 goal Garicano 05)
  • Envelope stuffers Pairs stuff more (221 vs 190)
    and within-pair variance falls (Falk-Ichino 06)
  • Supermarket cashiers More productive ones
    influence less productive (1? .23) if they can
    see slow ones (Mas-Moretti 06)

Examples in political economy
  • Menu effects on ballots (top is chosen more often
    in local elections)
  • Attention High news distraction (e.g.
    Olympics) lowers USAID foreign disaster relief
    (-30) (Eisensee-Stromberg 07)
  • Role of media in supplying information E.g. Fox
    News on local cable ? more Republican votes
    (Della Vigna-Kaplan 07)

  • Many areas remain
  • Extension to political science
  • Other types of psychology
  • Emotion, willpower, implicit discrimination,