Econ 219B Psychology and Economics: Applications (Lecture 11) - Stefano DellaVigna April 8, 2020 - Berkeley Economics

Page created by Brian Saunders
 
CONTINUE READING
Econ 219B Psychology and Economics: Applications (Lecture 11) - Stefano DellaVigna April 8, 2020 - Berkeley Economics
Econ 219B
   Psychology and Economics: Applications
                (Lecture 11)

                      Stefano DellaVigna

                           April 8, 2020

Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   1 / 92
Econ 219B Psychology and Economics: Applications (Lecture 11) - Stefano DellaVigna April 8, 2020 - Berkeley Economics
Outline

 1   Choice of Dominated Options
 2   Memory
 3   Mental Accounting
 4   Persuasion
 5   Emotions: Mood
 6   Emotions: Arousal
 7   Happiness

     Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   2 / 92
Econ 219B Psychology and Economics: Applications (Lecture 11) - Stefano DellaVigna April 8, 2020 - Berkeley Economics
Choice of Dominated Options

                                       Section 1

                Choice of Dominated Options

Stefano DellaVigna              Econ 219B: Applications (Lecture 11)   April 8, 2020   3 / 92
Choice of Dominated Options   Bhargava, Loewenstein, and Sydnor (2017)

Dominated Choice

   An especially strong case of non-standard decision making is the
   choice of a dominated option
   Bhargava, Loewenstein, and Sydnor (QJE 2017)
   Examine choice of health plans for employees of a large company
   Plans are such that the high-deductible plans tend to dominate
   the low-deductible plans

   Stefano DellaVigna              Econ 219B: Applications (Lecture 11)                   April 8, 2020   4 / 92
Choice of Dominated Options   Bhargava, Loewenstein, and Sydnor (2017)

Bhargava, Loewenstein,
               CHOOSE TOSydnor
                         LOSE                                                              1333

   Stefano DellaVigna                    FIGURE
                                   Econ 219B:     I
                                              Applications (Lecture 11)                   April 8, 2020   5 / 92
Choice of Dominated Options   Bhargava, Loewenstein, and Sydnor (2017)

Bhargava, Loewenstein, Sydnor
   Large costs of picking the wrong plan
 1340               QUARTERLY JOURNAL OF ECONOMICS

   Stefano DellaVigna              Econ 219B: Applications (Lecture 11)                   April 8, 2020   6 / 92
Choice of Dominated Options   Bhargava, Loewenstein, and Sydnor (2017)

Bhargava, Loewenstein, Sydnor
   Incidence of errors is much
                   CHOOSE       larger for low-income
                           TO LOSE                 1347people

   Stefano DellaVigna              Econ 219B: Applications (Lecture 11)                   April 8, 2020   7 / 92
Choice of Dominated Options   Bhargava, Loewenstein, and Sydnor (2017)

Bhargava, Loewenstein, Sydnor
   Large costs of picking the wrong plan for the poor

   Stefano DellaVigna              Econ 219B: Applications (Lecture 11)                   April 8, 2020   8 / 92
Choice of Dominated Options   Biases and Poverty

Other papers
   Are behavioral biases disproportionately hurting the poor
   (Mullainathan and Shafir)?
   Key variables in determining implications of behavioral
   economics for redistribution and inequality
   Two forces:
           Poor are likely less educated –> More bias
           Poor have lower cost of time –> Can in principle search harder
   In literature:
           Bhargava et al. (2017): first force clearly dominates
           Busse et al. (2013): also similar results for limited attention to
           odometer, much smaller magnitude
           Madrian and Shea (2001) – default effects larger for lower
           income
           Other papers?
   Incidence of behavioral biases is key emerging theme
   Stefano DellaVigna              Econ 219B: Applications (Lecture 11)    April 8, 2020   9 / 92
Memory

                           Section 2

                             Memory

Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   10 / 92
Memory

Introduction

   Limited memory likely to underlie several behavioral phenomena
           Procrastination
           Backward-looking reference points
           Overinference and Experience effects
   Start from Ericson (JEEA 2010) – Empirical evidence on
   limited memory and naivete’
                                      $
           Individuals will receive 20 if they remember to send an email in
           a particular week, months ahead
                                  $
           Or they can can get X unconditional
           What do they prefer, and how how often do they remember?

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)   April 8, 2020   11 / 92
Memory        Ericson (2010)

Ericson (2010)
               48                                                Journal of the European Economic Association

                                        TABLE 1. Excerpt from experiment instructions.

                                                                                                                         Downloaded from https://academic.oup.com/jeea/article-abstract/9/1/43/2298405 b
               Please indicate your choices between the two rewards listed on each line by checking your
               preferred box. Rewards in both columns will be mailed at the same time, on 11 January 2006.

                                                                        Rewards in this column will be mailed to
                         Rewards in this column will be                 you if you email us with your contact
                         automatically mailed to you.                   information between Jan. 5 and Jan. 10,
                                                                        2006. You must email us within this time
               Row       No further action is required on your part.    period to receive your reward.

               #0        ____ $20                                       ____ $20
               #1        ____ $19.25                                    ____ $20
               #2        ____ $18.50                                    ____ $20
               ...       ...                                            ...
               #19       ____ $5.75                                     ____ $20
               #20       ____ $5                                        ____ $20

               0 to 20, subjects received the election they made in the row that was thus numbered.
   What do
         Theypeople
             therefore hadchoose?
                           a 21% chance of receiving the election they had made in one of the
               21 rows. With 60% probability (dice rolls from 21 to 80), subjects were assigned to
               the $20 contingent payment condition, while with 19% probability (dice rolls from 81
               to 99) they were assigned to receive an automatic payment of $20. These probabilities
               were chosen to facilitate a simple description of the tasks to subjects. All subjects then
               filled out a sheet giving their contact information. Subjects receiving the automatic
   Stefano DellaVigna                     Econ 219B: Applications (Lecture 11)                               April 8, 2020                                                                                 12 / 92
54                                                            Journal of the European Economic Association

                                                                                                                           m/jeea/article-abstract/9/1/43/2298405
                                                                                    Memory            Ericson (2010)

                      0
              TABLE 3. Relationship between beliefs and behavior: probit model and means.
                             0                .25                 .5               .75                    1
                                         Subjective probability of claim implied by choices

Ericson (2010)

                                                                                                                                        Downloaded from https://academic.oup.com/jeea/article-abstract/9/1/43/2298405
                                 Dependent Variable: Remembered to Claim Payment
                                           Coefficient Marginal Effect Coefficient                 Marginal Effect
                      F IGURE 2. Distribution of subjective probability of claim ( p̂). (Analysis Sample.)
Subjective Probability of Claim p̂ 0.873∗                0.348∗                   1.009∗∗          0.402∗∗
                               TABLE 2.(0.353)
                                            Test for population
                                                         (0.140) overconfidence.  0.364            0.145)
In MBA Sample                            –               –                        0.584            0.228 ˆ
                         N      Fraction Claimed Average Implied Claim                               t-statistic
                                                                                      Probability (0.113)
                                                                                 (0.303)
Male                                     –             –                          0.137            0.055 ˆ
  Analysis Sample 186                   0.53                               0.76(0.202)                 5.61∗∗∗
                                                                                                  (0.080)

                                                                                                                                                                  by University of California Berkeley, Stefano DellaVignaby
Age in Years                                                                   −0.0155
                                                                          (0.04)                 −0.006
                                                                                                            ∗∗∗
  . . .College Only      36             0.39                               0.82(0.027)                 4.76
                                                                                                  (0.011)
N                                        186                              (0.04) 186
  . . .MBA
Pseudo    R2 Only       150             0.56
                                         0.025                             0.75 0.045                  4.01∗∗∗
                                                                          (0.02)
∗ p < 0.05, ∗∗ p < 0.01
  . . .Men Only         114             0.54                               0.72                        3.29∗∗
Analysis Sample. Standard errors in parentheses. Marginal effects are reported
                                                                          (0.03)at the mean of the independent
  . . .Women Only
variables.  The symbol ˆ indicates
                         72        a dichotomous
                                        0.50      variable, for which the effect
                                                                           0.82of a discrete change from  0 ∗∗∗
                                                                                                       4.99 to 1
is reported.                                                              (0.02)
 ∗∗ p < 0.01, ∗∗∗ p < 0.001.
 Standard errors in parentheses.
                 1

 result being driven by a misunderstanding of the instructions. Choosing a contingent
 payment over an automatic payment of the same amount is a weakly dominated action
                 .8

 that should only be chosen if p̂ = 1 and costs of claiming and trying to claim are zero.
 If the 16 subjects in the Analysis Sample (8.6%) who make this weakly dominated
         Fraction claimed
                      .6

 choice are excluded, the results are virtually unchanged: average p̂ is 0.74, the fraction

                                                                                                                                                                                                                           onUniversity
 claimed is 0.52, t = 4.91. Finally, this result is not driven by subjects attempting to

                                                                                                                                                                                                                              08 April 2020
 claim the payment outside the specified claim window. Only five subjects attempted to
           .4

 claim payment after the claim window had passed. The analysis thus far counts them
 as having failed to complete the memory task. However, they may not have found

                                                                                                                                                                                                                                        of California Berkeley, Stefano D
 credible the experimenters’ claim that payment would be denied if they did not fulfill
                 .2

                                  45° line
                 0

                            0            .2            .4              .6           .8                1
                                       Subjective probability of claim implied by choices

F IGURE 3. Average claim behavior by subjective probability of claim ( p̂). (Analysis Sample. Marker size
           Stefano DellaVigna
indicates the number of subjects in each p̂ value group.)
                                                            Econ 219B: Applications (Lecture                         11)                                                                                                                                                    April 8, 2020   13 / 92
Memory     Haushofer (2015)

More Evidence on Memory

   Haushofer (2015) presents more evidence on limited memory
           Experiment in Kenya via SMS messages
                        $
           Transfer of 3 through mobile system MPesa
                                      $
           Subject can get extra 13 by sending message on exact date
           $13 would be sent 5 weeks later (to keep time discounting out
           of picture)
   Can estimate extent of memory decay over time (not naivete)

   Stefano DellaVigna       Econ 219B: Applications (Lecture 11)   April 8, 2020   14 / 92
Memory            Haushofer (2015)

More Evidence on Memory
              Option 1:
                                  Get $3                                      Send request                                   Get $13
                                                   Immediately Today Tomorrow 1 week          2 weeks 3 weeks 4 weeks 5 weeks
                             Call 1
                               Call 2
             Option 2:
                                  Get $3                                        No request                                    No $13
                                                   Immediately Today Tomorrow 1 week          2 weeks 3 weeks 4 weeks 5 weeks
                             Call 1
                               Call 2
                                     1
                                     .9

                                                     .94
                                                             .92
                           .3 .4 .5 .6 .7 .8
                          Proportion remembering

                                                                      .75           .75              .74

                                                                                                                                   .65

                                                                                                                       .57
                                                                            y = .86^t (t in weeks)
                                                                                                           .48
                                     .2
                                     .1

                                                                                                           N = 50 at each time horizon
                                     0

                                               Immediately Today Tomorrow 1 week 2 weeks 3 weeks 4 weeks 5 weeks
   Stefano DellaVigna                                                         Delay
                                                         Econ 219B: Applications (Lecture 11)                  April 8, 2020             15 / 92
Memory     Zimmermann (2020)

Motivated Memory

   Above we discussed natural limitations to memory
   But memory can also work selectively
   Important part of several behavioral models, e.g.,
   Benabou-Tirole

   Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   16 / 92
Memory     Zimmermann (2020)

Motivated Memory

   Zimmermann (AER 2020) experiment on selective recall
           Two experimental sessions, one month apart
           Dictator game for 10 euro with German Red Cross (filler)
           IQ test (10 Raven matrices) with earnings of 4 euro
           Belief elicitation about performance, compared to 9 people in
           group – Are you in upper half?
           (Noisy) feedback about the ranking: select 3 out of 9 member
           and say how you did relative to each of them
           In ConfidenceDirect, second belief elicitation, with quadratic
           scoring rule
           In Confidence1Month, second belief elicitation is done 1 month
           later

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)   April 8, 2020   17 / 92
Memory                   Zimmermann (2020)

Motivated Memory    346                                             THE AMERICAN ECONOMIC REVIEW                                                 FEBRUARY 2020

                    Panel A. ConfidenceDirect                                                 Panel B. Confidence1month
                                               Positive                 Negative                                      Positive                   Negative
                              0.4                                                                     0.4

                              0.3                                                                     0.3

                                                                                           Fraction
                   Fraction
                              0.2                                                                     0.2

                              0.1                                                                     0.1

                               0                                                                       0
                               50

                                                             50

                                                                                                       50

                                                                                                                                     50
                                                                                                                                     50
                                                 0

                                                            50

                                                                          0

                                                                                      50

                                                                                                                       0

                                                                                                                                                  0

                                                                                                                                                             50
                              −

                                                           −

                                                                                                      −

                                                                                                                                    −
                                                      Belief adjustment                                                        Belief adjustment

                                                                                      Figure 1

                    Notes: Histograms of belief adjustments (posterior − prior) for treatments ConfidenceDirect (panel A) and
                    Confidence1month (panel B), separately for positive and negative feedback. Belief adjustments are censored at
                    +/− 50.

                              Panel A. ConfidenceDirect                                               Panel B. Confidence1month
                                         90                                                                     90
                                         80                                                                     80
                        Pr(upperhalf )

                                         70                                                    Pr(upperhalf )   70
                                         60                                                                     60
                                         50                                                                     50
                                         40                                                                     40
                                         30                                                                     30
                                          )

                                                          )

                                                                    )

                                                                                    )

                                                                                                                 )

                                                                                                                              )

                                                                                                                                             )

                                                                                                                                                            )
                                          37

                                                      34

                                                                   48

                                                                                    59

                                                                                                                 15

                                                                                                                             23

                                                                                                                                            33

                                                                                                                                                            37
                                    =

                                                     =

                                                                  =

                                                                                =

                                                                                                         =

                                                                                                                           =

                                                                                                                                        =

                                                                                                                                                       =
                               (N

                                                  (N

                                                              (N

                                                                              (N

                                                                                                      (N

                                                                                                                        (N

                                                                                                                                        (N

                                                                                                                                                      (N
                              5

                                                  5

                                                              6

                                                                              6

                                                                                                      5

                                                                                                                        5

                                                                                                                                        6

                                                                                                                                                      6
                        <

                                                                           >

                                                                                              <

                                                                                                                                                   >
                                                   Test performance                                                         Test performance
                                                                Prior                                     Posterior positive feedback
                                                                Posterior negative feedback

                                                                                      Figure 2
   Stefano DellaVigna                                      Econ 219B: Applications (Lecture 11)                                                                   April 8, 2020   18 / 92
Memory       Zimmermann (2020)

Motivated Memory
   Additional treatments: Two Recall treatments
            First recall treatment: one month later, ask how many
            comparisons to other 3 subjects were positive
      350                      THE AMERICAN ECONOMIC REVIEW    FEBRUARY 2020
            RecallHigh: Same, but 50 euros (!) to recall feedback
                                              1
                   Average recall accuracy

                                             0.8

                                             0.6

                                             0.4

                                             0.2

                                                                                                  Recall   RecallHigh
                                              0
                                                   0                 1                        2                         3
                                                                         Positive comparisons

   Stefano DellaVigna                                  Econ 219B: Applications (Lecture 11)                       April 8, 2020   19 / 92
Memory         Zimmermann (2020)

Motivated Memory

   Additional treatments: Announcement treatment
           Announce at end of session 1 that will ask about recall in
           session 2
         358                 THE AMERICAN ECONOMIC REVIEW      FEBRUARY 2020

                                 90

                                 80

                                 70
                    Pr(upperhalf )

                                 60

                                 50

                                 40                     Prior                           Posterior positive feedback
                                                        Posterior negative feedback
                                 30
                             6 (N = 49)

                                                                Test performance

                                                      Figure 4. Announcement
   Stefano DellaVigna                      Econ 219B: Applications (Lecture 11)                                April 8, 2020   20 / 92
Mental Accounting

                              Section 3

                     Mental Accounting

Stefano DellaVigna     Econ 219B: Applications (Lecture 11)   April 8, 2020   21 / 92
Mental Accounting

Introduction

   Thaler (1981): Mental Accounting is tendency of individuals to
   form special accounts for different expenditures, and keep
   inflows and outflow separated across accounts
                        $
           Example: 200/wk food budget and 100/wk entertaiment $
           budget
   Deviates from standard model with just one budget
   Why use mental accounting?
           Self control problems
           Simplicity
   What is the evidence for this?
   Until recently, quite weak. Rare component in Thaler agenda
   without too much support

   Stefano DellaVigna        Econ 219B: Applications (Lecture 11)   April 8, 2020   22 / 92
Mental Accounting    Hastings and Shapiro (2013)

Gas Prices

   Hastings and Shapiro (QJE 2013)
           Assume a mental account for gasoline
           Choice at the pump for regular gas, or premium (usually 10c
           more expensive)
           Mental accounting: Price of gasoline goes up –> switch to
           regular gasoline (from premium) to try to stay more in account
           Notice: Proportional thinking makes opposite prediction
           Standard model:
                   Makes same prediction based on income effect, but much
                   smaller impact
                   Can also look at 2009 when price of gasoline went down

   Stefano DellaVigna         Econ 219B: Applications (Lecture 11)             April 8, 2020   23 / 92
Mental Accounting     Hastings and Shapiro (2013)

Gas Prices, Data
                        FIGURE II: Regular share and the price of regular gasoline (retailer data)

                          .82

                                                                                                  4
                                                                                                  Price of regular gasoline (dollars)
                          .81

                                                                                                                             3.5
                     Regular share

                                                                                                                     3
                            .8

                                                                                                            2.5
                  .79

                                                                                                   2
                          .78

                                                                                                  1.5
                              2006w1         2007w1              2008w1               2009w1
                                                          Week

                                             Regular share             Price of regular

  Notes: Data are from the retailer. The plot shows the weekly share of transactions that go to regular gasoline and the
     Gas price and purchase of regular gasoline clearly move together
  weekly average transaction price of regular gasoline (in current US dollars).

     Notice: Also true in 2009 when income effects go the other way
     Stefano DellaVigna                    Econ 219B: Applications (Lecture 11)                                             April 8, 2020   24 / 92
Mental Accounting     Hastings and Shapiro (2013)

Gas Prices, Model Fit
                                       FIGURE VII: Fit of psychological models of decision-making

                                               Baseline                                             Category budgeting
                  .78 .79 .8 .81 .82

                                                                           .78 .79 .8 .81 .82
                     Regular share

                                                                              Regular share
                                2006w1     2007w1      2008w1     2009w1                 2006w1      2007w1      2008w1   2009w1
                                                    Week                                                      Week

                                               Salience                                               Loss aversion
                  .78 .79 .8 .81 .82

                                                                           .78 .79 .8 .81 .82
                     Regular share

                                                                              Regular share

                                2006w1     2007w1      2008w1     2009w1                 2006w1      2007w1      2008w1   2009w1
                                                    Week                                                      Week

                                                                Observed                          Predicted

Notes: Data are from the retailer. The line labeled “observed” shows the weekly share of transactions that go to
       Simple mental accounting model does good job of fit
regular gasoline. The line labeled “predicted” shows the average predicted probability of buying regular gasoline from
estimates of the model in equation (8) with different specifications of Gi jt . In the baseline model we set Gi jt = 08i, j,t.
       Stefano DellaVigna                              Econ 219B: Applications (Lecture 11)                                 April 8, 2020   25 / 92
Mental Accounting   Hastings and Shapiro (2018)

Food Stamps
   Hastings and Shapiro (AER 2018)
   What happens when food stamps come in?
           Large majority of individuals spend more on food than food
           stamp amount
           Standard model: Increase in food expenditure should equal the
           marginal propensity to consume on food from income shocks
           (about 0.1)
           Mental accounting: MPCF from food stamps will be high, since
           same account
   Use data from a retailer where can observe is spend with food
   stamps
           Three empirical strategies:
      1    Individuals enter food stamp program
      2    Exit from program most likely after 6, 12, 18... months
      3    Legislative changes in food stamp magnitude
   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)             April 8, 2020   26 / 92
.
        Share of household
                                                                                      Mental Accounting                         Hastings and Shapiro (2018)

                .2

     Food Stamps
       0

                                        −12                               0                             12
                                                         Months relative to SNAP adoption
                                                                                                             Figure 5: Monthly expenditure before and after SNAP adoption
                                                        Panel B: SNAP benefits                                             Panel A: SNAP-eligible spending

                                                                                                                              500
                      200
       Monthly SNAP benefit (dollars)

                                                                                                               Monthly expenditure (dollars)
                              150

                                                                                                                                   450
                 100

                                                                                                                  400
       50

                                                                                                                              350
                      0

                                        −12                               0                             12                                     −12                    0                         12
                                                         Months relative to SNAP adoption                                                            Months relative to SNAP adoption

                                                                                             Panel B: SNAP-ineligible spending
                                              Strategy 1: Identify entry into SNAP as 6 months of SNAP
  ample is the set of SNAP adopters. Panel A plots the share of households with positive SNAP
 each of the 12 months before and after the household’s first SNAP adoption. Panel B plots the
                                                                                                                              200

                                              spending, after 6 months of no SNAP
AP benefit in each of the 12 months before and after the first SNAP adoption.

                                              MPC of about 0.5/0.6
                                                                                                               hly expenditure (dollars)
                                                                                                                               150

                                              Stefano DellaVigna                            Econ 219B: Applications (Lecture 11)                                                April 8, 2020    27 / 92
                                                                                                               00
Mental Accounting                                                                        Hastings and Shapiro (2018)

Food Stamps
                                                                              Figure 6: Participation, benefits, and spending over the six-month SNAP clock
                                                                                                             Panel A: SNAP use

                                                                                                                                                  −.02
                                                                                                         Change in share of households using SNAP
                                                                                                              −.08          −.06−.1     −.04
                                                                                                                                                         7,13,..   8,14,..        9,15,..       10,16,..    11,17,..                                             12,18,..
                                                                                                                                                                             Months following SNAP adoption

                                                                               Panel B: SNAP benefits                                                                                                                                             Panel C: SNAP-eligible spending
                                                −5

                                                                                                                                                                                                                                      −2
                                                                                                                                                                                             Change in monthly SNAP−eligible spending ($)
 58

                              Change in monthly SNAP benefits ($)

                                                                                                                                                                                                                             −4
                                                   −10

                                                                                                                                                                                                           −8      −6
                                    −15

                                                                                                                                                                                                 −10
                                                −20

                                                                                                                                                                                                                    −12

                                                                    7,13,..    8,14,..        9,15,..       10,16,..    11,17,..                                              12,18,..                                                      7,13,..   8,14,..        9,15,..       10,16,..    11,17,..   12,18,..
                                                                                         Months following SNAP adoption                                                                                                                                         Months following SNAP adoption

       Notes: Each figure plots coefficients from a regression of the dependent variable on a vector of indicators for the position of the current month in a monthly clock
       that begins in the most recent adoption month and resets every six months or at the next SNAP adoption, whichever comes first. So, for example, the first month
      Strategy 2: Identify exit from SNAP every 6 months
       of the clock corresponds to months 7, 13, 19, etc. following SNAP adoption. The unit of observation for each regression is the household-month. The sample is
       the set of SNAP adopters. Error bars are ±2 coefficient standard errors. Standard errors are clustered by household. Each regression includes calendar month fixed
       effects. The omitted category consists of the first six months (inclusive of the adoption month) after the household’s most recent SNAP adoption, all months after
      MPC of about 0.5/0.6
       the first 24 months (inclusive of the adoption month) following the household’s most recent adoption, and all months for which there is no preceding adoption. In
       panel A, the dependent variable is the change in an indicator for whether the household-month is a SNAP month. In panel B, the dependent variable is the change
      Stefano
       in monthlyDellaVigna
                    SNAP benefits. In Panel C, the dependentEcon
                                                               variable219B:     Applications
                                                                        is the change              (Lecturespending.
                                                                                      in monthly SNAP-eligible    11)                                  April 8, 2020                                                                                                                                                 28 / 92
Mental Accounting                                                                     Hastings and Shapiro (2018)

Food Stamps
                      Figure 7: Monthly SNAP benefits and SNAP-eligible spending around benefit changes
                                         Panel A: Administrative data for retailer states
                                                                                                                     Farm Bill           ARRA

                                           Average monthly SNAP benefit per household (dollars)
                                                                                                  300
                                                                                                  280
                                                                                                  260
                                                                                                  240
                                                                                                  220

                                                                                                  Jan−2008           Oct−2008           Apr−2009              Dec−2009

                                                                                                             Panel B: Retailer data
                                                                                                                   Farm Bill           ARRA

                                                                                                                                                                380
                                                                                                  550
                                           Monthly SNAP−eligible spending (dollars)

                                                                                                                                                                      Monthly SNAP benefits (dollars)
                                                                                                                                                                350
                                                                                                  520

                                                                                                                                                                320
                                                                                                  490

                                                                                                                                                                290
                                                                                                  460

                                                                                                                                                                260
                                                                                                  430

                                                                                                  Jan−2008         Oct−2008        Apr−2009               Dec−2009

                                                                                                              SNAP−eligible spending      SNAP benefits

                Notes: Panel A plots the average monthly SNAP benefit per household between January 2008 and December 2009
   Strategy     3: Identify from legislative changes in levels of benefits
         from administrative data. The series was obtained by weighing the average monthly SNAP benefit per household
         for each state (according to data from the United States Department of Agriculture Food and Nutrition Service via
                                                                                                                                                                                                        as of May 2017) by
   Stefano DellaVigna                       Econ
             the number of retailer households with 219B:    Applications
                                                    at least one SNAP month.(Lecture  11) coefficients from a regression
                                                                              Panel B plots                          April
                                                                                                                         of 8, 2020                                                                                          29 / 92
Mental Accounting          Hastings and Shapiro (2018)

    Food Stamps
                                                Table 1: Estimated marginal propensities to consume
                                                                        (1)                (2)                 (3)                (4)
                                                                   SNAP-eligible      SNAP-eligible       SNAP-eligible      SNAP-ineligible
                                                                     spending           spending            spending           spending
                 MPC out of
                    SNAP benefits                                      0.5891              0.5495             0.5884               0.0230
                                                                      (0.0074)            (0.0360)           (0.0073)             (0.0043)
                      cash                                            -0.0019             -0.0013            -0.0020               0.0421
                                                                      (0.0494)            (0.0494)           (0.0494)             (0.0688)
                 p-value for equality of MPCs                          0.0000              0.0000             0.0000               0.7764
                 Instruments:
                 Change in price of regular gasoline                     Yes                Yes                 Yes                  Yes
                 ⇥(Household average gallons per month)
                 SNAP adoption                                          Yes                 No                 Yes                  Yes
                 First month of SNAP clock                              No                  Yes                Yes                  Yes
                 Number of household-months                           2005392             2005392            2005392              2005392
                 Number of households                                  24456               24456              24456                24456
 tes: The sample is the set of SNAP adopters. The unit of observation is the household-month. Each column reports coefficient estimates from a 2SLS regressi
 h standard errors in parentheses clustered by household and calendar month using the method in Thompson (2011). All models are estimated in first differen
                Estimated MPCF stable across the three strategies around 0.6
d include calendar month fixed effects. Endogenous regressors are SNAP benefits and the additive inverse of fuel spending; coefficients on these regress
  reported as marginal propensities to consume. The “price of regular gasoline” is the quantity-weighted average spending per gallon on regular grade gaso
                Estimated MPCF from other income skocks (gas prices) much
 ong all households before any discounts or coupons. “Household average gallons per month” is the average monthly number of gallons of gasoline purchased
 iven household during the panel. “SNAP adoption” is an indicator for whether the month is an adoption month as defined in section 3.5. “First month of SN
                smaller
 ck” is an indicator equal to one in the first month of a six-month clock that begins in the most recent adoption month. The indicator is set to zero in the first
 nths (inclusive of the adoption month) following the most recent adoption, in any month after the first 24 months (inclusive of the adoption month) follow
  most recent adoption,
                 Stefanoand  in any month for which there Econ
                           DellaVigna                        is no preceding adoption. (Lecture 11)
                                                                    219B: Applications                                        April 8, 2020          30 / 92
Persuasion

                           Section 4

                          Persuasion

Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   31 / 92
Persuasion

Introduction

   Persuasion: Change in opinion/action beyond prediction of
   Bayesian model

   Persuasion: Sender attempts to convince Receiver with
   words/images to take an action
           Rational persuasion through Bayesian updating
           Non-rational persuasion, i.e.: neglect of incentives of person
           presenting information
           Effect of persuasion directly on utility function
           (advertising/emotions)

   Compare to Social Pressure: Presence of Sender exerts pressure
   to take an action

   Stefano DellaVigna      Econ 219B: Applications (Lecture 11)   April 8, 2020   32 / 92
Persuasion    DellaVigna and Gentzkow (2010)

Overview on Persuasion
   DellaVigna and Gentzkow (Annual Review Econ, 2010):
           Persuading   consumers: Marketing
           Persuading   voters: Political Communication
           Persuading   donors: Fund-raising
           Persuading   investors: Financial releases
   First problem: How to measure when persuasion occurs?
   Treatment group T, control group C, Persuasion Rate is
                                               yT − yC 1
                           f = 100 ∗                           ,
                                               eT − eC 1 − y 0

           ei is the share of group i receiving the message,
           yi is the share of group i adopting the behavior of interest,
           y0 is the share that would adopt if there were no message
   Stefano DellaVigna       Econ 219B: Applications (Lecture 11)                 April 8, 2020   33 / 92
Persuasion              DellaVigna and Gentzkow (2010)

                                                                    TWLLE w8 PWRT W
                                                           PERSUWSION RWTES- SUMMWRY OF STU:IES
  Paper                                          Treatment                    Iontrol          Variable t       Time Treatment Iontrol Exposure Persuasion
                                                                                                               Horizon group t T group t C rate e T -e C rate f
                                                     %w*                         %f*              %5*            %G*      %D*      %wM*        %ww*       %wf*
Persuading Ionsumers
  Simester et al, %fMMG* %NE*             wG clothing catalogs sent          wf catalogs    Share Purchasing    w year     AF,Gy   AA,Dy     wMMy>      5,fy
                                                                                               Yk w item                   FD,wy   FF,Ry     wMMy>      F,Dy
  Lertrand8 Karlan8 Mullainathan8        Mailer with female photo         Mailer no photo Wpplied for loan     w month     D,wy    R,Zy      wMMy>      M,Gy
  Shafir8 and Zinman %fMwM* %FE*        Mailer with 5,Zy interest rate    Mailer F,Zy i,r,                                 D,wy    R,Zy      wMMy>      M,Gy
Persuading Voters
  Gosnell %wDfF*                       Iard reminding of registration         No card         Registration     Few days    5f,My   AA,My     wMM,My    wA,5y
  Gerber and Green %fMMM* %FE*        :oorBtoB:oor GOTV Ianvassing           No GOTV            Turnout        Few days    5G,fy   55,Ry     fG,Dy     wZ,Fy
                                       GOTV Mailing of wBA Iards             No GOTV                                       5f,Ry   5f,fy     wMMy>      w,My
  Green8 Gerber8                         :oorBtoB:oor Ianvassing             No GOTV            Turnout        Few days    Aw,My   fR,Fy     fD,Ay     ww,Zy
  and Nickerson %fMMA* %FE*
  Green and Gerber %fMMw* %FE*           Phone Ialls Ly Youth Vote           No GOTV            Turnout        Few days    Gw,wy   FF,My     GA,Gy     fM,5y
                                        Phone Ialls wRBAM YearBOlds          No GOTV            Turnout                    5w,Fy   5M,Zy     5w,5y      5,Zy
  :ellaVigna and Kaplan %fMMG* %NE*   Wvailab, of Fox News Via Iable      No F,N, via cable Rep, Vote Share    MB5 years   ZF,5y   ZF,My      A,Gy     ww,Fy =
  Enikolopov8 Petrova8 and            Wvailability of independent antiB                      Vote Share of
  Zhuravskaya %fMwM* %NE*                Putin TV station %NTV*               No NTV        antiBPutin parties A months    wG,My   wM,Gy     5G,My     G,Gy=

  Knight and Ihiang %fMwM* %NE*       Unsurprising :em, Endors, %NYT*        No endors,     Support for Gore     Few       GZ,Zy   GZ,My     wMM,My     f,My
                                      Surprising :em, Endors, %:enver*       No endors,                         weeks      ZZ,wy   Zf,My     wMM,My     F,Zy
  Gerber8 Karlan8 and Lergan %fMMD*     Free wMBweek subscription to                        :em, Vote Share
  %FE*                                        Washington Post                No Subscr,     %stated in survey* f months    FG,fy   ZF,My     D5,My     wD,Zy=
  Gentzkow %fMMF* %NE*                     Exposure to Television          No Television        Turnout        wM years    Z5,Zy   ZF,Zy     RM,My      5,5y
  Gentzkow and Shapiro %fMMD* %NE*         Read Local Newspaper            No local paper       Turnout        MB5 years   GM,My   FD,My     fZ,My     wf,Dy

  Stefano DellaVigna                                 Econ 219B: Applications (Lecture 11)                                                  April 8, 2020          34 / 92
Persuasion                      DellaVigna and Gentzkow (2010)

                                                                                                 TABLE M. PART B
                                                                                        PERSUASION RATESB SUMMARY OF STUDIES
   Paper                                                                     Treatment                                Control                 Variable t               Time Treatment Control Exposure Persuasion
                                                                                                                                                                      Horizon group t T group t C rate e T -e C rate f
                                                                                  hMN                                    h8N                        hDN                 hUN      hAN      hMvN        hMMN       hM8N
Persuading Donors
  List and Lucking%Reiley                                   Fund%raiser mailer with low seed                        No mailer                  Share                M%, weeks           ,*Uy            vy           MvvyS              ,*Uy
  h8vv8N hFEN                                               Fund%raiser mailer with high seed                       No mailer              Giving Money                                 O*8y            vy           MvvyS              O*8y
   Landry. Lange. List. Price.                                Door%To%Door Fund%raising                               No visit                 Share                immediate          Mv*Oy            vy            ,f*,y            8A*Uy
   and Rupp h8vvfN hFEN                                      Campaign for University Center                                                Giving Money
   DellaVigna. List. and Malmendier                          Door%To%Door Fund%raising                                No visit                 Share                immediate           D*fy            vy           DM*Uy             MM*vy
   h8vvAN hFEN                                             Campaign for Out%of%State Charity                                               Giving Money
   Falk h8vvUN hFEN                                           Fund%raiser mailer with no gift                       No mailer                  Share                M%, weeks          M8*8y            vy           MvvyS             M8*8y
                                                              Mailer with gift hD post%cardsN                       No mailer              Giving Money                                8v*fy            vy           MvvyS             8v*fy
Persuading Investors
  Engelberg and Parsons h8vvAN hNEN                             Coverage of Earnings News                         No coverage            Trading of Shares             , days         v*v8,y         v*vMUy           fv*vy           v*vMvy
                                                                      in Local Paper                                                     of Stock in News
Notes: Calculations of persuasion rates by the authors* The list of papers indicates whether the study is a natural experiment hkNEkN or a field experiment hkFEkN* Columns hAN and hMvN report the value of the behavior studied hColumn hDNN for the
Treatment and Control group* Column hMMN reports the Exposure Rate. that is. the difference between the Treatment and the Control group in the share of people exposed to the Treatment* Column hM8N computes the estimated persuasion rate f as
MvvShtT%tCNGhheT%eCNShM%tCNN* The persuasion rate denotes the share of the audience that was not previously convinced and that is convinced by the message* The studies where the exposure rate hColumn hMMN is denoted by kMvvySk are cases in
which the data on the differential exposure rate between treatment and control is not available* In these case. we assume eT%eCYMvvy. which implies that the persuasion rate is a lower bound for the actual persuasion rate* In the studies on
kPersuading Donorsk. even in cases in which an explicit control group with no mailer or no visit was not run. we assume that such a control would have yielded tCYvy. since these behaviors are very rare in absence of a fund%raiser* For studies

    Persuasion rate helps reconcile seemingly very different results,
    e.g. persuading voters

    Stefano DellaVigna                                                           Econ 219B: Applications (Lecture 11)                                                                                           April 8, 2020                            35 / 92
Persuasion   DellaVigna and Kaplan (2007)

More in Detail

   More in detail: DellaVigna-Kaplan (QJE, 2007), Fox News
   natural experiment
      1    Fast expansion of Fox News in cable markets
                   October 1996: Launch of 24-hour cable channel
                   June 2000: 17 percent of US population listens regularly to Fox
                   News (Scarborough Research, 2000)
      2    Geographical differentiation in expansion
                   Cable markets: Town-level variation in exposure to Fox News
                   9,256 towns with variation even within a county
      3    Conservative content
                   Unique right-wing TV channel (Groseclose and Milyo, 2004)

   Stefano DellaVigna         Econ 219B: Applications (Lecture 11)              April 8, 2020   36 / 92
Persuasion   DellaVigna and Kaplan (2007)

Empirical Results

   Selection. In which towns does Fox News select? (Table 3):
         FOX              R, Pres          R
        dk,2000 = α + βvk,1996    + βContrk,1996 + Γ2000 Xk,2000 +
                  Γ00−90 Xk,00−90 + ΓC Ck,2000 + εk .

   Controls X
           Cable controls (Number of channels and potential subscribers)
           US House district or county fixed effects
   Conditional on X , Fox News availability is orthogonal to
           political variables
           demographic variables

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)              April 8, 2020   37 / 92
Persuasion   DellaVigna and Kaplan (2007)

Fox News Availability

   Stefano DellaVigna   Econ 219B: Applications (Lecture 11)              April 8, 2020   38 / 92
Persuasion   DellaVigna and Kaplan (2007)

Baseline effect – Presidential races

    Effect on Presidential Republican vote share (Table 4):
            R, Pres    R, Pres                FOX
           vk,2000  − vk,1996      = α + βF dk,2000 + Γ2000 Xk,2000 +
                                     Γ00−90 Xk,00−90 + ΓC Ck,2000 + εk .

    Results:
            Significant effect of Fox News with district (Column 3) and
            county fixed effects (Column 4)
            .4-.7 percentage point effect on Republican vote share in Pres.
            elections
            Similar effect on Senate elections           
                                                 Effect is on ideology, not
            person-specific
            Effect on turnout

    Stefano DellaVigna      Econ 219B: Applications (Lecture 11)              April 8, 2020   39 / 92
Persuasion   DellaVigna and Kaplan (2007)

Presidential Vote Share

   Stefano DellaVigna   Econ 219B: Applications (Lecture 11)              April 8, 2020   40 / 92
Persuasion   DellaVigna and Kaplan (2007)

Generalizing the Effect
   Magnitude of effect: How do we generalize beyond Fox News?

   Estimate audience of Fox News in towns that have Fox News via
   cable (First stage)
           Use Scarborough micro data on audience with Zip code of
           respondent
           Fox News exposure via cable increases regular audience by 6 to
           10 percentage points
           How many people did Fox News convince?
           Heuristic answer: Divide effect on voting (.4-.6 percentage
           point) by audience measure (.6 to .10)

   Result: Fox News convinced 3 to 8 percent of audience (Recall
   measure) or 11 to 28 percent (Diary measure)
   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)              April 8, 2020   41 / 92
Persuasion   DellaVigna and Kaplan (2007)

Interpretation

    How do we interpret the results?

    Benchmark model:                                                        
       1    New media source with unknown bias β, with β ∼ N β0 , γ1β
       2    Media observes
                          (differential) quality of Republican politician,
                        1
            θt ∼ N 0, γθ , i.i.d., in periods 1, 2, . . . , T
       3    Media broadcast: ψt = θt + β. Positive β implies
            pro-Republican media bias
       4    Voting in period T . Voters vote Republican if θbT + α > 0,
            with α ideological preference

    Stefano DellaVigna      Econ 219B: Applications (Lecture 11)              April 8, 2020   42 / 92
Persuasion   DellaVigna and Kaplan (2007)

Signal extraction problem. New media (Fox News) says
Republican politician (George W. Bush) is great
        Is Bush great?
        Or is Fox News pro-Republican?

A bit of both, the audience thinks. Updated media bias after T
periods:
                            γβ β0 + T γθ ψ̄T
                     β̂T =                   .
                               γβ + T γθ
Estimated quality of Republican politician:
                          h         i       h        i
              γθ ∗ 0 + W ψT − β̂T         W ψT − β̂T
        θ̂T =                          =
                      γθ + W                γθ + W

Stefano DellaVigna    Econ 219B: Applications (Lecture 11)              April 8, 2020   43 / 92
Persuasion   DellaVigna and Kaplan (2007)

Persuasion. Voter with persuasion λ (0 ≤ λ ≤ 1) does not take
into account enough media bias:

                               W λ [ψT − (1 − λ) β̂T ]
                     θ̂Tλ =
                                     γθ + W λ

Vote share for Republican candidate.
P(α + θbTλ ≥ 0) = 1 − F (−θbTλ )

Proposition 1. Three results:
   1    Short-Run I: Republican media bias increases Republican vote
                           λ )]/∂β > 0.
        share: ∂[1 − F (−θbT
   2    Short-Run II: Media bias effect higher if persuasion (λ > 0).
   3    Long-run (T → ∞). Media bias effect ⇐⇒ persuasion λ > 0.

Stefano DellaVigna     Econ 219B: Applications (Lecture 11)              April 8, 2020   44 / 92
Persuasion   Cain, Loewenstein, and Moore (2005)

Evidence for Persuasion Bias

   Cain-Loewenstein-Moore (JLegalStudies, 2005).
   Psychology Experiment
           Pay subjects for precision of estimates of number of coins in a
           jar
           Have to rely on the advice of second group of subjects: advisors
           (Advisors inspect jar from close)
           Two experimental treatments:
                   Aligned incentives. Advisors paid for closeness of subjects’ guess
                   Mis-Aligned incentives, Common knowledge. Advisors paid for
                   how high the subjects’ guess is. Incentive common-knowledge
                   (Mis-Aligned incentives, Not Common knowledge.)

   Stefano DellaVigna          Econ 219B: Applications (Lecture 11)                     April 8, 2020   45 / 92
Persuasion   Cain, Loewenstein, and Moore (2005)

Payoffs

   Stefano DellaVigna   Econ 219B: Applications (Lecture 11)                     April 8, 2020   46 / 92
Persuasion   Cain, Loewenstein, and Moore (2005)

Result 1

   Advisors increase estimate in Mis-Aligned incentives treatment
   — Even more so when common knowledge

   Stefano DellaVigna   Econ 219B: Applications (Lecture 11)                     April 8, 2020   47 / 92
Persuasion   Cain, Loewenstein, and Moore (2005)

Result 2
   Estimate of subjects is higher in Treatment with Mis-Aligned
   incentives

   Subjects do not take sufficiently into account incentives of
   information provider
   Effect even stronger when incentives are known      Advisors feel    
   free(er) to increase estimate

   Applications to many settings
   Stefano DellaVigna   Econ 219B: Applications (Lecture 11)                     April 8, 2020   48 / 92
Persuasion   Malmendier and Shanthikumar (2007)

Application: Small Investors

   Application 1: Malmendier-Shanthikumar (JFE, 2007).
           Field evidence that small investors suffer from similar bias
           Examine recommendations by analysts to investors
           Substantial upward distortion in recommendations (Buy=Sell,
           Hold=Sell, etc)

   Higher distortion for analysis working in Inv. Bank affiliated with
   company they cover (through IPO/SEO)

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)                    April 8, 2020   49 / 92
Persuasion   Malmendier and Shanthikumar (2007)

Question

   Question: Do investors discount this bias?
           Analyze Trade Imbalance (essentially, whether trade is initiated
           by Buyer)
           Assume that
                   large investors do large trades
                   small investors do small trades
           See how small and large investors respond to recommendations

   Examine separately for affiliated and unaffiliated analysts

   Stefano DellaVigna          Econ 219B: Applications (Lecture 11)                    April 8, 2020   50 / 92
Persuasion   Malmendier and Shanthikumar (2007)

Analyst Recommendations

   Stefano DellaVigna   Econ 219B: Applications (Lecture 11)                    April 8, 2020   51 / 92
Persuasion   Malmendier and Shanthikumar (2007)

Results

   Results:
           Small investor takes analyst recommendations literally (buy
           Buys, sell Sells)
           Large investors discount for bias (hold Buys, sell Holds)
           Difference is particularly large for affiliated analysts
           Small investors do not respond to affiliation information

   Strong evidence of distortion induced by incentives

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)                    April 8, 2020   52 / 92
Emotions: Mood

                           Section 5

                     Emotions: Mood

Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   53 / 92
Emotions: Mood

Emotions Matter

   Emotions play a role in several of the phenomena considered so
   far:
           Self-control problems    Temptation
           Projection bias in food consumption      Hunger  
           Social preferences in giving      
                                          Empathy
           Gneezy-List (2006) transient effect of gift  Hot-Cold 
           gift-exchange

   Psychology: Large literature on emotions (Loewenstein and
   Lerner, 2003)
           Message 1: Emotions are very important
           Message 1: Different emotions operate very differently: anger 6=
           mood 6= joy

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)       April 8, 2020   54 / 92
Emotions: Mood

Consider two examples of emotions:
        Mood
        Arousal

Psychology: even minor mood manipulations have a substantial
impact on behavior and emotions
        On sunnier days, subjects tip more at restaurants (Rind, 1996)
        On sunnier days, subjects express higher levels of overall
        happiness (Schwarz and Clore, 1983)

Should this impact economic decisions?

Stefano DellaVigna     Econ 219B: Applications (Lecture 11)   April 8, 2020   55 / 92
Emotions: Mood     Hirshleifer and Shumway (2003)

Field Evidence

   Field: Impact of mood fluctuations on stock returns:
           Daily weather and Sport matches
           No effect on fundamentals
           However: If good mood leads to more optimistic expectations
             Increase in stock prices
   Evidence:
           Saunders (1993): Days with higher cloud cover in New York
           are associated with lower aggregate US stock returns
           Hirshleifer and Shumway (2003) extend to 26 countries
           between 1982 and 1997
                   Use weather of the city where the stock market is located
                   Negative relationship between cloud cover (de-trended from
                   seasonal averages) and aggregate stock returns in 18 of the 26
                   cities

   Stefano DellaVigna         Econ 219B: Applications (Lecture 11)                April 8, 2020   56 / 92
Emotions: Mood     Hirshleifer and Shumway (2003)

Weather and Stock Returns

   Stefano DellaVigna   Econ 219B: Applications (Lecture 11)                April 8, 2020   57 / 92
Emotions: Mood     Hirshleifer and Shumway (2003)

Weather and Stock Returns
   Magnitude:
           Days with completely covered skies have daily stock returns .11
           percent lower than days with sunny skies
           Five percent of a standard deviation
           Small magnitude, but not negligible

   After controlling for cloud cover, other weather variables such as
   rain and snow are unrelated to returns
   Edmans-Garcia-Norli, 2007: Evidence from international
   soccer matches (39 countries, 1973-2004)
   Interpretations:
           Mood impacts risk aversion or perception of volatility
           Mood is projected to economic fundamentals

   Stefano DellaVigna      Econ 219B: Applications (Lecture 11)                April 8, 2020   58 / 92
Emotions: Mood     Simonsohn (2007)

College Enrollment

   Simonsohn (2007): Subtle role of mood
           Weather on the day of campus visit to a prestigious university
           (CMU)
           Students visiting on days with more cloud cover are significantly
           more likely to enroll
           Higher cloud cover induces the students to focus more on
           academic attributes versus social attributes of the school
           Support from laboratory experiment

   Stefano DellaVigna      Econ 219B: Applications (Lecture 11)   April 8, 2020   59 / 92
Emotions: Arousal

                             Section 6

                     Emotions: Arousal

Stefano DellaVigna     Econ 219B: Applications (Lecture 11)   April 8, 2020   60 / 92
Emotions: Arousal

Separate impact of emotions: Arousal

   Josephson (1987): Arousal due to violent content
           Control group exposed to non-violent clip
           Treatment group exposed to violent clip
           Treatment group more likely to display more aggressive
           behavior, such as aggressive play during a hockey game
           Impact not due to imitation (violent movie did not involve sport
           scenes)
   Consistent finding from large set of experiments (Table 11)

   Dahl-DellaVigna (2009): Field evidence — Exploit timing of
   release of blockbuster violent movies

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)   April 8, 2020   61 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Model
   Consumer chooses between strongly violent movie av , mildly
   violent movie am , non-violent movie an , or alternative social
   activity as
           Utility depends on quality of movies                     Demand functions P(a )         j

   Heterogeneity:
           High taste for violence (Young): Ny consumers
           Low taste for violence (Old): No consumers
           Aggregate demand for group i: Ni P(aij )
   Production function of violence V (not part of utility fct.)
   depends on av , am , an , and as :
          X X j
   ln V =     [        αi Ni P(aij )+σi Ni (1−P(aiv )−P(aim )−P(ain ))]
               i=y ,o j=v ,m,n

   Stefano DellaVigna             Econ 219B: Applications (Lecture 11)            April 8, 2020   62 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Estimate (Aj is total attendance to movie of type j)

                     ln V = β0 + β v Av + β m Am + β n An + ε

Estimated impact of exposure to violent movies β v :

                     β v = x v (αyv − σy ) + (1 − x v )(αov − σo )

First point — Estimate of net effect
        Direct effect: Increase in violent movie exposure                    α   v

        Indirect effect: Decrease in Social Activity   σi                        i

Second point — Estimate on self-selected population:
        Estimate parameters for group actually attending movies
        Young over-represented: x v > N y / (N y + N o )

Stefano DellaVigna           Econ 219B: Applications (Lecture 11)            April 8, 2020   63 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Comparison with Psychology experiments
        Natural Experiment. Estimated impact of exposure to violent
        movies β v :

                     β v = x v (αyv − σy ) + (1 − x v )(αov − σo )

        Psychology Experiments. Manipulate a directly, holding
        constant as out of equilibrium

                      v          Ny                 Ny
                     βlab =            αyv + (1 −        )αv
                               Ny + No            Ny + No o

Two differences:
        ‘Shut down’ alternative activity, and hence σi does not appear
        Weights representative of (student) population, not of
        population that selects into violent movies

Stefano DellaVigna       Econ 219B: Applications (Lecture 11)            April 8, 2020   64 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Data
   Movie data
           Revenue data: Weekend (top 50) and Day (top 10) from The
           Numbers
           Violence Ratings from 0 to 10 from Kids In Mind (Appendix
           Table 1)
           Strong Violence Measure Avt : Audience with violence 8-10
           (Figure 1a)
           Mild Violence Measure Amt : Audience with violence 5-7 (Figure
           1b)
   Assault data
           Source: National Incident-Based Reporting System (NIBRS)
           All incidents of aggravated assault, simple assault, and
           intimidation from 1995 to 2004
           Sample: Agencies with no missing data on crime for > 7 days
           Sample: 1995-2004, days in weekend (Friday, Saturday, Sunday)
   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)            April 8, 2020   65 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Movie Attendance

   Stefano DellaVigna    Econ 219B: Applications (Lecture 11)            April 8, 2020   66 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Log Assault Residuals

   Stefano DellaVigna    Econ 219B: Applications (Lecture 11)            April 8, 2020   67 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Regression and Results
   Regression Specification. (Table 3)

                        log Vt = β v Avt + β m Am    n n
                                                t + β At + ΓXt + εt

           Coefficient β v is percent increase in assault for one million
           people watching strongly violent movies day t (Avt ) (Similarly
           β m and β n )
           Cluster standard errors by week

   Results.
           No effect of movie exposure in morning or afternoon (Columns
           1-2)
           Negative effect in the evening (Column 3)
           Stronger negative effect the night after (Column 4)
   Stefano DellaVigna           Econ 219B: Applications (Lecture 11)            April 8, 2020   68 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Stefano DellaVigna    Econ 219B: Applications (Lecture 11)            April 8, 2020   69 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Summary of Findings

 1   Violent movies lower same-day violent crime in the evening
     (incapacitation)
 2   Violent movies lower violent crime in the night after exposure
     (less consumption of alcohol in bars)
     No lagged effect of exposure in weeks following movie
                          
 3

     attendance     No intertemporal substitution
 4   Strongly violent movies have slightly smaller impact compared to
     mildly violent movies in the night after exposure

     Interpret Finding 4 in light of Lab-Field debate

     Stefano DellaVigna        Econ 219B: Applications (Lecture 11)            April 8, 2020   70 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Interpretation

    Finding 4. Non-monotonicity in Violent Content
            Night hours: β̂ v = −0.0192 versus β̂ m = −0.0205
            Odd if more violent movies attract more potential criminals
            Model above      Can estimate direct effect of violent movies if
            can control for selection
                                              xv − xn
                                                                  
                       v           v     n
                     α −α=β − β + m                     (βm − βn )
                                              x − xn

            Do not observe selection of criminals x j , but observe selection
            of correlated demographics (young males)

    Stefano DellaVigna      Econ 219B: Applications (Lecture 11)            April 8, 2020   71 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

                                                              
IMDB ratings data — Share of young males among raters
increases with movie violence (Figure 2) Use as estimate of
xj
        Compute α\ v − α = .011 (p = .08), about one third of total

        effect
        Pattern consistent with arousal induced by strongly violent
        movies (αv > αm )

Bottom-line 1: Can reconcile with laboratory estimates
Bottom-line 2: Can provide benchmark for size of arousal effect

Stefano DellaVigna     Econ 219B: Applications (Lecture 11)            April 8, 2020   72 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Share of Young Males vs. Movie Violence

   Stefano DellaVigna    Econ 219B: Applications (Lecture 11)            April 8, 2020   73 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

The Arousal Effect

   Stefano DellaVigna    Econ 219B: Applications (Lecture 11)            April 8, 2020   74 / 92
Emotions: Arousal   Dahl and DellaVigna (2009)

Lab vs. Field

   Differences from laboratory evidence (Levitt-List, 2007):
   Exposure to violent movies is
           Less dangerous than alternative activity (αv < σ)
           (Natural Experiment)
           More dangerous than non-violent movies (αv > αn )
           (Laboratory Experiments and indirect evidence above)

   Both types of evidence are valid for different policy evaluations
           Laboratory: Banning exposure to unexpected violence
           Field: Banning temporarily violent movies

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)            April 8, 2020   75 / 92
Happiness

                           Section 7

                          Happiness

Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   76 / 92
Happiness

Measuring Utility Through Happiness

   Is there a more direct way to measure utility?

   What about happiness questions?
           ‘Taken all together, how would you say things are these days,
           would you say that you are very happy, pretty happy, or not too
           happy?’
           or ‘How satisfied are you with your life as a whole?’
           Response on 1 to 7 or 0 to 10 scale

   Could average response measure utility?

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)   April 8, 2020   77 / 92
Happiness

There are a number of issues:
   1    (Noise I) Is the measure of happiness just noise?
   2    (Noise II) Even if valid, there are no incentives, how affected is
        it by irrelevant cues?
   3    (Scale) Happiness is measured on discrete intervals, with ceiling
        and floor effect
   4    (Content) What exactly does the measure capture?
        Instantaneous utility? Discounted utility?

Revealed preference approach remains heavily favored by
economists (myself included)

Still, significant progress in last 10-15 years on taking some role
in economics

Stefano DellaVigna      Econ 219B: Applications (Lecture 11)   April 8, 2020   78 / 92
Happiness    Oreopoulos (2006)

Issue 1: Noise I

    Issue 1 (Noise I). To address,
            Take happiness measure h
            Does it responds to well-identified, important shifters X which
            affect important economic outcomes?

    Oreopoulos (AER 2006). Exploit binding compulsory
    schooling laws to study returns to education

    Stefano DellaVigna      Econ 219B: Applications (Lecture 11)   April 8, 2020   79 / 92
Happiness    Oreopoulos (2006)

UK: 1947 increase in minimum schooling from 14 to 15

    Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   80 / 92
Happiness    Oreopoulos (2006)

Northern Ireland: 1957 increase from 14 to 15

    Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   81 / 92
Happiness    Oreopoulos (2006)

Clear impact on earnings: compare earnings for adults aged 32-64 as
a function of year of birth

    Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   82 / 92
Happiness    Oreopoulos (2006)

Implied returns to compulsory education: 0.148 (0.046)

    Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   83 / 92
Happiness    Oreopoulos (2007)

Did this affect happiness measures?

   Oreopoulos (JPubE 2007)
           Eurobarometer Surveys in UK and N. Ireland, 1973-1998
           Question on 1-4 scale

   Stefano DellaVigna    Econ 219B: Applications (Lecture 11)   April 8, 2020   84 / 92
Happiness    Oreopoulos (2007)

Results
   One year of additional (compulsory) education increases
   happiness somewhere between 2 and 8 percent
   In addition, large effects on health and wealth
   Reinforces puzzle: Why don’t people stay in school longer?

   Happiness response captures real information

   Happiness answer also responds to cues (Issue 2), has scale
   effects (Issue 3), but valid enough to use in combination with
   other measures

   However, Issue 4: How would we use happiness measure as part
   of economic research?
   Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   85 / 92
Happiness    Benjamin et al. (various)

Happiness in Economic Research

   Research agenda by Dan Benjamin, Ori Heffetz, Miles Kimball,
   Alex Rees-Jones
           Study Econ101a-type simple issues with happiness measures
           Critical to know how to correctly interpret these measures

   Paper 1. Benjamin, Heffetz, Kimball, and Rees-Jones
   (AER 2012)
           How does happiness (subjective well-being) relate to choice?
           Compare forecasted happiness with choice in several
           hypothetical scenarios
           Forecasts of happiness predict choice quite well, but other
           factors also play a role

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)           April 8, 2020   86 / 92
Happiness    Benjamin et al. (various)

Paper 2. Benjamin et al. (AER 2014)
        Medical students choosing match for residency
        Survey to elicit ranking of medical schools for residency + Ask
        anticipated happiness
        How well does happiness predict choice relative to other factors?

Stefano DellaVigna      Econ 219B: Applications (Lecture 11)           April 8, 2020   87 / 92
Happiness    Benjamin et al. (various)

Some evidence that one can also elicit intertemporal happiness

Stefano DellaVigna   Econ 219B: Applications (Lecture 11)           April 8, 2020   88 / 92
Happiness    Uses of Happiness Data

Uses of Happiness Data
   Di Tella, MacCulloch, and Oswald (AER 2001):
           How much do people dislike inflation versus unemployment?
           Life satisfaction from Euro-Barometer: “On the whole, are you
           very satisfied, fairly satisfied, not very satisfied, or not at all
           satisfied with the life you lead?”
           Aggregate to country-year and estimate panel regression

   Stefano DellaVigna      Econ 219B: Applications (Lecture 11)        April 8, 2020   89 / 92
Happiness    Other Work

Other Important Work on Happiness

   Luttmer (QJE 2005): Documents relative aspect of happiness:
   An increase in income of neighbors (appropriately instrumented)
   lower life satisfaction

   Stevenson and Wolfers (Brookings 2008):
           Debunks Easterlin paradox (income growth over time does not
           increase happiness)
           Clear link over time between log income and happiness

   Finkelstein, Luttmer, Notowidigdo (JEEA 2014):
           How does marginal utility of consumption vary with health?
           Needed for optimal policies
           Observe changes in happiness for varying health

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)   April 8, 2020   90 / 92
Next Lecture

                           Section 8

                       Next Lecture

Stefano DellaVigna   Econ 219B: Applications (Lecture 11)   April 8, 2020   91 / 92
Next Lecture

Next Lecture

   Market Response to Biases
           Employees: Behavioral Labor
           Investors: Behavioral Finance
           Voters: Behavioral Political Economy

   Stefano DellaVigna     Econ 219B: Applications (Lecture 11)   April 8, 2020   92 / 92
You can also read