1
00:00:00,000 --> 00:00:01,948
[SQUEAKING]

2
00:00:01,948 --> 00:00:03,409
[RUSTLING]

3
00:00:03,409 --> 00:00:05,357
[CLICKING]

4
00:00:10,720 --> 00:00:12,970
PROFESSOR: OK, so
what we're going

5
00:00:12,970 --> 00:00:17,360
to do in talking about social
preferences is the following.

6
00:00:17,360 --> 00:00:20,620
We're going to start discussing
what our social preferences.

7
00:00:20,620 --> 00:00:22,443
We're going to just
talk about what

8
00:00:22,443 --> 00:00:23,860
do we think social
preferences are

9
00:00:23,860 --> 00:00:25,780
and how do we think about
them, how do we measure them.

10
00:00:25,780 --> 00:00:27,550
This is partially
what we did already

11
00:00:27,550 --> 00:00:30,340
in class last week,
which gives you

12
00:00:30,340 --> 00:00:32,170
some sense of some
ways of measuring them.

13
00:00:32,170 --> 00:00:33,500
There are also other
ways of doing that.

14
00:00:33,500 --> 00:00:34,690
We're going to discuss that.

15
00:00:34,690 --> 00:00:35,770
We're going to
also discuss a bit

16
00:00:35,770 --> 00:00:37,120
kind of like the
measurements that we

17
00:00:37,120 --> 00:00:38,662
did in class that
seem sort of pretty

18
00:00:38,662 --> 00:00:40,120
contrived in various ways.

19
00:00:40,120 --> 00:00:42,520
Are they actually predictive
in some ways of real world

20
00:00:42,520 --> 00:00:44,488
behavior that we care about?

21
00:00:44,488 --> 00:00:46,030
Then we're going to
ask the question,

22
00:00:46,030 --> 00:00:48,520
well, it seems like, from
measuring social preferences,

23
00:00:48,520 --> 00:00:51,705
as we did in class, but also in
some experiments that we see,

24
00:00:51,705 --> 00:00:53,080
it looks like
people are actually

25
00:00:53,080 --> 00:00:54,220
quite nice to each other.

26
00:00:54,220 --> 00:00:55,720
But we're going to
ask the question,

27
00:00:55,720 --> 00:00:57,595
are people just genuinely
nice to each other?

28
00:00:57,595 --> 00:00:59,230
Do they actually
care about others?

29
00:00:59,230 --> 00:01:01,270
Or is it just
like, in some ways,

30
00:01:01,270 --> 00:01:03,760
perhaps, social image or
other issues about like they

31
00:01:03,760 --> 00:01:05,500
want to look nice
or they want to feel

32
00:01:05,500 --> 00:01:09,070
good by being nice to
others or look like they're

33
00:01:09,070 --> 00:01:11,410
nice to others, as opposed
to generally caring

34
00:01:11,410 --> 00:01:13,820
about the well-being of others.

35
00:01:13,820 --> 00:01:15,730
And we can talk through
some experiments

36
00:01:15,730 --> 00:01:17,560
and some evidence that
lets us disentangle

37
00:01:17,560 --> 00:01:19,055
between those kinds
of hypotheses--

38
00:01:19,055 --> 00:01:21,430
are people genuinely nice or
are they sort of just trying

39
00:01:21,430 --> 00:01:24,213
to look nice in various ways?

40
00:01:24,213 --> 00:01:25,630
And then we're
going to talk a bit

41
00:01:25,630 --> 00:01:27,640
about evidence towards the end.

42
00:01:27,640 --> 00:01:29,920
This is the paper by
Gautam Rao and others,

43
00:01:29,920 --> 00:01:32,560
which is the question about are
social preferences malleable

44
00:01:32,560 --> 00:01:35,440
and can policies
change pro-sociality.

45
00:01:35,440 --> 00:01:37,750
That is to say, can
we do certain things,

46
00:01:37,750 --> 00:01:40,420
can we mix people in certain
ways-- it can be like this will

47
00:01:40,420 --> 00:01:45,833
be how roommates are chosen,
and the specific people

48
00:01:45,833 --> 00:01:47,500
we're going to talk
about is from India,

49
00:01:47,500 --> 00:01:50,590
where poor students are
mixed with rich students.

50
00:01:50,590 --> 00:01:52,180
And does that affect
how nice people

51
00:01:52,180 --> 00:01:55,780
are, the rich students, in this
case, towards poor students,

52
00:01:55,780 --> 00:01:57,860
but also towards other
people in general.

53
00:01:57,860 --> 00:02:00,520
So are there some policies
that we can perhaps implement

54
00:02:00,520 --> 00:02:05,140
that help people, society as
a whole, being more pro-social

55
00:02:05,140 --> 00:02:07,300
and people being
nicer to each other

56
00:02:07,300 --> 00:02:09,910
or doing things that
are good for the greater

57
00:02:09,910 --> 00:02:12,437
good, if you want, overall.

58
00:02:12,437 --> 00:02:14,770
As before, we're going to
look at two types of evidence.

59
00:02:14,770 --> 00:02:17,440
We're going to look at lab
experiments and field evidence.

60
00:02:17,440 --> 00:02:19,420
Lab experiments are the
types of experiments

61
00:02:19,420 --> 00:02:22,000
that you saw so far.

62
00:02:22,000 --> 00:02:25,420
And field evidence are sort
of the type of experiments

63
00:02:25,420 --> 00:02:28,750
where we look at students being
mixed, rich and poor students.

64
00:02:28,750 --> 00:02:30,922
What happens then to
people's social preferences

65
00:02:30,922 --> 00:02:31,630
when you do that?

66
00:02:36,030 --> 00:02:38,700
OK, so now let's sort
of back up a little

67
00:02:38,700 --> 00:02:44,350
bit into review, very quickly,
of what we did last week.

68
00:02:44,350 --> 00:02:46,440
I showed you sort of
an action, three games

69
00:02:46,440 --> 00:02:47,890
to measure social preferences.

70
00:02:47,890 --> 00:02:49,690
These are not randomly
selected games,

71
00:02:49,690 --> 00:02:53,130
these are the most prevalent,
the most commonly-used games

72
00:02:53,130 --> 00:02:55,555
in behavioral and
experimental economics,

73
00:02:55,555 --> 00:02:57,930
which is the dictator game,
ultimatum game, and the trust

74
00:02:57,930 --> 00:02:58,530
game.

75
00:02:58,530 --> 00:03:01,020
You can read descriptions and
discussions of these games

76
00:03:01,020 --> 00:03:05,160
in the background readings,
which is Camerer and Fehr 2004.

77
00:03:05,160 --> 00:03:06,640
Now what are these games?

78
00:03:06,640 --> 00:03:08,833
I'm going to briefly
review them right now.

79
00:03:08,833 --> 00:03:10,500
And then I'm going
to go to the evidence

80
00:03:10,500 --> 00:03:13,368
and go back to evidence that
actually uses these games.

81
00:03:13,368 --> 00:03:14,910
So what are the
games, just to recap?

82
00:03:14,910 --> 00:03:17,700
There's the dictator game,
which is very simple,

83
00:03:17,700 --> 00:03:21,720
where the dictator gets some
money allocated to him or her.

84
00:03:21,720 --> 00:03:24,300
And there's another person
who is the recipient.

85
00:03:24,300 --> 00:03:25,800
And a dictator
just gets to choose

86
00:03:25,800 --> 00:03:27,412
how much money does
he or she wants

87
00:03:27,412 --> 00:03:28,620
to give to this other person.

88
00:03:28,620 --> 00:03:29,700
That's it.

89
00:03:29,700 --> 00:03:32,250
You can think of this as
some raw concern for others,

90
00:03:32,250 --> 00:03:35,288
essentially to say, when
you don't know anything

91
00:03:35,288 --> 00:03:37,080
about this other person--
often these games

92
00:03:37,080 --> 00:03:38,970
are anonymous-- here's
an anonymous person whom

93
00:03:38,970 --> 00:03:40,000
you could give some money.

94
00:03:40,000 --> 00:03:41,310
And often it's another student.

95
00:03:41,310 --> 00:03:45,553
It could be like a poor
person in Kenya or elsewhere.

96
00:03:45,553 --> 00:03:47,220
Here's a person with
whom you don't have

97
00:03:47,220 --> 00:03:48,853
any interactions whatsoever.

98
00:03:48,853 --> 00:03:50,520
How much money do you
want for yourself?

99
00:03:50,520 --> 00:03:53,040
How much money do you want
for this other person?

100
00:03:53,040 --> 00:03:55,200
The original version
is just sort of 1 to 1.

101
00:03:55,200 --> 00:03:57,570
Any dollar that you
give is given directly

102
00:03:57,570 --> 00:03:58,723
to this other person.

103
00:03:58,723 --> 00:04:01,140
There's different versions
where essentially any dollar is

104
00:04:01,140 --> 00:04:03,630
converted, where essentially
the exchange rate is varied,

105
00:04:03,630 --> 00:04:05,310
where it's cheaper
or more expensive,

106
00:04:05,310 --> 00:04:08,310
if you give a dollar, the other
person gets $2.00 or $0.50

107
00:04:08,310 --> 00:04:09,840
or the like.

108
00:04:09,840 --> 00:04:11,760
OK, so that's the dictator game.

109
00:04:11,760 --> 00:04:12,930
What's the ultimatum game?

110
00:04:12,930 --> 00:04:14,930
It's sort of like a
bargaining game if you want.

111
00:04:14,930 --> 00:04:18,870
It's sort of very, very simple
game theory if you want.

112
00:04:18,870 --> 00:04:20,940
You can use that, which
is essentially used again

113
00:04:20,940 --> 00:04:22,350
by experimental economists.

114
00:04:22,350 --> 00:04:24,480
What they do is there's
two players again.

115
00:04:24,480 --> 00:04:28,050
There's a proposer or the
sender that gives, again, like,

116
00:04:28,050 --> 00:04:29,570
some divisible pie.

117
00:04:29,570 --> 00:04:31,290
Think of this as
$10 or the like.

118
00:04:31,290 --> 00:04:32,340
Often it's money.

119
00:04:32,340 --> 00:04:35,160
And then you say a portion
of x is given, then,

120
00:04:35,160 --> 00:04:36,420
to the responder.

121
00:04:36,420 --> 00:04:38,370
The responder then
has the chance

122
00:04:38,370 --> 00:04:40,290
to either accept or reject.

123
00:04:40,290 --> 00:04:42,660
If the responder accepts,
then the division

124
00:04:42,660 --> 00:04:43,830
is just implemented.

125
00:04:43,830 --> 00:04:47,670
If the responder rejects,
then essentially both players

126
00:04:47,670 --> 00:04:50,568
get nothing.

127
00:04:50,568 --> 00:04:51,610
And why is it a game now?

128
00:04:51,610 --> 00:04:55,860
Well, it's because
essentially the proposer

129
00:04:55,860 --> 00:04:57,510
needs to sort of take
into account what

130
00:04:57,510 --> 00:04:58,920
the other person is
going to do and have

131
00:04:58,920 --> 00:05:01,290
some beliefs about that and
anticipating that, then sort

132
00:05:01,290 --> 00:05:05,047
make that choice of how much
money would you like to give.

133
00:05:05,047 --> 00:05:06,630
Finally, there's the
trust game, which

134
00:05:06,630 --> 00:05:08,520
is, in fact, very
similar to the ultimatum

135
00:05:08,520 --> 00:05:12,150
game except for that the
amount sent by the sender

136
00:05:12,150 --> 00:05:15,930
is tripled before the
decider is supposed to say--

137
00:05:19,650 --> 00:05:21,780
before they decide
how much to return,

138
00:05:21,780 --> 00:05:25,600
the recipient decides how
much return, if anything.

139
00:05:25,600 --> 00:05:29,250
So now essentially, it's like
a high-return opportunity

140
00:05:29,250 --> 00:05:35,508
to give money, essentially
in a situation of high trust.

141
00:05:35,508 --> 00:05:37,050
In particular,
there's communication.

142
00:05:37,050 --> 00:05:39,342
If I'm playing with somebody,
I would tell that person,

143
00:05:39,342 --> 00:05:40,620
yes, give me all of the money.

144
00:05:40,620 --> 00:05:42,840
And if you are nice to
me, that money is tripled,

145
00:05:42,840 --> 00:05:44,970
and I give you half of
it back, or the like.

146
00:05:44,970 --> 00:05:46,800
So both of us are better off.

147
00:05:46,800 --> 00:05:49,080
If there's lots of trust
between us, that's actually

148
00:05:49,080 --> 00:05:51,288
implementable, and a sense
of, like, I'm telling you,

149
00:05:51,288 --> 00:05:53,205
give me that money,
you'll give me that money,

150
00:05:53,205 --> 00:05:55,110
and then I'm actually
returning that money.

151
00:05:55,110 --> 00:05:56,568
But of course, if
there's no trust,

152
00:05:56,568 --> 00:05:58,985
then that's what the game name
comes from, the trust game.

153
00:05:58,985 --> 00:06:01,110
If there's no trust, I'm
going to tell you, give me

154
00:06:01,110 --> 00:06:03,100
a bunch of money, you
might actually do that,

155
00:06:03,100 --> 00:06:05,010
and then I'll just
keep it all for myself.

156
00:06:05,010 --> 00:06:07,320
Or you might not even believe
me in the first place.

157
00:06:07,320 --> 00:06:11,080
You might just say, yeah, I'd
love to, but I don't trust you,

158
00:06:11,080 --> 00:06:11,580
Frank.

159
00:06:11,580 --> 00:06:13,050
I give you nothing.

160
00:06:13,050 --> 00:06:15,330
And then essentially,
nothing is tripled.

161
00:06:15,330 --> 00:06:19,620
And that's meant to then measure
some form of trust in society

162
00:06:19,620 --> 00:06:22,333
or that people have
with each other.

163
00:06:22,333 --> 00:06:24,000
We're going to go
back to that in trying

164
00:06:24,000 --> 00:06:27,725
to see what does it actually
measure in the real world.

165
00:06:27,725 --> 00:06:29,100
So just sort of
having said that,

166
00:06:29,100 --> 00:06:30,660
those are those
three things you saw.

167
00:06:30,660 --> 00:06:31,890
And then there's
different versions

168
00:06:31,890 --> 00:06:33,848
of these games that could
be sort of like-- you

169
00:06:33,848 --> 00:06:35,610
can vary the privacy.

170
00:06:35,610 --> 00:06:37,680
Are these games
private versus public?

171
00:06:37,680 --> 00:06:38,640
Is it anonymous?

172
00:06:38,640 --> 00:06:40,920
Do you know who the other
person is versus not?

173
00:06:40,920 --> 00:06:42,150
Is there communication?

174
00:06:42,150 --> 00:06:44,610
Can you talk to this
other person or not?

175
00:06:44,610 --> 00:06:46,950
And what are the stakes?

176
00:06:46,950 --> 00:06:49,650
Are we talking about
hypothetical choices?

177
00:06:49,650 --> 00:06:50,910
Are we talking about money?

178
00:06:50,910 --> 00:06:51,990
High stakes with money.

179
00:06:51,990 --> 00:06:56,940
Are we talking about other
goods like apples or chocolate?

180
00:06:56,940 --> 00:06:58,350
Any questions so far?

181
00:06:58,350 --> 00:07:00,150
We'll get back to
the actual evidence,

182
00:07:00,150 --> 00:07:02,820
but I should make sure
that these games are

183
00:07:02,820 --> 00:07:03,640
clear to everyone.

184
00:07:03,640 --> 00:07:05,265
That's sort of what
happened last time.

185
00:07:09,210 --> 00:07:10,950
OK.

186
00:07:10,950 --> 00:07:14,260
Now, backing up
a little bit now,

187
00:07:14,260 --> 00:07:17,550
then the question is,
like, what is, in fact,

188
00:07:17,550 --> 00:07:19,020
social preferences?

189
00:07:19,020 --> 00:07:23,840
And so most economic
analysis in some way

190
00:07:23,840 --> 00:07:27,200
assumes away social
preferences and sort of assumes

191
00:07:27,200 --> 00:07:29,510
self-interest, very
narrowly defined.

192
00:07:29,510 --> 00:07:33,380
And so that means that
people essentially only care

193
00:07:33,380 --> 00:07:36,210
about their own interests
and their own outcomes,

194
00:07:36,210 --> 00:07:38,730
and then the market is sort
of taking care of the rest.

195
00:07:38,730 --> 00:07:42,530
And this is at the heart of
what Adam Smith originally

196
00:07:42,530 --> 00:07:44,708
was saying, I think, in
The Wealth of Nations,

197
00:07:44,708 --> 00:07:47,250
which is like, "It's not from
the benevolence of the butcher,

198
00:07:47,250 --> 00:07:50,400
the brewer, or the baker
that we expect our dinner,

199
00:07:50,400 --> 00:07:52,610
but from their regard
for their own interest.

200
00:07:52,610 --> 00:07:54,890
We address ourselves
not to their humanity,

201
00:07:54,890 --> 00:07:56,660
but to their self-love,
and never talk

202
00:07:56,660 --> 00:08:00,020
to them of our necessity,
but of their advantage."

203
00:08:00,020 --> 00:08:03,620
That is to say that when
everybody is very selfish

204
00:08:03,620 --> 00:08:06,800
and cares about themselves,
just comparative advantage

205
00:08:06,800 --> 00:08:08,630
and the ability to
trade with each other

206
00:08:08,630 --> 00:08:12,440
will make sure that other
people do useful things for us.

207
00:08:12,440 --> 00:08:15,350
So you can do the work
you do, make some money,

208
00:08:15,350 --> 00:08:16,610
and then go to the bakery.

209
00:08:16,610 --> 00:08:19,160
And the baker will give
you some bread in exchange

210
00:08:19,160 --> 00:08:22,048
for some money, not because
the baker likes you or wants

211
00:08:22,048 --> 00:08:24,590
to be friends with you or the
like, but because you pay them,

212
00:08:24,590 --> 00:08:26,060
and it's in their
interest to give you

213
00:08:26,060 --> 00:08:27,530
good bread because
you're going to pay them,

214
00:08:27,530 --> 00:08:28,820
and you're going to come
back perhaps and buy

215
00:08:28,820 --> 00:08:30,370
more bread in the future.

216
00:08:30,370 --> 00:08:32,370
So that's sort of a pretty
reasonable assumption

217
00:08:32,370 --> 00:08:34,789
in saying people are doing
useful things for each other

218
00:08:34,789 --> 00:08:36,450
as part of a market transaction.

219
00:08:36,450 --> 00:08:37,880
And that's what a
lot of economics

220
00:08:37,880 --> 00:08:41,299
has assumed and done
for a long time.

221
00:08:41,299 --> 00:08:44,030
Now, that is not
a bad assumption

222
00:08:44,030 --> 00:08:46,430
at all in the sense
of, like, it's actually

223
00:08:46,430 --> 00:08:50,330
pretty realistic in various
ways in thinking about humans.

224
00:08:50,330 --> 00:08:54,140
The question is not
necessarily whether there

225
00:08:54,140 --> 00:08:56,820
are some instances of whether
people care about others.

226
00:08:56,820 --> 00:08:59,240
The question that we're
going to ask is, like, well,

227
00:08:59,240 --> 00:09:02,730
what do we miss by ignoring
social preferences?

228
00:09:02,730 --> 00:09:06,290
And so it's pretty
clear that preferences

229
00:09:06,290 --> 00:09:08,678
depart from pure self-interest
in non-trivial ways.

230
00:09:08,678 --> 00:09:10,970
And the question we're going
to ask is, like, are there

231
00:09:10,970 --> 00:09:12,553
some concrete settings
where we really

232
00:09:12,553 --> 00:09:14,670
care about this and
sort of say, well,

233
00:09:14,670 --> 00:09:16,350
we should model
things differently?

234
00:09:16,350 --> 00:09:20,120
And if we do that, we can
understand certain phenomena

235
00:09:20,120 --> 00:09:20,905
better.

236
00:09:20,905 --> 00:09:22,280
So the goal is
here to understand

237
00:09:22,280 --> 00:09:24,155
how common and important
these departures are

238
00:09:24,155 --> 00:09:28,280
and what their nature is, like,
how can we best model these.

239
00:09:28,280 --> 00:09:34,217
And here's a nice example
of perhaps how one

240
00:09:34,217 --> 00:09:36,050
might want to think
about social preferences

241
00:09:36,050 --> 00:09:38,930
and how perhaps people think
about social preferences.

242
00:09:38,930 --> 00:09:39,920
So this is a picture--

243
00:09:39,920 --> 00:09:42,380
I don't know how well
you can see this--

244
00:09:42,380 --> 00:09:47,450
taken in Hawaii,
which is a box where

245
00:09:47,450 --> 00:09:48,590
banana bread is being sold.

246
00:09:48,590 --> 00:09:50,190
It's freely available
to everybody,

247
00:09:50,190 --> 00:09:51,540
so you can just take it.

248
00:09:51,540 --> 00:09:53,720
I think there's some
price tag somewhere.

249
00:09:53,720 --> 00:09:58,550
There's also a lockbox next to
it on the left of the picture

250
00:09:58,550 --> 00:10:00,128
where you can deposit payments.

251
00:10:00,128 --> 00:10:01,670
So the banana bread
costs some money.

252
00:10:01,670 --> 00:10:04,460
You can deposit your payments.

253
00:10:04,460 --> 00:10:07,447
There's nobody that
enforces it, but the lockbox

254
00:10:07,447 --> 00:10:09,280
is such that it's
actually locked in a sense

255
00:10:09,280 --> 00:10:12,420
that you can't
take away the box.

256
00:10:12,420 --> 00:10:14,660
So when you see
that picture, what

257
00:10:14,660 --> 00:10:18,260
perception of social preferences
or humanity does that reflect?

258
00:10:18,260 --> 00:10:21,430
How do we think
about this picture?

259
00:10:21,430 --> 00:10:22,184
Yes?

260
00:10:22,184 --> 00:10:24,851
AUDIENCE: That we're going to be
honest and give the money based

261
00:10:24,851 --> 00:10:26,212
on how much [INAUDIBLE].

262
00:10:26,212 --> 00:10:26,920
PROFESSOR: Right.

263
00:10:26,920 --> 00:10:28,450
So one assumption,
to some degree,

264
00:10:28,450 --> 00:10:31,190
is, like, well, people seem
pretty nice in general.

265
00:10:31,190 --> 00:10:33,483
It seems in a sense
of, like, we allow

266
00:10:33,483 --> 00:10:34,900
them to just take
the banana bread

267
00:10:34,900 --> 00:10:37,940
and run away and
not put in anything.

268
00:10:37,940 --> 00:10:39,370
We sort of ask them nicely.

269
00:10:39,370 --> 00:10:41,680
It costs, like, $1
or $5 or whatever.

270
00:10:41,680 --> 00:10:42,940
Please put in the money.

271
00:10:42,940 --> 00:10:45,190
It seems like the
conception to some degree

272
00:10:45,190 --> 00:10:49,550
is, like, people are going
to be nice and just do that.

273
00:10:49,550 --> 00:10:51,240
Is that all, or is
there something else?

274
00:10:51,240 --> 00:10:52,320
Yes.

275
00:10:52,320 --> 00:10:54,060
AUDIENCE: Maybe
there's an expectation

276
00:10:54,060 --> 00:10:56,750
not that everyone would
do the right thing,

277
00:10:56,750 --> 00:11:03,120
but that the loss from
some people who would just

278
00:11:03,120 --> 00:11:07,310
take the bread and run will be
maybe less than, for example,

279
00:11:07,310 --> 00:11:09,260
the cost of labor or
staffing the booth.

280
00:11:09,260 --> 00:11:09,980
PROFESSOR: Right.

281
00:11:09,980 --> 00:11:10,490
Exactly.

282
00:11:10,490 --> 00:11:12,645
So there's some--
the expectation,

283
00:11:12,645 --> 00:11:15,020
to be more precise, is not
everybody is going to be nice,

284
00:11:15,020 --> 00:11:18,020
but on average, most people
are going to be pretty nice

285
00:11:18,020 --> 00:11:21,950
or adhering to this normal rule
that you should pay some money.

286
00:11:21,950 --> 00:11:23,990
And some people might
run, but as you say,

287
00:11:23,990 --> 00:11:27,080
it might not be worth standing
there in the heat all day

288
00:11:27,080 --> 00:11:31,670
to make up for that.

289
00:11:31,670 --> 00:11:32,720
Exactly.

290
00:11:32,720 --> 00:11:34,440
And then what else is there?

291
00:11:34,440 --> 00:11:34,940
Yes.

292
00:11:34,940 --> 00:11:36,398
AUDIENCE: It gives
a perception of,

293
00:11:36,398 --> 00:11:38,594
like, you don't expect
them to steal banana bread,

294
00:11:38,594 --> 00:11:41,080
but you expect them to steal
the money, so you lock it.

295
00:11:41,080 --> 00:11:42,500
PROFESSOR: Right,
there's a lock.

296
00:11:42,500 --> 00:11:44,417
So it's like, in some
sense, you sort of think

297
00:11:44,417 --> 00:11:47,180
that, on the one hand, the
assumption is most people are

298
00:11:47,180 --> 00:11:48,690
going to be pretty nice.

299
00:11:48,690 --> 00:11:53,150
But if we allow them to run
away with all the money,

300
00:11:53,150 --> 00:11:55,940
at least there are some
people who, if allowed,

301
00:11:55,940 --> 00:11:58,320
would not be nice at all.

302
00:11:58,320 --> 00:12:00,403
In fact, they would take
away the money and not

303
00:12:00,403 --> 00:12:01,820
only not pay for
the banana bread,

304
00:12:01,820 --> 00:12:04,920
but also take away the lockbox.

305
00:12:04,920 --> 00:12:09,877
So that's exactly what
I've written down here,

306
00:12:09,877 --> 00:12:11,960
which is to say people in
general are nice enough.

307
00:12:11,960 --> 00:12:14,750
Most will probably pay
for the banana bread,

308
00:12:14,750 --> 00:12:17,870
so it's not worth sort of
standing around and ensuring

309
00:12:17,870 --> 00:12:18,590
that.

310
00:12:18,590 --> 00:12:20,840
But people also can be selfish.

311
00:12:20,840 --> 00:12:23,060
At least, some people
will be quite selfish.

312
00:12:23,060 --> 00:12:24,800
So if a lot of cash
were easy to take,

313
00:12:24,800 --> 00:12:26,360
someone at least would take it.

314
00:12:26,360 --> 00:12:28,878
Notice that's not an assumption
about everybody, or even

315
00:12:28,878 --> 00:12:29,420
the majority.

316
00:12:29,420 --> 00:12:31,295
It's just to say there's
a fraction of people

317
00:12:31,295 --> 00:12:36,850
who, if given the opportunity,
might take away our money

318
00:12:36,850 --> 00:12:38,948
and run from it.

319
00:12:38,948 --> 00:12:40,490
So that's actually
a pretty good sort

320
00:12:40,490 --> 00:12:44,150
of perspective for thinking
about social preferences, which

321
00:12:44,150 --> 00:12:46,880
is to say that
self-interest is probably

322
00:12:46,880 --> 00:12:50,990
a major driver of behavior
in many economic contexts.

323
00:12:50,990 --> 00:12:53,240
In some situations,
self-interest is not

324
00:12:53,240 --> 00:12:54,570
the main motive, in some sense.

325
00:12:54,570 --> 00:12:56,840
Again, we don't need
to necessarily enforce

326
00:12:56,840 --> 00:12:57,720
this for everybody.

327
00:12:57,720 --> 00:12:59,030
People will be quite nice.

328
00:12:59,030 --> 00:13:02,793
But there will be-- some
seemingly small departures

329
00:13:02,793 --> 00:13:04,460
from pure self-interest
can dramatically

330
00:13:04,460 --> 00:13:06,050
influence economic outcomes.

331
00:13:06,050 --> 00:13:10,550
That is to say, if we
couldn't lock the box,

332
00:13:10,550 --> 00:13:14,510
if you were not doing that,
somebody would take away

333
00:13:14,510 --> 00:13:15,020
this box.

334
00:13:15,020 --> 00:13:17,103
And now we couldn't offer
banana bread to anybody,

335
00:13:17,103 --> 00:13:19,100
and the whole transaction
would go down.

336
00:13:19,100 --> 00:13:24,150
Put differently, if we had
things like a theft or robbery

337
00:13:24,150 --> 00:13:26,150
and so on, if there's
some people who are really

338
00:13:26,150 --> 00:13:28,850
mean and essentially take
away all of our money,

339
00:13:28,850 --> 00:13:30,290
now we can't do any transaction.

340
00:13:30,290 --> 00:13:35,240
I can't have a shop trying
to sell stuff to people.

341
00:13:35,240 --> 00:13:39,110
This is the examples [INAUDIBLE]
positive and negative.

342
00:13:39,110 --> 00:13:42,620
Small things can actually make
huge differences for others.

343
00:13:42,620 --> 00:13:44,900
This is like helping a
stranger with directions

344
00:13:44,900 --> 00:13:47,180
or helping a stranger who
wants to go to the hospital

345
00:13:47,180 --> 00:13:47,450
and so on.

346
00:13:47,450 --> 00:13:49,450
That could make huge
differences in their lives,

347
00:13:49,450 --> 00:13:52,070
even if the-- or calling an
ambulance for somebody, that

348
00:13:52,070 --> 00:13:53,810
can make huge differences
people's lives,

349
00:13:53,810 --> 00:13:55,460
even though the
action from your side

350
00:13:55,460 --> 00:13:57,530
is actually relatively small.

351
00:13:57,530 --> 00:13:59,528
Helping a fellow student
with computer trouble

352
00:13:59,528 --> 00:14:01,070
or other issues that
they might have,

353
00:14:01,070 --> 00:14:04,280
helping somebody who
is sad on a given day

354
00:14:04,280 --> 00:14:05,910
feel better about themselves.

355
00:14:05,910 --> 00:14:08,540
In fact, as you might
know, it's the Random Acts

356
00:14:08,540 --> 00:14:13,010
of Kindness Week this week, so
doing random acts of kindness

357
00:14:13,010 --> 00:14:15,390
might actually make a huge
difference in people's lives.

358
00:14:15,390 --> 00:14:18,560
In fact, the problem
said, well, we'll

359
00:14:18,560 --> 00:14:21,230
ask you to do some
of that, in fact.

360
00:14:21,230 --> 00:14:23,990
Washing your hands is
particularly pertinent

361
00:14:23,990 --> 00:14:25,760
right now.

362
00:14:25,760 --> 00:14:28,610
Again, you might think you're
doing that to protect yourself,

363
00:14:28,610 --> 00:14:29,720
and surely you are.

364
00:14:29,720 --> 00:14:33,740
But in fact, there's also a huge
externality potentially that,

365
00:14:33,740 --> 00:14:37,160
like, by protecting
yourself, you protect others

366
00:14:37,160 --> 00:14:38,630
from potentially getting sick.

367
00:14:38,630 --> 00:14:42,510
And that could make a huge
difference in people's lives.

368
00:14:42,510 --> 00:14:43,740
Any questions on this?

369
00:14:49,010 --> 00:14:49,530
OK.

370
00:14:49,530 --> 00:14:51,530
So now we're getting back
to the ultimatum game.

371
00:14:51,530 --> 00:14:54,520
I'm going to ask the question
about how are people actually

372
00:14:54,520 --> 00:14:56,980
behaving in real situations.

373
00:14:56,980 --> 00:14:59,200
How do people behave in general?

374
00:14:59,200 --> 00:15:00,730
And perhaps, what
does game theory

375
00:15:00,730 --> 00:15:02,570
say how people should behave?

376
00:15:02,570 --> 00:15:04,700
How are people
behaving in reality,

377
00:15:04,700 --> 00:15:09,410
and how does that compare to how
you guys behaved here in class?

378
00:15:09,410 --> 00:15:12,650
So again, the ultimatum game,
you should be familiar with.

379
00:15:12,650 --> 00:15:14,770
There's a proposer, a responder.

380
00:15:14,770 --> 00:15:18,100
Proposer proposes
some split of the pie.

381
00:15:18,100 --> 00:15:20,990
Responder can accept or reject.

382
00:15:20,990 --> 00:15:24,850
Now, what does game theory
predict what people should do?

383
00:15:24,850 --> 00:15:27,323
Like, in the absence
of social preferences?

384
00:15:38,170 --> 00:15:39,020
Yes?

385
00:15:39,020 --> 00:15:40,400
AUDIENCE: [INAUDIBLE].

386
00:15:43,282 --> 00:15:43,990
PROFESSOR: Right.

387
00:15:43,990 --> 00:15:45,698
If you're the recipient,
you'll care only

388
00:15:45,698 --> 00:15:49,150
about your own output
or your own outcomes.

389
00:15:49,150 --> 00:15:51,370
Even if I offer
you $0.01 or $0.05

390
00:15:51,370 --> 00:15:53,680
or whatever, whatever is
worth the transaction cost,

391
00:15:53,680 --> 00:15:55,220
you're going to
accept everything.

392
00:15:55,220 --> 00:15:57,520
So if I'm playing with you,
I'm going to anticipate--

393
00:15:57,520 --> 00:15:59,500
I know that you will
accept essentially

394
00:15:59,500 --> 00:16:02,410
any offer that's positive.

395
00:16:02,410 --> 00:16:04,180
Even zero, you
might even accept,

396
00:16:04,180 --> 00:16:06,880
because it's sort of
like, why do you care?

397
00:16:06,880 --> 00:16:11,170
But you strictly prefer a
positive offer over no offer,

398
00:16:11,170 --> 00:16:12,800
so you're going to accept that.

399
00:16:12,800 --> 00:16:16,420
And then if I'm playing
with you, anticipating that,

400
00:16:16,420 --> 00:16:19,340
I will just give you essentially
the lowest possible amount,

401
00:16:19,340 --> 00:16:21,190
and that will be the outcomes.

402
00:16:21,190 --> 00:16:22,330
Right?

403
00:16:22,330 --> 00:16:23,590
So here it is.

404
00:16:23,590 --> 00:16:26,080
The prediction is very clear.

405
00:16:26,080 --> 00:16:27,970
The responder cares
only-- this is, again,

406
00:16:27,970 --> 00:16:30,220
game theory, but game
theory in the absence

407
00:16:30,220 --> 00:16:31,240
of social preferences.

408
00:16:31,240 --> 00:16:32,690
I should be more precise.

409
00:16:32,690 --> 00:16:35,920
So the responder cares
only about money,

410
00:16:35,920 --> 00:16:38,500
so she will accept
any x that's smaller

411
00:16:38,500 --> 00:16:41,230
than 0, any amount that's
given, because that's better

412
00:16:41,230 --> 00:16:43,150
than not having any money.

413
00:16:43,150 --> 00:16:46,580
The proposer will
understand that,

414
00:16:46,580 --> 00:16:49,670
will only care
about money as well,

415
00:16:49,670 --> 00:16:51,700
so offers as little as possible.

416
00:16:51,700 --> 00:16:55,060
The proposer doesn't care about
how the recipient is doing.

417
00:16:55,060 --> 00:16:57,130
So therefore, the
proposer will just

418
00:16:57,130 --> 00:17:04,213
offer as little as possible
to maximize their own payout.

419
00:17:04,213 --> 00:17:06,130
And then an equilibrium,
that will essentially

420
00:17:06,130 --> 00:17:09,640
be the zero or smallest
possible offer will be made,

421
00:17:09,640 --> 00:17:11,730
and that offer will be accepted.

422
00:17:11,730 --> 00:17:14,230
So there will be no rejections,
either because, essentially,

423
00:17:14,230 --> 00:17:16,930
there's no reason to
reject, and there will also

424
00:17:16,930 --> 00:17:19,990
be no large offers of the
like because there's no reason

425
00:17:19,990 --> 00:17:23,680
to give more than
minimal amounts.

426
00:17:23,680 --> 00:17:24,190
OK.

427
00:17:24,190 --> 00:17:26,859
So then what are typical
results in the ultimatum game?

428
00:17:26,859 --> 00:17:28,670
What do people actually find?

429
00:17:28,670 --> 00:17:33,310
Well, so most people offer
between 40% and 50% of the pie.

430
00:17:33,310 --> 00:17:35,230
Such offers are mostly accepted.

431
00:17:35,230 --> 00:17:37,000
The acceptance
rate is increasing

432
00:17:37,000 --> 00:17:39,970
in the offer, which is to
say if you offer, like, 10%,

433
00:17:39,970 --> 00:17:42,010
you have a pretty good
chance of being rejected.

434
00:17:42,010 --> 00:17:44,315
If you offer, like,
30%, 40%, 50%,

435
00:17:44,315 --> 00:17:45,940
you're much more
likely to be accepted.

436
00:17:45,940 --> 00:17:48,340
And you offer, like,
70%, 80%, almost surely

437
00:17:48,340 --> 00:17:49,570
you'll be accepted.

438
00:17:49,570 --> 00:17:52,600
And an offer that's below
20% are mostly rejected.

439
00:17:52,600 --> 00:17:53,230
Why is that?

440
00:17:53,230 --> 00:17:57,290
Or what's the reason
for that, presumably?

441
00:17:57,290 --> 00:17:58,638
Yes?

442
00:17:58,638 --> 00:18:00,120
AUDIENCE: [INAUDIBLE].

443
00:18:05,442 --> 00:18:06,150
PROFESSOR: Right.

444
00:18:06,150 --> 00:18:10,712
And why are you sort of upset
if somebody offers you 10%?

445
00:18:10,712 --> 00:18:12,920
AUDIENCE: Because it shows
they're disrespecting you.

446
00:18:12,920 --> 00:18:17,610
PROFESSOR: Yeah, something about
fairness or disrespect, or it's

447
00:18:17,610 --> 00:18:19,260
just not a nice thing to do.

448
00:18:19,260 --> 00:18:22,080
And now you'd rather
not have any money

449
00:18:22,080 --> 00:18:24,780
or not take $1 or whatever
they offer to you.

450
00:18:24,780 --> 00:18:26,280
You'd rather reject
that, so you're

451
00:18:26,280 --> 00:18:29,820
willing to give up $1
to essentially get back

452
00:18:29,820 --> 00:18:31,410
at the other person.

453
00:18:31,410 --> 00:18:32,190
Yeah?

454
00:18:32,190 --> 00:18:35,170
AUDIENCE: It could also
be that for some people,

455
00:18:35,170 --> 00:18:39,648
the thought of settling for
such small amounts of money

456
00:18:39,648 --> 00:18:43,660
is embarrassing, or it
feels like it's demeaning

457
00:18:43,660 --> 00:18:45,757
or [INAUDIBLE].

458
00:18:45,757 --> 00:18:47,340
PROFESSOR: Yeah, so
it's one of those.

459
00:18:47,340 --> 00:18:48,632
Some of it is-- yeah, exactly--

460
00:18:48,632 --> 00:18:50,670
I think, demeaning,
or just feels like--

461
00:18:50,670 --> 00:18:54,030
so one way to put this is
you have inequality aversion

462
00:18:54,030 --> 00:18:58,230
in some ways in the sense
of you just don't want to be

463
00:18:58,230 --> 00:19:00,360
the person who has $1 when
the other person has 9,

464
00:19:00,360 --> 00:19:03,190
or you don't want to have $0.10
when the other has, like, 9.90.

465
00:19:03,190 --> 00:19:07,010
So you really don't like
inequality in some ways.

466
00:19:07,010 --> 00:19:09,510
There are some versions of this
game where the computer sort

467
00:19:09,510 --> 00:19:11,708
of allocates these choices.

468
00:19:11,708 --> 00:19:13,125
And you can look
at how people are

469
00:19:13,125 --> 00:19:14,590
rejecting those offers as well.

470
00:19:14,590 --> 00:19:17,220
And they tend to not do
that, which sort of suggests

471
00:19:17,220 --> 00:19:21,960
that it's really coming
from some sense of people

472
00:19:21,960 --> 00:19:24,060
don't care so much about
the outcome, per se.

473
00:19:24,060 --> 00:19:26,970
They care about the
fairness or the intention

474
00:19:26,970 --> 00:19:28,252
of the other person.

475
00:19:28,252 --> 00:19:28,960
But you're right.

476
00:19:28,960 --> 00:19:31,680
There's other potential
considerations at play.

477
00:19:31,680 --> 00:19:34,600
Now, what do things
look like in our class?

478
00:19:34,600 --> 00:19:37,560
In fact, this looks
quite similar to what

479
00:19:37,560 --> 00:19:38,940
people do in general.

480
00:19:38,940 --> 00:19:40,590
I know you're very
special, but when

481
00:19:40,590 --> 00:19:42,250
it comes to the
ultimatum game, you're

482
00:19:42,250 --> 00:19:44,010
actually pretty ordinary.

483
00:19:44,010 --> 00:19:49,150
So what you see is, essentially,
most people offer 40%, 50%.

484
00:19:49,150 --> 00:19:51,900
Some people offer somewhat more.

485
00:19:51,900 --> 00:19:54,660
You see in-- you can't
see this that well.

486
00:19:54,660 --> 00:19:56,970
You see it in red
are the rejections.

487
00:19:56,970 --> 00:19:59,400
So rejections tend to
be on the left side,

488
00:19:59,400 --> 00:20:05,670
in particular when people
offer between 1% and 10%.

489
00:20:05,670 --> 00:20:08,130
There seem to be around
some rejections, so 50%,

490
00:20:08,130 --> 00:20:09,930
or maybe it's just below 50%.

491
00:20:09,930 --> 00:20:13,620
But usually, if you look
at, essentially, about 50%,

492
00:20:13,620 --> 00:20:19,410
or 40%, 50%, 60%, people tend
to not reject any offers.

493
00:20:19,410 --> 00:20:20,460
Any questions on that?

494
00:20:24,580 --> 00:20:25,434
Yeah?

495
00:20:25,434 --> 00:20:28,030
AUDIENCE: Could it
also be that people

496
00:20:28,030 --> 00:20:29,590
are anticipating future games?

497
00:20:29,590 --> 00:20:32,170
Or maybe not [INAUDIBLE],,
but sending a signal,

498
00:20:32,170 --> 00:20:34,080
like there's any
similar [INAUDIBLE]

499
00:20:34,080 --> 00:20:35,583
be better off [INAUDIBLE]?

500
00:20:35,583 --> 00:20:37,250
PROFESSOR: Yes, that's
a great question.

501
00:20:37,250 --> 00:20:38,470
So I kind of--

502
00:20:38,470 --> 00:20:40,090
all of what I was
doing, and I should

503
00:20:40,090 --> 00:20:42,790
emphasize this a bit more--

504
00:20:42,790 --> 00:20:45,760
the way I implemented some
of these games in class,

505
00:20:45,760 --> 00:20:48,130
some experimental economists
would get a heart attack

506
00:20:48,130 --> 00:20:51,490
in terms of how sloppy that
was in a sense of, like,

507
00:20:51,490 --> 00:20:55,390
a lot of these games, usually
when I show you these results,

508
00:20:55,390 --> 00:20:57,790
these are usually very careful
games, where, essentially,

509
00:20:57,790 --> 00:20:58,900
usually it's anonymous.

510
00:20:58,900 --> 00:21:00,160
Usually it's a one-shot game.

511
00:21:00,160 --> 00:21:02,590
Usually it's very clear
that it's a one-shot game.

512
00:21:02,590 --> 00:21:04,180
You will just never
see that person.

513
00:21:04,180 --> 00:21:06,100
You never even interact
with that person.

514
00:21:06,100 --> 00:21:09,040
It'll be clear whether
it's public versus private,

515
00:21:09,040 --> 00:21:09,950
and so on.

516
00:21:09,950 --> 00:21:13,180
So usually in those games,
things are extremely clear.

517
00:21:13,180 --> 00:21:15,373
There is no strategic
interactions, and so on.

518
00:21:15,373 --> 00:21:17,290
All of these things are
essentially shut down.

519
00:21:17,290 --> 00:21:20,290
And then you can look
at, randomly vary,

520
00:21:20,290 --> 00:21:23,313
some of these characteristics
and look at deviations of that.

521
00:21:23,313 --> 00:21:23,980
So you're right.

522
00:21:23,980 --> 00:21:26,710
In our game, in fact,
you might sort of

523
00:21:26,710 --> 00:21:28,493
think that, let's
just be nice now,

524
00:21:28,493 --> 00:21:30,160
in part, because the
other person, maybe

525
00:21:30,160 --> 00:21:31,240
I play with that person again.

526
00:21:31,240 --> 00:21:33,430
And I wasn't really very
precise about that either.

527
00:21:33,430 --> 00:21:36,280
I think actually they're
all like randomly allocated.

528
00:21:36,280 --> 00:21:38,410
So in fact, that was
not a consideration,

529
00:21:38,410 --> 00:21:39,910
but it wasn't entirely clear.

530
00:21:39,910 --> 00:21:42,430
Or for example, some
people were selected,

531
00:21:42,430 --> 00:21:44,920
and it happened to be that
it was 50/50 selection.

532
00:21:44,920 --> 00:21:47,570
That may have affected others
in the class, and so on.

533
00:21:47,570 --> 00:21:49,510
So a lot of these things
were fairly sloppy

534
00:21:49,510 --> 00:21:51,700
in terms of the
actual implementation.

535
00:21:51,700 --> 00:21:54,220
In the real games, that
would be all shut down

536
00:21:54,220 --> 00:21:56,832
and done much more carefully.

537
00:21:56,832 --> 00:21:58,540
There were some other
questions, I think?

538
00:21:58,540 --> 00:22:00,820
No.

539
00:22:00,820 --> 00:22:01,540
OK.

540
00:22:01,540 --> 00:22:02,050
Yeah?

541
00:22:02,050 --> 00:22:04,090
AUDIENCE: Was this
just for the chocolate?

542
00:22:04,090 --> 00:22:06,310
PROFESSOR: I think this is--

543
00:22:06,310 --> 00:22:07,450
I looked at all of these.

544
00:22:07,450 --> 00:22:09,560
The ultimatum games all
looked very similar.

545
00:22:09,560 --> 00:22:12,460
So this is, I think, for--

546
00:22:12,460 --> 00:22:14,320
this is the money one.

547
00:22:14,320 --> 00:22:16,250
But even for others, it
looked fairly similar.

548
00:22:16,250 --> 00:22:19,060
I'll show you some
variation in the dictator

549
00:22:19,060 --> 00:22:24,160
game in a second, where we have,
like, do the stakes matter?

550
00:22:24,160 --> 00:22:26,470
Does it matter whether
it's private versus public?

551
00:22:26,470 --> 00:22:29,230
Does it matter whether
it's chocolate or money?

552
00:22:29,230 --> 00:22:30,910
And perhaps, why
might it matter?

553
00:22:30,910 --> 00:22:32,830
We'll get back to
that in a second.

554
00:22:32,830 --> 00:22:34,420
Turns out, it does matter.

555
00:22:34,420 --> 00:22:37,360
I'll say a little bit
about that in a second.

556
00:22:37,360 --> 00:22:38,136
Yeah?

557
00:22:38,136 --> 00:22:41,520
AUDIENCE: So the
fraction of rejection

558
00:22:41,520 --> 00:22:46,420
in the green [INAUDIBLE]
others, and then [INAUDIBLE]

559
00:22:46,420 --> 00:22:48,730
is much higher.

560
00:22:48,730 --> 00:22:51,617
The number of rejections
in, like, $1 to $10

561
00:22:51,617 --> 00:22:53,433
much higher than [INAUDIBLE]?

562
00:22:53,433 --> 00:22:54,100
PROFESSOR: Yeah.

563
00:22:54,100 --> 00:22:56,750
AUDIENCE: [INAUDIBLE] explain
the notion of fairness

564
00:22:56,750 --> 00:23:00,178
or the equitable distribution
of the pie you just mentioned,

565
00:23:00,178 --> 00:23:02,016
because [INAUDIBLE]
accept when they

566
00:23:02,016 --> 00:23:05,305
are being given 90% of the
pie, whereas they are just not

567
00:23:05,305 --> 00:23:07,912
ready to accept [INAUDIBLE].

568
00:23:07,912 --> 00:23:08,620
PROFESSOR: Right.

569
00:23:08,620 --> 00:23:09,578
So that's a good point.

570
00:23:09,578 --> 00:23:11,680
So you were saying
earlier, maybe some people

571
00:23:11,680 --> 00:23:16,828
have some issues about inequity
or inequality across people.

572
00:23:16,828 --> 00:23:18,370
So if that were the
only explanation,

573
00:23:18,370 --> 00:23:21,100
then you would rejections
not only at, like,

574
00:23:21,100 --> 00:23:25,240
if I give you $1, and I keep 9,
but also if they give you $9,

575
00:23:25,240 --> 00:23:27,370
and I keep 1.

576
00:23:27,370 --> 00:23:29,050
So that means there's
something else.

577
00:23:29,050 --> 00:23:33,070
I think the fairness, people
don't object to inequality.

578
00:23:33,070 --> 00:23:35,855
If I say, OK, I'll
give you a lot,

579
00:23:35,855 --> 00:23:38,230
and you have more than I do,
people don't object to that.

580
00:23:38,230 --> 00:23:41,740
People object to, if I keep a
lot and give you very little,

581
00:23:41,740 --> 00:23:45,130
presumably because I'm selfish,
and I'm not nice to you.

582
00:23:45,130 --> 00:23:47,510
So that's kind of the
sense of fairness.

583
00:23:47,510 --> 00:23:50,115
So it's a fairness that
goes beyond the outcomes.

584
00:23:50,115 --> 00:23:51,490
So we're going to
talk about this

585
00:23:51,490 --> 00:23:54,073
for a bit, which is to say you
could just look at the outcomes

586
00:23:54,073 --> 00:23:56,200
and say, how much is the
inequality of outcomes?

587
00:23:56,200 --> 00:23:57,928
And I care about that.

588
00:23:57,928 --> 00:23:59,470
But that doesn't
seem to be the case.

589
00:23:59,470 --> 00:24:00,928
What seems to be
rather the case is

590
00:24:00,928 --> 00:24:02,440
that people care
about intentions

591
00:24:02,440 --> 00:24:03,350
one way or the other.

592
00:24:03,350 --> 00:24:07,690
It's like, what was I thinking,
or was I a mean person when

593
00:24:07,690 --> 00:24:08,750
I was doing this?

594
00:24:08,750 --> 00:24:11,110
And if you get the sense
that I was not very nice,

595
00:24:11,110 --> 00:24:14,230
you might reject almost
regardless of what

596
00:24:14,230 --> 00:24:15,970
the outcome will be eventually.

597
00:24:15,970 --> 00:24:19,240
And of course, you will
reject less if it's

598
00:24:19,240 --> 00:24:21,530
more costly for you to do so.

599
00:24:21,530 --> 00:24:24,380
But perhaps it's also the
case if I do, like, 70-30,

600
00:24:24,380 --> 00:24:27,070
you might say, well,
that's maybe not nice,

601
00:24:27,070 --> 00:24:30,280
but it's also not
that unfair either.

602
00:24:30,280 --> 00:24:30,780
OK.

603
00:24:33,770 --> 00:24:34,600
Right.

604
00:24:34,600 --> 00:24:37,990
So now, one thing you can
ask is how do these games

605
00:24:37,990 --> 00:24:40,570
look like across countries?

606
00:24:40,570 --> 00:24:44,110
And as I said, this game
has been extensively played.

607
00:24:44,110 --> 00:24:46,060
It's a game that like
lots and lots of people

608
00:24:46,060 --> 00:24:49,030
have played in all
sorts of situations.

609
00:24:49,030 --> 00:24:53,505
And the patterns are remarkably
stable across countries.

610
00:24:53,505 --> 00:24:54,880
And this is also
the case-- like,

611
00:24:54,880 --> 00:24:58,690
every time I play this
in class, these patterns

612
00:24:58,690 --> 00:24:59,887
look exactly the same.

613
00:24:59,887 --> 00:25:01,720
So whether people are
playing in Pittsburgh,

614
00:25:01,720 --> 00:25:06,670
in Ljubljana, in Yogyakarta,
Tokyo, play all essentially

615
00:25:06,670 --> 00:25:09,460
the same way.

616
00:25:09,460 --> 00:25:13,090
And even increasing people's
stakes to, like, several months

617
00:25:13,090 --> 00:25:16,818
pay by running
essentially the experiment

618
00:25:16,818 --> 00:25:18,610
with poorer populations,
where you can pay,

619
00:25:18,610 --> 00:25:23,530
essentially, days' or weeks'
or even a month's pay,

620
00:25:23,530 --> 00:25:30,220
people seem to play very
similar across places?

621
00:25:30,220 --> 00:25:30,850
Why is that?

622
00:25:30,850 --> 00:25:32,830
It seems to be like
somehow rejecting

623
00:25:32,830 --> 00:25:35,350
becomes a lot more
costly when I offer you,

624
00:25:35,350 --> 00:25:37,600
like, a day's wage or the like.

625
00:25:37,600 --> 00:25:40,140
But why do you think it
looks fairly similar as well?

626
00:25:40,140 --> 00:25:42,730
Why do people still reject?

627
00:25:42,730 --> 00:25:43,571
Yes?

628
00:25:43,571 --> 00:25:45,735
AUDIENCE: If your
goal is to get back

629
00:25:45,735 --> 00:25:48,862
at the other person, if
the pool is just bigger,

630
00:25:48,862 --> 00:25:51,598
then you're hurting them
more by denying them.

631
00:25:51,598 --> 00:25:52,390
PROFESSOR: Exactly.

632
00:25:52,390 --> 00:25:53,470
So it's exactly as you said.

633
00:25:53,470 --> 00:25:54,670
There's two things going on.

634
00:25:54,670 --> 00:25:57,860
On the one hand, rejection
becomes really expensive to me.

635
00:25:57,860 --> 00:26:01,930
So if there's a month's wage,
and you give me 10% of that

636
00:26:01,930 --> 00:26:07,737
and keep 90% of that, rejecting
10% off a month's wage,

637
00:26:07,737 --> 00:26:09,320
that's three or four
days or whatever.

638
00:26:09,320 --> 00:26:12,070
That's really
costly for me to do.

639
00:26:12,070 --> 00:26:15,532
However, it's also really
mean of you to do that,

640
00:26:15,532 --> 00:26:16,990
and I can really
screw you over now

641
00:26:16,990 --> 00:26:20,590
by sort of rejecting that offer.

642
00:26:20,590 --> 00:26:23,590
So those things essentially seem
to be more or less [INAUDIBLE]..

643
00:26:23,590 --> 00:26:25,180
They go into
opposite directions.

644
00:26:25,180 --> 00:26:29,300
Empirically it seems to be
that cancels each other out.

645
00:26:29,300 --> 00:26:31,660
That's exactly right.

646
00:26:31,660 --> 00:26:36,370
So now, there's a very
interesting paper then

647
00:26:36,370 --> 00:26:41,120
that does this ultimatum game
in hunter-gatherer societies

648
00:26:41,120 --> 00:26:44,560
and asks the question
about, like, while it

649
00:26:44,560 --> 00:26:46,510
seems to be that
people behave fairly

650
00:26:46,510 --> 00:26:50,410
similar in these games in
industrialized societies--

651
00:26:50,410 --> 00:26:52,480
whether they're
relatively rich or not,

652
00:26:52,480 --> 00:26:54,770
people seem to behave
very similarly.

653
00:26:54,770 --> 00:26:57,220
But perhaps if you look at
hunter-gatherer society that

654
00:26:57,220 --> 00:26:59,350
are quite different in
various ways, perhaps

655
00:26:59,350 --> 00:27:02,870
there we can look at interesting
differences across places.

656
00:27:02,870 --> 00:27:06,850
And so on the one
hand, we see there's

657
00:27:06,850 --> 00:27:08,560
lots of uniformity of play.

658
00:27:08,560 --> 00:27:12,470
In lots of places, people
seem to behave very similarly.

659
00:27:12,470 --> 00:27:14,660
But there are some
important differences.

660
00:27:14,660 --> 00:27:19,240
And perhaps what this line
of research is doing--

661
00:27:19,240 --> 00:27:22,310
it's often called
cultural economics--

662
00:27:22,310 --> 00:27:24,520
is to try to see what
is the role of culture.

663
00:27:24,520 --> 00:27:27,220
And there's Nathan Nunn
and other people at Harvard

664
00:27:27,220 --> 00:27:30,700
are doing this in particular,
inspired by Joe Henrich, who

665
00:27:30,700 --> 00:27:34,870
is the leader of that, is
trying to look at, like,

666
00:27:34,870 --> 00:27:38,410
what is the role of culture in
shaping people's preferences?

667
00:27:38,410 --> 00:27:40,690
And so one thing they had
done-- and at the time,

668
00:27:40,690 --> 00:27:44,050
they looked at 15 different
hunter-gatherer societies.

669
00:27:44,050 --> 00:27:45,460
You can see them
here on the map.

670
00:27:45,460 --> 00:27:49,837
Some of them are
in South America.

671
00:27:49,837 --> 00:27:50,920
Some are in South America.

672
00:27:50,920 --> 00:27:52,960
Some are in Africa.

673
00:27:52,960 --> 00:27:57,700
And some are in Asia, Oceania.

674
00:27:57,700 --> 00:28:01,180
And so you could look at
these different societies.

675
00:28:01,180 --> 00:28:06,430
And by looking at the way
they are living and producing,

676
00:28:06,430 --> 00:28:09,490
you can try to think about
potential predictions of how

677
00:28:09,490 --> 00:28:13,600
they might behave and
social preferences behavior.

678
00:28:13,600 --> 00:28:15,980
And there's two examples
here that we have.

679
00:28:15,980 --> 00:28:22,150
One is the society that you
can see here, the Machiguenga,

680
00:28:22,150 --> 00:28:24,850
which are, I guess, in--

681
00:28:24,850 --> 00:28:25,976
which country is this?

682
00:28:25,976 --> 00:28:26,728
AUDIENCE: Peru.

683
00:28:26,728 --> 00:28:27,520
PROFESSOR: In Peru?

684
00:28:27,520 --> 00:28:28,300
AUDIENCE: Peru.

685
00:28:28,300 --> 00:28:31,300
PROFESSOR: Who are in
Peru in Latin America.

686
00:28:31,300 --> 00:28:32,080
You can see them.

687
00:28:32,080 --> 00:28:34,090
They essentially are
independent families.

688
00:28:34,090 --> 00:28:36,010
They do a lot of cash cropping.

689
00:28:36,010 --> 00:28:39,880
Their way of living appears to
be slash-and-burn agriculture,

690
00:28:39,880 --> 00:28:41,795
gathering food,
fishing, and hunting.

691
00:28:41,795 --> 00:28:44,170
So it's overall-- and you can
correct me if that's wrong.

692
00:28:44,170 --> 00:28:46,870
It seems to be fairly
individualistic in the sense

693
00:28:46,870 --> 00:28:48,640
of the way people produce.

694
00:28:48,640 --> 00:28:51,160
Essentially, they live
individualistically.

695
00:28:51,160 --> 00:28:53,830
They are not really relying
on a lot of cooperation

696
00:28:53,830 --> 00:28:59,590
in the way they produce
their food or their living.

697
00:28:59,590 --> 00:29:05,180
In contrast, the
Lamaleras in Indonesia,

698
00:29:05,180 --> 00:29:06,310
these are whale hunters.

699
00:29:06,310 --> 00:29:08,650
Essentially, when you
try to hunt a whale--

700
00:29:08,650 --> 00:29:09,850
I don't quite know.

701
00:29:09,850 --> 00:29:11,380
I imagine, at least--

702
00:29:11,380 --> 00:29:13,960
it requires lots of
cooperation in the sense of,

703
00:29:13,960 --> 00:29:16,090
like, people have
to work together

704
00:29:16,090 --> 00:29:23,620
by nature of essentially finding
food or hunting these whales.

705
00:29:23,620 --> 00:29:28,730
And so they on a daily basis are
really relying on cooperation.

706
00:29:28,730 --> 00:29:32,410
And so now this line of
research by Henrich and others

707
00:29:32,410 --> 00:29:38,530
is trying to see, by looking
at people's way of living

708
00:29:38,530 --> 00:29:40,390
and way of producing
on any given life,

709
00:29:40,390 --> 00:29:44,830
is that predictive of people's
behavior in ultimatum games?

710
00:29:44,830 --> 00:29:47,350
And what you see
is the following.

711
00:29:47,350 --> 00:29:49,790
I'll just show you
what results are.

712
00:29:49,790 --> 00:29:54,550
So the society in Peru,
the slash-and-burn

713
00:29:54,550 --> 00:29:57,850
horticulturalists, seem
to have little concerns

714
00:29:57,850 --> 00:30:01,360
for others outside of the
family or for social status.

715
00:30:01,360 --> 00:30:06,400
What you see essentially
is relatively low offers.

716
00:30:06,400 --> 00:30:10,240
Most of these offers
are being accepted.

717
00:30:10,240 --> 00:30:13,120
There's essentially very
low rejection rates.

718
00:30:13,120 --> 00:30:14,965
And that seems to
be essentially--

719
00:30:14,965 --> 00:30:19,370
and the rejections of
amounts that are below 20%

720
00:30:19,370 --> 00:30:20,950
is only, like, 1 out of 10.

721
00:30:20,950 --> 00:30:24,460
So it seems to be the
case, this society

722
00:30:24,460 --> 00:30:26,860
that looks a lot
more individualistic

723
00:30:26,860 --> 00:30:30,460
is, in fact, much closer
to perhaps the assumption,

724
00:30:30,460 --> 00:30:33,460
the neoclassical assumptions of
no social preferences mattering

725
00:30:33,460 --> 00:30:36,790
in the sense of
they seem to be--

726
00:30:36,790 --> 00:30:40,000
in the way, at least, they
play the ultimatum game,

727
00:30:40,000 --> 00:30:42,340
they seem to be happy
to offer low amounts,

728
00:30:42,340 --> 00:30:46,540
and they seem to also be not
upset or annoyed or getting

729
00:30:46,540 --> 00:30:49,780
back to others who
offer them low amounts.

730
00:30:49,780 --> 00:30:52,240
In contrast, if you
look at the Lamaleras,

731
00:30:52,240 --> 00:30:54,970
the whale hunting culture based
on strong and meticulously

732
00:30:54,970 --> 00:31:01,450
detailed corporation, what you
see is the mean offer is 57%,

733
00:31:01,450 --> 00:31:04,120
or the mode is 50%.

734
00:31:04,120 --> 00:31:06,610
There are quite
a few rejections,

735
00:31:06,610 --> 00:31:09,700
in particular, rejections
of small amounts.

736
00:31:09,700 --> 00:31:12,430
And that's to say, like,
there's really an expectation

737
00:31:12,430 --> 00:31:14,470
of people cooperating.

738
00:31:14,470 --> 00:31:16,690
And if you don't
cooperate, if you are not

739
00:31:16,690 --> 00:31:18,430
playing by those
kinds of rules, people

740
00:31:18,430 --> 00:31:21,007
are being punished accordingly.

741
00:31:21,007 --> 00:31:22,090
And there are some others.

742
00:31:22,090 --> 00:31:29,680
There's sort of another
culture of the Au/Gnau,

743
00:31:29,680 --> 00:31:32,710
which is a culture where gift
giving is an avenue to status

744
00:31:32,710 --> 00:31:35,837
and gives the right to ask
for reciprocity at any time.

745
00:31:35,837 --> 00:31:38,170
So that's the thing that you
often see in India as well,

746
00:31:38,170 --> 00:31:42,610
that people tend
to reject offers

747
00:31:42,610 --> 00:31:45,760
that are very large
in part because they

748
00:31:45,760 --> 00:31:46,990
think that's inappropriate.

749
00:31:46,990 --> 00:31:49,365
They think that's just a weird
thing to do in a sense of,

750
00:31:49,365 --> 00:31:51,550
like, I offer you,
like, 50% or 60%,

751
00:31:51,550 --> 00:31:53,890
they just think, like, if
I accept this offer, now

752
00:31:53,890 --> 00:31:58,360
I owe you something and
have to give you stuff back.

753
00:31:58,360 --> 00:32:00,034
You are nodding.

754
00:32:00,034 --> 00:32:00,930
AUDIENCE: Oh, yes.

755
00:32:00,930 --> 00:32:01,555
PROFESSOR: Yes.

756
00:32:01,555 --> 00:32:06,540
So that's in India as well
to some degree as far as--

757
00:32:06,540 --> 00:32:07,580
yeah.

758
00:32:07,580 --> 00:32:08,615
Any comments on this?

759
00:32:08,615 --> 00:32:09,115
Yeah.

760
00:32:09,115 --> 00:32:11,840
AUDIENCE: Is that last
experiment anonymous?

761
00:32:11,840 --> 00:32:14,240
Because if it's
anonymous, then there's

762
00:32:14,240 --> 00:32:18,102
no sense of this specific
person [INAUDIBLE]..

763
00:32:18,102 --> 00:32:19,560
PROFESSOR: That's
a great question.

764
00:32:19,560 --> 00:32:21,680
I think it may not have been.

765
00:32:21,680 --> 00:32:23,180
I need to check.

766
00:32:23,180 --> 00:32:27,260
I think-- so there's some
question on how closely in some

767
00:32:27,260 --> 00:32:28,010
sense to the--

768
00:32:30,650 --> 00:32:32,290
these are 15 societies.

769
00:32:32,290 --> 00:32:34,760
So in some sense,
these are not strictly

770
00:32:34,760 --> 00:32:37,940
speaking statistical
tests of, like,

771
00:32:37,940 --> 00:32:40,460
does culture cause
this behavior?

772
00:32:40,460 --> 00:32:42,830
Because you only have
essentially 15 observations

773
00:32:42,830 --> 00:32:46,260
of certain ways people behave.

774
00:32:46,260 --> 00:32:49,845
So in some sense, the mapping
between the description

775
00:32:49,845 --> 00:32:51,470
of what I was telling
you, which surely

776
00:32:51,470 --> 00:32:54,440
was inaccurate in terms
of how people really

777
00:32:54,440 --> 00:32:57,110
live in those societies,
and their behavior

778
00:32:57,110 --> 00:32:59,220
is more a descriptive mapping.

779
00:32:59,220 --> 00:33:02,480
So in some ways, it could be
that even if it were anonymous,

780
00:33:02,480 --> 00:33:04,790
people behave in certain
ways that reflects

781
00:33:04,790 --> 00:33:09,155
a culture of cooperation is to
say, it's just inappropriate--

782
00:33:09,155 --> 00:33:10,910
or reflect, essentially,
when they say,

783
00:33:10,910 --> 00:33:13,760
if somebody gives you a
large amount or a large gift,

784
00:33:13,760 --> 00:33:15,380
you're not supposed
to accept it.

785
00:33:15,380 --> 00:33:18,260
You might sort of not accept
it even in an anonymous game

786
00:33:18,260 --> 00:33:21,200
because you feel bad
about it afterwards

787
00:33:21,200 --> 00:33:26,150
because you can't reciprocate,
if that makes sense.

788
00:33:26,150 --> 00:33:29,000
But I have to look it up, what
the specific details were.

789
00:33:29,000 --> 00:33:31,700
The paper is also-- and
it's a very short paper.

790
00:33:31,700 --> 00:33:34,280
So it's also on the course
website if you want to look.

791
00:33:36,800 --> 00:33:39,440
Any other comments?

792
00:33:39,440 --> 00:33:41,710
Are you from Peru?

793
00:33:41,710 --> 00:33:42,960
Do you know about the society?

794
00:33:42,960 --> 00:33:45,200
Is that remotely accurate
what I was saying?

795
00:33:45,200 --> 00:33:46,980
AUDIENCE: I'm not really sure.

796
00:33:46,980 --> 00:33:49,290
PROFESSOR: All right, well,
you can look into this

797
00:33:49,290 --> 00:33:51,570
and report back.

798
00:33:51,570 --> 00:33:52,140
OK.

799
00:33:52,140 --> 00:33:55,260
But anyway, it's a very
interesting application

800
00:33:55,260 --> 00:33:58,410
of the ultimatum game because
it essentially tries to say,

801
00:33:58,410 --> 00:34:00,390
we're trying to look
at how do people

802
00:34:00,390 --> 00:34:01,590
relate in the real world.

803
00:34:01,590 --> 00:34:03,840
And looking at these
different societies

804
00:34:03,840 --> 00:34:05,580
gives you some
sense of, like, it

805
00:34:05,580 --> 00:34:08,820
does seem to capture something
in the real world of how

806
00:34:08,820 --> 00:34:12,670
people are, in fact, behaving.

807
00:34:12,670 --> 00:34:13,650
OK.

808
00:34:13,650 --> 00:34:17,040
So now we can sort
think about what's

809
00:34:17,040 --> 00:34:18,306
going on in ultimatum games.

810
00:34:18,306 --> 00:34:19,889
And we talked about
this a little bit,

811
00:34:19,889 --> 00:34:21,681
but I want to be a
little bit more precise.

812
00:34:21,681 --> 00:34:24,659
It's like, why do responders
reject low offers?

813
00:34:24,659 --> 00:34:26,413
What's going on here?

814
00:34:26,413 --> 00:34:27,330
What are people doing?

815
00:34:32,710 --> 00:34:33,433
Yes?

816
00:34:33,433 --> 00:34:35,620
AUDIENCE: [INAUDIBLE] spite.

817
00:34:35,620 --> 00:34:37,570
PROFESSOR: Yes, and
so the spite is--

818
00:34:37,570 --> 00:34:39,489
so what exactly are
you objecting to?

819
00:34:39,489 --> 00:34:43,690
I asked this before,
but let me ask again.

820
00:34:43,690 --> 00:34:46,664
AUDIENCE: So I think a lot of
things have already been said.

821
00:34:46,664 --> 00:34:51,750
If you will reject, you
think it's unfair, so much so

822
00:34:51,750 --> 00:34:53,719
that you would rather
that they get nothing,

823
00:34:53,719 --> 00:34:55,605
you both get nothing,
than [INAUDIBLE]

824
00:34:55,605 --> 00:34:57,032
take this pity offer.

825
00:34:57,032 --> 00:34:57,740
PROFESSOR: Right.

826
00:34:57,740 --> 00:34:58,910
So there's two things.

827
00:34:58,910 --> 00:35:01,920
Let me put them out here.

828
00:35:01,920 --> 00:35:02,920
There's two things here.

829
00:35:02,920 --> 00:35:04,390
You're talking about the
second thing, I think,

830
00:35:04,390 --> 00:35:05,680
which is the fairness part.

831
00:35:05,680 --> 00:35:09,460
So one is about a procedural
thing, which is to say,

832
00:35:09,460 --> 00:35:11,440
I object to procedure.

833
00:35:11,440 --> 00:35:15,430
I object to you not treating
me nicely or fairly.

834
00:35:15,430 --> 00:35:16,960
I'm just mad at
you for doing that.

835
00:35:16,960 --> 00:35:18,980
Therefore, I reject.

836
00:35:18,980 --> 00:35:20,500
Hypothesis number
one-- this is what

837
00:35:20,500 --> 00:35:22,690
we were talking about
earlier, which is to say,

838
00:35:22,690 --> 00:35:24,280
I just dislike being behind.

839
00:35:24,280 --> 00:35:26,150
You offer me-- you keep 6.

840
00:35:26,150 --> 00:35:26,830
I have 4.

841
00:35:26,830 --> 00:35:30,730
I just don't like having
less than somebody else,

842
00:35:30,730 --> 00:35:33,670
and I might reject that as well.

843
00:35:33,670 --> 00:35:36,580
Can we disentangle
those explanations?

844
00:35:41,100 --> 00:35:42,870
From dictator game data?

845
00:35:52,020 --> 00:35:52,808
Yes?

846
00:35:52,808 --> 00:35:57,390
AUDIENCE: You can
have some [INAUDIBLE]..

847
00:35:57,390 --> 00:36:03,030
I guess, reject [INAUDIBLE].

848
00:36:03,030 --> 00:36:03,810
PROFESSOR: Right.

849
00:36:03,810 --> 00:36:06,120
So from the data
that I just showed,

850
00:36:06,120 --> 00:36:08,760
or from the pure,
simple dictator game,

851
00:36:08,760 --> 00:36:10,380
it's very hard to
disentangle these

852
00:36:10,380 --> 00:36:12,990
because it looks the same.

853
00:36:12,990 --> 00:36:15,720
But what you're saying now is
we can have different versions

854
00:36:15,720 --> 00:36:19,185
of that where you could have
either the computer implement

855
00:36:19,185 --> 00:36:23,430
it, or you could choose for
other people, and so on.

856
00:36:23,430 --> 00:36:24,930
I think that's what
you were saying.

857
00:36:24,930 --> 00:36:27,250
And then from that,
you could say, well,

858
00:36:27,250 --> 00:36:30,720
if you're also rejecting
that, then perhaps it's

859
00:36:30,720 --> 00:36:32,550
not about the procedure per se.

860
00:36:32,550 --> 00:36:36,270
If the computer chose 6 and
4 and you reject it anyway,

861
00:36:36,270 --> 00:36:38,370
then it's not to
do with fairness,

862
00:36:38,370 --> 00:36:40,800
because the computer was
perhaps just randomizing.

863
00:36:40,800 --> 00:36:42,750
But it's rather to
do with, like, I

864
00:36:42,750 --> 00:36:46,660
don't like to have less than
somebody else in the same game.

865
00:36:46,660 --> 00:36:48,397
Right?

866
00:36:48,397 --> 00:36:48,980
AUDIENCE: Yes.

867
00:36:48,980 --> 00:36:50,480
So if you used a
computer, you would

868
00:36:50,480 --> 00:36:54,530
be able to [INAUDIBLE] just
disliking [INAUDIBLE] less.

869
00:36:54,530 --> 00:37:01,416
If you have an individual
respond to [INAUDIBLE] accept

870
00:37:01,416 --> 00:37:04,083
or reject an offer that
somebody else made,

871
00:37:04,083 --> 00:37:05,215
it's quite [INAUDIBLE]--

872
00:37:05,215 --> 00:37:05,360
PROFESSOR: Right.

873
00:37:05,360 --> 00:37:07,813
AUDIENCE: Then you would
see whether they were just

874
00:37:07,813 --> 00:37:08,722
trying to punish.

875
00:37:08,722 --> 00:37:09,430
PROFESSOR: I see.

876
00:37:09,430 --> 00:37:10,130
Yes, I see.

877
00:37:10,130 --> 00:37:10,630
I see.

878
00:37:10,630 --> 00:37:12,170
So yes, you can do that.

879
00:37:12,170 --> 00:37:14,780
There's a bit of an
issue, a little bit like--

880
00:37:14,780 --> 00:37:19,250
another version of that would
be just I don't like inequality,

881
00:37:19,250 --> 00:37:21,590
but I just might want to--

882
00:37:21,590 --> 00:37:25,082
I just really like 5-5
or over 50-50 outcomes.

883
00:37:25,082 --> 00:37:27,540
There, I think we can look at
what you were saying earlier,

884
00:37:27,540 --> 00:37:31,010
which is to say I could see
if there's a third party that

885
00:37:31,010 --> 00:37:34,160
looks at somebody
choosing 60 for themselves

886
00:37:34,160 --> 00:37:37,700
and 40 for somebody else,
versus somebody choosing 40

887
00:37:37,700 --> 00:37:39,860
for themselves and
60 for somebody else.

888
00:37:39,860 --> 00:37:41,510
If you really
dislike inequality,

889
00:37:41,510 --> 00:37:43,970
you would just reject
both of those things.

890
00:37:43,970 --> 00:37:48,230
And if you sort of
instead think this

891
00:37:48,230 --> 00:37:50,700
is like-- you
dislike inequality.

892
00:37:50,700 --> 00:37:55,940
If the person only does that for
themselves in some selfish way,

893
00:37:55,940 --> 00:38:00,660
you would only reject the
60-40, but not the 40-60.

894
00:38:00,660 --> 00:38:01,160
Great.

895
00:38:04,990 --> 00:38:05,490
Right.

896
00:38:05,490 --> 00:38:09,553
So the answer is no, we cannot
distinguish these motives based

897
00:38:09,553 --> 00:38:10,470
on the ultimatum game.

898
00:38:10,470 --> 00:38:13,290
So we need some additional
evidence to do that.

899
00:38:13,290 --> 00:38:15,750
We're going to do
that mostly next time.

900
00:38:15,750 --> 00:38:19,180
We're going to try to make
some progress towards that.

901
00:38:19,180 --> 00:38:22,380
Now, if you now think about
the proposer's motive--

902
00:38:22,380 --> 00:38:25,230
if a proposer gives a large
or a small amount, what can

903
00:38:25,230 --> 00:38:26,100
we learn from that?

904
00:38:26,100 --> 00:38:28,767
Or why is it difficult to figure
out what the proposal is doing?

905
00:38:32,460 --> 00:38:34,800
If you see somebody gives
a large amount of money,

906
00:38:34,800 --> 00:38:36,200
what have you learned from that?

907
00:38:36,200 --> 00:38:36,934
Yeah?

908
00:38:36,934 --> 00:38:39,825
AUDIENCE: It could be that
they're just being optimistic,

909
00:38:39,825 --> 00:38:44,040
or it could be that they
have beliefs about what

910
00:38:44,040 --> 00:38:48,347
the responder would reject
versus what the responder would

911
00:38:48,347 --> 00:38:49,048
not reject.

912
00:38:49,048 --> 00:38:51,138
And they could
reject [INAUDIBLE]..

913
00:38:51,138 --> 00:38:51,930
PROFESSOR: Exactly.

914
00:38:51,930 --> 00:38:54,087
So it could be that I'm
just trying to be nice.

915
00:38:54,087 --> 00:38:55,170
I'm trying to be friendly.

916
00:38:55,170 --> 00:38:55,980
I care about you.

917
00:38:55,980 --> 00:39:00,420
I offer you a 50-50 because
I want you to have half.

918
00:39:00,420 --> 00:39:02,010
Or it could be that
I just actually

919
00:39:02,010 --> 00:39:03,420
don't care about you at all.

920
00:39:03,420 --> 00:39:05,753
I just think if I
offer you only 30 or 20

921
00:39:05,753 --> 00:39:07,170
or whatever, there's
a good chance

922
00:39:07,170 --> 00:39:08,610
that you might reject it.

923
00:39:08,610 --> 00:39:12,510
Therefore, I just
optimize over my--

924
00:39:12,510 --> 00:39:14,490
so it depends on my
beliefs about what

925
00:39:14,490 --> 00:39:16,110
the rejection rates are.

926
00:39:16,110 --> 00:39:20,790
And if the beliefs are that
below 30% or 40% people's

927
00:39:20,790 --> 00:39:23,520
rejection rates are
positive and sort of higher,

928
00:39:23,520 --> 00:39:25,770
I might essentially just
maximize depending on my risk

929
00:39:25,770 --> 00:39:26,858
preferences or whatever.

930
00:39:26,858 --> 00:39:27,900
Suppose I'm risk-neutral.

931
00:39:27,900 --> 00:39:30,930
I might just maximize
expected payout.

932
00:39:30,930 --> 00:39:33,240
It might be that
essentially 40% or 50%

933
00:39:33,240 --> 00:39:37,320
is just the optimal solution
to that maximization problem,

934
00:39:37,320 --> 00:39:40,240
and I just don't care
about you whatsoever.

935
00:39:40,240 --> 00:39:42,210
So if I'm looking at this
behavior essentially,

936
00:39:42,210 --> 00:39:44,670
if I'm looking at
an ultimatum game

937
00:39:44,670 --> 00:39:48,120
and looking at a proposer
offering large amounts

938
00:39:48,120 --> 00:39:50,730
or positive amounts, you can
learn nothing about their--

939
00:39:50,730 --> 00:39:52,200
or it's very hard
to learn anything

940
00:39:52,200 --> 00:39:55,110
about their social preferences,
because essentially there's

941
00:39:55,110 --> 00:39:58,860
strategic motives
potentially involved as well.

942
00:39:58,860 --> 00:39:59,790
Is that clear?

943
00:39:59,790 --> 00:40:01,154
Or any questions?

944
00:40:05,050 --> 00:40:06,300
So what do we need to do then?

945
00:40:06,300 --> 00:40:11,260
Well, now we need to sort
of have some other evidence.

946
00:40:11,260 --> 00:40:13,110
So we talked about this already.

947
00:40:13,110 --> 00:40:15,030
When you think about
social preferences,

948
00:40:15,030 --> 00:40:18,120
you can think about three broad
categories of preferences.

949
00:40:18,120 --> 00:40:20,410
And we're going to talk
about this a lot more.

950
00:40:20,410 --> 00:40:22,530
So one is distributional
preferences.

951
00:40:22,530 --> 00:40:26,100
This is to say that you
care about outcomes.

952
00:40:26,100 --> 00:40:28,245
You care about how
much each person gets.

953
00:40:28,245 --> 00:40:30,390
So that's to say we can
sort of represent this to,

954
00:40:30,390 --> 00:40:33,450
like, I put some weight on
my utility, on my outcomes.

955
00:40:33,450 --> 00:40:35,310
I put some weight
on your outcomes.

956
00:40:35,310 --> 00:40:37,230
And then depending on
what that weight is,

957
00:40:37,230 --> 00:40:40,360
I'm going to decide about
how to distribute outcomes.

958
00:40:40,360 --> 00:40:42,443
These are sort of
distributional preferences.

959
00:40:42,443 --> 00:40:43,860
Then there are
things like what we

960
00:40:43,860 --> 00:40:46,470
want to call face-saving
concerns, which

961
00:40:46,470 --> 00:40:51,120
is people don't want to
look bad in front of others.

962
00:40:51,120 --> 00:40:53,370
You can also think about
this like social image, which

963
00:40:53,370 --> 00:40:56,250
is to say, I want to
look like I'm a nice guy.

964
00:40:56,250 --> 00:40:59,070
So in particular, if I'm in
public making certain choices,

965
00:40:59,070 --> 00:41:01,650
if I'm being observed,
I'm going to behave

966
00:41:01,650 --> 00:41:03,630
in a very friendly way.

967
00:41:03,630 --> 00:41:07,840
If I could secretly be
mean, I would do that.

968
00:41:07,840 --> 00:41:10,170
But as long as my social
image is at stake,

969
00:41:10,170 --> 00:41:11,810
I might look like a nice person.

970
00:41:11,810 --> 00:41:14,310
We're going to talk about some
experimental evidence looking

971
00:41:14,310 --> 00:41:15,580
at that.

972
00:41:15,580 --> 00:41:17,820
And then there's some
which we want call

973
00:41:17,820 --> 00:41:20,260
intentions-based preferences.

974
00:41:20,260 --> 00:41:21,930
These are things
like reciprocity,

975
00:41:21,930 --> 00:41:27,055
procedural justice, and other
facets of social preferences.

976
00:41:27,055 --> 00:41:28,680
Broadly speaking,
you can think of this

977
00:41:28,680 --> 00:41:32,580
as fairness, which is
to say people don't care

978
00:41:32,580 --> 00:41:34,440
about the outcomes, per se.

979
00:41:34,440 --> 00:41:38,310
They care about essentially
the process in which

980
00:41:38,310 --> 00:41:40,500
the outcomes were generated.

981
00:41:40,500 --> 00:41:43,080
So they might be perfectly
happy with one person getting

982
00:41:43,080 --> 00:41:45,660
a lot and another person
getting nothing as

983
00:41:45,660 --> 00:41:47,790
long as the computer
randomized and [INAUDIBLE],,

984
00:41:47,790 --> 00:41:51,040
the probability of getting money
was the same for everybody.

985
00:41:51,040 --> 00:41:54,270
So that would be
somebody who said--

986
00:41:54,270 --> 00:41:57,570
these would be intention-based
preferences, where people say,

987
00:41:57,570 --> 00:41:58,800
this is procedural justice.

988
00:41:58,800 --> 00:42:00,060
This is a fair thing to do.

989
00:42:03,240 --> 00:42:07,620
And people are
motivated by that,

990
00:42:07,620 --> 00:42:11,100
or essentially, they
might reciprocate.

991
00:42:11,100 --> 00:42:13,680
In cases when they
think things are unfair,

992
00:42:13,680 --> 00:42:15,210
they might not do
so in case they

993
00:42:15,210 --> 00:42:17,850
think things are
fairly generated, even

994
00:42:17,850 --> 00:42:21,420
if things are unequal ex-post.

995
00:42:21,420 --> 00:42:22,350
OK.

996
00:42:22,350 --> 00:42:25,300
So now we're going to talk
about all three of those.

997
00:42:25,300 --> 00:42:27,420
So let's start with
distributional preferences.

998
00:42:27,420 --> 00:42:30,180
This is, again, the simplest
kinds of social preferences.

999
00:42:30,180 --> 00:42:32,700
This is very much
a natural extension

1000
00:42:32,700 --> 00:42:35,350
of how economists
think about the world.

1001
00:42:35,350 --> 00:42:36,840
This is going back
to Gary Becker,

1002
00:42:36,840 --> 00:42:39,750
I think, in the '60s, where
essentially, instead of saying,

1003
00:42:39,750 --> 00:42:43,170
does a utility function that
just has my own outcomes,

1004
00:42:43,170 --> 00:42:45,330
as my own consumption
as an argument,

1005
00:42:45,330 --> 00:42:48,120
instead, have a utility function
that has my own utility,

1006
00:42:48,120 --> 00:42:50,892
how much I get, a
function of that,

1007
00:42:50,892 --> 00:42:52,350
plus how much does
the other person

1008
00:42:52,350 --> 00:42:54,940
get, and a weight on
that other person.

1009
00:42:54,940 --> 00:42:58,830
And so again, we
sort of look at,

1010
00:42:58,830 --> 00:43:01,320
like, the person cares not
about only how much they get,

1011
00:43:01,320 --> 00:43:02,192
but also the other.

1012
00:43:02,192 --> 00:43:03,900
There's two versions
of that, and this is

1013
00:43:03,900 --> 00:43:05,108
what you were saying earlier.

1014
00:43:05,108 --> 00:43:07,170
There are interested
distributional preferences

1015
00:43:07,170 --> 00:43:09,690
and disinterested
distributional preferences.

1016
00:43:09,690 --> 00:43:12,000
The interested one
is essentially--

1017
00:43:12,000 --> 00:43:14,040
the version is, like,
where I'm involved,

1018
00:43:14,040 --> 00:43:16,440
and I get a certain
outcome, and there's

1019
00:43:16,440 --> 00:43:18,540
another person involved,
and I put some weight

1020
00:43:18,540 --> 00:43:19,800
on that other person.

1021
00:43:19,800 --> 00:43:21,600
So it's about me
versus another person,

1022
00:43:21,600 --> 00:43:25,440
and how these resources
are being divided.

1023
00:43:25,440 --> 00:43:28,080
Could be also multiple
people, of course.

1024
00:43:28,080 --> 00:43:29,700
And then there's a
disinterested part,

1025
00:43:29,700 --> 00:43:32,700
which is to say, how
do I want resources

1026
00:43:32,700 --> 00:43:35,160
to be divided across others?

1027
00:43:35,160 --> 00:43:38,220
This is questions like, how
do we want society to look?

1028
00:43:38,220 --> 00:43:42,510
How do people feel about
inequality and the like?

1029
00:43:42,510 --> 00:43:45,180
Or, like what public goods--
how much should the government

1030
00:43:45,180 --> 00:43:47,460
engage in redistribution
and so on?

1031
00:43:47,460 --> 00:43:49,550
These would be often--

1032
00:43:49,550 --> 00:43:51,300
to some degree, of
course, it's interested

1033
00:43:51,300 --> 00:43:56,340
because people have to pay
taxes or receive benefits

1034
00:43:56,340 --> 00:43:57,390
from the government.

1035
00:43:57,390 --> 00:43:59,100
But often, it's you
can ask questions

1036
00:43:59,100 --> 00:44:00,600
people have opinions about.

1037
00:44:00,600 --> 00:44:01,920
Suppose you live in the US.

1038
00:44:01,920 --> 00:44:04,470
You could have opinions about
how should Europe's tax system

1039
00:44:04,470 --> 00:44:05,220
look like.

1040
00:44:05,220 --> 00:44:06,780
What is fair and what is unfair?

1041
00:44:06,780 --> 00:44:09,600
Those would be, essentially,
disinterested distributional

1042
00:44:09,600 --> 00:44:10,705
preferences?

1043
00:44:13,820 --> 00:44:16,315
Any questions on that?

1044
00:44:16,315 --> 00:44:17,690
So we're going to
start with this

1045
00:44:17,690 --> 00:44:21,020
and see how far can you get
distributional preferences.

1046
00:44:21,020 --> 00:44:22,730
What kinds of things
can be explained?

1047
00:44:22,730 --> 00:44:24,950
And then we're going
to move towards saving

1048
00:44:24,950 --> 00:44:29,310
face or social image
type of concerns.

1049
00:44:29,310 --> 00:44:31,970
So the dictator game, I
already explained to you

1050
00:44:31,970 --> 00:44:33,810
what it looks like.

1051
00:44:33,810 --> 00:44:39,920
So here's what the dictator
game looked like in class

1052
00:44:39,920 --> 00:44:41,190
with no stakes.

1053
00:44:41,190 --> 00:44:42,860
This is very typical
to how people

1054
00:44:42,860 --> 00:44:46,380
behave in the real world.

1055
00:44:46,380 --> 00:44:47,440
What do we see here?

1056
00:44:47,440 --> 00:44:52,160
What are patterns of behaviors?

1057
00:44:52,160 --> 00:44:54,800
This is no stakes, and it's just
hypothetical questions here.

1058
00:45:06,670 --> 00:45:07,240
Yeah?

1059
00:45:07,240 --> 00:45:08,782
AUDIENCE: There are
two large groups.

1060
00:45:08,782 --> 00:45:11,080
One group that
doesn't give anything,

1061
00:45:11,080 --> 00:45:14,468
and a group that gives half.

1062
00:45:14,468 --> 00:45:15,260
PROFESSOR: Exactly.

1063
00:45:15,260 --> 00:45:19,690
And this is very typical,
again, in many situations.

1064
00:45:19,690 --> 00:45:21,220
There's a group
of people, often,

1065
00:45:21,220 --> 00:45:24,670
like, 30%, 40% of people
who just essentially keep

1066
00:45:24,670 --> 00:45:26,350
the money.

1067
00:45:26,350 --> 00:45:28,300
There's a group of
people that thinks

1068
00:45:28,300 --> 00:45:30,620
50-50 is the right thing to do.

1069
00:45:30,620 --> 00:45:33,250
That's often also something,
like, 30% of people.

1070
00:45:33,250 --> 00:45:35,320
And then there's people
often in between,

1071
00:45:35,320 --> 00:45:36,970
often between 0 and 50.

1072
00:45:36,970 --> 00:45:41,350
And there are some people
who tend to give a lot.

1073
00:45:41,350 --> 00:45:44,110
That's often a
minority of people.

1074
00:45:44,110 --> 00:45:47,290
Now, what can we infer
from those choices?

1075
00:45:47,290 --> 00:45:49,300
What can we infer from
people who choose 0?

1076
00:45:52,250 --> 00:45:53,110
Yes?

1077
00:45:53,110 --> 00:45:56,400
AUDIENCE: They give
themselves [INAUDIBLE]..

1078
00:45:56,400 --> 00:46:00,590
PROFESSOR: And do we have
any objections to that?

1079
00:46:00,590 --> 00:46:02,570
Or what other interpretation
could we have?

1080
00:46:08,820 --> 00:46:09,640
Yes?

1081
00:46:09,640 --> 00:46:12,370
AUDIENCE: In the case of
money, we can't necessarily

1082
00:46:12,370 --> 00:46:14,530
assume they're
self-interested, I guess,

1083
00:46:14,530 --> 00:46:17,998
because they could be
giving it to charity.

1084
00:46:17,998 --> 00:46:19,040
PROFESSOR: Yeah, exactly.

1085
00:46:19,040 --> 00:46:22,010
So in the case of money,
you could keep the money.

1086
00:46:22,010 --> 00:46:26,420
You could say, I'm
choosing the $10.

1087
00:46:26,420 --> 00:46:28,250
And instead, I'm
giving it to somebody

1088
00:46:28,250 --> 00:46:30,230
who is in higher
need, who has higher

1089
00:46:30,230 --> 00:46:33,260
marginal value of this money.

1090
00:46:33,260 --> 00:46:35,960
That could be you could
give it to charity

1091
00:46:35,960 --> 00:46:38,930
or give it to somebody on the
street or any other person

1092
00:46:38,930 --> 00:46:40,400
you think is in need.

1093
00:46:40,400 --> 00:46:43,340
So that person who looks
like they're fairly selfish

1094
00:46:43,340 --> 00:46:45,680
might, in fact, be quite
generous in the sense

1095
00:46:45,680 --> 00:46:48,530
of giving the money to others.

1096
00:46:48,530 --> 00:46:50,960
That's a common issue
with a lot of these games,

1097
00:46:50,960 --> 00:46:53,960
is essentially it's often an
outside option in a sense of,

1098
00:46:53,960 --> 00:46:57,087
like, we can look at these games
as if this is the only game.

1099
00:46:57,087 --> 00:46:58,670
That's the only thing
that's happening

1100
00:46:58,670 --> 00:47:01,220
in the world, and nothing
else, and look at--

1101
00:47:01,220 --> 00:47:03,200
try to explain
behavior through that.

1102
00:47:03,200 --> 00:47:05,660
But often the issue is
there's other things that

1103
00:47:05,660 --> 00:47:07,410
happen in the world.

1104
00:47:07,410 --> 00:47:09,500
So when I think
about, am I behind,

1105
00:47:09,500 --> 00:47:13,700
or do I get more money
versus somebody else, when

1106
00:47:13,700 --> 00:47:16,340
you play with somebody who is
a lot richer or a lot poorer

1107
00:47:16,340 --> 00:47:21,665
than you are, in fact, giving
them $10 versus not or $5

1108
00:47:21,665 --> 00:47:23,840
or not or will not change
that fact that you're

1109
00:47:23,840 --> 00:47:25,190
a lot richer than they are.

1110
00:47:25,190 --> 00:47:27,530
So perhaps you should give a
lot more, a lot less money,

1111
00:47:27,530 --> 00:47:28,560
and so on.

1112
00:47:28,560 --> 00:47:30,320
So in anything that
we're assuming,

1113
00:47:30,320 --> 00:47:32,255
we look at this
through the lens of,

1114
00:47:32,255 --> 00:47:35,415
like, there's only this game
that happens, and nothing else.

1115
00:47:35,415 --> 00:47:37,790
It turns out that's actually
a pretty good approximation.

1116
00:47:37,790 --> 00:47:39,830
People sort of, what we
call, are narrowly framed.

1117
00:47:39,830 --> 00:47:41,247
They look at
essentially this game

1118
00:47:41,247 --> 00:47:43,370
as if this is completely
separate from anything

1119
00:47:43,370 --> 00:47:45,540
else in the world.

1120
00:47:45,540 --> 00:47:47,660
So people tend to not
integrate these games

1121
00:47:47,660 --> 00:47:49,910
with the rest of life
or the rest of utility.

1122
00:47:49,910 --> 00:47:51,762
So expected utilities
would say, you

1123
00:47:51,762 --> 00:47:53,720
should integrate this
game with everything else

1124
00:47:53,720 --> 00:47:55,230
that happens in life.

1125
00:47:55,230 --> 00:47:57,530
So you should not look
at, like, 6 versus 4.

1126
00:47:57,530 --> 00:47:59,300
I'm earning more
than somebody else.

1127
00:47:59,300 --> 00:48:02,090
You should look at, like,
what is my income overall,

1128
00:48:02,090 --> 00:48:02,930
a lifetime income?

1129
00:48:02,930 --> 00:48:05,600
What is your lifetime income
or the expected lifetime

1130
00:48:05,600 --> 00:48:07,460
income of somebody
else in the experiment?

1131
00:48:07,460 --> 00:48:10,010
And depending on that, you
should give money or not.

1132
00:48:10,010 --> 00:48:12,540
Or differently, you
might say, well,

1133
00:48:12,540 --> 00:48:14,690
should I give money to
somebody else in class?

1134
00:48:14,690 --> 00:48:16,940
Well, perhaps not because
most people in class

1135
00:48:16,940 --> 00:48:21,470
will eventually make a lot
of money once they graduate.

1136
00:48:21,470 --> 00:48:24,140
And instead, let's
just use that money

1137
00:48:24,140 --> 00:48:29,390
and give it to somebody else
who has higher need of that.

1138
00:48:29,390 --> 00:48:29,930
OK.

1139
00:48:29,930 --> 00:48:31,450
So here's no stakes.

1140
00:48:31,450 --> 00:48:34,410
It turns out when you look
at stakes, not much happens.

1141
00:48:34,410 --> 00:48:37,100
So this is answers-- if you
look at-- this is no stakes.

1142
00:48:37,100 --> 00:48:37,910
Here's stakes.

1143
00:48:37,910 --> 00:48:41,220
Answers look, in
fact, very similar.

1144
00:48:41,220 --> 00:48:43,280
There's not much
of a difference.

1145
00:48:43,280 --> 00:48:46,068
Even when you have
higher stakes--

1146
00:48:46,068 --> 00:48:48,110
I think this is, like,
$20, and the like-- people

1147
00:48:48,110 --> 00:48:49,600
seem to behave very similarly.

1148
00:48:49,600 --> 00:48:52,592
So it seems to be, in fact,
asking hypothetical questions--

1149
00:48:52,592 --> 00:48:54,050
and this is how
some of these games

1150
00:48:54,050 --> 00:48:56,870
started, just asking
hypothetically,

1151
00:48:56,870 --> 00:48:59,412
what would you do if somebody
gave you $10 and so on?

1152
00:48:59,412 --> 00:49:01,370
Turns out, you actually
get pretty good answers

1153
00:49:01,370 --> 00:49:04,078
for how people
behave in reality.

1154
00:49:04,078 --> 00:49:05,870
That's true for this
class, but that's also

1155
00:49:05,870 --> 00:49:07,040
true in actual experiments.

1156
00:49:07,040 --> 00:49:08,957
In actual experiments,
if you just ask people,

1157
00:49:08,957 --> 00:49:10,020
like, what would you do?

1158
00:49:10,020 --> 00:49:11,510
And then when you
actually implement it

1159
00:49:11,510 --> 00:49:13,260
with the same people
or with other people,

1160
00:49:13,260 --> 00:49:17,640
you get actually pretty
similar patterns across people.

1161
00:49:17,640 --> 00:49:19,740
Now, we do find that--

1162
00:49:19,740 --> 00:49:21,740
so we had a version of
that that was essentially

1163
00:49:21,740 --> 00:49:25,250
in private, which is
to say you did not

1164
00:49:25,250 --> 00:49:27,750
have to reveal if selected
to the entire class,

1165
00:49:27,750 --> 00:49:30,555
essentially, what you chose,
but only to the other person.

1166
00:49:30,555 --> 00:49:32,930
Notice that that's not really
private in the sense of you

1167
00:49:32,930 --> 00:49:35,450
still have to tell the other
person, ha, I chose 10,

1168
00:49:35,450 --> 00:49:37,950
and you get 0, so
that's not very nice.

1169
00:49:37,950 --> 00:49:39,950
There's other versions
that would be in private,

1170
00:49:39,950 --> 00:49:43,190
would just be the other
person might never find out,

1171
00:49:43,190 --> 00:49:45,650
or the other person might
at least never find out

1172
00:49:45,650 --> 00:49:49,280
that it was you who
chose 10, and they got 0.

1173
00:49:49,280 --> 00:49:52,490
And so but what we do
see, if you look at--

1174
00:49:52,490 --> 00:49:54,350
I think this is with
monetary stakes.

1175
00:49:54,350 --> 00:49:55,790
This is with $10.

1176
00:49:55,790 --> 00:49:57,830
This is with $10 in private.

1177
00:49:57,830 --> 00:50:00,920
And you do see, for
example, that when

1178
00:50:00,920 --> 00:50:03,980
you look at the fraction
who gives essentially 100%,

1179
00:50:03,980 --> 00:50:07,010
that fraction tends to go down.

1180
00:50:07,010 --> 00:50:11,420
So it looks like people are
somewhat less nice in private.

1181
00:50:11,420 --> 00:50:13,097
We're going to get
back to that again

1182
00:50:13,097 --> 00:50:14,930
in the sense of trying
to see-- when we look

1183
00:50:14,930 --> 00:50:16,820
at the question of,
are people really nice

1184
00:50:16,820 --> 00:50:19,460
when they give money, and so on.

1185
00:50:19,460 --> 00:50:22,010
We're going to look at cleaner
versions of that, where people

1186
00:50:22,010 --> 00:50:26,150
have the ability to
essentially hide in some ways

1187
00:50:26,150 --> 00:50:27,470
that they're not nice.

1188
00:50:27,470 --> 00:50:28,490
How might one do that?

1189
00:50:28,490 --> 00:50:30,530
How could you-- what
kind of game would you

1190
00:50:30,530 --> 00:50:33,680
design to hide the fact
that the person is not nice?

1191
00:50:36,400 --> 00:50:38,080
Yes?

1192
00:50:38,080 --> 00:50:40,890
AUDIENCE: For example,
if the recipient

1193
00:50:40,890 --> 00:50:44,620
is only given the
money and not told

1194
00:50:44,620 --> 00:50:47,452
what percentage of the
original money [INAUDIBLE]..

1195
00:50:47,452 --> 00:50:48,160
PROFESSOR: Right.

1196
00:50:48,160 --> 00:50:50,260
So one version of that
would be the recipient just

1197
00:50:50,260 --> 00:50:52,090
doesn't even know
what the game is.

1198
00:50:52,090 --> 00:50:53,260
They just get some money.

1199
00:50:53,260 --> 00:50:56,050
Here's some money
for whatever reason.

1200
00:50:56,050 --> 00:50:58,610
You were just given it because
you're in the experiment.

1201
00:50:58,610 --> 00:51:00,370
And so you could
just essentially hide

1202
00:51:00,370 --> 00:51:02,865
the entire game
from the recipient.

1203
00:51:02,865 --> 00:51:04,240
What's another
thing you could do

1204
00:51:04,240 --> 00:51:06,560
where you don't hide the
entire game but sort of say,

1205
00:51:06,560 --> 00:51:07,477
still, there's a game?

1206
00:51:07,477 --> 00:51:09,030
But what other ways could you--

1207
00:51:09,030 --> 00:51:10,780
in what other ways
could you hide the fact

1208
00:51:10,780 --> 00:51:13,870
that people are not nice?

1209
00:51:13,870 --> 00:51:18,380
What mechanisms
could you implement?

1210
00:51:18,380 --> 00:51:19,692
Yeah?

1211
00:51:19,692 --> 00:51:22,948
AUDIENCE: If you mix robots
and [INAUDIBLE] people.

1212
00:51:22,948 --> 00:51:23,740
PROFESSOR: Exactly.

1213
00:51:23,740 --> 00:51:24,430
And that's exactly--

1214
00:51:24,430 --> 00:51:26,430
I'm going to show you that,
I think, on Wednesday.

1215
00:51:26,430 --> 00:51:28,763
That's exactly what people
have done, where are you say,

1216
00:51:28,763 --> 00:51:32,258
with 50% chance there's a robot
who is not very nice or very

1217
00:51:32,258 --> 00:51:34,050
nice or whatever, and
some chance the robot

1218
00:51:34,050 --> 00:51:37,110
is really mean, with some
chance the robot is really nice.

1219
00:51:37,110 --> 00:51:38,580
And now we can look at--

1220
00:51:38,580 --> 00:51:40,770
so with 50%, the robot decides.

1221
00:51:40,770 --> 00:51:43,107
With 50% chance, you decide.

1222
00:51:43,107 --> 00:51:44,940
And the recipient will
never know whether it

1223
00:51:44,940 --> 00:51:48,210
was a robot versus you.

1224
00:51:48,210 --> 00:51:50,490
And so now you can essentially
hide behind the robot.

1225
00:51:50,490 --> 00:51:53,340
You can say, well, I really
wanted to give you a lot.

1226
00:51:53,340 --> 00:51:55,800
But sorry, the robot
was really mean.

1227
00:51:55,800 --> 00:51:57,270
Too bad.

1228
00:51:57,270 --> 00:51:59,970
And the other person
will never find out.

1229
00:51:59,970 --> 00:52:02,610
And it turns out when you
do that kind of game, people

1230
00:52:02,610 --> 00:52:08,251
tend to be a lot less nice
than when there's no robot.

1231
00:52:08,251 --> 00:52:09,280
OK.

1232
00:52:09,280 --> 00:52:13,820
So then we have chocolate.

1233
00:52:13,820 --> 00:52:15,570
Looks like people are
giving somewhat more

1234
00:52:15,570 --> 00:52:17,460
when it comes to chocolate.

1235
00:52:17,460 --> 00:52:21,570
They tend to give a lot more
when it comes to apples.

1236
00:52:24,630 --> 00:52:26,955
So now, why is that?

1237
00:52:26,955 --> 00:52:27,830
Why does that matter?

1238
00:52:27,830 --> 00:52:28,860
Yes?

1239
00:52:28,860 --> 00:52:33,570
AUDIENCE: If you model
algorithms, your utility

1240
00:52:33,570 --> 00:52:37,220
plus the other person's
utility [INAUDIBLE],,

1241
00:52:37,220 --> 00:52:42,400
it could be that people
with marginal utility,

1242
00:52:42,400 --> 00:52:45,875
in chocolate apples decline
more than [INAUDIBLE]..

1243
00:52:45,875 --> 00:52:46,500
PROFESSOR: Yes.

1244
00:52:46,500 --> 00:52:49,530
It turns out getting the fifth
or sixth or seventh apple

1245
00:52:49,530 --> 00:52:52,620
of the day, maybe
not as enjoyable

1246
00:52:52,620 --> 00:52:55,720
as the first or second.

1247
00:52:55,720 --> 00:52:56,400
So exactly.

1248
00:52:56,400 --> 00:52:58,500
So well, it's really
reasonable to think

1249
00:52:58,500 --> 00:53:00,150
that the marginal
utility of money

1250
00:53:00,150 --> 00:53:03,330
is linear for small amounts.

1251
00:53:03,330 --> 00:53:06,370
If I give you $1, 2, or 3,
or 4, or 5, in some sense,

1252
00:53:06,370 --> 00:53:08,370
that shouldn't be very concave.

1253
00:53:08,370 --> 00:53:09,900
We discussed that at length.

1254
00:53:09,900 --> 00:53:12,330
For apples, they're perishable.

1255
00:53:12,330 --> 00:53:14,730
There's only so many apples
you can eat on a given day.

1256
00:53:14,730 --> 00:53:16,772
So really what we should
look at is, like, what's

1257
00:53:16,772 --> 00:53:17,950
the utility of apples?

1258
00:53:17,950 --> 00:53:21,180
And if the marginal utility
of apples is decreasing,

1259
00:53:21,180 --> 00:53:24,910
that might matter quite a bit.

1260
00:53:24,910 --> 00:53:26,190
OK.

1261
00:53:26,190 --> 00:53:30,210
So maybe let me
summarize a little bit.

1262
00:53:30,210 --> 00:53:32,220
What did we learn in
those kinds of games?

1263
00:53:32,220 --> 00:53:38,070
So first is people look fairly
generous in dictator games,

1264
00:53:38,070 --> 00:53:41,190
even when the game
is largely private.

1265
00:53:41,190 --> 00:53:43,800
We're going to challenge that
assumption a little bit more,

1266
00:53:43,800 --> 00:53:46,770
exactly as you suggested,
with other ways in which

1267
00:53:46,770 --> 00:53:49,920
either people can be more
private, they can just

1268
00:53:49,920 --> 00:53:52,650
hide the game entirely,
or when they perhaps

1269
00:53:52,650 --> 00:53:56,340
can hide behind a machine or
a computer or in other ways,

1270
00:53:56,340 --> 00:54:01,110
or when they perhaps can opt
out of the game in certain ways.

1271
00:54:01,110 --> 00:54:03,000
People generally seem
to give something

1272
00:54:03,000 --> 00:54:04,890
like 30% of the total.

1273
00:54:04,890 --> 00:54:05,880
That's quite common.

1274
00:54:05,880 --> 00:54:08,795
20% to 30% is quite
common overall.

1275
00:54:08,795 --> 00:54:10,920
It doesn't seem to matter
that much whether there's

1276
00:54:10,920 --> 00:54:13,650
hypothetical versus
actual choices.

1277
00:54:13,650 --> 00:54:17,820
The size of the stakes also
seem not to matter that much.

1278
00:54:17,820 --> 00:54:20,100
We seem to see somewhat
different behaviors

1279
00:54:20,100 --> 00:54:21,208
of chocolates and apples.

1280
00:54:21,208 --> 00:54:23,250
There's also some people
who told me that they're

1281
00:54:23,250 --> 00:54:24,790
lactose intolerant and so on.

1282
00:54:24,790 --> 00:54:27,330
So essentially saying,
like, their marginal utility

1283
00:54:27,330 --> 00:54:28,780
of chocolate is quite low.

1284
00:54:28,780 --> 00:54:30,960
So it's much easier-- it's
much less costly to give

1285
00:54:30,960 --> 00:54:31,938
to somebody else.

1286
00:54:31,938 --> 00:54:34,230
Or put differently, it's
actually efficient potentially

1287
00:54:34,230 --> 00:54:35,938
because the marginal
utility of chocolate

1288
00:54:35,938 --> 00:54:38,743
is much higher for that other
person than it is for you.

1289
00:54:38,743 --> 00:54:40,410
Of course, there's a
problem now, again,

1290
00:54:40,410 --> 00:54:42,480
that you could just choose
the chocolates for yourself

1291
00:54:42,480 --> 00:54:44,220
and then just give
it to your friends.

1292
00:54:44,220 --> 00:54:48,450
So you look quite selfish,
but in fact, you might not be.

1293
00:54:48,450 --> 00:54:51,100
So anyway, but sort of
what really matters here--

1294
00:54:51,100 --> 00:54:52,410
and this is perhaps--

1295
00:54:52,410 --> 00:54:55,000
in some sense, when we talk
about apples and chocolates,

1296
00:54:55,000 --> 00:54:56,053
it's a little bit silly.

1297
00:54:56,053 --> 00:54:57,720
But when you think
about a dictator game

1298
00:54:57,720 --> 00:55:01,810
where you play with somebody in
Kenya whose income or lifetime

1299
00:55:01,810 --> 00:55:04,117
income is, like, orders of
magnitude lower than yours,

1300
00:55:04,117 --> 00:55:05,700
then it matters,
actually, quite a bit

1301
00:55:05,700 --> 00:55:08,452
what's the marginal
utility of the $10 that you

1302
00:55:08,452 --> 00:55:10,410
could keep for yourself
versus for that person.

1303
00:55:10,410 --> 00:55:12,060
For that person,
that's a lot of money.

1304
00:55:12,060 --> 00:55:14,850
For you, a lot less.

1305
00:55:14,850 --> 00:55:17,220
And so you might
be more generous

1306
00:55:17,220 --> 00:55:20,130
when playing with a person
whose marginal utility of money

1307
00:55:20,130 --> 00:55:23,700
or consumption is a lot higher.

1308
00:55:23,700 --> 00:55:25,620
This is what I was
mentioning as well.

1309
00:55:25,620 --> 00:55:28,350
You want to think about,
what is the outside option?

1310
00:55:28,350 --> 00:55:29,910
So some people might
give low amount

1311
00:55:29,910 --> 00:55:32,310
and decide to give the
money or the apples

1312
00:55:32,310 --> 00:55:34,740
or whatever to somebody else.

1313
00:55:34,740 --> 00:55:37,140
Those people will look
selfish in dictator games,

1314
00:55:37,140 --> 00:55:39,360
while, in fact,
they're quite generous.

1315
00:55:39,360 --> 00:55:40,740
That's a problem
with these games

1316
00:55:40,740 --> 00:55:43,900
that are hard to deal with.

1317
00:55:43,900 --> 00:55:47,760
We're not going to be able
to deal with that overall.

1318
00:55:47,760 --> 00:55:49,960
You can think about it a
little bit and sort of see.

1319
00:55:49,960 --> 00:55:53,580
Can we give people options
to opt out and so on?

1320
00:55:53,580 --> 00:55:55,110
But it will always be an issue.

1321
00:55:55,110 --> 00:55:57,360
It's less of an issue than
you perhaps think about it.

1322
00:55:57,360 --> 00:55:59,350
Conceptually, that's
a huge problem.

1323
00:55:59,350 --> 00:56:01,470
But when we look at
people's actual behavior,

1324
00:56:01,470 --> 00:56:03,540
it looks like people really
behave as if this was

1325
00:56:03,540 --> 00:56:04,950
the only thing in the world.

1326
00:56:04,950 --> 00:56:06,930
And then, in some ways,
it's quite predictive

1327
00:56:06,930 --> 00:56:09,780
of actual behavior in the world,
even perhaps, in some ways,

1328
00:56:09,780 --> 00:56:11,670
it shouldn't be.

1329
00:56:11,670 --> 00:56:14,365
Any other comments or things we
learned perhaps from the game

1330
00:56:14,365 --> 00:56:15,240
or that you observed?

1331
00:56:18,180 --> 00:56:18,956
Yeah?

1332
00:56:18,956 --> 00:56:22,050
AUDIENCE: Is there a possibility
that the people that give 0,

1333
00:56:22,050 --> 00:56:27,140
it's just that they're
interested distribution

1334
00:56:27,140 --> 00:56:29,370
[INAUDIBLE] is the same
as your disinterested one,

1335
00:56:29,370 --> 00:56:31,912
and they actually think, like,
oh, the best thing for society

1336
00:56:31,912 --> 00:56:37,090
is for me to have those $10
and not the other person?

1337
00:56:37,090 --> 00:56:41,123
PROFESSOR: So because you put
a lot of weight on yourself or?

1338
00:56:41,123 --> 00:56:42,540
AUDIENCE: Sort of
like, regardless

1339
00:56:42,540 --> 00:56:45,390
if it's correct or
not, but you think

1340
00:56:45,390 --> 00:56:47,576
that the way you'll
use these $10

1341
00:56:47,576 --> 00:56:51,000
is better for society
than the other person?

1342
00:56:51,000 --> 00:56:53,610
PROFESSOR: Yeah, I think you
have to justify that somehow

1343
00:56:53,610 --> 00:56:55,320
in the sense of,
like, if you are very

1344
00:56:55,320 --> 00:56:57,630
similar to the other students,
then you might sort of

1345
00:56:57,630 --> 00:57:00,850
think that dividing it--

1346
00:57:00,850 --> 00:57:04,005
so depends what do you think is
the marginal utility of money.

1347
00:57:04,005 --> 00:57:05,880
So if you think the
marginal utility of money

1348
00:57:05,880 --> 00:57:08,820
is essentially the same for
everybody, in some sense then,

1349
00:57:08,820 --> 00:57:10,800
it doesn't matter very
much whether you keep it

1350
00:57:10,800 --> 00:57:12,330
to yourself versus
somebody else,

1351
00:57:12,330 --> 00:57:14,740
in some sense because, in
the grand scheme of things,

1352
00:57:14,740 --> 00:57:16,107
it doesn't look very different.

1353
00:57:16,107 --> 00:57:18,690
It could be that, for example,
you think, for whatever reason,

1354
00:57:18,690 --> 00:57:21,060
you just really had a bad
day, and you can use the $10

1355
00:57:21,060 --> 00:57:22,647
to buy something nice.

1356
00:57:22,647 --> 00:57:24,480
It could be that you
think you're relatively

1357
00:57:24,480 --> 00:57:26,268
poor to others in class.

1358
00:57:26,268 --> 00:57:28,560
For whatever reason, you just
don't have a lot of money

1359
00:57:28,560 --> 00:57:29,430
right now.

1360
00:57:29,430 --> 00:57:32,220
The marginal utility of money
for you might be really high.

1361
00:57:32,220 --> 00:57:35,580
So again, and in some sense,
that's socially efficient.

1362
00:57:35,580 --> 00:57:38,070
You would actually--
even if it weren't you,

1363
00:57:38,070 --> 00:57:40,450
you would give the
money to yourself.

1364
00:57:40,450 --> 00:57:42,510
Now, I think it is
the case that when

1365
00:57:42,510 --> 00:57:45,570
you look at these
distributions, it

1366
00:57:45,570 --> 00:57:49,000
tends to be that 30% of
people keep all of that money.

1367
00:57:49,000 --> 00:57:55,160
So it tends to be that it's hard
to believe that that's really--

1368
00:57:55,160 --> 00:57:57,660
I mean, you'd have to look at,
like, how does that correlate

1369
00:57:57,660 --> 00:57:59,160
with other characteristics?

1370
00:57:59,160 --> 00:58:01,170
It probably does,
and we'll get back

1371
00:58:01,170 --> 00:58:03,360
to that in the sense
of when we see now,

1372
00:58:03,360 --> 00:58:05,790
when people have the chance
to hide money and so on, when

1373
00:58:05,790 --> 00:58:08,670
you see now the fraction
of people who choose

1374
00:58:08,670 --> 00:58:10,530
$10 for themselves
and 0 for others

1375
00:58:10,530 --> 00:58:13,740
goes up, that, in some ways,
compared to when you can't

1376
00:58:13,740 --> 00:58:17,010
hide, that perhaps is an
identifying selfishness and so

1377
00:58:17,010 --> 00:58:17,640
on.

1378
00:58:17,640 --> 00:58:19,710
Here, when you see
people keeping 10,

1379
00:58:19,710 --> 00:58:22,442
you can come up with lots
of explanations, including

1380
00:58:22,442 --> 00:58:24,900
the one that you mentioned,
but also including the ones you

1381
00:58:24,900 --> 00:58:26,700
keep it for yourself and
then give it to your friends

1382
00:58:26,700 --> 00:58:28,230
and share it with
them and so on.

1383
00:58:28,230 --> 00:58:30,600
And we can't really
infer very much.

1384
00:58:30,600 --> 00:58:32,400
But in these
cleaner experiments,

1385
00:58:32,400 --> 00:58:35,665
where you can
really say, OK, when

1386
00:58:35,665 --> 00:58:37,290
you don't have the
outside option, when

1387
00:58:37,290 --> 00:58:40,890
you don't have the chance to
hide, you look quite nice.

1388
00:58:40,890 --> 00:58:43,260
That goes away when you
have the chance to hide.

1389
00:58:43,260 --> 00:58:46,860
Presumably, that is because
you just are selfish and not

1390
00:58:46,860 --> 00:58:48,770
for other reasons.

1391
00:58:48,770 --> 00:58:50,210
OK.

1392
00:58:50,210 --> 00:58:51,212
Yes?

1393
00:58:51,212 --> 00:58:52,920
AUDIENCE: Have people
looked at if things

1394
00:58:52,920 --> 00:58:54,762
change if you
actually get the money

1395
00:58:54,762 --> 00:58:55,970
or get the apples beforehand?

1396
00:58:55,970 --> 00:58:57,416
Like, [INAUDIBLE]?

1397
00:59:03,942 --> 00:59:07,930
PROFESSOR: Surely
they-- surely they have.

1398
00:59:07,930 --> 00:59:09,820
What I don't know--
so surely [INAUDIBLE]..

1399
00:59:09,820 --> 00:59:12,620
So [INAUDIBLE] is very
robust in various situations.

1400
00:59:12,620 --> 00:59:15,790
So my guess is they were
less likely to give.

1401
00:59:15,790 --> 00:59:16,750
I don't have--

1402
00:59:16,750 --> 00:59:20,350
I can look it up, but I don't
have like a great experiment

1403
00:59:20,350 --> 00:59:22,480
that I can tell you about.

1404
00:59:22,480 --> 00:59:24,430
But my sense is that--

1405
00:59:24,430 --> 00:59:27,970
so the question is, if you give
people the $10 into their hand,

1406
00:59:27,970 --> 00:59:29,830
are they going to
choose differently?

1407
00:59:29,830 --> 00:59:34,210
My guess is they will
look a lot less selfish.

1408
00:59:34,210 --> 00:59:36,310
In the experiment that
you played, in some sense

1409
00:59:36,310 --> 00:59:37,810
you are sort of
endowed with a game.

1410
00:59:37,810 --> 00:59:38,962
So in some ways--

1411
00:59:38,962 --> 00:59:40,420
but you could do
deviation of that.

1412
00:59:40,420 --> 00:59:42,030
You could look at,
like, here's $10.

1413
00:59:42,030 --> 00:59:43,780
It's neither yours nor
the other person's.

1414
00:59:43,780 --> 00:59:45,250
How would you like to divide it?

1415
00:59:45,250 --> 00:59:47,333
My guess is that person
will be more generous when

1416
00:59:47,333 --> 00:59:49,990
they do that compared to
saying, like, here's $10.

1417
00:59:49,990 --> 00:59:53,110
They give it to you in
$1 bills, and now you

1418
00:59:53,110 --> 00:59:54,670
have to give me back some.

1419
00:59:54,670 --> 00:59:56,947
People will be more likely
to keep the money when

1420
00:59:56,947 --> 00:59:59,530
they actually have the money in
hand, when they feel like it's

1421
00:59:59,530 --> 01:00:02,950
yours or theirs,
as opposed to have

1422
01:00:02,950 --> 01:00:07,840
to divide money that doesn't
really belong to any person.

1423
01:00:07,840 --> 01:00:09,940
So my guess is that's
true, but I don't--

1424
01:00:09,940 --> 01:00:12,820
there's literally thousands
of experiments of people doing

1425
01:00:12,820 --> 01:00:14,510
these kinds of variations.

1426
01:00:14,510 --> 01:00:16,510
So I'm sure there is
somebody who has done this.

1427
01:00:16,510 --> 01:00:21,190
I can look up and see
whether I find that.

1428
01:00:21,190 --> 01:00:21,700
OK.

1429
01:00:21,700 --> 01:00:23,440
Any other thoughts
of what we learned?

1430
01:00:27,980 --> 01:00:28,820
OK.

1431
01:00:28,820 --> 01:00:31,070
So when you now think
about how should you

1432
01:00:31,070 --> 01:00:33,110
model these distributional
preferences--

1433
01:00:33,110 --> 01:00:36,560
and there's a classic paper
by Charness and Rabin.

1434
01:00:36,560 --> 01:00:38,870
Rabin is the guy that you
know from the calibration

1435
01:00:38,870 --> 01:00:41,870
theorem about risk preferences.

1436
01:00:41,870 --> 01:00:44,930
And the way their
model is very simple.

1437
01:00:44,930 --> 01:00:47,450
They essentially
have the preferences

1438
01:00:47,450 --> 01:00:49,550
over outcomes x1 and x2.

1439
01:00:49,550 --> 01:00:50,710
Think of this as money.

1440
01:00:50,710 --> 01:00:52,760
It could be also apples
or something else.

1441
01:00:52,760 --> 01:00:54,560
Money is perhaps a
better approximation

1442
01:00:54,560 --> 01:00:56,520
because the marginal
utility of money--

1443
01:00:56,520 --> 01:00:59,250
think of this as
rather constant.

1444
01:00:59,250 --> 01:01:03,080
And now player 2
is the dictator.

1445
01:01:03,080 --> 01:01:05,240
So x2 is how much player 2 gets.

1446
01:01:05,240 --> 01:01:08,360
So player 2 has a
utility function u2.

1447
01:01:08,360 --> 01:01:11,592
The utility of player
2 has a function of x1.

1448
01:01:11,592 --> 01:01:13,550
This is how much the
other person gets, and x2,

1449
01:01:13,550 --> 01:01:16,490
how much the player 2 gets.

1450
01:01:16,490 --> 01:01:19,160
And the utility function
looks like rho times

1451
01:01:19,160 --> 01:01:21,650
x1 plus 1 minus rho times x2.

1452
01:01:21,650 --> 01:01:23,990
If x2 is larger
or equal than x1--

1453
01:01:23,990 --> 01:01:26,430
that is to say that
the person is ahead,

1454
01:01:26,430 --> 01:01:28,370
if x2 gets more than x1--

1455
01:01:28,370 --> 01:01:32,990
and it's sigma times x1
plus 1 minus sigma x2

1456
01:01:32,990 --> 01:01:37,580
if x2 is smaller
or equal than x1.

1457
01:01:37,580 --> 01:01:39,980
So what do these
parameters measure?

1458
01:01:39,980 --> 01:01:41,540
What a rho and
sigma measure here?

1459
01:01:52,243 --> 01:01:53,410
Why might they be different?

1460
01:01:53,410 --> 01:01:54,430
Yeah?

1461
01:01:54,430 --> 01:01:55,630
AUDIENCE: How nice they are?

1462
01:01:55,630 --> 01:01:57,513
PROFESSOR: Yes, exactly.

1463
01:01:57,513 --> 01:01:58,180
What does it do?

1464
01:01:58,180 --> 01:02:01,390
How much weight do you put on
yourself or the other person?

1465
01:02:01,390 --> 01:02:03,615
AUDIENCE: So if
it's a lower value,

1466
01:02:03,615 --> 01:02:06,972
then you don't really care
about the other person,

1467
01:02:06,972 --> 01:02:08,430
so you put more
weight on yourself.

1468
01:02:08,430 --> 01:02:08,482
PROFESSOR: Right.

1469
01:02:08,482 --> 01:02:09,482
So in the extreme case--

1470
01:02:09,482 --> 01:02:12,250
I guess, in one case, I
guess, if you choose--

1471
01:02:12,250 --> 01:02:14,830
if this is 0, then
essentially, you're

1472
01:02:14,830 --> 01:02:19,690
back in the neoclassical case of
not caring about others at all.

1473
01:02:19,690 --> 01:02:22,630
So if you choose, like, rho
sigma equals 0, then it's just,

1474
01:02:22,630 --> 01:02:23,830
like, how much do you get?

1475
01:02:23,830 --> 01:02:27,440
You should get 0, and you
should accept any offer, right?

1476
01:02:27,440 --> 01:02:33,550
And now as rho and sigma
goes up, 0.5 or whatever,

1477
01:02:33,550 --> 01:02:35,860
you give more weight
on the other person.

1478
01:02:35,860 --> 01:02:38,170
So 0.5, for example,
is the case where

1479
01:02:38,170 --> 01:02:43,120
you care equally about the other
person and yourself, right?

1480
01:02:43,120 --> 01:02:45,190
Because the weights
are then the same.

1481
01:02:45,190 --> 01:02:47,460
Now, why might rho and
sigma be different?

1482
01:02:53,570 --> 01:02:54,874
Yes?

1483
01:02:54,874 --> 01:02:59,850
AUDIENCE: So probably
rho is larger than sigma.

1484
01:02:59,850 --> 01:03:02,020
PROFESSOR: And why is that?

1485
01:03:02,020 --> 01:03:05,340
AUDIENCE: So if you're
person 2 and you already

1486
01:03:05,340 --> 01:03:07,680
know you're getting more
than the other person,

1487
01:03:07,680 --> 01:03:11,160
then you don't feel such
a need to do so much--

1488
01:03:11,160 --> 01:03:14,008
you don't feel a need to
have so much more than them.

1489
01:03:14,008 --> 01:03:15,050
PROFESSOR: Yeah, exactly.

1490
01:03:15,050 --> 01:03:15,960
It's just easier.

1491
01:03:15,960 --> 01:03:19,430
That seems to be a very
intuitive feature of the world.

1492
01:03:19,430 --> 01:03:21,230
It seems to be much
easier to be generous

1493
01:03:21,230 --> 01:03:22,857
when you're ahead anyway.

1494
01:03:22,857 --> 01:03:24,440
If you're very rich,
you might as well

1495
01:03:24,440 --> 01:03:26,780
give some money to the
poorer person, and so on.

1496
01:03:26,780 --> 01:03:30,740
If you are poorer
than the other person,

1497
01:03:30,740 --> 01:03:33,290
it's much harder to be
generous in some ways,

1498
01:03:33,290 --> 01:03:35,150
perhaps because you
feel like you deserve

1499
01:03:35,150 --> 01:03:37,520
to get as much as
the other person,

1500
01:03:37,520 --> 01:03:40,970
so why would you give
the other person more?

1501
01:03:40,970 --> 01:03:43,790
You put essentially less
weight on that other person

1502
01:03:43,790 --> 01:03:45,950
if you're behind.

1503
01:03:45,950 --> 01:03:48,800
What's the case of
sigma smaller than 0?

1504
01:03:57,610 --> 01:03:59,140
Yeah?

1505
01:03:59,140 --> 01:04:02,790
AUDIENCE: Would it
be then the person

1506
01:04:02,790 --> 01:04:06,404
who want to hurt
the other person,

1507
01:04:06,404 --> 01:04:10,708
so then they put more weight
on decreasing their value?

1508
01:04:10,708 --> 01:04:11,750
PROFESSOR: Yeah, exactly.

1509
01:04:11,750 --> 01:04:14,250
So if you're behind-- so it's
the case where you are behind,

1510
01:04:14,250 --> 01:04:16,110
you're the x2, the player 2.

1511
01:04:16,110 --> 01:04:19,480
If player 2 is behind,
has less than player 1.

1512
01:04:19,480 --> 01:04:23,950
Now, you might get
positive utility

1513
01:04:23,950 --> 01:04:28,010
from reducing
player 1's outcome.

1514
01:04:28,010 --> 01:04:29,590
So if you're
behind, you'd rather

1515
01:04:29,590 --> 01:04:31,840
have the other person get
less, and you'll feel better

1516
01:04:31,840 --> 01:04:36,520
about that, presumably because
you're then less behind, even

1517
01:04:36,520 --> 01:04:40,340
holding constant
your own outcomes.

1518
01:04:40,340 --> 01:04:42,520
So if you keep your own
outcome, your own payment,

1519
01:04:42,520 --> 01:04:43,870
the same-- suppose you get 5.

1520
01:04:43,870 --> 01:04:45,490
The other person gets 20--

1521
01:04:45,490 --> 01:04:47,020
you feel better if
the other person

1522
01:04:47,020 --> 01:04:49,510
gets 15 or 10 because
you're less behind now

1523
01:04:49,510 --> 01:04:51,510
than the other person.

1524
01:04:51,510 --> 01:04:53,470
OK, so you're willing to
hurt the other person.

1525
01:04:53,470 --> 01:04:56,620
You might be even willing to pay
to hurt the other person, which

1526
01:04:56,620 --> 01:04:59,290
is kind of what happens
in the dictator game

1527
01:04:59,290 --> 01:05:03,380
when the person rejects
the dictator game.

1528
01:05:03,380 --> 01:05:04,390
Any questions on that?

1529
01:05:08,470 --> 01:05:12,370
So I have an example of
sigma smaller than 1.

1530
01:05:12,370 --> 01:05:16,000
Suppose you have
sigma being minus 1/3.

1531
01:05:16,000 --> 01:05:18,032
You have to choose
between 0, 0 and 9, 1.

1532
01:05:18,032 --> 01:05:19,240
What are you going to choose?

1533
01:05:24,710 --> 01:05:26,810
So this is what this
looks like, sigma times

1534
01:05:26,810 --> 01:05:28,460
x1 plus 1 minus sigma x2.

1535
01:05:39,600 --> 01:05:40,100
Yes?

1536
01:05:40,100 --> 01:05:41,435
AUDIENCE: You choose 0, 0.

1537
01:05:41,435 --> 01:05:42,560
PROFESSOR: And why is that?

1538
01:05:42,560 --> 01:05:44,733
AUDIENCE: Because it
multiplies the other numbers.

1539
01:05:44,733 --> 01:05:45,400
PROFESSOR: Yeah.

1540
01:05:45,400 --> 01:05:46,635
AUDIENCE: [INAUDIBLE].

1541
01:05:46,635 --> 01:05:48,510
PROFESSOR: Right, so
essentially-- the reason

1542
01:05:48,510 --> 01:05:51,360
is that essentially, if
you look at this situation,

1543
01:05:51,360 --> 01:05:54,000
and you put minus 1/3
weight, essentially,

1544
01:05:54,000 --> 01:05:55,320
on the other person.

1545
01:05:55,320 --> 01:05:57,960
So you really don't like
the other person getting 9.

1546
01:05:57,960 --> 01:06:03,300
So you're really
unhappy about that.

1547
01:06:03,300 --> 01:06:07,423
Despite the fact that you value
getting some money yourself--

1548
01:06:07,423 --> 01:06:09,090
you have positive
value on that as well.

1549
01:06:09,090 --> 01:06:12,480
It's, I guess, 4/3 times 1.

1550
01:06:12,480 --> 01:06:14,430
But sort of putting
together, that's negative.

1551
01:06:14,430 --> 01:06:16,080
So essentially, you're
willing to pay--

1552
01:06:16,080 --> 01:06:19,320
and this is, I guess, one
explanation potentially

1553
01:06:19,320 --> 01:06:20,550
why people might reject.

1554
01:06:20,550 --> 01:06:22,550
And this is, I guess,
what I was saying earlier.

1555
01:06:22,550 --> 01:06:25,710
You might reject the
ultimatum game simply

1556
01:06:25,710 --> 01:06:26,940
for distributional reasons.

1557
01:06:26,940 --> 01:06:29,490
You might just say, I don't
like being behind in this game.

1558
01:06:29,490 --> 01:06:32,760
If I'm behind, I put negative
weight on this other person,

1559
01:06:32,760 --> 01:06:35,250
and then I'm now
rejecting it because I

1560
01:06:35,250 --> 01:06:40,470
feel happier if the person who
is in front of me gets less,

1561
01:06:40,470 --> 01:06:45,360
even if that comes at the cost
of me having to pay some money.

1562
01:06:45,360 --> 01:06:48,080
Now, notice, you are not going
to reject all unequal offers.

1563
01:06:48,080 --> 01:06:51,390
So I think 6-4, you
would probably accept.

1564
01:06:51,390 --> 01:06:55,920
But you will reject
really uneven offers.

1565
01:06:55,920 --> 01:06:58,650
So in some sense, if I ask you
questions using the strategy

1566
01:06:58,650 --> 01:06:59,910
method, as we did--

1567
01:06:59,910 --> 01:07:02,250
if I asked you questions
about at what threshold

1568
01:07:02,250 --> 01:07:05,460
would you reject an
offer, I can essentially--

1569
01:07:05,460 --> 01:07:07,860
if I assume that your
preferences look like this,

1570
01:07:07,860 --> 01:07:12,750
I can back out,
essentially, your sigma.

1571
01:07:12,750 --> 01:07:14,430
Questions on that?

1572
01:07:14,430 --> 01:07:15,213
Yeah?

1573
01:07:15,213 --> 01:07:18,540
AUDIENCE: What's the
utility [INAUDIBLE]

1574
01:07:18,540 --> 01:07:21,915
for the dictator, not the person
who gets to reject the offer?

1575
01:07:21,915 --> 01:07:23,540
PROFESSOR: Sorry,
this is just-- sorry,

1576
01:07:23,540 --> 01:07:24,665
I should have been clearer.

1577
01:07:24,665 --> 01:07:28,700
So this could be for--

1578
01:07:28,700 --> 01:07:32,520
these are preferences that are
generic for any types of games

1579
01:07:32,520 --> 01:07:34,180
that are being played.

1580
01:07:34,180 --> 01:07:37,270
So you could apply
this to any game.

1581
01:07:37,270 --> 01:07:40,740
So Charness and Rabin,
essentially what they do

1582
01:07:40,740 --> 01:07:43,080
is-- in their paper,
and it's a long paper

1583
01:07:43,080 --> 01:07:44,580
that covers a lot of ground.

1584
01:07:44,580 --> 01:07:46,247
But essentially, what
they're saying is,

1585
01:07:46,247 --> 01:07:48,940
like, here's what distributional
preferences might look like.

1586
01:07:48,940 --> 01:07:51,060
And you can then choose
different parameters.

1587
01:07:51,060 --> 01:07:53,490
And then the question is, like,
what kinds of behavior now

1588
01:07:53,490 --> 01:07:54,523
can we explain?

1589
01:07:54,523 --> 01:07:56,190
Being the dictator
game, potentially you

1590
01:07:56,190 --> 01:07:58,360
can try to explain sort
of people's behavior.

1591
01:07:58,360 --> 01:08:01,377
But not I'm trying to apply
it to the ultimatum game.

1592
01:08:01,377 --> 01:08:03,210
Or you could also apply
it to the trust game

1593
01:08:03,210 --> 01:08:05,430
and say, for what
kinds of behavior,

1594
01:08:05,430 --> 01:08:10,260
what kinds of preferences might
people do certain behavior?

1595
01:08:10,260 --> 01:08:16,090
So what kinds of behaviors
can be explained with that?

1596
01:08:16,090 --> 01:08:19,660
So now what's the experimental
evidence on rho and sigma?

1597
01:08:19,660 --> 01:08:22,310
Just to remind you,
rho is the parameter,

1598
01:08:22,310 --> 01:08:24,810
the weight that you put on the
other person if you're ahead.

1599
01:08:24,810 --> 01:08:28,380
Sigma is the parameter on the
other person when you're behind

1600
01:08:28,380 --> 01:08:30,250
in a situation.

1601
01:08:30,250 --> 01:08:32,640
So we think that when
people are ahead,

1602
01:08:32,640 --> 01:08:35,229
most people seem to
have a positive rho.

1603
01:08:35,229 --> 01:08:37,319
So people usually tend to--

1604
01:08:37,319 --> 01:08:39,810
when ahead, they're willing
to sacrifice some money

1605
01:08:39,810 --> 01:08:42,038
to increase the other
person's payout.

1606
01:08:42,038 --> 01:08:44,580
So this is essentially-- when
you think about dictator games,

1607
01:08:44,580 --> 01:08:46,229
when they have 10,
essentially you

1608
01:08:46,229 --> 01:08:48,779
think of this as being,
like, 10-0, you're ahead.

1609
01:08:48,779 --> 01:08:51,720
So you're willing to sacrifice
at least some money, like 7-3

1610
01:08:51,720 --> 01:08:53,640
or 6-4, even 5-5.

1611
01:08:53,640 --> 01:08:55,080
You like that better.

1612
01:08:55,080 --> 01:08:58,290
You're happy to give the
other person some amount.

1613
01:08:58,290 --> 01:09:00,330
A minority are even
willing to sacrifice money

1614
01:09:00,330 --> 01:09:02,372
to give the other person
an equal amount of money

1615
01:09:02,372 --> 01:09:04,140
or even less--

1616
01:09:04,140 --> 01:09:07,439
sorry, even more.

1617
01:09:07,439 --> 01:09:10,319
When deciding to
split $10, subjects

1618
01:09:10,319 --> 01:09:13,170
tend to give about
20% to 25% on average.

1619
01:09:13,170 --> 01:09:14,910
In class, that was
actually pretty close.

1620
01:09:14,910 --> 01:09:18,600
The 28%, I think, are
about the $10 amount.

1621
01:09:18,600 --> 01:09:20,640
You guys gave it
something like 28%.

1622
01:09:20,640 --> 01:09:22,682
That's actually pretty
close to what people would

1623
01:09:22,682 --> 01:09:24,120
do in normal dictator games.

1624
01:09:24,120 --> 01:09:26,939
The estimated rho
tends to be about 0.4.

1625
01:09:26,939 --> 01:09:28,050
Yeah?

1626
01:09:28,050 --> 01:09:30,960
AUDIENCE: How can rho be
estimated from the dictator

1627
01:09:30,960 --> 01:09:31,460
game?

1628
01:09:31,460 --> 01:09:33,961
The linearity of the
preferences here,

1629
01:09:33,961 --> 01:09:37,730
shouldn't it predict
that player 2 will always

1630
01:09:37,730 --> 01:09:43,134
either give 0 or give 50%?

1631
01:09:43,134 --> 01:09:44,547
PROFESSOR: No,
it's just to say--

1632
01:09:44,547 --> 01:09:47,130
I think this is just to say you
have the same marginal utility

1633
01:09:47,130 --> 01:09:47,970
than I have.

1634
01:09:47,970 --> 01:09:51,240
So depending on when I ask
you, like, how much do you

1635
01:09:51,240 --> 01:09:58,320
give, that pins down your rho.

1636
01:09:58,320 --> 01:10:03,710
AUDIENCE: But if
x1 is 10 minus x2,

1637
01:10:03,710 --> 01:10:09,860
and we want to maximize rho
x1 plus 1 minus rho times

1638
01:10:09,860 --> 01:10:14,982
x2, and you're
maximizing that over x2.

1639
01:10:14,982 --> 01:10:17,710
[INAUDIBLE] equation,
shouldn't it be [INAUDIBLE]??

1640
01:10:17,710 --> 01:10:20,592
PROFESSOR: No, no, but the
unknown here is the rho.

1641
01:10:20,592 --> 01:10:22,300
It's essentially to
say, for example, you

1642
01:10:22,300 --> 01:10:25,570
choose 6-4, that means
you prefer 6-4 over 7-3,

1643
01:10:25,570 --> 01:10:28,120
and you prefer 6-4 over 5-5.

1644
01:10:28,120 --> 01:10:29,740
And those two
inequalities essentially

1645
01:10:29,740 --> 01:10:32,950
give you a bound on rho.

1646
01:10:32,950 --> 01:10:36,190
We can talk afterwards, but
I'm pretty sure that's true.

1647
01:10:36,190 --> 01:10:37,010
Yeah.

1648
01:10:37,010 --> 01:10:37,510
Yeah.

1649
01:10:41,392 --> 01:10:43,850
But yeah, that's essentially
just saying, like, essentially

1650
01:10:43,850 --> 01:10:45,440
what you do is you have--

1651
01:10:45,440 --> 01:10:47,510
since you chose
a certain option,

1652
01:10:47,510 --> 01:10:50,750
now essentially you'd prefer
not to choose a higher option.

1653
01:10:50,750 --> 01:10:52,520
You prefer not to
choose a lower option.

1654
01:10:52,520 --> 01:10:54,437
When you write that down,
that should give you

1655
01:10:54,437 --> 01:10:58,760
inequalities as a function
of rho in this case.

1656
01:10:58,760 --> 01:11:01,970
But we'll talk about
it in two minutes.

1657
01:11:01,970 --> 01:11:05,180
What's the experimental
evidence of sigma?

1658
01:11:05,180 --> 01:11:08,300
Only about 10% to 20% of
players have sigma strong enough

1659
01:11:08,300 --> 01:11:10,580
to pay a non-trivial amount
to hurt the other player.

1660
01:11:10,580 --> 01:11:16,850
These are people essentially
who reject offers.

1661
01:11:16,850 --> 01:11:19,010
About 30% of people
will sacrifice

1662
01:11:19,010 --> 01:11:20,490
to help a player ahead of them.

1663
01:11:20,490 --> 01:11:22,782
So even if they're behind
and have the option of, like,

1664
01:11:22,782 --> 01:11:26,240
can you sort of help the person
in front of you, 30% of people

1665
01:11:26,240 --> 01:11:27,483
will do that.

1666
01:11:27,483 --> 01:11:29,150
If a person is behind
you, often neither

1667
01:11:29,150 --> 01:11:31,010
wants to help or hurt
the other person by much.

1668
01:11:31,010 --> 01:11:32,780
Essentially, sort of
like, whatever, I don't

1669
01:11:32,780 --> 01:11:33,988
care about this other person.

1670
01:11:33,988 --> 01:11:35,600
I don't feel inclined
to do very much.

1671
01:11:35,600 --> 01:11:40,310
But also, I'm not
necessarily hurting them.

1672
01:11:40,310 --> 01:11:42,830
And now I want to
just preview what

1673
01:11:42,830 --> 01:11:45,230
we're going to talk
about next time, which

1674
01:11:45,230 --> 01:11:47,930
is to say it looks
like people are pretty

1675
01:11:47,930 --> 01:11:49,055
nice in various situations.

1676
01:11:49,055 --> 01:11:51,222
It looks like in the dictator
game, they give money.

1677
01:11:51,222 --> 01:11:52,760
It looks like in
the ultimatum game,

1678
01:11:52,760 --> 01:11:54,260
they give money, and so on.

1679
01:11:54,260 --> 01:11:57,110
Looks like people
give money to charity.

1680
01:11:57,110 --> 01:12:00,920
This is what we discussed
earlier about 2% of GDP.

1681
01:12:00,920 --> 01:12:04,820
People also do a bunch
of volunteering or other

1682
01:12:04,820 --> 01:12:06,090
contributions.

1683
01:12:06,090 --> 01:12:10,310
So they give some time, about
15 hours per month on average

1684
01:12:10,310 --> 01:12:12,050
when you do surveys.

1685
01:12:12,050 --> 01:12:16,550
So it looks like people seem
fairly generous overall.

1686
01:12:16,550 --> 01:12:21,080
Now, when you think about social
recognition or social image,

1687
01:12:21,080 --> 01:12:24,350
perhaps that picture
gets less rosy overall.

1688
01:12:24,350 --> 01:12:27,050
It seems to be that
essentially, people

1689
01:12:27,050 --> 01:12:30,620
care a lot about what others
think of them, and not just

1690
01:12:30,620 --> 01:12:32,480
about the outcomes, per se.

1691
01:12:32,480 --> 01:12:34,760
So one example would be
gifts to organizations.

1692
01:12:34,760 --> 01:12:36,380
If you look at
thresholds of-- like,

1693
01:12:36,380 --> 01:12:38,338
when you look at the
Boston Symphony Orchestra,

1694
01:12:38,338 --> 01:12:40,520
and you look at the
distribution of people giving,

1695
01:12:40,520 --> 01:12:42,590
it tends to be there's
these thresholds where

1696
01:12:42,590 --> 01:12:44,780
you can give a certain amount,
and then you recognize them.

1697
01:12:44,780 --> 01:12:47,197
Like, when you go to the Boston
Symphony, in this booklet,

1698
01:12:47,197 --> 01:12:50,210
you can who's like a certain
donor of which category.

1699
01:12:50,210 --> 01:12:52,340
People tend to bunch
just above the thresholds

1700
01:12:52,340 --> 01:12:53,240
for these categories.

1701
01:12:53,240 --> 01:12:57,050
And presumably, that's because
they want to look good.

1702
01:12:57,050 --> 01:12:58,590
That's just one example.

1703
01:12:58,590 --> 01:13:01,400
So I think the next question
we're going to ask on Wednesday

1704
01:13:01,400 --> 01:13:06,080
is to say, is social recognition
a major motivation for giving?

1705
01:13:06,080 --> 01:13:08,720
So people seem to not only
care about what others get,

1706
01:13:08,720 --> 01:13:12,900
but what others think of
them when they give or not.

1707
01:13:12,900 --> 01:13:15,770
So in some ways, that's a
philosophical discussion

1708
01:13:15,770 --> 01:13:18,815
in saying, like, are people
really nice versus not?

1709
01:13:18,815 --> 01:13:20,690
And you might say, it's
sort of disappointing

1710
01:13:20,690 --> 01:13:24,530
if they're not nice in certain
situations compared to others.

1711
01:13:24,530 --> 01:13:26,570
The flip side of
that is, well, if you

1712
01:13:26,570 --> 01:13:29,390
know what kinds of
situations generate

1713
01:13:29,390 --> 01:13:31,730
altruistic or prosocial
behavior, well,

1714
01:13:31,730 --> 01:13:33,980
if you know that, then you
can essentially also create

1715
01:13:33,980 --> 01:13:37,370
situations to
foster prosociality.

1716
01:13:37,370 --> 01:13:39,860
So once we kind of know
under which structure

1717
01:13:39,860 --> 01:13:42,890
or in which kind
of circumstances

1718
01:13:42,890 --> 01:13:45,140
people are nice,
well, that gives us

1719
01:13:45,140 --> 01:13:48,050
some lever or some ways for
policy and saying, like,

1720
01:13:48,050 --> 01:13:49,820
well, you have to
design your institutions

1721
01:13:49,820 --> 01:13:52,580
or your organizations or
your company in certain ways

1722
01:13:52,580 --> 01:13:55,280
to get people to behave
nice to each other.

1723
01:13:55,280 --> 01:13:56,930
In contrast, if
you thought people

1724
01:13:56,930 --> 01:13:59,930
are either generally nice or
not, well, then in some sense,

1725
01:13:59,930 --> 01:14:01,080
that's good to know.

1726
01:14:01,080 --> 01:14:03,920
But then in some sense, we
can't do very much about that.

1727
01:14:03,920 --> 01:14:07,370
And that's what we're going
to talk about next time.

1728
01:14:07,370 --> 01:14:09,130
Thank you.