1
00:00:00,000 --> 00:00:01,992
[SQUEAKING]

2
00:00:01,992 --> 00:00:03,984
[RUSTLING]

3
00:00:03,984 --> 00:00:05,976
[CLICKING]

4
00:00:11,200 --> 00:00:14,750
PROFESSOR: So let me
very briefly recap

5
00:00:14,750 --> 00:00:19,770
what we discussed last time.

6
00:00:19,770 --> 00:00:23,820
So we started off thinking
about choices and risk.

7
00:00:23,820 --> 00:00:27,500
How do people think
about economic behavior

8
00:00:27,500 --> 00:00:31,290
when things are uncertain,
when there's risk involved?

9
00:00:31,290 --> 00:00:33,380
I sort of discussed
and showed you

10
00:00:33,380 --> 00:00:37,220
sort of the main workhorse
model that economists

11
00:00:37,220 --> 00:00:38,120
use to study risk.

12
00:00:38,120 --> 00:00:40,340
That's sort of the
expected utility.

13
00:00:40,340 --> 00:00:44,000
That is a very commonly
extremely useful

14
00:00:44,000 --> 00:00:45,530
model for many situations.

15
00:00:45,530 --> 00:00:47,270
So it's like very widely used.

16
00:00:47,270 --> 00:00:50,250
If you ask 100
economists randomly

17
00:00:50,250 --> 00:00:52,250
which model explains
choices [INAUDIBLE]

18
00:00:52,250 --> 00:00:54,110
the majority will
surely tell you expected

19
00:00:54,110 --> 00:00:55,850
utility is what you should use.

20
00:00:55,850 --> 00:00:59,000
It explains a wide range of
phenomena in very useful ways.

21
00:00:59,000 --> 00:01:01,040
You can think about lots
of different things.

22
00:01:01,040 --> 00:01:03,380
You can, for example, think
about investment behavior,

23
00:01:03,380 --> 00:01:04,920
finance.

24
00:01:04,920 --> 00:01:08,150
When you think about what
should you invest in,

25
00:01:08,150 --> 00:01:14,060
if things become more risky,
if assets are more volatile,

26
00:01:14,060 --> 00:01:16,363
you need to have a
higher return for that,

27
00:01:16,363 --> 00:01:17,780
or you need to be
offered a higher

28
00:01:17,780 --> 00:01:20,760
return to invest in those
assets and so on and so forth.

29
00:01:20,760 --> 00:01:23,180
There's lots of useful
applications in finance

30
00:01:23,180 --> 00:01:24,725
using expected utility.

31
00:01:24,725 --> 00:01:26,600
You can think of a range
of different issues.

32
00:01:26,600 --> 00:01:28,880
You can think about, for
example, criminal behavior,

33
00:01:28,880 --> 00:01:30,590
about sort of the risk
of getting caught.

34
00:01:30,590 --> 00:01:32,600
What happens when the
risk of caught goes up?

35
00:01:32,600 --> 00:01:35,282
People engage in less crime
and so on and so forth.

36
00:01:35,282 --> 00:01:36,740
There's lots of
different behaviors

37
00:01:36,740 --> 00:01:39,780
that you can think about and
explain using expected utility.

38
00:01:39,780 --> 00:01:43,250
So I do want you to sort of take
away the expected utility model

39
00:01:43,250 --> 00:01:46,075
is a very useful model
for various applications.

40
00:01:46,075 --> 00:01:47,450
What we're trying
to do is trying

41
00:01:47,450 --> 00:01:50,660
to understand are there some
applications for which perhaps

42
00:01:50,660 --> 00:01:54,530
the expected utility model
has some limitation, perhaps

43
00:01:54,530 --> 00:01:57,080
because of its simplicity
or parsimony because there's

44
00:01:57,080 --> 00:01:58,730
only one parameter in there?

45
00:01:58,730 --> 00:02:01,130
Can we sort of alter
that in some ways

46
00:02:01,130 --> 00:02:05,790
and try to make it more
realistic in some situations?

47
00:02:05,790 --> 00:02:07,815
So the key parameter
of interest when

48
00:02:07,815 --> 00:02:09,440
you try to sort of
estimate this model,

49
00:02:09,440 --> 00:02:12,035
trying to sort of match
the data in some ways,

50
00:02:12,035 --> 00:02:13,910
the parameter that you'd
estimate here is you

51
00:02:13,910 --> 00:02:15,950
need to sort of assume
some functional form.

52
00:02:15,950 --> 00:02:18,420
This is what I did last time.

53
00:02:18,420 --> 00:02:19,460
I can show you this.

54
00:02:23,650 --> 00:02:26,850
So this here, one very
commonly functional form

55
00:02:26,850 --> 00:02:30,120
is the CRRA utility function
that's very widely used

56
00:02:30,120 --> 00:02:32,530
in a lot of range of settings.

57
00:02:32,530 --> 00:02:35,460
The feature of that is it
has a constant relative risk

58
00:02:35,460 --> 00:02:38,640
aversion, and that has a
bunch of useful properties

59
00:02:38,640 --> 00:02:41,423
for estimating things
or making predictions.

60
00:02:41,423 --> 00:02:42,840
So what you then
need to do is you

61
00:02:42,840 --> 00:02:45,220
try to sort of estimate
somebody's risk preferences.

62
00:02:45,220 --> 00:02:47,508
How do people behave under risk?

63
00:02:47,508 --> 00:02:49,800
What you need to do then sort
of assume some functional

64
00:02:49,800 --> 00:02:52,052
form, for example, this
CRRA utility function.

65
00:02:52,052 --> 00:02:53,760
And then the question
is kind of like how

66
00:02:53,760 --> 00:02:56,910
do we estimate gamma the
risk aversion parameter?

67
00:02:56,910 --> 00:02:59,280
That's the key
parameter in this model.

68
00:02:59,280 --> 00:03:02,338
Now, how do you do that?

69
00:03:02,338 --> 00:03:04,380
I showed you some different
choices [INAUDIBLE]..

70
00:03:04,380 --> 00:03:05,430
Essentially we
reveal preference.

71
00:03:05,430 --> 00:03:07,305
Economists believe when
we reveal preference,

72
00:03:07,305 --> 00:03:09,160
I give you some choices
that involve risk.

73
00:03:09,160 --> 00:03:10,618
And depending what
you choose, that

74
00:03:10,618 --> 00:03:12,390
reveals what your gamma is.

75
00:03:12,390 --> 00:03:14,592
So you can do like
small scale gambles,

76
00:03:14,592 --> 00:03:17,050
which is just like small choices
between different options.

77
00:03:17,050 --> 00:03:18,790
Some entail more
risks than others,

78
00:03:18,790 --> 00:03:20,165
and then you can
essentially just

79
00:03:20,165 --> 00:03:22,440
sort of estimate
using those choices,

80
00:03:22,440 --> 00:03:25,740
or people certainty
equivalent for such choices,

81
00:03:25,740 --> 00:03:27,270
what's people's gamma is.

82
00:03:27,270 --> 00:03:29,820
Now, what we found is that when
you have small-scale gambles,

83
00:03:29,820 --> 00:03:32,100
people look or appear
very risk-averse.

84
00:03:32,100 --> 00:03:39,917
They'll often decline gambles
with positive expected value,

85
00:03:39,917 --> 00:03:42,000
which makes them appear
quite risk-averse once you

86
00:03:42,000 --> 00:03:43,830
sort of estimate
this parameter gamma.

87
00:03:43,830 --> 00:03:47,580
Gamma looks like gamma is
above 10, above 20, above 30,

88
00:03:47,580 --> 00:03:48,990
really, really high.

89
00:03:48,990 --> 00:03:52,510
Now, at the same time, you
can look at large-scale risk.

90
00:03:52,510 --> 00:03:55,278
There, when you look
at large-scale choices,

91
00:03:55,278 --> 00:03:57,570
when you sort of think about
what's a reasonable gamma,

92
00:03:57,570 --> 00:04:00,780
people actually only appear
moderately risk-averse.

93
00:04:00,780 --> 00:04:03,430
It doesn't look like they're
particularly risk-averse.

94
00:04:03,430 --> 00:04:05,940
When you sort of think for
those large-scale choices

95
00:04:05,940 --> 00:04:08,610
and we look at finance or
housing or other applications

96
00:04:08,610 --> 00:04:11,010
that people have
estimated such models,

97
00:04:11,010 --> 00:04:15,030
you get like gamma sort of
between 0 and 2 roughly.

98
00:04:15,030 --> 00:04:17,970
Now, what that then
implies is if your gamma is

99
00:04:17,970 --> 00:04:21,240
between 0 and 2, that
means essentially

100
00:04:21,240 --> 00:04:23,083
for small-scale
gambles you should

101
00:04:23,083 --> 00:04:24,250
be essentially risk neutral.

102
00:04:24,250 --> 00:04:26,083
You should not care
about really small risks

103
00:04:26,083 --> 00:04:28,443
that are about a dollar or two.

104
00:04:28,443 --> 00:04:30,360
So that sort of poses a
problem because now we

105
00:04:30,360 --> 00:04:32,220
have sort of two
contradicting answers.

106
00:04:32,220 --> 00:04:34,950
We have for small-scale
risk, it looks like people

107
00:04:34,950 --> 00:04:36,180
are really risk-averse.

108
00:04:36,180 --> 00:04:38,550
For a large-scale gamble it
looks like people are not

109
00:04:38,550 --> 00:04:39,570
so risk-averse.

110
00:04:39,570 --> 00:04:41,130
Now, we only have
like one parameter,

111
00:04:41,130 --> 00:04:43,290
is this gamma, which is
coming from the concavity

112
00:04:43,290 --> 00:04:44,727
of the utility function.

113
00:04:44,727 --> 00:04:46,560
And when we only have
one parameter and sort

114
00:04:46,560 --> 00:04:48,510
of two contradicting
pieces of evidence,

115
00:04:48,510 --> 00:04:50,280
we can't sort of
match both, right?

116
00:04:50,280 --> 00:04:52,440
Because if you match
one, then essentially

117
00:04:52,440 --> 00:04:55,240
you can match the
other and vice versa.

118
00:04:55,240 --> 00:04:59,142
Now, I showed you a little
bit Matthew Rabin's--

119
00:04:59,142 --> 00:05:01,475
what he calls the calibration
theorem, which essentially

120
00:05:01,475 --> 00:05:04,590
is sort of calibrating, showing
in a fairly compelling way

121
00:05:04,590 --> 00:05:09,030
that, in fact, you
can formally show it's

122
00:05:09,030 --> 00:05:11,220
not about sort of assumptions
of a specific utility

123
00:05:11,220 --> 00:05:12,360
function or the like.

124
00:05:12,360 --> 00:05:15,930
For these very sort of
minimal assumptions, which

125
00:05:15,930 --> 00:05:20,640
is just the utility
function is weakly concave,

126
00:05:20,640 --> 00:05:24,390
you can essentially show that
declining small-scale gambles

127
00:05:24,390 --> 00:05:28,140
with positive
expected value implies

128
00:05:28,140 --> 00:05:35,767
that people make absurd choices
when the gambles become larger.

129
00:05:35,767 --> 00:05:37,350
The recitation will
discuss this a bit

130
00:05:37,350 --> 00:05:40,320
in more detail and sort
of walking you exactly,

131
00:05:40,320 --> 00:05:43,870
in somewhat slower speeds,
through the specific example.

132
00:05:43,870 --> 00:05:46,080
Now then, where we
started then last time

133
00:05:46,080 --> 00:05:48,330
was thinking about
insurance choices.

134
00:05:48,330 --> 00:05:52,320
This is a very nice
paper by Justin Sydnor,

135
00:05:52,320 --> 00:05:55,770
and one very nice
feature of this paper

136
00:05:55,770 --> 00:05:57,420
is that it involves
real-world choices.

137
00:05:57,420 --> 00:05:59,250
So it's not like
some lab experiments

138
00:05:59,250 --> 00:06:00,840
with some experiments.

139
00:06:00,840 --> 00:06:03,365
Of course, some people care
a lot about undergrads.

140
00:06:03,365 --> 00:06:04,740
Some people might
say, well, what

141
00:06:04,740 --> 00:06:06,210
are these undergrads
choosing anyway?

142
00:06:06,210 --> 00:06:08,293
What does this have to do
with real-world choices?

143
00:06:08,293 --> 00:06:10,440
I think undergrads are great.

144
00:06:10,440 --> 00:06:13,140
But one might wonder,
like, if you recruit people

145
00:06:13,140 --> 00:06:16,200
into some experiments and you
see some choices, like what

146
00:06:16,200 --> 00:06:17,580
do these choices really reveal?

147
00:06:17,580 --> 00:06:20,190
Like, do we really find that
these choices are predictive

148
00:06:20,190 --> 00:06:21,498
of real-world behaviors?

149
00:06:21,498 --> 00:06:23,040
So one answer to
that is, well, let's

150
00:06:23,040 --> 00:06:24,550
find some data from
the real world.

151
00:06:24,550 --> 00:06:26,133
Let's look at real
choices that people

152
00:06:26,133 --> 00:06:27,600
have made in
real-world settings,

153
00:06:27,600 --> 00:06:30,480
and this is exactly
what Sydnor does.

154
00:06:30,480 --> 00:06:34,380
So what he does is he has this
data set of from a large home

155
00:06:34,380 --> 00:06:35,220
insurance provider.

156
00:06:35,220 --> 00:06:37,290
This is sort of 50,000
standard policies

157
00:06:37,290 --> 00:06:39,210
that are sort of like
representative of what

158
00:06:39,210 --> 00:06:41,490
people choose overall.

159
00:06:41,490 --> 00:06:45,180
And the key outcome of
interest in this study

160
00:06:45,180 --> 00:06:47,257
is like people's
deductible choices.

161
00:06:47,257 --> 00:06:48,090
What's a deductible?

162
00:06:48,090 --> 00:06:50,010
Again, these are expenses
paid out-of-pocket

163
00:06:50,010 --> 00:06:52,900
before the insurer
pays any expenses.

164
00:06:52,900 --> 00:06:55,180
So you have like a
deductibles of $500,

165
00:06:55,180 --> 00:06:58,413
you have a damage
of like $200, you

166
00:06:58,413 --> 00:06:59,580
have to pay it all yourself.

167
00:06:59,580 --> 00:07:02,160
If you have a damage of
$1,000, you pay the 500

168
00:07:02,160 --> 00:07:05,080
and then the insurer
pays the rest.

169
00:07:05,080 --> 00:07:07,500
And so what he has
is he has choices

170
00:07:07,500 --> 00:07:14,490
of a menu of four deductibles
for each customer or client.

171
00:07:14,490 --> 00:07:16,470
So you can see both
people's choice sets,

172
00:07:16,470 --> 00:07:18,570
and you can see people's
preferred options.

173
00:07:18,570 --> 00:07:20,320
And that allows them
to sort of say, well,

174
00:07:20,320 --> 00:07:23,310
if you have four options,
you picked one of them.

175
00:07:23,310 --> 00:07:26,740
That means you preferred that
one over all three others.

176
00:07:26,740 --> 00:07:29,430
So we can sort of
essentially put some bounds

177
00:07:29,430 --> 00:07:31,470
on people's risk aversion.

178
00:07:31,470 --> 00:07:33,260
And so we looked
at this already.

179
00:07:33,260 --> 00:07:35,318
This is kind of what
this roughly looks like.

180
00:07:35,318 --> 00:07:37,860
There are different deductibles,
which is essentially, again,

181
00:07:37,860 --> 00:07:42,570
like how much how much
you have to pay yourself

182
00:07:42,570 --> 00:07:45,342
in case of damage until the
insurance payments kick in.

183
00:07:45,342 --> 00:07:47,550
You have the premium, which
is like how much for sure

184
00:07:47,550 --> 00:07:49,200
you have to pay every year.

185
00:07:49,200 --> 00:07:53,280
And then there is the premium
relative to the $1,000 policy.

186
00:07:53,280 --> 00:07:54,735
How much more expensive is it?

187
00:07:54,735 --> 00:07:55,860
That's in the third column.

188
00:07:55,860 --> 00:07:58,860
How much more expensive is it
to choose a lower deductible

189
00:07:58,860 --> 00:08:01,170
relative to the $1,000 premium?

190
00:08:01,170 --> 00:08:03,840
And then we have people's
choices which is, in this case,

191
00:08:03,840 --> 00:08:08,400
I guess, policyholder one
shows a deductible of $250.

192
00:08:08,400 --> 00:08:11,100
It was a premium
of $661, which is

193
00:08:11,100 --> 00:08:17,430
$157 more expensive than the
$1,000 deductible option.

194
00:08:17,430 --> 00:08:18,180
OK?

195
00:08:18,180 --> 00:08:22,710
And for each
policyholder, the company

196
00:08:22,710 --> 00:08:26,730
was, in fact, sort of
providing individual prices.

197
00:08:26,730 --> 00:08:29,530
So essentially they
were looking at where

198
00:08:29,530 --> 00:08:32,669
do they live, what's the housing
value, and so on and so forth.

199
00:08:32,669 --> 00:08:33,900
Sydnor knows all of that.

200
00:08:33,900 --> 00:08:35,490
So he knows the
full set of options

201
00:08:35,490 --> 00:08:37,950
that people had available
and their actual choices,

202
00:08:37,950 --> 00:08:40,200
and the options
available vary by person.

203
00:08:44,133 --> 00:08:45,800
How do we learn now
about risk aversion?

204
00:08:45,800 --> 00:08:47,810
Well, so the losses
to the customers

205
00:08:47,810 --> 00:08:49,610
are capped by the
deductible, right?

206
00:08:49,610 --> 00:08:52,910
So any loss you have from
any damage that you get,

207
00:08:52,910 --> 00:08:56,400
the losses are only
up to the deductible.

208
00:08:56,400 --> 00:08:58,550
So if you have a
deductible of $500,

209
00:08:58,550 --> 00:09:01,670
the most you can lose or
have to pay in any case

210
00:09:01,670 --> 00:09:04,800
if any loss occurs is $500.

211
00:09:04,800 --> 00:09:06,410
So choosing a lower
deductible then,

212
00:09:06,410 --> 00:09:08,810
what it does it amounts
to essentially reducing

213
00:09:08,810 --> 00:09:10,620
that loss in case
you have a damage.

214
00:09:10,620 --> 00:09:13,370
So if you have a
deductible of $500

215
00:09:13,370 --> 00:09:17,035
and decide to instead choose
a deductible of $250, that

216
00:09:17,035 --> 00:09:18,410
means essentially
in case there's

217
00:09:18,410 --> 00:09:21,050
a damage, in case you
have to pay something,

218
00:09:21,050 --> 00:09:22,490
you don't have to pay 500.

219
00:09:22,490 --> 00:09:25,490
You have to only pay 250.

220
00:09:25,490 --> 00:09:28,080
But of course, if you
lower your deductible,

221
00:09:28,080 --> 00:09:32,270
the price of your insurance
goes up, the premium goes up,

222
00:09:32,270 --> 00:09:34,703
and the premium you
have to pay for sure.

223
00:09:34,703 --> 00:09:36,370
So the way you can
think about this then

224
00:09:36,370 --> 00:09:38,750
is like, if you choose a
lower deductible, for sure

225
00:09:38,750 --> 00:09:40,580
you have to pay more money.

226
00:09:40,580 --> 00:09:42,920
But in case there's
some damage to you

227
00:09:42,920 --> 00:09:45,775
with some probability
that happens,

228
00:09:45,775 --> 00:09:47,150
if you have some
claims, you have

229
00:09:47,150 --> 00:09:50,880
to like pay less because
your deductible is now lower.

230
00:09:50,880 --> 00:09:52,268
OK?

231
00:09:52,268 --> 00:09:53,810
So now what info do
we actually need?

232
00:09:53,810 --> 00:09:55,542
We need the available
deductibles.

233
00:09:55,542 --> 00:09:58,000
Like, essentially what are the
deductibles for each choice?

234
00:09:58,000 --> 00:10:00,050
We need the premium
for each option.

235
00:10:00,050 --> 00:10:03,740
We need the claim probabilities
and people's wealth levels

236
00:10:03,740 --> 00:10:05,610
because you have a
utility function where

237
00:10:05,610 --> 00:10:06,610
there's wealth in there.

238
00:10:06,610 --> 00:10:08,550
I'll talk about
this in one second.

239
00:10:08,550 --> 00:10:09,520
Any questions so far?

240
00:10:16,160 --> 00:10:16,940
OK.

241
00:10:16,940 --> 00:10:19,400
So now one important
feature in this I think

242
00:10:19,400 --> 00:10:21,830
was asked like last
time about like, well,

243
00:10:21,830 --> 00:10:23,060
what about the claim rates?

244
00:10:23,060 --> 00:10:24,935
Well if the claim
rates are really high

245
00:10:24,935 --> 00:10:26,810
or if people think the
claim rates are really

246
00:10:26,810 --> 00:10:29,630
high, in some sense then,
having very low deductibles

247
00:10:29,630 --> 00:10:31,550
makes a lot of
sense because then--

248
00:10:31,550 --> 00:10:33,950
and very often it happens
that you have to pay.

249
00:10:33,950 --> 00:10:38,930
Then it makes lots of sense to
have like lower deductibles.

250
00:10:38,930 --> 00:10:41,960
But it turns out claims
rates are actually very low.

251
00:10:41,960 --> 00:10:44,173
So you can see overall--

252
00:10:44,173 --> 00:10:45,590
this is like the
full sample, this

253
00:10:45,590 --> 00:10:47,962
is everybody-- people's
claim rates is 4.2%.

254
00:10:47,962 --> 00:10:49,170
These are yearly claim rates.

255
00:10:49,170 --> 00:10:52,850
This is like out of 100
customers, 4.2 per year

256
00:10:52,850 --> 00:10:55,640
actually claim any damage.

257
00:10:55,640 --> 00:10:59,490
And then it varies a little bit
also by choice of deductible.

258
00:10:59,490 --> 00:11:01,340
So there's the
people who happened

259
00:11:01,340 --> 00:11:05,880
to choose in the end like
$1,000, $500, $250, and $100.

260
00:11:05,880 --> 00:11:07,880
But for each of them
essentially, the claim rate

261
00:11:07,880 --> 00:11:09,290
is below 5%.

262
00:11:09,290 --> 00:11:11,120
So it's very low.

263
00:11:11,120 --> 00:11:12,530
OK?

264
00:11:12,530 --> 00:11:16,070
The second factor from this data
is that reducing the deductible

265
00:11:16,070 --> 00:11:17,460
is very expensive.

266
00:11:17,460 --> 00:11:21,620
So for example, this is
the full sample again.

267
00:11:21,620 --> 00:11:24,860
On average, purchasing
the insurance

268
00:11:24,860 --> 00:11:28,760
where the deductible
of $1,000 cost $615,

269
00:11:28,760 --> 00:11:31,310
we can't say very
much about that choice

270
00:11:31,310 --> 00:11:33,882
because who knows how much the
actual damages are and so on.

271
00:11:33,882 --> 00:11:35,840
In some sense, that's
sort of irrelevant for us

272
00:11:35,840 --> 00:11:36,950
what that number is.

273
00:11:36,950 --> 00:11:38,840
What we're interested
in is like what

274
00:11:38,840 --> 00:11:41,270
are the differences in costs
of different deductibles?

275
00:11:41,270 --> 00:11:43,670
How much do you have to pay
to lower your deductible

276
00:11:43,670 --> 00:11:46,320
to like $500, $250, and so on?

277
00:11:46,320 --> 00:11:48,710
Now what you see here
is like on average,

278
00:11:48,710 --> 00:11:51,840
reducing the deductible
from $1,000 to $500,

279
00:11:51,840 --> 00:11:54,290
which is sort of what this
column shows that's shown

280
00:11:54,290 --> 00:11:59,720
in red, costs $999.91.

281
00:11:59,720 --> 00:12:02,840
So if you choose then
like $500, is this

282
00:12:02,840 --> 00:12:04,220
is a risk-averse choice or not?

283
00:12:04,220 --> 00:12:07,200
How do we think about that?

284
00:12:07,200 --> 00:12:09,165
Suppose your claim
rate is like, say, 5%.

285
00:12:17,430 --> 00:12:19,120
Yes.

286
00:12:19,120 --> 00:12:22,030
AUDIENCE: Well, I think it
would be a risk-averse decision

287
00:12:22,030 --> 00:12:25,500
because you're paying $100
more and your deductible

288
00:12:25,500 --> 00:12:27,965
has gone down by $500.

289
00:12:27,965 --> 00:12:31,356
So claim rate for that back
of the envelope calculation

290
00:12:31,356 --> 00:12:33,628
would need to be about 20.

291
00:12:33,628 --> 00:12:34,420
PROFESSOR: Exactly.

292
00:12:34,420 --> 00:12:37,000
So what you're saying is
you're reducing the deductible

293
00:12:37,000 --> 00:12:39,880
from $1,000 to $500.

294
00:12:39,880 --> 00:12:44,260
Now, if you think that happens
with a 5% chance, on average

295
00:12:44,260 --> 00:12:47,600
you're going to reduce
your payments by $25.

296
00:12:47,600 --> 00:12:54,310
So 5% times $500,
which is a $25.

297
00:12:54,310 --> 00:12:57,260
But people are willing to
pay about $100 for that.

298
00:12:57,260 --> 00:13:01,930
So for sure they're paying $100,
and the benefit that they get

299
00:13:01,930 --> 00:13:04,450
is with 5% chance, at
least the average customer,

300
00:13:04,450 --> 00:13:09,920
with 5% chance they're going
to pay $500 less in case

301
00:13:09,920 --> 00:13:10,960
there's some damage.

302
00:13:10,960 --> 00:13:13,323
That looks already
pretty risk-averse.

303
00:13:13,323 --> 00:13:14,740
Because as Ben
says, surely you're

304
00:13:14,740 --> 00:13:16,948
not risk-neutral, because
then you would not do that.

305
00:13:16,948 --> 00:13:19,000
You would choose the $1,000.

306
00:13:19,000 --> 00:13:21,160
It looks fairly risk-averse.

307
00:13:21,160 --> 00:13:26,410
Now if you go down then, if you
go to like from $250 to $200,

308
00:13:26,410 --> 00:13:30,840
there's an additional $133.22.

309
00:13:30,840 --> 00:13:35,100
So that's to say reducing your
deductible by another $150,

310
00:13:35,100 --> 00:13:39,930
from $250 to $100, makes you
for sure you have to pay $133.

311
00:13:39,930 --> 00:13:42,990
And now if your chance is like
5% of getting like essentially

312
00:13:42,990 --> 00:13:47,430
a damage, that is for a
5% chance of saving $150,

313
00:13:47,430 --> 00:13:51,200
people are willing
to pay $133 for sure.

314
00:13:51,200 --> 00:13:51,900
OK?

315
00:13:51,900 --> 00:13:54,090
So now if you try to calibrate
this, what we already

316
00:13:54,090 --> 00:13:55,830
know from this,
the simple example

317
00:13:55,830 --> 00:13:58,020
is that people look
extremely risk-averse.

318
00:13:58,020 --> 00:13:58,980
OK?

319
00:13:58,980 --> 00:14:05,182
So that's kind of like the
exercise that Sydnor is doing,

320
00:14:05,182 --> 00:14:07,640
just saying, look, let's take
these choices very seriously.

321
00:14:07,640 --> 00:14:09,580
Let's look what people have
done in real-world situations.

322
00:14:09,580 --> 00:14:10,997
These are repeat
customers, people

323
00:14:10,997 --> 00:14:13,080
who have done this for
a long time and so on.

324
00:14:13,080 --> 00:14:14,320
What are people choosing?

325
00:14:14,320 --> 00:14:19,550
And if you sort of assume
expected utility, what would

326
00:14:19,550 --> 00:14:21,180
people's gamma need
to look like to be

327
00:14:21,180 --> 00:14:23,460
able to explain this data?

328
00:14:23,460 --> 00:14:26,670
And we can do this customer--
this is like the average rates.

329
00:14:26,670 --> 00:14:29,780
They can do this sort
of customer by customer.

330
00:14:29,780 --> 00:14:33,240
Now, what he then finds
is the majority of people

331
00:14:33,240 --> 00:14:34,950
choose small deductibles.

332
00:14:34,950 --> 00:14:39,090
Lots of people
choose $250, $500.

333
00:14:39,090 --> 00:14:42,990
Very few people choose
$1,000, even among people,

334
00:14:42,990 --> 00:14:45,720
and this is on the x-axis,
who have been at the company

335
00:14:45,720 --> 00:14:47,493
for 15-plus years.

336
00:14:47,493 --> 00:14:49,410
You would say like the
first time you do this,

337
00:14:49,410 --> 00:14:50,970
maybe you don't understand
your claim rate,

338
00:14:50,970 --> 00:14:53,100
you don't understand what's
going on or whatever.

339
00:14:53,100 --> 00:14:56,400
But there's people who have been
at this company for 15 years.

340
00:14:56,400 --> 00:14:58,170
They should kind of
know at some point

341
00:14:58,170 --> 00:15:02,830
that claim rates are pretty
low, at least on average.

342
00:15:02,830 --> 00:15:04,950
And so like if you have
15 years at this company,

343
00:15:04,950 --> 00:15:08,280
it's hard to believe that you'd
still think that your claim

344
00:15:08,280 --> 00:15:10,260
rate is, say, about
10% or the like

345
00:15:10,260 --> 00:15:13,470
since it just doesn't
happen very often.

346
00:15:13,470 --> 00:15:16,160
OK.

347
00:15:16,160 --> 00:15:19,310
Now, how do we think about
people choosing a deductible?

348
00:15:19,310 --> 00:15:22,250
Again, what you need is like
the following parameter.

349
00:15:22,250 --> 00:15:24,320
You need to have
the yearly premium.

350
00:15:24,320 --> 00:15:27,350
You need to have the
deductible D. You're

351
00:15:27,350 --> 00:15:29,483
assuming no other risks
to lifetime wealth, which

352
00:15:29,483 --> 00:15:31,400
is a bit of a distraction,
but essentially you

353
00:15:31,400 --> 00:15:35,270
can diversify risk and
so on and so forth.

354
00:15:35,270 --> 00:15:38,427
You also can assume that's
at most one risk per year.

355
00:15:38,427 --> 00:15:40,010
This is, again, sort
of simplification

356
00:15:40,010 --> 00:15:44,900
and doesn't really matter very
much for the probability pi.

357
00:15:44,900 --> 00:15:48,110
And then for now
at least, we assume

358
00:15:48,110 --> 00:15:52,150
accurate, subjective beliefs
about the likelihood of a loss.

359
00:15:52,150 --> 00:15:56,200
Now then, what is then the
indirect utility function

360
00:15:56,200 --> 00:15:56,700
of wealth?

361
00:15:56,700 --> 00:15:58,158
What does the
utility function look

362
00:15:58,158 --> 00:16:00,290
like depending on
these parameters?

363
00:16:00,290 --> 00:16:02,290
Can somebody explain this
what I'm showing here?

364
00:16:02,290 --> 00:16:08,010
What is this equation?

365
00:16:08,010 --> 00:16:09,324
Yes.

366
00:16:09,324 --> 00:16:13,116
AUDIENCE: The first part
says pi [INAUDIBLE] w

367
00:16:13,116 --> 00:16:15,380
minus P minus D. The
w minus P minus D

368
00:16:15,380 --> 00:16:18,290
is your wealth if
something happens.

369
00:16:18,290 --> 00:16:19,270
[INAUDIBLE] pi.

370
00:16:19,270 --> 00:16:23,200
And the 1 minus pi is on
your utility of your wealth

371
00:16:23,200 --> 00:16:25,014
if nothing bad happens.

372
00:16:25,014 --> 00:16:26,448
[INAUDIBLE]

373
00:16:26,448 --> 00:16:27,240
PROFESSOR: Exactly.

374
00:16:27,240 --> 00:16:31,260
So for sure, so with probability
1 minus pi, nothing happens.

375
00:16:31,260 --> 00:16:34,080
You have your wealth
W that you had before.

376
00:16:34,080 --> 00:16:35,730
You have to pay the
premium for sure.

377
00:16:35,730 --> 00:16:37,813
So in that case, you also
have to pay the premium.

378
00:16:37,813 --> 00:16:40,950
So you're going to end up
with W minus P, the premium.

379
00:16:40,950 --> 00:16:43,380
And then with
probability pi, you also

380
00:16:43,380 --> 00:16:47,065
have to pay the premium,
which is W minus P,

381
00:16:47,065 --> 00:16:48,690
but also you have to
pay the deductible

382
00:16:48,690 --> 00:16:50,920
because some damage occurred.

383
00:16:50,920 --> 00:16:51,420
All right.

384
00:16:51,420 --> 00:16:54,600
And then your indirect
utility-- your expected utility

385
00:16:54,600 --> 00:16:56,718
for that year is
essentially then

386
00:16:56,718 --> 00:16:58,260
the weighted average
of these things.

387
00:16:58,260 --> 00:17:01,270
And pi is essentially
the weight on that,

388
00:17:01,270 --> 00:17:03,880
which is the subjective
or, in this case,

389
00:17:03,880 --> 00:17:06,910
assumed actual probability
of a damage occurring.

390
00:17:06,910 --> 00:17:11,670
Now, I sort of said the
indirect utility function,

391
00:17:11,670 --> 00:17:14,430
utility of wealth
function, what is that?

392
00:17:14,430 --> 00:17:16,170
What's A, an indirect
utility function?

393
00:17:16,170 --> 00:17:18,630
And B, why is there wealth
in it and not consumption?

394
00:17:18,630 --> 00:17:20,510
Usually we think
people eat stuff

395
00:17:20,510 --> 00:17:23,770
and there should be
like consumption here.

396
00:17:23,770 --> 00:17:25,020
Why do we have wealth in here?

397
00:17:25,020 --> 00:17:25,829
What is this?

398
00:17:35,840 --> 00:17:36,920
Yes.

399
00:17:36,920 --> 00:17:46,290
AUDIENCE: I think they might
be like assuming [INAUDIBLE]

400
00:17:46,290 --> 00:17:47,100
PROFESSOR: Exactly.

401
00:17:47,100 --> 00:17:50,014
What is the indirect
utility function?

402
00:17:50,014 --> 00:17:51,928
AUDIENCE: [INAUDIBLE]

403
00:17:51,928 --> 00:17:52,720
PROFESSOR: Exactly.

404
00:17:52,720 --> 00:17:56,280
So usually you think what
you do is if you go back

405
00:17:56,280 --> 00:17:59,520
to like 14.01 notes or
what was done in the first

406
00:17:59,520 --> 00:18:01,470
I think recitation,
usually what you

407
00:18:01,470 --> 00:18:05,430
do is you maximize
consumption with several goods

408
00:18:05,430 --> 00:18:07,630
or one good or
whatever over time,

409
00:18:07,630 --> 00:18:09,300
and usually there's
a budget constraint

410
00:18:09,300 --> 00:18:11,592
and wealth is usually in your
budget constraint, right?

411
00:18:11,592 --> 00:18:14,160
You can only consume as much
as how much money you have.

412
00:18:14,160 --> 00:18:16,050
Could be like your
income or your wealth

413
00:18:16,050 --> 00:18:17,700
if it's over your lifetime.

414
00:18:17,700 --> 00:18:19,510
Now when you do that
and maximize it,

415
00:18:19,510 --> 00:18:20,760
then you end up at an optimum.

416
00:18:20,760 --> 00:18:23,280
What you can then do is
like essentially express

417
00:18:23,280 --> 00:18:24,810
the optimum.

418
00:18:24,810 --> 00:18:26,520
Assuming that you
have chosen optimally,

419
00:18:26,520 --> 00:18:28,170
your consumption is
saying you already

420
00:18:28,170 --> 00:18:31,130
chose whether you wanted
apples or bananas or whatever.

421
00:18:31,130 --> 00:18:34,130
If we assume that you optimize,
I can then essentially just say

422
00:18:34,130 --> 00:18:36,390
assuming that you're
optimizing, what

423
00:18:36,390 --> 00:18:39,960
is your optimized utility for
different levels of wealth?

424
00:18:39,960 --> 00:18:43,140
And usually it's a function
of wealth and prices,

425
00:18:43,140 --> 00:18:48,120
and that's what the indirect
utility function is.

426
00:18:48,120 --> 00:18:50,920
We can very briefly also
go over that in recitation.

427
00:18:50,920 --> 00:18:55,720
But if you go back to
your 1401 or other notes,

428
00:18:55,720 --> 00:18:57,870
you will see
essentially that it's

429
00:18:57,870 --> 00:18:59,760
the outcome of a
maximization problem.

430
00:18:59,760 --> 00:19:01,710
Usually it's like for
two goods or whatever.

431
00:19:01,710 --> 00:19:03,540
Like, it's like income.

432
00:19:03,540 --> 00:19:06,980
In this case, it's wealth
because it's like over--

433
00:19:06,980 --> 00:19:10,320
Yeah, it's wealth, but it
could be available income

434
00:19:10,320 --> 00:19:11,850
as well if you wanted.

435
00:19:11,850 --> 00:19:14,280
Now, what the person
is then going to do

436
00:19:14,280 --> 00:19:19,740
is like each contract gives you
an expected utility in terms

437
00:19:19,740 --> 00:19:22,500
of how much do you expect your
utility to be if you choose

438
00:19:22,500 --> 00:19:24,450
that specific contract.

439
00:19:24,450 --> 00:19:26,220
And now the maximization
problem is now

440
00:19:26,220 --> 00:19:29,520
you choose the contract J
that maximizes the expected

441
00:19:29,520 --> 00:19:33,750
utility or the expected
indirect utility as a function

442
00:19:33,750 --> 00:19:36,180
of these parameters.

443
00:19:36,180 --> 00:19:38,960
Any questions on this?

444
00:19:38,960 --> 00:19:41,260
So for each contract,
we can write down what's

445
00:19:41,260 --> 00:19:43,795
the indirect utility function.

446
00:19:43,795 --> 00:19:45,273
It depends on people's wealth.

447
00:19:45,273 --> 00:19:46,690
So we have to make
some assumption

448
00:19:46,690 --> 00:19:47,982
of how wealthy people are.

449
00:19:47,982 --> 00:19:49,690
And it depends on
these other parameters.

450
00:19:49,690 --> 00:19:53,200
It depends on the premium,
it depends on the deductible,

451
00:19:53,200 --> 00:19:57,970
and it depends on the subjective
probability of a claim

452
00:19:57,970 --> 00:19:59,200
occurring in that year.

453
00:19:59,200 --> 00:20:01,158
We assume that there's
only one claim per year.

454
00:20:05,250 --> 00:20:06,090
OK.

455
00:20:06,090 --> 00:20:11,040
So now what we can do is then
we can back out the implied risk

456
00:20:11,040 --> 00:20:14,190
aversion from people's choices.

457
00:20:14,190 --> 00:20:15,720
And in fact, what
we can do is we

458
00:20:15,720 --> 00:20:18,150
can get upper and lower bounds
on people's risk aversion

459
00:20:18,150 --> 00:20:19,275
from what they have chosen.

460
00:20:19,275 --> 00:20:21,990
Let me sort of give you
an example for that.

461
00:20:21,990 --> 00:20:26,460
Suppose a person chooses
a $100 deductible.

462
00:20:26,460 --> 00:20:27,340
What does this mean?

463
00:20:27,340 --> 00:20:30,630
Well, this means essentially
that he or she preferred

464
00:20:30,630 --> 00:20:34,650
the $100 deductible over all
the other deductibles that

465
00:20:34,650 --> 00:20:35,470
were available.

466
00:20:35,470 --> 00:20:35,970
Right?

467
00:20:35,970 --> 00:20:37,990
So essentially, if
you choose the 100,

468
00:20:37,990 --> 00:20:39,180
you get three inequalities.

469
00:20:39,180 --> 00:20:41,430
You get like the $100
deductible is better

470
00:20:41,430 --> 00:20:43,890
than the $250
deductible, the $100

471
00:20:43,890 --> 00:20:46,650
is better than $500
deductible, and $100 is

472
00:20:46,650 --> 00:20:49,500
better than $1,000 deductibles.

473
00:20:49,500 --> 00:20:51,870
Now, this gives us a bound
on people's risk aversion.

474
00:20:51,870 --> 00:20:55,690
Is it the lower or an
upper bound and why?

475
00:21:15,690 --> 00:21:16,462
Yes.

476
00:21:16,462 --> 00:21:21,900
AUDIENCE: I think it's a lower
bound because $100 [INAUDIBLE]

477
00:21:21,900 --> 00:21:29,940
is like the lowest you can
go in a lower deductible

478
00:21:29,940 --> 00:21:32,470
can cause more risk-aversion.

479
00:21:32,470 --> 00:21:33,660
PROFESSOR: Right, exactly.

480
00:21:33,660 --> 00:21:34,952
Sort of like a corner solution.

481
00:21:34,952 --> 00:21:36,930
So like, if the person
who chooses $100--

482
00:21:36,930 --> 00:21:39,097
that's the person that I
just showed you previously.

483
00:21:39,097 --> 00:21:42,370
This is the example
that I showed you here.

484
00:21:42,370 --> 00:21:43,020
Where was it?

485
00:21:45,720 --> 00:21:49,170
This is a person who looks
extremely risk-averse, right?

486
00:21:49,170 --> 00:21:50,790
This is a person
essentially saying

487
00:21:50,790 --> 00:21:54,390
like, I'm choosing the
lowest possible deductible.

488
00:21:54,390 --> 00:21:56,640
I'm for sure paying
quite a bit of money

489
00:21:56,640 --> 00:21:59,850
compared to all these other
options, 4% or 5% chance

490
00:21:59,850 --> 00:22:02,740
of not having a damage.

491
00:22:02,740 --> 00:22:05,240
So this person will
look very risk-averse.

492
00:22:05,240 --> 00:22:06,620
It's the lowest possible option.

493
00:22:06,620 --> 00:22:10,350
So maybe the person, if there
had been a $50 or zero dollars

494
00:22:10,350 --> 00:22:12,300
option, would have even
chosen that option.

495
00:22:12,300 --> 00:22:14,940
We don't know because
that's not available.

496
00:22:14,940 --> 00:22:18,420
What you can then
do is, however, you

497
00:22:18,420 --> 00:22:20,190
can just write down
these inequalities.

498
00:22:20,190 --> 00:22:22,410
And so if you choose
the $100 deductible

499
00:22:22,410 --> 00:22:24,870
compared to the
$250 deductible, it

500
00:22:24,870 --> 00:22:27,480
will be the case that
if you solve for gamma,

501
00:22:27,480 --> 00:22:30,030
this gives you a lower
bound for or gamma.

502
00:22:30,030 --> 00:22:31,320
So what does that mean?

503
00:22:31,320 --> 00:22:33,480
We know that gamma
is at least as high

504
00:22:33,480 --> 00:22:37,650
as the solution of this
inequality will tell us,

505
00:22:37,650 --> 00:22:40,200
but in fact, their gamma
could be even higher.

506
00:22:40,200 --> 00:22:43,650
We just don't know it because we
don't have additional choices.

507
00:22:43,650 --> 00:22:47,200
Now, if you choose the $1,000
deductible on the other hand,

508
00:22:47,200 --> 00:22:49,830
you'll get like an upper bound.

509
00:22:49,830 --> 00:22:52,500
The reasoning is
exactly the same.

510
00:22:52,500 --> 00:22:54,840
Essentially, if you choose
the $1,000 deductible,

511
00:22:54,840 --> 00:22:56,310
that's like the
riskiest option you

512
00:22:56,310 --> 00:22:58,620
can choose because
essentially you're

513
00:22:58,620 --> 00:23:01,590
choosing something
you will choose not

514
00:23:01,590 --> 00:23:03,450
to reduce your risk in any way.

515
00:23:03,450 --> 00:23:05,330
So you're not willing
to pay to do that.

516
00:23:05,330 --> 00:23:08,790
It's kind of like
accepting a gamble

517
00:23:08,790 --> 00:23:10,208
and not choosing
the safe options.

518
00:23:10,208 --> 00:23:12,000
But we don't know
whether this person would

519
00:23:12,000 --> 00:23:15,000
have chosen like $1,000
on one deductible,

520
00:23:15,000 --> 00:23:17,640
or $2,000, or $5,000,
what deductible would

521
00:23:17,640 --> 00:23:19,980
have been chosen because
there's no other options

522
00:23:19,980 --> 00:23:22,540
of deductibles.

523
00:23:22,540 --> 00:23:24,913
And then in between, we
have essentially lower

524
00:23:24,913 --> 00:23:26,580
and upper bounds,
because essentially we

525
00:23:26,580 --> 00:23:28,680
know that if you
choose 500, we know

526
00:23:28,680 --> 00:23:30,960
that you didn't choose
1,000, and we know also

527
00:23:30,960 --> 00:23:32,610
that you didn't choose 250.

528
00:23:32,610 --> 00:23:35,250
So your gamma must be in
between those options whatsoever

529
00:23:35,250 --> 00:23:38,970
as implied by those two options.

530
00:23:38,970 --> 00:23:41,550
There's a previous problem
set that we posted that sort

531
00:23:41,550 --> 00:23:43,200
of walks you through that.

532
00:23:43,200 --> 00:23:46,890
We'll also go through
that in recitation

533
00:23:46,890 --> 00:23:50,200
to go through the
mechanics of that.

534
00:23:50,200 --> 00:23:51,240
Any questions on this?

535
00:23:54,990 --> 00:23:55,808
Yes.

536
00:23:55,808 --> 00:23:57,950
AUDIENCE: [INAUDIBLE]

537
00:24:06,270 --> 00:24:10,590
PROFESSOR: I think that's not
a problem because it's just

538
00:24:10,590 --> 00:24:12,180
sort of flipped.

539
00:24:12,180 --> 00:24:14,280
So we have the gamma--

540
00:24:14,280 --> 00:24:17,350
that's why you have the gamma
in the denominator as well.

541
00:24:17,350 --> 00:24:19,590
So you get a problem
if your gamma is 1,

542
00:24:19,590 --> 00:24:21,750
because in dividing
by 1, usually people

543
00:24:21,750 --> 00:24:24,796
use a log utility for that.

544
00:24:24,796 --> 00:24:26,260
Yeah.

545
00:24:26,260 --> 00:24:28,700
AUDIENCE: [INAUDIBLE]

546
00:24:39,703 --> 00:24:41,370
PROFESSOR: Yes, that's
a great question.

547
00:24:41,370 --> 00:24:43,440
So I'll get to this in a second.

548
00:24:43,440 --> 00:24:46,050
So what I have done
right now, and this

549
00:24:46,050 --> 00:24:48,350
is the typical way economists
think about these things,

550
00:24:48,350 --> 00:24:50,990
is sort of say, let's take
a model very seriously.

551
00:24:50,990 --> 00:24:52,760
Here's the model
that I'm using to try

552
00:24:52,760 --> 00:24:56,210
to explain people's
choices, and I'm essentially

553
00:24:56,210 --> 00:24:57,920
assuming everything else away.

554
00:24:57,920 --> 00:25:00,440
I'm essentially assuming
that the person optimizes.

555
00:25:00,440 --> 00:25:01,310
Those no mistakes.

556
00:25:01,310 --> 00:25:02,910
There's no framing of facts.

557
00:25:02,910 --> 00:25:05,570
There's no other stuff going
on, liquidity constraints

558
00:25:05,570 --> 00:25:06,322
and so on.

559
00:25:06,322 --> 00:25:07,780
And I'm taking this
very seriously.

560
00:25:07,780 --> 00:25:09,380
I'm estimating gamma.

561
00:25:09,380 --> 00:25:11,900
Now, the typical-- and this
is section four of the paper,

562
00:25:11,900 --> 00:25:12,590
in fact--

563
00:25:12,590 --> 00:25:14,117
the typical thing
then that happens

564
00:25:14,117 --> 00:25:15,950
is other people are
going to say like, well,

565
00:25:15,950 --> 00:25:19,040
what about if people don't
understand what they're doing?

566
00:25:19,040 --> 00:25:20,850
What about people
misperceiving the risk?

567
00:25:20,850 --> 00:25:23,840
What about people framing
effects in terms of the way

568
00:25:23,840 --> 00:25:25,880
you present the choices?

569
00:25:25,880 --> 00:25:28,130
People like to not choose
extremes, but like to choose

570
00:25:28,130 --> 00:25:29,240
and the middle.

571
00:25:29,240 --> 00:25:33,060
Can that explain the results?

572
00:25:33,060 --> 00:25:35,360
There's some
concerns about that.

573
00:25:35,360 --> 00:25:37,790
I think one thing that your
explanation, for example,

574
00:25:37,790 --> 00:25:41,450
would be able perhaps to explain
is people choosing $500 over

575
00:25:41,450 --> 00:25:42,560
like $1,000.

576
00:25:42,560 --> 00:25:44,690
It's hard for the
framing effects

577
00:25:44,690 --> 00:25:47,850
to explain why people
are choosing 250.

578
00:25:47,850 --> 00:25:51,950
So if you look at this figure
in particular, the people who

579
00:25:51,950 --> 00:25:55,010
are in the company for
15 years, lots of people

580
00:25:55,010 --> 00:25:57,770
choose deductibles of 250.

581
00:25:57,770 --> 00:26:00,170
It's a little bit harder to
explain with framing effects.

582
00:26:00,170 --> 00:26:04,830
Why wouldn't you choose
like 500 as opposed to 250?

583
00:26:04,830 --> 00:26:07,395
Sydnor argues that's
really not what's going on.

584
00:26:07,395 --> 00:26:09,020
I think at the end
of the day, probably

585
00:26:09,020 --> 00:26:11,420
it's the case that people
do not want these extremes

586
00:26:11,420 --> 00:26:13,033
and, to some
degree, I think some

587
00:26:13,033 --> 00:26:14,450
of what's going
on in this perhaps

588
00:26:14,450 --> 00:26:17,000
at least contributed by some
form of framing effect, maybe

589
00:26:17,000 --> 00:26:17,510
marketing.

590
00:26:17,510 --> 00:26:21,660
People who sell the insurance
choices really want people--

591
00:26:21,660 --> 00:26:24,933
they get paid presumably
if they sell essentially

592
00:26:24,933 --> 00:26:27,350
these low deductibles because
that's how the company makes

593
00:26:27,350 --> 00:26:28,930
a lot of money.

594
00:26:28,930 --> 00:26:30,680
What Sydnor says there
is like, well, it's

595
00:26:30,680 --> 00:26:33,138
actually hard to sell people
on stuff that they don't like.

596
00:26:33,138 --> 00:26:35,390
It seems like people really
seem to want these things,

597
00:26:35,390 --> 00:26:37,973
and maybe some of that is sort
of sales pressure, but probably

598
00:26:37,973 --> 00:26:38,942
not everything.

599
00:26:38,942 --> 00:26:40,400
So I think some of
what is going on

600
00:26:40,400 --> 00:26:44,040
is a little bit hard to rule
out all of those things,

601
00:26:44,040 --> 00:26:46,943
but if you read the
paper it's reasonable.

602
00:26:46,943 --> 00:26:48,860
And since it's what I'm
going to show you next

603
00:26:48,860 --> 00:26:51,590
is like the implied
gamma's like so large

604
00:26:51,590 --> 00:26:54,050
that even if you sort of
said, OK, half of this effect

605
00:26:54,050 --> 00:26:56,240
is driven by other things,
you would get still

606
00:26:56,240 --> 00:27:00,263
like really absurdly large
estimates of risk aversion.

607
00:27:00,263 --> 00:27:01,430
But that's a great question.

608
00:27:01,430 --> 00:27:02,195
Yes.

609
00:27:02,195 --> 00:27:04,520
AUDIENCE: [INAUDIBLE]

610
00:27:22,160 --> 00:27:24,420
PROFESSOR: Well, to some
degree, in some sense,

611
00:27:24,420 --> 00:27:26,030
I think the way they sort of--

612
00:27:26,030 --> 00:27:28,340
The question was like, is
the company deliberately

613
00:27:28,340 --> 00:27:31,580
giving people choices that leads
them to choose low deductibles?

614
00:27:31,580 --> 00:27:36,020
And therefore, do we sort of
overestimate people's gamma?

615
00:27:36,020 --> 00:27:37,460
To some degree,
yes, but it's not

616
00:27:37,460 --> 00:27:40,460
like just we have low
deductibles available.

617
00:27:40,460 --> 00:27:42,560
There's a $1,000
option available.

618
00:27:42,560 --> 00:27:44,090
I think maybe what
you're alluding

619
00:27:44,090 --> 00:27:46,945
to is like there's some sales
pressure and so on going on.

620
00:27:46,945 --> 00:27:48,320
That may well be
true, and people

621
00:27:48,320 --> 00:27:51,620
sort of might emphasize risk
and make it particularly salient

622
00:27:51,620 --> 00:27:54,210
and make customers nervous and
say like, look, these floods

623
00:27:54,210 --> 00:27:56,540
and so on are going on,
and really, low deductibles

624
00:27:56,540 --> 00:27:57,800
are good.

625
00:27:57,800 --> 00:27:59,730
I think to some
degree, that's true,

626
00:27:59,730 --> 00:28:03,920
but you have to be pretty
compelling in your reasoning.

627
00:28:03,920 --> 00:28:06,230
One other comment
is like, there's

628
00:28:06,230 --> 00:28:08,990
lots of other examples of
people choosing low deductibles

629
00:28:08,990 --> 00:28:10,850
and sort of extended
warranties and so on.

630
00:28:10,850 --> 00:28:12,690
And in lots of
cases, for example,

631
00:28:12,690 --> 00:28:16,160
if you look at like iPhones or
iPads and so on and so forth,

632
00:28:16,160 --> 00:28:19,580
laptops, et cetera,
Apple in particular,

633
00:28:19,580 --> 00:28:23,000
but other companies try to
sell extended warranties

634
00:28:23,000 --> 00:28:26,420
that, if you actually did
this exact same calculation,

635
00:28:26,420 --> 00:28:30,020
are not worth
engaging in or that

636
00:28:30,020 --> 00:28:33,740
reveal essentially extreme
risk aversion among customers.

637
00:28:33,740 --> 00:28:35,953
I myself am looking at
this kind of research.

638
00:28:35,953 --> 00:28:38,370
There's lots of research that
argues that people shouldn't

639
00:28:38,370 --> 00:28:39,830
choose extended warranties.

640
00:28:39,830 --> 00:28:46,298
Of course, then when I bought
a laptop the last time,

641
00:28:46,298 --> 00:28:48,590
I was like, of course I don't
need extended warranties.

642
00:28:48,590 --> 00:28:55,610
And then, of course, soon after,
the laptop broke and so on,

643
00:28:55,610 --> 00:28:57,390
and I didn't have a
warranty and so on.

644
00:28:57,390 --> 00:29:00,140
So of course, in specific
examples that may happen,

645
00:29:00,140 --> 00:29:05,780
but on average it's
not a good idea to do.

646
00:29:05,780 --> 00:29:06,280
OK.

647
00:29:06,280 --> 00:29:09,560
So now what does Sydnor find?

648
00:29:09,560 --> 00:29:14,080
So now here in this table, you
can see the implied estimates

649
00:29:14,080 --> 00:29:15,065
of gamma.

650
00:29:15,065 --> 00:29:16,690
Remember, what we
said is what we think

651
00:29:16,690 --> 00:29:19,270
is a reasonable gamma is
somewhere between 0 and 2

652
00:29:19,270 --> 00:29:22,150
for large-scale choices.

653
00:29:22,150 --> 00:29:23,980
He has different
types of assumptions.

654
00:29:23,980 --> 00:29:27,520
He has like lower bounds
and upper bounds of gammas.

655
00:29:27,520 --> 00:29:30,970
And what essentially you
see is the gammas are like,

656
00:29:30,970 --> 00:29:33,220
depending what the assumptions
are, in the hundreds

657
00:29:33,220 --> 00:29:34,405
or in the thousands.

658
00:29:34,405 --> 00:29:36,490
It depends a little bit
what people's wealth is.

659
00:29:36,490 --> 00:29:39,110
So essentially depending on
how much you assume people--

660
00:29:39,110 --> 00:29:40,810
So the data that
he does not have

661
00:29:40,810 --> 00:29:43,990
is what is people's
wealth actually like.

662
00:29:43,990 --> 00:29:45,880
How much money do
people actually have?

663
00:29:45,880 --> 00:29:48,050
So he makes some reasonable
assumptions by saying,

664
00:29:48,050 --> 00:29:50,755
look, these are people whose
houses are like $200,000,

665
00:29:50,755 --> 00:29:52,960
$300,000, $400,000 worth.

666
00:29:52,960 --> 00:29:54,880
So presumably, and we
know from other data

667
00:29:54,880 --> 00:29:57,800
sets roughly how much money
people have available.

668
00:29:57,800 --> 00:30:00,160
So then depending on essentially
what utility function

669
00:30:00,160 --> 00:30:04,390
you assume, he is mostly
using like CRRA utility,

670
00:30:04,390 --> 00:30:09,910
like then is wealth
of like $1,000,000.

671
00:30:09,910 --> 00:30:15,050
You can look at like $100,000,
$50,000, $5,000, and so on.

672
00:30:15,050 --> 00:30:17,890
You can also use CARA utility,
which is constant absolute risk

673
00:30:17,890 --> 00:30:18,970
aversion, and so on.

674
00:30:18,970 --> 00:30:21,370
And essentially, what
you have to assume

675
00:30:21,370 --> 00:30:25,690
is like extremely low levels of
wealth or something like $5,000

676
00:30:25,690 --> 00:30:27,670
to get into like
sort of single digits

677
00:30:27,670 --> 00:30:29,980
or double digits of gamma.

678
00:30:29,980 --> 00:30:32,620
It's extremely
hard to sort of get

679
00:30:32,620 --> 00:30:34,960
estimates that are
reasonable in the sense

680
00:30:34,960 --> 00:30:39,940
that we think are actually
reasonable parameters of gamma.

681
00:30:39,940 --> 00:30:40,990
Any questions on this?

682
00:30:45,500 --> 00:30:46,420
OK.

683
00:30:46,420 --> 00:30:50,560
So now, why do people choose
those small deductibles?

684
00:30:50,560 --> 00:30:52,360
And this is what you
were saying before.

685
00:30:52,360 --> 00:30:56,257
So one classical
explanation would

686
00:30:56,257 --> 00:30:58,340
be, well, they must be
really, really risk-averse.

687
00:30:58,340 --> 00:31:00,580
Their gamma must be really high.

688
00:31:00,580 --> 00:31:02,920
There you might
sort of say, well,

689
00:31:02,920 --> 00:31:04,780
we know already that
gamma shouldn't be that

690
00:31:04,780 --> 00:31:06,100
high from some other choices.

691
00:31:06,100 --> 00:31:07,600
So that's sort of
hard to reconcile.

692
00:31:07,600 --> 00:31:08,560
Was there a question?

693
00:31:08,560 --> 00:31:09,060
No.

694
00:31:11,498 --> 00:31:13,290
Second, you could say,
well, there's really

695
00:31:13,290 --> 00:31:15,035
high objective
probability of claims.

696
00:31:15,035 --> 00:31:16,910
We know that the objectives
for probabilities

697
00:31:16,910 --> 00:31:19,063
are only 4% or 5%.

698
00:31:19,063 --> 00:31:20,730
You'd have to sort
of have probabilities

699
00:31:20,730 --> 00:31:23,550
of claims that are
like 20%, 30%, 40%

700
00:31:23,550 --> 00:31:25,470
to be able to match these data.

701
00:31:25,470 --> 00:31:27,570
Now, it be that people
have risk misperception.

702
00:31:27,570 --> 00:31:29,280
It could be that
people really think

703
00:31:29,280 --> 00:31:32,660
the probability is actually 20%
when it's, at the end, only 5

704
00:31:32,660 --> 00:31:33,580
or 4.

705
00:31:33,580 --> 00:31:35,790
Now, what's sort of
inconsistent with that

706
00:31:35,790 --> 00:31:37,710
is that like repeat
customers, people that

707
00:31:37,710 --> 00:31:40,410
have been at the company
for 10, 15, 20 years,

708
00:31:40,410 --> 00:31:42,510
are making very similar choices.

709
00:31:42,510 --> 00:31:45,360
In some sense, it's hard
to reconcile that everybody

710
00:31:45,360 --> 00:31:48,180
is misperceiving this risk
year after year after year

711
00:31:48,180 --> 00:31:50,202
and spending lots of money.

712
00:31:50,202 --> 00:31:52,410
There's some questions on
is it borrowing constraint?

713
00:31:52,410 --> 00:31:54,750
Is it like people just
don't have enough money?

714
00:31:54,750 --> 00:32:00,400
Are they worried about having
to pay these deductibles?

715
00:32:00,400 --> 00:32:02,430
That also seems quite
unlikely because in fact,

716
00:32:02,430 --> 00:32:04,740
the deductibles
are particularly--

717
00:32:04,740 --> 00:32:06,600
we're not talking
about like $5,000.

718
00:32:06,600 --> 00:32:08,100
People could sort
of, if they really

719
00:32:08,100 --> 00:32:09,030
faced these borrowing
constraints,

720
00:32:09,030 --> 00:32:10,450
they could save and so on.

721
00:32:10,450 --> 00:32:13,830
So Sydnor also argues
that that's not going on.

722
00:32:13,830 --> 00:32:16,110
There's some questions about
marketing social pressure

723
00:32:16,110 --> 00:32:17,153
and so on, which is--

724
00:32:17,153 --> 00:32:19,320
of course, the company has
very much like incentives

725
00:32:19,320 --> 00:32:22,430
to sell people these
kinds of deductibles.

726
00:32:22,430 --> 00:32:24,180
I think some of that
is probably going on,

727
00:32:24,180 --> 00:32:26,340
and it's hard to
rule out entirely.

728
00:32:26,340 --> 00:32:28,810
Again, like, it's hard to
actually sell people stuff they

729
00:32:28,810 --> 00:32:31,230
don't really want to.

730
00:32:31,230 --> 00:32:34,750
So there must be lots of social
pressure to, in fact, do that.

731
00:32:34,750 --> 00:32:35,470
Menu effects.

732
00:32:35,470 --> 00:32:38,400
Already talked about
it a little bit.

733
00:32:38,400 --> 00:32:41,730
There we think that
maybe menu effects

734
00:32:41,730 --> 00:32:44,550
can explain why
people choose to 500

735
00:32:44,550 --> 00:32:47,100
or, like the interior choices,
why they don't choose 1,000.

736
00:32:47,100 --> 00:32:48,810
It's hard to explain
with menu effects

737
00:32:48,810 --> 00:32:52,260
why people choose 50 over 500.

738
00:32:52,260 --> 00:32:54,240
So then Sydnor's
preferred explanation

739
00:32:54,240 --> 00:32:56,670
is then reference-dependent
preferences and loss aversion,

740
00:32:56,670 --> 00:32:58,560
which we're going
to talk about next.

741
00:32:58,560 --> 00:32:59,470
Yes.

742
00:32:59,470 --> 00:33:02,298
AUDIENCE: Is there any data
on whether the probability

743
00:33:02,298 --> 00:33:07,040
of the claim is correlated with
which menu option the people

744
00:33:07,040 --> 00:33:08,340
chose?

745
00:33:08,340 --> 00:33:10,320
PROFESSOR: Yes, there is.

746
00:33:10,320 --> 00:33:13,500
So you have it here.

747
00:33:13,500 --> 00:33:17,610
So it's sort of weakly
correlated in a sense.

748
00:33:17,610 --> 00:33:19,200
Like, if you choose--

749
00:33:19,200 --> 00:33:21,210
or weekly negatively correlated.

750
00:33:21,210 --> 00:33:32,700
So people who choose $1,000 have
2.5% claim rates, and the $100

751
00:33:32,700 --> 00:33:34,060
are 4.7.

752
00:33:34,060 --> 00:33:36,730
So that's kind of not quite
explaining things either.

753
00:33:36,730 --> 00:33:38,165
Yeah.

754
00:33:38,165 --> 00:33:40,290
AUDIENCE: I think the use
of [INAUDIBLE] regression

755
00:33:40,290 --> 00:33:43,700
to control for fact that
those with lower deductibles

756
00:33:43,700 --> 00:33:45,612
may claim multiple times.

757
00:33:45,612 --> 00:33:46,320
PROFESSOR: I see.

758
00:33:46,320 --> 00:33:48,480
Yes.

759
00:33:48,480 --> 00:33:49,200
Great, yes.

760
00:33:54,670 --> 00:33:56,510
OK.

761
00:33:56,510 --> 00:33:57,760
So what do we learn from this?

762
00:33:57,760 --> 00:34:00,400
I think this is very
much sort of confirming

763
00:34:00,400 --> 00:34:04,030
some of the lab evidence data on
relatively small-scale gambles.

764
00:34:04,030 --> 00:34:06,670
These are not like small-scale
gambles of like a dollar

765
00:34:06,670 --> 00:34:08,199
or 2 or 5 or 10.

766
00:34:08,199 --> 00:34:10,989
These are about several
hundreds of dollars,

767
00:34:10,989 --> 00:34:13,900
but they're not about hundreds
of thousands of dollars.

768
00:34:13,900 --> 00:34:16,840
So these are relatively small
relative to like people's

769
00:34:16,840 --> 00:34:17,584
lifetime wealth.

770
00:34:17,584 --> 00:34:19,459
And what we see for
those kinds of choices is

771
00:34:19,459 --> 00:34:24,219
it really looks
like people appear

772
00:34:24,219 --> 00:34:27,280
to be very, very risk-averse.

773
00:34:27,280 --> 00:34:29,262
This is what I was saying
already previously.

774
00:34:29,262 --> 00:34:31,179
When looking at sort of
reasonably small-scale

775
00:34:31,179 --> 00:34:31,780
choices--

776
00:34:31,780 --> 00:34:34,505
and I count the Sydnor evidence
as reasonably small-scale

777
00:34:34,505 --> 00:34:36,880
choices because it's not about
like hundreds of thousands

778
00:34:36,880 --> 00:34:37,690
of dollars--

779
00:34:37,690 --> 00:34:42,550
you essentially see
that such choices imply

780
00:34:42,550 --> 00:34:43,900
people seem very risk-averse.

781
00:34:43,900 --> 00:34:47,080
These choices imply
enormous risk aversion

782
00:34:47,080 --> 00:34:49,000
for large-scale risks.

783
00:34:49,000 --> 00:34:51,940
But people are not avoiding
all sorts of large-scale risks.

784
00:34:51,940 --> 00:34:53,780
People take on lots
of large-scale risk.

785
00:34:53,780 --> 00:34:56,989
So in some sense, that
can't be true in some ways.

786
00:34:56,989 --> 00:34:59,080
We also find that
individuals are moderately

787
00:34:59,080 --> 00:35:00,470
risk-averse a large-scale risk.

788
00:35:00,470 --> 00:35:02,740
People are taking on
some risk, as I said.

789
00:35:02,740 --> 00:35:04,750
Now, if you sort of
take that seriously

790
00:35:04,750 --> 00:35:08,500
and say, well, people might not
be risk-averse, that in turn

791
00:35:08,500 --> 00:35:10,870
implies that people are
nearly risk-neutral.

792
00:35:10,870 --> 00:35:12,670
They should be
nearly risk-neutral

793
00:35:12,670 --> 00:35:14,633
for small-scale risk.

794
00:35:14,633 --> 00:35:16,300
So it can't be that
both of these things

795
00:35:16,300 --> 00:35:18,060
are true at the same time.

796
00:35:18,060 --> 00:35:21,790
Now, in fact, there's
a much older paper

797
00:35:21,790 --> 00:35:25,750
that's a very famous and seminal
paper by Kahneman and Tversky.

798
00:35:25,750 --> 00:35:29,000
Kahneman got the
Nobel Prize, in fact,

799
00:35:29,000 --> 00:35:32,440
for this work and similar work.

800
00:35:32,440 --> 00:35:34,210
Kahneman is a
psychologist and they

801
00:35:34,210 --> 00:35:36,100
were doing psychology
experiments,

802
00:35:36,100 --> 00:35:39,080
and it just turned out that
these psychology experiments

803
00:35:39,080 --> 00:35:42,610
they're extremely influential in
affecting how economists think

804
00:35:42,610 --> 00:35:46,330
about risk and risk
preferences, and in particular,

805
00:35:46,330 --> 00:35:49,390
sort of preference-dependent
preferences.

806
00:35:49,390 --> 00:35:52,708
And so what Kahneman and Tversky
were doing at the time way

807
00:35:52,708 --> 00:35:54,250
before a lot of this
other literature

808
00:35:54,250 --> 00:35:56,160
that I just showed you--

809
00:35:56,160 --> 00:35:58,540
A, they showed even
more sort of evidence

810
00:35:58,540 --> 00:36:01,325
against the expected
utility model.

811
00:36:01,325 --> 00:36:02,950
But perhaps more
importantly, they also

812
00:36:02,950 --> 00:36:04,840
proposed an alternative
model in saying,

813
00:36:04,840 --> 00:36:06,370
look, here's some
choices that we

814
00:36:06,370 --> 00:36:10,330
think are hard to explain
and hard to rationalize

815
00:36:10,330 --> 00:36:11,930
using expected utility.

816
00:36:11,930 --> 00:36:16,510
Now here's a different model
that can explain things perhaps

817
00:36:16,510 --> 00:36:19,400
better in some situations.

818
00:36:19,400 --> 00:36:21,400
So what did Kahneman
and Tversky actually do?

819
00:36:21,400 --> 00:36:23,320
The experiments are
actually extremely

820
00:36:23,320 --> 00:36:24,490
simple in various ways.

821
00:36:24,490 --> 00:36:26,808
They're very clever
and very clean,

822
00:36:26,808 --> 00:36:27,850
but actually very simple.

823
00:36:27,850 --> 00:36:29,710
And so what do these
experiments look like?

824
00:36:29,710 --> 00:36:31,245
These are survey responses.

825
00:36:31,245 --> 00:36:32,620
Like, essentially
they just asked

826
00:36:32,620 --> 00:36:35,050
people about what would you
do in different situations.

827
00:36:35,050 --> 00:36:36,820
This is not what economists
were doing at the time.

828
00:36:36,820 --> 00:36:38,410
Economists at the
time were saying

829
00:36:38,410 --> 00:36:40,450
like, revealed
preferences are important.

830
00:36:40,450 --> 00:36:44,032
I need to make you
do actual choices.

831
00:36:44,032 --> 00:36:45,490
Whatever you say
in surveys doesn't

832
00:36:45,490 --> 00:36:48,922
matter because who knows whether
you actually mean what you say.

833
00:36:48,922 --> 00:36:51,130
It turns out that the survey
responses are actually--

834
00:36:51,130 --> 00:36:53,290
these are hypothetical
stakes, but it turns out

835
00:36:53,290 --> 00:36:54,970
that if you do this
with actual stakes,

836
00:36:54,970 --> 00:36:57,940
you find very similar results.

837
00:36:57,940 --> 00:37:01,630
And the experiments
were as follows.

838
00:37:01,630 --> 00:37:05,230
There were things like questions
like, which of the following

839
00:37:05,230 --> 00:37:08,740
would you prefer, kind of like
I showed you in the first class

840
00:37:08,740 --> 00:37:11,962
at the end of the
survey that you did.

841
00:37:11,962 --> 00:37:13,420
Would you prefer
option A, which is

842
00:37:13,420 --> 00:37:17,800
a 50% of winning $1,000 or
50% chance of winning nothing,

843
00:37:17,800 --> 00:37:22,550
versus option B is
like $450 for sure?

844
00:37:22,550 --> 00:37:25,130
And so they did a series of
these types of questions.

845
00:37:25,130 --> 00:37:29,340
Now, one of the
things that then they

846
00:37:29,340 --> 00:37:32,220
showed is one key prediction
of expected utility

847
00:37:32,220 --> 00:37:34,830
is that, as I said
before, people only

848
00:37:34,830 --> 00:37:39,592
care about final outcomes and
their associated probabilities.

849
00:37:39,592 --> 00:37:41,010
Kahneman and
Tversky show a bunch

850
00:37:41,010 --> 00:37:44,013
of different striking
contradictions of that.

851
00:37:44,013 --> 00:37:46,680
I want you to focus on the first
row-- problem three and problem

852
00:37:46,680 --> 00:37:47,700
three prime.

853
00:37:47,700 --> 00:37:51,930
And look at that and
tell me what of that

854
00:37:51,930 --> 00:37:55,745
example is contradicting
expected utility.

855
00:37:55,745 --> 00:37:57,120
I don't know if
you can see this.

856
00:38:02,770 --> 00:38:05,010
So let me just read this
for you if it's hard to see,

857
00:38:05,010 --> 00:38:07,050
but think of them for a second.

858
00:38:07,050 --> 00:38:12,780
Problem three says,
"Would you prefer

859
00:38:12,780 --> 00:38:18,435
an 80% chance of $4,000
over $3,000 for sure?"

860
00:38:18,435 --> 00:38:19,560
What do you see below then?

861
00:38:19,560 --> 00:38:21,210
These are always 100 people.

862
00:38:21,210 --> 00:38:23,010
You see the number of
people who preferred

863
00:38:23,010 --> 00:38:24,580
one option over the other.

864
00:38:24,580 --> 00:38:27,810
So 80 people preferred
the 3,000 option.

865
00:38:27,810 --> 00:38:32,560
20 people said I'd rather
have the 80% chance of $4,000.

866
00:38:32,560 --> 00:38:33,930
That's problem three.

867
00:38:33,930 --> 00:38:37,110
Problem three
prime is about what

868
00:38:37,110 --> 00:38:39,090
they call negative prospects.

869
00:38:39,090 --> 00:38:43,920
Would you prefer an 80%
chance of losing $4,000,

870
00:38:43,920 --> 00:38:47,970
or for sure losing $3,000?

871
00:38:47,970 --> 00:38:55,650
And there you see 92% choose
the first option and 80%

872
00:38:55,650 --> 00:39:01,790
choose the $3,000 loss for sure.

873
00:39:01,790 --> 00:39:05,840
Let's start with
the left example.

874
00:39:05,840 --> 00:39:08,150
What did we learn
from the left example?

875
00:39:08,150 --> 00:39:10,610
Are people risk-averse,
risk-neutral, risk-loving,

876
00:39:10,610 --> 00:39:13,963
or what have we
learned from that?

877
00:39:13,963 --> 00:39:14,880
AUDIENCE: Risk-averse.

878
00:39:14,880 --> 00:39:17,610
PROFESSOR: Risk-averse,
and why is that?

879
00:39:17,610 --> 00:39:22,306
AUDIENCE: If you calculate
the expected monetary value,

880
00:39:22,306 --> 00:39:25,470
I guess you would
expect to [INAUDIBLE]..

881
00:39:29,056 --> 00:39:29,850
PROFESSOR: Right.

882
00:39:29,850 --> 00:39:31,860
So we said before
somebody is risk neutral

883
00:39:31,860 --> 00:39:34,770
if the person is indifferent
between two options

884
00:39:34,770 --> 00:39:37,830
when they have the same
expected monetary value.

885
00:39:37,830 --> 00:39:40,530
And so in this case,
an 80% chance of $4,000

886
00:39:40,530 --> 00:39:44,477
is $3,200 in expectation, right?

887
00:39:44,477 --> 00:39:46,060
But there's a bunch
of people who say,

888
00:39:46,060 --> 00:39:51,120
I'd rather get the $3,000, which
is less than 3,200, for sure.

889
00:39:51,120 --> 00:39:52,920
Which means essentially
they take a surer--

890
00:39:52,920 --> 00:39:55,950
like an option that's
for sure they get.

891
00:39:55,950 --> 00:39:57,690
They prefer that
over the uncertainty

892
00:39:57,690 --> 00:40:01,230
of getting 0 versus 4,000,
which on average gets them more,

893
00:40:01,230 --> 00:40:03,210
3,200.

894
00:40:03,210 --> 00:40:05,580
Expected utility would say
if you choose that option,

895
00:40:05,580 --> 00:40:09,690
it must be that you
are risk-averse.

896
00:40:09,690 --> 00:40:12,510
Now, not everybody
seems risk-averse.

897
00:40:12,510 --> 00:40:15,120
There's 80 people out
of 100 choose that,

898
00:40:15,120 --> 00:40:20,520
or I think it's 80% choose
that and the remaining ones

899
00:40:20,520 --> 00:40:21,870
choose the other option.

900
00:40:21,870 --> 00:40:25,170
These other people we
don't know much about.

901
00:40:25,170 --> 00:40:26,900
They're not very
risk-averse, but they

902
00:40:26,900 --> 00:40:28,650
could be also risk-averse
just like it was

903
00:40:28,650 --> 00:40:30,310
revealed in their choice there.

904
00:40:30,310 --> 00:40:30,810
OK.

905
00:40:30,810 --> 00:40:32,352
So from the left
side, we say we know

906
00:40:32,352 --> 00:40:36,485
at least 80 people, or 80% of
the sample, is risk-averse.

907
00:40:36,485 --> 00:40:37,860
Now let's look at
the right side.

908
00:40:37,860 --> 00:40:39,277
What do you see
on the right side?

909
00:40:53,800 --> 00:40:54,567
Yes.

910
00:40:54,567 --> 00:40:56,650
AUDIENCE: Well, they seem
to be risk [INAUDIBLE],,

911
00:40:56,650 --> 00:41:03,570
because unexpected risk minus
3,200 versus minus 3,000,

912
00:41:03,570 --> 00:41:07,585
but they prefer the minus
3,200 on expectation.

913
00:41:07,585 --> 00:41:09,260
PROFESSOR: Right.

914
00:41:09,260 --> 00:41:13,740
So if you look at the left
option, it's minus 3,200.

915
00:41:13,740 --> 00:41:16,240
So the left option
has more risk,

916
00:41:16,240 --> 00:41:18,958
and it also has a lower
expected monetary value.

917
00:41:18,958 --> 00:41:20,750
As you said in the
expected monetary value,

918
00:41:20,750 --> 00:41:23,920
it's just a flip of
the problem three.

919
00:41:23,920 --> 00:41:29,440
It's minus 3,200, and the
other one is minus 3,000.

920
00:41:29,440 --> 00:41:31,540
So if you're in a
risk-neutral, surely you

921
00:41:31,540 --> 00:41:33,640
would choose the minus 3,000.

922
00:41:33,640 --> 00:41:36,860
If instead you choose
the minus 3,200, well,

923
00:41:36,860 --> 00:41:38,860
it must be that you really
appreciate that there

924
00:41:38,860 --> 00:41:40,280
is additional risk there.

925
00:41:40,280 --> 00:41:43,270
So it looks like
you're risk-loving.

926
00:41:43,270 --> 00:41:43,990
OK.

927
00:41:43,990 --> 00:41:47,170
And now we find here
that 92% of the sample

928
00:41:47,170 --> 00:41:50,800
choose the first option,
but at the same time,

929
00:41:50,800 --> 00:41:55,720
with the same people, when
they're offered or given

930
00:41:55,720 --> 00:41:58,480
the choice between problem
three, 80% of people

931
00:41:58,480 --> 00:41:59,760
choose the other option.

932
00:41:59,760 --> 00:42:01,510
So there are a bunch
of people essentially

933
00:42:01,510 --> 00:42:05,350
that choose the $3,000
for sure in problem three.

934
00:42:05,350 --> 00:42:07,600
At the same time, they
choose the minus 4,000

935
00:42:07,600 --> 00:42:12,050
with an 80% chance when
given problem three prime.

936
00:42:12,050 --> 00:42:14,350
So that means essentially
we have two choices here.

937
00:42:14,350 --> 00:42:18,160
One choice says people
are risk-averse.

938
00:42:18,160 --> 00:42:21,810
The other choice says
people are risk-loving.

939
00:42:21,810 --> 00:42:25,840
In expected utility terms in
our world, this cannot happen.

940
00:42:25,840 --> 00:42:28,560
The reason being that we have
one parameter, which is gamma.

941
00:42:28,560 --> 00:42:31,260
Gamma tells us how
risk-averse or how risk-loving

942
00:42:31,260 --> 00:42:32,970
or the risk-neutral alike.

943
00:42:32,970 --> 00:42:34,800
That parameter
tells us everything

944
00:42:34,800 --> 00:42:37,140
about your risk
preferences for all choices

945
00:42:37,140 --> 00:42:38,340
that I'm giving you.

946
00:42:38,340 --> 00:42:40,800
It cannot be that you're
simultaneously risk-loving

947
00:42:40,800 --> 00:42:41,970
and risk-averse.

948
00:42:41,970 --> 00:42:45,750
So this evidence essentially
sort of rejecting the expected

949
00:42:45,750 --> 00:42:49,195
utility model or cannot
explain this behavior.

950
00:42:49,195 --> 00:42:50,070
Does this make sense?

951
00:42:52,978 --> 00:42:54,520
I sort of wrote this
down here, but I

952
00:42:54,520 --> 00:42:57,370
think I said everything
that's to be said.

953
00:43:00,730 --> 00:43:01,720
Any questions on this?

954
00:43:08,070 --> 00:43:10,080
So where we're going
to go is essentially A,

955
00:43:10,080 --> 00:43:12,330
these people seem to
be behaving differently

956
00:43:12,330 --> 00:43:15,630
for gains versus
losses and, in addition

957
00:43:15,630 --> 00:43:18,630
to that, so not only do people
seem to dislike losses--

958
00:43:18,630 --> 00:43:20,530
I'll show you some
evidence of that--

959
00:43:20,530 --> 00:43:23,550
but in addition, people seem
to be risk-loving for losses

960
00:43:23,550 --> 00:43:25,193
and risk-averse for gains.

961
00:43:25,193 --> 00:43:27,610
And that's kind of what Kahneman
and Tversky are claiming.

962
00:43:27,610 --> 00:43:29,130
And once you make
that claim, you'll

963
00:43:29,130 --> 00:43:33,390
be able to explain these
patterns in the data.

964
00:43:33,390 --> 00:43:37,080
Now, a second thing that
they show is the following.

965
00:43:37,080 --> 00:43:38,970
They show in problem
11 and problem 12.

966
00:43:38,970 --> 00:43:43,480
I'll let you read
it for yourself.

967
00:43:43,480 --> 00:43:45,780
And essentially,
you see that 84%--

968
00:43:45,780 --> 00:43:50,885
so the choice here is in
addition to whatever you have,

969
00:43:50,885 --> 00:43:52,260
for sure you're
going to be given

970
00:43:52,260 --> 00:43:55,230
$1,000, or shekels I guess.

971
00:43:55,230 --> 00:43:59,040
You are now asked to choose
between 1,000 with a 50% chance

972
00:43:59,040 --> 00:44:01,410
and 500 for sure.

973
00:44:01,410 --> 00:44:01,980
OK?

974
00:44:01,980 --> 00:44:04,290
84% say option B.

975
00:44:04,290 --> 00:44:06,690
Problem two is, in addition
to whatever your own,

976
00:44:06,690 --> 00:44:09,480
you have been given 2,000.

977
00:44:09,480 --> 00:44:11,670
You're now asked to
choose between option

978
00:44:11,670 --> 00:44:14,490
C, which is minus
1,000 with 50% chance,

979
00:44:14,490 --> 00:44:17,928
and option D,
which is minus 500.

980
00:44:17,928 --> 00:44:25,670
69% of people here say
they choose option C. OK?

981
00:44:25,670 --> 00:44:27,200
So what's the problem with this?

982
00:44:32,320 --> 00:44:33,130
Yes.

983
00:44:33,130 --> 00:44:35,470
AUDIENCE: I mean,
at the end of it,

984
00:44:35,470 --> 00:44:38,056
you would say that
problem 11 option

985
00:44:38,056 --> 00:44:41,210
B, you end up with 1,500.

986
00:44:41,210 --> 00:44:46,200
Whereas problem 12 option B,
you still end up with 1,500.

987
00:44:46,200 --> 00:44:49,210
People were inconsistent
with these decisions

988
00:44:49,210 --> 00:44:51,068
based on how the
question's framed.

989
00:44:51,068 --> 00:44:51,860
PROFESSOR: Exactly.

990
00:44:51,860 --> 00:44:55,910
So framing matters, or
reference to points matter.

991
00:44:55,910 --> 00:44:58,330
And so what assumption of
the expected utility model

992
00:44:58,330 --> 00:45:01,805
is that rejecting?

993
00:45:01,805 --> 00:45:03,588
AUDIENCE: [INAUDIBLE]

994
00:45:03,588 --> 00:45:04,380
PROFESSOR: Exactly.

995
00:45:04,380 --> 00:45:10,640
So we postulated that only
final outcomes matter.

996
00:45:10,640 --> 00:45:13,280
Here, if you write
this down, it turns out

997
00:45:13,280 --> 00:45:15,770
option A and option C are
exactly the same in terms

998
00:45:15,770 --> 00:45:17,690
of final outcomes.

999
00:45:17,690 --> 00:45:20,480
Turns out option B and D
are also exactly the same.

1000
00:45:20,480 --> 00:45:23,051
I'll let you look at
this for a second.

1001
00:45:23,051 --> 00:45:26,150
So A and C are the same in
terms of the final outcomes.

1002
00:45:26,150 --> 00:45:29,360
B and D are also
exactly the same

1003
00:45:29,360 --> 00:45:31,380
in terms of final outcomes.

1004
00:45:31,380 --> 00:45:34,370
So that means essentially when
you compare A versus B and C

1005
00:45:34,370 --> 00:45:37,400
versus D, If you only
cared about final outcomes,

1006
00:45:37,400 --> 00:45:41,180
you cannot choose different
things here in this lottery.

1007
00:45:41,180 --> 00:45:44,150
It cannot be that your utility
is defined over just final

1008
00:45:44,150 --> 00:45:46,610
outcomes, because you just
told me you like two different

1009
00:45:46,610 --> 00:45:49,980
things for the exact same thing.

1010
00:45:49,980 --> 00:45:54,000
And there's 84% who choose
option B while 69% choose

1011
00:45:54,000 --> 00:45:57,000
option C. So there's a bunch of
people who essentially switch.

1012
00:46:00,107 --> 00:46:00,690
Is that clear?

1013
00:46:03,400 --> 00:46:04,090
OK.

1014
00:46:04,090 --> 00:46:06,430
So now, what is that?

1015
00:46:06,430 --> 00:46:07,670
What's going on here?

1016
00:46:07,670 --> 00:46:09,490
Well, where we're going
to go is like well,

1017
00:46:09,490 --> 00:46:11,733
it seems like, exactly as
you say, framing matters.

1018
00:46:11,733 --> 00:46:13,150
How you frame the
problem matters,

1019
00:46:13,150 --> 00:46:14,270
but why does it matter?

1020
00:46:14,270 --> 00:46:16,840
Well, it's because we setting
like a reference point.

1021
00:46:16,840 --> 00:46:19,180
In the first place,
in the first example,

1022
00:46:19,180 --> 00:46:22,420
the reference point is like
1,000 and there is like a gain

1023
00:46:22,420 --> 00:46:23,720
relative to 1,000.

1024
00:46:23,720 --> 00:46:27,340
We look at like people evaluate
this gamble as like gains.

1025
00:46:27,340 --> 00:46:29,350
In the second example,
the reference point

1026
00:46:29,350 --> 00:46:32,080
is set to like 2,000, and
now people essentially

1027
00:46:32,080 --> 00:46:35,800
evaluate this gamble as
losses relative to the 2,000.

1028
00:46:35,800 --> 00:46:37,390
And of depending
on how people think

1029
00:46:37,390 --> 00:46:39,860
about gains versus
losses, it turns out

1030
00:46:39,860 --> 00:46:42,625
that people make
different choices.

1031
00:46:42,625 --> 00:46:44,920
I'll be more formal about
this, but that's essentially

1032
00:46:44,920 --> 00:46:47,140
the idea of, like in
Kahneman and Tversky,

1033
00:46:47,140 --> 00:46:49,730
their prospect
theory they proposed,

1034
00:46:49,730 --> 00:46:55,600
why can explain these behaviors
as opposed to expected utility?

1035
00:46:55,600 --> 00:46:56,170
OK.

1036
00:46:56,170 --> 00:46:58,570
So now what are sort of
the most important points?

1037
00:46:58,570 --> 00:46:59,750
There's three of them.

1038
00:46:59,750 --> 00:47:01,333
I'm going to show
you two and then get

1039
00:47:01,333 --> 00:47:03,910
to the third one at the end.

1040
00:47:03,910 --> 00:47:07,570
So one is like what matters
a lot for people's behaviors

1041
00:47:07,570 --> 00:47:09,920
is changes rather than levels.

1042
00:47:09,920 --> 00:47:13,450
So what they argue is utility
is not defined by people's final

1043
00:47:13,450 --> 00:47:15,910
status-- how much
they end up with--

1044
00:47:15,910 --> 00:47:19,660
but as opposed to
like how much changes

1045
00:47:19,660 --> 00:47:21,820
relative to some
reference point.

1046
00:47:21,820 --> 00:47:24,650
That could be changes
relative to like a status quo.

1047
00:47:24,650 --> 00:47:27,400
How much do I have
right now and how much

1048
00:47:27,400 --> 00:47:29,800
does it change
positively or negatively?

1049
00:47:29,800 --> 00:47:32,050
Or, it could be relative
to some expectation

1050
00:47:32,050 --> 00:47:33,490
or some other reference point.

1051
00:47:33,490 --> 00:47:35,350
What I showed you
here in this example

1052
00:47:35,350 --> 00:47:38,140
is like this is not the status
quo that people evaluate

1053
00:47:38,140 --> 00:47:41,440
their utility against because
the outcome is always the same.

1054
00:47:41,440 --> 00:47:43,930
The reference point here is
kind of like the expectation

1055
00:47:43,930 --> 00:47:46,450
I sent you and say,
OK, here's 1,000.

1056
00:47:46,450 --> 00:47:48,490
And now what seems
to be happening

1057
00:47:48,490 --> 00:47:51,820
is that people's expectations,
in fact, become 1,000.

1058
00:47:51,820 --> 00:47:53,410
And then relative
to that expectation,

1059
00:47:53,410 --> 00:47:57,220
people are going to
evaluate gains and losses.

1060
00:47:57,220 --> 00:48:00,250
And similarly, if I give
you-- like, say you get 2,000,

1061
00:48:00,250 --> 00:48:02,560
again, people will
evaluate the utility

1062
00:48:02,560 --> 00:48:08,080
or the outcomes as gains
and losses relative to that.

1063
00:48:08,080 --> 00:48:12,400
Second, there seems
to be loss aversion.

1064
00:48:12,400 --> 00:48:14,500
Losses loom larger than gains.

1065
00:48:14,500 --> 00:48:16,870
That is to say people
dislike a loss that's

1066
00:48:16,870 --> 00:48:21,850
as large as a gain by a lot
more than they like the gain.

1067
00:48:21,850 --> 00:48:22,370
OK?

1068
00:48:22,370 --> 00:48:26,470
So it's like if you say you
lose 100 versus you gain 100,

1069
00:48:26,470 --> 00:48:28,840
people really dislike
losing 100 a lot more

1070
00:48:28,840 --> 00:48:32,350
than they like gaining 100.

1071
00:48:32,350 --> 00:48:34,160
And once you sort
of postulate that,

1072
00:48:34,160 --> 00:48:37,000
well then, that can explain why
you reject a bunch of gambles.

1073
00:48:37,000 --> 00:48:41,260
Because if say if the gamble is
like minus 10 with 50% chance

1074
00:48:41,260 --> 00:48:44,410
and plus 11 with
50% chance, well,

1075
00:48:44,410 --> 00:48:46,810
if you dislike the losses
a lot more, if you dislike

1076
00:48:46,810 --> 00:48:51,430
losing $10 a lot more than
gaining $10 or $11, well then,

1077
00:48:51,430 --> 00:48:54,130
you're going to
reject this gamble

1078
00:48:54,130 --> 00:48:59,440
even if the expected
value is positive.

1079
00:48:59,440 --> 00:49:00,192
OK?

1080
00:49:00,192 --> 00:49:02,650
We'll write down sort of like
a utility function next time.

1081
00:49:02,650 --> 00:49:04,750
That's sort of does
this more formally,

1082
00:49:04,750 --> 00:49:07,270
but essentially those
are the two key areas.

1083
00:49:07,270 --> 00:49:10,370
There's a third one, which I'm
going to get to in a second.

1084
00:49:10,370 --> 00:49:13,300
Now, the key part here is
like there is essentially

1085
00:49:13,300 --> 00:49:14,260
reference dependence.

1086
00:49:14,260 --> 00:49:18,880
People evaluate their outcomes
relative to some reference

1087
00:49:18,880 --> 00:49:20,038
point.

1088
00:49:20,038 --> 00:49:21,580
What kinds of examples
do we actually

1089
00:49:21,580 --> 00:49:22,540
have a reference point?

1090
00:49:22,540 --> 00:49:23,920
When you look in the
world-- and again,

1091
00:49:23,920 --> 00:49:26,110
that's kind of very much
what I'd like you to do--

1092
00:49:26,110 --> 00:49:28,510
when you think about things
that you see in the world,

1093
00:49:28,510 --> 00:49:31,270
what kinds of examples do
we have that people care

1094
00:49:31,270 --> 00:49:34,270
about reference points as
opposed to final outcomes?

1095
00:49:46,140 --> 00:49:47,354
Yeah.

1096
00:49:47,354 --> 00:49:51,448
AUDIENCE: My first thought was,
you know how we have to pay

1097
00:49:51,448 --> 00:49:54,900
$0.10 for shopping bags?

1098
00:49:54,900 --> 00:49:57,226
Where I was from a
couple years ago,

1099
00:49:57,226 --> 00:50:00,653
we did it so that if
you used a reusable bag,

1100
00:50:00,653 --> 00:50:02,006
you got $0.05 back.

1101
00:50:02,006 --> 00:50:05,470
That was like not effective.

1102
00:50:05,470 --> 00:50:09,970
[INAUDIBLE]

1103
00:50:09,970 --> 00:50:11,350
PROFESSOR: Right.

1104
00:50:11,350 --> 00:50:13,540
Yeah.

1105
00:50:13,540 --> 00:50:14,200
So exactly.

1106
00:50:14,200 --> 00:50:16,420
When you think about pricing
of different options,

1107
00:50:16,420 --> 00:50:17,830
it matters a lot to people.

1108
00:50:17,830 --> 00:50:20,440
It seems that when
you add $0.05,

1109
00:50:20,440 --> 00:50:23,410
were to subtract $0.05
or $0.10 or whatever,

1110
00:50:23,410 --> 00:50:26,530
the same changes about like
it's just $0.05 at the end

1111
00:50:26,530 --> 00:50:29,392
of the day that you have
to pay more or less,

1112
00:50:29,392 --> 00:50:30,850
depending on whether
you use a bag,

1113
00:50:30,850 --> 00:50:33,400
your choice of using a bag
versus not should not be

1114
00:50:33,400 --> 00:50:37,990
affected by whether it's framed
or put as like a loss versus

1115
00:50:37,990 --> 00:50:42,045
a gain or like a whatever rebate
or whatever that you might get.

1116
00:50:42,045 --> 00:50:44,170
And I think that's true
for many different-- that's

1117
00:50:44,170 --> 00:50:48,900
true for shopping plastic bags
or any other shopping bags,

1118
00:50:48,900 --> 00:50:50,650
but it's also true for
many other options.

1119
00:50:50,650 --> 00:50:53,050
It depends a lot whether you
sort of add on that option

1120
00:50:53,050 --> 00:50:54,730
or whether you get
sort of essentially

1121
00:50:54,730 --> 00:50:57,025
subtracted the option
as some discount.

1122
00:50:59,660 --> 00:51:01,535
Yes.

1123
00:51:01,535 --> 00:51:03,510
AUDIENCE: [INAUDIBLE]

1124
00:51:12,992 --> 00:51:13,700
PROFESSOR: Right.

1125
00:51:13,700 --> 00:51:15,710
I think some of it
is not exactly--

1126
00:51:15,710 --> 00:51:18,020
I think there's some
legal constraints

1127
00:51:18,020 --> 00:51:20,565
to that specific behavior
because you're not

1128
00:51:20,565 --> 00:51:22,940
supposed to sort of trick
people, but surely some of that

1129
00:51:22,940 --> 00:51:24,530
is going on.

1130
00:51:24,530 --> 00:51:25,770
People love discounts.

1131
00:51:25,770 --> 00:51:27,950
Like, they love
making good deals.

1132
00:51:27,950 --> 00:51:29,990
They love sort of
getting things cheaper

1133
00:51:29,990 --> 00:51:32,342
in various ways compared
to some reference point.

1134
00:51:32,342 --> 00:51:33,800
And the reference
point, since it's

1135
00:51:33,800 --> 00:51:35,810
hard to tell what actually
should the price be,

1136
00:51:35,810 --> 00:51:37,280
is often like the
previous price.

1137
00:51:37,280 --> 00:51:39,500
It's something that
seems really expensive

1138
00:51:39,500 --> 00:51:42,590
and now they're getting
it for less money.

1139
00:51:42,590 --> 00:51:44,532
And so if you get it
for half the price,

1140
00:51:44,532 --> 00:51:45,990
even if it's still
quite expensive,

1141
00:51:45,990 --> 00:51:50,030
you think like you saved
half of the price somehow.

1142
00:51:50,030 --> 00:51:51,530
AUDIENCE: So how
you feel about what

1143
00:51:51,530 --> 00:51:54,482
you have may depend on what
you see your neighbor having.

1144
00:51:54,482 --> 00:51:55,190
PROFESSOR: Right.

1145
00:51:55,190 --> 00:51:56,810
So reference points
could not just

1146
00:51:56,810 --> 00:52:00,980
be prices or sort of
things that affect yourself

1147
00:52:00,980 --> 00:52:04,100
in certain ways, and price
is the key example overall.

1148
00:52:04,100 --> 00:52:06,440
The reference point could
just be your environment,

1149
00:52:06,440 --> 00:52:07,520
your social environment.

1150
00:52:07,520 --> 00:52:09,228
We kind of talk about
this to some degree

1151
00:52:09,228 --> 00:52:11,120
also when we talk about
social preferences.

1152
00:52:11,120 --> 00:52:12,920
People care a lot
about what others do,

1153
00:52:12,920 --> 00:52:14,628
and their reference
point might very much

1154
00:52:14,628 --> 00:52:16,490
be formed by what
their neighbors do.

1155
00:52:16,490 --> 00:52:17,760
Like, how big is their house?

1156
00:52:17,760 --> 00:52:19,430
How big is their car?

1157
00:52:19,430 --> 00:52:21,065
Do they have a
swimming pool or not?

1158
00:52:21,065 --> 00:52:22,940
It could be also like
your neighbor's-- like,

1159
00:52:22,940 --> 00:52:24,107
how good are they at school?

1160
00:52:24,107 --> 00:52:26,000
Are they smart,
how good looking?

1161
00:52:26,000 --> 00:52:27,220
And so on and so forth.

1162
00:52:27,220 --> 00:52:29,030
So when people evaluate
certain outcomes,

1163
00:52:29,030 --> 00:52:32,390
they often don't evaluate
the levels, but rather

1164
00:52:32,390 --> 00:52:34,940
kind of how much
another person makes

1165
00:52:34,940 --> 00:52:37,160
or whatever other
people's outcomes are.

1166
00:52:37,160 --> 00:52:38,750
And part of the
reason might be that

1167
00:52:38,750 --> 00:52:40,857
like evaluating absolute
outcomes is really hard.

1168
00:52:40,857 --> 00:52:42,440
It's very hard to
actually understand.

1169
00:52:42,440 --> 00:52:45,920
Is $10,000, $50,000,
$100,000, a million--

1170
00:52:45,920 --> 00:52:49,850
how much money do you
actually need to be happy

1171
00:52:49,850 --> 00:52:53,330
or that you like, or
what kinds of outcomes

1172
00:52:53,330 --> 00:52:55,400
are you excited about and so on?

1173
00:52:55,400 --> 00:52:57,350
But it's much easier to
say you have something

1174
00:52:57,350 --> 00:52:59,270
and another person does not
or the other way around.

1175
00:52:59,270 --> 00:53:01,130
It's much easier to
compare with others.

1176
00:53:01,130 --> 00:53:04,040
It's very hard to actually
evaluate absolute outcomes,

1177
00:53:04,040 --> 00:53:07,890
because who knows how you
should feel about that.

1178
00:53:07,890 --> 00:53:09,440
Much easier to say
I have this, they

1179
00:53:09,440 --> 00:53:12,600
don't, or the other way around.

1180
00:53:12,600 --> 00:53:13,860
Yeah.

1181
00:53:13,860 --> 00:53:15,660
AUDIENCE: People
sometimes don't want

1182
00:53:15,660 --> 00:53:20,730
to let go of fallen
stock they own

1183
00:53:20,730 --> 00:53:27,390
because they don't want to see
that they sold at a good price.

1184
00:53:27,390 --> 00:53:28,450
PROFESSOR: Right.

1185
00:53:28,450 --> 00:53:28,950
Exactly.

1186
00:53:28,950 --> 00:53:31,950
That's called what people
call the disposition effect.

1187
00:53:31,950 --> 00:53:34,200
There's actually
pretty large literature

1188
00:53:34,200 --> 00:53:35,725
studying this and
lots of debates

1189
00:53:35,725 --> 00:53:37,350
on whether it's really
going on or not,

1190
00:53:37,350 --> 00:53:40,420
how important it is,
and so on and so forth.

1191
00:53:40,420 --> 00:53:42,000
But sort of one
very basic stylized

1192
00:53:42,000 --> 00:53:45,060
fact is that when people are
looking at stocks that they

1193
00:53:45,060 --> 00:53:48,420
might want to sell, they
are much more likely to sell

1194
00:53:48,420 --> 00:53:50,440
winners compared to losers.

1195
00:53:50,440 --> 00:53:54,070
Now, why is that bad, or
should you not do that?

1196
00:53:57,120 --> 00:54:04,957
AUDIENCE: Is that because
the winners [INAUDIBLE]

1197
00:54:04,957 --> 00:54:06,540
PROFESSOR: There's
a bit of a question

1198
00:54:06,540 --> 00:54:09,300
of is there like momentum
or reversal or the like.

1199
00:54:09,300 --> 00:54:11,610
But if you believe in
efficient markets, which

1200
00:54:11,610 --> 00:54:14,670
many economists do,
or at least some

1201
00:54:14,670 --> 00:54:18,600
of them who are in Chicago--

1202
00:54:18,600 --> 00:54:20,280
if you believe in
efficient markets,

1203
00:54:20,280 --> 00:54:24,630
then the current stock
price should essentially

1204
00:54:24,630 --> 00:54:26,700
not be-- or like
the previous losses

1205
00:54:26,700 --> 00:54:28,920
and so on should not be
informative of what's

1206
00:54:28,920 --> 00:54:30,220
going on in the future.

1207
00:54:30,220 --> 00:54:32,730
So in expectations, your
losers and your winners,

1208
00:54:32,730 --> 00:54:35,130
if there's any information
about the future,

1209
00:54:35,130 --> 00:54:37,320
about the future valuation,
that should already

1210
00:54:37,320 --> 00:54:39,065
be incorporated in the price.

1211
00:54:39,065 --> 00:54:40,440
So if you look at
two stocks, one

1212
00:54:40,440 --> 00:54:42,660
lost some money and
one gained some money,

1213
00:54:42,660 --> 00:54:45,730
they're equally likely
to make you money.

1214
00:54:45,730 --> 00:54:48,210
And so if anything, what
they show in these papers

1215
00:54:48,210 --> 00:54:49,440
is like there's momentum.

1216
00:54:49,440 --> 00:54:51,732
This is what you're saying
is that the winners actually

1217
00:54:51,732 --> 00:54:55,020
are, in fact, going to be more
likely to try to increase value

1218
00:54:55,020 --> 00:54:56,250
compared to the losers.

1219
00:54:56,250 --> 00:55:00,720
But people tend to essentially
want to sort of realize gains.

1220
00:55:00,720 --> 00:55:02,790
People seem to be happy
about making gains.

1221
00:55:02,790 --> 00:55:06,960
They seem to be very
reluctant to sell the losers,

1222
00:55:06,960 --> 00:55:09,900
and there's some questions on
how costly is that actually,

1223
00:55:09,900 --> 00:55:13,022
but that's a very robust
pattern in the data.

1224
00:55:13,022 --> 00:55:13,980
The same is also true--

1225
00:55:13,980 --> 00:55:16,050
I'm going to show you
on Monday, the same

1226
00:55:16,050 --> 00:55:17,520
is also true for houses.

1227
00:55:17,520 --> 00:55:20,970
People are much less
likely to sell their house

1228
00:55:20,970 --> 00:55:24,420
when it has lost value compared
to when it gained value.

1229
00:55:24,420 --> 00:55:27,120
I'm controlling for
a bunch of things.

1230
00:55:27,120 --> 00:55:28,600
Yeah.

1231
00:55:28,600 --> 00:55:30,839
AUDIENCE: [INAUDIBLE]

1232
00:55:38,118 --> 00:55:39,910
PROFESSOR: Yeah, so
that's a little tricky.

1233
00:55:39,910 --> 00:55:42,170
There might be other things
going on, but exactly.

1234
00:55:42,170 --> 00:55:45,500
It could just be that like
if you don't go to a movie

1235
00:55:45,500 --> 00:55:47,390
when you have bought
a ticket yourself,

1236
00:55:47,390 --> 00:55:50,300
it's sort of perceived as a loss
and you really don't like that.

1237
00:55:50,300 --> 00:55:51,800
It's a little
complicated to think

1238
00:55:51,800 --> 00:55:53,883
about this in simple terms,
because in some sense,

1239
00:55:53,883 --> 00:55:55,697
the loss is still there.

1240
00:55:55,697 --> 00:55:56,780
But anyway, yeah, exactly.

1241
00:55:56,780 --> 00:55:57,990
It's like you lost money.

1242
00:55:57,990 --> 00:56:00,230
It's almost like as if
you lost the movie ticket

1243
00:56:00,230 --> 00:56:01,965
and didn't get
some value from it.

1244
00:56:01,965 --> 00:56:04,340
There's a bit of a question
whether you sort of integrate

1245
00:56:04,340 --> 00:56:06,233
these two things or not
because people think

1246
00:56:06,233 --> 00:56:08,150
about monetary terms
versus other things often

1247
00:56:08,150 --> 00:56:10,400
in separation, but I think
some of that is exactly.

1248
00:56:10,400 --> 00:56:13,460
People feel like they have
a loss if they don't take

1249
00:56:13,460 --> 00:56:14,877
advantage of the movie tickets.

1250
00:56:14,877 --> 00:56:17,210
So I'm going to show you some
examples that are actually

1251
00:56:17,210 --> 00:56:19,610
much more basic in some ways.

1252
00:56:19,610 --> 00:56:22,360
When you think about
like visual illusions,

1253
00:56:22,360 --> 00:56:24,860
this is what's called like
the size contrast illusion.

1254
00:56:24,860 --> 00:56:26,030
So one of those
things is like when

1255
00:56:26,030 --> 00:56:27,530
you look at circles
or things that's

1256
00:56:27,530 --> 00:56:30,257
supposed to look the same,
and in fact are the same size,

1257
00:56:30,257 --> 00:56:31,340
they look quite different.

1258
00:56:31,340 --> 00:56:33,263
Depending on what you
contrast things with,

1259
00:56:33,263 --> 00:56:34,430
things look quite different.

1260
00:56:34,430 --> 00:56:36,770
For example, if you look at
these circles in the middle,

1261
00:56:36,770 --> 00:56:38,540
they're in fact
exactly the same size,

1262
00:56:38,540 --> 00:56:40,800
but in fact, they don't
really look like that.

1263
00:56:40,800 --> 00:56:42,290
Similarly, if you look
at these two circles,

1264
00:56:42,290 --> 00:56:43,290
it's perhaps more stark.

1265
00:56:43,290 --> 00:56:45,560
These circles are, in
fact, exactly the same size

1266
00:56:45,560 --> 00:56:46,530
that are in the middle.

1267
00:56:46,530 --> 00:56:47,840
Every time I'm
teaching this class,

1268
00:56:47,840 --> 00:56:50,173
I sort of have to convince
myself that they are actually

1269
00:56:50,173 --> 00:56:50,760
the same size.

1270
00:56:50,760 --> 00:56:52,370
So I print it out
and sort of measure

1271
00:56:52,370 --> 00:56:54,870
to make sure that it's actually
the same size, and they are.

1272
00:56:54,870 --> 00:56:56,720
I checked.

1273
00:56:56,720 --> 00:56:59,190
But when you see this, even
if you know it's an illusion,

1274
00:56:59,190 --> 00:57:02,090
it's very hard to convince
yourself that it's not.

1275
00:57:02,090 --> 00:57:03,535
But they are the same size.

1276
00:57:03,535 --> 00:57:04,910
When you look at
these bars, when

1277
00:57:04,910 --> 00:57:07,243
you look at the upper black
bar and the lower black bar,

1278
00:57:07,243 --> 00:57:10,467
again, it seems like somehow
the upper one is like wider,

1279
00:57:10,467 --> 00:57:12,050
but in fact, it's
not, and again, it's

1280
00:57:12,050 --> 00:57:15,050
all about the contrast
to the other side.

1281
00:57:15,050 --> 00:57:17,240
There's this one which
throws me off the most.

1282
00:57:17,240 --> 00:57:22,190
When you look at fields A and
B, they are the same color.

1283
00:57:22,190 --> 00:57:24,595
That's hard to believe
once you see it.

1284
00:57:24,595 --> 00:57:25,970
They are actually
the same color,

1285
00:57:25,970 --> 00:57:27,710
and again, I would print
it out and actually

1286
00:57:27,710 --> 00:57:28,877
put them next to each other.

1287
00:57:28,877 --> 00:57:30,140
They are the exact same color.

1288
00:57:30,140 --> 00:57:30,865
I checked.

1289
00:57:30,865 --> 00:57:32,240
There's also a
video that you can

1290
00:57:32,240 --> 00:57:34,460
watch that sort of shows this.

1291
00:57:34,460 --> 00:57:36,650
And exactly what's
happening is the way

1292
00:57:36,650 --> 00:57:41,360
we perceive colors is
coming from contrast, right?

1293
00:57:41,360 --> 00:57:45,440
So like black does not or gray
doesn't look as gray when white

1294
00:57:45,440 --> 00:57:46,730
or darker gray is next to it.

1295
00:57:50,090 --> 00:57:53,030
Similarly here, if you look
at the gray of this bar,

1296
00:57:53,030 --> 00:57:54,410
this is the exact same gray.

1297
00:57:54,410 --> 00:57:56,120
Again, you can print it
out and sort of look at it.

1298
00:57:56,120 --> 00:57:56,960
It's the exact same.

1299
00:57:56,960 --> 00:57:58,335
When you look at
it, it just does

1300
00:57:58,335 --> 00:58:01,580
look like on the right side
it's darker than on the left.

1301
00:58:01,580 --> 00:58:04,730
Now, there's plenty of examples
of reference dependence

1302
00:58:04,730 --> 00:58:05,685
for vision.

1303
00:58:05,685 --> 00:58:06,560
There's tons of them.

1304
00:58:06,560 --> 00:58:08,540
They're kind of
quite interesting.

1305
00:58:08,540 --> 00:58:10,820
At some level, it tells us
something about the brain.

1306
00:58:10,820 --> 00:58:14,210
Like, in some ways, we can learn
about essentially the way we

1307
00:58:14,210 --> 00:58:17,030
evaluate certain outcomes
overall is essentially--

1308
00:58:17,030 --> 00:58:19,370
it's much easier for us, or
we think about in contrast,

1309
00:58:19,370 --> 00:58:22,332
it's very hard for us often
to think about levels.

1310
00:58:22,332 --> 00:58:23,540
But of course, that's vision.

1311
00:58:23,540 --> 00:58:26,510
So in some sense, what did we
learn really about utility?

1312
00:58:26,510 --> 00:58:31,370
One thing you can look at is
bronze and silver medal winners

1313
00:58:31,370 --> 00:58:32,270
at the Olympics.

1314
00:58:32,270 --> 00:58:35,090
Presumably or arguably,
winning a silver medal

1315
00:58:35,090 --> 00:58:37,370
is better than a
bronze medal winner.

1316
00:58:37,370 --> 00:58:39,710
If you look at these
two women, one of them

1317
00:58:39,710 --> 00:58:43,003
won the silver medal,
one of them won bronze.

1318
00:58:43,003 --> 00:58:45,170
So what people have done
and psychologists have done

1319
00:58:45,170 --> 00:58:46,940
is like took actually
a bunch of pictures

1320
00:58:46,940 --> 00:58:50,600
from like ceremonies of
bronze and silver winners

1321
00:58:50,600 --> 00:58:52,680
and just looked at
who looks happier.

1322
00:58:52,680 --> 00:58:55,180
And when you do
that, essentially

1323
00:58:55,180 --> 00:58:57,680
the bronze medalists look on
average happier than the silver

1324
00:58:57,680 --> 00:58:58,580
medalists.

1325
00:58:58,580 --> 00:59:01,160
Presumably it's because
they just missed gold

1326
00:59:01,160 --> 00:59:03,860
while the other person
sort of won bronze

1327
00:59:03,860 --> 00:59:09,080
because it's great to be
third as opposed to fourth.

1328
00:59:09,080 --> 00:59:09,950
And there's more.

1329
00:59:09,950 --> 00:59:12,620
You can read more about this,
but that's sort of like one

1330
00:59:12,620 --> 00:59:13,610
finding.

1331
00:59:13,610 --> 00:59:19,790
Now, it's true for lots
of different feelings,

1332
00:59:19,790 --> 00:59:22,330
perceptions,
judgment, and so on.

1333
00:59:22,330 --> 00:59:25,413
People compare stimuli
in various ways

1334
00:59:25,413 --> 00:59:26,830
when it comes to
temperature, when

1335
00:59:26,830 --> 00:59:30,800
it comes to all sorts of things
relative to reference levels.

1336
00:59:30,800 --> 00:59:34,970
It's very hard for them to
evaluate it in absolute terms.

1337
00:59:34,970 --> 00:59:37,220
So it's easy to say--
when you look at water,

1338
00:59:37,220 --> 00:59:39,160
it's very hard to say
what's the temperature

1339
00:59:39,160 --> 00:59:41,080
even if you sort of
practice it a lot.

1340
00:59:41,080 --> 00:59:43,480
It's very easy to say
one bucket of water

1341
00:59:43,480 --> 00:59:44,690
is warmer than another.

1342
00:59:44,690 --> 00:59:48,760
It's very hard to sort of say
it's like 70 degrees or 60

1343
00:59:48,760 --> 00:59:49,270
or whatever.

1344
00:59:49,270 --> 00:59:50,770
It's very hard to
sort of understand

1345
00:59:50,770 --> 00:59:52,630
what absolute
temperatures are, and I

1346
00:59:52,630 --> 00:59:59,770
think that's the same in some
ways for a lot of consumption

1347
00:59:59,770 --> 01:00:00,650
or other decisions.

1348
01:00:00,650 --> 01:00:02,858
It's very hard to say how
much would you pay for this

1349
01:00:02,858 --> 01:00:05,590
or how happy should I be
with certain outcomes,

1350
01:00:05,590 --> 01:00:07,090
because people need
some reference,

1351
01:00:07,090 --> 01:00:08,470
and often then the
reference are either

1352
01:00:08,470 --> 01:00:10,210
some expectations,
or their neighbors,

1353
01:00:10,210 --> 01:00:15,500
or just comparing between
different options.

1354
01:00:15,500 --> 01:00:18,470
So that's the example
I mentioned previously,

1355
01:00:18,470 --> 01:00:20,050
which is it's much
easier to compare

1356
01:00:20,050 --> 01:00:21,980
your income or any
outcomes-- your grades,

1357
01:00:21,980 --> 01:00:24,580
et cetera-- compared to
what your friend has.

1358
01:00:24,580 --> 01:00:28,780
It's much harder to say how
much an extra $1,000 or having

1359
01:00:28,780 --> 01:00:31,870
$50,000 per year,
is that good or bad?

1360
01:00:31,870 --> 01:00:36,472
It depends a lot on what
you compare it against.

1361
01:00:36,472 --> 01:00:37,930
So what people tend
to do, and this

1362
01:00:37,930 --> 01:00:40,990
is exactly what Kahneman and
Tversky were postulating,

1363
01:00:40,990 --> 01:00:43,270
is that people
compare the outcomes

1364
01:00:43,270 --> 01:00:46,452
relative to reference points.

1365
01:00:46,452 --> 01:00:48,160
And again, we're going
to write this down

1366
01:00:48,160 --> 01:00:51,055
in more detail at the
beginning of next class,

1367
01:00:51,055 --> 01:00:53,980
but what Kahneman and
Tversky were postulating

1368
01:00:53,980 --> 01:00:55,960
were essentially two things.

1369
01:00:55,960 --> 01:00:58,420
They're postulating A,
there's a reference level

1370
01:00:58,420 --> 01:01:00,385
of consumption or any outcomes.

1371
01:01:00,385 --> 01:01:03,280
We can talk a little bit about
what actually is this reference

1372
01:01:03,280 --> 01:01:04,600
level, where it's coming from.

1373
01:01:04,600 --> 01:01:08,320
For now, we just assume there's
a reference level that people

1374
01:01:08,320 --> 01:01:10,430
compare the outcomes against.

1375
01:01:10,430 --> 01:01:14,440
And then against that,
outcomes are compared.

1376
01:01:14,440 --> 01:01:17,770
And in particular,
the function is

1377
01:01:17,770 --> 01:01:20,890
like steeper on the left
compared to the right, right?

1378
01:01:20,890 --> 01:01:24,730
Essentially losses
loom larger than gains.

1379
01:01:24,730 --> 01:01:28,990
So like going down by 10
units on the left or 1 unit

1380
01:01:28,990 --> 01:01:32,650
on the left is more
painful than going up

1381
01:01:32,650 --> 01:01:35,400
on the right relative to
those reference points.

1382
01:01:35,400 --> 01:01:36,580
OK?

1383
01:01:36,580 --> 01:01:39,460
And so that's all said.

1384
01:01:39,460 --> 01:01:43,540
So then what experimental
evidence do we in fact

1385
01:01:43,540 --> 01:01:45,480
have for that?

1386
01:01:45,480 --> 01:01:49,612
So one example-- so we have some
experimental evidence of this.

1387
01:01:49,612 --> 01:01:51,070
And I'm going to
show you next time

1388
01:01:51,070 --> 01:01:54,220
a bunch of different
examples of real choices,

1389
01:01:54,220 --> 01:01:56,500
starting from golfing,
to selling houses,

1390
01:01:56,500 --> 01:01:58,000
to lots of other outcomes.

1391
01:01:58,000 --> 01:02:00,443
But sort of the experimental
evidence, the earlier

1392
01:02:00,443 --> 01:02:02,110
experimental evidence
that people showed

1393
01:02:02,110 --> 01:02:04,510
were preferences
over risky gambles.

1394
01:02:04,510 --> 01:02:07,430
I showed you some of those
already, and then in particular

1395
01:02:07,430 --> 01:02:10,297
unwillingness to trade
different options compared

1396
01:02:10,297 --> 01:02:12,130
to like an alternative
option that I showed.

1397
01:02:12,130 --> 01:02:15,178
That's what people refer
to as the endowment effect.

1398
01:02:15,178 --> 01:02:16,720
So let me show you
first the gambles.

1399
01:02:16,720 --> 01:02:17,990
We already had
that in some sense.

1400
01:02:17,990 --> 01:02:19,448
These are like
essentially gambles.

1401
01:02:19,448 --> 01:02:21,640
People seem really
risk averse when

1402
01:02:21,640 --> 01:02:24,070
they are offered these gambles.

1403
01:02:24,070 --> 01:02:26,890
Kahneman and Tversky would
say people are essentially

1404
01:02:26,890 --> 01:02:27,490
loss averse.

1405
01:02:27,490 --> 01:02:30,760
People really dislike
the loss of $10

1406
01:02:30,760 --> 01:02:32,170
relative to the gain of $11.

1407
01:02:32,170 --> 01:02:33,580
We can explain that behavior.

1408
01:02:33,580 --> 01:02:36,247
And that's a very robust finding
that people decline these kinds

1409
01:02:36,247 --> 01:02:37,130
of types of gambles.

1410
01:02:37,130 --> 01:02:39,310
Kahneman and Tversky
would say that's evidence

1411
01:02:39,310 --> 01:02:40,557
of loss aversion.

1412
01:02:40,557 --> 01:02:42,640
Now, you might say these
are really small gambles.

1413
01:02:42,640 --> 01:02:43,990
Do we really care about them?

1414
01:02:43,990 --> 01:02:46,803
Well, once you do
this with $500, $550

1415
01:02:46,803 --> 01:02:48,970
or whatever, large amounts,
people also do the same.

1416
01:02:48,970 --> 01:02:50,890
It's hard to do this with
like real-world money

1417
01:02:50,890 --> 01:02:52,240
because it's quite
a bit of money.

1418
01:02:52,240 --> 01:02:53,860
It turns out there's
actually some studies who

1419
01:02:53,860 --> 01:02:54,527
do that as well.

1420
01:02:54,527 --> 01:02:57,670
They did this for real with MBA
students, financial analysts,

1421
01:02:57,670 --> 01:02:58,810
and rich investors.

1422
01:02:58,810 --> 01:03:02,992
And even those people like
tend to then turn down

1423
01:03:02,992 --> 01:03:03,950
those kinds of gambles.

1424
01:03:03,950 --> 01:03:05,410
So there seems to
be quite evidence

1425
01:03:05,410 --> 01:03:08,928
of lots of loss aversion.

1426
01:03:08,928 --> 01:03:10,720
Given the choices
between gains and losses,

1427
01:03:10,720 --> 01:03:13,455
people seem to really
dislike losses.

1428
01:03:13,455 --> 01:03:14,830
Now, what's the
endowment effect?

1429
01:03:14,830 --> 01:03:20,410
That's sort of the perhaps most
famous evidence in this domain.

1430
01:03:20,410 --> 01:03:24,250
It essentially is people, when
endowed with a certain item,

1431
01:03:24,250 --> 01:03:26,380
they really do
not like to trade.

1432
01:03:26,380 --> 01:03:30,235
And so the way you
would do an experiment--

1433
01:03:30,235 --> 01:03:32,620
you would give people
an item, and then

1434
01:03:32,620 --> 01:03:37,420
you ask them essentially-- and
a randomly selected fraction

1435
01:03:37,420 --> 01:03:39,910
of people are given an item.

1436
01:03:39,910 --> 01:03:41,980
And then either people
are offered the choice--

1437
01:03:41,980 --> 01:03:43,480
you would you either
keep your item,

1438
01:03:43,480 --> 01:03:45,938
or would you get another item
that's sort of the same value

1439
01:03:45,938 --> 01:03:46,960
overall?

1440
01:03:46,960 --> 01:03:50,770
Or, what people do is like
they have some experiments--

1441
01:03:50,770 --> 01:03:52,750
there's many of
these experiments--

1442
01:03:52,750 --> 01:03:56,470
you give some people one
item and some other people

1443
01:03:56,470 --> 01:03:57,970
another item, and
then see would you

1444
01:03:57,970 --> 01:03:59,380
like to trade with each other.

1445
01:03:59,380 --> 01:04:01,040
And what you see is
essentially people

1446
01:04:01,040 --> 01:04:03,560
are extremely
reluctant to trade.

1447
01:04:03,560 --> 01:04:06,190
There's many of these
kinds of examples.

1448
01:04:06,190 --> 01:04:12,170
Like, for example, so
one example-- sorry,

1449
01:04:12,170 --> 01:04:13,800
I skipped that.

1450
01:04:13,800 --> 01:04:16,037
One example is like if you
just give people an item--

1451
01:04:16,037 --> 01:04:18,120
usually it's like a mug--
if you give people a mug

1452
01:04:18,120 --> 01:04:21,030
and asked like, how much
do I have to pay you

1453
01:04:21,030 --> 01:04:24,270
to give me to sell me this mug?

1454
01:04:24,270 --> 01:04:25,510
People say large amounts.

1455
01:04:25,510 --> 01:04:27,110
They would say like $5.

1456
01:04:27,110 --> 01:04:28,860
If, instead, you asked
them, here's a mug.

1457
01:04:28,860 --> 01:04:30,360
Would you like to
purchase this mug?

1458
01:04:30,360 --> 01:04:33,432
And sort of controlling for how
much money they have and so on.

1459
01:04:33,432 --> 01:04:35,640
If you ask them like, would
you like to buy this mug,

1460
01:04:35,640 --> 01:04:37,380
people would say
like $2 or something.

1461
01:04:37,380 --> 01:04:40,110
It's the same mug, and
essentially their choice

1462
01:04:40,110 --> 01:04:43,410
depends on whether you give
them the mug versus whether you

1463
01:04:43,410 --> 01:04:45,910
endow them with the mug, which
is where the endowment effect

1464
01:04:45,910 --> 01:04:48,420
name comes from, or whether
you just sort of like

1465
01:04:48,420 --> 01:04:51,040
ask their willingness to
pay when they don't have it.

1466
01:04:51,040 --> 01:04:52,590
There's tons of
different experiments

1467
01:04:52,590 --> 01:04:54,630
that do exactly that.

1468
01:04:54,630 --> 01:04:56,340
Now, a different
version of that is like

1469
01:04:56,340 --> 01:04:59,940
so this one is about like
buying a mug, selling a mug,

1470
01:04:59,940 --> 01:05:01,660
buying and selling prices.

1471
01:05:01,660 --> 01:05:03,600
Different versions
of that is to say

1472
01:05:03,600 --> 01:05:07,823
you have a population of
students or different people,

1473
01:05:07,823 --> 01:05:09,240
and what you do
is essentially you

1474
01:05:09,240 --> 01:05:12,230
find two items that on average
have the same valuation.

1475
01:05:12,230 --> 01:05:13,870
Knetsch was doing that.

1476
01:05:13,870 --> 01:05:16,470
And so he had like
mugs and pens,

1477
01:05:16,470 --> 01:05:19,170
and he sort of calibrated such
that on average, when you just

1478
01:05:19,170 --> 01:05:21,828
asked, people's willingness
to pay for these mugs and pens

1479
01:05:21,828 --> 01:05:23,370
are like on average
roughly the same.

1480
01:05:23,370 --> 01:05:26,610
Of course, there's variation,
but on average it's the same.

1481
01:05:26,610 --> 01:05:28,620
And then you offered
half the students mugs

1482
01:05:28,620 --> 01:05:31,740
and half the students pens
and then offered in exchange.

1483
01:05:31,740 --> 01:05:35,915
He also offered in exchange
for an addition of $0.05

1484
01:05:35,915 --> 01:05:37,290
and saying like,
OK, maybe you're

1485
01:05:37,290 --> 01:05:38,332
just exactly indifferent.

1486
01:05:38,332 --> 01:05:40,898
So I'm giving you $0.05
in case you exchange.

1487
01:05:40,898 --> 01:05:43,440
And it turns out that the mug
people like to keep their mugs,

1488
01:05:43,440 --> 01:05:47,190
and the pen people like to keep
the pens, and 90% of people

1489
01:05:47,190 --> 01:05:47,910
do that.

1490
01:05:47,910 --> 01:05:50,670
That's one of the
most robust findings

1491
01:05:50,670 --> 01:05:54,650
in experimental economics.

1492
01:05:54,650 --> 01:05:56,660
There are some questions
on expectations.

1493
01:05:56,660 --> 01:05:59,130
Do people expect to keep the
mugs and so on and so forth?

1494
01:05:59,130 --> 01:06:00,630
There's some
complications for that,

1495
01:06:00,630 --> 01:06:02,570
but the basic result
is very robust

1496
01:06:02,570 --> 01:06:05,510
and has been shown in
many different settings.

1497
01:06:05,510 --> 01:06:07,400
Any questions on this?

1498
01:06:14,200 --> 01:06:14,730
OK.

1499
01:06:14,730 --> 01:06:15,870
So what's going on here?

1500
01:06:15,870 --> 01:06:18,560
Well, essentially,
the reference point

1501
01:06:18,560 --> 01:06:22,330
seems to be essentially
affected by ownership.

1502
01:06:22,330 --> 01:06:24,360
So if you own a mug,
your reference point

1503
01:06:24,360 --> 01:06:25,800
is owning a mug.

1504
01:06:25,800 --> 01:06:29,040
So now if I asked you, would
you like to sell me this mug?

1505
01:06:29,040 --> 01:06:30,750
Well, now you have
a loss of a mug.

1506
01:06:30,750 --> 01:06:32,860
So essentially, you're in
the lost domain of mugs.

1507
01:06:32,860 --> 01:06:36,540
So I have to pay you more money
to compensate you for that.

1508
01:06:36,540 --> 01:06:39,330
If, in contrast, you do not own
the mug, so like you have zero

1509
01:06:39,330 --> 01:06:41,910
mugs, your reference point is
zero mugs, and I'm asking you,

1510
01:06:41,910 --> 01:06:46,050
would you like to receive a
mug or like buy a mug from me,

1511
01:06:46,050 --> 01:06:48,240
then you're essentially
on the gain domain.

1512
01:06:48,240 --> 01:06:51,630
And sort of what I showed
you here previously--

1513
01:06:51,630 --> 01:06:52,690
one second.

1514
01:06:52,690 --> 01:06:55,080
So when I'm asking you
to purchase your mug,

1515
01:06:55,080 --> 01:06:57,357
essentially you're on the
left side of this figure.

1516
01:06:57,357 --> 01:06:58,440
You're in the loss domain.

1517
01:06:58,440 --> 01:07:00,600
Your marginal utility
of mugs is very high.

1518
01:07:00,600 --> 01:07:04,530
I have to pay you a lot of money
to receive the mug from you.

1519
01:07:04,530 --> 01:07:06,750
In contrast, if you have
zero mugs, I'm asking you,

1520
01:07:06,750 --> 01:07:08,125
would you like to
purchase a mug,

1521
01:07:08,125 --> 01:07:10,125
you're on the right side
of the suffering point.

1522
01:07:10,125 --> 01:07:12,130
So essentially now you're
in the gain domain.

1523
01:07:12,130 --> 01:07:15,480
Now you're not willing
to pay a lot of money

1524
01:07:15,480 --> 01:07:17,940
because you're gaining a mug.

1525
01:07:17,940 --> 01:07:20,670
On top of that, you also on
the gain loss domain of money,

1526
01:07:20,670 --> 01:07:23,950
but like I'm setting
that sort of aside.

1527
01:07:23,950 --> 01:07:25,450
Does that make any sense?

1528
01:07:27,750 --> 01:07:28,250
OK.

1529
01:07:32,740 --> 01:07:35,125
And so people hate losses
more than they like the gain.

1530
01:07:35,125 --> 01:07:36,250
So they stick with the mug.

1531
01:07:36,250 --> 01:07:40,610
And similarly, that's the
same thing for the pen owners.

1532
01:07:40,610 --> 01:07:42,140
There's lots of
different examples

1533
01:07:42,140 --> 01:07:44,300
of those kinds of behaviors.

1534
01:07:44,300 --> 01:07:46,610
For example, law
school students were

1535
01:07:46,610 --> 01:07:50,000
asked to assess compensation
for pain and suffering damages

1536
01:07:50,000 --> 01:07:51,710
in one study.

1537
01:07:51,710 --> 01:07:53,600
This is expected to last--

1538
01:07:53,600 --> 01:07:56,210
or in this example-- is
expected to last three years

1539
01:07:56,210 --> 01:07:58,040
and be quite unpleasant.

1540
01:07:58,040 --> 01:08:01,740
There's no impact on
earnings capacity.

1541
01:08:01,740 --> 01:08:04,390
For example, it would be extreme
stiffness in the upper back

1542
01:08:04,390 --> 01:08:04,890
and neck.

1543
01:08:04,890 --> 01:08:07,530
I think that would probably
affect your earning capacity,

1544
01:08:07,530 --> 01:08:12,040
but anyway let's assume
that's not the case.

1545
01:08:12,040 --> 01:08:13,740
So then some students
are led to imagine

1546
01:08:13,740 --> 01:08:15,060
they were being injured.

1547
01:08:15,060 --> 01:08:19,210
How much would you be
willing to pay to get better?

1548
01:08:19,210 --> 01:08:19,710
OK.

1549
01:08:19,710 --> 01:08:21,750
So your reference point
is like you're injured.

1550
01:08:21,750 --> 01:08:25,470
How much are you willing
to pay to get better?

1551
01:08:25,470 --> 01:08:30,270
And people said
151,448 on average.

1552
01:08:30,270 --> 01:08:31,770
Now, another group
of students was--

1553
01:08:31,770 --> 01:08:35,250
this is randomized-- was led
to imagine being uninjured.

1554
01:08:35,250 --> 01:08:39,600
Like, how much would I need to
pay you to accept the injury?

1555
01:08:39,600 --> 01:08:40,920
You're not injured.

1556
01:08:40,920 --> 01:08:43,535
Now I'm asking you, how
much are you willing to pay,

1557
01:08:43,535 --> 01:08:44,910
or how much do I
have to pay you,

1558
01:08:44,910 --> 01:08:48,390
how much do I have to compensate
you to accept this injury?

1559
01:08:48,390 --> 01:08:52,529
Now, you would think this
is like the same thing,

1560
01:08:52,529 --> 01:08:54,870
but what's your price
of not being injured?

1561
01:08:54,870 --> 01:08:57,600
The price of health should be
independent on which way I'm

1562
01:08:57,600 --> 01:08:58,380
asking you.

1563
01:08:58,380 --> 01:09:01,859
Turns out, when people
have their good health,

1564
01:09:01,859 --> 01:09:05,319
they're willing to pay
a lot more to keep it

1565
01:09:05,319 --> 01:09:08,490
or to not lose it compared to
like when they have lost it

1566
01:09:08,490 --> 01:09:11,729
and they want to regain it.

1567
01:09:11,729 --> 01:09:13,890
There's another
quite nice example.

1568
01:09:13,890 --> 01:09:15,930
I'm showing you here.

1569
01:09:15,930 --> 01:09:16,930
One second.

1570
01:09:16,930 --> 01:09:20,189
So in case you didn't
see, that was Dan Ariely,

1571
01:09:20,189 --> 01:09:22,260
who has several nice books.

1572
01:09:22,260 --> 01:09:24,160
One of them is
predictably irrational,

1573
01:09:24,160 --> 01:09:27,580
which is a quite nice read.

1574
01:09:27,580 --> 01:09:32,736
So there's lots of these kinds
of examples of loss aversion

1575
01:09:32,736 --> 01:09:34,319
or what's called the
endowment effect.

1576
01:09:34,319 --> 01:09:36,808
When people have
stuff essentially,

1577
01:09:36,808 --> 01:09:38,100
they're not willing to part it.

1578
01:09:38,100 --> 01:09:41,760
When they don't have it,
they're willing to pay less.

1579
01:09:41,760 --> 01:09:44,279
One nice thing that Ariely was
also talking about-- a little

1580
01:09:44,279 --> 01:09:45,840
bit about how
people then explain

1581
01:09:45,840 --> 01:09:46,950
these kinds of behaviors.

1582
01:09:46,950 --> 01:09:49,260
We're going to get
back to that in I

1583
01:09:49,260 --> 01:09:50,837
think lecture 20
or something, which

1584
01:09:50,837 --> 01:09:52,920
is about kind of like I
think what's going on here

1585
01:09:52,920 --> 01:09:53,670
is very obvious.

1586
01:09:53,670 --> 01:09:55,200
Potentially, people
are loss-averse

1587
01:09:55,200 --> 01:09:57,150
and they're randomized
into like gains or losses,

1588
01:09:57,150 --> 01:09:59,317
and that's kind of what's
explaining their behavior.

1589
01:09:59,317 --> 01:10:01,058
But people don't
necessarily understand

1590
01:10:01,058 --> 01:10:03,600
that they're being randomized
in one of the other conditions,

1591
01:10:03,600 --> 01:10:05,700
and they then sort of try
to explain their behavior

1592
01:10:05,700 --> 01:10:08,158
in various ways and saying, I
want to tell my grandchildren

1593
01:10:08,158 --> 01:10:08,783
about this.

1594
01:10:08,783 --> 01:10:10,200
But in some sense,
they're sort of

1595
01:10:10,200 --> 01:10:12,150
like rationalizing
their preferences

1596
01:10:12,150 --> 01:10:14,688
and they don't necessarily sort
of understand these stories,

1597
01:10:14,688 --> 01:10:16,230
and they don't
necessarily understand

1598
01:10:16,230 --> 01:10:19,320
where these preferences are
coming from even though we kind

1599
01:10:19,320 --> 01:10:21,450
of know, because the
person has been manipulated

1600
01:10:21,450 --> 01:10:23,040
into those kinds of choices.

1601
01:10:23,040 --> 01:10:24,430
We'll get back to that.

1602
01:10:24,430 --> 01:10:28,440
Now, the third part of Kahneman
and Tversky's prospect theory

1603
01:10:28,440 --> 01:10:29,940
of their paper is
essentially what's

1604
01:10:29,940 --> 01:10:31,615
called diminishing sensitivity.

1605
01:10:31,615 --> 01:10:33,990
I'm going to get back to this
and sort of write this down

1606
01:10:33,990 --> 01:10:36,720
in more detail on Monday,
but essentially it's

1607
01:10:36,720 --> 01:10:40,290
like people are risk-averse
in the gain domain,

1608
01:10:40,290 --> 01:10:42,600
but risk-loving and
the loss region.

1609
01:10:42,600 --> 01:10:43,720
What does it look like?

1610
01:10:43,720 --> 01:10:45,930
Essentially the
utility function is

1611
01:10:45,930 --> 01:10:48,390
a little bit different from
what I showed you before.

1612
01:10:48,390 --> 01:10:50,370
It's not only like
more steep on the left

1613
01:10:50,370 --> 01:10:52,800
relative to the reference
point compared to the right,

1614
01:10:52,800 --> 01:10:57,090
but it's also concave on the
right and convex on the left.

1615
01:10:57,090 --> 01:10:59,490
And what that gets you and
what that buys you essentially

1616
01:10:59,490 --> 01:11:05,070
is that people are risk
averse in the gain domain,

1617
01:11:05,070 --> 01:11:08,430
but people are risk-loving
in the loss domain,

1618
01:11:08,430 --> 01:11:10,890
and that sort of is needed
to be able to explain some

1619
01:11:10,890 --> 01:11:12,930
of the behaviors
that I showed you

1620
01:11:12,930 --> 01:11:16,320
early on in the Kahneman
and Tversky evidence.

1621
01:11:16,320 --> 01:11:19,440
So again, I'll tell you
about this in more detail

1622
01:11:19,440 --> 01:11:22,043
and sort of write this
down more precisely.

1623
01:11:22,043 --> 01:11:23,460
And then we're
going to talk about

1624
01:11:23,460 --> 01:11:25,230
like many different
applications.

1625
01:11:25,230 --> 01:11:27,180
And particularly, there's
the endowment effect

1626
01:11:27,180 --> 01:11:29,760
which is kind of what I talked
to you already about before.

1627
01:11:29,760 --> 01:11:31,808
There's labor
supply, employment,

1628
01:11:31,808 --> 01:11:33,600
and effort depending
on what people expect,

1629
01:11:33,600 --> 01:11:34,590
how much they earn.

1630
01:11:34,590 --> 01:11:40,110
They make different choices
about how many hours

1631
01:11:40,110 --> 01:11:42,120
they actually work depending
on the expectation,

1632
01:11:42,120 --> 01:11:44,460
whether it's above or
below a reference point.

1633
01:11:44,460 --> 01:11:46,230
People are reluctant
to sell their houses

1634
01:11:46,230 --> 01:11:47,850
if they're at a loss
compared to again

1635
01:11:47,850 --> 01:11:50,040
even for very similar houses.

1636
01:11:50,040 --> 01:11:52,260
And marathon-running
people are trying

1637
01:11:52,260 --> 01:11:54,480
to reach essentially
certain targets.

1638
01:11:54,480 --> 01:11:56,580
People like to be below
four hours or three

1639
01:11:56,580 --> 01:11:58,470
hours and the like.

1640
01:11:58,470 --> 01:12:00,580
There's the disposition
effect that we mentioned,

1641
01:12:00,580 --> 01:12:03,060
which essentially is
people like to sell winners

1642
01:12:03,060 --> 01:12:05,280
compared to losers.

1643
01:12:05,280 --> 01:12:08,230
And the insurance choice that
I already showed you before--

1644
01:12:08,230 --> 01:12:13,800
there's some evidence on
violence in the household.

1645
01:12:13,800 --> 01:12:17,940
The particular example is about
football or games essentially--

1646
01:12:17,940 --> 01:12:20,310
I think football games
where essentially depending

1647
01:12:20,310 --> 01:12:24,210
on whether people expected
a loss or win of their team

1648
01:12:24,210 --> 01:12:29,490
and when then a loss or a win
of the team actually happens,

1649
01:12:29,490 --> 01:12:32,220
there's more violence when there
are unexpected losses compared

1650
01:12:32,220 --> 01:12:35,380
to expected losses
which is, again,

1651
01:12:35,380 --> 01:12:38,590
consistent with loss aversion.

1652
01:12:38,590 --> 01:12:39,090
What's next?

1653
01:12:39,090 --> 01:12:41,215
So next we're going to talk
about many applications

1654
01:12:41,215 --> 01:12:44,310
of reference
dependence on Monday.

1655
01:12:44,310 --> 01:12:47,400
I'd like you to sort of read
the Kahneman and Tversky

1656
01:12:47,400 --> 01:12:49,830
paper, at least the
first pages to get

1657
01:12:49,830 --> 01:12:51,083
some sense of their work.

1658
01:12:51,083 --> 01:12:52,500
And then on
Wednesday, we're going

1659
01:12:52,500 --> 01:12:54,282
to start talking about
social preferences.

1660
01:12:54,282 --> 01:12:56,490
In particular, we can have
some experiments in class.

1661
01:12:56,490 --> 01:13:00,210
There's going to be
no readings for that.

1662
01:13:00,210 --> 01:13:02,030
There's the opportunity
to make money

1663
01:13:02,030 --> 01:13:04,720
in class, at least some of it.