1
00:00:02,290 --> 00:00:05,945
♪ [music] ♪

2
00:00:14,295 --> 00:00:15,400
- [Professor Tyler Cowen] 
In the previous video,

3
00:00:15,400 --> 00:00:18,280
we introduced the ideas 
of asymmetric information,

4
00:00:18,280 --> 00:00:21,280
and adverse selection 
and we applied those ideas

5
00:00:21,280 --> 00:00:22,679
to the used car market.

6
00:00:22,880 --> 00:00:25,040
Let's take those same 
basic concepts,

7
00:00:25,040 --> 00:00:28,630
and build a basic model 
of health insurance.

8
00:00:28,630 --> 00:00:31,420
Suppose that potential 
health insurance consumers

9
00:00:31,420 --> 00:00:33,660
come in a range of states of health.

10
00:00:33,660 --> 00:00:35,680
For instance, 
the least healthy people

11
00:00:35,680 --> 00:00:38,920
might cost about $30,000 a year.

12
00:00:38,920 --> 00:00:40,610
That's these folks here.

13
00:00:40,610 --> 00:00:43,980
The most healthy might cost 
nothing in healthcare.

14
00:00:43,980 --> 00:00:46,120
That's these folks over here.

15
00:00:46,120 --> 00:00:48,540
Now consumers know this information,

16
00:00:48,540 --> 00:00:51,180
but by assumption, insurers don't.

17
00:00:51,180 --> 00:00:53,160
From the insurer point of view,

18
00:00:53,160 --> 00:00:56,010
everyone is of the same 
average health.

19
00:00:56,010 --> 00:00:58,964
Here again, we have 
asymmetric information.

20
00:00:59,110 --> 00:01:01,460
That is consumers of healthcare

21
00:01:01,460 --> 00:01:02,780
have more information about

22
00:01:02,780 --> 00:01:05,100
their health status 
than insurers do.

23
00:01:05,480 --> 00:01:08,880
In this scenario, insurers have to 
price the coverage

24
00:01:08,880 --> 00:01:12,000
based on the average cost 
among all consumers,

25
00:01:12,000 --> 00:01:13,975
namely, $15,000.

26
00:01:14,530 --> 00:01:18,960
But if the insurance costs $15,000,
then a portion of the market,

27
00:01:18,960 --> 00:01:20,830
the relatively healthy people,

28
00:01:20,830 --> 00:01:23,460
they will choose not 
to buy insurance as

29
00:01:23,460 --> 00:01:26,130
the cost of that insurance 
is greater to them

30
00:01:26,130 --> 00:01:27,543
than the expected benefit.

31
00:01:27,880 --> 00:01:31,300
So only part of this market 
will buy insurance.

32
00:01:31,300 --> 00:01:33,950
The average cost of those 
who actually will buy

33
00:01:33,950 --> 00:01:39,420
is then not $15,000 but $22,500.

34
00:01:39,420 --> 00:01:41,830
In that case, 
the insurance company,

35
00:01:41,830 --> 00:01:45,580
if it tries to price at $15,000, 
loses money.

36
00:01:45,580 --> 00:01:51,400
If the insurance company instead 
raises the price to $22,500,

37
00:01:51,400 --> 00:01:55,217
well, the same dynamic
is actually going to kick in again.

38
00:01:55,217 --> 00:01:56,677
That is relatively healthy people

39
00:01:56,677 --> 00:01:58,900
won't find it worth 
paying that price.

40
00:01:58,900 --> 00:02:01,290
The sicker people still will buy,

41
00:02:01,290 --> 00:02:03,030
and that will raise 
the expected costs

42
00:02:03,030 --> 00:02:06,423
to the insurer, 
and thus the price even further.

43
00:02:06,800 --> 00:02:10,400
This dynamic continues 
until the individual insurance firm

44
00:02:10,400 --> 00:02:13,200
finds there is no price 
at which it can attract

45
00:02:13,200 --> 00:02:16,020
a set of customers with 
healthcare costs

46
00:02:16,020 --> 00:02:18,312
lower than the price of insurance.

47
00:02:18,540 --> 00:02:23,140
This is the same death spiral 
we saw before with used cars

48
00:02:23,140 --> 00:02:24,833
and it leads to a market failure.

49
00:02:26,080 --> 00:02:27,540
As we saw in the used car market,

50
00:02:27,540 --> 00:02:30,540
there are several reasons 
why reality may differ

51
00:02:30,540 --> 00:02:32,350
from the simple model.

52
00:02:32,350 --> 00:02:35,500
First, the model we laid out would 
predict that the healthy people,

53
00:02:35,500 --> 00:02:38,050
those who exercise, 
eat their veggies,

54
00:02:38,050 --> 00:02:41,370
and buckle their seatbelts would 
not buy insurance,

55
00:02:41,370 --> 00:02:45,370
while the model is predicting that 
the smokers, the mountain climbers,

56
00:02:45,370 --> 00:02:48,550
and the motorcycle riders would
buy insurance.

57
00:02:48,550 --> 00:02:51,240
Is this true? Mostly no.

58
00:02:51,240 --> 00:02:52,960
The people who buy health insurance

59
00:02:52,960 --> 00:02:56,490
actually turn out to be 
the healthier people as well.

60
00:02:56,490 --> 00:02:57,830
Why is that?

61
00:02:57,830 --> 00:03:01,400
Well, those who try to avoid risk 
by eating well

62
00:03:01,400 --> 00:03:04,870
also try to avoid risk 
by buying health insurance.

63
00:03:04,870 --> 00:03:06,220
Our initial assumption that

64
00:03:06,220 --> 00:03:08,480
everyone calculates 
costs and benefits

65
00:03:08,480 --> 00:03:10,720
in exactly the same way 
is too simple.

66
00:03:10,720 --> 00:03:12,680
Once you account for the fact that

67
00:03:12,680 --> 00:03:15,880
people have differential 
tolerances for risk,

68
00:03:15,880 --> 00:03:18,390
you can end up having 
the healthier people be


69
00:03:18,390 --> 00:03:21,100
those who choose to buy 
the health insurance.

70
00:03:21,100 --> 00:03:24,050
This is called 
“propitious selection”


71
00:03:24,050 --> 00:03:27,280
where the people who buy 
the health insurance are healthier,

72
00:03:27,280 --> 00:03:29,320
not sicker than average.

73
00:03:29,320 --> 00:03:32,512
This can keep costs low, 
and prevent the death spiral.

74
00:03:33,540 --> 00:03:36,630
Another possible response 
to the adverse selection problem

75
00:03:36,630 --> 00:03:39,440
in health insurance 
might seem familiar.

76
00:03:39,440 --> 00:03:42,440
If you recall, 
we saw that services such as

77
00:03:42,440 --> 00:03:45,170
CARFAX and Certified Inspections

78
00:03:45,170 --> 00:03:47,830
can alleviate 
the asymmetric information problem

79
00:03:47,830 --> 00:03:49,750
when buying a used car.

80
00:03:49,750 --> 00:03:52,640
These services allow 
the buyer of the car

81
00:03:52,640 --> 00:03:54,000
to have similar information

82
00:03:54,000 --> 00:03:57,000
to that possessed
by the seller of the car.

83
00:03:57,620 --> 00:04:00,080
The result of this information 
is that better cars

84
00:04:00,080 --> 00:04:03,530
can sell for more, 
and lemons can sell for less.

85
00:04:03,530 --> 00:04:06,710
Is there an analogous approach 
for people in health insurance?

86
00:04:06,710 --> 00:04:07,900
Well, yes.

87
00:04:07,900 --> 00:04:09,750
The health of people 
can be inspected

88
00:04:09,750 --> 00:04:11,910
just as cars are inspected.

89
00:04:11,910 --> 00:04:14,989
So while consumers initially 
may have more information

90
00:04:14,989 --> 00:04:18,450
about their health than what 
the insurance companies have,

91
00:04:18,450 --> 00:04:20,650
a checkup will allow 
the insurance firms

92
00:04:20,650 --> 00:04:23,530
to get a better idea of 
the consumer's expected

93
00:04:23,530 --> 00:04:24,960
healthcare costs.

94
00:04:24,960 --> 00:04:27,120
And that allows 
the insurance companies

95
00:04:27,120 --> 00:04:31,630
to charge healthy consumers less 
and sicker consumers more.

96
00:04:32,100 --> 00:04:33,160
In the used car market,

97
00:04:33,160 --> 00:04:35,180
that seemed like 
a pretty good solution.

98
00:04:35,180 --> 00:04:37,600
After all, better cars 
should sell for more,

99
00:04:37,600 --> 00:04:40,180
and lemons should sell for less.

100
00:04:40,180 --> 00:04:41,510
In the health insurance market,

101
00:04:41,510 --> 00:04:43,380
that solution might work,

102
00:04:43,380 --> 00:04:45,630
but some people feel it is 
doubly unfair.

103
00:04:45,630 --> 00:04:47,860
Not only are the sick sick,

104
00:04:47,860 --> 00:04:49,710
but now they also have 
to pay more

105
00:04:49,710 --> 00:04:51,570
for their health insurance.

106
00:04:51,570 --> 00:04:53,120
Another problem 
with inspection is that

107
00:04:53,120 --> 00:04:55,610
it might reveal 
too much information,

108
00:04:55,610 --> 00:04:59,160
thereby rendering health insurance
no longer viable.

109
00:04:59,160 --> 00:05:02,270
For instance, let's say there's 
a very good diagnostic test,

110
00:05:02,270 --> 00:05:05,350
and it determines that 
a patient A has cancer

111
00:05:05,350 --> 00:05:09,540
and then B we know that cancer
will cost $1 million to treat.

112
00:05:09,540 --> 00:05:11,550
Well, to insure against that cancer,

113
00:05:11,550 --> 00:05:14,750
the price of the policy 
has to be about $1 million,

114
00:05:14,750 --> 00:05:16,520
but that's no longer insurance.

115
00:05:16,520 --> 00:05:19,110
That's just presenting the patient 
with the bill.

116
00:05:19,110 --> 00:05:23,070
Insurance is protecting against 
unexpected states of affairs,

117
00:05:23,070 --> 00:05:25,150
and it's a kind of risk pooling,

118
00:05:25,150 --> 00:05:27,740
a kind of protecting yourself
against the high bill.

119
00:05:27,740 --> 00:05:30,048
But if you're getting the high bill 
no matter what when you're sick,

120
00:05:30,048 --> 00:05:34,060
well, then we've lost 
those benefits of insurance.

121
00:05:34,660 --> 00:05:37,260
Another solution to 
the adverse selection problem

122
00:05:37,260 --> 00:05:39,630
when used extensively 
in the United States

123
00:05:39,630 --> 00:05:42,110
is group health insurance 
through employers.

124
00:05:42,110 --> 00:05:46,150
Most people in America 
don't purchase insurance directly.

125
00:05:46,150 --> 00:05:48,290
Instead, their employer 
purchases it for them

126
00:05:48,290 --> 00:05:49,685
as part of a group plan.

127
00:05:50,490 --> 00:05:53,650
The benefit of the system is that 
the insurance company

128
00:05:53,650 --> 00:05:57,080
doesn't have to worry about 
adverse selection so much

129
00:05:57,080 --> 00:05:59,875
The employer doesn't know much
more about its employees' health

130
00:05:59,875 --> 00:06:02,232
than does the insurance firm.

131
00:06:02,232 --> 00:06:04,209
Furthermore, the employer is
going to be buying

132
00:06:04,209 --> 00:06:08,052
health insurance for the employees
regardless of their health.

133
00:06:08,052 --> 00:06:11,420
So for these reasons,
the adverse selection problem is

134
00:06:11,420 --> 00:06:14,310
much weaker with 
group health insurance.

135
00:06:14,310 --> 00:06:17,770
Group health insurance, however, 
does cause other problems.

136
00:06:17,770 --> 00:06:20,810
If you lose your job, 
you can lose your health insurance.

137
00:06:20,810 --> 00:06:23,220
And what we do about retirees?

138
00:06:23,860 --> 00:06:26,220
In the United States, 
various laws have made

139
00:06:26,220 --> 00:06:28,520
health insurance more affordable,

140
00:06:28,520 --> 00:06:31,170
and furthermore retirees are insured 
by the government

141
00:06:31,170 --> 00:06:32,690
under Medicare.

142
00:06:32,690 --> 00:06:37,140
So, there are some solutions, 
albeit imperfect ones as usual.

143
00:06:37,580 --> 00:06:40,930
The most recent approach 
to the adverse selection problem

144
00:06:40,930 --> 00:06:43,390
was implemented 
in the Affordable Care Act,

145
00:06:43,390 --> 00:06:45,558
otherwise known as Obamacare.

146
00:06:46,210 --> 00:06:47,830
Under the Affordable Care Act,

147
00:06:47,830 --> 00:06:50,590
everyone is supposed to buy
health insurance.

148
00:06:50,590 --> 00:06:53,550
If you don't, 
you will be fined by law.

149
00:06:53,550 --> 00:06:57,660
The idea here is to force 
all the healthy people into the pool

150
00:06:57,660 --> 00:06:59,300
of those who buy insurance

151
00:06:59,300 --> 00:07:01,820
that will moderate the cost 
of health insurance,

152
00:07:01,820 --> 00:07:04,280
and we will avoid the death spiral.

153
00:07:04,280 --> 00:07:06,120
As you can see, although 


154
00:07:06,120 --> 00:07:08,550
the adverse selection model 
is pretty simple,

155
00:07:08,550 --> 00:07:10,150
it has lots of applications

156
00:07:10,150 --> 00:07:13,230
to some pretty complex 
real-world problems.

157
00:07:13,230 --> 00:07:17,000
Next up we'll tackle moral
hazard. See you then.

158
00:07:17,000 --> 00:07:18,790
♪ [music] ♪

159
00:07:18,790 --> 00:07:19,790
- [Announcer] If you want 
to test yourself,

160
00:07:19,790 --> 00:07:21,945
click “Practice Questions."

161
00:07:22,315 --> 00:07:26,279
Or, if you're ready to move on,
just click “Next Video.”