ABOUT THE SPEAKER
Emily Oster - Assumption-busting economist
Emily Oster, a University of Chicago economist, uses the dismal science to rethink conventional wisdom, from her Harvard doctoral thesis that took on famed economist Amartya Sen to her recent work debunking assumptions on HIV prevalence in Africa.

Why you should listen

Emily Oster, an Assistant Professor of Economics at the University of Chicago, has a history of rethinking conventional wisdom.

Her Harvard doctoral thesis took on famed economist Amartya Sen and his claim that 100 million women were statistically missing from the developing world. He blamed misogynist medical care and outright sex-selective abortion for the gap, but Oster pointed to data indicating that in countries where Hepetitis B infections were higher, more boys were born. Through her unorthodox analysis of medical data, she accounted for 50% of the missing girls. Three years later, she would publish another paper amending her findings, stating that, after further study, the relationship between Hepetitis B and missing women was not apparent. This concession, along with her audacity to challenge economic assumptions and her dozens of other influential papers, has earned her the respect of the global academic community. 

She's also investigated the role of bad weather in the rise in witchcraft trials in Medieval Europe and what drives people to play the Powerball lottery. Her latest target: busting assumptions on HIV in Africa.

And she's an advice columnist too >>

 

More profile about the speaker
Emily Oster | Speaker | TED.com
TED2007

Emily Oster: Flip your thinking on AIDS in Africa

Filmed:
921,618 views

Emily Oster re-examines the stats on AIDS in Africa from an economic perspective and reaches a stunning conclusion: Everything we know about the spread of HIV on the continent is wrong.
- Assumption-busting economist
Emily Oster, a University of Chicago economist, uses the dismal science to rethink conventional wisdom, from her Harvard doctoral thesis that took on famed economist Amartya Sen to her recent work debunking assumptions on HIV prevalence in Africa. Full bio

Double-click the English transcript below to play the video.

00:26
So I want to talk to you today about AIDS in sub-Saharan Africa.
0
1000
3000
00:29
And this is a pretty well-educated audience,
1
4000
2000
00:31
so I imagine you all know something about AIDS.
2
6000
3000
00:34
You probably know that roughly 25 million people in Africa
3
9000
2000
00:36
are infected with the virus, that AIDS is a disease of poverty,
4
11000
4000
00:40
and that if we can bring Africa out of poverty, we would decrease AIDS as well.
5
15000
4000
00:44
If you know something more, you probably know that Uganda, to date,
6
19000
3000
00:47
is the only country in sub-Saharan Africa
7
22000
2000
00:49
that has had success in combating the epidemic.
8
24000
3000
00:52
Using a campaign that encouraged people to abstain, be faithful, and use condoms --
9
27000
4000
00:56
the ABC campaign -- they decreased their prevalence in the 1990s
10
31000
4000
01:00
from about 15 percent to 6 percent over just a few years.
11
35000
4000
01:04
If you follow policy, you probably know that a few years ago
12
39000
3000
01:07
the president pledged 15 billion dollars to fight the epidemic over five years,
13
42000
4000
01:11
and a lot of that money is going to go to programs that try to replicate Uganda
14
46000
3000
01:14
and use behavior change to encourage people and decrease the epidemic.
15
49000
6000
01:20
So today I'm going to talk about some things
16
55000
2000
01:22
that you might not know about the epidemic,
17
57000
2000
01:24
and I'm actually also going to challenge
18
59000
2000
01:26
some of these things that you think that you do know.
19
61000
2000
01:28
To do that I'm going to talk about my research
20
63000
3000
01:31
as an economist on the epidemic.
21
66000
2000
01:33
And I'm not really going to talk much about the economy.
22
68000
2000
01:35
I'm not going to tell you about exports and prices.
23
70000
3000
01:38
But I'm going to use tools and ideas that are familiar to economists
24
73000
4000
01:42
to think about a problem that's more traditionally
25
77000
2000
01:44
part of public health and epidemiology.
26
79000
2000
01:46
And I think in that sense, this fits really nicely with this lateral thinking idea.
27
81000
4000
01:50
Here I'm really using the tools of one academic discipline
28
85000
3000
01:53
to think about problems of another.
29
88000
2000
01:55
So we think, first and foremost, AIDS is a policy issue.
30
90000
3000
01:58
And probably for most people in this room, that's how you think about it.
31
93000
3000
02:01
But this talk is going to be about understanding facts about the epidemic.
32
96000
4000
02:05
It's going to be about thinking about how it evolves, and how people respond to it.
33
100000
3000
02:08
I think it may seem like I'm ignoring the policy stuff,
34
103000
3000
02:11
which is really the most important,
35
106000
2000
02:13
but I'm hoping that at the end of this talk you will conclude
36
108000
2000
02:15
that we actually cannot develop effective policy
37
110000
2000
02:17
unless we really understand how the epidemic works.
38
112000
3000
02:20
And the first thing that I want to talk about,
39
115000
2000
02:22
the first thing I think we need to understand is:
40
117000
2000
02:24
how do people respond to the epidemic?
41
119000
2000
02:26
So AIDS is a sexually transmitted infection, and it kills you.
42
121000
4000
02:30
So this means that in a place with a lot of AIDS,
43
125000
2000
02:32
there's a really significant cost of sex.
44
127000
2000
02:34
If you're an uninfected man living in Botswana, where the HIV rate is 30 percent,
45
129000
4000
02:38
if you have one more partner this year -- a long-term partner, girlfriend, mistress --
46
133000
4000
02:42
your chance of dying in 10 years increases by three percentage points.
47
137000
4000
02:46
That is a huge effect.
48
141000
2000
02:48
And so I think that we really feel like then people should have less sex.
49
143000
3000
02:51
And in fact among gay men in the US
50
146000
2000
02:53
we did see that kind of change in the 1980s.
51
148000
2000
02:55
So if we look in this particularly high-risk sample, they're being asked,
52
150000
4000
02:59
"Did you have more than one unprotected sexual partner in the last two months?"
53
154000
3000
03:02
Over a period from '84 to '88, that share drops from about 85 percent to 55 percent.
54
157000
6000
03:08
It's a huge change in a very short period of time.
55
163000
2000
03:10
We didn't see anything like that in Africa.
56
165000
2000
03:12
So we don't have quite as good data, but you can see here
57
167000
3000
03:15
the share of single men having pre-marital sex,
58
170000
2000
03:17
or married men having extra-marital sex,
59
172000
2000
03:19
and how that changes from the early '90s to late '90s,
60
174000
3000
03:22
and late '90s to early 2000s. The epidemic is getting worse.
61
177000
3000
03:25
People are learning more things about it.
62
180000
2000
03:27
We see almost no change in sexual behavior.
63
182000
2000
03:29
These are just tiny decreases -- two percentage points -- not significant.
64
184000
4000
03:33
This seems puzzling. But I'm going to argue that you shouldn't be surprised by this,
65
188000
4000
03:37
and that to understand this you need to think about health
66
192000
3000
03:40
the way than an economist does -- as an investment.
67
195000
3000
03:43
So if you're a software engineer and you're trying to think about
68
198000
3000
03:46
whether to add some new functionality to your program,
69
201000
3000
03:49
it's important to think about how much it costs.
70
204000
2000
03:51
It's also important to think about what the benefit is.
71
206000
2000
03:53
And one part of that benefit is how much longer
72
208000
2000
03:55
you think this program is going to be active.
73
210000
2000
03:57
If version 10 is coming out next week,
74
212000
2000
03:59
there's no point in adding more functionality into version nine.
75
214000
3000
04:02
But your health decisions are the same.
76
217000
2000
04:04
Every time you have a carrot instead of a cookie,
77
219000
2000
04:06
every time you go to the gym instead of going to the movies,
78
221000
3000
04:09
that's a costly investment in your health.
79
224000
2000
04:11
But how much you want to invest is going to depend
80
226000
2000
04:13
on how much longer you expect to live in the future,
81
228000
2000
04:15
even if you don't make those investments.
82
230000
2000
04:17
AIDS is the same kind of thing. It's costly to avoid AIDS.
83
232000
3000
04:20
People really like to have sex.
84
235000
3000
04:23
But, you know, it has a benefit in terms of future longevity.
85
238000
6000
04:29
But life expectancy in Africa, even without AIDS, is really, really low:
86
244000
4000
04:33
40 or 50 years in a lot of places.
87
248000
3000
04:36
I think it's possible, if we think about that intuition, and think about that fact,
88
251000
4000
04:40
that maybe that explains some of this low behavior change.
89
255000
3000
04:43
But we really need to test that.
90
258000
2000
04:45
And a great way to test that is to look across areas in Africa and see:
91
260000
3000
04:48
do people with more life expectancy change their sexual behavior more?
92
263000
4000
04:52
And the way that I'm going to do that is,
93
267000
2000
04:54
I'm going to look across areas with different levels of malaria.
94
269000
3000
04:57
So malaria is a disease that kills you.
95
272000
3000
05:00
It's a disease that kills a lot of adults in Africa, in addition to a lot of children.
96
275000
3000
05:03
And so people who live in areas with a lot of malaria
97
278000
3000
05:06
are going to have lower life expectancy than people who live in areas with limited malaria.
98
281000
4000
05:10
So one way to test to see whether we can explain
99
285000
2000
05:12
some of this behavior change by differences in life expectancy
100
287000
3000
05:15
is to look and see is there more behavior change
101
290000
3000
05:18
in areas where there's less malaria.
102
293000
2000
05:20
So that's what this figure shows you.
103
295000
2000
05:22
This shows you -- in areas with low malaria, medium malaria, high malaria --
104
297000
4000
05:26
what happens to the number of sexual partners as you increase HIV prevalence.
105
301000
4000
05:30
If you look at the blue line,
106
305000
2000
05:32
the areas with low levels of malaria, you can see in those areas,
107
307000
3000
05:35
actually, the number of sexual partners is decreasing a lot
108
310000
3000
05:38
as HIV prevalence goes up.
109
313000
2000
05:40
Areas with medium levels of malaria it decreases some --
110
315000
2000
05:42
it doesn't decrease as much. And areas with high levels of malaria --
111
317000
3000
05:45
actually, it's increasing a little bit, although that's not significant.
112
320000
5000
05:50
This is not just through malaria.
113
325000
2000
05:52
Young women who live in areas with high maternal mortality
114
327000
3000
05:55
change their behavior less in response to HIV
115
330000
3000
05:58
than young women who live in areas with low maternal mortality.
116
333000
3000
06:01
There's another risk, and they respond less to this existing risk.
117
336000
4000
06:06
So by itself, I think this tells a lot about how people behave.
118
341000
3000
06:09
It tells us something about why we see limited behavior change in Africa.
119
344000
3000
06:12
But it also tells us something about policy.
120
347000
2000
06:14
Even if you only cared about AIDS in Africa,
121
349000
3000
06:17
it might still be a good idea to invest in malaria,
122
352000
3000
06:20
in combating poor indoor air quality,
123
355000
2000
06:22
in improving maternal mortality rates.
124
357000
2000
06:24
Because if you improve those things,
125
359000
2000
06:26
then people are going to have an incentive to avoid AIDS on their own.
126
361000
4000
06:30
But it also tells us something about one of these facts that we talked about before.
127
365000
4000
06:34
Education campaigns, like the one that the president is focusing on in his funding,
128
369000
4000
06:38
may not be enough, at least not alone.
129
373000
2000
06:40
If people have no incentive to avoid AIDS on their own,
130
375000
2000
06:42
even if they know everything about the disease,
131
377000
2000
06:44
they still may not change their behavior.
132
379000
2000
06:46
So the other thing that I think we learn here is that AIDS is not going to fix itself.
133
381000
3000
06:49
People aren't changing their behavior enough
134
384000
2000
06:51
to decrease the growth in the epidemic.
135
386000
3000
06:54
So we're going to need to think about policy
136
389000
2000
06:56
and what kind of policies might be effective.
137
391000
2000
06:58
And a great way to learn about policy is to look at what worked in the past.
138
393000
3000
07:01
The reason that we know that the ABC campaign
139
396000
2000
07:03
was effective in Uganda is we have good data on prevalence over time.
140
398000
3000
07:06
In Uganda we see the prevalence went down.
141
401000
2000
07:08
We know they had this campaign. That's how we learn about what works.
142
403000
3000
07:11
It's not the only place we had any interventions.
143
406000
2000
07:13
Other places have tried things, so why don't we look at those places
144
408000
4000
07:17
and see what happened to their prevalence?
145
412000
3000
07:20
Unfortunately, there's almost no good data
146
415000
2000
07:22
on HIV prevalence in the general population in Africa until about 2003.
147
417000
5000
07:27
So if I asked you, "Why don't you go and find me
148
422000
2000
07:29
the prevalence in Burkina Faso in 1991?"
149
424000
3000
07:32
You get on Google, you Google, and you find,
150
427000
3000
07:35
actually the only people tested in Burkina Faso in 1991
151
430000
3000
07:38
are STD patients and pregnant women,
152
433000
2000
07:40
which is not a terribly representative group of people.
153
435000
2000
07:42
Then if you poked a little more, you looked a little more at what was going on,
154
437000
3000
07:45
you'd find that actually that was a pretty good year,
155
440000
3000
07:48
because in some years the only people tested are IV drug users.
156
443000
3000
07:51
But even worse -- some years it's only IV drug users,
157
446000
2000
07:53
some years it's only pregnant women.
158
448000
2000
07:55
We have no way to figure out what happened over time.
159
450000
2000
07:57
We have no consistent testing.
160
452000
2000
07:59
Now in the last few years, we actually have done some good testing.
161
454000
5000
08:04
In Kenya, in Zambia, and a bunch of countries,
162
459000
3000
08:07
there's been testing in random samples of the population.
163
462000
3000
08:10
But this leaves us with a big gap in our knowledge.
164
465000
3000
08:13
So I can tell you what the prevalence was in Kenya in 2003,
165
468000
3000
08:16
but I can't tell you anything about 1993 or 1983.
166
471000
3000
08:19
So this is a problem for policy. It was a problem for my research.
167
474000
4000
08:23
And I started thinking about how else might we figure out
168
478000
4000
08:27
what the prevalence of HIV was in Africa in the past.
169
482000
2000
08:29
And I think that the answer is, we can look at mortality data,
170
484000
4000
08:33
and we can use mortality data to figure out what the prevalence was in the past.
171
488000
4000
08:37
To do this, we're going to have to rely on the fact
172
492000
2000
08:39
that AIDS is a very specific kind of disease.
173
494000
2000
08:41
It kills people in the prime of their lives.
174
496000
2000
08:43
Not a lot of other diseases have that profile. And you can see here --
175
498000
3000
08:46
this is a graph of death rates by age in Botswana and Egypt.
176
501000
4000
08:50
Botswana is a place with a lot of AIDS,
177
505000
2000
08:52
Egypt is a place without a lot of AIDS.
178
507000
2000
08:54
And you see they have pretty similar death rates among young kids and old people.
179
509000
3000
08:57
That suggests it's pretty similar levels of development.
180
512000
3000
09:00
But in this middle region, between 20 and 45,
181
515000
3000
09:03
the death rates in Botswana are much, much, much higher than in Egypt.
182
518000
4000
09:07
But since there are very few other diseases that kill people,
183
522000
4000
09:11
we can really attribute that mortality to HIV.
184
526000
3000
09:14
But because people who died this year of AIDS got it a few years ago,
185
529000
4000
09:18
we can use this data on mortality to figure out what HIV prevalence was in the past.
186
533000
5000
09:23
So it turns out, if you use this technique,
187
538000
2000
09:25
actually your estimates of prevalence are very close
188
540000
2000
09:27
to what we get from testing random samples in the population,
189
542000
3000
09:30
but they're very, very different than what UNAIDS tells us the prevalences are.
190
545000
5000
09:35
So this is a graph of prevalence estimated by UNAIDS,
191
550000
3000
09:38
and prevalence based on the mortality data
192
553000
2000
09:40
for the years in the late 1990s in nine countries in Africa.
193
555000
4000
09:44
You can see, almost without exception,
194
559000
2000
09:46
the UNAIDS estimates are much higher than the mortality-based estimates.
195
561000
4000
09:50
UNAIDS tell us that the HIV rate in Zambia is 20 percent,
196
565000
4000
09:54
and mortality estimates suggest it's only about 5 percent.
197
569000
4000
09:58
And these are not trivial differences in mortality rates.
198
573000
3000
10:01
So this is another way to see this.
199
576000
2000
10:03
You can see that for the prevalence to be as high as UNAIDS says,
200
578000
2000
10:05
we have to really see 60 deaths per 10,000
201
580000
2000
10:07
rather than 20 deaths per 10,000 in this age group.
202
582000
4000
10:11
I'm going to talk a little bit in a minute
203
586000
2000
10:13
about how we can use this kind of information to learn something
204
588000
3000
10:16
that's going to help us think about the world.
205
591000
2000
10:18
But this also tells us that one of these facts
206
593000
2000
10:20
that I mentioned in the beginning may not be quite right.
207
595000
3000
10:23
If you think that 25 million people are infected,
208
598000
2000
10:25
if you think that the UNAIDS numbers are much too high,
209
600000
3000
10:28
maybe that's more like 10 or 15 million.
210
603000
2000
10:30
It doesn't mean that AIDS isn't a problem. It's a gigantic problem.
211
605000
4000
10:34
But it does suggest that that number might be a little big.
212
609000
4000
10:38
What I really want to do, is I want to use this new data
213
613000
2000
10:40
to try to figure out what makes the HIV epidemic grow faster or slower.
214
615000
4000
10:44
And I said in the beginning, I wasn't going to tell you about exports.
215
619000
3000
10:47
When I started working on these projects,
216
622000
2000
10:49
I was not thinking at all about economics,
217
624000
2000
10:51
but eventually it kind of sucks you back in.
218
626000
3000
10:54
So I am going to talk about exports and prices.
219
629000
3000
10:57
And I want to talk about the relationship between economic activity,
220
632000
3000
11:00
in particular export volume, and HIV infections.
221
635000
4000
11:04
So obviously, as an economist, I'm deeply familiar
222
639000
4000
11:08
with the fact that development, that openness to trade,
223
643000
2000
11:10
is really good for developing countries.
224
645000
2000
11:12
It's good for improving people's lives.
225
647000
3000
11:15
But openness and inter-connectedness, it comes with a cost
226
650000
2000
11:17
when we think about disease. I don't think this should be a surprise.
227
652000
3000
11:20
On Wednesday, I learned from Laurie Garrett
228
655000
2000
11:22
that I'm definitely going to get the bird flu,
229
657000
2000
11:24
and I wouldn't be at all worried about that
230
659000
3000
11:27
if we never had any contact with Asia.
231
662000
3000
11:30
And HIV is actually particularly closely linked to transit.
232
665000
4000
11:34
The epidemic was introduced to the US
233
669000
2000
11:36
by actually one male steward on an airline flight,
234
671000
4000
11:40
who got the disease in Africa and brought it back.
235
675000
2000
11:42
And that was the genesis of the entire epidemic in the US.
236
677000
3000
11:45
In Africa, epidemiologists have noted for a long time
237
680000
4000
11:49
that truck drivers and migrants are more likely to be infected than other people.
238
684000
4000
11:53
Areas with a lot of economic activity --
239
688000
2000
11:55
with a lot of roads, with a lot of urbanization --
240
690000
3000
11:58
those areas have higher prevalence than others.
241
693000
2000
12:00
But that actually doesn't mean at all
242
695000
2000
12:02
that if we gave people more exports, more trade, that that would increase prevalence.
243
697000
4000
12:06
By using this new data, using this information about prevalence over time,
244
701000
4000
12:10
we can actually test that. And so it seems to be --
245
705000
4000
12:14
fortunately, I think -- it seems to be the case
246
709000
2000
12:16
that these things are positively related.
247
711000
2000
12:18
More exports means more AIDS. And that effect is really big.
248
713000
4000
12:22
So the data that I have suggests that if you double export volume,
249
717000
4000
12:26
it will lead to a quadrupling of new HIV infections.
250
721000
5000
12:31
So this has important implications both for forecasting and for policy.
251
726000
3000
12:34
From a forecasting perspective, if we know where trade is likely to change,
252
729000
4000
12:38
for example, because of the African Growth and Opportunities Act
253
733000
3000
12:41
or other policies that encourage trade,
254
736000
2000
12:43
we can actually think about which areas are likely to be heavily infected with HIV.
255
738000
5000
12:48
And we can go and we can try to have pre-emptive preventive measures there.
256
743000
6000
12:54
Likewise, as we're developing policies to try to encourage exports,
257
749000
3000
12:57
if we know there's this externality --
258
752000
2000
12:59
this extra thing that's going to happen as we increase exports --
259
754000
2000
13:01
we can think about what the right kinds of policies are.
260
756000
3000
13:04
But it also tells us something about one of these things that we think that we know.
261
759000
3000
13:07
Even though it is the case that poverty is linked to AIDS,
262
762000
3000
13:10
in the sense that Africa is poor and they have a lot of AIDS,
263
765000
3000
13:13
it's not necessarily the case that improving poverty -- at least in the short run,
264
768000
4000
13:17
that improving exports and improving development --
265
772000
2000
13:19
it's not necessarily the case that that's going to lead
266
774000
2000
13:21
to a decline in HIV prevalence.
267
776000
2000
13:24
So throughout this talk I've mentioned a few times
268
779000
2000
13:26
the special case of Uganda, and the fact that
269
781000
2000
13:28
it's the only country in sub-Saharan Africa with successful prevention.
270
783000
4000
13:32
It's been widely heralded.
271
787000
2000
13:34
It's been replicated in Kenya, and Tanzania, and South Africa and many other places.
272
789000
6000
13:40
But now I want to actually also question that.
273
795000
4000
13:44
Because it is true that there was a decline in prevalence
274
799000
3000
13:47
in Uganda in the 1990s. It's true that they had an education campaign.
275
802000
4000
13:51
But there was actually something else that happened in Uganda in this period.
276
806000
6000
13:57
There was a big decline in coffee prices.
277
812000
2000
13:59
Coffee is Uganda's major export.
278
814000
2000
14:01
Their exports went down a lot in the early 1990s -- and actually that decline lines up
279
816000
5000
14:06
really, really closely with this decline in new HIV infections.
280
821000
4000
14:10
So you can see that both of these series --
281
825000
3000
14:13
the black line is export value, the red line is new HIV infections --
282
828000
3000
14:16
you can see they're both increasing.
283
831000
2000
14:18
Starting about 1987 they're both going down a lot.
284
833000
2000
14:20
And then actually they track each other
285
835000
2000
14:22
a little bit on the increase later in the decade.
286
837000
2000
14:24
So if you combine the intuition in this figure
287
839000
2000
14:26
with some of the data that I talked about before,
288
841000
3000
14:29
it suggests that somewhere between 25 percent and 50 percent
289
844000
4000
14:33
of the decline in prevalence in Uganda
290
848000
2000
14:35
actually would have happened even without any education campaign.
291
850000
4000
14:39
But that's enormously important for policy.
292
854000
2000
14:41
We're spending so much money to try to replicate this campaign.
293
856000
2000
14:43
And if it was only 50 percent as effective as we think that it was,
294
858000
3000
14:46
then there are all sorts of other things
295
861000
2000
14:48
maybe we should be spending our money on instead.
296
863000
2000
14:50
Trying to change transmission rates by treating other sexually transmitted diseases.
297
865000
4000
14:54
Trying to change them by engaging in male circumcision.
298
869000
2000
14:56
There are tons of other things that we should think about doing.
299
871000
2000
14:58
And maybe this tells us that we should be thinking more about those things.
300
873000
4000
15:02
I hope that in the last 16 minutes I've told you something that you didn't know about AIDS,
301
877000
5000
15:07
and I hope that I've gotten you questioning a little bit
302
882000
2000
15:09
some of the things that you did know.
303
884000
2000
15:11
And I hope that I've convinced you maybe
304
886000
2000
15:13
that it's important to understand things about the epidemic
305
888000
2000
15:15
in order to think about policy.
306
890000
2000
15:18
But more than anything, you know, I'm an academic.
307
893000
2000
15:20
And when I leave here, I'm going to go back
308
895000
2000
15:22
and sit in my tiny office, and my computer, and my data.
309
897000
3000
15:25
And the thing that's most exciting about that
310
900000
2000
15:27
is every time I think about research, there are more questions.
311
902000
3000
15:30
There are more things that I think that I want to do.
312
905000
2000
15:32
And what's really, really great about being here
313
907000
2000
15:34
is I'm sure that the questions that you guys have
314
909000
2000
15:36
are very, very different than the questions that I think up myself.
315
911000
3000
15:39
And I can't wait to hear about what they are.
316
914000
2000
15:41
So thank you very much.
317
916000
2000

▲Back to top

ABOUT THE SPEAKER
Emily Oster - Assumption-busting economist
Emily Oster, a University of Chicago economist, uses the dismal science to rethink conventional wisdom, from her Harvard doctoral thesis that took on famed economist Amartya Sen to her recent work debunking assumptions on HIV prevalence in Africa.

Why you should listen

Emily Oster, an Assistant Professor of Economics at the University of Chicago, has a history of rethinking conventional wisdom.

Her Harvard doctoral thesis took on famed economist Amartya Sen and his claim that 100 million women were statistically missing from the developing world. He blamed misogynist medical care and outright sex-selective abortion for the gap, but Oster pointed to data indicating that in countries where Hepetitis B infections were higher, more boys were born. Through her unorthodox analysis of medical data, she accounted for 50% of the missing girls. Three years later, she would publish another paper amending her findings, stating that, after further study, the relationship between Hepetitis B and missing women was not apparent. This concession, along with her audacity to challenge economic assumptions and her dozens of other influential papers, has earned her the respect of the global academic community. 

She's also investigated the role of bad weather in the rise in witchcraft trials in Medieval Europe and what drives people to play the Powerball lottery. Her latest target: busting assumptions on HIV in Africa.

And she's an advice columnist too >>

 

More profile about the speaker
Emily Oster | Speaker | TED.com