ABOUT THE SPEAKER
Roger Stein - Financial management expert
Roger Stein wants to bring financial engineering to the world of drug funding.

Why you should listen

Roger Stein is a senior lecturer in finance at MIT's Sloan School of Management and a research affiliate at the MIT Laboratory for Financial Engineering. He is also the former chief analytics officer at State Street Global Exchange. He has been working in risk modeling and financial prediction for almost 25 years; his products and services are used widely in industry and have become benchmarks in banking and finance. With MIT colleagues, he is currently collaborating on a new model that uses modern risk management methods and financial engineering techniques to change the way new drug research is funded.

Previously he was managing director of research and academic relations globally for Moody’s Corporation, and prior to that was president of Moody’s Research Labs. He has a Ph.D. from New York University, has coauthored two full-length texts on applied analytics, and has written more than 50 academic articles and papers. He has also been practicing Aikido since 1980.

More profile about the speaker
Roger Stein | Speaker | TED.com
TED@State Street Boston

Roger Stein: A bold new way to fund drug research

Filmed:
952,929 views

Believe it or not, about 20 years' worth of potentially life-saving drugs are sitting in labs right now, untested. Why? Because they can't get the funding to go to trials; the financial risk is too high. Roger Stein is a finance guy, and he thinks deeply about mitigating risk. He and some colleagues at MIT came up with a promising new financial model that could move hundreds of drugs into the testing pipeline.
- Financial management expert
Roger Stein wants to bring financial engineering to the world of drug funding. Full bio

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

00:12
So this is a picture of my dad and me
0
639
3765
00:16
at the beach in Far Rockaway,
1
4404
2590
00:18
or actually Rockaway Park.
2
6994
1670
00:20
I'm the one with the blond hair.
3
8664
2690
00:23
My dad's the guy with the cigarette.
4
11354
2108
00:25
It was the '60s. A lot of people smoked back then.
5
13462
2367
00:27
In the summer of 2009,
6
15829
1843
00:29
my dad was diagnosed with lung cancer.
7
17672
4496
00:34
Cancer is one of those things
that actually touches everybody.
8
22168
3630
00:37
If you're a man in the United States of the America,
9
25798
2504
00:40
you've got about a one in two chance
10
28302
2078
00:42
of being diagnosed with cancer during your lifetime.
11
30380
3153
00:45
If you're a woman, you've got
about a one in three chance
12
33533
2778
00:48
of being diagnosed with cancer.
13
36311
1921
00:50
Everybody knows somebody
14
38232
1499
00:51
who's been diagnosed with cancer.
15
39731
3403
00:55
Now, my dad's doing better today,
16
43134
1603
00:56
and part of the reason for that is that
17
44737
3312
01:00
he was able to participate in the trial
18
48049
2059
01:02
of an experimental new drug
19
50108
1611
01:03
that happened to be specially formulated
20
51719
1425
01:05
and very good for his particular kind of cancer.
21
53144
1794
01:06
There are over 200 kinds of cancer.
22
54938
3553
01:10
And what I want to talk about today is
23
58491
2346
01:12
how we can help more people like my dad,
24
60837
2859
01:15
because we have to change the way
we think about raising money
25
63696
2861
01:18
to fund cancer research.
26
66557
2510
01:21
So a while after my dad was diagnosed,
27
69067
2055
01:23
I was having coffee with my friend Andrew Lo.
28
71122
2908
01:26
He's the head of the Laboratory
for Financial Engineering at MIT,
29
74030
3426
01:29
where I also have a position,
30
77456
2014
01:31
and we were talking about cancer.
31
79470
1507
01:32
And Andrew had been doing
his own bits of research,
32
80977
2551
01:35
and one of the things that he had been told
33
83528
1889
01:37
and that he'd learned from studying the literature
34
85417
4608
01:42
was that there's actually a big bottleneck.
35
90043
2068
01:44
It's very difficult to develop new drugs,
36
92111
2278
01:46
and the reason it's difficult to develop new drugs
37
94389
1779
01:48
is because in the early stages of drug development,
38
96168
2401
01:50
the drugs are very risky,
39
98569
1251
01:51
and they're very expensive.
40
99820
1646
01:53
So Andrew asked me if I'd want to
41
101466
1655
01:55
maybe work with him a bit,
42
103121
1212
01:56
work on some of the math and the analytics
43
104333
1634
01:57
and see if we could figure out
something we could do.
44
105967
3944
02:01
Now I'm not a scientist.
45
109911
1698
02:03
You know, I don't know how to build a drug.
46
111609
1898
02:05
And none of my coauthors, Andrew Lo
47
113507
2401
02:07
or Jose Maria Fernandez or David Fagnan --
48
115908
2692
02:10
none of those guys -- are scientists either.
49
118600
2554
02:13
We don't know the first thing about
how to make a cancer drug.
50
121154
3418
02:16
But we know a little bit about risk mitigation
51
124590
2128
02:18
and a little bit about financial engineering,
52
126718
2248
02:20
and so we started thinking, what could we do?
53
128966
2093
02:23
What I'm going to tell you about is some work
54
131059
1949
02:25
we've been doing over the last couple years
55
133008
2006
02:27
that we think could fundamentally change the way
56
135014
2893
02:29
research for cancer and lots
of other things gets done.
57
137907
3843
02:33
We want to let the research drive the funding,
58
141750
2413
02:36
not the other way around.
59
144163
2282
02:38
So in order to get started, let me tell you
60
146445
1230
02:39
how you get a drug financed.
61
147675
2063
02:41
Imagine that you're in your lab --
62
149738
1622
02:43
you're a scientist, you're not like me --
63
151360
1663
02:45
you're a scientist, and you've developed
64
153023
1854
02:46
a new compound that you think might be
65
154877
2339
02:49
therapeutic for somebody with cancer.
66
157216
2899
02:52
Well, what you do is, you test in animals,
67
160115
2208
02:54
you test in test tubes,
68
162323
1500
02:55
but there's this notion of going from the bench
69
163823
1743
02:57
to the bedside,
70
165566
1410
02:58
and in order to get from the bench, the lab,
71
166976
1575
03:00
to the bedside, to the patients,
72
168551
1773
03:02
you've got to get the drug tested.
73
170324
2058
03:04
And the way the drug gets tested is through a series
74
172382
1752
03:06
of, basically, experiments,
75
174134
1860
03:07
through these large, they're called trials,
76
175994
1848
03:09
that they do to determine whether the drug is safe
77
177842
2199
03:12
and whether it works and all these things.
78
180041
2188
03:14
So the FDA has a very specific protocol.
79
182229
2725
03:16
In the first phase of this testing,
80
184954
1412
03:18
which is called testing for toxicity,
81
186366
2395
03:20
it's called Phase I.
82
188761
1542
03:22
In the first phase, you give
the drug to healthy people
83
190303
2781
03:25
and you see if it actually makes them sick.
84
193084
3249
03:28
In other words, are the side effects just so severe
85
196333
1887
03:30
that no matter how much good it does,
86
198220
1364
03:31
it's not going to be worth it?
87
199584
1475
03:33
Does it cause heart attacks, kill people,
88
201059
2000
03:35
liver failure, this kind of thing?
89
203059
1611
03:36
And it turns out, that's a pretty high hurdle.
90
204670
1949
03:38
About a third of all drugs drop out at that point.
91
206619
2812
03:41
In the next phase, you test
to see if the drug's effective,
92
209431
2764
03:44
and what you do there is you
give it to people with cancer
93
212195
1904
03:46
and you see if it actually makes them better.
94
214099
1619
03:47
And that's also a higher hurdle. People drop out.
95
215718
2019
03:49
And in the third phase, you actually
test it on a very large sample,
96
217737
2820
03:52
and what you're trying to determine
is what the right dose is, and also,
97
220557
3196
03:55
is it better than what's available today?
98
223753
1894
03:57
If not, then why build it?
99
225647
2039
03:59
When you're done with all that,
100
227686
2068
04:01
what you have is a very small percentage of drugs
101
229754
2471
04:04
that start the process actually
come out the other side.
102
232225
2939
04:07
So those blue bottles, those blue bottles save lives,
103
235164
3560
04:10
and they're also worth billions,
104
238724
1462
04:12
sometimes billions a year.
105
240186
2372
04:14
So now here's a question:
106
242558
1682
04:16
if I were to ask you, for example,
107
244240
3316
04:19
to make a one-time investment of, say,
108
247556
2069
04:21
200 million dollars
109
249625
1734
04:23
to buy one of those bottles,
110
251359
1669
04:25
so 200 million dollars up front, one time,
111
253028
2199
04:27
to buy one of those bottles,
112
255227
1122
04:28
I won't tell you which one it is,
113
256349
1753
04:30
and in 10 years, I'll tell you whether
you have one of the blue ones.
114
258102
3931
04:34
Does that sound like a good deal for anybody?
115
262033
2043
04:36
No. No, right?
116
264076
1466
04:37
And of course, it's a very, very risky trial position,
117
265542
1966
04:39
and that's why it's very hard to get funding,
118
267508
1720
04:41
but to a first approximation,
119
269228
1712
04:42
that's actually the proposal.
120
270940
1651
04:44
You have to fund these things
121
272591
1567
04:46
from the early stages on. It takes a long time.
122
274158
2991
04:49
So Andrew said to me, he said,
123
277149
3441
04:52
"What if we stop thinking about these as drugs?
124
280590
2349
04:54
What if we start thinking about
them as financial assets?"
125
282939
2674
04:57
They've got really weird payoff
structures and all that,
126
285613
2077
04:59
but let's throw everything we know
127
287690
1458
05:01
about financial engineering at them.
128
289148
1667
05:02
Let's see if we can use all the tricks of the trade
129
290815
1785
05:04
to figure out how to make these drugs
130
292600
2804
05:07
work as financial assets?
131
295404
2726
05:10
Let's create a giant fund.
132
298130
1684
05:11
In finance, we know what to do
133
299814
1495
05:13
with assets that are risky.
134
301309
1326
05:14
You put them in a portfolio
135
302635
1473
05:16
and you try to smooth out the returns.
136
304108
1680
05:17
So we did some math, and it turned out
137
305788
1351
05:19
you could make this work,
but in order to make it work,
138
307139
2807
05:21
you need about 80 to 150 drugs.
139
309946
2569
05:24
Now the good news is, there's plenty of drugs
140
312515
1590
05:26
that are waiting to be tested.
141
314105
1281
05:27
We've been told that there's a backlog
142
315386
2337
05:29
of about 20 years of drugs
143
317723
2289
05:32
that are waiting to be tested but can't be funded.
144
320012
2655
05:34
In fact, that early stage of the funding process,
145
322667
2106
05:36
that Phase I and pre-clinical stuff,
146
324773
2236
05:39
that's actually, in the industry,
called the Valley of Death
147
327009
2332
05:41
because it's where drugs go to die.
148
329359
1675
05:43
It's very hard to for them to get through there,
149
331034
1577
05:44
and of course, if you can't get through there,
you can't get to the later stages.
150
332611
4203
05:48
So we did this math, and we figured out, okay,
151
336814
1596
05:50
well, you know, you need about 80 to, say, 150,
152
338410
2599
05:53
or something like that, drugs.
153
341009
1877
05:54
And then we did a little more math, and we said,
154
342886
1875
05:56
okay, well that's a fund of about
155
344761
1610
05:58
three to 15 billion dollars.
156
346371
3314
06:01
So we kind of created a new problem
157
349685
2356
06:04
by solving the old one.
158
352041
1331
06:05
We were able to get rid of the risk,
but now we need a lot of capital,
159
353372
2615
06:07
and there's only one place to get that kind of capital,
160
355987
1938
06:09
the capital markets.
161
357925
1164
06:11
Venture capitalists don't have it.
Philanthropies don't have it.
162
359089
2814
06:13
But we have to figure out how we can
get people in the capital markets,
163
361903
2004
06:15
who traditionally don't invest in this stuff,
164
363907
1543
06:17
to want to invest in this stuff.
165
365450
1607
06:19
So again, financial engineering was helpful here.
166
367057
2965
06:22
Imagine the megafund actually starts empty,
167
370022
2983
06:25
and what it does is it issues some debt
168
373005
2783
06:27
and some equity,
169
375788
1321
06:29
and that generates cash flow.
170
377109
1679
06:30
That cash flow is used, then, to buy
171
378788
1240
06:32
that big portfolio of drugs that you need,
172
380028
2017
06:34
and those drugs start working their way
173
382045
1536
06:35
through that approval process,
174
383581
1812
06:37
and each time they go through
a next phase of approval,
175
385393
1634
06:39
they gain value.
176
387027
964
06:39
And most of them don't make it,
177
387991
1272
06:41
but a few of them do,
178
389263
1299
06:42
and with the ones that gain value, you can sell some,
179
390562
2589
06:45
and when you sell them,
180
393151
1426
06:46
you have money to pay the interest on those bonds,
181
394577
1539
06:48
but you also have money to
fund the next round of trials.
182
396116
2478
06:50
It's almost self-funding.
183
398594
2109
06:52
You do that for the course of the transaction,
184
400703
2718
06:55
and when you're done, you liquidate the portfolio,
185
403421
1681
06:57
pay back the bonds, and you can
186
405102
1665
06:58
give the equity holders a nice return.
187
406767
1952
07:00
So that was the theory, and we talked about it
188
408719
2243
07:02
for a bit, we did a bunch of experiments,
189
410962
1672
07:04
and then we said, let's really try to test it.
190
412634
2621
07:07
We spent the next two years doing research.
191
415255
3429
07:10
We talked to hundreds of experts in drug financing
192
418684
3112
07:13
and venture capital.
193
421796
1644
07:15
We talked to people who have developed drugs.
194
423440
2203
07:17
We talked to pharmaceutical companies.
195
425643
1566
07:19
We actually looked at the data for drugs,
196
427209
3364
07:22
over 2,000 drugs that had been approved or denied
197
430573
2680
07:25
or withdrawn,
198
433253
1415
07:26
and we also ran millions of simulations.
199
434668
3308
07:29
And all that actually took a lot of time.
200
437976
2426
07:32
But when we were done, what we found was something that was sort of surprising.
201
440402
3033
07:35
It was actually feasible to structure that fund
202
443435
2019
07:37
such that when you were done structuring it,
203
445454
2338
07:39
you could actually produce low-risk bonds
204
447792
2335
07:42
that would be attractive to bond holders,
205
450127
1397
07:43
that would give you yields of
about five to eight percent,
206
451524
2784
07:46
and you could produce equity
207
454308
1673
07:47
that would give equity holders
about a 12-percent return.
208
455981
2311
07:50
Now those returns aren't going to be attractive
209
458292
1526
07:51
to a venture capitalist.
210
459818
1569
07:53
Venture capitalists are those guys
who want to make those big bets
211
461387
2186
07:55
and get those billion dollar payoffs.
212
463573
2136
07:57
But it turns out, there are lots of other folks
that would be interested in that.
213
465709
2823
08:00
That's right in the investment sweet spot
214
468532
1751
08:02
of pension funds and 401(k) plans
215
470283
2653
08:04
and all this other stuff.
216
472936
1585
08:06
So we published some articles
in the academic press.
217
474521
2285
08:08
We published articles in medical journals.
218
476806
2599
08:11
We published articles in finance journals.
219
479405
2340
08:13
But it wasn't until we actually
220
481745
1268
08:15
got the popular press interested in this
221
483013
1978
08:16
that we began to get some traction.
222
484991
2662
08:19
We wanted to do something more than
just make people aware of it, though.
223
487653
2483
08:22
We wanted people to get involved.
224
490136
1481
08:23
So what we did was, we actually
took all of our computer code
225
491617
2090
08:25
and made that available online
226
493707
1554
08:27
under an open-source license
227
495261
1665
08:28
to anybody that wanted it.
228
496926
1351
08:30
And you guys can download it today
229
498277
1734
08:32
if you want to run your own experiments
to see if this would work.
230
500011
2235
08:34
And that was really effective, because people
231
502246
1415
08:35
that didn't believe our assumptions
232
503661
1734
08:37
could try their own assumptions
and see how it would work.
233
505395
2706
08:40
Now there's an obvious problem, which is,
234
508101
2638
08:42
is there enough money in
the world to fund this stuff?
235
510739
2172
08:44
I've told you there's enough drugs,
236
512911
1414
08:46
but is there enough money?
237
514325
1831
08:48
There's 100 trillion dollars of capital
238
516156
2725
08:50
currently invested in fixed-income securities.
239
518881
3541
08:54
That's a hundred thousand billion.
240
522422
3603
08:58
There's plenty of money.
241
526025
1926
08:59
(Laughter)
242
527951
3223
09:03
But what we realized was that it's
more than just money that's required.
243
531174
3037
09:06
We had to get people motivated,
people to get involved,
244
534211
1829
09:08
and people had to understand this.
245
536040
2524
09:10
And so we started thinking about all
the different things that could go wrong.
246
538564
2900
09:13
What are all the challenges to doing
this that might get in the way?
247
541486
3329
09:16
And we had a long list, and so what we did was
248
544815
2047
09:18
we assigned a bunch of people, including ourselves,
249
546862
2565
09:21
different pieces of this problem,
250
549427
2185
09:23
and we said, could you start
a work stream on credit risk?
251
551612
2474
09:26
Could you start a work stream
on the regulatory aspects?
252
554086
2376
09:28
Could you start a work stream on
253
556462
1454
09:29
how you would actually manage so many projects?
254
557916
2954
09:32
And we had all these experts get together
255
560870
1618
09:34
and do these different work streams,
256
562488
1291
09:35
and then we held a conference.
257
563779
2593
09:38
The conference was held over
the summer, this past summer.
258
566372
2622
09:40
It was an invitation-only conference.
259
568994
1377
09:42
It was sponsored by the American Cancer Society
260
570371
2023
09:44
and done in collaboration
261
572394
1609
09:46
with the National Cancer Institute.
262
574003
2388
09:48
And we had experts from every field
that we thought would be important,
263
576391
2755
09:51
including the government, including
264
579146
2095
09:53
people that run research centers and so on,
265
581241
1557
09:54
and for two days they sat around
and heard the reports
266
582798
1989
09:56
from those five work streams,
and they talked about it.
267
584787
1916
09:58
It was the first time
268
586703
1431
10:00
that the people who could actually make this happen
269
588134
1549
10:01
sat across the table from each other
270
589683
1644
10:03
and had these conversations.
271
591327
2308
10:05
Now these conferences, it's typical to have a dinner,
272
593635
3509
10:09
and at that dinner,
273
597144
1328
10:10
you kind of get to know each other,
274
598472
1602
10:12
sort of like what we're doing here.
275
600074
1456
10:13
I happened to look out the window,
276
601530
1606
10:15
and hand on my heart, I looked out the window
277
603136
1878
10:17
on the night of this conference --
it was the summertime --
278
605014
1898
10:18
and that's what I saw, it was a double rainbow.
279
606912
2382
10:21
So I'd like to think it was a good sign.
280
609294
3000
10:24
Since the conference, we've got people working
281
612294
1986
10:26
between Paris and San Francisco,
282
614280
1932
10:28
lots of different folks working on this
283
616212
1636
10:29
to try to see if we can really make it happen.
284
617848
2370
10:32
We're not looking to start a fund,
285
620218
1494
10:33
but we want somebody else to do this.
286
621712
2178
10:35
Because, again, I'm not a scientist.
287
623890
2663
10:38
I can't build a drug.
288
626553
1737
10:40
I'm never going to have enough money
289
628290
1354
10:41
to fund even one of those trials.
290
629644
2463
10:44
But all of us together, with our 401(k)'s,
291
632107
2975
10:47
with our 529 plans, with our pension plans,
292
635082
2492
10:49
all of us together can actually fund hundreds of trials
293
637574
2870
10:52
and get paid well for doing it
294
640444
1770
10:54
and save millions of lives like my dad.
295
642214
2776
10:56
Thank you.
296
644990
2330
10:59
(Applause)
297
647320
3729

▲Back to top

ABOUT THE SPEAKER
Roger Stein - Financial management expert
Roger Stein wants to bring financial engineering to the world of drug funding.

Why you should listen

Roger Stein is a senior lecturer in finance at MIT's Sloan School of Management and a research affiliate at the MIT Laboratory for Financial Engineering. He is also the former chief analytics officer at State Street Global Exchange. He has been working in risk modeling and financial prediction for almost 25 years; his products and services are used widely in industry and have become benchmarks in banking and finance. With MIT colleagues, he is currently collaborating on a new model that uses modern risk management methods and financial engineering techniques to change the way new drug research is funded.

Previously he was managing director of research and academic relations globally for Moody’s Corporation, and prior to that was president of Moody’s Research Labs. He has a Ph.D. from New York University, has coauthored two full-length texts on applied analytics, and has written more than 50 academic articles and papers. He has also been practicing Aikido since 1980.

More profile about the speaker
Roger Stein | Speaker | TED.com
THE ORIGINAL VIDEO ON TED.COM