1
00:00:10,740 --> 00:00:14,370
Hello, everyone, and welcome to the Oxford Martin School.

2
00:00:14,370 --> 00:00:21,930
My name's Charles Godfrey, I'm the director of the Oxford Martin School and we're here because of the visionary bequest by James Martin.

3
00:00:21,930 --> 00:00:30,270
I'm delighted to say that James widow, Jim Martens widow Lillian, is in the audience this afternoon with her daughter Gero,

4
00:00:30,270 --> 00:00:39,990
and I'm delighted to welcome our speaker, Diane Coyle, this afternoon in our series on evolving economic thought.

5
00:00:39,990 --> 00:00:45,750
Diane is currently the inaugural Bennett Professor of Public Policy and co-director of the Bennett Institute for

6
00:00:45,750 --> 00:00:54,390
Public Policy at the University of Cambridge and has been there about 15 months was previously at Manchester.

7
00:00:54,390 --> 00:00:59,820
Don has done a huge number of things, both in the academic world and in the public policy world.

8
00:00:59,820 --> 00:01:03,780
I'll just mention a couple of the things that Dan is involved at the moment.

9
00:01:03,780 --> 00:01:05,340
Member of the Natural Capital Committee,

10
00:01:05,340 --> 00:01:12,840
chaired by the town expert advice to the National Infrastructure Commission and a member of the very influential Council of Economic

11
00:01:12,840 --> 00:01:22,680
Advisers that was awarded a CVA in 2018 for contributions towards the public understanding of economics and how fields of public policy,

12
00:01:22,680 --> 00:01:27,870
economics, technology, industrial strategy and global inequality.

13
00:01:27,870 --> 00:01:36,210
And Dan is a fabulous author, and if you haven't read it, I do recommend GDP a brief but affectionate history,

14
00:01:36,210 --> 00:01:40,380
and it takes real skill to write so wonderfully on that topic.

15
00:01:40,380 --> 00:01:45,810
And I have to say I before I even read it, I was invited by the fact it doesn't have a picture of you in it,

16
00:01:45,810 --> 00:01:52,890
but it does have a picture of Utah cabbage who contributes not at all to GDP, but greatly to welfare.

17
00:01:52,890 --> 00:02:04,380
Sit down, please come and give your lecture. Good evening, everybody.

18
00:02:04,380 --> 00:02:10,710
My first challenge is to make sure I've got this the right way up because although I write about technology, I'm not very good at operating it.

19
00:02:10,710 --> 00:02:14,340
So hello all of you, and can you hear me properly at the back?

20
00:02:14,340 --> 00:02:23,130
OK. It's a great pleasure to be here today, and thank you very much for the invitation to take part in this series.

21
00:02:23,130 --> 00:02:32,430
And amongst my other interests, the state of economics has been one of them and I've been quite involved in the core curriculum project as well.

22
00:02:32,430 --> 00:02:35,020
And so I was delighted to accept the invitation,

23
00:02:35,020 --> 00:02:42,270
and I'm going to talk about my own work on digital technology, which I've been doing for many years now.

24
00:02:42,270 --> 00:02:51,690
In fact, my first book about it came out 22 years ago, and when I started getting interested in the mid 1990s and digital technology,

25
00:02:51,690 --> 00:02:55,650
I said to a very distinguished economist he was already very distinguished.

26
00:02:55,650 --> 00:03:00,270
He's now is more distinguished. I can't see him in the room this evening, but I'd better not name him.

27
00:03:00,270 --> 00:03:03,900
I said, the internet is wonderful. It's going to change all kinds of things.

28
00:03:03,900 --> 00:03:07,260
And he said, Why is the internet interesting?

29
00:03:07,260 --> 00:03:12,600
It's going to reduce transactions costs a bit, but we know how to handle transactions costs in our model.

30
00:03:12,600 --> 00:03:20,310
So why are you bothering wasting your time on this, on this new technology? And in a way that sets my theme this evening?

31
00:03:20,310 --> 00:03:25,350
Because what I want to argue is that the the characteristics,

32
00:03:25,350 --> 00:03:34,410
the economic characteristics of digital technology mean that the way we need to think about doing economics has to change.

33
00:03:34,410 --> 00:03:38,940
So that's the theme, and I'm going to start with this man.

34
00:03:38,940 --> 00:03:44,490
Some of you will know him. Hal Varian, chief economist at Google.

35
00:03:44,490 --> 00:03:51,510
And why does he lose sleep at night? And I ask this because I've heard him give a talk a couple of times,

36
00:03:51,510 --> 00:03:57,360
claiming that there is nothing to worry about in terms of competition in digital markets.

37
00:03:57,360 --> 00:04:02,460
Sure. Google has 98 percent of search outside China.

38
00:04:02,460 --> 00:04:06,660
It's quite a lot of it out of 100 percent, but there's nothing to worry about,

39
00:04:06,660 --> 00:04:12,600
he says, because any new technological upstarts that had a better search engine,

40
00:04:12,600 --> 00:04:20,340
better search technology could very quickly knock Google out of that position and grab that market for itself.

41
00:04:20,340 --> 00:04:24,390
And he says this is exactly what Google did to Yahoo back in the day.

42
00:04:24,390 --> 00:04:28,590
And so he loses sleep over, not over.

43
00:04:28,590 --> 00:04:32,310
Department of Justice investigations or European Commission investigations.

44
00:04:32,310 --> 00:04:40,290
But over the fear that the next Google is out there and it's going to eat the existing Google now.

45
00:04:40,290 --> 00:04:43,140
That's why he loses sleep. A lot of sceptics.

46
00:04:43,140 --> 00:04:49,170
There are a lot of sceptics about this claim, of course, and you might have heard the term gaffer or Google,

47
00:04:49,170 --> 00:04:53,820
Apple, Facebook, Amazon, Microsoft is sometimes there and sometimes not.

48
00:04:53,820 --> 00:04:57,810
Uber is sometimes there, but you can't make a nice acronym with Uber.

49
00:04:57,810 --> 00:05:05,940
So we call it gaffer and concerned about the dominance of digital companies in some of these markets has increased everywhere.

50
00:05:05,940 --> 00:05:10,440
European Commission has been particularly active in pursuing some of the cases,

51
00:05:10,440 --> 00:05:17,520
and the even the American antitrust authorities are now getting somewhat interested in it in this country.

52
00:05:17,520 --> 00:05:23,520
We had a digital competition expert panel chaired by Jason Furman, who's an economist at the Harvard Kennedy School,

53
00:05:23,520 --> 00:05:33,720
and I was a member of that as evidence of some trends towards concentration more widely in the US economy and maybe this one as well,

54
00:05:33,720 --> 00:05:41,160
but especially in digital, because these markets do tend to be dominated by one company.

55
00:05:41,160 --> 00:05:49,000
And there are reasons for that, and the reasons lie in the underlying economic characteristics of the technology.

56
00:05:49,000 --> 00:05:59,620
And these are what they are. First of all, superstar features, this superstar name came from a paper by Sherman Rosen in the early or mid 80s,

57
00:05:59,620 --> 00:06:06,280
where he was trying to explain the very high pay of a particular movie Star Wars food star.

58
00:06:06,280 --> 00:06:16,510
And it operates on the demand side and the supply side. On the supply side, there are very high fixed costs and very low or zero marginal cost.

59
00:06:16,510 --> 00:06:25,780
So increasing returns to scale and particularly depending on the level that those fixed costs, that that scale can be quite large.

60
00:06:25,780 --> 00:06:31,270
And on the demand side, you often talking about experience goods so people don't know what they're like until

61
00:06:31,270 --> 00:06:36,580
they have actually already consumed a movie watching a baseball match and so on.

62
00:06:36,580 --> 00:06:42,910
And so once people start to hear by word of mouth or other forms of contact that something is good,

63
00:06:42,910 --> 00:06:47,320
then even if by any objective characteristics is not better,

64
00:06:47,320 --> 00:06:53,500
particularly than anything else in the market, the demand will congregate on those particular suppliers.

65
00:06:53,500 --> 00:06:59,470
So those are the superstar features and they operate in these digital markets.

66
00:06:59,470 --> 00:07:06,130
They're also what are called indirect network effects in the jargon. Network effects are familiar for everyday life.

67
00:07:06,130 --> 00:07:11,190
If you want to make a phone call, then the more people on the phone network, the better it is for you.

68
00:07:11,190 --> 00:07:18,910
Indirect just refers to the fact that in many digital markets, they are matching suppliers to consumers.

69
00:07:18,910 --> 00:07:24,730
And so if you're a consumer of one of these goods, if you want to hire an Airbnb apartment,

70
00:07:24,730 --> 00:07:28,510
the more people supplying the Airbnb apartments, the better it is for you.

71
00:07:28,510 --> 00:07:34,270
And if you're a supplier who wants to rent out your apartment, the more consumers on the platform, the better it is for you.

72
00:07:34,270 --> 00:07:41,800
And so these indirect network effects are also mutually reinforcing and encouraging of scale.

73
00:07:41,800 --> 00:07:45,910
These platforms are very often matching their matching, quite varied supply,

74
00:07:45,910 --> 00:07:51,050
like the range of apartments that you can get in the city with quite varied demand.

75
00:07:51,050 --> 00:07:59,470
So they're quite heterogeneous and there's a matching offer going on there, and they have a particular price structure.

76
00:07:59,470 --> 00:08:07,300
If one side of the market is more likely to go to other suppliers, so usually it's the consumer side of the market.

77
00:08:07,300 --> 00:08:11,830
Then the other side has to subsidise them to stay on the platform.

78
00:08:11,830 --> 00:08:17,140
So many of these cases, consumers don't pay anything and supplies the platform,

79
00:08:17,140 --> 00:08:24,580
appetisers and search or apartment owners visiting their apartments, pay a commission or fee and a commission to the platform.

80
00:08:24,580 --> 00:08:32,590
So once one side subsidises the other, all of these mean there's what what's called the chicken and egg problem, which is exactly what it sounds like.

81
00:08:32,590 --> 00:08:36,130
It's really hard to get one of these platforms going and succeeding.

82
00:08:36,130 --> 00:08:43,180
And there's some new work out by Annabel GAWA and Michael Cusumano suggesting that as many as four and five platforms fail.

83
00:08:43,180 --> 00:08:48,580
I wouldn't be surprised if actually it was more than that. It's really hard to get to viable scale,

84
00:08:48,580 --> 00:08:59,090
and the big American ones have very large sums of venture capital money covering their losses for a long period, and those losses can be large indeed.

85
00:08:59,090 --> 00:09:03,430
And so all of this adds up to you need to be big.

86
00:09:03,430 --> 00:09:08,270
You've got to get to a certain critical mass. Once you do, there's a characteristic dynamic.

87
00:09:08,270 --> 00:09:13,810
So you've been bumping along for a while, you get to the critical point and then all of a sudden you're huge you.

88
00:09:13,810 --> 00:09:18,490
That takes off very quickly. And if you're not big, then you're dead.

89
00:09:18,490 --> 00:09:23,080
So those are the basic characteristics of these markets.

90
00:09:23,080 --> 00:09:30,610
So on the supply side, increasing returns to scale on the demand side that often experience good information goods.

91
00:09:30,610 --> 00:09:35,140
There are these large externalities in consumption in this form of these network effects.

92
00:09:35,140 --> 00:09:40,720
If you do basic economics, we focus on the externalities in production. But these are in consumption.

93
00:09:40,720 --> 00:09:48,160
There's great heterogeneity in the product. Every search is different or they're not, but you get the idea.

94
00:09:48,160 --> 00:09:49,990
And even in a competitive market,

95
00:09:49,990 --> 00:09:58,360
price is not going to equal marginal cost because marginal cost is so low and you've got to cover the fixed cost as well.

96
00:09:58,360 --> 00:10:02,260
So that's what they're like. There's the basics.

97
00:10:02,260 --> 00:10:08,330
There are also other features of these digital platforms that make it very hard to think about them,

98
00:10:08,330 --> 00:10:13,660
certainly in the competition context, in the way that you do any other kind of business.

99
00:10:13,660 --> 00:10:17,350
And one of this is the strategy called Envelopment. I love this.

100
00:10:17,350 --> 00:10:24,610
I took this picture in Manchester of Uber deliverers Uber Eats deliveries and Deliveroo deliveries,

101
00:10:24,610 --> 00:10:30,140
walking alongside each other as a form of healthy competition, obviously.

102
00:10:30,140 --> 00:10:37,550
So if you've got a lot of uses on one side of one market, if you've got a lot of consumers using your taxi service,

103
00:10:37,550 --> 00:10:45,080
which is what Uber is, then you can use those consumers to cross sell a completely different product.

104
00:10:45,080 --> 00:10:49,700
So this isn't quite like the bundling that we're used to in oligopolistic market and economics.

105
00:10:49,700 --> 00:10:55,190
The products have nothing to do with each other. It just happens to be that Uber decided in this case,

106
00:10:55,190 --> 00:11:02,180
it would get into the food delivery business because it had a lot of users already on its app, and that was easy.

107
00:11:02,180 --> 00:11:04,640
And so that means if you are thinking about competition,

108
00:11:04,640 --> 00:11:10,580
if a company is dominant in one market or a large animal market, you not only need to think about that market,

109
00:11:10,580 --> 00:11:19,070
you need to think about all the other ones that could get into and is scale is having a lot of uses,

110
00:11:19,070 --> 00:11:23,510
a bull market and a barrier to entry to other people getting into other markets.

111
00:11:23,510 --> 00:11:29,840
Or can you expect the dominant players to dive into anything that they think looks promising?

112
00:11:29,840 --> 00:11:35,270
Uber's latest thing is dark kitchens capitalising on the delivery market.

113
00:11:35,270 --> 00:11:43,580
They now want to buy up dark kitchens in big cities so that they can produce all different kinds of food and capture the whole supply chain there.

114
00:11:43,580 --> 00:11:50,540
So they do. That's very common. And the other thing that's really striking is the data barrier,

115
00:11:50,540 --> 00:11:57,440
and people sometimes call this the data flywheel or the data loop, and it's a self-reinforcing feedback mechanism.

116
00:11:57,440 --> 00:12:02,360
If you are a big company and you've got a great service, you get a lot of customers.

117
00:12:02,360 --> 00:12:10,490
If you have a lot of customers and you accumulate their data, you can use that data to improve your service and so it feeds on itself.

118
00:12:10,490 --> 00:12:19,300
And so one of the things that we looked at in our report was whether there are ways in particular to break down this data loop.

119
00:12:19,300 --> 00:12:30,440
But what all of these features add up to is that the kind of conventional analysis that we do in competition policy is much harder and in fact,

120
00:12:30,440 --> 00:12:38,600
it doesn't work at all because in a conventional competition enquiry and I spent eight years doing this and you look at

121
00:12:38,600 --> 00:12:46,700
the possibility for a merged company or a dominant company to increase its price by a small but significant amount.

122
00:12:46,700 --> 00:12:54,560
And you see what substitution possibilities that all can your customers easily switch to something that's very similar or not?

123
00:12:54,560 --> 00:13:02,150
And so it's called a snip test, which stands for small but significant.

124
00:13:02,150 --> 00:13:06,560
Increase in price, and I can't remember the instance for the minute.

125
00:13:06,560 --> 00:13:13,400
You can't do these in these markets, partly because you've got this phenomenon operating across different markets,

126
00:13:13,400 --> 00:13:19,910
so you need to figure out exactly what the boundaries of the markets that you're looking at.

127
00:13:19,910 --> 00:13:32,360
But also because of the asymmetry in price and a perfectly competitive platform may still be charging a zero price to its consumers.

128
00:13:32,360 --> 00:13:36,620
And so that standard kind of test is really difficult to do.

129
00:13:36,620 --> 00:13:44,400
An alternative would be to look at profitability, so the companies dominating a market, you would expect it to have higher than average profitability.

130
00:13:44,400 --> 00:13:52,670
And sometimes you do this analysis in an enquiry. But these are companies which even until they're very large, are losing money.

131
00:13:52,670 --> 00:13:57,970
So. It's a losing money and then charging zero prices, what's the problem?

132
00:13:57,970 --> 00:14:02,200
You have to ask as a competition analyst, why is that a problem at all?

133
00:14:02,200 --> 00:14:10,900
And there are these coaches that dynamics as well, that you're looking at potential competitors to a dominant company.

134
00:14:10,900 --> 00:14:15,220
And it's the Halliburton question, the Hal Varian challenge. You know,

135
00:14:15,220 --> 00:14:22,720
I'm really worried about this small guy because at some stage they could grow very quickly and capture this market conventional

136
00:14:22,720 --> 00:14:30,580
competition and don't handle these contexts very well and have never really handled the dynamics of markets really well.

137
00:14:30,580 --> 00:14:37,930
If you want to think about what the home of Google dominance might be, it's that there's innovation that's not going to happen because of it,

138
00:14:37,930 --> 00:14:41,800
that it will dominate the market such that the better search engine can't ever

139
00:14:41,800 --> 00:14:50,050
get into that market because of the data barrier or because of the sheer scale.

140
00:14:50,050 --> 00:14:52,870
So we've got to ask what the dynamic costs of this kind of dominance.

141
00:14:52,870 --> 00:14:58,510
And we are in a world not of static competition, which we're used to looking at in competition policy.

142
00:14:58,510 --> 00:15:09,790
But the world of sugar and competition, the creative destruction that he wrote about so famously and there are indeed examples of digital overthrow.

143
00:15:09,790 --> 00:15:17,020
So rather than look at competition in the market, we're looking at competition for the market in the terminology that will be introduced,

144
00:15:17,020 --> 00:15:22,900
and there hasn't been examples of this kind of overthrow. I've given you a chart here of Facebook.

145
00:15:22,900 --> 00:15:32,680
And if you look carefully, you'll see that actually it ends in 2014. So even before the recent Facebook scandals, there was some engagement metrics.

146
00:15:32,680 --> 00:15:39,940
This suggests that Facebook has passed its peak and may not be the dominant social media platform.

147
00:15:39,940 --> 00:15:48,370
And apparently, young people are using all kinds of things these days that we old folks sitting in this room have never even heard of.

148
00:15:48,370 --> 00:15:50,050
So maybe this will be the same.

149
00:15:50,050 --> 00:15:56,950
And there are earlier examples, too, like Microsoft Internet Explorer being the dominant browser and then being overturned,

150
00:15:56,950 --> 00:16:02,130
or as this chart shows, Facebook itself overthrowing MySpace.

151
00:16:02,130 --> 00:16:11,320
Look at MySpace. This deep research for this, and it turns out that Kim Kardashian is still on MySpace, so it isn't completely dead yet.

152
00:16:11,320 --> 00:16:16,810
But this kind of dynamic analysis is difficult for competition authorities to do

153
00:16:16,810 --> 00:16:22,270
because you have to think about what might happen to the sector in the future.

154
00:16:22,270 --> 00:16:28,000
You got to try to forecast the technology and the consumer response to that technology.

155
00:16:28,000 --> 00:16:32,080
So digital is different. It's got these increasing returns to scale. He's not alone.

156
00:16:32,080 --> 00:16:38,380
You need dynamics, these externalities. It's also a public good is it's not rivalry.

157
00:16:38,380 --> 00:16:44,770
In many cases, a piece of software is in technical terms of public good because once it's been written,

158
00:16:44,770 --> 00:16:51,820
it can be used by any number of people, and that doesn't detract from the first user.

159
00:16:51,820 --> 00:16:55,810
This is all a roundabout way of saying through the example of competition policy,

160
00:16:55,810 --> 00:17:03,040
that the conventional ways we have of thinking about production in the economy don't really apply in this sector.

161
00:17:03,040 --> 00:17:15,380
And for all that, academic economists do all kinds of research looking at digital companies and increasing returns to scale and so on.

162
00:17:15,380 --> 00:17:19,870
And they're not used in policy economics. The applied tools are not there yet.

163
00:17:19,870 --> 00:17:21,640
And we, as a research community,

164
00:17:21,640 --> 00:17:29,920
haven't given competition authorities the way to think about how do I know whether Facebook is really past his best and somebody

165
00:17:29,920 --> 00:17:38,230
else is coming in with a better search engine that in five years time might have 98 percent of the market and six billion users?

166
00:17:38,230 --> 00:17:52,040
We don't know how to answer that question. And a lot of what we revert to is our instinct about how markets operate doesn't apply and these markets.

167
00:17:52,040 --> 00:18:01,460
So this is one of an example of a general purpose technology, and they fulfil this definition.

168
00:18:01,460 --> 00:18:08,240
They enable radical innovation in both products and services and in processes of production.

169
00:18:08,240 --> 00:18:13,250
And these apply across a wide range of activities in the economy. They start out in a particular sector.

170
00:18:13,250 --> 00:18:16,580
And before you know it, they're applying to a wide range.

171
00:18:16,580 --> 00:18:23,270
They lead to a major reorganisation of the structure of the economy because they change input costs dramatically.

172
00:18:23,270 --> 00:18:27,440
They also require a lot of additional investments. So it's the same kind of dynamic.

173
00:18:27,440 --> 00:18:33,020
They start out very slowly. And then eventually they start to have quite dramatic impacts.

174
00:18:33,020 --> 00:18:40,610
And part of the reason for the slow start is micro-scale is because they need a lot of investments in infrastructure and organisation,

175
00:18:40,610 --> 00:18:42,800
in the skills to operate them and so on.

176
00:18:42,800 --> 00:18:51,950
So they're slow to diffuse the when really well known example of this is a paper Paul David wrote in 1980 about the spread of computer technology,

177
00:18:51,950 --> 00:18:57,590
digital technology and its first wave, comparing it to the electric dynamo.

178
00:18:57,590 --> 00:19:04,610
And he pointed out that that innovation had required new kinds of factories to be built because you

179
00:19:04,610 --> 00:19:09,800
only get the productivity benefits of using electricity to power the factory rather than steam.

180
00:19:09,800 --> 00:19:16,550
If you have low, flat factories with assembly lines rather than the tall ones,

181
00:19:16,550 --> 00:19:22,460
with the shaft being driven by one steam engine driving lots of machines on different stories of the building,

182
00:19:22,460 --> 00:19:29,510
so whole building's had to come along, the grid had to be built. The because the factories were in greenfield sites.

183
00:19:29,510 --> 00:19:37,700
Ultimately, you ended up with so suburbanisation and the electrification of the economy and the creation of household electronic goods.

184
00:19:37,700 --> 00:19:43,940
So that had an effect on women's activities and their ability to go out and work and paid employment.

185
00:19:43,940 --> 00:19:51,740
So the ultimate impacts were substantial, but the impact on real GDP growth.

186
00:19:51,740 --> 00:19:55,580
First of all, to the long time shows through in productivity figures.

187
00:19:55,580 --> 00:20:05,210
And even then, it was actually quite a small number. And the economic history looking at the GDP impact of general purpose technologies

188
00:20:05,210 --> 00:20:11,000
shows that it's not large numbers in terms of the increment to growth,

189
00:20:11,000 --> 00:20:14,720
but they have very dramatic impacts. So this is doing this,

190
00:20:14,720 --> 00:20:21,860
and before long it's going to be really stupid to talk about the digital economy and the ways that we wouldn't talk about the electricity economy.

191
00:20:21,860 --> 00:20:29,670
It just is the economy and you get the slow and fast pace of change.

192
00:20:29,670 --> 00:20:32,310
How dramatic is it going to be? Well, the answer is very,

193
00:20:32,310 --> 00:20:42,480
and to see that you need to look just at the prices and the top left panel here shows you the price of computation calculated by William,

194
00:20:42,480 --> 00:20:46,950
not a very well-known paper, and it's a log scale.

195
00:20:46,950 --> 00:20:55,800
So it shows you the issues in Moore's law because of that dramatic fall that begins around 1940.

196
00:20:55,800 --> 00:21:04,080
Most famously, that you get a doubling of the computing power or halving of the price every every 18 months or two years.

197
00:21:04,080 --> 00:21:12,060
How does that translate into actual prices? People pay for things. The other charts come from some of my recent research with co-authors,

198
00:21:12,060 --> 00:21:19,440
and the top right hand one is the price that you pay for access to cloud computing facilities.

199
00:21:19,440 --> 00:21:26,700
And you can see that it's come down quite dramatically in this chart, which runs from 2010 up to last year.

200
00:21:26,700 --> 00:21:35,580
This is not a log scale, but if you look at the figures, something that started out at 40 cents has is now less than 10 cents.

201
00:21:35,580 --> 00:21:39,960
And what this means is that companies that used to invest in servers and hire

202
00:21:39,960 --> 00:21:45,180
people to operate those and have I.T. departments don't need to do that now.

203
00:21:45,180 --> 00:21:53,340
If they've got one already, they might still have kept it. But more and more companies, pretty much all start ups, don't invest in their own servers,

204
00:21:53,340 --> 00:22:01,650
and they use cloud computing companies like Amazon Web Services or Microsoft Azure or Google's Cloud Services.

205
00:22:01,650 --> 00:22:07,470
You might recognise these names from the earlier part of my talk this evening.

206
00:22:07,470 --> 00:22:16,380
So it's cheap to do, and I have talked to executives who say, Well, we used to have this department and it cost tens of thousands of pounds a year.

207
00:22:16,380 --> 00:22:21,900
But now, if I want to do something, I put it on the company credit card and it's just a few pounds.

208
00:22:21,900 --> 00:22:26,880
So it's a dramatic fall in price, an increase in capability,

209
00:22:26,880 --> 00:22:36,300
and it means that companies don't need to have skilled data scientists to use A.I. they can use algorithms that are provided by these cloud providers.

210
00:22:36,300 --> 00:22:44,160
The second example is the price of telecommunications services. And the darkness runs from 2010 to 2015.

211
00:22:44,160 --> 00:22:56,400
We've updated it a bit pattern hasn't changed. The top line The Blue Line shows the official price index for telecommunication services over

212
00:22:56,400 --> 00:23:04,650
a period when communications technology has changed dramatically speeds the compression,

213
00:23:04,650 --> 00:23:14,130
the latency and the amount of data that people are using and transmitting have fundamentally changed over that period.

214
00:23:14,130 --> 00:23:21,930
So telecoms engineers came to own us and my colleagues and said, What is going on here?

215
00:23:21,930 --> 00:23:32,490
Why is this index showing no decrease in the price? And one answer was that the index left out new data services.

216
00:23:32,490 --> 00:23:41,940
So we put those in and that gives you the red line. So rather than having zero decrease, you get a decrease for third in the price over five years.

217
00:23:41,940 --> 00:23:46,440
And the bottom line, the purple one, shows a thought experiment of saying,

218
00:23:46,440 --> 00:23:52,080
Suppose we take the revenues of the companies and we divide that by the volume of data in bits,

219
00:23:52,080 --> 00:23:58,860
and that shows you a 90 percent decline over that five year period, and that's gone down even further since.

220
00:23:58,860 --> 00:24:04,800
And this is pretty logical thing to do because there will be bits of data.

221
00:24:04,800 --> 00:24:11,190
The component services actually do have different prices per bit, but they're converging quite rapidly,

222
00:24:11,190 --> 00:24:15,330
and the more they converge, the closer the price index will get to the purple line.

223
00:24:15,330 --> 00:24:24,690
So here too. You've got communications services falling substantially in price about 90 percent over a five year period.

224
00:24:24,690 --> 00:24:34,170
And so when things are almost free or really cheap, people use them a lot and that's what companies have been doing.

225
00:24:34,170 --> 00:24:38,190
I might maybe next and there's a lot of hype about A.I.

226
00:24:38,190 --> 00:24:43,080
There are many economists. I'm one of them who argue it's the next general purpose technology.

227
00:24:43,080 --> 00:24:47,400
It's not just more digital. There's a lot of hype about it.

228
00:24:47,400 --> 00:24:51,600
And you've probably seen things like this on the internet,

229
00:24:51,600 --> 00:25:01,030
which show you that it's actually really hard for a computer to distinguish between a chihuahua and a blueberry muffin.

230
00:25:01,030 --> 00:25:05,230
Actually, it's not that funny, because it turns out that for some specific tasks,

231
00:25:05,230 --> 00:25:10,900
including image recognition, the air has been better than we have since 2015.

232
00:25:10,900 --> 00:25:16,960
So this is one of the images. This is the image that competition and by 2015.

233
00:25:16,960 --> 00:25:21,190
The error rate for the algorithm is much lower and the error rate for humans.

234
00:25:21,190 --> 00:25:27,160
So it turns out that we're not very good at the blueberry muffin chihuahua test either.

235
00:25:27,160 --> 00:25:32,140
In fact, you can get some quite sophisticated algorithms for free.

236
00:25:32,140 --> 00:25:39,940
And I found a load of adverts for a code to build your own robot dogs.

237
00:25:39,940 --> 00:25:48,130
So these are this is number eight, 11 12 in the catalogue. As a teenager near you making robots like this right now.

238
00:25:48,130 --> 00:25:52,450
The parts are cheap, the algorithms are free. You can put them together. They will be roaming the streets.

239
00:25:52,450 --> 00:26:00,490
Folks are before we know it. So there is a lot of hype. The abilities of A.I. currently are quite narrow.

240
00:26:00,490 --> 00:26:04,450
But it's also really cheap and it's going to change again.

241
00:26:04,450 --> 00:26:10,150
Change what people do in some ways. That's not all.

242
00:26:10,150 --> 00:26:20,110
If you look at other areas outside the immediate digital territory like drug discovery, prices are falling faster there.

243
00:26:20,110 --> 00:26:25,450
And this is a chart from the National Institutes of Health showing that the costs of

244
00:26:25,450 --> 00:26:32,110
sequencing the genome are falling faster than predicted the trajectory of Moore's law.

245
00:26:32,110 --> 00:26:36,850
So this is like digital, but even more so, and this is a logarithmic scale.

246
00:26:36,850 --> 00:26:45,640
So this is really dramatic change in the price. And in a sense, drug discovery is becoming an AI technology.

247
00:26:45,640 --> 00:26:53,820
A lot of it's being done by A.I. All of this is about the production side of the economy and how that changes with digital.

248
00:26:53,820 --> 00:27:00,340
What about the consumption side? What about us? How do we feel about it? Is this how we feel about it?

249
00:27:00,340 --> 00:27:05,530
And judging by popular literature and a lot of the news?

250
00:27:05,530 --> 00:27:07,390
Yeah, we're pretty upset about this.

251
00:27:07,390 --> 00:27:17,650
And there are all kinds of stories about aling bias and the use of algorithms in the criminal justice system leading to undesired outcomes.

252
00:27:17,650 --> 00:27:24,580
But, you know, all new technologies make people afraid. And this is a cartoon from the age of electricity.

253
00:27:24,580 --> 00:27:30,670
And it wasn't Frankenstein in a vacuum. People were generally pretty sure that electricity was going to kill you.

254
00:27:30,670 --> 00:27:36,910
So here's somebody in Victorian London caught up in the wires and killed by the light.

255
00:27:36,910 --> 00:27:46,600
And so there's a fear about new technologies. And once people become familiar enough, they just can't get enough of them.

256
00:27:46,600 --> 00:27:49,990
And there's some work that's been done by Eric Brynjolfsson and his colleagues,

257
00:27:49,990 --> 00:27:59,590
and we're repeating this now for the UK and testing different methodologies about how much people really, really love digital technology.

258
00:27:59,590 --> 00:28:09,760
These are estimates of the willingness to accept giving up Facebook for a certain period of time or search engines or email.

259
00:28:09,760 --> 00:28:18,370
So this is probably too small to read search engines willingness to accept giving up all search engines for a year in 2016,

260
00:28:18,370 --> 00:28:23,050
almost $15000 or email almost $6000.

261
00:28:23,050 --> 00:28:27,310
I mean, some of us might pay to give up email, but we're not the average and so on.

262
00:28:27,310 --> 00:28:30,340
So it doesn't say that you can ask these up.

263
00:28:30,340 --> 00:28:37,600
It's not clear how you aggregate these or what kinds of time budget constraints you need to think about, which is what we want to explore.

264
00:28:37,600 --> 00:28:42,100
But these are big numbers. And if you could add them up, you'd have, I don't know,

265
00:28:42,100 --> 00:28:52,000
two thirds of median US income as the value that people say that they place on these technologies, just reflecting the fact that we use them a lot.

266
00:28:52,000 --> 00:28:59,710
So I'm not switching to UK data, but this is Ofcom's tracker of how much time people spend online.

267
00:28:59,710 --> 00:29:04,420
It goes from 2007 to 2017, and we are.

268
00:29:04,420 --> 00:29:12,790
The question is how many hours of a typical week would you say you spend online at home, at your workplace or place of education or anywhere else?

269
00:29:12,790 --> 00:29:19,360
So it's everything. It's looking at Twitter. When you're on your mobile phone answering emails at work, it's everything,

270
00:29:19,360 --> 00:29:30,410
but it's 24 hours a week online, so we're using it a lot and that's doubled over that 10 year period.

271
00:29:30,410 --> 00:29:41,960
So we're doing a lot of things online now we are doing banking, travel agency, education, entertainment, social media communication,

272
00:29:41,960 --> 00:29:54,110
accessing information as a syndrome and lot of what we would default to assuming about consumption of goods and services doesn't hold either,

273
00:29:54,110 --> 00:29:59,090
because if you think about the basic economic theory that we have our indifference curves.

274
00:29:59,090 --> 00:30:03,110
I was just a local school safety and I'm told they don't teach in different schools anymore.

275
00:30:03,110 --> 00:30:10,850
It's a sign of how old I am, but we have this construct of utility and individuals have utility derive from their preferences.

276
00:30:10,850 --> 00:30:15,500
We have seen think of some fixed preferences that are known in some way as I know,

277
00:30:15,500 --> 00:30:19,970
or we can figure it out from food preferences, from their behaviour.

278
00:30:19,970 --> 00:30:25,280
So we've got the utility concept and we aggregate individual utilities to get social welfare in some way.

279
00:30:25,280 --> 00:30:32,210
And there's a whole huge literature about how you might do that aggregation, but is the utility constraint?

280
00:30:32,210 --> 00:30:41,200
I've never seen the utility. I don't know what one is. This is a constraint that doesn't really allow for new goods and big changes in behaviour.

281
00:30:41,200 --> 00:30:49,040
And there's an assumption that we know now will we knew in 2005 when our preferences for buying smartphones was going to be?

282
00:30:49,040 --> 00:30:55,820
And that's obviously untrue. It's also not the case that preferences are individual and not fixed,

283
00:30:55,820 --> 00:31:01,730
and they're not an individual either, as these guys would have told you in the 1950s.

284
00:31:01,730 --> 00:31:10,820
The whole advertising industry is based on the malleability of preferences and also social influence and preferences.

285
00:31:10,820 --> 00:31:16,670
So the social aspect isn't just that there are network effects for some of these technologies, such as talking about before,

286
00:31:16,670 --> 00:31:22,280
but that people like Jony Ive persuade us that we really want an iPhone as a job

287
00:31:22,280 --> 00:31:31,710
category of YouTube influences and social media have become absolutely pervasive.

288
00:31:31,710 --> 00:31:38,520
I think, anyway, economics is in a bit of a mess about how we think about choice, and on the one hand,

289
00:31:38,520 --> 00:31:50,640
we assume that people are better than the A.I., the algorithms because we are assumed to optimise in the same way in a reasoning way.

290
00:31:50,640 --> 00:31:57,540
You can restrict information search for sure, but there's a rational calculation in some form going on,

291
00:31:57,540 --> 00:32:02,070
and we're better than the robots because they can only do it in their narrow domains in which they're programmed.

292
00:32:02,070 --> 00:32:06,330
Whereas we operate in one domains and on the other hand, you've got the behave.

293
00:32:06,330 --> 00:32:12,720
That's the standard economic assumption, and the competing behavioural assumption in effect assumes that we're not as smart

294
00:32:12,720 --> 00:32:17,730
as pigeons because there's a whole branch of study biological markers theory,

295
00:32:17,730 --> 00:32:31,590
which finds that creatures from even fungi to pigeons to monkeys, sea creatures or any creatures in certain contexts behave like rational,

296
00:32:31,590 --> 00:32:37,800
calculating individuals optimising in the information environment to which they find themselves.

297
00:32:37,800 --> 00:32:43,260
And yet, the behavioural economics assumes that we don't do that in lots of domains,

298
00:32:43,260 --> 00:32:50,100
so we don't really understand in which domains to apply which kind of model.

299
00:32:50,100 --> 00:32:52,010
So on the production side,

300
00:32:52,010 --> 00:32:59,700
are defaults set of assumptions and hold on the consumption side or default set of assumptions is a bit of a mess in this digital economy,

301
00:32:59,700 --> 00:33:02,850
which is becoming increasingly pervasive.

302
00:33:02,850 --> 00:33:09,000
And there's a sense in which economists know that, of course, we know that the economy is not like the Phillips machine.

303
00:33:09,000 --> 00:33:14,370
This is the Cambridge one. They they kinda opened the COVID doors so I could photograph it for you.

304
00:33:14,370 --> 00:33:16,920
We know that the economy is not like that.

305
00:33:16,920 --> 00:33:27,810
We've put the machine in the cupboard, but we haven't updated the everyday mental furniture in a really important sense.

306
00:33:27,810 --> 00:33:37,080
And it's in the sense a lot of public policy economics is based in some way on the basic welfare theorems in economics,

307
00:33:37,080 --> 00:33:42,540
and they are these people irrational, self-interested, Typekit fixed preferences.

308
00:33:42,540 --> 00:33:48,480
They use all the available information. There are no increasing returns to scale.

309
00:33:48,480 --> 00:33:57,420
There are no externalities. There are complete markets, including for future goods, good survival, not non-renewable.

310
00:33:57,420 --> 00:34:03,740
And they are owned by a stranger, their property rights and transactions can be enforced.

311
00:34:03,740 --> 00:34:09,540
And so from what I've been saying about digital markets, the digital economy,

312
00:34:09,540 --> 00:34:13,140
maybe the Russian self-interested, we don't have fixed preferences, though.

313
00:34:13,140 --> 00:34:20,790
Maybe we use all available information, but we've got increasing returns to scale across a really wide swathe of the economy,

314
00:34:20,790 --> 00:34:26,610
loads of externalities even more than we were used to thinking about in the old economy.

315
00:34:26,610 --> 00:34:33,260
We don't have complete markets for future goods, and many digital goods are non-removable.

316
00:34:33,260 --> 00:34:34,670
You might even question the last one,

317
00:34:34,670 --> 00:34:42,110
if you think about the contestation of intellectual property and the way that the concept of property and physical markets has

318
00:34:42,110 --> 00:34:51,380
been transferred to intangible intellectual property in digital markets and the some of the strange consequences of that.

319
00:34:51,380 --> 00:34:59,630
Now some economists in the room might be twitching now and saying, We know this is nothing new in this, and in a sense, that's absolutely true.

320
00:34:59,630 --> 00:35:04,880
There are lots of people in academic economics who know that these assumptions don't

321
00:35:04,880 --> 00:35:10,670
hold and are researching digital markets and increasing returns to scale and so on.

322
00:35:10,670 --> 00:35:15,360
But in public policy. They're not operating on the academic front here.

323
00:35:15,360 --> 00:35:21,360
They're operating on what they were taught in their courses. A number of years ago.

324
00:35:21,360 --> 00:35:26,430
And that matters because there are lots of economists in government.

325
00:35:26,430 --> 00:35:33,210
This is a Treasury chart going up to 15, you can see that lovely upward trend in the number of economists employed in the government economic service.

326
00:35:33,210 --> 00:35:36,330
I've put in a Red Cross to represent the Brexit surge.

327
00:35:36,330 --> 00:35:46,440
It's up to about 18:00 now, and a lot of other people who are not economists and government has studied some economics and they have internalised.

328
00:35:46,440 --> 00:35:52,980
They've been socialised into thinking in that conventional way,

329
00:35:52,980 --> 00:36:02,730
which ultimately says the default is that markets work that quotes free markets are the default way to organise the economy.

330
00:36:02,730 --> 00:36:11,040
And if you spot next another team, you can address that in an individual way and devise a government policy to fix it.

331
00:36:11,040 --> 00:36:15,450
So my argument is that the interdependence in the economy,

332
00:36:15,450 --> 00:36:23,970
which has always been there to some degree but is getting much greater thanks to digital, means that that should not be the default assumption.

333
00:36:23,970 --> 00:36:32,640
And we should switch it around. Well, Elmo wrote his Ph.D. thesis got published in 1950, and he made this very point.

334
00:36:32,640 --> 00:36:37,020
And actually a lot of what I do involves reading things people wrote in 1940 or 1950 and going, you know,

335
00:36:37,020 --> 00:36:42,210
they've really had a point with that simplifying premise that these types of

336
00:36:42,210 --> 00:36:48,030
interdependence in supply and demand a negligible or non-existent is misleading.

337
00:36:48,030 --> 00:36:51,210
Such an assumption is not neutral, he said.

338
00:36:51,210 --> 00:37:00,300
And he made this very point that the assumption that the absence of interdependence points you towards the benchmark free market model.

339
00:37:00,300 --> 00:37:06,270
And that's circular reasoning because you've assumed interdependence and therefore

340
00:37:06,270 --> 00:37:12,090
you conclude that markets of people operating you've assumed independence.

341
00:37:12,090 --> 00:37:17,220
So you conclude that markets people operating independently are the way to organise the economy.

342
00:37:17,220 --> 00:37:23,490
If you think about the interdependence, you might come to different types of conclusion.

343
00:37:23,490 --> 00:37:29,340
So I just want to think a bit about what it would mean to take economic welfare

344
00:37:29,340 --> 00:37:35,250
seriously and ditch that framework that we're all taught and our courses.

345
00:37:35,250 --> 00:37:39,450
And there are some really good straws in the wind, I think.

346
00:37:39,450 --> 00:37:45,840
So the top left hand one is an example from a fantastic paper in the Oxford Review of Economic Policy.

347
00:37:45,840 --> 00:37:56,190
I can see at least one of the co-authors here in the room, an example of using market design to take account of interdependencies at the start.

348
00:37:56,190 --> 00:38:00,810
So you're using a choice mechanism, a market process.

349
00:38:00,810 --> 00:38:06,480
But you are right from the start taking account of externalities and interdependencies and people's behaviour.

350
00:38:06,480 --> 00:38:11,010
So that's a very exciting area of research.

351
00:38:11,010 --> 00:38:19,860
The top right hand example comes from Jim O'Neill, who was appointed by the government to look at the problem of antimicrobial resistance.

352
00:38:19,860 --> 00:38:25,530
The report came out a couple of years ago, and it was praised by the pharmaceutical industry,

353
00:38:25,530 --> 00:38:32,400
which has since made absolutely zero progress on developing new generation antibiotics.

354
00:38:32,400 --> 00:38:38,100
And so early this year, Jim O'Neill came out and said the market is not going to do this.

355
00:38:38,100 --> 00:38:45,180
We need a public provider who will produce this kind of pharmaceutical product and that is

356
00:38:45,180 --> 00:38:52,680
taking the high fixed costs and the non-life rivalry of many of these types of good seriously.

357
00:38:52,680 --> 00:38:59,700
So I would argue that once there is a public policy we should be thinking about very seriously is public provision of digital goods.

358
00:38:59,700 --> 00:39:05,040
Why would we bother to leave it to Google and Facebook? Why would you not think about some public provision?

359
00:39:05,040 --> 00:39:09,540
Why would you not think about nationalising Facebook if only it were ours?

360
00:39:09,540 --> 00:39:16,200
So the characteristics that I start out by describing actually make public ownership, which of course has lots of problems.

361
00:39:16,200 --> 00:39:22,740
Public provision a very sensible alternative to think about bottom left hand corner

362
00:39:22,740 --> 00:39:29,700
is an Austrian and homework was about non-state non-market governance structures,

363
00:39:29,700 --> 00:39:39,540
institutional structures for dealing with the production and allocation, and exactly the kinds of goods that I've been talking about.

364
00:39:39,540 --> 00:39:44,910
Where trust is, trust needs to be high, when the goods are known rival,

365
00:39:44,910 --> 00:39:52,080
whether interdependencies between people and it's not a free market, it's not Whitehall telling you what to do.

366
00:39:52,080 --> 00:39:57,390
Maybe you should be thinking about different models of economic institutions.

367
00:39:57,390 --> 00:40:03,150
And then finally, and I can't resist talking about measurement, which is what I do in my day job.

368
00:40:03,150 --> 00:40:12,690
Thinking about economic statistics and GDP seems to me fundamentally the wrong measure anyway,

369
00:40:12,690 --> 00:40:20,220
because it's a value added construct and anybody who's done economics will know that you add up all the production in the.

370
00:40:20,220 --> 00:40:24,120
On me, all of the income, all expenditure, but on the production side,

371
00:40:24,120 --> 00:40:31,170
you edit all up and you net out the intermediate stages of consumption because the argument is you double counting.

372
00:40:31,170 --> 00:40:38,220
You don't want to count both the flour going into the bread and the bread at the same time.

373
00:40:38,220 --> 00:40:43,770
You've got a net out the intermediate products. What this means is that in a sense,

374
00:40:43,770 --> 00:40:50,310
the whole process of disintermediation and breaking up the value chains has gone on over the past 20 years and is

375
00:40:50,310 --> 00:40:58,530
continuing apace with things like cloud computing is sort of invisible in the statistics when letting it all out.

376
00:40:58,530 --> 00:41:05,220
The structure of the economy has changed in terms of those extended production chains has changed profoundly.

377
00:41:05,220 --> 00:41:14,310
And actually, that's what Adam Smith said. It was the driving force of capitalism that you the greater the scope of your market people to specialise.

378
00:41:14,310 --> 00:41:19,230
The more narrow the specialisation, then the better off it would be.

379
00:41:19,230 --> 00:41:26,940
And we've kind of taken that out. And so I'm very interested in thinking about just entirely different ways to think about what's making us better.

380
00:41:26,940 --> 00:41:34,230
How do we take account of those really high values? People seem to be placing on things that they do have to pay for, and that in some sense,

381
00:41:34,230 --> 00:41:40,590
don't get captured in a very obvious way in the GDP and conventional statistics.

382
00:41:40,590 --> 00:41:46,830
There's lots of stuff we're not measuring about the economy. We're not measuring how much consumers use these digital services,

383
00:41:46,830 --> 00:41:52,500
the time they spend and maybe a new time use survey that covers that, but it's not there yet.

384
00:41:52,500 --> 00:41:56,970
We don't know how many people use digital platforms they know, but we don't.

385
00:41:56,970 --> 00:42:05,010
We don't know the extent to which companies are adopting cloud computing. We know if they use any, but not how much or what services they using.

386
00:42:05,010 --> 00:42:09,960
We don't know how much companies are using AI. We don't know how much data there is.

387
00:42:09,960 --> 00:42:14,040
There's no data on data. Talks about it being a fundamental economic asset.

388
00:42:14,040 --> 00:42:20,760
I to right at the start about the way it was a barrier to entry in digital markets,

389
00:42:20,760 --> 00:42:27,450
but we didn't know what there is and we don't know how to value it, even if we didn't know what there is in terms of volume, in terms of bits of data.

390
00:42:27,450 --> 00:42:34,470
So I'm starting a project trying to think about how would you categorise the valuation of different types of data set?

391
00:42:34,470 --> 00:42:37,680
We have no idea about cross-border flows.

392
00:42:37,680 --> 00:42:47,310
If a manufacturer in this country emails a blueprint to a contract manufacturer in Malaysia, we don't know what the value of that is.

393
00:42:47,310 --> 00:42:52,080
We don't know what they're doing. We don't know what the transfer pricing is. We don't know how much data is crossing borders.

394
00:42:52,080 --> 00:42:56,220
If a company here uses a cloud computing service,

395
00:42:56,220 --> 00:43:00,540
we don't know if there's any export or import involved because we don't know we think the century goes to.

396
00:43:00,540 --> 00:43:06,750
It could be here in the UK or it could be in Belgium. We don't know what prices people are paying for things.

397
00:43:06,750 --> 00:43:10,020
The price of a digital camera is still recorded.

398
00:43:10,020 --> 00:43:14,910
Its weight in the consumer price index has been going down because people aren't buying them very much anymore.

399
00:43:14,910 --> 00:43:20,310
But nowhere are we putting the zero price that we're all paying for taking photographs and looking at them on our smartphone.

400
00:43:20,310 --> 00:43:28,020
So the price indices that we use to calculate real GDP and real productivity are completely wrong.

401
00:43:28,020 --> 00:43:30,000
We're not even doing some of the basics,

402
00:43:30,000 --> 00:43:44,460
and I thought I'd try to look up a trade in robots because robots are becoming thing and I found in the tariff schedule actually as a heading.

403
00:43:44,460 --> 00:43:47,700
Industrial robots, not elsewhere, specified or included.

404
00:43:47,700 --> 00:43:53,640
So if you're a car manufacturer and you've got a robot arm, you might describe it as a machine tool for making cars.

405
00:43:53,640 --> 00:43:58,140
If it's not specified anywhere else, you will put it in this category of industrial robots.

406
00:43:58,140 --> 00:44:07,560
We also count toy robots really carefully. We've got really good statistics on these little guys in the trade figures.

407
00:44:07,560 --> 00:44:13,410
And the reason is because non-human figures have different tariff rate than human figures.

408
00:44:13,410 --> 00:44:20,250
And it's been changed. That shows if you know the story that used to be human figures for toys versus animal figures for toys,

409
00:44:20,250 --> 00:44:25,980
and it was a famous court case fought over Star Wars characters and whether they were humans or animals.

410
00:44:25,980 --> 00:44:32,520
So the category is now being extended. For example, robots, devils or angels or aliens should be in that as well.

411
00:44:32,520 --> 00:44:40,140
So we don't have some quite basic stuff. So what's the new agenda for economics?

412
00:44:40,140 --> 00:44:45,960
We need some badly, need some policy and empirical approaches to these issues.

413
00:44:45,960 --> 00:44:52,230
Theoretically, the people research them, but the competition authority doesn't know how to do it.

414
00:44:52,230 --> 00:44:57,360
Somebody needs to start doing that applied work and figuring out, you know,

415
00:44:57,360 --> 00:45:04,250
what is the potential harm to consumers of a new search engine not being able to get into the market?

416
00:45:04,250 --> 00:45:07,710
And we need to think a lot about the public provision.

417
00:45:07,710 --> 00:45:16,650
As I was arguing an alternative the regulation of information goods, the public good and non rival the market will under provide them.

418
00:45:16,650 --> 00:45:20,150
How are we going to regulate this? How are we going to think about regulating and.

419
00:45:20,150 --> 00:45:26,780
Lateral property better. We need to think more about social capital and governance.

420
00:45:26,780 --> 00:45:35,360
As I was asking about Alan Ostrom was the right business models and governance and governance mechanisms for these types of companies.

421
00:45:35,360 --> 00:45:43,130
What do you think about the dynamics and what gets the economy onto one kind of dynamic rather than another?

422
00:45:43,130 --> 00:45:48,920
Both Shiller has an interesting book out this autumn called Narrative Economics, and it's exactly about this point.

423
00:45:48,920 --> 00:45:56,960
If you've got self-fulfilling phenomena or averting phenomena, what is it that tips you from one to the other determines which path you get onto?

424
00:45:56,960 --> 00:46:05,300
And the argument is, the narrative is part of the story. You've got to align people's behaviour, and the narrative is what helps to do that.

425
00:46:05,300 --> 00:46:11,450
And I think you really fundamentally in my mind about what is it that makes people better.

426
00:46:11,450 --> 00:46:16,970
We've got this wealth economic construct that actually we've known for a long time isn't fit for purpose and

427
00:46:16,970 --> 00:46:22,580
less and less so with the economic characteristics that are spreading across the economy through digital.

428
00:46:22,580 --> 00:46:30,660
So that would be my economic research agenda. So to sum up, we've got an economics that is all the things on the left, linear static,

429
00:46:30,660 --> 00:46:35,760
constant returns, et cetera, and an economy that's exactly the opposite.

430
00:46:35,760 --> 00:46:43,080
We've got a construct of economics, the point you towards the market thinking about the market as your default, your benchmark.

431
00:46:43,080 --> 00:46:48,330
Whereas we should be thinking about institutional structures as our benchmarks,

432
00:46:48,330 --> 00:46:56,850
whether that be market design or Ostrom style institutions or public provision, let's think seriously about institutional structures.

433
00:46:56,850 --> 00:47:05,310
This involves some changing of methods as well, and economists who hear about big data will always reply We we do.

434
00:47:05,310 --> 00:47:11,250
Data is what we do, and we might have not have millions and millions of records, but we've had a very large datasets.

435
00:47:11,250 --> 00:47:19,140
But there were new data sources. The digital economy itself is a source of data is quite hard to access the data generating process.

436
00:47:19,140 --> 00:47:24,930
The statistical properties aren't always well understood. A lot of it's held in these companies silos.

437
00:47:24,930 --> 00:47:30,030
We need to figure out how to get that through the top for the public good because there's public

438
00:47:30,030 --> 00:47:35,160
value being left on the table by the fact that these silos are completely separate from each other.

439
00:47:35,160 --> 00:47:42,090
I want to be able to join up my smartwatch and my shopping data online GDP data for my own purposes.

440
00:47:42,090 --> 00:47:48,000
I want something to find an app for me that does that matching and tells me how I can make myself better off.

441
00:47:48,000 --> 00:47:55,170
We need computational approaches. You can't do a lot of this and analytically the way that we've been used to in economics.

442
00:47:55,170 --> 00:48:01,470
So to give one example, we have people maximise their utility subject to budget constraint of money.

443
00:48:01,470 --> 00:48:07,710
If you want to add a budget constraint of time, which is key with these digital goods, it becomes algebraically intractable.

444
00:48:07,710 --> 00:48:14,400
You've got to do some other way when you reintroduce more economic history because we are in a technological

445
00:48:14,400 --> 00:48:21,600
revolution and looking back at past ones is a good way to think about how to approach the one we're going through.

446
00:48:21,600 --> 00:48:23,190
They'll be huge differences,

447
00:48:23,190 --> 00:48:32,010
but it points you towards thinking about those much more distant changes that I was talking about with the interesting example.

448
00:48:32,010 --> 00:48:38,790
I think narratives is an idea of really worth taking seriously. We have very quantitative and economics.

449
00:48:38,790 --> 00:48:42,420
I think we need to recognise the merits of qualitative methods.

450
00:48:42,420 --> 00:48:48,270
Going to talk to companies is absolutely illuminating and you find out lots of things about how they operate,

451
00:48:48,270 --> 00:48:53,430
and they've got no idea what their marginal costs are and don't know what you're talking about.

452
00:48:53,430 --> 00:48:57,720
But above all, I think it's a new mindset. It's a new default that we go to.

453
00:48:57,720 --> 00:49:01,650
And in this room, we can pat ourselves on the back if they're already thinking about this.

454
00:49:01,650 --> 00:49:05,200
But for the policy world, this is quite a big change in mindset.

455
00:49:05,200 --> 00:49:14,490
Still, too many economists in in government who who default to the market, as I said, I've been writing about this for a long time.

456
00:49:14,490 --> 00:49:18,900
This is just the advertising interlude at the end.

457
00:49:18,900 --> 00:49:26,010
So the way this world came out in 1997, and it's free on my website if anybody wants to read it and a lot of these issues,

458
00:49:26,010 --> 00:49:32,160
I think, were clear already back then the measurement issues, the sustainability issues I haven't really talked about.

459
00:49:32,160 --> 00:49:41,340
And a new book coming out early in 2020, which tries to apply some of the things I've been talking about this evening to public policy.

460
00:49:41,340 --> 00:49:48,120
And hopefully by the time we get to the next generation of students having their jobs and government, things will be a little bit different.

461
00:49:48,120 --> 00:50:02,100
Probably going a bit over my time. So I'm going to stop now. Thank you all very much. I think we've got a bit of time for Q&A.

462
00:50:02,100 --> 00:50:08,340
Just to remind you before questions, this is being videos I do remember when you ask a question that is going out live,

463
00:50:08,340 --> 00:50:14,080
who would like to ask the first question? Question there.

464
00:50:14,080 --> 00:50:21,040
Thank you so much for a great talk and for really focussing on an area that I think is we're going through such a transition.

465
00:50:21,040 --> 00:50:24,760
And I know in hindsight we'll look back and go, Wow, I can't believe we went through that,

466
00:50:24,760 --> 00:50:28,870
but you know, we didn't feel it while we were going through it. That's true generally, anyway.

467
00:50:28,870 --> 00:50:32,290
So we're moving from a tangible world to an intangible world.

468
00:50:32,290 --> 00:50:39,040
And one of the things that strikes me is economics is a huge thing and whether capitalism works and how it works within that,

469
00:50:39,040 --> 00:50:47,440
but it interacts with all these other things like laws. You're talking about IP laws and competition laws and things, taxation, right?

470
00:50:47,440 --> 00:50:52,690
And how we tax intangible assets versus tangible pensions, education systems,

471
00:50:52,690 --> 00:50:58,000
health care systems, they're all systems and they all interact and we can fix one.

472
00:50:58,000 --> 00:51:03,610
But if we do it in isolation of the others, we're going to have an asymmetry and it's not going to work.

473
00:51:03,610 --> 00:51:09,670
So how do you embed this into this greater kind of multi system?

474
00:51:09,670 --> 00:51:14,650
My goodness, if I had a quick answer to that, I'd be making a fortune from advising governments about it.

475
00:51:14,650 --> 00:51:16,300
It's obviously really hard.

476
00:51:16,300 --> 00:51:23,820
Any changing one system is hard, and we're in a context where we've got an overlapping set of systems, including the environmental system.

477
00:51:23,820 --> 00:51:30,070
And so one answer is that if you look back at the historical examples of what happens

478
00:51:30,070 --> 00:51:36,190
of conflict and period of chaos and contestation and quite a political conflict,

479
00:51:36,190 --> 00:51:41,500
and then eventually new arrangements and systems emerge out of that.

480
00:51:41,500 --> 00:51:47,500
The worry about apart from the discomfort of waiting for that this time is that some of

481
00:51:47,500 --> 00:51:52,550
those systems seem to be approaching catastrophic for the environment in particular.

482
00:51:52,550 --> 00:51:59,270
So you want to change the energy system pretty pronto. And I just don't know how you do it.

483
00:51:59,270 --> 00:52:02,950
It's really hard in government. We don't have government structures that enable it.

484
00:52:02,950 --> 00:52:06,100
When we did our report and the competition aspect of digital economy,

485
00:52:06,100 --> 00:52:16,240
we recommended a new regulatory body that would have particular oversight over companies deemed to have strategic market power at the same time.

486
00:52:16,240 --> 00:52:22,210
Frances Cairncross did a report on the news industry and the way it's been hollowed out by digital.

487
00:52:22,210 --> 00:52:27,640
And she recommended a new regulator. Ofcom already has some responsibilities.

488
00:52:27,640 --> 00:52:35,800
Competition and Markets Authority already has some responsibilities. DCMS put out a white paper saying, We need a regulator for online harms.

489
00:52:35,800 --> 00:52:43,090
You've already got six digital regulators there, and it's not at all obvious to me how Whitehall,

490
00:52:43,090 --> 00:52:48,250
never mind the Conservative Party that's in government at the moment, Will will settle on them.

491
00:52:48,250 --> 00:52:55,300
How will they arrange that and actually what is the vital management for that? So part of the question isn't about economics at all.

492
00:52:55,300 --> 00:53:02,680
It's about the political science fundamentally of what's the right level of governance for certain decisions and how do you do that?

493
00:53:02,680 --> 00:53:04,780
Joining up that we are profoundly bad at doing.

494
00:53:04,780 --> 00:53:09,430
We have government silos and I don't know how you bring this on is you've got to change the budget lines.

495
00:53:09,430 --> 00:53:13,480
It's tough. Thank you. Good, thank you, John.

496
00:53:13,480 --> 00:53:18,480
I've got a couple of questions that relate to the point that so much of what we buy now is free.

497
00:53:18,480 --> 00:53:22,450
Okay. The first one is what is the inflation rate generally in this environment?

498
00:53:22,450 --> 00:53:28,590
I know you've asked before. What does it mean for central bank policy? Should central banks be sending negative interest rates?

499
00:53:28,590 --> 00:53:32,250
But clearly, if they do that, the whole financial system doesn't work.

500
00:53:32,250 --> 00:53:37,590
On the other question relates to a similar theme, but basically, I saw some estimates coming out of the U.S.,

501
00:53:37,590 --> 00:53:46,650
which said that if you try and value the internet and you put a user charge on it, the value of the internet is something like $500 billion.

502
00:53:46,650 --> 00:53:52,170
And to put that in context, world GDP is 80 on the value of the entire energy industry.

503
00:53:52,170 --> 00:53:56,610
20. So if that's true, surely geopolitics are going to come in here.

504
00:53:56,610 --> 00:54:01,440
Governments are going to try and grab this for taxpayers or to do something with it.

505
00:54:01,440 --> 00:54:04,650
Well, my government would love to start taxing some of this,

506
00:54:04,650 --> 00:54:13,350
but digital companies are just as good as any other multinational and moving their profits to places where they don't get taxed.

507
00:54:13,350 --> 00:54:18,420
So I'm not a macro economist and I'm no expert on what the Bank of England ought to be doing.

508
00:54:18,420 --> 00:54:21,930
We still have money. You can still count nominal GDP.

509
00:54:21,930 --> 00:54:31,590
And I think part of the error is to think not only that, you can calculate real GDP in a meaningful way given your point about price indices,

510
00:54:31,590 --> 00:54:36,210
but that you can then compare that to potential real GDP.

511
00:54:36,210 --> 00:54:43,560
I don't think we've got any idea really what what that is. For all of these intangible sectors.

512
00:54:43,560 --> 00:54:44,580
So if I were them,

513
00:54:44,580 --> 00:54:52,680
I would be looking at nominal GDP and actually I would be raising interest rates because the banking sector isn't functioning with low interest rates,

514
00:54:52,680 --> 00:54:59,670
never mind negative ones. And it point seatbelts, not using monetary policy as a means of demand management, I think at the moment.

515
00:54:59,670 --> 00:55:05,480
That was a huge caveat. That is probably 30 years since I've done any macroeconomics.

516
00:55:05,480 --> 00:55:18,300
Eric, terrific talk. Diane, as one of our leading experts on metrics, I'd like to ask you about a metric that policymakers obsess over productivity.

517
00:55:18,300 --> 00:55:25,560
You know, it has a clear meaning and is useful concept in a world of goods that you can draw upon your foot in this world.

518
00:55:25,560 --> 00:55:31,260
What does it mean? Does it mean anything? Do we need a new concept? We need a new way of thinking about that whole notion.

519
00:55:31,260 --> 00:55:38,010
I think you're right. We need a new concept, and the one that I've been playing with is about time.

520
00:55:38,010 --> 00:55:46,020
Because in a service based economy, never mind digital and intangible and a service based economy, you either want something to happen really quickly.

521
00:55:46,020 --> 00:55:50,940
You want your train to take you as fast as possible on the journey.

522
00:55:50,940 --> 00:55:56,040
You've got to have a blood test done. You want that to be done as quickly as possible or you want the opposite.

523
00:55:56,040 --> 00:56:00,300
You want it to be as long as it needs to be in really high quality.

524
00:56:00,300 --> 00:56:07,800
So if you're in the ICU, you want to have the devoted attention of a skilled nurse for 24 hours a day.

525
00:56:07,800 --> 00:56:11,670
So those would be examples of productivity. Not at all. Not at all.

526
00:56:11,670 --> 00:56:17,880
Captured by the current productivity metrics and services. So I think the concept needs to change and even physical goods,

527
00:56:17,880 --> 00:56:25,310
the quality change that gets embedded in those now means that it's very hard to think about what productivity figures need.

528
00:56:25,310 --> 00:56:33,220
Thank you. Over there, David.

529
00:56:33,220 --> 00:56:40,630
Hi, Dan, you managed to do the whole top without mentioning the greatest externality facing us, which obviously we need to think about pricing.

530
00:56:40,630 --> 00:56:43,780
We need to think about how we deal with it. It's a global problem.

531
00:56:43,780 --> 00:56:51,710
It's got huge numbers of and then I'd thought about consequences that would be happening over time.

532
00:56:51,710 --> 00:57:00,080
I mean, it's very hard to think that these are the key problems, given that we've got a much bigger problem overlying it.

533
00:57:00,080 --> 00:57:09,020
And I do wonder about how you would think about integrating climate change into your kind of analysis here.

534
00:57:09,020 --> 00:57:13,580
It's a good challenge, and I think about it in a separate mental bucket is the honest answer.

535
00:57:13,580 --> 00:57:20,120
So as well as trying to measure digital and GDP, I try to think about the balance sheet of the economy,

536
00:57:20,120 --> 00:57:24,530
including natural assets and the health of the climate.

537
00:57:24,530 --> 00:57:29,420
And so we have a programme looking at the moment of natural capital and social capital.

538
00:57:29,420 --> 00:57:35,810
And you know, if you don't have a balance sheet in some broad sense, you're not thinking about sustainability.

539
00:57:35,810 --> 00:57:39,020
It doesn't answer the detail of your question in any way.

540
00:57:39,020 --> 00:57:46,730
But if you think about it in terms of metrics, then thinking about assets seems to be one way to go about it.

541
00:57:46,730 --> 00:57:50,720
But you know, as you note, people in the room are much more expert than I am on this.

542
00:57:50,720 --> 00:57:59,660
A lot of thought has been going into thinking about policy approaches to solving the climate change problem.

543
00:57:59,660 --> 00:58:03,860
And it's not really an economic question, it seems to me now it's the political, again,

544
00:58:03,860 --> 00:58:11,480
a political science question about why it's so hard for any of those policies to gain traction.

545
00:58:11,480 --> 00:58:18,620
And the one that looks most appealing to every politician I've ever spoken to is that we find some wonderful new technology that fixes it.

546
00:58:18,620 --> 00:58:23,600
Don, one of the things I didn't say in my introduction is that you've been a member of the BBC Trust.

547
00:58:23,600 --> 00:58:29,930
Does the argument about the public provision of information and some of the digital goods?

548
00:58:29,930 --> 00:58:36,440
Does it have traction when applied to the BBC in policy circles in this country?

549
00:58:36,440 --> 00:58:41,060
Yes. And that's one of the experiences that informs my thinking about this and whether we

550
00:58:41,060 --> 00:58:46,820
should have a public service search engine or public service social media company.

551
00:58:46,820 --> 00:58:52,640
Because although the BBC has been attacked by free market politicians and funding has been cut,

552
00:58:52,640 --> 00:58:59,570
so that isn't the same solid consensus about it as I would have been 30 years ago.

553
00:58:59,570 --> 00:59:04,880
It's very popular, it's very high quality. Its services are really widely used.

554
00:59:04,880 --> 00:59:10,340
And I spent my eight years on the BBC Trust being concerned about its impact on the commercial market,

555
00:59:10,340 --> 00:59:15,920
and it's always been very careful not to foreclose markets to commercial rivals.

556
00:59:15,920 --> 00:59:20,840
But we've got a really healthy media market with a really big public player in it,

557
00:59:20,840 --> 00:59:29,330
and I think it means that the competition hasn't been about large number of viewers and the lowest common denominator programming.

558
00:59:29,330 --> 00:59:34,130
There's been competition about quality as well. And so an about innovation.

559
00:59:34,130 --> 00:59:39,860
We are one of the few net exporters of popular and classical music around the world.

560
00:59:39,860 --> 00:59:48,500
Sweden is one of the US is the other. And that's partly because we've got a public service broadcaster with a mission to encourage new material,

561
00:59:48,500 --> 00:59:52,970
new creation inside this country so that it gets done and then it gets exported.

562
00:59:52,970 --> 00:59:56,220
So I think it's a great parallel, actually. Thank you.

563
00:59:56,220 --> 01:00:02,390
Question over there. Yeah. You talked a lot about costs going down with these new technologies.

564
01:00:02,390 --> 01:00:06,860
My sense is that if you think of ICTs overall,

565
01:00:06,860 --> 01:00:12,650
then the household budget or individual budgets have just been steadily growing

566
01:00:12,650 --> 01:00:18,590
in terms of the amount of money that these services costs to individuals.

567
01:00:18,590 --> 01:00:23,870
And you also talked about kind of how we change this or how what kind of choices people make.

568
01:00:23,870 --> 01:00:28,880
It seems to me that that is not going to change. I mean, it's not going to increase vastly,

569
01:00:28,880 --> 01:00:34,460
but driving down the costs for these large parts of a household or individual

570
01:00:34,460 --> 01:00:38,750
budgets that have kind of grown over time isn't going to change anytime quickly,

571
01:00:38,750 --> 01:00:44,930
even if it's not going to go up very much. I think there are two things going on.

572
01:00:44,930 --> 01:00:50,540
One is that the price households pay per byte data.

573
01:00:50,540 --> 01:00:58,490
Has been going down substantially, but the amount of the budget that they spend on information goods has been rising.

574
01:00:58,490 --> 01:01:03,140
And that's the kind of shift in consumption patterns that we've seen over and over again over time.

575
01:01:03,140 --> 01:01:05,150
And you know, when people have low incomes,

576
01:01:05,150 --> 01:01:11,540
they spend more on food and shelter and that diversifies into other kinds of goods and then into other kinds of services.

577
01:01:11,540 --> 01:01:16,170
And so there's no reason to expect that that share will go down.

578
01:01:16,170 --> 01:01:21,640
But I think that's different from the price per byte of information, if you like.

579
01:01:21,640 --> 01:01:28,490
Distinguish between two of them. Thank you. Other any further questions, one,

580
01:01:28,490 --> 01:01:36,570
they can you can you make some comments on how this relates to our thinking about inequality in the sense that you know,

581
01:01:36,570 --> 01:01:38,750
the way a lot of this technology is working?

582
01:01:38,750 --> 01:01:45,900
On one hand, you're getting ever more billionaires and millionaires and a lot of the sort of wealth going towards them.

583
01:01:45,900 --> 01:01:51,650
And and but on the other hand, lots of poor people are getting access to lots of technology very, very cheaply.

584
01:01:51,650 --> 01:02:00,560
How do we sort of think about global inequality and the impact of all this discussion on the inequality we care about you?

585
01:02:00,560 --> 01:02:06,640
So, so clearly, these are both going on at the same time, access to goods plus plus and inequality.

586
01:02:06,640 --> 01:02:12,130
And so the short answer is it's complicated.

587
01:02:12,130 --> 01:02:16,210
One is that there are genuine things like superstar effects.

588
01:02:16,210 --> 01:02:21,340
So for some individuals, those will operate and make them very well off.

589
01:02:21,340 --> 01:02:34,070
That is also. A sort of locational phenomenon in that the digital economy is enhancing forces of geographic agglomeration into big cities.

590
01:02:34,070 --> 01:02:41,510
So productivity and incomes are higher in big cities, and that seems to be accelerating, at least if the US data is anything to go by.

591
01:02:41,510 --> 01:02:46,700
So that's happening. There's also the thing that we often overlook in the economic discussion,

592
01:02:46,700 --> 01:02:50,810
which is the politics of it and a lot of the inequalities is driven by politics,

593
01:02:50,810 --> 01:03:02,720
by tax cuts for the wealthy, by bashing the unions and weakening labour market power by all kinds of institutional structures that can be changed.

594
01:03:02,720 --> 01:03:07,730
And I have a Ph.D. student who's looking at exactly this issue about the difference in the incomes

595
01:03:07,730 --> 01:03:13,880
of workers who were in industrial jobs got downsized and how that compares to green countries.

596
01:03:13,880 --> 01:03:15,230
And it's very different.

597
01:03:15,230 --> 01:03:21,680
So we have in our heads that the US model, there's something kind of deterministic about what the technology and trade effects are doing.

598
01:03:21,680 --> 01:03:26,300
It gets refracted through politics as well, and that's a big part of the story, I think.

599
01:03:26,300 --> 01:03:36,950
Thank you, gentlemen in a blue jumper. You mentioned the importance or perhaps the new importance of economic history.

600
01:03:36,950 --> 01:03:45,170
I wonder whether there's any parallel or you can draw any between the the way in which public policy responded at the various stages of,

601
01:03:45,170 --> 01:03:54,770
say, the industrial revolution. I'm thinking, say, canals, speed, electricity, late 19th century when there's a lot of disarray about all of that,

602
01:03:54,770 --> 01:04:02,860
as you as you said, how quickly did public policy respond to that?

603
01:04:02,860 --> 01:04:11,830
It depends which bit you're talking about, and some of it was quite variable and some areas of deregulation,

604
01:04:11,830 --> 01:04:16,390
financial deregulation, financial changes and regulation seem to happen relatively quickly.

605
01:04:16,390 --> 01:04:21,760
Things like the limited liability company came in relatively early in the period.

606
01:04:21,760 --> 01:04:32,410
Things like safety, regulation or standards, which involve coordinating a lot of people in places much more slowly.

607
01:04:32,410 --> 01:04:40,090
The one interesting panel is the data collection, and it took a long time for the government to start collecting data on the

608
01:04:40,090 --> 01:04:45,790
industrial economy and for a very long period during the industrial revolution,

609
01:04:45,790 --> 01:04:51,650
there were fantastic agricultural statistics. But all of the information people had about the industrial economy child labour were

610
01:04:51,650 --> 01:04:55,930
number of mines and so on came through blue book reports commissioned by parliaments.

611
01:04:55,930 --> 01:05:03,160
They were sort of one-off reports on what was happening. So even for policymakers to understand what's going on by having statistics is pretty.

612
01:05:03,160 --> 01:05:11,890
It's pretty slow. The categories people have categories in, they had to collect the data and new technologies, often dangerous to start with.

613
01:05:11,890 --> 01:05:17,530
I showed the cartoons for electricity, but it was dangerous. Air travel was dangerous to start with.

614
01:05:17,530 --> 01:05:24,490
And I think an interesting question now is whether a tolerance for any kind of danger has reduced and whether autonomous vehicles will ever

615
01:05:24,490 --> 01:05:34,780
really happen outside quite tightly due fenced areas because we don't have a tolerance for machines killing people in the way that we used to.

616
01:05:34,780 --> 01:05:48,910
Thank you. That, Sarah. So thank you, Diana, for just a wonderful talk.

617
01:05:48,910 --> 01:05:56,380
Your slide of the 1500 civil servants in Britain that learnt and old economics,

618
01:05:56,380 --> 01:06:03,670
which you think can't address these, these problems, perhaps let alone the problems that David would add.

619
01:06:03,670 --> 01:06:07,030
How do you address that? I mean, obviously, they should all buy your book and read it.

620
01:06:07,030 --> 01:06:18,670
But but seriously, you you know you're Moore's law slides have their equivalent for the need for policy innovation to deal with this.

621
01:06:18,670 --> 01:06:26,590
So any ideas on on what we do with government? I think the training need isn't so much my new book or that.

622
01:06:26,590 --> 01:06:30,820
Of course, I hope we'll buy it as understanding the technologies that they're dealing with.

623
01:06:30,820 --> 01:06:37,000
And there's not a lot of expertise in even regulators on these technologies,

624
01:06:37,000 --> 01:06:43,180
and partly because the digital companies themselves buy so much of the expertise.

625
01:06:43,180 --> 01:06:47,530
I think it's about the climate of ideas more than targeting the individuals.

626
01:06:47,530 --> 01:06:55,540
And we've had a long period where a climate of ideas has been dominated by a free market benchmark.

627
01:06:55,540 --> 01:07:05,500
It was actually a process encouraged by certain economists and think tanks and embedded by the Thatcher and Reagan revolutions.

628
01:07:05,500 --> 01:07:09,340
And that has to change, I think, and I think it is, I see all kinds of straws in the wind.

629
01:07:09,340 --> 01:07:18,970
Maybe I'm a deluded optimist, but I think that is changing and that will cause all those policymakers to through social influence,

630
01:07:18,970 --> 01:07:25,050
change the way they think as well and reflect differently on the way they approach policy issues.

631
01:07:25,050 --> 01:07:30,990
Thank you. Before asking you to thank done again. Two announcements, first of all.

632
01:07:30,990 --> 01:07:35,130
Please join us for a glass of wine next door. In a moment.

633
01:07:35,130 --> 01:07:38,010
And also just a bit of advertising.

634
01:07:38,010 --> 01:07:46,920
Next week, we have a double act Colin Meyer and Paul Collier, both from Oxford, talking about the future of the corporation, economy and society.

635
01:07:46,920 --> 01:07:51,810
And the following week we have Eric Fine, hooker Eric. I don't know the title of your talk.

636
01:07:51,810 --> 01:08:02,100
Do you yet have new economic and moral foundations for the new moral and economic foundations for the Anthropocene?

637
01:08:02,100 --> 01:08:06,750
So please come along to both of them, but please join me now for thanking God.

638
01:08:06,750 --> 01:08:21,201
Thank you for the great.