CORRECTED TRANSCRIPT OF ORAL EVIDENCE To be published as HC 784-i

HOUSE OF COMMONS

HOUSE OF LORDS

ORAL EVIDENCE

TAKEN BEFORE THE

PARLIAMENTARY COMMITTEE ON BANKING STANDARDS

SUB-COMMITTEE G-PANEL ON THE OPERATION OF WHOLESALE MARKETS

COMPUTER-BASED TRADING

MONDAY 26 NOVEMBER 2012

PROFESSOR DAVE CLIFF, PROFESSOR OLIVER LINTON and DR JEAN-PIERRE ZIGRAND

ALEXANDER JUSTHAM

USE OF THE TRANSCRIPT

Oral Evidence

Taken before the Parliamentary Commission on Banking Standards

Sub-Committee G-Panel on the operation of wholesale markets

on Monday 26 November 2012

Members present:

The Lord Bishop of Durham

Mark Garnier

Examination of Witnesses

Witnesses: Professor Dave Cliff, Dr Jean-Pierre Zigrand and Professor Oliver Linton FBA, lead experts on Foresight report on computer-based trading in financial markets, gave evidence.

Q1 Chair: Thank you very much for coming to speak to us today. It is very much appreciated. Mark Garnier and I are on this panel, and a report will be written for the full panel of the Banking Standards Commission. It is fair to say that although Mark has considerable background experience in this area and I have a lot of background experience in markets, high-frequency trading, dark pools and the algorithmic approach are relatively new to me, so I hope you will excuse some fairly dumb questions as we go through the evidence.

You were all on the lead expert group tasked with overseeing the Foresight project on computer-based trading. How did that project come about, and what were the perceived issues you were seeking to address?

Professor Cliff: In late 2009, I was in discussion with Sandy Thomas, head of Foresight, whom I had by then known for two or three years. I run a big research project in my job at the university, which looks at systemic risk and the resilience of large-scale software-intensive systems. I happened to mention that I had also worked for Deutsche Bank in the City on automated trading. Foresight was coming to the end of the cycle on one of its projects, and was looking for a new project to pursue, so I wrote a one-page summary of a project that might take place exploring the future of computer-based trading in the financial markets. John Beddington read it, and I then had a meeting with him where he asked me expand it. Over a number of weeks, it was expanded, and John sent it out to various other people to read and comment on. I finished the proposal in its final form on 22 April 2010, and less than two weeks later, the Flash crash occurred.

In my proposal, I did not actually say it would be 6 May and it would be New York, but mention was made of the fact that computer-based trading might have given rise to instabilities that could lead to market dynamics that were unattractive, and after the Flash crash-

Q2 Mark Garnier: They felt that that was unattractive.

Professor Cliff: Yes.

Dr Zigrand: That was combined with the fact that the MiFID II proposal was coming up, and the new infrastructure would be decided in Brussels and Strasbourg, so there was an additional reason to have discourse at that time to influence and help to guide practitioners and policy makers.

Q3 Chair: The key thing for me is that it started before the Flash crash. It was not in response to a particular event.

Professor Cliff: No.

Q4 Chair: In your report, which is commendably even-handed-it is very difficult to tell what you really think-you identify both the benefits of and the concerns about computer-based trading. I suppose because I started trading in the markets some 30 years ago, I was slightly questioning what value this provides beyond value within the lives of the computer-based traders themselves. What benefit is there in this to society? Thomas Peterffy said that higher speed in trading "has absolutely no social value." Do you agree with him?

Professor Linton: Maybe I could answer that first, and others may then have something to say. I think the main social benefit of computer-based high-frequency trading is in the two main sides of that activity: market making and statistical arbitrage across markets and across securities. The market-making side provides liquidity for people who want to buy and sell, and the arbitrage side makes sure that price is accurate across exchanges and securities, so that accurate information goes to the wider investment community. Those are the standard benefits that intermediaries perform, and the computer-based high-frequency traders just do that in a more cost-effective way so that they are able to carry out this function more efficiently and contribute to timing of spreads and reduction of trading costs for investors. That is one side of it.

In terms of the social benefit and social cost side of things, one must look at the overall picture of what is going on in the trading market, and what is the scale of computer-based trading in relation to other activities. For example, for Knight Capital, which has been in the news recently for different reasons, this was and still is a relatively large high-frequency, market-making activity. In 2011, it contributed about 17% of the trading volume on the NYSE and NASDAQ in the US, so it is quite a big part of the total market trading that goes on. The total revenue of that company in 2011 was about $1.5 billion, and its net income was $115 million. In terms of the social cost, you can try to extrapolate from those figures to see what is the total dead-weight, or the weight of this activity that is being put in the market, and if you compare that with some of the other figures around, such as Exxon’s profits in the second quarter of this year, which were $16 billion just in one quarter, you see that they are a relatively small part of the picture. Even Goldman Sachs is a much bigger participant in markets, and has much more turnover. The social cost of that activity is also relatively small in comparison with those bigger things that are going on.

Q5 Chair: But in his speech on "The race to zero", which you have doubtless seen, Andy Haldane talked about the way in which this creates very fat tails in the market, and effectively a less stable market that is more susceptible to significant shock. This may be a question for Dr Zigrand. Is that something that would make you question the value of this sort of activity?

Dr Zigrand: That is a very difficult question. My interest is in systemic risk and things that can go wrong. Before I go there, one advantage of computer-based trading in modern markets stems from the fact, which you cannot forget, that markets today are fragmented. There are many different exchanges and different places where the same securities-regular securities-are traded today. One could not have today’s markets without computers; the markets have to be linked together. In the Flash crash we saw what can happen if the link between certain markets goes wrong: derivatives get mispriced; ETFs no longer reflect the prices of the baskets they are composed of, and when that happens panic breaks out and people do not understand the market any more. So computers play an advantageous role in the sense that they link all markets together. But there are risks. I fully agree.

In the report we have identified a whole series of feedback loops-what we call endogenous risks. Sometimes algorithms can get into these feeding frenzies where one algorithm sells and that makes the second sell, which makes the first one sell, like in the Flash crash where they were passing securities around. There was an algorithm that said, "The more volume there is in markets, the more I will sell because I can hide myself behind this volume." So the more that algorithm sold, the more that happened, to just pass the buck around. Volume went up; therefore the original algorithm sold more again and the whole thing completely exploded. There are a lot of situations where algorithms, which cannot really think quite as much as humans, can get into this frenzy and these feedback loops. There can be fatter tails in the end if one is not careful.

One of the big outcomes of our report was to think carefully about what can possibly go wrong: to identify all the possible feedback loops and then come up with suggestions as to what can be done to cut through those loops. Much of the Foresight report is about cutting various feedback loops by having circuit breakers and various means to slow down markets if necessary. Bad things can happen with computers, so the report suggests that one has to monitor markets better than before and have intelligent circuit breakers and things like that.

Professor Cliff: I would agree, and I also agree with Oliver. Although the amounts involved are very large sums by the standards of how much someone earns in a lifetime, in the context of corporate earnings they are not necessarily particularly large. Some academics in America did an empirical study, in which they replayed the tape and said, "Okay, if we had an absolutely perfect high-frequency trading system, how much money could it extract from the North American markets?" It was of the order of $10 billion per year. So it is not a huge amount of money in the context of the global economy.

I guess it is important to make the distinction between the primary market and the secondary market. The primary market exists when a company first makes its public offering. That allows the company to raise capital so that it can expand. It rewards early investors in that company. But you will only have the primary market, in equities at least, if there is also a liquid secondary market. People will only buy shares if they believe they are going to be able to trade them thereafter, rather than being required to hold them for an extremely long time. High-frequency trading is just a logical conclusion of the application of computerised technology to the secondary market. I don’t think it is necessarily intrinsically bad but, as with any application of advanced technology, if it is done badly it can introduce risks.

Q6 Mark Garnier: You are absolutely right. The primary function of a market is to raise capital for industry and for jobs and all the rest of it. We all agree, I think, that that is a very good thing. But that has always been the case. We have always had a certain amount of liquidity in equity markets for a long period of time. That was certainly before you had computerised trading. My time in the Stock Exchange was spent as a dealer on the floor of the exchange where it was a face-to-face market. Part of my question is how has it changed that? Secondly, there are huge numbers of shares that are traded electronically that have appallingly low liquidity. Certainly if you look at the AIM market, that is also an electronically traded system and more often than not it is by appointment only when you want to trade on the AIM market. By the time you get out of the FTSE 350, you are getting pretty tenuous liquidity, and even within the FTSE 350 it is pretty dodgy. So actually this high-frequency trading is ultimately only limited to a number of stocks that were liquid before they came in. I question whether that is providing the liquidity you are talking about to the wider market or whether it is just condensing that liquidity into a smaller place and amplifying it. Does anyone want to comment on that?

Professor Linton: That is a difficult question to answer. I cannot give a strong answer to that. In the last 10 or 15 years, financial markets have really suffered a lot, as the general macro-economy has done. Lots of forces have been contributing to bad liquidity in markets and decreasing liquidity reduction in initial public offerings and so forth. It is difficult to extract the contribution of high frequency trading to the overall liquidity pool while you control for all these other things that have been going on. When you do that, the evidence that we present in our report is that they do make a positive contribution to liquidity and to other features of markets, such as efficiency and transaction costs for investors.

It is true that a lot of value is concentrated in the larger capitalised stocks, where there is already an active market. There is some concern, obviously, about what is going on with the smaller stocks. We discuss market-making programs in our report and the different sorts of programs that have been considered around the world for the small stocks in Sweden and places such as that. So it is a different issue, but it is something that we have considered, at least briefly, in the report.

Q7 Mark Garnier: If I may, I will press you a bit more on this. The problem is, you talk about treating it briefly. I have met a number of companies that had a main listing, and one of the things that has been incredibly disappointing is that the shares trade, if they are lucky, once a year, sometimes twice a year. Clearly, high frequency trading in these market-making programs is not helping there. Again, I come back to my original point, which is that all that is really happening is that the people who are writing the programs are concentrating on where there is already liquidity.

I completely buy the arbitrage case. If you have 25 markets all quoting slightly different prices in the same shares, this will iron out those anomalies. That is a good thing. I can see that. But I am still struggling to see how this is bringing wider market benefit. It seems to be a certain amount of benefit in a small number of stocks. Specifically how many stocks is it? Are we talking about the FTSE 100? The top 25?

Professor Linton: It does prevail quite broadly across the FTSE 350 in the studies that we have seen. That represents 99% of the market capitalisation of the UK market. Obviously, the small companies are very important.

Q8 Mark Garnier: How many companies are listed outside the 350? How many companies are not benefiting from this?

Professor Linton: I do not think there is any prohibition for high frequency trading to get involved with smaller companies. As you say, to the extent that they are on electronic order books, they can participate. As I understand it, there have been quite a few issues with initial public offerings, both in the United States and in Europe over the last 10 or 15 years, and a number of explanations have come along. In terms of investment and asset managers, they tend to concentrate on the big indexes, so you will want to get a broadly diversified portfolio of assets. Do you really need to go beyond the FTSE 350? That is 99% of the value of the UK economy.

Professor Cliff: I want to strike a note of balance. In the report, we were pleasantly surprised that there are these instances in the data. The evidence suggests that high frequency trading can bring some benefits, but we did not say that high frequency trading is a panacea that injects liquidity into any market. For me the surprise was that there were these apparently non-negative effects of high frequency trading in certain specific instances if you look in certain datasets, so it is not universally bad. It might not be a universal good.

Q9 Mark Garnier: It might be useful to step back a bit and look at what we are actually talking about. As I understand it, you have, essentially, three types. You have the black box traders, where you have a proprietary trading system putting on loads of trades using, as I remember, a Bayesian probability model. That is a prop trading thing. You then have the order of execution. As an investment manager, I can go to a broker and they can put on a TWAP or a VWAP trade. Then you have the market-making side of it. For the record, can you give a brief description of how each of those work?

Professor Cliff: Gosh.

Professor Linton: To some extent, one of the difficulties is that there is a bit of a blurring of these activities. That is one point that has come out in the study. For example, when you think of market making, you traditionally think of something like Travelex at Heathrow airport, which stands there to buy and sell at a price, and they make quite a big spread out of that. Nowadays, the function of market making is also done across trading venues, because, as JP said, there are fragmented venues, and also across securities. Not only do they supply liquidity to the market in terms of passive bid and ask combinations, but they are also doing that across multiple venues. Some of the time they are being aggressive in their orders-so market orders as opposed to just purely limit orders. There is quite a bit of blurring between these functions, and we would acknowledge that.

Dr Zigrand: The same companies do a bit of each. Some of the papers we have seen show that some of these high frequency traders could be 51% passive orders providing liquidity and 49% aggressive orders. So they do a bit of everything; they just profit wherever they can using whatever strategies or algorithms they can extract money from. There is nobody who is a market maker as in the old days, who gets rewarded for putting liquidity up in return.

Q10 Mark Garnier: Sorry, did you say there is no such a thing as that?

Dr Zigrand: No. Some countries, such as the Scandinavian countries, have mechanisms by which, if a computer or a market maker offers liquidity over the course of the day in small stocks, that person gets rewarded. You get paid so many million a year, and in return for that amount of money, you have to make markets for that small company and get those markets liquid. These models could exist, but they do not exist for all securities.

Professor Linton: Or at least they are part of the market, and then there would be competition because market makers do not have exclusive control of the order book, so other people can come in and cut into their business on the London Stock Exchange.

Q11 Mark Garnier: If I were to go to an institution and have a look at their computerised trading system, would it be the same system being used by their prop traders, market makers and the executors on behalf of the client business? Is it the same thing that is being used?

Professor Linton: I think that a small company, such as Knight Capital, would have a central risk management procedure that brings in different strategies. They may have different explicit strategies in different markets, but they would control all that risk at the same time and there would be a central function that allocated the risk across these different activities.

Professor Cliff: It is difficult to answer your question in general, but I can think of some specific examples. Around 2004-05, Credit Suisse investment bank, which had been building up a significant expertise in algorithmic trading, released their algorithms to their clients. The algorithms that had been developed by the prop desk and had been running internal to the bank were now available to the bank’s clients. That was quite an innovative move, which allowed Credit Suisse to capture quite a lot of business, and various of its competitors then followed suit. In one sense, that is an answer to your question, because exactly the same algorithms that were internal are now shared by the client.

There is another answer, which is that eventually they are all sharing the same system, because they will all connect through the same data pipes to the market. But very many large investment banks are quite heavily siloed, and they have often grown by acquisition. Certainly, seven or eight years ago, when I was working for an investment bank, the equity trading floor was pretty much entirely separate from the foreign exchange trading floor, which was separate again from fixed income. It could be that, even as recently as five or six years ago, there were difficulties in plugging together in a bank the different asset class trading systems because they had been developed separately.

Q12 Chair: Could we move on to cultural implications and issues around the standards and market abuse in this area? That is one of the core areas for the Banking Standards Commission. It is clear from some of the evidence we have heard in other areas that we have been working on-evidence from people such as Paul Volcker, Martin Taylor and others-that the demise of human relationships in banking has led to a decline in standards. Things just happen automatically and there is no sense that you are dealing with people at the other end. This is where we get into the question-I am struggling to put an intelligent face on. I suppose what I am trying to ask is, does the dehumanisation of the markets make it more likely for market abuse to happen and for that not to be seen as anything wrong, because it is just one algorithm beating up another algorithm rather than something more personal?

Dr Zigrand: That was a question that we asked ourselves at the beginning. We decided to have one survey of end users done by Oliver Wyman. They went around and asked everyone they knew exactly those questions: are you worried about this? What are your main worries? Do you think you are front-running the markets-all these questions about ethics, worries and trust.

Separately, we hired three LSE professors-sociologists-who went out and had in-depth interviews with a variety of traders. They tried to address exactly the question of depersonalisation. We don’t have a centralised trading place-a pit where everyone would trade with each other. Now we have one man, a dog and a computer in various places, and they do not know why people are trading. They do not see who is trading. There are no trading floors any more where you can see a person buying, selling, getting frantic or going to the loo more often than otherwise. The question they tried to answer is: how would you deal with the fact that you no longer have information about what everyone else is doing and thinking-the idea of common knowledge?

What they found was that people in market stress environments would call each other up to get some market flavour: "Do you see that? Are you calling the investment banks? Do you see any flow? Are your traders busy or not busy?" They go to social networks twittering or blogging about various aspects. There is a lack of human observation.

What we have found with end users is that they said, "I can’t trust the markets quite as much as before because there is less of everyone observing everyone else and keeping everyone else in check." If everyone sees what everyone else is doing, trust and certain standards can be upheld more easily than now with computers, where people can make some trading patterns and no one can see who did what to whom and so forth.

Even though the report did not find an increase in market abuse-we were asked specifically to write papers on that-that is quite irrelevant. If final users think that there is scope for manipulation to happen, whether or not it happens, they will just not trade or trust the market as much as before. If that happens, markets will be less liquid than before, and then they can be manipulated even more easily. People will realise that the markets are less liquid and that they can be manipulated even more easily, so people will trade even less than before. There is a feedback loop that leads to less and less trust.

One of the suggestions in the report is whether we can cut the feedback loop by allowing regulators to have a better picture of markets. Market participants told us that they do not see the regulators as being able to monitor all the markets and to have accurate time stamps to find out if there was an instance where someone front-ran a client or sniffed out an order coming in. People think that there is no big policeman who has the big picture. So we have common time stamps allowing regulators to go back and find out who traded before whom and in and across which markets. We have many standard suggestions in our report, precisely to get trust back into markets.

Professor Cliff: I agree. The findings from our survey seem to be summarisable as: everyone thinks that someone else is engaging in market abuse.

There is the issue of whether we are talking about the legal definition of market abuse-are there specific rules, laws or regulations that are being broken? Or does the technology enable people to act in ways that, although they do not break the letter of the law, distort the market or give those people an unfair advantage, or increase the perception that the markets are unfair? I think that there is anecdotal evidence to suggest that that is the case.

There is the issue of trust in whether the markets are appropriately regulated and policed. The story of Flash crash is not a happy one. I know that that is not a British market-it is an American market-but it took the combined efforts of the SEC-the Securities and Exchange Commission-and the CFTC five months to come up with a report, which was largely discredited within two or three days of its release. It took them a very long time to gather and analyse the data, and their analysis of the data is, essentially, wrong. So there isn’t a coherent official account of what happened on the afternoon of 6 May. In circumstances like that, you would not expect investor confidence to be increased.

Q13 Chair: No, I think that’s fair.

Professor Linton: I want to add a few things to that. Of course, I share those concerns. On the simple point of whether people are noticing things going wrong, one of the things with market abuse in a computerised environment is that there has not been a large number of cases brought. The two specific cases we are aware of are: one in the United States, involving Trillium Capital, and one that is still in process in the UK, involving a company that I will not mention. As I understand it, they start with tip-offs from the industry that things are not right in this security on this day, and we have seen this over a period of time. So there is still a crucial human element involved in identifying market abuse. At the London Stock Exchange, they have tools to try to identify whether market abuse of certain types is going on, and then it would go down the line to the FSA for potential prosecution. There is still a human involvement, and it is still possible to identify abuse, but nowadays it becomes much more difficult when you think about abuse across markets. You can identify it if there is a lot of bad stuff going on at the London Stock Exchange, but if things are going on across different venues, it is possible that regulators and exchanges are not able to identify what is going on, since it is happening in different people’s jurisdictions. They are only seeing part of what is going on.

Dr Zigrand: There is an ethical point as well, which is intent. Does an algorithm have intent? Does an algorithm want to violate a certain law? The interesting question is not only that, but, if one has a big, high frequency trading company, and one trades in lots of markets all the time, purely by chance it is possible that the algorithms have done trades, whereby, if one picked various trades out in the sequence, they would show an abusive strategy, violating one regulation or another. But there was not one algorithm trying to accomplish precisely that. It is just a fact that, if you trade a lot across various markets, how can you prove intent in that case? It just happens randomly.

Q14 Chair: But it sounds as though the people questioned by the sociologists were concerned, so there is a concern about standards. They do not see this as an irrelevant question, which is quite an important point.

Dr Zigrand: I agree.

Q15 Mark Garnier: Can I ask about your glossary? It is fantastic, particularly at L-"Layering and spoofing." I will briefly remind you: "Layering refers to entering hidden orders on one side of the book (for example, a sell) and simultaneously submitting visible orders on the other side of the book (e.g. buys). The visible buys orders are intended only to encourage others in the market to believe there is strong price pressure on one side, thereby moving prices up. If this occurs, the hidden sell order executes, and the trader then cancels the visible orders. Similarly, spoofing involves using and immediately cancelling limit orders in an attempt to lure traders to raise their own limits, again for the purpose of trading at an artificially inflated price." That has got to be market abuse, hasn’t it?

Professor Linton: It is abuse, and those are actually at the heart of the two cases I mentioned-Trillium Capital and the Swift Trade case in the UK. A process has gone through-they were trying first to identify what was going on, then getting confirmation of that; that it was a pervasive phenomenon with an intent to manipulate the price. Of course, that would be by a small amount, not huge amounts. That is what makes finding out and establishing it difficult.

Q16 Mark Garnier: Market abuse is market abuse. It is the intent to abuse the market, whether you achieve it or not. Looking at that, is it not the case that if you are a computer programmer who is coming up with this, or a trader, you may not necessarily be the world’s leading expert on interpreting the manipulating transactions rule of the market abuse regime? You are looking for all sorts of clever ways, and you could-I am being very generous here-unwittingly create an algorithm that is actually abusive to the market and that gets into the system. Being uncharitable, you are deliberately going out and trying to keep one step ahead of the regulators.

Dr Zigrand: You mentioned Bayesian updating, and I think that is a problem. What if algorithms learn to make money? You programme them in a very neutral, harmless way, and they just learn what happens to the market and learn to interact with other algorithms and develop over time.

Q17 Mark Garnier: This is robot trading?

Dr Zigrand: Yes, and robots learn through Bayesian updating, as you mentioned. They filter things out and change their behaviour. What if they then come to a point where they do an abusive strategy? How can that be prevented?

Q18 Chair: Who do you hold accountable?

Dr Zigrand: Exactly. I think that is a good question.

Professor Cliff: Indeed.

Q19 Chair: I am asking the question.

Professor Linton: On a more positive note, the issue of market abuse is that it is harder to achieve these objectives in large stocks, so it has to be at the smaller end of the market. I do not think you can manipulate the price of Vodafone or Apple or something like that. It is very difficult when you have a huge scale of market that is not necessarily on your side. It seems expensive.

Q20 Mark Garnier: It is interesting. That reinforces this idea that there is a polarisation in the market, where you have the existing big-cap stocks, which have a certain amount of trade in them anyway, and therefore are getting more liquidity, because that is where the trading systems can work. It can then get even worse. Not only are you not getting the liquidity that you want outside the FTSE 350, but on top of that you are getting an opportunity where robot trading, or otherwise, can find opportunities for quite significant market abuse.

Dr Zigrand: I agree, but it can be easily detected, given that there is not much noise to hide behind.

Q21 Mark Garnier: How do you detect this? Is there a computer program that is out there?

Professor Linton: You will probably hear it from the London Stock Exchange. They will be able to give you more details about the tools that they have. All I know is that there has to be finger-pointing-"Somebody is stealing my lunch." If that is a pervasive activity and enough people point credible fingers at what is going on, there is an investigation by, in the first instance, London Stock Exchange. Warnings are then given to stop doing it. If it keeps on, I guess it gets passed on to the regulators, who then try to make a case.

As I said, there have been only two cases of this type of market abuse that I know of that have been in the public spotlight. There is the case of Trillium Capital in the US, which did not go through a court procedure; the industry regulator, FINRA, issued a $1 million fine. So presumably the scale of what was going on was relatively small if it justifies a $1 million fine. In the UK, the Swift Trade case was an £8 million fine. It has not yet been awarded and they most likely will not get an £8 million fine. It is hard to detect the overall scale, and the cases that have come up have been relatively small in terms of economic value.

I did not want to squeeze out this discussion about market abuse, but the report does find that computer-based high frequency trading is not just manipulating markets. There are cases where it actually reduces manipulation, because of the arbitrage thing. If somebody is trying to push the price away from fundamentals and high frequency traders are then bringing them back together, that will tend to reduce the opportunities for market abuse. Two of the reports in our paper look specifically at the end-of-day manipulation, which people have worried about for many years. They report evidence that higher participation by high frequency traders in markets actually reduces the amount of end-of-day price manipulation.

Q22 Mark Garnier: This is fixing the closing price, is it not?

Professor Linton: Yes.

Mark Garnier: That goes on outside the FTSE 350 quite a lot.

Professor Cliff: One more thing to say is that, since May 2010 there have been a number of adverse events in several financial instruments, which are now colloquially referred to as mini Flash crashes, where you see a price trending on a drunkard’s walk and then suddenly it will drop down. That is generally not a good thing, but, to the best of my knowledge, none of those events have involved anything that would classify naturally as market abuse. If market abuse is where people are acting with malign intent to maximise their profits, most of these adverse events are actually the unforeseen, unexpected consequences of benignly motivated systems interacting with each other in ways that the designers did not anticipate. I would flag that as something that should perhaps be as much of a concern as market abuse.

Q23 Chair: I am glad you raised that as that was my next question. Your report refers-I presume that you are talking here about the normalisation of deviance, as you referred to it?

Professor Cliff: Yes.

Q24 Chair: And the problem that affected Apollo 13, Three Mile Island, Chernobyl and, for that matter, Piper Alpha-as it happens, that was something that I was aware of-is that, in engineered systems, major, system-level failures become inevitable owing to complexity and our incapacity to predict the interaction of the system’s different aspects. Does this not say to us that, without some way of at least slightly slowing it down, and compulsory pauses, it is highly probable that computer-based trading will be behind the next major financial crisis?

Professor Cliff: I think that that is not impossible, but I am not sure that slowing everything down will necessarily make it better. The issue with normalisation of deviance is where you have a system which is not fully understood, so people are learning on the job. They have a preconception of what the safe operating limits of the system are, but every now and again they step outside those limits. Then, because a disaster does not ensue, they think, "Oh well, it must be safe to operate it in that zone" and, eventually, mother nature comes to tell them that they are wrong.

I think that it is instructive to look at engineering culture and practices in other industries that are safety-critical. A lot of work has been done in defence and aerospace, nuclear power engineering and other various aspects of engineering where you have complicated systems, which are themselves composed of large numbers of subcomponent constituents which are self-contained systems that have been designed and are operated by different entities and groups, which are not necessarily all in concert; sometimes they are in competition as opposed to in co-operation. Those practices developed in safety-critical engineering are, to the best of my knowledge, not yet widespread in the engineering of advanced computer-based trading systems. In fact, one of the things that we discuss in our report is the potential for better practice.

Dr Zigrand: I would like to add something to that. Contrary to explosions in nuclear accidents, where the explosion happens for physical reasons, in financial markets, the explosions happen because people do crazy things that allow other people to do crazy things-there is no natural asteroid hitting the financial markets. If we have huge crashes and crises in the market, it is because the market itself feeds back on itself and sells more: the more that is sold, the more I have to sell, and the more you sell, the more I have to sell because prices go down, I have less capital and risk is up. So I have to sell, but if I sell, prices fall further and then you have to sell. The explosion in the markets happens because of what the markets get themselves into rather than there being a point where, suddenly, boom-the fuse blows and everything explodes. It helps us, in some sense, to cut again these feedback loops. One may not have it in other industries, but in finance there is a hope that if one had a way to prevent all of us from selling, one would cut the loop and prevent the explosion. Here again, circuit breakers that buy people time to reflect, come up with capital and replenish the margins-accounts and things like that-show that there are things that can be done. It is not quite as unavoidable as perhaps a nuclear accident is.

Q25 Chair: Circuit breakers are going to have to be designed by the regulators, or am I wrong?

Professor Linton: I do not think that is necessarily a good direction to go. They are a technological solution to a partially technological problem. In the United States, they were mandated after the 1987 crash of the Stock Exchange, but the London Stock Exchange, for example, has had circuit breakers for quite a few years, and they operate on a different basis from those in the US. There has been innovation in how circuit breakers are used and the features that they have. In particular, the new generation of circuit breakers work ex ante so that they anticipate. Instead of waiting until a transaction completely blows out so that it is on the record that there is a bad transaction, they use the order book information to stop potential bad events happening. So they can work ex ante before the actual transaction has gone through. They are more intelligent and more forward looking than the old generation of circuit breakers. I think that is coming through within the industry.

One issue that we raised in our report was about the potential need for co-ordination of circuit breakers. During the flash crash that was an issue. Things started in Chicago and gravitated to New York. Within Europe we have many different trading venues. The London Stock Exchange, for many securities, is not the dominant place for trading, so you might think that there needs to be some sort of co-ordination across trading venues in the event that some bad things happen in one venue or in one security that could cause contagion in other parts of the market.

Q26 Chair: But the speed at which they are working means that the circuit breakers have to be even quicker.

Professor Linton: Well, they are all automated, too. There is nobody there pushing the on/off button. It is all completely automated.

Q27 Chair: I realise that, but given that you are talking about moving your servers five yards nearer to wherever it is, I am thinking that the Stock Exchange-in order to be able to save yourself a tiny proportion of time-if the circuit breakers are going to operate ex ante, is talking about the circuit breakers operating in minuscule periods of time.

Professor Linton: Right, but they are perfectly capable of doing that.

Q28 Chair: That is doable?

Professor Linton: There is an issue of peak-load messaging, which was an issue in the flash crash with the flooding of servers, and so forth. But there are pricing schemes in place, I believe, in the London Stock Exchange, and there are also constraints on the volume of orders that people can submit to the market. It seems quite a natural market development to control spam and to try to reduce the volume that goes through their servers. There has to be some sensible way of allocating space across investors. I think that exchanges have quite a few programs on those issues.

Q29 Mark Garnier: On the circuit breakers, are they an exchange-driven thing, as opposed to a prop-trader thing, so the exchange will pick it up? Presumably, it is in the interest of proprietary traders to have some sort of circuit breaker, too. If they suddenly find that they are selling Vodafone down to one cent, that would be quite a bad thing to be doing, wouldn’t it?

Professor Linton: Yes. In principle they have their own risk limits, and if they breach them, they would shut down or take some other action to reduce the risk, but that does not always work.

Professor Cliff: Knight Capital took 35 minutes.

Q30 Mark Garnier: Can I talk about something else with which I am not at all familiar? Can you expand on the general equity of the market in terms of people getting access to it? In particular, I understand that you can trade from news feeds. How does that work? Dr Zigrand, you were nodding first, so why don’t you answer the question?

Professor Linton: He is a Twitter fan.

Dr Zigrand: There are entities that have developed semantic algorithms that can read. These days, when the news agencies broadcast a news item, they attach various tags to it. Dave can tell me if I am wrong, but various entities and traders now have systems that read those tags and try to make up stories. The computers will read the tags and see, for example, the unemployment numbers. They look at the number, and they have these pre-fixed numbers, so they put, say, 3 million inside. The news comes in and they read that unemployment is 2.5 million and, automatically, the algorithm will say, "Well, that is less than what we thought. That is good news, so I will buy the market." Therefore markets are much quicker today at bringing information into prices, because they are read by computers in real time and traded upon.

Q31 Mark Garnier: So you are looking for certain things. For example, you have got ICI results coming out. You are expecting it to be up 15% and it is up 14%-that is bad news; but you have to tell the program first, so there is a lot of human interaction.

Dr Zigrand: They have the shadow prices, but even the shadow prices can be computed by models, which are inside the computer itself.

Q32 Mark Garnier: And these are straightforward news feeds? Or they are specialist-provided news feeds, via Bloomberg or someone?

Professor Cliff: Yes, and also social networking sites; so various people have written systems which analyse-they harvest tweets on Twitter and just count the number that are positively or negatively disposed towards a particular company or a particular event occurring, and use that to bias the machine’s trading activity. But the machine understanding of linguistic semantic data is nowhere near as advanced as the machine processing of numeric data. There has been a very long history of computers sucking in numbers and doing maths on the numbers and spitting out a number as a result; but it will take a very long time-a while at least.

Q33 Mark Garnier: But how long is long? Are we talking about half a day or two years?

Professor Cliff: The point I was trying to make was just that right now an elementary machine-based trading system can do mathematics that I am not physically capable of, because I cannot receive data quickly enough and process it and reach forward and press a button on a trader terminal quickly enough; but through the benefits of 4 billion years of evolution I am quite good at processing linguistic information, and better than most computers. So IBM’s Watson, this kind of artificial intelligence thing that famously won "Jeopardy!", is an indication that probably within a decade or two close approximations to human ability at processing information that is in linguistic form might be employable in the field; but at the moment it is still a research issue.

These things that mine sentiment are really just trying to work out which way the populace is leaning. Are they broadly saying we should buy, or broadly saying we should sell? What is still very hard is any kind of logical reasoning. The famous example is buying the London Rubber Company shortly after the advent of HIV being identified. When it was clear that there was going to be a big public health education push to increase the use of condoms, the kind of causal reasoning that you then go through, which then means, "OK, we’ll buy London Rubber Company, because they are producers of that product," is something that it will take some time before computers can engage in.

Chair: Although that would be a bad decision, because actually the management was not very efficient at dealing with the upgrade of their machinery.

Professor Cliff: Quite possibly.

Q34 Chair: It seems to me that what you are describing is the development of a complete asymmetry of information, because the ordinary investor who does not have access to this and who is reading the daily paper or following a stock on the web might as well be back in the 17th century in a coffee shop.

Professor Cliff: Well, JP has got something to say, but my response to that would be, I guess, yes, in principle, but let us not forget that Warren Buffett has been an extraordinarily successful investor, buying and holding for very long periods of time. He is one of the richest people on the planet and he got rich doing something that is still entirely consistent with today’s markets.

Dr Zigrand: In the old times the market makers had a monopoly of making markets. They were the first ones. They saw the news coming. They had the screens in front of them. They could trade immediately, whereas we could not. If anything, it is probably-and this is my personal view, not the Foresight view-easier for everybody today to trade quickly than it used to be in the old times, just because there is no monopoly any more. If I am tucked in at a broker, my trades will just go straight through the exchange, through the broker, and I can be really quick, and setting up a high-frequency trading company, or at least something quite good, is no longer impossible-a few hundred thousand pounds, I am told-and one can be really fast. One could not have done it in the old times.

Chair: Thank you. That is a very valid question. Let us move on to policy measures.

Q35 Mark Garnier: Just following on from that point. What is the point of it? Warren Buffett is one of the most successful traders. You are suggesting that it is still possible. When I was first in the Stock Exchange, a great story went around that of all the investment funds that traded reasonably frequently, the ones that did best were, I think, an Oxford or Cambridge college that met once a year and had a policy meeting in January and set it every year. [Interruption.] Trinity College. There you go. The important point is that, no matter how hard you try to beat the market, actually, you cannot. The market is still the market, which is reassuring, because as you said, it probably is equitable and it probably does provide the liquidity, and all the rest of it. But what advantage are they actually getting out of it? Presumably, these institutions are spending huge amounts of money on their kit. They are presumably paying vast amounts of money-we will find out in the next hour or so-on getting their servers next to the Stock Exchange server. Are they actually making money? Are they being competitive against each other?

Professor Linton: With the Twitter feed tools at the moment, I would imagine that most of the time they make very little money. The quality-the predictive ability––of these tools is pretty weak. There is also the possibility that people do not write the truth. That can happen. I am not so impressed with the Twitter feed. It will be a few years before that will really have a big impact. It still goes back to this intermediary function. Ultimately, they are potentially providing this link between fundamental buyers and fundamental sellers, doing this function in this electronic way and, by most accounts, the amount of profit that they are making from that is relatively small, so that the social cost of this activity is relatively limited. It is not going to take over the world, because the amount of profits that are available for doing this sort of thing are relatively small.

Q36 Mark Garnier: Although greed would always mean that people believe they can beat the market. That is why people trade, gamble and do this kind of stuff. Somebody always believes that they can get-

Professor Linton: Yes, and those people will lose money. If they buy high and sell low, they are going to go out of business. Knight Capital almost did. Sometimes, as we have discovered with these things, there can be systemic issues where these things cascade out of control and that can cause a problem. But the general ongoing operation of markets, as we found in our project, was that the interaction of all these people with their selfish motives and their high technology generally has led to more liquid markets, more efficient markets and better lower transaction costs for people who want to buy for the long term, for their pension or education.

Professor Cliff: It is definitely worth the while of the banks investing in this technology, because in my experience most banks and fund management companies see IT as an unpleasant cost and will not spend money on it unless they believe it brings them advantage. But the dynamic between the companies is one of an arms race. Company A innovates and that innovation gives it a competitive advantage for as long as the other companies do not have that technology. So it then either develops its own in-house technology or buys it from third parties and eventually the playing field is levelled and everyone has spent a bit more and then some other company will innovate and gain an advantage. For as long as no one else has that technology, the technology gives them an advantage, but then everyone else spends to re-level the playing field. I think they believe that the technology does give them advantages, but quite often the advantages are quickly shaved away to almost zero.

Dr Zigrand: The profits went down. If you compare the profits made by high-frequency traders today to the profits made by market makers in the old times, they would be much less. Markets are more efficient. There is less gravy left on the table.

Q37 Mark Garnier: Can I turn to international competitiveness and regulation, which has to be looked at internationally? You talk about the arbitrage between markets across Europe, and there are obviously lots of different exchanges, and all the rest of it. First, if there is going to be any regulation on this, it presumably has to be global. You talk in your report about Europe-wide regulation, but surely it has to take into account the whole world, and also harmonisation in terms of spreads and that kind of stuff.

Dr Zigrand: All the linked markets would have to be looked at as one. If one can trade the same security in London, Paris, New York and so forth, one really would have to have common rules. And one has to think about the circuit breakers-should they all break at the same time, in all places where the same securities are traded or where related securities are traded? One would have to look at where the related securities are situated and then find common rules.

If the circuit breakers are not co-ordinated, what can happen is that one market fails and all the sellers go to the other market, which could fail, and there could be a sequence of such events. The problem of linking everything always is this: what happens if, on a very small exchange that does not have many trades at all, somebody has a fat finger and that exchange gets triggered? Should we then stop the main exchange as a result? Probably the answer is no. So there needs to be some intelligent, international, cross-market co-ordination of various policy devices.

Professor Linton: But it may not be necessary to involve the whole international community to set identical rules, because I think one size fits all does not necessarily work for different countries; the London market is quite different from Slovenia, Thailand or somewhere like that. So there is obviously something to learn about best practice-about how to intervene and constrain risks in markets-but I think that it is more of an informal discussion type of thing at the global level.

Q38 Mark Garnier: All these markets are competitive with each other. You rightly say that a market like Thailand is not sophisticated enough and has its own set of stops. But at some point, Thailand wants to be an international player, I would imagine, and does that not give it the opportunity to start trading these stocks without necessarily having the circuit breakers?

Professor Cliff: Yes, in principle. In some of the evidence that we gathered before we wrote the report, there was discussion of the fact that technology development has lowered barriers to entry. So, if you have a big enough pile of money, you can now build a data centre and a data network, which means you can set yourself up as a national or international exchange in a way that probably was not possible 20 or 30 years ago.

Q39 Mark Garnier: How much money would you need, just out of interest?

Professor Cliff: I just would not want to answer that question.

Q40 Mark Garnier: About £100 million-that sort of thing?

Professor Cliff: I think the next witness would be better placed to answer that question.

Q41 Mark Garnier: There is an important point. Obviously we are extremely interested in the competitiveness of London. What threats do you see to the UK Stock Exchanges through this, if any at all?

Professor Cliff: Here I am speaking personally and definitely not on behalf of the Foresight project, although in my role for Foresight I was involved in reviewing some of the MiFID draft legislation, and I think that badly thought-out regulations are a serious concern. There was some press coverage of the project’s activities, which seemed to imply that we were supporters of HFT because we were critical of MiFID. My response to that would be that we were critical of MiFID because MiFID was not written very well and it was not very well thought out. I think that there is space for good regulation-appropriate regulation-but having politicians saying, "Something must be done", and then drafting legislation that is poorly thought out or impossible to execute-

Mark Garnier: Not us, I can assure you. [Laughter.]

Professor Cliff: So, as I was saying, that is a concern, because under European law and our current membership of the EU, if MiFID is enacted, each national legislature has to introduce laws that are compliant with MiFID. We could find ourselves having to operate all the markets in the UK in a way that is compliant with MiFID’s regulations.

Professor Linton: A separate macro-economic point is that over the longer term, with the growth of Asia, Asian economies and their exchanges have also grown in value, and they are replacing European Stock Exchanges in the top 10 list of venues. If that continues for the next 10 to 20 years, obviously a lot more action will be taking place in Asia relative to Europe and in particular the UK, but that is probably not something-

Dr Zigrand: There are externalities, I think. The more liquid an exchange is, the harder it is to move liquidity away; we have seen that in futures exchanges. If lots of people trade on the London Stock Exchange, for instance, it is very hard to move them, because everybody says, "I will move if everybody else moves", but then no one will ever move. Secondly, for the UK, stamp duty is always an option that can be played with. If London is in danger, stamp duty can be lowered.

Q42 Mark Garnier: I was going to ask about the financial transaction tax, because many European countries are talking about bringing it in. Do you see it as a potential benefit to the UK, in terms of the high-frequency trading? A financial transaction tax on a unilateral basis would completely destroy that exchange, would it not?

Professor Linton: Yes. It would damage the business model. We have a stamp duty tax at present.

Q43 Mark Garnier: We were struggling with this a bit earlier. That is only charged when you register the stock at the end of the day, is that right? Intra-day trading is not liable to stamp duty.

Professor Linton: Registered intermediaries are exempt from the stamp duty and if you are not a registered intermediary, you can simulate this by using contracts for difference. Even if you are not explicitly a registered intermediary, like many high-frequency trading companies, you could do it that way.

Q44 Mark Garnier: They are using CFDs?

Professor Linton: Yes. Stamp duty on falls on the long-term investors-the people who actually hold the shares.

Q45 Mark Garnier: So it is when you register the stock as opposed to when you transact it.

Professor Linton: Yes.

Q46 Chair: We have just two or three minutes left. Looking at the long-term future, which is obviously easy to predict, where do you see this going? Are we getting towards a limit in computer-based trading? If we were 10 years down the line, what would you expect to be looking at?

Dr Zigrand: Some markets seem to be saturated. Some of the papers that we commissioned found that some of the markets, such as Euronext, seemed to be saturated. The extent of computer-based trading seems to have peaked, although one does not know. The point is that one has to think about computer-based traders versus high-frequency traders. Since everybody in the future will probably trade with computers, everything may be computer-based trading, but the numbers raised usually are about high-frequency traders. How many transactions have a high-frequency trader on one of the two sides? There must be a limit, since for them it is a zero-sum game. It cannot grow over a certain limit, which is determined by the flow of buy and hold investors at the end of the day.

For the equity markets, there will be a natural limit at some stage, but most markets have not been influenced all that much by computer-based trading. Dave probably knows more about this than I do, but there is pressure from regulators to move markets from OTC, which is contracts over the counter, towards exchange. That needs clearing first, but once things are cleared and standardised, it can be put to those exchanges, so there will be many products that will be ready to be computer-traded. Many securities will become more HFT.

Professor Cliff: I would agree with all of that. For as long as there have been traders, latency has been an issue. Messages used to pass on horseback and then on pigeons and then via telephones. As long as there is technology coming onstream that reduces latency, there will be brief opportunities to make money from low-latency solutions, until everyone else has got the solution.

Professor Linton: And I would say that there will be a continuing improvement in latency, because computer technology will increase. At some point, however, it comes up against the granularity of tick sizes and other kinds of limits that will prevent trading going to an extreme. There is also the limit of the speed of light, in terms of latency. They cannot go beyond that. As we approach that speed, the extra value that is available for high-frequency traders diminishes quite rapidly. There is a limit to the size of the number of shares that you can trade. You cannot trade a fraction of a share and so on, so there are natural limits that will prevent it going to the speed of light.

Dr Zigrand: Banking regulation-this is a banking commission-will probably affect that as well, because banks make markets. They have market-making entities and prop trading entities. Depending on which way the wholesale market is regulated when the report goes through, there will be segmentation between the retail side of things and the wholesale side of things, and the wholesale part of the banks will no longer have access to retail deposits and capital. Therefore, there may be less intimidation by banks, which would leave the door open for more computers to step in. That may be purely on the basis that market making becomes onerous to banks. There may be more high-frequency traders who will take that job.

Chair: Thank you for your time and for a very interesting session. We are extremely grateful to you. We will have a three or four-minute break.

Examination of Witnesses

Witness: Alexander Justham, CEO, London Stock Exchange plc, gave evidence.

Q47 Chair: Thank you very much for coming in. You have been sitting there watching and listening carefully, and obviously with some interest, so we will now reverse the process. As I warned you, my opening question is this. Is there anything you would like to challenge, change, amend, add to or agree with in what you have heard over the past hour? What struck you in particular?

Alexander Justham: First of all, it is a pleasure to be here, Chair, and thank you for inviting me. I would agree broadly with most of the evidence that you heard from the previous three participants. Hopefully, they provided an insightful view of the process. One reason why we have been tremendous sponsors of the Foresight project, both in my previous role-I should probably declare my previous role; I used to be the regulator-

Chair: Yes.

Alexander Justham: You have probably clocked that.

Q48 Chair: And you were at JP Morgan.

Alexander Justham: I was at JP Morgan a long time ago-I will put that into the dim, distant history-but I did my five years’ tour of duty in public service at the FSA, where I was also the European regulator as well, by the way. I was on the board of the European Securities and Markets Authority and chaired the secondary markets standing committee. I produced the guidelines around this space, so it is quite a familiar space.

The Foresight project was incredibly important, because it helped start the journey to addressing some of the issues that you highlighted-where is all this taking us? What are the issues around it?-and to addressing some of the perceptions versus reality around this space. Both as a regulator in my former life and now, running the Stock Exchange here in London, I strongly welcomed and supported the Foresight project. I think it has done a great job of starting to look at the issues versus the perceptions, and the realities versus the concerns, which may or may not be founded. Putting good scientific evidence around that has been a great step forward in taking everyone’s understanding of this market space forward.

Q49 Chair: Following up from that, given that you are essentially talking to a fairly vague bishop, in terms of the questions that I was asking-I certainly would not speak for Mark Garnier-what perceptions did you hear that you were uncomfortable with or would want to challenge?

Alexander Justham: I was not uncomfortable with any of them. I think that in many ways-

Q50 Chair: We won’t be offended, I assure you.

Alexander Justham: I think the questions are all very relevant, which does not surprise me, given that both of you, from your backgrounds, are very experienced within this context. These are the same questions that we had as a regulator and that we have as a market operator, without any doubt. We think about these issues all the time. Both as a regulator and now as an operator, my key, principal view and role in life is to ensure that we operate an efficient, functioning market for a core, central economic purpose, which is to allow companies to raise capital and investors to invest in capital.

The Kay report has a very good discussion of this, and it is incredibly important. The one comment that I would make, in light of your very first question, was that the Kay report makes a good distinction and explores quite well the sometimes conflated views of investors and traders. It is important to draw the key conclusion. There are core constituents: companies who need to raise capital; and investors or savers, for want of a better word-people who need to invest in, be a part of and obtain long-term economic growth from economic activity.

The marketplace in the middle essentially serves one basic purpose: to provide intermediation between those two core constituents as efficiently and effectively as possible. As Kay describes, it performs, critically, probably the most important function of a marketplace: it allows those two entities to invest-they are both investing, in a way-over different time horizons, and allows those different time horizons to function. That is essentially what we are talking about. It is essentially my job every day to ensure that we as an organisation provide, for those two core constituents of our economy, the most efficient marketplace to allow the functioning of their investment through different time horizons.

On the trading community that sits inside that-the liquidity provision that we have been referring to-the issue that we are all talking about is how efficient and how effective that intermediation is. To go back to old money, we would never have called the jobber an investor: they are not; they never were; and they never shall be.

To go back, Mr Chairman, to your oil days, no one would describe the Chicago pit as the investors, the producers or users of oil, but they were an essential form of intermediation and of changing that time horizon. They themselves made money, and that is the economic rent extraction from that intermediation. The interesting thing in that dynamic is the economic rent, and it is what is probably most socially important-to come back to the social usefulness and other things. The most important debate to have is, what is the economic cost of intermediation between those two constituents? This is the debate that Kay, again, has. The interesting debate, which we have heard today, is that the economic cost of this intermediation and the technological change has actually, strangely enough, been going down, rather than going up. That is therefore socially useful.

Q51 Chair: And is it sufficiently useful to justify the risks we heard? I think that phrase was, "It is not impossible that the next major crisis will be caused by computer-based trading."

Alexander Justham: I have slightly different views-there are two different, balanced views. I think that clearly, as has already been demonstrated through a couple of key events in the US, any market development and market change, particularly technological development, which in essence this is, comes with new challenges and issues. As the marketplace and dynamics change, so do the rules and the regulations, and critically-I used to use this terminology-the brakes. As your car gets faster-you build a bigger engine and you can drive down the autobahn at 150 mph-at the same time you also need to think, "Do I have a good set of disc brakes?"

As a regulator we felt-and put out very clear guidelines to try to keep abreast of those developments-that as an operator it is absolutely essential that we have the capacity in the various checks and balances within the system. You were asking whether that lies at the exchange, at the first level of intermediation or ultimately at the origination of the process, and I think that the answer is yes to all of them. It must do and it should do.

Coming to the arguments of complex engineering and checks and balances within the space, the system overall needs to start developing all those key elements. We, as the operator, almost have to catch something when it has failed through all the others. If they miswrote the algorithm in the first place, or if the systems of the entity receiving the trade or intermediating it-the broker or whoever-do not check on it or do not call it, ultimately, the operator has to be the last quarterback, to use the American expression, in the process. That is why we have over the years pioneered circuit breakers-static and dynamic ones-pioneered ex-ante as much as ex-post. We have pioneered volume thresholds as well, because sometimes it is not only the price movement that is causing it, but actually the sheer volume coming through the system.

Everyone has to be constantly vigilant and constantly keep up to date with the dynamics within the marketplace. Even though we, or regulators, impose controls further up the chain-we do so, and we ensure that people test their processes before they can come on to our market or anything else-as a previous participant said, that can never be foolproof. You have to ultimately have the quarterback process at the end of it, which basically says, "If the volatility goes out of whack, you stop it or shut it down. If the volume coming through is too much, you shut it down."

To answer another one of your questions-you were saying, "Hang on, can you do that?"-the key thing is that the very same technology that can make people go very fast, is the very same technology that can stop it very quickly. It is an absolute mirror, and quite essential. In fact, you can buy systems that can do pre-trade controls further up the chain-they can check the credit limit of the individual member or whatever-and they can be super, super, superfast. There is the technology, so in some senses, the very technology that potentially creates the issue, is the very same technology that can protect you.

Q52 Chair: I have one last question, before I hand over to Mark. I was struggling in the last session, which you were listening to, to see where there is responsibility for standards and culture, because all I see is computers and algorithms. How would you answer that question? It sounds to me that that is you.

Alexander Justham: It is multiple layers. We are definitely a part of it; of course, we are. We are part of it in terms of how we are thinking, what our governance processes are, what controls we are building and what requirements we put on people who want to trade on our markets. If you look at the ESMA guidelines that we produce, we have said that that has got to feed up the chain: there has to be accountability all the way through the various components of that chain. At one level, there are humans in all of this, as a previous participant said, and a computer program is not written by itself-not yet, anyway.

Q53 Chair: That is one of the things that is worrying me.

Alexander Justham: I appreciate that. At that stage-well, I will not comment on that.

At this stage, obviously there is human intervention, and human accountability and responsibility-these firms are human firms, who are using computers. There is a whole regulatory framework that should probably go over that accountability and responsibility, and the regulatory world is getting further along the line of that, by thinking about what the accountability is, what the responsibility is, who is checking these things, who is signing off on them, who is accountable when they go wrong, and all that kind of regulatory structure for the various entities. For example, is everyone in that chain regulated? Historically, that was not the case, but it is absolutely moving towards being the case.

It seems to me illogical that you cannot have some form of regulatory oversight if you are a major participant within these marketplaces. There are changes in the responsibilities coming through, which I think are very important and essential in trying to get that very key element of, basically, integrity.

Q54 Mark Garnier: This whole question of integrity is very difficult for you, isn’t it? Ultimately, you are the London Stock Exchange, which started in Jonathan’s coffee shop in 16- or 17-whatever, as you know. You have been around for an awful long time and, as a result of that, you have a huge amount of intrinsic trade that is going to come with you. You have that volume and everything that makes you desirable but, at the same time, you have new trading systems that are coming to you, and you have to accommodate those.

You have to continue to be at the heart of share trading-huge numbers of shares, internationally, are listed on the London Stock Exchange-but at the same time you have to potentially open yourself up to this incredibly frontier type of technology that is coming in and which does not yet seem to have very clear written rules. You have talked about robotic programs that can invent all sorts of potentially market-abuse types of trades. How do you deal with that conflict?

Alexander Justham: We have probably always lived with a variety of stakeholders. A market, by its definition, is a broad group of stakeholders. Our obligation, as we have always seen it, is to operate fair and orderly trading. That is the heart of what we are trying to do, and integrity within that is obviously paramount. In the same way that, as you said, we have been here since 16-whatever, our objective in life is to be here an equally long time into the future. We are only going to achieve that if we operate high-quality and critically-coming to some of the issues raised earlier-perceived high-quality marketplaces that offer a variety of participants that fair and orderly environment in which to intermediate and trade.

We are regulated, absolutely for that reason, and we ourselves have spent a lot of time and invested a huge amount in technology. In fact, we own a technology company, and that has definitely helped us stay on top of these types of issues. It has helped our surveillance of those types of issues, so that builds our surveillance capacity. Just for the record, we employ about 600 or 700 people in Sri Lanka who do all this. We own a company called MillenniumIT, based in Colombo. So we take that responsibility very seriously.

As the previous participant said, technological change and market dynamics have changed within the marketplace. MiFID has opened up this fragmentation which really did not occur prior to the introduction of MiFID I and, with that fragmentation, high-frequency trading almost became the necessity in order to reduce the arbitrage. That, again, is a benefit and not a disadvantage. We have had to stay apace with that. We have operated in a very competitive environment. We, in essence, pretty much lost 40% or 50% market share, so while on the one hand you say that it is a given that everyone will always come and trade with us, the reality is that that is not the case and we have lost significant market share since the opening up of competition within that base.

We have to adapt, we have to maintain being competitive in that context but, critically, we have to maintain our paramount objective which is to operate a fair and orderly market, so we put a lot of time and effort into our regulatory side and our surveillance side and into ensuring that the treatment of stakeholders across the board is fair and orderly. It will always be so. It will always be a challenge. I am not telling you that we do nothing on it; in fact we devote a lot of resource to it. It is a very important thing that we need to maintain.

Q55 Mark Garnier: In a practical sense, you obviously get an awful lot of people who are writing programs that will interact directly with your programs. What check do you have and what right do you have to go and look into those programs to see if there is anything in there that will cause problems, or are they allowed just to plug in?

Alexander Justham: No, they are not allowed to. When someone comes on to the market, they have to test on to our market, so we do a huge number of tests and stress tests within that context. So no, people are not allowed to do that. Under the contract by which they come on to the market, if they change the code, they have a contractual obligation to retest with us. We are vigilant within that perspective. Again, that is something that the London Stock Exchange does; that is something that quite frankly-as the ESMA guidelines provide-we think should be a universal position across all marketplaces. It is essential that these checks and balances are built through the entire chain.

Q56 Mark Garnier: Do you have a right to see their proprietary models?

Alexander Justham: I will probably have to come back to you with the precise answer. We have the absolute right to enforce the checking.

Q57 Mark Garnier: The other thing that is coming up is the proximity of the servers to your servers. I gather you rent out space next to your servers-actually, where are your servers? Are you allowed to say?

Alexander Justham: I do not know if I am allowed to say-probably not really. Put it like this, given who the Chair is, they were not by St Paul’s.

Q58 Mark Garnier: Presumably you have a redundant set of servers as well.

Alexander Justham: We do have a back-up site as well, yes. We have our main facility-

Mark Garnier: In place A, and a back-up in place B.

Alexander Justham: And a back-up in place B. Actually, if absolute worst comes to worst, we can move it to Milan, because we run the Italian Stock Exchange, but that presents issues.

Q59 Mark Garnier: I come back to my point, which is that with your primary server presumably there is an opportunity for people to buy space next to it. Does that not give an unfair playing field to people who have a lot more money than others?

Alexander Justham: You started to have those debates in previous conversations as well. The cost of co-location is not a hugely expensive barrier to entry. If you are interested in the number, it roughly costs about £60,000 to rent a box near our server, and well over 100 of our members do so. As we experience in many aspects of life, strangely enough, the technology has caused greater democratisation than existed in the old world. In the old world, you probably had to build your offices near the Stock Exchange, and an office was probably a lot more expensive than putting a computer next door to our computer.

We are obviously conscious of that, but one of the interesting aspects when we debate the variety of stakeholders we are trying to offer that fair and orderly market to, is that we look at the ability of those stakeholders or participants to choose and perform a trading service that they would prefer and that would not disadvantage them. There is no real evidence at all that the co-location environment is somehow disadvantaging a whole group of investors, compared with others. Most people, if they are trading through one of the major members of our exchange, have access to those facilities at very effective costs.

Q60 Mark Garnier: Okay. Let me come back to you on an earlier question, about people changing the code in their computer programs, because we interrupted you. How do you know they do not do it? How can you check whether they change their code? Do they tell you about it?

Alexander Justham: If they don’t tell us about it, they have breached their contract.

Q61 Mark Garnier: Well, they may lie.

Alexander Justham: Generally, if it looks different, and starts behaving differently, we will observe it.

Q62 Mark Garnier: How would you know?

Alexander Justham: Because we can track, obviously, the origin of trading, so if something looks very different, or suddenly starts to behave in a very different manner, we would be able to observe that.

Q63 Mark Garnier: Are you running diagnostics that look at the patterns, trends and activities of a certain counter-party?

Alexander Justham: Absolutely, yes. We run the front-line market surveillance. This comes back to one of your earlier questions: we are still very heavily regulated, and we take on regulatory responsibility, as an exchange. Our principal regulatory responsibility is that we take on that front line of ensuring orderliness and fair trading in the marketplace. We run a surveillance division, who are literally monitoring trading all day. We use a lot of technology-again, developed by us-to survey that and monitor that. You were talking earlier about monitoring spoofing, layering and other such jargon, and those are the kinds of things we look out for. We have a variety of mechanisms to do that, and, again, we are constantly evolving, as one has to; that is a very important function of what we do, and we dedicate a lot of time and resource to it.

Q64 Mark Garnier: But a robot trader that is writing-if that is the right word for it-its own trading system as it goes along is presumably a dynamic system, and it is going to be changing, so it will be flagging up as a new system the whole time.

Alexander Justham: It may be flagging up as a new and different type of trading. The question is whether it is abusive, or whether it is not orderly.

Q65 Mark Garnier: That is important, and I am not trying to tie you in knots, but I am trying to get an understanding of this from our point of view, because we are just humble politicians and bishops, and it is important to understand this. I understand that if somebody writes a system that just sits there and trades, and then they change it, you might pick that up, and they will be in breach of contract and all the rest of it. But if somebody has a system that updates itself in terms of its trading approach, are you, first of all, picking that up? Presumably you are. Secondly, it could be in breach of all sorts of-

Alexander Justham: We are monitoring the outcome; I think that is quite key. We monitor the outcome of what is trading and happening on our order books, so if we see behaviour that we are uncomfortable with, or we think is not in line with creating an orderly and fair market-whether it is abusive, or just creating too many orders and too many trades and clogging up the system-we monitor that and have the capacity, therefore, to go back and say, "So what’s going on?" Ultimately, we are monitoring the outcome of that marketplace to ensure its resilience and its fairness.

Q66 Mark Garnier: Some of the best computer programmers, presumably in the world, are writing this stuff. They are looking all the time at ways to outperform their competitors to be better. Surely, at some point, they are also trying to outperform you. Can they not game you as the LSE?

Alexander Justham: We are not an economic actor in that sense; they are not gaming us.

Q67 Mark Garnier: But they are trying to game the rules; they are trying to game the market and their competitors.

Alexander Justham: If they are trying to be abusive, that comes, again, down to how good the systems and the regulatory environment are in trying to track abusive behaviour. Without any doubt, one has to keep on adapting and moving with the times. The type of abuse that used to occur has always modified itself and it has done for many years.

Q68 Mark Garnier: But you could ultimately have three or four separate trading systems working together in concert, and you would not necessarily pick that up.

Alexander Justham: We would pick up the outcome. In the end, the outcome sits on our-

Q69 Mark Garnier: Why is one outcome obvious and not just a lucky trade?

Alexander Justham: In what context?

Q70 Mark Garnier: Presumably the outcome that you or the trader is looking for is that you buy a stock and it goes up, and then you sell it when it is higher. You are looking for market abuse. You are looking for patterns, for example, of layering and spoofing, but why is it not possible simply to have a number of different trading systems that have a link together and are working, as I say, as a sort of computerised concert party, trying to create an outcome that you would not necessarily pick up? In itself, each trading system is not doing anything that looks like anything other than a normal trade, but between the three or four of them, they are doing something that is market abuse, and you just cannot see it.

The point is-this is at the hub of it-is that technology is now so clever that it is incredibly difficult to keep ahead of it. That is what we are trying to get to the bottom of. We can sit down until we are old and grey, as some of us already are, trying to work out ways of gaming the system, but it is now moving so fast. So much is going on. I am absolutely convinced of your integrity. I am absolutely convinced that the London Stock Exchange is, without a shadow of a doubt, definitely doing everything that it possibly can to keep ahead of this, but it does not take long to think of ways that the system can try and get ahead of you. I cannot see how you can keep up.

Alexander Justham: As you say, we make every effort to keep up and we have clever people as well. We dedicate a lot of time and resource to that. The previous participants made a very good observation, which, as the former head of the FSA division dealing with market abuse, I can absolutely verify. There are a couple of key things about algorithmic trading or computerised trading that are worth observing. First, two algorithms cannot go down the pub and agree something. That is the advantage of it. They do, by definition-

Q71 Mark Garnier: They can go down to the internet café.

Alexander Justham: Yes, they can go down to the internet café-the allusions could carry on, but they obviously do leave a record. That is one of the key things and one of the issues that has therefore been brought up.

Secondly, the core source of all information, on both sides of the Atlantic-I have spent a lot of time working with FINRA as well, having exactly the kind of debates you have, and the debates on cross-market surveillance are pertinent. We have spent, certainly as the regulatory world, a lot of time thinking about it, which is almost a variation of your concert party. The core source of insight more often than not, on pretty much all market abuse, is that someone is the loser, and someone-the loser-is as much the eyes and ears as anyone else. In fact, for the FSA, SEC and FINRA and everyone else, the core source of market abuse insight comes from suspicious transaction reports. It comes from people looking at it-but almost being on the other side.

You probably saw me slightly grimace when one of the previous participants named the FSA case, but then I remembered that it is public now. I left the FSA 11 months ago and it was not then. The insight into that came from someone whose algorithm was losing on the other side. It was quite an interesting insight. They were running an algorithm. It is always a fine technological debate about whether it is smart technology, and someone is just cleverer, or whether they are being abusive. We constituted that it was abusive. It has been challenged by the entity, but we-the FSA-deemed that it was abusive behaviour. It was the other algo who came to us and said, "Look, I have written this nice algorithm but I lose money all the time, and the reason I am losing money all the time is that these guys are beating me." So that is where that started, in that respect.

The fact is that the eyes and ears in the marketplace are multiple. We are a core component, and we will continue to do everything possible, as is our regulatory obligation, to get ahead of this game and to stay there. There are certain advantages. Technology makes some aspects of traditional market abuse much harder, as the previous participant said. But actually, the monitoring and managing of market abuse is a market-wide affair, and that is where the collective source ultimately manages this.

Q72 Chair: Could we move on to dark pools and off-market trading? There seems to be growing evidence that large orders are migrating to dark pools because of predatory algorithmic trading. This, if it continues, will obviously affect price discovery, as any use of dark pools does. Is there a danger that the ultimate losers from this will be the participants in the real economy or longer-term investors? You might like to comment a bit about the impact of the Turquoise dark pool, which you operate, and why you are confident it is not to the detriment to your main exchange.

Alexander Justham: The first thing we have to clarify is that dark trading has been a feature of the marketplace for years and years.

Chair: Indeed.

Alexander Justham: We sometimes used to have this debate. The word "dark" obviously comes with lots of connotations-Darth Vader and all those kind of things. There are quite a few bad words in the financial world-I will not go into naked short selling and all those kind of things-that conjure up certain connotations. Some people would say that dark trading is basically non-pre-trade transparency lit trading. That is too much of a jargon phrase so people adopted "dark".

The fact is that it has always been an element of trading. There is a careful balance between the collective whole, which generally would like, and benefits from, transparency, and the individual right of an investor to buy or sell shares without clearly indicating their entire position or their entire sense of direction and then be aggressively traded against. That is the origin of all of this. It is old as markets and the day; there is nothing new in this. There has always been a careful balance between the collective transparency desire and the right of the individual not to be significantly disadvantaged because they need to liquidate a large stake or anything else.

Ultimately, a lot of dark trading is designed to benefit significant investors. I think that is quite important. It is not designed for managing HFT or other such things-it is designed for investors to manage their positions. What MiFID did was codify some of the practices that had been in place for a long time on a variety of European exchanges, whether it was trading at a volume-weighed average price; whether it was trading at a reference price to the exchange, where you agreed a deal but just agreed it with reference to the main market; whether it was a historical, negotiated trade, which was a bilateral trade between two parties who agree to trade at a certain price away from the main market. Those were all things that were in place and codified through MiFID with the things called pre-trade transparency waivers. You can waive that pre-trade transparency and it becomes dark.

You are absolutely right to say that there is always, and always has been, a careful balance between that broad market good of transparency and that private right. Clearly, you cannot travel to both extremes in either direction; that is quite important. When we look at the developments that have occurred and the level of dark trading that exists today-we did a full survey of this at CESR and ESMA, so these data are not just for the London Stock Exchange, but for the broad European market-the proportion of dark trading through dark pools run by exchanges or MTFs was growing and was roughly reaching about 8% or 9% of the marketplace. At the London Stock Exchange today in the FTSE 100, our broad numbers are roughly 93 lit, seven dark. We look at our core MTFs competitors, whether it is the Chi-X BATS dark pool, whether it is the UBS dark pool, or the Goldman Sachs one. When we look at those entities and the dark trading for the more liquid stocks it is about 93/7. Going to many of Mr Garnier’s earlier questions, when you go down the curve, that gets less and less and less.

Q73 Chair: Which, the seven?

Alexander Justham: Yes. Because people’s dark pools are not operating in the liquid stocks. The liquid stock, as we can discuss later, trades in a very different manner. That probably would have been one of the things I would have slightly corrected from the earlier debate.

Has that grown from zero? No, because there clearly was existing dark trading. Has it grown a bit? It has probably grown a bit as it has been formalised. Before I came here, I looked at the numbers we had gathered at ESMA, based on 2009-10, which had reached that level and I looked at numbers we are trading at now. It has actually been static for the past couple of years. You are absolutely right to say that at either extreme this would not be a healthy environment. Are we in that position at this stage? No. Do we monitor that and think about that? I think that for us, the regulatory authorities and others, the answer must be yes.

I am sorry I did not answer the question on Turquoise. We inherited a very small dark pool from Turquoise when we bought our stake in Turquoise. It is pretty small.

Q74 Mark Garnier: Let’s talk about liquid stocks, because you went back to it. We are obviously talking about HFT and the top 350 shares or thereabouts, which covers-what is it?-90?

Alexander Justham: I wasn’t too sure about the precise number. You quoted 99. It is a very high percentage. It is in the 90s, in terms of market cap. You can appreciate that in terms of trading it is obviously very substantial. The vast concentration of trading in market capitalisation lives in the 350.

Q75 Mark Garnier: How many shares are listed?

Alexander Justham: We have roughly about 2,300 between our AIM market and main market.

Q76 Mark Garnier: And 350 represent 90-plus per cent. of the volume and the market cap?

Alexander Justham: We can provide you with the precise numbers, but it is all very substantially within those top 350.

Q77 Mark Garnier: It would be quite useful, because we are talking about just under 2,000 shares, which is an awful lot, which have come to the market because they want to have a primary listing. That is fantastic and that is exactly what it is there for and that is very good news. The problem is that once they have been sold, some of these stocks trade on an annual basis-one trade a year-and the order books I have seen of some of the smaller liquid ones have stuck around for a very long time. What are you doing ultimately to try to get some liquidity into these things?

Alexander Justham: We are thinking about it a lot. I spent quite a lot of my first few months at the Exchange working on exactly this. Just to clear up one issue, this is not high frequency. If you are trading once a year or even five times a day, you could take all the concerns of high-frequency trading, algo-trading, computer-based trading, right off the table.

What we are talking about is liquidity provision in the old-fashioned way. What we are talking about is a challenge between liquidity provision from an order book, which is in essence a democratic version where all participants can come in and make bids and offers, which is a very traditional way of market making. There is the core balance we have to contemplate and manage and, to be honest, it is very important, we manage this better than any other market anywhere else in the world. We are the only successful small cap growth market anywhere.

That is a genuine fact. They have tried it numerous times in Europe. They have all failed. Quite a lot of that core liquidity comes from traditional old market makers. You will probably recognise some of the names: Winterflood; Peel Hunt; Shore Capital. That dynamic-what the optimal liquidity provision is between an order book type structure and the traditional market maker structure, which is off the order book and in essence becomes much more of a bilateral-type trade-is the constant debate that we have. We offer different marketplaces and different market trading systems for that. We have SETS, which is in essence the order book, we have SETSqx, which is a combination of an order book and an auction-type programme, and we have the old SEAQ, which is, in essence, how life used to be once upon a time, before we had an order book. That brings me back to market makers.

It is a key focus for us, Mr Garnier. We recognise the importance. Coming back to our key role, that is to help corporates raise capital and help investors take part in their economic activity. It is critical to ensure that liquidity provision in the middle works. It is, without doubt, always a challenge within the smaller structures, as you go down the train there, but it is something that we have focused on. We spend a lot of time with the various liquidity providers. We are constantly trying to get more investors into that space. We are having a lot of debate with Government right now: stamp duty is a key issue, and conduct of business rules, again, are a key issue. If you classify a named stock as too risky to buy, people do not buy it. It is Catch-22: do you want to protect investors or do you want to let investors invest in the growth part of your economy? These are all the kind of difficult issues that need to be thought through and addressed. That is something that we are very actively engaged in, with the regulatory world and with Government.

Q78 Mark Garnier: I am going to change the subject and talk about the standards-this is the Parliamentary Commission on Banking Standards-of your members and the people who participate in the market. As you know, I go back to a time many years before big bang, when there was a very different type of Stock Exchange. Was your experience that long ago?

Alexander Justham: I did an internship before I went to university, which was before big bang. I left university and joined JP Morgan just after big bang. I am familiar with the world of the 1980s, but it is starting to be distant, even for me.

Q79 Mark Garnier: Over the years, have you seen a marked change in standards? Have you got a comment on general standards, as you see them from the point of view of the Stock Exchange, and how you think investment banks-what were merchant banks and stock brokers and jobbers and such-have changed?

Alexander Justham: I think that, in terms of relativity, it is an impossible comment to make. Maybe in the history of time someone will be able to make that judgment. Without any doubt-I have lived it quite directly, principally on the regulatory side for the past five years-there have clearly been a series of very high-profile issues within our markets. Actually, that has been less so within the trading markets with equities, to be honest; there has been, in fact, very little in that context. There are plenty of challenges, which we have been talking about, and plenty of keeping apace of change; but there has been no noticeable or visible change of the standards of ethics. We were talking about market abuse earlier; there has been no change of those standards within the trading markets.

You talked about our members: clearly, in the broader context there have been some very high-profile events, and your Commission, the regulatory changes and everything else are setting about addressing that. Can one make a time-relative judgment on that? I do not think that I can, to be honest.

Mark Garnier: All right. I will not push you.

Q80 Chair: Just a couple of specific questions on regulation that have come up. In the MiFID proposals, the European Parliament voted for minimum resting times for orders, on the basis-no, I will not say that.

Alexander Justham: You can say what the basis is.

Q81 Chair: I shall refrain from controversial comments. Do you agree with that?

Alexander Justham: No, we do not, and nor does the Commission. When the Council produce their answer, we do not anticipate that they will agree with it either. MiFID at this stage is very important, and it is producing its three different answers. We have had the Commission’s answer, we have had the Parliament’s answer and we have just had, sadly, a delay to the governmental Council answer, and then all three get mashed into one. We are still producing the various different answers.

The Commission did not propose that, and we do not anticipate that the Council will propose it either. This has come from just one of the three components of the legislative process. Why do we not support that? I am anticipating your next question. We do not support it on the basis that, as highlighted in the Foresight report, in essence one of the key things we are trying to help support is liquidity in our marketplace, coming back to all the issues Mr Garnier was referring to when you go down to more junior stocks. Passive liquidity-people who voluntarily put up prices-is a key component of that. It is akin to the old-fashioned market-making sense. What a minimum resting period does is actually to punish passive liquidity in favour of aggressive liquidity. It forces someone to hold out a price and lets people aggressively hit it. In the very thing that we have been debating today, this actually creates an even worse environment, so we would not support that at all. It is a proposal that favours one type of trading over the other at a time when we are all trying quite constructively to ensure that there is passive liquidity in our marketplaces.

Q82 Chair: What about standardised minimum tick sizes?

Alexander Justham: We do support that, broadly, for a number of reasons. Part of it is trying to manage some of the fragmentation across the marketplaces, and also trying to focus the points of liquidity a little better. If you have very fragmented, granular trading, where people are putting their orders in-it can be across the board-if you slightly force those steps it means there is more of a concentration of liquidity at a certain price point. In trying to balance the various constituents and components, in what we are trying to do-to provide the most liquid and effective marketplace for trading-we instinctively feel that harmonising that tick basis would be constructive.

Q83 Chair: Is there anything you would like to raise that we have not asked you?

Alexander Justham: No, thank you, Chairman.

Chair: That’s great. Thank you very much indeed for your time. It is very good to see you.