n+1 = innovation
John Fingleton CBE on how to make a success of the Regulatory Innovation Office
From novel business models that reorganise entire sectors, to breakthrough technologies that turn the unimaginable into the unremarkable – innovation is the driving force behind modern economic growth.
For decades, there was a strong case to be made that advances in science and technology were slowing down. Yet, recent progress in artificial intelligence, engineering biology and energy technologies – to name just a few – should give us hope that ‘the Great Stagnation’ really is coming to an end.
Welcome as new technologies can be, however, it is not always the case that our regulatory frameworks are able to appropriately govern them. Moreover, this deficiency can itself serve as an obstacle to the very innovation we seek to encourage.
In Britain, new initiatives like the establishment of the Regulatory Innovation Office (RIO) – which this week announced former Universities and Science Minister Lord Willetts as its first Chair – aim to provide a better system for properly regulating emerging technologies. Beyond these, there are also questions for how cross-cutting regulators can impact the rate of innovation.
I sat down with John Fingleton CBE, Chair of Fingleton and the former CEO of the Office for Fair Trading (the precursor to the Competition and Markets Authority) to get his perspectives on how Britain should go about regulating markets in a way that encourages more innovation and economic growth.
What we discussed:
Why an ‘n+1 regulator’ could unlock innovation in hard-to-define markets.
How the RIO can hit the ground running – and how it should work with other bodies.
Whether new market entrants can drive inter- and intra-market innovation.
What government can do to avoid different types of regulatory capture.
How regulators can embrace risk in a risk-averse society.
Ideas to improve competition policy.
How regulators should approach the AI sector.
Lessons for ensuring the RIO is set up for success:
Pick your priorities. The RIO won’t be able to cover everything, so it needs to prioritise sectors where it can have the most impact. These should focus on complex regulatory questions which blend the ‘supply side’ (where Britain has advantages), the ‘demand side’ (where outcomes will benefit consumers most) and what’s going on internationally (and where lessons from abroad could be learnt).
Start up, then scale up. It’s early doors, and right now the focus for the RIO should be getting up and running. The Chair will need to know how to navigate Whitehall and the other regulators, and be more focused on providing immediate leadership on establishing the necessary pathways, conventions, protocols and relationships than being across the minutiae of innovation policy.
Think local. The RIO should consider the whole range of players in the innovation-regulatory ecosystem, including Metro Mayors to pilot test cases.
Move quickly. What can often frustrate innovation the most are regulatory processes that drag on for months and years and don’t give clear answers. The RIO should place a premium on being agile. Speed can also be a way to circumvent society’s growing tendency for risk-aversion.
Full interview
You have called before for an ‘n+1’ regulator that could support disruptive market entrants – could you explain what that means, and whether you think the Regulatory Innovation Office is a step towards that vision?
The idea of the n+1 regulator goes back to about 2012, when I worked in the Cabinet Office and was advising on supply side reforms. The essential idea was that new business models come along, and the existing regulatory framework doesn’t suit them. That could be because incumbents have captured it, or it could be because what they’re doing is just more risky or has a different profile of risk.
You can think of examples like Uber and Airbnb in the sharing economy, as well as some of the fintech businesses that were trying to do different things like crypto, et cetera. But the point was that there are always new things coming along, and if the existing model doesn’t deal with them, this n+1 regulator could give them a temporary licence – say, for five years – or underwrite or monitor the risk, or check that liability insurance takes on the risk. Ultimately, it would be there to help them to navigate their convergence into the existing regulatory system over time.
Talking to ministers at the time about it, I quickly realised it was conceptually more interesting than it was practical. But then two years ago I was asked to give a speech at Oxford on regulation and productivity, where I set out what an innovative regulatory agency could do and how it would work in practice – and there are a few things in there.
One is a general view I’ve developed looking at the regulatory system in the UK, which is that we do too much in a sector-specific way and not enough in a cross-sectoral way. A good example of a cross-sector regulator is the Health and Safety Executive, or the Competition and Markets Authority (CMA) – they’re there to apply consistency of rules across all sectors.
Then you have all the sector-specific regulators, such as the Medicines and Healthcare products Regulatory Agency, Ofgem or Ofcom – and the problem with them is that they tend to become very wedded to things in their own sector and not look outside. But also, a lot of the biggest innovations that come along are often cross-sectoral.
The other issue is about the risk. Anything new that comes along that’s properly disruptive is going to have a different risk-reward profile from what currently exists. A really easy, tangible example is electric scooters on the streets. On the one hand, they can help address urban congestion. On the other hand, they can be a dangerous irritant. Both can be true at the same time – you can enjoy using scooters one day, and think “why did that scooter just almost knock me off the pavement?” and curse the person using it, even though you might have done the same yourself yesterday.
With these technologies, there is a question about how to manage that risk. Some people will get injured, some people might die, but I worry that if we invented, say, cars today, we would never get them onto the roads because the risk could be too great. We have to tolerate a certain level of risk for a certain level of reward.
This presents a big challenge for any individual regulator. If you’re the Civil Aviation Authority and you’re thinking about drones, you have one job – keep airplane passengers safe and keep cost down. Against those objectives, banning drones makes a lot of sense – or at least banning drones out of line-of-sight, or anywhere near an airport – because while the wider benefits accrue to others, the risks are very much owned by the regulator.
Part of what I hope the RIO will be able to do is to allow government to underwrite some of those risks and internalise some of the externalities that exist across regulators.
Does the RIO need more teeth to do that?
It probably does, and there are some low-hanging fruit that can be picked while it’s on a non-statutory basis – such as ensuring the right conversations are happening, or that the system is more agile. In many of these areas it’s more important that we make quick decisions – if you’re an entrepreneur, you’d probably prefer a quick “yes” or “no” decision. Even if it is a “no,” at least if you know it’s a “no,” you can get on and do something else. What people find very frustrating are processes that drag on for months and years and don’t give them clear answers. Accelerating the decision-making processes would be a good thing to do.
Legislating takes a long time, so setting the RIO up on a non-statutory basis and then sorting out the framework later is a no-regret policy. If you decide you want to put it on a statutory basis later, you will have learnt a lot about what it can and can’t do, and what exactly you need the statute for. That would be a practical way forward.
The Government is saying it will focus on four areas – engineering biology, space, drones and AI in healthcare. You mentioned before that it’s pretty hard to pin down specific sectors when we’re talking about innovation, so might that contradict the current approach?
Picking a number of pilot areas to focus on when you’re getting up and running is the right approach. It is important to select areas that can provide good pilots or testbeds for what might later be rolled out more widely.
I can see why people who are not in those four areas might think “oh, why not my area?” but ultimately you have to pick some somewhere to start with. It is always better to prioritise earlier, even if you get it a bit wrong.
Given limited resources, how would you approach selecting those sectors?
There’s probably a number of things that are important to the prioritisation.
One is “do our universities produce technology and do we have innovative companies trying to do things that are commercialising research out of our universities?” That’s the supply side – and something we haven’t always been good at in the UK. We have fantastic research and we haven’t always been good at commercialising it.
In AI, two of the recent Nobel Prize winners went to British universities and a lot of the research is done here. In engineering biology, we have some of the top research in the world. In drones, we had a fantastic Industrial Strategy challenge and we have a lot of technology on drones in the country. In space, as well, there’s some interesting stuff happening.
The second part is the demand side – “will people see a difference from this?” and “what’s that going to mean in practice at the end for consumers?” You don’t have to be a consumer of drones to see the impact they’ve had in the war in Ukraine, and the positive impact they could have for delivery of medicines in remote areas and other applications.
Think about something like climate change – one of the problems we have with nuclear power is that it has become very costly because of safety regulations. But drones could help bring down some of those costs, and anything that’s going to take out the cost out of other processes can be presented to the public as “this is having a positive benefit in your life.” As an aside, energy costs in the UK are a big drag on the economy – and even if they’re incremental, lots of incremental things to reduce energy costs that are innovative will be quite good.
Third, there has to be a regulatory challenge between the supply and the demand sides that we haven’t already met. For all of the sectors RIO is focusing on, there’s quite a lot of interesting regulatory complexity.
A fourth factor is “what’s happening internationally and what can we learn from what others are doing?” Regardless of one’s views on Elon Musk and Donald Trump, I think the space sector in America is going to become much more interesting. When you look at what SpaceX and others like Boeing are doing, I think that’s going to make for interesting learnings for the UK.
That’s probably also true in drones, especially given the dual military use that’s emerging – dual-use technologies are incredibly important, and in all of those areas you probably have quite a lot of international learning as well.
Those would be the four criteria I would use to prioritise something – and all four of those areas at least seem to be plausibly aligned with the type of priorities that have been agreed. I don’t know what prioritisation criteria are actually used, but I imagine they were something like those.
Have you got any thoughts on how the RIO – either now or as it potentially becomes more powerful – interacts with the CMA, the Financial Conduct Authority, other organisations or government bodies?
How the RIO interacts with sector-specific regulators could be that there’s not just one player but rather a set of issues that innovative players face. There’s a regulator engaging with them at some level – maybe with a sandbox or another mechanism – but in some sense, the sum is less than the parts.
Then there’s the question of “what needs to be done to cut through this?” and “what is the obstacle to making regulatory decisions?” – and there’s a little bit of adjudication in this space. Very often, the regulator will say “we’d like to do more, but we don’t feel we have the mandate,” and very often the firms will say “the regulator could do more and what you’re saying doesn’t actually stack up.”
So there’s something to be said for digging into that and just understanding “is there a solution? What’s the solution? And who needs to own it?” That’s the sector-specific piece – and the Financial Conduct Authority would be in that category.
With the CMA it’s very different, because the CMA’s role is to think about how markets work generally – within a consistent, horizontal framework. A large part of the role in the CMA’s thinking is “how do we ensure that competition works to get these technologies to market?” and recognising that competition in markets actually is one of the best drivers of adoption for new technologies.
We know from the work of John van Reenen and Nick Bloom and others that we have this long tail of less productive, even inefficient, businesses that don’t adopt new technology. It has always been a conundrum in my mind as to why competition doesn’t drive the between-firm effects that we know drives productivity growth, where we have more efficient firms replacing less efficient firms. After all, we do have a pretty competitive economy in the UK and the CMA doesn’t often find big cartels.
This is an area that the CMA is interested in – it’s directly relevant to growth, and I don’t actually see any trade-offs between competition policy and growth. The question then is “should the CMA be looking at different markets or different areas?” It’s already looking at AI, for example, but shouldn’t it be thinking about “how do we make sure we have greater competition in this area? Or greater adoption and diffusion?” And then on the consumer protection side “are we ensuring that contracts are fair and reasonable?”
But for regulators like the CMA and some other horizontal regulators like the Health and Safety Executive the question is “how do you get this right across the economy as a whole?” And there should be more allies of the Regulatory Innovation Office than obstacles.
The AI White Paper has set out a very clear sector-specific approach to regulation which the UK has been praised for, so how do you think the government should approach AI regulation?
AI is a general purpose technology just like the internet was. We now have systems that have regulated various online aspects of life and they have coped quite well. Take online shopping – we had very generous rules on the return of goods bought online that we didn’t have in store, which was to try and facilitate and protect the customer in those relationships. Section 75 on credit cards also really helped because it meant that if you spent more than £100, your credit card protected you. Regulations like that can be helpful with the adoption of the new technology because they build consumer confidence.
It’s almost a trivial thing to suggest now in retrospect, but there are parallels there with just making sure that consumer protection and the adoption of AI in the individual areas where it gets applied is going to support the adoption of things by consumers where there might be extra risk for them. Outside of that, I think the sort of existential issues around AI destroying humanity are not ones I’d be trying to legislate for now.
The CMA is obviously looking at questions around how the market might evolve, and dominance by a small number of players. But it’s quite difficult for a regulator to decide they want to have a five-firm competitive market in a technology like AI at this stage of the process. What the CMA has done so far is not unhelpful in terms of just understanding the vertical stack in AI, where the bottlenecks might be, informing how they might look at mergers in that space.
In competition policy there’s a great deal of ex post regret about Facebook and Instagram, Facebook and WhatsApp, Google and Waze, Google and Fitbit, Google and DoubleClick and so on. Though even going back, the CMA did commission research to ask “given the information at the time, could we have ever made a different decision?” and the answer was “probably no.”
There’s also a question in my mind as to what competition policy can do in some of those areas. But there’s now greater vigilance about new technologies and regulation and we have to make sure that that vigilance doesn’t stop innovation progressing. If I look at the way the EU has regulated GDPR [the General Data Protection Regulation] – which is important to protect privacy but is largely an irritant for most people – it has also had profoundly negative effects on competition for two reasons.
First, the big tech companies were all well positioned to comply with GDPR. They had IT systems in place to manage data requests and to delete data on request. In contrast, smaller companies did not have those systems and, to make matters worse, often had to go to the big tech companies to re-engage with their customers. So GDPR had a negative effect on competition and [Mario] Draghi more or less says that in his report.
I’m worried that the approach that the European Commission is taking towards AI will impose more costs in the short-term than the benefits it will gain. It’s not an area where I would rush to legislation – I would learn from others, and then if I’m legislating, keep it as flexible as possible.
The RIO’s going to get a new Chair. What advice or what skills do you think they will need to have? [N.B. Our interview took place prior to David Willetts’ announcement as the new Chair of the RIO.]
It is helpful for them to understand how Whitehall works, and how regulators work. This is not about running it in ‘steady state’ – this is about getting it up and running and establishing the pathways, the conventions, the protocols, the relationships and so on.
Understanding something about the innovation ecosystem is important, but that’s something the Chair can draw on from others’ expertise. What matters right now is more about leadership and vision, prioritising the right things and getting some early successes.
Reflecting on the success of sandboxes, or some of the limitations you found when you were involved in setting those up, what lessons could be learnt for the RIO?
I have not seen an evaluation of what sandboxes have done well and haven’t done well.
Would you have liked to see that?
Yes – certainly for the Financial Conduct Authority one that was inspired by the n+1 regulator back in 2012. Now, having a sandbox is sort of a no-brainer for a regulator, but there are two big questions in my mind.
First of all, “are the businesses coming through really innovative, or just smaller and a little bit different?” I’m not totally informed, so this is a non-scientific, non-rigorous, non-evidence-based view – or just based on my own slice of it – but in financial services we’ve been incredibly successful globally with B2B fintech. There are a lot of really good examples of people doing interesting, clever things in supply chains that nobody knows about. On the B2B fintech side, we have this invisible, bottom of the iceberg that’s big and successful – but in B2C fintech, we have the tip of the iceberg that’s smaller and more visible, but in my mind slightly less successful.
Take banking in the UK – I do think the fintechs have had a hugely positive effect on improving the existing banks. I was advising one of the big four banks during the CMA banking enquiry ten years ago, and I saw how the people running their digital app and customer service were obsessed with what the fintechs like Monzo, Starling and Revolut were beginning to do to improve their user interface.
The fintechs were getting people looking at their bank accounts three or four times a week, whereas the banks up until then had people looking at their accounts once or twice a month. People looking at the bank accounts on a very frequent basis is very good because they’re less likely to go overdrawn, they’re going to manage their money better and they’re going to get better interest rates because they’re moving their money around. So, consumer engagement with apps went up enormously, and the legacy banks’ apps really improved as a result.
So you can point to big productivity benefits from new entry – and the question in fintech is “how big has it been?” and “is there more, or is that it?” and the jury is still out on that in the B2C side. It’s a little early to say on B2C fintech, but I do wonder if we could have done more over the last ten years in that space.
On all of this, I draw a parallel with when Ryanair, EasyJet and other airlines entered the market. Air fares used to be very high, and during the 1990s there was quite a lot of competition but not much happened.
The European Commission had always caved in to national governments on giving state aid to airlines because it was inconceivable that the national flag carrier would be allowed to go bankrupt. So the likes of Aer Lingus, Sabena, Air France, et cetera were all very badly run, and they all just got bailed out every few years by their national governments.
Then 9/11 happened, and the European Commission said to the Belgian Government “we’re not allowing you to bail out Sabena,” and Sabena went bankrupt. I think Aer Lingus – which was then run by Willy Walsh – managed to take out 30% of its cost in the next two years.
Now one of the big benefits of new entry into markets, and particularly any type of efficient, innovative new entry, is not just the fact that people switch – the within-firm effects – but actually the efficiency effect it can have on incumbents.
But outside of fintech – for example, let’s take defence – I don’t think that the competition and innovation that’s happening in the startup world has had an effect on innovation that’s happening within incumbents or MoD [Ministry of Defence].
I’m not an expert on that, but I would highlight the role that government procurement plays in doing this – and this is something the last Government and this Government is trying to get better at. How do you get big, government IT systems to be innovative? I suspect the same thing is true of the MoD – how do you bring in innovation? DARPA [the Defense Advanced Research Projects Agency] in the US has been very, very successful at pioneering some of that, and it may be an area that ARIA [the Advanced Research and Invention Agency] here can make progress.
Going back to drones and space, it’s good that those areas are in the immediate priorities for the RIO because they will have interesting defence applications, and if you can think about drones, defence, and the right regulatory and procurement framework, then maybe that gives you a little bit of a pathway or blueprint for thinking “how would we generalise this to other areas?”
Defence is also one of those areas where because government is the main buyer, procurement may matter more than regulation – but regulation is still in there as well. So an interesting question which I don’t know the answer to is “what is the RIO’s ability to deal with procurement issues in an area like defence when it comes to drones?” As they start to formulate which government departments will get involved with the RIO, it wouldn’t surprise me if the MoD was going to be part of the discussion with space and defence in there.
How would you think about regulatory capture, and how can we avoid that with the RIO? There’s a traditional view of regulatory capture in areas like defence, but even innovative firms can still find a way of capturing politicians and civil servants because they’ve raised a lot of funding or so on.
On the ‘capture by the incumbents’ piece, you have to apply a very critical lens to claims that are made. For example, the US government has introduced a regulation that requires airlines to make automatic refunds to passengers when flights are delayed or cancelled. One of the obstacles to doing this in the UK is that airlines say “oh, well, people sometimes change their credit card before they’re due to fly, so we wouldn’t be able to do the refund.” To my mind, that counts as a spurious argument – there has to be a way of dealing with that. You can see this in rail too. But if we had a better automatic compensation system, it would change the incentives of the people running those services to supply them on time, and we’d get better outcomes for consumers.
People will always play the safety card and some spurious consumer protection card to protect an existing regulatory regime that maybe is no longer fit for purpose. That’s where the RIO could push back on regulators and say “have you challenged that properly? Surely there’s some solution?”
On the ‘capture by the entrance of specific technologies’ piece, it’s incredibly important that you’re thinking about the technology and the set of future innovators, rather than one specific one. It’s fine to use one or two specific ones as the test case for that technology, but you also have to be wise to the idea that this person wants to sell a very specific solution and that the regulatory gates only get opened for that solution.
That’s a bigger risk for politicians than for regulators, because somebody who has a good story to tell and a stakeholder network may be very compelling to a politician in a way that they may not be to a regulator who’s dug into the evidence. That has lots of positives as well, because when a politician says “I’ve met these guys, I think what they’re trying to do is great”, it’s good to be able to reply “yes it is, and we agree with it, but we need to do it in the way that keeps things open for everybody.”
When there’s risk involved, there’s a question about where the risk is owned and where the buck stops, because with a lot of things that are truly innovative, they have to do something different – you don’t make an omelette without breaking eggs.
If there’s a new technology that’s going to save 10,000 lives a year, but there’s a downside risk that it’s going to kill five people, the real risk is that you obsess with the five people that might die and you forget about the others who’ll benefit. I suspect this may happen with drug trials – which have become slower and more cumbersome because we’ve become more risk-averse as a society in the last 50 years.
We’re not easily going to change society’s preferences on risk, but sometimes getting decisions made more quickly, quantifying the risk and understanding how to mitigate it is part of the solution. It’s not that we’re not going to bring the technology to the market, it’s just that we end up delaying it by five or ten years rather than just getting on with it and saying “okay let’s do it, and let’s learn from it.”
How can the RIO or other regulators better understand what’s going on in innovation? It sometimes feels like people in government are scared to go and talk to entrepreneurs and innovators because they might get accused of doing the wrong thing.
When I was on the board of UKRI [UK Research and Innovation], I thought that UKRI, Innovate UK and the research councils could play a much greater role in identifying some of the regulatory barriers that they see, and particularly seeing them in advance. When you’re looking at things that are at early TRL [technology readiness level] stages, you should be thinking “what’s going to be the obstacle down the line here in terms of regulation?” Getting early insight on that is less risky than looking at it from the point of view of an individual company.
Stakeholders always come to government looking for money, probably because it’s really easy to ask for. I’m always a bit sceptical about that, because as an economist, I always worry about rent-seeking. But if government could change regulations as easily as it can hand out grants, people would come to it and ask for change. The whole system requires people who know about this – whether that’s ARIA, Innovate UK, research councils that are part of UKRI, Catapults – to have the ability to come forward and say “we want to do this.”
We should also be thinking about how in the UK we can make better use of local pilots. For example, we talked about electric scooters – it may be trivial but it’s an example that most people can relate to because they can see it on the streets. John Kingman, who used to run growth at the Treasury, wrote a great piece about economic growth – and he talks in particular about the Metro Mayors and what they could be doing. You could imagine a Mayor saying “we’re going to pilot autonomous vehicles, or drones, or some aspect of drone delivery in our area – and we’re going to try and create that regulatory framework at a local level.”
So how can you try and have those kinds of conversations with the RIO as well? If you wanted to completely revolutionise local transport delivery logistics at the level of a town, where would you start? This could be a big role for the RIO, but also for local Mayors and everybody in between – but the RIO shouldn’t just become a talking shop. There’s a question about how much to engage with that level of broader discussion and debate, and how much to say “we are very focused on these issues. If you’ve got a concrete proposal on this, come and speak to us.”
How should the CMA approach competition concerns in AI markets, particularly regarding vertical mergers, and what lessons can be drawn from past regulatory challenges?
It’s unclear to me whether the fundamental economics of the AI supply chain have the same sort of tipping effect towards monopoly that you get with something like a social network or search. There are clearly parts of the supply stack that are always going to be highly concentrated – chips, for example – but it is possible that a lot of the various components can be kept fairly separate.
This poses an interesting question for the CMA, which is “what’s your approach to vertical mergers in that setting, as opposed to purely horizontal ones? Is that something you want to try and encourage, or discourage?” Unfortunately, that’s quite fact specific and anything which is that fact specific is going to have that lack of clarity up front – and it’s only by doing some cases that you get the clarity, but cases are very expensive for the parties involved. One solution to that might be that if there are interesting test cases with small firms, could some of the costs of that be borne by the state because of the clarity it would give?
Back in 2011 we had this problem with local newspaper mergers. It was quite a political problem, but basically lots of local newspapers were looking to merge with the other local newspaper. Often one was a paid newspaper and the other was a free newspaper. Therefore for the OFT [Office of Fair Trading], which I was then running, there was a question of “was there a reduction of competition in the market for advertising?” Of course there were categories of things – death notices for instance – that people only wanted to advertise in local newspapers, and if you only had one newspaper, the prices for advertising could go up.
The problem was that the cases were all so poor. They could get to Phase One and we’d say we need to do an in-depth investigation to get to the bottom of it, but they could never go to Phase Two. I remember Jeremy Hunt, who was the Secretary of State for Culture, Olympics, Media and Sport, asked me “what are we going to do about this?” So we had a roundtable of the industry and so on, but then we had this nice case called Nottingham-Topper where we actually managed at Phase One to do quite an in-depth analysis and clear the deal.
That’s the same conundrum we have here. Can the CMA do an in-depth Phase One investigation on some of these complex issues, and can it get it right earlier? That’s a reasonable question to ask, particularly in some of these fast moving areas. The other reasonable question is “if there’s a set of cases where a real precedent can be set, is there some way of insulating a smaller company from the effects of that financially?” Because you want them to go through the process, because there’s a positive externality that we don’t often think about.
We ask all our guests the same closing question: what’s one interesting thing you’ve read or listened to recently that you’d like to share with our readers?
While I have not read the book, I did like Jason Furman’s review of Marc J. Dunkelman’s Why Nothing Works: Who Killed Progress―and How to Bring It Back.
 | A guest post by
|