Dr Jonathan Cave on his latest book on ‘next generation’ free trade agreements and why he became involved in economic regulation

What projects are you working on at the moment?

I’m one of the more externally engaged members of the department – I have at least two other jobs – maybe three! As a Turing Fellow I do a lot of things on data analytics, ethics and algorithms. I’m also affiliated with various think tanks – I used to work for RAND/RAND Europe and still do some projects with them, for a policy consultancy in the Netherlands and directly with the European Commission and Parliament. I also participate in Warwick-based research with colleagues from e.g. the Medical School, Mathematics and the Cyber Security Centre.

I’m just starting a project on product safety standards in a globalised environment - the regulatory picture is confused because online purchases can appear to be located within a particular regulatory space, such as the Single Market, when they are not.

Other projects with which I am involved concern ‘internet of things’ devices, where product safety refers both to the physical device and to the way it is used. What cannot easily be observed or monitored – and a major headache for market surveillance and other authorities - are the software that runs on those devices and the data they collect and process. Software, for instance, is push updated without users’ knowledge. But that’s not the end of the story - Alexa, Siri and their ilk listen to you and learn from what they hear in ways that are not audited or explained and which may not be audited or explicable.

I’m also an economist member of the government’s Regulatory Policy Committee, which is tasked with scrutinising and validating impact assessments of new regulations – expected and attained burdens on business, consumers and society as a whole. The objective is “Better Regulation” – decisions that take into transparent and proportionate account complete and robust evidence, objectively analysed. I’m in my second four-year term and get to see all significant regulations and activities of regulators. I don’t do the impact assessments themselves, but scrutinise them to see whether they are fit for purpose. As a result, I also work across government and internationally to develop and improve better regulation frameworks and impact assessment methods. Recently, this has involved international collaborative governance and impacts on trade, competition, innovation and the environment.

I’m just editing a book (in which I wrote 8 chapters) based on an EU-US project looking at collaboration on 5G, Internet of Things/cyberphysical systems and big data and AI.

Another line of work concerns ‘evidence-based’ policy; even before Covid-19, many important bits of legislation are highly controversial precisely because of their expected impacts and the quality of the analysis supporting them.. I’m currently writing a paper on how practice came to differ from the ideal picture assumed in the frameworks and how things might be rebalanced. One example concerns (ex-post) evaluation; there is a tendency only to collect and analyse evidence on things that were initially expected to be significant and many emergent or simply overlooked aspects are missed. Unintended consequences become unmeasured consequences. Additionally, political reality ensures that the people who knew and cared about the policy have been rotated out or focus on the current crisis or big idea, so institutional dementia sets in. The same mistakes get made time and time again.

I’m particularly interested in the impact of algorithmic decision-making in economic contexts from financial trading to price-setting e.g. for airlines or holidays (remember those?). One example is algorithmic collusion; if you and I are competing and pricing dynamically, I will use feed information about your prices to a revenue management algorithm to set my prices. One approach is reinforcement learning, which uses a model plane and a data plane. Exploitation phases when I use the model to set (currently) optimal prices are interspersed with exploration phases when I vary my prices to learn about the current state of the market, adjusting the model to improve a specific objective . If only one firm uses this approach, it will learn its residual demand curve; if many firms do this (ignoring each others’ learning), even fairly simple models can converge to super-competitive prices (albeit often short of the monopoly price). This is not generally illegal, since there is neither communication nor intent to collude. We just use algorithms that prove profitable. But regulatory economics should consider whether this should be illegal, how it might be detected and what can be done to prevent associated harms.

This convergence is most likely – in a model where firms ignore each others’ learning -if they all use profit as the objective that guides reinforcement learning, though this depends on the network structure of firms’ market (and observational) interactions – which firms’ prices do I observe and which firms’ behaviour affects my profits (these need not be the same). But a really interesting phenomenon is that changing the objectives of only a few firms (in critical network positions) to include consumer surplus as well as profit can lead the whole system to converge to much more competitive behaviour. We don’t have to force virtue on everyone – we can work with evolutionary and learning processes to improve compliance and competitive health. New research on trophic coherence, led by the Warwick Mathematics Department, will further develop methods for understanding this automated interplay of structure, conduct and performance.

This has lessons for e.g. banking regulation, where standards for network structure and algorithmic practices can promote resilience and stability while inhibiting tacit collusion – and do this using fairly effective, light-touch and ethically robust alternatives to coercive controls. [This is not a panacea; models in which firms take account of each others’ learning are considerably less reassuring]. But ultimately, the direction of travel is towards innovative, light touch regulation, and a better way of understanding how policy is likely to play out in practice, which is – or should be - of direct use to those who design and implement policy.

I’m just editing a book (in which I wrote 8 chapters) based on an EU-US project looking at collaboration on 5G, Internet of Things/cyberphysical systems and big data and AI. One interesting aspect concerns ‘next generation’ free trade agreements that pay considerable attention to technical and non-tariff barriers to trade (which include regulatory alignment). Many domestic regulations and standards (e.g. on product or food safety) can create market barriers – at least initially. But mandating adherence or creating strong mutual recognition arrangements can affect both domestic and international standards. The stock of domestic regulation should gradually be cleared of disguised protectionist measures. Other domestic regulations embed important societal values (e.g. on privacy or IPR), some of which we’ve adopted when forced to by trading partners. These values could spread out across the world, to the extent that our industries trading abroad and foreign firms seeking to sell into UK markets find it too expensive to create multiple versions of the same good meeting different standards. Regulatory regimes find an equilibrium mediated through the marketplace – the trick is to race to the top rather than the bottom.

What interested you about this area of research?

I am fascinated by governance as a property of economic interactions, and the impact of information and cognition or analysis.

When I went on the job market after my PhD I had three job offers; a fellowship at the Hebrew University Institute for Advanced Studies to do game theory research, an Assistant Professorship at the University of Illinois and a staff economist job with the Federal Trade Commission (FTC) in Washington DC. I also had an offer from RAND in Santa Monica, but I could not imagine living in LA. I planned to decide during a total eclipse in Oregon – the moment of totality passed without a decision, which I took as a sign that I take all three jobs (in series, not parallel).

I got to the FTC in time for some fascinating confluences of policy and theory relating to facilitating practices and information remedies for product safety; I got to work on automobile safety, the ‘Ethyl case’ (most favoured customer clauses facilitating collusion) and cigarette labelling. It was a really exciting time where economic theory and experimental economics could actually influence policy and through that, people’s lives.

I became fascinated with regulation as a complement to help markets to identify and facilitate efficient behaviour. This interest in policy ultimately led me back to California to work for RAND – first to work on self-enforcing agreements (like the international energy agreement and fisheries treaties), where theory could lead practice rather than giving an account after the fact. This led to my work on cartels that vote; using game theory to design mechanisms that would make the world a better place or tell you more about the world as it was.

I love regulation, because like industrial organisation it’s a clear application of game theory, which appeals to the lazy person in me who wants to have one way of thinking about all these issues rather than many different specific models.

Why did you decide to become an economist?

There’s a sense in which I am fulfilling my mother’s ambition – she was an economist of the John R. Commons (institutional economics) school. As far as she was concerned, however, my theoretical economics wasn’t real economics because (as she often said) “it had no people in it, just mathematical representations).

I went to Yale as an undergraduate and initially majored in molecular biophysics and biochemistry. But this changed – perhaps because my dad was a chemist and my mother was an economist, I was attracted by Yale’s combined sciences programme, which let you create your own degree if you could create a coherent topic, recruit supervisors and take enough credits to satisfy both sets of requirements. The overlap I found was systems that organise themselves. As a result, I wound up doing degrees in chemistry and economics.

In that era, education strove to nurture the desire to solve problems. And that requires a range of perspectives. And I can trace that through to game theory, especially communications games on networks, which I do almost as a side-line, but also to involvement with policy. Part of it is because I believe in the issues they address – a climate that sustains us, an economy that promotes efficiency and equity, etc. – but part of it is that I cannot really understand how human systems function unless I understand how people whose decisions move the needle think.

Why did you join the Economics Department at Warwick?

After Yale I went to Turkey and Munich for a sort of gap year. When I finally got to England, my economics adviser from Yale (Ross Starr), was at the LSE on sabbatical. He said ‘You like general equilibrium - why don’t you work with Frank Hahn at Cambridge?’

Uncle Frank suggested that I do a degree. Cambridge did not recognise Yale degrees, so I enrolled as an undergraduate for the Tripos. Overseas student fees were quite high, so I spent a year working at the Bank of England, thanks to Marcus Miller. The following year, when studying at Cambridge, I became aware of Warwick; Marcus and my BoE colleague Mark Salmon were affiliated with Warwick, and I also used to drive across to attend Zeeman’s lectures on differential topology.

I found the place was delightful - intellectually delightful, because you can have conversations that I would pay to be part of, and they don’t mind if your interest doesn’t go all the way to applications

I formed an attachment to Warwick at that point as a place for conversations that I would pay to be part of, on all sorts of topics. An additional advantage was possibly the nicest proper campus in the UK. Later, when I had a sabbatical from UCLA, I immediately thought of Warwick as an intellectually vibrant place, where I had friends and could ignore departmental stovepipes. So I asked to spend my sabbatical year here, and they said yes.

I found the place delightful not least because you could develop ideas theoretically, or in applications without having to pigeonhole things too soon. The students are also quite lively and there are some absolutely first-class minds. I thought this was a good place to be; as I began to get drawn more and more into the world of policy, I realised this was a good base – if I needed to find out about law, politics, macroeconomics, complexity science, etc. there are people here I can talk to and – as interest and opportunity allow – work with. I wish to continue to engage with students, I can hire them to work on my projects. So it seemed like a perfectly lovely base academically and intellectually, and in an open and unconstructed environment.

And finally there’s the physical environment – back when it was open, I loved the fact I could reach the RSC in 15 minutes; I live both in the country and just outside of town, and cycle to work along the bike path. When I go to London, it is a short cycle ride to Kenilworth train station. It’s a perfect place to spend time. And when I cross the moat that leads to Kenilworth Castle and head up the lane home, that feeling of leaving one part of the world behind and joining quite another is absolutely wonderful (even in lockdown). I can’t imagine a nicer place to be – and I’ve been in a lot of places.

You can learn more about Dr. Jonathan Cave, including a list of his recent publications here.