New thinking in regulatory economics

Would you run for your life or stand still if you came face-to-face with a bear? FCA chief economist Peter Andrews looks at how hard it is to understand human decision making and assesses the implications for policy makers.

Location: The European Securities and Markets Authority, Paris, France.

I am very grateful to Verena Ross for giving me the opportunity to speak to this expert group. The views I express are my own not those of the FCA, and I will start with a personal statement.

From my past work with Verena, from people I have seconded to work here, from people I ask to come here now to co-operate on CEMA’s cutting-edge research projects, and from my dealings at IOSCO with your Chief Economist, Steffen Kern, I have a lot of confidence in the European Securities and Markets Authority (ESMA).

I am sure you can do great work in your mission to help safeguard the stability of the EU's financial system by enhancing investor protection and promoting orderly financial markets.

In fact I believe that regulation designed, like yours, to help financial markets allocate capital fairly and efficiently can do far more for economic growth and stability than is generally appreciated. And there is plenty of evidence in Joseph Stiglitz’s 2015 book ‘The Roaring Nineties’ to make anyone pause for thought before claiming that financial markets will do this all by themselves.

The scope for macro benefits of markets conduct regulation is real, including through measures to enhance competition. In the case of over-the-counter (OTC) derivatives, a 2013 report by the Bank for International Settlements estimated net macro benefits arising from a mixture of conduct requirements, meaning here changes to clearing and collateral practices in the market, and capital requirements to be about 0.12 percentage points of GDP. I am going to address a key topic today: How do people take decisions and what does this mean for regulation?

In other words, I am going to consider how we might influence the demand side of markets and the supply side of markets to make choices that further our objectives as regulators. The supply side material is not included here. Given the weight of regulatory process and the challenge of dealing with multiple stakeholders, it is worth reminding ourselves that the central task of market regulators is to design remedies that truly correct or, at least, offset market failures. Our success depends on making markets work well.

How do people take decisions and what does this mean for regulation?

First, an obvious question with an obvious answer. The obvious question is: what is the importance of people’s decision-making for regulators? And the obvious answer is: people’s decision-making is of critical importance to regulators because most regulation is designed to change decisions which are self-harming, harmful to specific others or harmful to society as a whole.

The valuable contribution of behavioural and traditional disclosures

The answer just given has implications for regulators. In particular, if we want to change decisions for the better, we need to understand how they are made and what might influence them.

The FCA has been very active on this front. An especially useful document on influencing decisions that you can see is our Occasional Paper 23. This summarises the results of a range of behavioural interventions designed to improve consumers’ decision-making.

Changing the minds of a high proportion of consumers through interventions based on today’s ‘behavioural economics’ is typically rather hard.

It shows that changing the minds of a high proportion of consumers through interventions based on today’s ‘behavioural economics’ is typically rather hard. One can see similar results in Stirling University’s excellent ‘nudge database’, which summarises the outcomes of one hundred high-quality (randomised, controlled) trials of behaviourally informed interventions.

The broader position is more challenging than these documents narrowly suggest. The reason why we invested significantly in behavioural work is that we had good reason to doubt the efficacy of traditional disclosures designed to be read by Homo Economicus.

This does not mean that traditional disclosures are entirely unhelpful. They were often based on extensive consumer testing and contained valuable information that some people used. And in some markets influencing the marginal consumer protects all consumers. FSA Occasional Paper 32 (2009) uses differences between sales channels to provide evidence of increased shopping around following the introduction of meaningful price disclosure.

Also, it is hard to imagine that prospectus and listing disclosures are not extremely important to the functioning of the relevant markets. It is just that information disclosures did not deal with all the harm we could see.

Nor do our behaviourally informed disclosures, as FSA Occasional Paper 32 mentioned above clearly shows. To be sure, the results in Occasional Paper 23 are not due to any kind of poor practice. Far from it, in fact. We work with global experts and take a strict empirical approach to setting behavioural remedies. And the science on which the remedies are based is clearly a useful advance on the idea that the representative agent is somebody who parses contracts and uses the results to optimise their set of choices.

But, equally clearly, what we have done so far is more a beginning than an end. We need somehow to find more impactful remedies. So I will say a little about the nature of the problem of getting people to make better choices. Also, since my topic is new thinking in regulatory economics, I will say a little about the fresh avenues that can be explored. Avenues which one day may give rise to fresh and better remedies for regulators to use.

Trying to get a realistic perspective on consumer decision-making in finance

Given that people’s decision-making really does matter to the regulator, I am going to explore the problem of getting people to make better choices by using a story about decision making that is based on personal experience. 

My aim is to show just how hard it may be to understand decision making – because of individual characteristics and intrinsic complexities - and why we regulators might under estimate this problem. Suppose I say that you are an intelligent group of people with excellent technical qualifications in finance. I am not flattering you. Rather, I am recording an important fact.

Let’s assume that I also fit into this group. It’s a defensible proposition. I have qualifications and experience in accounting, taxation, finance, law and economics. When I consider consumers’ financial decisions, these things can help, and perhaps also hinder, me.

Moreover, I am aware of Kahneman’s ‘Thinking fast and slow’ and that sometimes I might be prey to biases or utilise heuristics in System 1 thinking. But I have a comfort blanket. Not only do I know that there is a difference between System 1 and System 2, and that in some circumstances System 2 can be really useful, I have the tools consciously to switch to System 2 whenever it suits me to do so. And so do you.

The result is that I – and perhaps you? - struggle to imagine financial decisions as decisions often, even typically, made by people experiencing strong emotions arising from a fundamental sense of uncertainty, lack of relevant knowledge, and strong fear of very negative consequences if a mistake is made. Luckily, I met a hunting Grizzly Bear while above the tree line in the Canadian Rockies last year and suddenly the complexity of decision-making under uncertainty, ignorance and fear of mistakes became much clearer to me.

This experience, coupled with an increased awareness of new work in emotional finance, anthropology, sociology and neurobiology, has helped me to understand that, while the System 1 versus System 2 analysis of decision making tells us something important, it is also too simple to capture everything important that is going on. At least for regulators who want to understand and influence real-world financial decisions made.

For example, Kruglanski and Gigerenzer in ‘Intuitive and deliberate judgments are based on common principles’ challenge dual process theories of decision making, which they characterise as involving intuitive and deliberative judgements. They provide evidence and arguments that both approaches are rules based and that the rules in both cases are the same. Thus a unified theoretical approach to both is warranted.

Beierholm et al in ‘Separate encoding of model-based and model-free valuations in the human brain’ also say that neurobiological foundations for the existence of a dual system have not been established but claim to have found the first neuroscientific evidence for a dual system. This was in 2011. I understand, though, from a recent lecture by a top academic psychologist, that the claim is not considered to be decisive and from my own exploration of relevant, new literature I can see that the scientific consensus is increasingly that the System 1/System 2 approach is a weak reflection of reality.

Materially for our purposes, given the degree of uncertainty in financial decisions, Beierholm et al also found ‘that subjects tended to deviate from the System1 and System 2 optimal strategies when valuation uncertainty was highest’.

Perhaps then we should not be surprised that information disclosures and behavioural interventions other than default laws tend to have limited effects. I will take a look at default laws later. They are a special case.

The story of the tribesman and the bear

For now, I invite you to join me in a thought experiment about, let’s say, a primitive tribesman in the Rockies who is walking across the side of a mountain and sees a bear walking towards him. How does he react? What does he do?

The first reaction, of course, is a shudder of primal fear. The animal is powerful, unpredictable and not open to debate. Needless to say, a primitive tribesman can at this point use a true heuristic - something learnt from one’s own or others’ experience: bears chase and catch people who run away. It’s obviously not a good idea to run.

But what if the individual lacks self-control? He makes a mistake, runs, and is eaten. Presumably it is people like this who gave rise to the ‘don’t run’ heuristic. One might think of this individual as a consumer who does something patently wrong, like falling for an investment scam that looks far too good to be true, and is indeed not true.

Now let’s suppose that the individual has enough self-control not to run, not to fall for the investment scam. The initial wave of excitement is wearing off. He knows that he needs to move away as unobtrusively as he can. He is now thinking rationally. He cannot go forward or back along the slope. But he can go up the slope or down the slope. He starts to calculate…

Somewhere up the slope are fellow tribesmen who can drive off the bear. But what if the bear follows and catches him first? Down the slope is a narrow ravine where he can hide between rocks in a gap too small for the bear to enter. But can he get there before being caught? Perhaps he has time to finish the calculations, the answer is clear, and a rational choice is made. In investment terms, one might think of this as optimising a portfolio.

Perhaps just as likely, our tribesman is finding the calculations hard due to agonising about the unavoidable uncertainties. Will the bear give chase? How much faster is it: will it catch him before he reaches safety?

And the bear is getting closer, distracting his attention from the problem. Unable to work out the permutations in the available time, and increasingly unwilling to try to do so as stress concerning the not yet calculated, the unknown and the unknowable increases with the approaching deadline, our tribesman rationally decides to use another heuristic.

Bears have huge hindquarters and he and his fellow tribesmen have seen them running uphill at great speed to catch deer. But they have also seen that the huge hindquarters are almost a disadvantage when running downhill. So he decides to go down the slope. In investment terms, this might be considered analogous to the choice of someone who uses the heuristic that typically the cost of charges exceeds the benefit of alpha.

Or does he move off down the slope? He is an individual, a real person, not somebody restricted to the limited ideas and choices of a representative agent in a simple model. Perhaps he experiences strong emotions on seeing the bear. It ate his family’s supply of food for the winter. Or it ate his wife.

Let’s say he hates the bear. Or perhaps he is just angry about something completely unconnected. He decides to try to stand and fight and kill the bear, even though the odds on success are very poor. In investment terms, this is a case of ‘emotional finance’ such as the person who gets caught up in bubbles and buys near the top of an unsustainable market.

Or, again, perhaps our primitive tribesman feels that he has a position in society that requires him to stand and fight, however bad an idea it is, because people he leads are watching or may get to know what happened. A great example of this could be Savonarola, the Franciscan Monk who succeeded in ousting the Medicis from Florence and in setting up a religious republic in the city state… only to agree to the first trial by fire there in 400 years, to prove that God was on his side.

This, predictably, destroyed his credibility, and his story did not end well. In investment terms, we might think of people who have boasted about taking big risks and winning, then feeling social pressure to generate new exciting stories to demonstrate their investment acumen.

And there are other realistic possibilities. Perhaps our tribesman is in thrall to a village elder with unusual and unhelpful views about how to deal with bears. An extreme example of this kind of misplaced trust, to show just how powerful it can be, is the suicide of over 900 Americans in Guyana in 1978 under the guidance of supposedly visionary leader Jim Jones.

In investment markets, no such extremity is needed for harm to arise. An adviser may well be trusted and if the adviser has disproportionate faith in certain companies or markets, badly performing portfolios are the likely result.

Again, anthropological analysis may tell us that the tribesman’s village has developed a particular culture about what one must do when facing large and dangerous animals. And, whatever this is, our tribesman may do it, and it may not align with the evidence-based heuristics. 

In terms of investment, we can think of this as roughly analogous to the received wisdom to be found on social media or in the wider family group. I shall not easily forget the reaction I received after making a perfectly reasonable statistical point on a semi-scientific web discussion forum. Unfortunately, my point contradicted the consensus about which other contributors were misguidedly empathising with each other.

Finally, our tribesman may have deep seated personal beliefs and act in accordance with these whatever the cost. This seems to be the case of John Proctor in Miller’s play ‘The Crucible’, an account of the historical Salem Witch trials.

The Salem witch trials are a great example of a society building a false consensus.

By the way, the Salem witch trials are a great example of a society building a false consensus about something that, like finance, is not easily observable and then taking decisions that reflected this consensus.

The point about individual beliefs is captured in Caplan’s ‘The Myth of the Rational Voter’, a fascinating book on how economic misconceptions lead to costly choices. Again, following deep seated personal beliefs as principles whose costs are immaterial could lead to very unhelpful investment strategies.

An important implication of the story

To spell out one important message from this story: if regulators see investors experiencing poor outcomes in an investment market, the choices that drive these outcomes are likely to have diverse drivers. This is an extremely challenging state of affairs for regulators. Let me explain.

Traditional disclosures are based on the idea that if we gave people the information required to perform a rational calculus, they would do so. Such disclosures are in principle easy to design. We just need to know what the required information is, and finance textbooks tell us this.

The behavioural approach to disclosure, including very simple information such as a reminder given in the form of a nudge, rightly makes the regulator’s task more complicated. It does not seek to provide a complete set of information. It seeks to provide a subset of the relevant information in a certain way that the policy-designers judge will lead people to the ‘correct’ choice. I say ‘rightly’ because this approach better, though still imperfectly, reflects reality. A thorough explanation of the application of behavioural thinking to financial regulation is inOccasional Paper 1.

The behavioural approach might be said to take two broad forms. One involves providing the information required for rational calculus in a form that will cause people to pay attention to it. The other is to attempt a short-cut in influencing decision making. For example, if people tend to pay back ‘too little’ when given a choice to rollover debt, devise a quantitative anchor that will tend to cause them to pay back more. Both approaches typically rely on identifying a single psychological driver or bias and seeking to exploit it.

The results of these approaches can be valuable. In principle, it matters that the initial outcome is that a statistically significant proportion of people choose better. The results can also be unexpected. For example, Laibson et al sought to increase savings amounts by providing an anchor in the form of the amounts saved by others. In fact, the target group saved less, possibly because they were intimidated by the achievements of others and gave up.

If decision making is multi-faceted and individually diverse in the way I have described, it is unsurprising that correcting a market failure for most people cannot easily be achieved through disclosing facts or through single nudges or attempts to exploit single biases. I will discuss the implications of this in terms of new analytical possibilities and different types of remedy.

What else does the story mean?

Overall, from our simple story of the primitive man meeting a bear, we see that in reaching a single decision people might shift from System 1 to System 2 and back, and be influenced by a wide range of sociological, anthropological, relationship, personal and emotional factors. Quite possibly all at once or in very quick sequence. And these influences apply equally in the context of financial decisions, as I have shown by using investment as an example.

The process of an individual considering all these different angles perhaps aligns with the ideas of ‘narrative economics’: this is about how people build stories that lead to or justify choices. It could be that sifting through a set of possible influences may help build a story that enables one to choose which influence shall be the clincher in a particular case. I will discuss narrative economics later.

For regulators considering real world markets, it is hard to know, case by case, which of the many possible decision making influences described above we are dealing with, and in what proportions. And then there is the challenge of finding ways to alter the outcome. The job is hard.

Even when people are – or should be - applying rational models of decision-making and are apparently well-qualified to do so, outcomes may be very diverse. See, for example Haghani et al ‘Rational Decision-Making Under Uncertainty: Observed Betting Patterns on a Biased Coin’ (2016).

The experiment was designed to be similar to stock market investing and participants were told exactly how biased the coin was to enable them to place informed bets. But 28% of them went bankrupt in the ‘game’, appropriate analytical tools were not employed and a wide range of behavioural biases was observed. The authors conclude: 

"If a high fraction of quantitatively sophisticated, financially trained individuals have so much difficulty in playing a simple game with a biased coin, what should we expect when it comes to the more complex and long-term task of investing one’s savings?"

If we regulators don’t know enough about decision making to design informational and behavioural remedies that lead a high proportion of consumers to change their decisions and avoid harm – or achieve their goals – one implication we might draw is that we should consider defaults.

Are defaults the answer?

The first point to make on defaults is that they tend to achieve very powerful results. Outcomes really are different for large numbers of people. One example is much increased recruitment to US pension schemes, where the upward ratchet in contributions is a further and clever default.

Perhaps the most striking example is organ donation schemes. Here we see that in neighbouring countries of mainland Europe with similar religious and cultural values organ donation may be below 5% in a country without default and above 95% in a country with default.

But what does it mean to set defaults? Are they really, as sometimes characterised, just another ‘behavioural remedy’? And do they have significant downsides as well as significant strengths?

First, let’s be clear about what we mean. When a body with statutory authority sets a default, it in effect passes a law which says that unless individual citizens take action to exclude themselves from the default then the default shall apply to them. This is very different from typical behavioural remedies which seek to offset unhelpful biases through improved disclosure or choice architecture, reminders, nudges and the like.

Unsurprisingly, given inertia, inattention, adherence to social norms, procrastination and whatever other ‘biases’ are said to be in play, let alone explanatory variables which are not ‘biases’ such as time poverty, simply forgetting or genuine indifference, defaults tend to produce big changes.

For present purposes, though, I am going to assume that defaults are not the answer to all of the regulator’s problems. There are at least three powerful reasons for this.

One reason is that to set a ‘benevolently paternalistic’ default, we need to know what the ‘right’ default is. But how much do we know about what people really, and reasonably, want in different markets? Let alone what people want ‘across’ markets: for example, people may well have views about the critical issue of the size of the household balance sheet – when they want to reduce debt rather than invest more – but, for better or worse, we regulate the markets we observe and they help little with such questions. And how much variety and reliability might there be in these preferences anyway?

A possible response to this is to be merely ‘paternalistic’. That is, to set defaults that reflect our view of what is right regardless of what people want.

The second reason for doubt about defaults is that, while proponents of defaults have characterised them as cases of ‘libertarian paternalism’ – because in principle they allow choice - it is not only extreme libertarians who struggle with this notion. There is a debate about whether the term is an oxymoron. Also, as the statistics on organ donation strongly suggest, there is doubt about the extent to which in reality choices are being made by the citizens affected. Rather, they are de facto imposed.

If so, one might say that typically defaults are really just paternalism because most people will end up doing what the central planner wants, regardless of their own views. Charitably, we could say that this is a case of ‘benevolent paternalism’ because the central planner setting a default usually does care about what some people want and tries to take this into account.

But of course the limitations of the central planner have been widely discussed, for example in Usher’s ‘Political Economy’. And the term ‘benevolent paternalism’ itself has a chequered history. It was used in connection with slavery in America’s south. See for example ‘The Outrage of Benevolent Paternalism’ in the Harvard Crimson.

The third and perhaps most important reason for doubt about defaults is that they may well prevent competition from delivering even better outcomes. To be sure, I do not advocate that more competition is invariably the answer. It has great strengths but in some circumstances causes problems. Stucke’s paper ‘Is competition always good?’ in the Journal of Antitrust Enforcement provides a balanced account.

You will have realised from these points that I am unconvinced that defaults are always the answer when information disclosure and true behavioural remedies are unlikely to enable consumers to make markets work well.

Ways forward

I mentioned earlier that narrative economics might help us to understand more about decision making, and therefore increase our ability to influence decisions to further our objectives. It is an area of increased current interest in economics, starting from a much lower base than in sciences such as anthropology, sociology and political science, where the role of narrative has been perceived as more important.

An example of narrative economics of particular relevance to financial services is Falk and Tirole’s ‘Narratives, imperatives and moral reasoning’ (2016). They describe narratives that allow individuals to maintain a positive image when in fact acting in a morally questionable way. Interestingly for regulators, they also identify the conditions under which Kantian (ethical) behaviour will emerge in an otherwise fully utilitarian environment, and explore how collective decision making and organizational design produces a sub-additivity of responsibility. This is important background for the second part of my talk today: getting firms to comply with regulations.

Again, one could be infected by a contagious external narrative. This is the determinant of choices described by Shiller in Narrative Economics (2017). He uses cases such as the Great Depression of the 1930s and the Great Recession of 2007-9 to show how positions develop across networks. The anthropological analysis of the tribesman’s village above may be a case like this.

Related work between the Bank of England and UCL, Nyman et al, had the self-explanatory title: ‘News and narratives in financial systems: exploiting big data for systemic risk assessment’ (2014). This used text analytics to assess how narratives and emotions play a role in driving developments in the financial system.

Given my earlier discussion of the tribesman and the bear, it is interesting to see that this work treats emotions and narratives in tandem.

Other work in this field of particular interest to regulators is Akerlof and Shiller’s ‘Phishing for Phools’ (2015). A ‘phool’ is somebody susceptible to ‘phishing’, and in this case phishing includes setting up a false narrative that phools accept. This leads to a phishing equilibrium, for example when investment bankers established a market for instruments which depended on the idea that, like modern day alchemists, they could input base metal materials and produce Triple A gold.

It may be, for example, that regulatory information campaigns can help to cut off the development of unhelpful public narratives that could adversely affect the objectives of markets and conduct regulators. In a sense, this is what central banks already do when seeking to calm the fears of markets or of queuing depositors.

The meaning and usefulness of narrative economics for regulatory policy in general is mostly a field waiting to be explored.

But the meaning and usefulness of narrative economics for regulatory policy in general is mostly a field waiting to be explored. The potential complexity may well undermine the tractability of models. For example, central banks and international organisations have struggled for years to add a meaningful set of agents to their models.

Other analytical avenues to explore could be much more use of ex post analysis and engagement with the UK government’s What Works Centres. These seek to bring philosophical rigour about inference and truth to public policy. The practical challenge this poses – and the implication that it is all too easy to rely on things that are not or may not be true – is very well described in Cartwright and Hardie’s book ‘Evidence Based Policy: a practical guide to doing it better’.

Another area of research that could help us understand how firms and consumers take decisions, and therefore influence them, is the field of decision making under uncertainty. Much past work in this field has been actuarial or other applications of probability theory and calculus.

An interesting current development is that the UK’s EPSRC, which funds research, has just given a large grant to the LSE and UCL to apply a multi-disciplinary approach with significant inputs from psychology. This starts with definitions of four major types of uncertainty: imprecision, ambiguity, intractability and indeterminacy (which are taken from Smith and Stern, ‘Uncertainty in science and its role in climate policy’, 2011), all of which are relevant in one financial market or another.

We are hoping to collaborate with this project in some way but it launches later in March so I have nothing specific to report on how it may help us to devise better remedies. I will, though, share two points from the above paper by Smith and Stern.

The first is to show how useful their typology may be for thinking through how uncertainty affects the different markets we regulate. This matters because people may well approach decisions differently in the face of different types of uncertainty:

  • Imprecision (Knightian risk, conditional probability): related to outcomes which we do not know precisely, but for which we believe robust, decision-relevant probability statements can be provided. 
  • Ambiguity (Knightian uncertainty): related to outcomes (be they known, unknown or disputed), for which we are not in a position to make probability statements.
  • Intractability: related to computations known to be relevant to an outcome, but lying beyond the current mathematical or computational capacity to formulate or to execute faithfully.
  • Indeterminacy: related to quantities relevant to policy-making for which no precise value exists. This may arise from the honest diversity of views among people, regarding the desirability of obtaining or avoiding a given outcome. 

The second is a quote to encourage all regulators who have been frustrated by the residual or even fundamental uncertainty left by most impact assessments.

‘Science can be certain that the impacts would be huge even when it cannot quantify those impacts. Communicating this fact by describing what those impacts might be can be of value to the policy maker. Thus, for the scientist supporting policy-making, the immediate aim may not be to reduce uncertainty, but first to better quantify, classify and communicate both the uncertainty and the potential outcomes in the context of policy-making.’

I suppose the final implication of my story could be that influencing decisions is just too hard. It may make more sense to impose product standards than attempt to persuade people through disclosures and nudges to choose the right thing, given their goals.

For example, the FCA is developing an Occasional Paper on complexity. In financial services, we see that complexity can be a tool for blunting competition and establishing phishing equilibria. As stands, firms are free to ratchet up product complexity. This may interact unhelpfully with the decision making complexity already described. Or the product complexity may be designed to confuse the buyers given the process and information prescribed by regulators.

Under this analysis, there is a case for revisiting the CAT standards set by the UK government in the early years of this century. ‘CAT’ stood for conditions, access and terms’. The idea was that the regular, individual consumer has financial needs broadly similar to those of most other consumers, so there was no material loss from lack of elaboration. And the standards were set such that prices were low and pernicious exclusions or hidden risks not allowed. Thus there was some confidence that CAT products were safe, good value for money and suitable for most people’s circumstances.

It is worth noting that industry responses to CAT products were highly negative. They achieved only low sales until the regulator introduced a rule saying that they must be sold unless the adviser could demonstrate the superiority for a consumer of another product. It seems that demonstrating this was not very easy and sales increased. But firms lobbied against the rule. It was withdrawn, and sales fell sharply.

As we do not have any silver bullets for changing choices on the demand side, there is a strong case for considering decisions on the supply side and how to influence them. In particular, how to incentivise substantive compliance so that demand side weaknesses matter less. But this is beyond the scope of this note.


Assuming that regulators are resource constrained, as surely we must be, there is a question about whether it makes more sense to intervene on the demand side or the supply side. There is no easy answer to this question. On one hand, we sometimes see weak demand side responses to regulatory incentives to make better choices. On the other hand, we sometimes see systematic non-compliance that must in part reflect wilful misconduct.

Some argue that we can have more success by trying to change the decisions of firms, especially the decision not to comply in substance, than by trying to change the decisions of consumers. This is because decision makers in firms are in principle subject to constraint by governance and have somewhat predictable incentives. Firms can also be disciplined or shamed by us in ways that consumers cannot.

Moreover, if firms do the right thing, it no longer matters much that consumers make mistakes.  But of course, just as it is hard to engage consumers to make good choices, it is hard to stop firms altering their business to reduce the impacts of regulation.

As ever, much depends on the particular markets under consideration. Standardised products offer protection from product risk and value for money but have drawbacks in terms of competition. But this cost may be low if scope for innovation is low, as in cases where most consumers’ true needs are broadly homogeneous and much of the product differentiation we see is a shrouding or price discrimination device that has little or nothing to do with sensible preferences.

I hope that you found something useful in my remarks today, whether about how individuals actually take decisions and what this implies for regulators’ remedies or about culture and compliance in firms.