Ethics and actuaries: issues and opportunities

I was invited by the “London Markets Actuarial Group”, made up of general insurance actuaries, to speak at an ethics workshop. This post is about some observations I have from this experience. My observations are critical. This does not mean I believe actuaries are not well-intentioned, rather there might be gaps in their understanding that inhibit clear ethical reasoning. The purpose is to highlight how ethical assumptions are always questionable and it is by asking questions that an individual becomes ethical.

The context is that the professional body of UK actuaries, the IFoA, requires members to undertake two hours CPD on ethics a year and the workshop was a means to accomplish this requirement. I had been asked to comment on, amongst other topics: whether ethics just comes down to personal opinion; what is meant by “the public interest” and some comment on how ethics relates to managing risks, the core competence of actuaries. The organizer came up with some simple “cases” to make my theoretical session more interactive.

I began by asserting that ethics was not personal with the explanation that ethical norms were socially constructed and rooted in culture. This basis does not imply that ethical norms are necessarily relative. I gave the examples of truth telling and promise keeping as being essential to enable communication (without mentioning higher order moral absolutism and token absolutism). I explained that “emotivism”, the attitude that moral statements are simply expressions of personal preferences developed in the context of twentieth century logical-positivism and is, today, rather peripheral. I observed that emotivism, as a branch of non-cognitivism, is based in Hume’s idea that moral judgements are essentially passions (desires that motivate action). I suggest that this is an error, in that the motivating desire is a well-running society and moral norms are thus a means to fulfil that desire. The norm of truth telling is not the aim; effective communication is, truth telling is a rational means of achieving communication.

I also made the observation that “being ethical” was not the same as being a “good person”. The point here is that ethics requires thought and reason. Kant made the point, which is contested, that an impulsive act might be good but cannot be ethical. The importance of distinguishing being “good” from being “ethical” is that if questions are approached on the basis of “good” and “bad” the discussion can quickly become polarised, inhibiting discussion, which actually inhibits ethical consideration. A common experience is when politics comes down to being “good” rather than seeking an effective means of attaining policy ends (which are typically common across the political spectrum).

I then, very briefly, summarise the three main approaches to ethics: consequentialist, deontological and character. I began with consequentialism because it is so dominant in finance, mentioning that there are fundamental issues with the foundations of consequentialism; such as, how are consequences identified and assessed and how are they weighted for different people. I gave the classical example of someone taking a lonely neighbour out for a coffee; if the neighbour was killed by a car on the way home, as a consequence of the act, consequentialism would be critical of the actions, though most would think the act of kindness was morally right. I then moved on to a similarly brief explanation of deontological ethics, emphasising their basis in promulgated rules before discussing Kant’s categorical imperative, though I don’t think I mentioned the requirement to treat others as ends in themselves, and not means to an end.

There followed a discussion of the following “case”

• Your firm uses a sophisticated renewal pricing algorithm, so that the least price-sensitive customers will be charged a higher price.
• You are aware of the regulatory concerns about this practice, and so the algorithm has been adjusted to cap the prices for those customers that could be categorised as “vulnerable”​.
• This approach means that your firm offers insurance to a far wider customer base than would otherwise be the case.

I noted that the second point seems to address deontological concerns while the third appears to justify the practice on consequentialist terms. I then mentioned that I had recently seen my car insurance quote, from the same provider, drop 30% by asking for quotes from a price comparison website. I observed that, as a customer, I had issues with this type of practice. 

An initial comment was that customers who shop around were lower risk, and so should pay lower premiums. My observation here is that ethical reasoning must be based on matters of fact. For example, I cannot justify sacrificing children because the sacrificial act appeases the great god and will improve crop yields; this justification is not factual. In the case presented, I would challenge whether there was a causal relationship between “shopping around” and being a lower risk driver. There may be a correlation, but actuaries should be competent at understanding correlation is not causation. The correlation between “shopping around” and “low risk” is likely a consequence of a common underlying factor, such as education. This is covered by quoting based on occupation meaning that the explanation that “shopping around” is relevant information is a bit dubious.

The fundamental ethical issue here is to what extent are insurers treating their customers as ends in themselves, rather than as a means to a personal (corporate) end? The response would probably be that customers are free to shop around and since the insurers are not inhibiting the customers autonomy there is no problem. I think this is open to discussion.

I was left wondering how familiar actuaries are with emergent issues around algorithmic decision-making being highly discriminatory. Two, of many, books on the topic are: The Black Box Society: The Secret Algorithms That Control Money and Information written by a lawyer (the HUP page has lots of additional resources); Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy written by a mathematician. Actuaries are free to disagree with these accounts, but they should be aware that these debates are important and might impact actuarial practice in the future.

Another comment made by a participant, I can’t recall if it was at this point, was that insurers need to be flexible in pricing to allow cross-subsidisation. I think the context was that margins on car insurance were very tight and so profits had to be sought elsewhere. Well established financial theory (for example, Risk Uncertainty and Profit) would suggest that the margins on car insurance, a mature business with lots of relevant data, should be almost non-existent; the market is working well if the margins are tight. What is less clear is why there should be cross-subsidies in car insurance?

Cross subsidising, the practice of charging higher prices to one type of consumers to artificially lower prices for another group, is not the same as pooling, as pooling is amongst similar consumers and a core competency of actuaries is to discriminate different types of customers. Since there are legal and regulatory issues around cross-subsidies, I was surprised to hear this being offered as a justification for specific pricing policies, since they may create operational (regulatory or legal) risks. One moral issue is: what gives insurers the right to determine which type of consumer should be being subsidised and which type should be being penalised?

I think another justification offered (it might have come up elsewhere) was that as long as a contract was transparent there were no ethical issues with offering that contract: the customer is free to accept or decline a clearly presented contract. I am not sure of the ethical basis of this approach, it seems to be rooted in “ethical egoism”, the view that someone acting in their own self-interest is acting morally. This ethical approach is associated with libertarian politics. If this is the ethical basis then it would be incoherent with the apparent fact that insurers are being paternalistic in determining who subsidises who. A general criticism of ethical egoism is that people often do not know what is actually in their “self-interest” for various reasons: is it long-term or short-term self-interest; do individuals have the cognitive/psychological capability to establish their self-interest; is their tacit coercion involved.

Whether or not the “transparency” argument can be justified ethically, that is rationally, is a moot point since there is legal principle in English law that a person must have “capacity” to enter a contract. In US law, “capacity” is assumed if the contractor is old enough, it is not so clear in English law. This issue played a part in the PPI mis-selling scandal and is reflected in the change to the Actuaries’ Code 6.4

Where Members identify that a user of their work has, or is reasonably likely to have, misunderstood or misinterpreted the information or advice provided by them in a way which could have a material impact, Members should draw the user’s attention to this.

There is a lot of judgement involved in “has, or is reasonably likely to have, misunderstood or misinterpreted the information”. The requirement is not simply that the terms are clear and transparent in the eyes of a well-educated actuary, the “reasonably likely” clause is open to interpretation and a court is likely to regard what is “clear” to a well-educated actuary might not be clear to their client.

Returning to the case under discussion, I observed that while it might appear to conform to some consequentialist or deontological ethical frameworks it did appear to be “unfair”. The concept of “fairness” is closely related to the third major ethical framework: character or virtue ethics.

Aristotle is usually recognised as the key proponent of virtue ethics. I explained his approach was based on the idea that everything has an “ends” or purpose. This apples to inanimate physical objects, the purpose of a chair is to support a sitting person; different chairs have different specific purposes (dining, arm, desk, …). “Virtues” lead to the successful attainment of the purpose. Aristotle argued that the purpose of humans was eudaimonia, often translated as “happiness” though I think “equilibrium” is better. The Greeks identified the four cardinal virtues: prudence (wisdom), courage (fortitude) and temperance (decorum) and to clarify I pointed out that the three central characters of the Harry Potter novels represented these three core virtues.

The fourth Greek virtue is justice. In this context, justice is seen as the virtue (quality) of a functionally differentiated system that enables the system to work well (to achieve its purpose). For example, a functioning computer system can be seen as possessing the virtue of justice. Justice is the virtue of a community that enables different people to work well together to deliver a well-functioning society. Just as truth-telling and promise-keeping are essential to justice, the well running of a society, reciprocity, or fairness, is essential to all but the most primitive societies. The significance of reciprocity only emerges in societies based on commercial exchange, reinforcing the idea that ethical norms are socially constructed in order to support the coherence of that society. A significant proportion of my research is concerned with understanding how the ethical norm of reciprocity is embedded in financial economics.

Having reviewed the different ethical frameworks I spoke a bit about “public interests”. I feel this was rather weak/irrelevant in the context. My aim was to introduce the idea that Adam Smith “created” capitalism by suggesting that “passions” (private interests) and “interests” (public interests) could be balance in monetary terms. This is the ethical justification of the modern financial imperative to maximise expected utility (of money).

Following this presentation the following case was introduced

• You are asked to price a new product
• This will be provided on a wholesale, net priced basis to your distribution partner
• Although the product is expected to generate very few claims, the consumer perception of risk is much greater than this
• Your distribution partner is therefore expecting to sell the product for a very high mark up.
• They are also happy for you to set a relatively conservative initial net price for the product, as long as there is a 100% profit commission to the distributor once you have made your agreed margi

A comment about this case was along the lines that all companies “mark up” and the example of the mark-up on men’s razors was offered. I thought about this comment for a while, the reason I did this was considering whether a full response would be possible in the time available. I decided it would not have been but I can outline the issues here.

Firstly, I believe there is a false equivalence, in general, between selling razors and providing insurance. My evidence for this is in the fact that actuaries are licensed professionals, where as razor manufacturers are not. There are two aspects to this. Licensing, or regulating, professions means that those engaged in the profession must demonstrate technical skills and meet standards of behaviour. In compensation, licensing limits competition and so they can charge premium rates. Economic libertarians (see before) are fundamentally opposed to licensing on the basis that there should be a free market in goods and services. To assert that there is an equivalence between being an actuary and razor manufacturer/retailer misses this point and the fact that actuaries’ fees are already inflated: charging high mark-ups could be seen rent seeking.

Secondly, as can be deduced by my appearance, razors are not a necessity where as I am required to take out insurance if I wish to drive or borrow money. This would suggest that mark-ups in insurance contravene Kant’s injunction to treat people as ends not means. Therefore, I do not think it is tenable to justify high mark-ups in insurance on the basis of high mark-ups on the basis of razors.

Passing over the specifics of the comment there is the issue as to whether it is legitimate to “take more than what is given”, for a commercial transaction not to conform to the norm of reciprocity or “fairness in exchange”. The ethical justification for charging high mark-ups is the main moral imperative of economics: to maximise profit. This is rooted in a synthesis of Mill’s utilitarianism: right action is the action that produces the greatest good to the greatest number, synthesised with Smith’s claim that wealth can measure “passions” and “interests”. There are significant philosophical issues with both these supporting principles and this economic imperative is losing its influence, though I recognise I am in a minority of those sympathetic to finance in rejecting this imperative (in favour of the norm of reciprocity).

From the early 1950s the profit maximisation principle was established in English case law in Buttle v Saunders, where it was ruled that profit maximisation superseded any concept of “commercial morality”. This case was later employed to prevent the National Union of Miners investing the miners pension fund in support of miners “political”, rather than financial, interests (Cowan v Scargill). This century, concerns over the ability of investment funds to invest “ethically”, and divest in hydrocarbons, for example, led to a re-think and there is widespread doubt today that “profit maximisation” is the economic imperative in finance.

An actuary might well argue that arguing against the ethical legitimacy of mark-ups is utopian and impractical. However, there is good evidence to support the belief that not imposing mark-ups is fundamental to sound finance. The evidence is in the practice of "dual quoting", simultaneously giving bid and ask prices without knowing which position the counterparty will take. This was the basis of jobbing the London markets for centuries. Dual quoting imposes sincerity on jobbers; they are unable to set high mark-ups without them being clearly obvious in a wide spread. A razor retailer could not impose a 4000% mark-up if they were required to simultaneously quote a price at which they would buy razors. This might seem rather absurd, but there is  evidence that Quaker retailers would "dual quote" as part of their commitment to "honest dealing". It is worth noting that the influence on global finance and English commerce of a small number of Quakers is profound: Barclays, Lloyds, Waterhouse, Friends Provident, Boots, Clarks and the British industrial revolution was built on the Quaker Coalbrookdale Iron Company and Stockton & Darlington Railway. Being honest in pricing seems to promote persistence in business.

Dual quoting has a mathematical relevance. Modern subjectivist approaches to statistics, including Bayesian methods, were initially justified using the example of dual quoting by Frank Ramsey, the "Dutch Book Argument".

The point I went on to make is that ethics is fundamentally concerned with questioning one’s behaviours. In this sense it is as “scientific” as physics, which questions phenomena in the physical world rather than the social. Ethical norms are nothing more, or less, than the norms that ensure a particular society “works”.

On this basis, I finished my presentation by drawing links between ethics and risk management. I pointed out that the three pillars of Basel II corresponded to the three main ethical frameworks (Minimum capital requirements are deontological rules; Supervisory review relates to the character of the firm; Market discipline is concerned with consequences). I also described the failure of LTCM and the success of J.P. Morgan through the Credit Crisis in terms of ethics.

The failure of LTCM is widely understood as being a consequence of reckless trading. Donald MacKenzie has a different explanation that suggests the firm’s failure was more to do with their social isolation, rooted in their success and consequent envy of others. Their downfall was precipitated by the poor drafting of a fax and a lack of support when beliefs about their solvency started to spread. The success of J.P. Morgan through the Credit Crisis was based on the fact that they questioned assumptions of prices their models produced, this self-examination is fundamentally ethical.

These closing remarks were designed to emphasise the practical and serious utility of ethical behaviour. Ethics can be viewed as vacuous philosophical introspection irrelevant to daily life, or mundane cases that address clear situations. Ethics comes into its own when people work out what behaviours will prove beneficial in their daily lives.

Plato argued it was impossible to be knowingly un-ethical, bad behaviour was a consequence of ignorance. Pushing this, and the line taken by Aristotle and Kant, unethical behaviour is always a consequence of people not thinking about their behaviour. The problem with ethics, and all social, as opposed to physical, phenomena, is that the rules of the game are fluid. Most of the audience in the workshop would have been taught that profit maximisation was the moral imperative, and on this basis they offered the comments presented. A problem will develop if society shifts its consensus away from profit maximisation and towards social cohesion, and so actuaries become isolated from the society they inhabit. This is a risk management problem and mirrors the type of situation investment managers face: will the fundamentals change. It is, and should be, a responsibility of the IFoA to horizon scan these shifts and reflect them in its approach to professionalism. I perceive changes in the Actuaries’ Code, such as in regard to Communication, being a response to this foresight. If actuaries, or finance in general, become disconnected from society they will find their support network, potentially including their status as a licensed profession, evaporating, just as LTCM saw their status as “masters of the universe” disappear. This is why ethics are important.

P.S. One participant commented that the workshop had been Euro-centric. This is true for the following reasons. Time did not allow me to point out links between European and Chinese philosophical approaches. Actuarial science is a product of European thought, if an actuaries employed techniques that had emerged out of Buddhist societies the workshop would have placed greater emphasis on Buddhist ethics. To try and apply Buddhist ethics to Euro-centric actuarial practices is going to be tough.

Mathematician, heal thyself

I attended the second “Ethics in Mathematics” conference held at the University of Cambridge and organised by Maurice Chiodo. I had come across Maurice about twelve months ago when I was working with a Norwegian academic on the issue of ethics in maths. Maurice has developed into the Henry Oldenburg or Maurice Fréchet, though I suspect being a central node in a network is not as career enhancing for Maurice as it was for Oldenburg and Fréchet because the concern is ethics. In my experience mathematicians don’t really want to face up to ethical issues associated with their discipline.

Ethics in mathematics is a broad issue, and Maurice’s interests extend well beyond my focus. For example, there are significant issues of professional ethics (diversity of the profession; treatment of junior academics); incorporating ethics into the curricula (school, undergraduate, postgraduate) as well as the “ethical content” of mathematics and the use of mathematics in society (my interests).

The first hurdle in talking about ethics in mathematics is to clarify what is being talked about. Generally, “ethical behaviour” is taken to mean being “good”. But this does not stand up to examination since it is usually impossible to define what is being meant by good, for example the Mitchell and Webb sketch that has become an internet meme of David Mitchel as an SS officer asking the question “Maybe we’re the bad guys?” Socrates addressed the problem and saw a solution in being self-reflective: the SS officer asking the question is the start of ethics. Too often, the assumption is “I am a good person, therefore what I think is ethical”, all too often this results in people calling each other fascists.

I frequently refer to Gillian Tett’s Fools’ Gold as an account of ethical mathematical practice. Tett explains how J.P. Morgan came out of the 2008-2009 Financial Crisis because it used mathematics critically rather than blindly accepting the outputs of “black boxes”. I felt the approach Tett described was oddly discordant with the attitude of mathematicians. During the crisis, I co-ordinated a response from UK mathematicians, through the Council of Mathematical Sciences, to criticism of the use of mathematics in finance, this information was also passed onto the UK Science Minister of the time.

The standard response from (senior) UK mathematicians was along the lines that finance hadn’t used mathematics but abused it. The solution was to have “more” and “better” mathematicians. This was underpinned by some adopting a logical positivist line, attributed to Hume, that the role of mathematicians is to describe the world as it is, not as it ought to be. At the time I felt mathematicians were not examining the role of their discipline in the crisis; they were not behaving ethically. This was the start of my journey that transformed me from an “uncritical” (unethical?) mathematician to someone who feels mathematics is vital, so long as it is critical.

I was reminded of my experience in 2009-2010 at the conference when Maurice highlighted that of the hundreds of Cambridge maths faculty, only the two or three involved in organising the meeting were listening to the talks: UK mathematics (as distinct from statistics) seems disinterested in ethics, it does not wish to be critical. After making this observation, the session was on “Calling out bad mathematics”. There were two talks, on by Sam Marsh on issues with the valuation of UK academics’ valuation and the second about the use contemporary use of mathematics in support of “neo-eugenics” (“contemporary” because there is a long history of this). I had to come home and so was unable to participate in the discussion and what follows are the comments I would have made.

During the workshop I had made the point not to demonise finance, it is not intrinsically unethical (see the example in Fools’ Gold) though I did not comment in a discussion about funding sources that governments are rarely benign. I would have made this point in respect to Sam Marsh’s presentation. The pensions dispute is generally framed as academics (unionised workers) against the Universities Superannuation Scheme (agents of the bosses). What was pointed out to me by academic actuaries early on in the dispute was that the “gold plated” academic pension was doomed not because of the employers but because the UK Government’s “Pensions Regulator” was running a campaign to close these types of pensions because they were high risk if the sponsoring company failed. Academics do not tend to think universities will fail financially, the government takes a different view. As a result, the pension is not as sound, looking forward, as it appears. My question to Dr Marsh would have been “Where did the boss of the USS work before the USS”. The answer is, the Pensions Regulator. My point is, there was more in the USS valuation than the mathematics, the USS were enacting policy principles from the Pensions Regulator (a government agency), not necessarily the employers. This highlights that governments are not always paragons of virtue.

Clement Mouhout’s presentation highlighted the “abuse” of mathematics in support of eugenics. Mouhout categorised three types of “bad maths”: not being internally rigorous (not being valid); wrong application of axiomatic method (I interpret this as correct logic but incorrect premises); “math-science washing” (the area I am interest in). After I had left, a report on the discussion was

Also really like Mouhot's response to a question about what if some actual legit mathematics happens to validate or be useful to awful ideas/people. Answer: it's not a real hypothetical; to the extent it's validating awful things it is not a legit result of math. #EiM2

— Michael J. Barany (@MBarany) April 4, 2019

In the sense that mathematics cannot legitimate untrue theories, I agree. I am more concerned if the claim is mathematics cannot support “unpalatable” ideas. My concern is an issue I have with the use of mathematics to support left-wing ideology, not just right-wing ideology. My battle calling out mathematics was about a theory criticising finance. I have asked the authors to explain a saltus in their mathematical argument; their response was it has bee peer reviewed and they do not have to explain to me. I note this author/correspondent was on the editorial board of the journal who published the paper. Mathematics is unconcerned with politics, in this sense.

One issue that came out of Marsh's presentation and resonated with me was that people recognise the authority of mathematics and people will generally listen to mathematicians more than sociologists. Dr Marsh's experience, and confusion around the fact that the USS were unperturbed my criticism by mathematicians of mathematical ideas is that, while they might listen to mathematicians they might be more inclined to take notice of social scientists. I think this is a real ethical, in the sense that it relates to the "ethos", problem for mathematics: mathematics is authoritative, but widely irrelevant. The theories are rigorous, but vacuous.

This isn't a new issue, related to the widespread ignorance of expertise. In Feller's influential paper On the Theory of Stochastic Processes, with Particular Reference to Applications there is a rather innocuous discussion of the Pareto's law of wealth distribution (p 418). The Pareto Law is sometimes used to justify inequality (in a similar manner to mathematical models of race/sex being used to justify inequality) and is colloquially the "80-20" rule: 80% of wealth is held by 20% of the population. Feller raises the question whether the rule is valid in non-ergodic situations. This is substantive as it implies the 80-20 rule need not always hold, it is not natural. Economists do not worry about this mathematical concern.

What is the problem? I think the starting point is that mathematics is neutral, just as a rifle is neutral. There are a proportion of the population who believe mathematics is intrinsically evil (because it enables inhumane instrumental reasoning) just as there are a proportion who believe all firearms are intrinsically evil. There are those who feel that mathematics is intrinsically good, just as there are those who feel the solution to mass shootings is wider gun ownership. I tend to think mathematics is neither good nor evil but the use of mathematics an be ethical or unethical, just as a rifle can be used ethically or not. Mathematics is ethical if it is being used critically, meaning that those using it are being self-reflective in what they do. Mathematics is not ethical if it is being used instrumentally, that is used as a means to a pre-defined ends (justifying risky investments, closing pension schemes, eugenics, restraining finance).

Moreover, mathematicians need to articulate this and then take steps to be listened to in serious policy discussions. Politicians don't listen to mathematicians because mathematicians (as scientists) tell the politicians that their work is not political. This is a lie, mathematics is constantly being used to justify political ends, and a useless lie, in that it means mathematicians are not taken notice of by politicians when they challenge the instrumental use of mathematics.