Archives for uncertainty

Why I think about Insurance 2.0

April 12, 2013 by tim in Insurance 2.0 | Leave a comment

I have just up-dated my definitions or risk and uncertainty and, in so doing, have clarified why I am writing all this about insurance 2.0.

Here is my new definition of uncertainty:

I am slightly lazy about how I use the term uncertainty – I suspect, like most people. I don’t, for example, stick strictly to either of the definitions I use when I compare the theoretical differences between risk and uncertainty in my definition of risk.

Without trying to excuse my laziness, I do this because, in most of my thinking, I have a working assumption that something like uncertainty is what most of us are actually talking about when we refer to risk. By ‘something like uncertainty’, I mean that I conflate two theoretical views of what uncertainty is compared to risk. In this hybrid view, I suspect that most of us have too little idea of all that could happen in a given situation and, were something bad to happen, how bad it could be.

I mash up the ‘measurable/not measurable’ and ‘multiple options/downside of the options’ theoretical approaches for practical reasons associated with what I hope Insurance 2.0 will deliver.

On one level, I would like to alter the asymmetry that currently sees insurance buyers generally ignorant of the extent of their possible outcomes and the potential costs in a way that insurers are not. By the way, that is not to say that insurers are omniscient; far from it. They have no idea who is going to be loss free and who is not but they have a pretty good idea how many policyholders will have losses and how big those losses will be.

Altering the asymmetry is however, no more than a by-product of the main aim. I would like to help buyers to have fewer losses and for those to cost them less.

3 reasons I don’t want big data insurance

April 2, 2013 by tim in Insurance 2.0 | Leave a comment

I was recently talking to a good friend of mine, who pointed out that a lot of what I have been talking about here is encapsulated in the simple idea that new technology means a whole load of clever new stuff can now be done.

As far as that statement goes, he was absolutely right but, in seeking to further clarify what he thought I have been talking about, he went on to describe a ‘big data’ centric view of what all this clever new stuff could be, as far as insurance is concerned. He specifically described how insurers will be able to segment risk into ever-finer categories and sub-categories, so allowing them more effectively to exclude the risks they prefer not to accept and to price more accurately the risks they are happy to accept.

<p>His comments made me realise that I haven’t explained clearly enough the specific type of clever new stuff I am thinking about. So, to try to clarify, there seem to me to be diverging paths ahead of us.

In the ‘big data’ scenario, insurers will develop ever-more detailed information about us that they will use to segment the market into ever-finer and more accurately priced gradations. The information will be algorithm developed, will be spookily accurate about us but will forever remain the property of the insurer; it won’t be shared with us and so it won’t be available to us to let us use it to manage our uncertainty. {See this post to see what I mean by ‘managing uncertainty’.}

In this scenario, big data will increasingly be seen as something that is done or used against us, even if we tacitly accept the trade-off between the information gathering by the insurer and the products this allows them to provide to us.

In the ‘network scenario’ on the other hand, I will consciously share my information with a utility of my choice, that is designed to help me make sense of my current uncertainty and to give me access to the resources to help me reduce it. I will be able to choose the connections I want to make to develop improvement information and processes; the utility will also provide the platform from which I can implement the improvements. I will not only own my information in the formal legal sense but I will also own the improvement process and the resulting improvements.

In this scenario, I will also be able to use the information about the quantity and quality of my risk (not uncertainty anymore) to contract with insurers also accessible via the platform for a volatility management service that will be unrecognisable from current insurance. Apart from anything else, the likelihood and cost of my losses will be lower and I will be able to demonstrate full disclosure; these are the three main components of every pure premium calculation. By ‘pure’, I mean the cost of the risk to the insurer, without their own costs added.

The key difference between the 2 scenarios is that, in the first, risk management is directed by the insurer in their interest; in the second, I direct risk management in my interest. Scenario 2 won’t be for everyone but scenario 2 is my choice, for 3 reasons.

The first is simply that I will own my own information. Information known about me, accurate or not, dictates how everyone behaves towards me. In a bar, if someone thinks I am an arsehole, that is their prerogative. Of course, that hardly ever happens… But, if an insurer were to have the wrong information about me, which can happen in all sorts of ways and for all sorts of reasons, I might desperately need to correct the information – assuming I can even discover they have erroneous information. If it is their information, I may struggle to get them even to disclose it, much less change it.

The second is that scenario 1 is too financially based. There is nothing wrong with a financial perspective; it is just not enough perspective when talking about managing either uncertainty or risk. From a financial perspective, my insurer will be asked to pay my losses, whatever my losses happen to be. They will naturally seek to use risk selection and pricing tools to coerce me to behave in ways that reduce the likelihood of their having to pay claims. They won’t however, develop or share information about how I might behave outside their preferred behaviours but remain an acceptable risk.

This will partly be because they will rightly not see it as any of their business to tell me how to behave. The obvious irony of course, is that is precisely the effect they will have; in seeking to reduce their costs, they will also reduce my options. More important to them, in the highly competitive insurance business, the quality of your data is your greatest source of competitive advantage; you just don’t give that stuff away.

Taking the broader perspective of scenario 2, I expect to be able to choose how I behave in the knowledge that there are different potential outcomes – positive and negative – that can arise from my choice of behaviour. There will then be all kinds of options to choose from – or not – at my choice. So, I might choose to improve the chance of a positive outcome. On the other hand, other issues beyond the specific choice of behaviour might mean it is more important ‘for now’ to reduce the chance of a negative outcome. I may also be able to hedge the negative outcomes by acting in advance of the losses I can’t prevent, to make sure both their effects and costs are kept as low as possible.

The point is that risk management is not just about money but insurance is.

The third reason, however, is the most important, dealing as it does with what I see as the real digital divide. I understand that corporations will try to influence my behaviour when it comes to trying to convince me to buy their car, gadget, shoes or whatever (3 of my particular weaknesses…). But at the heart of this overall process is risk management – how I optimize the positive and minimize the negative effects of my behaviours, choices and activities. I don’t want to sound overly apocalyptic but it is the very essence of my ‘me-ness’ that I have the free will to make decisions about how I behave for good or ill. I don’t want an insurance company making those kinds of decisions for me.

My own intermediary between uncertainty and risk?

March 12, 2013 by tim in Insurance 2.0 | Leave a comment

When I am asked what insurance is, I have until recently thought it was sufficient to trot out one of two old faithfuls. I would either say that it is was the mechanism by which the premiums of the many pay the losses of the few or that it was a promise made to pay losses by an entity that can afford to pay them, to one that cannot.

As far as they go, in “insurance as usual” terms, these descriptions work. But I have now realised that they are typical of much shorthand; they have made me think too lazily about what insurance is and disguised what I am coming to think of as the inadequacy of “insurance as usual”.

Thanks to two recent conversations, I have had something of an epiphany; though at the time, for someone who thought he had been thinking about Insurance 2.0 for some time, it felt more like being slapped around the face with a wet fish…

I would be interested in hearing if this – with some clarification – gets a bit closer to the core of what insurance is?

Insurance is the mechanism that converts uncertainty into risk.

In thinking about insurance this way, I have also come to realise what Insurance 2.0 should be.

In Insurance 2.0, I will become my own intermediary between uncertainty and risk.

First, a couple of definitions and clarifications. Here I am using risk and uncertainty as the difference between being able to measure, or not, the possible outcomes in any given situation. In one situation – the uncertain one – there is, for example, insufficient information to measure the possible outcome(s). In the other, one can measure the risk of one outcome – or even a number of possible outcomes. Risk is measurable, uncertainty is not; it is as simple and as complicated as that.

As I have come to think about the inadequacy of insurance, the uncertainty I am thinking about is your and my uncertainty if, like me, you have a family, a job, a mortgage, you drive a car and are healthy. It applies equally well however, for a business with assets and resources, customers and debts, let’s say.

Both the business and I have a general idea of what could go wrong with any of the things that matter to us but neither of us – on our own – can measure the probability of a material detrimental event occurring, how bad such an event could be or what really works – and is worthwhile to spend money or time on – to improve either situation, whether before, during or after such an event. I am not sure which is the chicken and which the egg but I think the reasons may lie in and around the ideas that we all have too many other things to think about and, even if we wanted to, we have no yardstick against which we can make any kind of calculated risk judgement.

Risk on the other hand, as I am using it, is a very different concept. On the basis of detailed calculations using data on the past performance of portfolios and projecting forward by factoring changes to the data, insurers aim to put themselves into a situation where they have (say) just a 0.01% chance of loosing more than the premium they collect and their capital from losses arising from the multiple portfolios of insurance policies they write. This is carefully calculated risk.

So, the conversion process occurs when insurers make promises to buyers to relieve them of a little of their uncertainty in return for a premium, and in so doing, they accept the risk of how the sum of those promises will work out. The specific processes are:

1. Pick one of the things someone is uncertain about;
2. Issue a promise that covers some of it; and
3. Aggregate lots of as similar promises as possible.

Then, go out and pick something else someone is uncertain about, that is preferably uncorrelated to the last thing you picked, and repeat. And keep repeating until you have what you believe is the optimum balance of risk, premium and capital. Then adjust as environment changes require.

It is a bit more complex than that – but not by much – and I accept that there is a genuine and valuable promise made by the insurer, though some seem more willing to fulfill their promises than others. The problem is that the amount of uncertainty removed by any one promise or policy (and so the value of its removal) is tiny (a highly technical, quantitative term) and cognitively, the value might even be negative given the known difficulty of dealing with insurers – but I digress a bit.

The more central point is that every policy is tightly focused on a particular peril and has limits, deductibles and other qualifications that make it, by design, incomplete even for the peril it addresses. I haven’t worked out the math (pun intended…) but even if I bought every single insurance policy I could possibly buy, I couldn’t insure (even if I could afford to) everything, so I would still be left with residual exposure. And I would still have done precisely nothing to make any of the events insured against any less likely to occur or any less damaging if they did.

I have no problem with insurers – or anyone else – making a profit and I expect profit to be earned but, with insurers making a profit by turning my uncertainty into their risk without making any appreciable difference to my uncertainty, I am no longer sure they are doing enough to justify my premium or the profit they make from it.

I want a service that delivers more than that. And I think the tools are now available to deliver more in the form of Insurance 2.0.

Insurance is known as a risk management tool – but it is really only that when looked at from the perspective of the insurer, who creates risk – and its corollary, reward – by making many little, similar but very selective promises. Insurers turn uncertainty into risk for their benefit and, to the extent there hasn’t so far been any alternative, I have accepted what has been offered.

What I really want though, is a suite of tools to help me manage uncertainty. If, by using them, I can become my own intermediary between uncertainty and risk, that would be fine… I am sure it would involve some work but, for example, to have a clear understanding of what risks I have, whether I can take more or less of them and to understand the many different ways I can increase or reduce them – these would all be valuable. And I am sure it would be cheaper and more effective to buy risk protection than uncertainty protection, which is really all I can buy now.

Risk, unlike insurance, is no pig; an introduction to gamification

October 11, 2011 by tim in Social technology | Leave a comment

I understand why people get excited when Facebook makes a change.  Here is a good set of examples.  Quite apart from some changes – like default privacy settings – having been extremely poorly conceived and others poorly implemented, some people just don’t like change.

But it is a testament to Facebook’s success that their many changes – and they change things a lot and often – have generally been readily accepted, if not without initial ‘sound and fury’.  And I like the fact that Facebook changes a lot.  Indeed, I suspect Facebook would be much less successful without frequent change.

Why is lots of change so important?

I certainly enjoy sharing my friends and relatives lives and doings on Facebook; for me and almost everyone else this is, of course, Facebook’s primary feature.  But let’s be honest, much of what we read about some of our friends and family is less than scintillating, so the sharing process itself has to be engaging.  Facebook isn’t successful simply because it allows us to connect with people we know who are otherwise too far away to see and hear from everyday; nor is it successful just because it is an application we all enjoy using – an enjoyment that is kept fresh by change.  I think it is the combination of the connections and the process – in different proportions for different people - that explains Facebook’s success.

It is as much the medium as the message that engages us.  Another day, I might ponder on what I see as Facebook’s biggest challenge for its future – how it maintains an engaging process in the face of plans to become a utility, at least as that word is traditionally understood, and in the face of the “seen that, got the tee-shirt, move along” attitude towards sharing others’ lives – but for now, that is a digression.

What I am curious about now, and what I see as offering significant potential to the risk and insurance industries, is a process I am seeing mentioned more and more often; ‘gamification‘ – the use of game design techniques to engage audiences.  This post (by Confused of Calcutta) is a good discussion of the topic.  This TED video is good too.

Change is the biggest gamification example of them all – “a new game every week” – and Wikipedia lists the following further gamification examples:

  • achievement “badges”
  • achievement levels
  • “leader boards”
  • a progress bar or other visual meter to indicate how close people are to completing a task a company is trying to encourage, such as completing a social networking profile or earning a frequent shopper loyalty award.
  • systems for awarding, redeeming, trading, gifting, and otherwise exchanging points
  • challenges between users
  • embedding small casual games within other activities.

Back to the Confused of Calcutta link above, and where this all hopefully starts to make sense when it concerns risk, JP suggests something along the lines of: gamification used on its own is like “putting lipstick on a pig”…

He is talking about the fact that, if an underlying subject or process is fundamentally dull, no amount of gaming layered on top is going to make the slightest difference.  He is specifically talking about work – he now works for salesforce.com – and rather candidly suggesting, I think, that just sticking a salesforce platform under a dull business won’t make it any less dull – but I am getting back to the medium and the message again and away from my point.

Gamification won’t make the process of buying (or selling) insurance any more “fun” but it will be one of the tactics used to encourage people to share their knowledge about uncertainty – something I believe people are genuinely interested and concerned about, and which has nothing to do with insurance, right?

So, to the extent uncertainty about particular things (for example, activities {like driving}, events {a party for example} or processes {like banking}) is one core feature of the more general term “risk”, I expect gamification will be a key element used to encourage people to start sharing knowledge about specific uncertainties.  Such collaborative processes will begin to cause reductions in some specific uncertainties, and lead all of us to be able to start thinking differently about risk generally, and produce new ideas for dealing with it.

And risk, unlike insurance, is no pig.