Electronic Insurance
by Paul Hazelden

We introduce computers to speed up our work, and then discover that the computers can change our work. This process of change has been given various names, such as ‘Re-inventing the company’ and ‘Business Process Re-engineering’ but the meaning behind the words has stayed constant: to focus your attention on the core of your business, and ask yourself what is the best way to do that.

So often in business, as in life, we find that we focus on methods, processes and techniques instead of goals. The processes were designed to help us achieve the goals, so we assume that if we improve them we will achieve the goals more effectively. This assumption fails for at least three fundamental reasons: the internal constraints inherent in all processes, the paradox of optimisation of complex systems, and the unpredictable effect of changes to such systems.

There is no point in introducing EDI in an attempt to automate the existing system if the inefficiencies of the existing system are a major part of the problem we are attempting to solve. EDI would only make matters worse by continuing to support a structure which needs, in the long run, to be demolished. EDI is a major investment and only makes sense if we use it as part of our process of changing the industry. It is one of the foundations of the new system, and not until it is in place can we start to save money by cutting back on the old system.

The rest of this document suggests a possible role for EDI to play in a revised insurance industry.

If Insurance had only just been invented and we had no companies, organisations and systems already in place, what would we do? If we were given five years to create an industry, what would we build?

Of course, this is not our situation. But answering this question should give us some idea of where we are heading. We can then work out the best way of getting there.

If were to start with a blank sheet, we would probably end up with something like the following.

• A range of Insurers: we need competition.
• Simple but flexible products. Insurance is essentially simple, as I try to demonstrate in the final section below. Centuries of tradition mask the underlying simplicity, which point leads on to the next...
• Little or no special terminology and jargon.
• No combination products. The work of establishing insurance contracts is done by computer, so the Client does not care if there is one contract or a thousand. The Client is only interested in, and only needs to know, the cost and the cover.
• Access to the products directly from the public and through accountants. There will always be people who want to handle all or some of this themselves. But financial planning is never likely to be simple, and most people will probably want to trust a human specialist rather than a program written by someone they do not know. Insurance is only one part of your entire financial situation, and determining the best course of action can only be done with a complete knowledge of your financial, family and other related circumstances, plans and concerns. It makes sense if all this information is only held by one professional person or organisation.
• Commission, if it is required at all, is small and determined by the amount of work needed to set up the contract, not by the value of the premium or the amount of the sum assured. People are paid for the work they do - in other words, the amount of work they have saved the Insurer, but because it is done electronically there is next to no cost in setting up or updating normal insurance cover. If you require specialist advice, you would pay your accountant like you pay your solicitor, for their time.
• Products offered and sales agreed electronically, using computers communicating by EDI. The next section describes this in more detail.

Is that all? I have already suggested that insurance, in essence, is really quite simple. Are there any other significant features of the insurance industry we have left out?

It is generally accepted that general insurance - for houses and cars - is bought, while life assurance is sold. Do we not need to add some salesmen to this picture? Perhaps. I am not sure on this point. It would be interesting to know what percentage of sales in the life industry are generated from cold calling by salesmen. And, of those sales, how many are to people who were already considering the need for some additional cover and would have contacted their existing insurer or an IFA in the next couple of months anyway?

I expect Accountants to become more important to normal people as taxation becomes more complex and the Inland Revenue move to self-assessment. By ‘Accountants’ I mean in the first place real people, but as time goes on they will increasingly be replaced by software packages designed to provide the same type of advice for the 90% of ‘normal’ people. Those who need specialist advice will probably continue for some time to go to real experts in their particular field.

Your Accountant will tell you when you need to take out more life cover. Now an Accountant doesn’t want you to sink all your money into a Pension Fund - he wants you to have lots of spare cash so you have to decide what to do with it, and also so you can afford to pay his bills. So if the Accountant advises you to put money into a Pension Fund, it is probably in your best interest to do so. You pay the Accountant to give you good advice. It is in the Accountant’s best interest to give you the best advise possible because they want to develop your financial position, their relationship with you, and your trust in their ability, over the years they see ahead. The software packages want to give you the best advice because they rely on selling you the annual upgrades. In either case, they are more interested in building a lifetime commitment than in getting commission from a quick sale today. That means the Clients are more likely to take their advice as they see less conflict of interest. Hence there is less need to ‘sell’ in the traditional sense.

There is also scope for different methods of selling. ‘Educational’ videos are now nearly as cheap as junk mail. When many homes are equipped with multimedia PCs, you can distribute multimedia presentations at next to no cost. You can incentivise people to use interactive ‘let’s explore what would happen if’ packages to bring home the consequences of under-insurance. They might be given five pounds off entry to Alton Towers if they email the results of working through the exercises to the relevant address.

Insurers will always produce ‘you can rely on us’ adverts to ensure brand recognition. They will continue to target existing customers with special offers and use mailing lists to establish new contacts.

The government has an interest in ensuring the population is properly insured. They could decide to place a duty of care on employers to ensure that the employee is making adequate provision for their retirement. Perhaps mortgage lenders would have a similar duty concerning household insurance. Car insurance is already a legal obligation. Perhaps your employer and mortgage lender will each need a letter - or email - each year from your accountant saying that your circumstances have been reviewed and you are in their opinion adequately covered.

Of course, ‘adequate cover’ would need to be defined fairly carefully, but would that not be in everybody’s interests to do? It would be helpful to have some objective yardstick to compare my situation against. Of course, it would be far too difficult and complicated to produce anything like this without computers, but that is no longer a barrier.

In this new environment, how would we use EDI? It seems likely that insurance would be negotiated through on-line dialogues between two computers, each representing the interests of their owners, with the conversation taking place through interactive EDI and going three main stages like this:

Broadcast to the industry: Who is willing to offer me life cover lasting 20-30 years for a male aged 35-40 with no details falling outside the normal 90% ranges?

Replies come back, either ‘Yes’ or ‘Yes, if he is a non-smoker’, ‘Yes, if he passes our medical test’, etc.

To those who replied: please quote for a person with these details who wants maybe this cover or maybe that cover.

Replies come back from each company for each cover requested: ‘No offer’, ‘This amount’, ‘That amount, and we throw in this for free’, ‘We need to know this as well’, or ‘We cannot offer what you ask for, but we can offer this instead’.

To those who provided a quote or a request for more information: how about this cover instead?

The same type of replies come back. Repeat this cycle until the best combination of cost and benefit has been found.

To the company offering the best deal: we like the figures you quote, are you willing to supply this cover to this person?

Reply either ‘Yes’ or ‘No, for this reason’. Maybe he has taken out three life policies in the last month and we are suspicious, or he defaulted on payment for a previous policy, or he came up on our ‘warning’ list so we may be willing to supply the cover but have to investigate and will come back to you as soon as possible. If the reply is ‘Yes,’ the full set of details are sent back for confirmation, plus a reference, in a signed, non-repudiatable message.

If the answer is ‘No,’ perhaps ask for another combination of company and cover.

To the company who said ‘Yes’: we accept your offer with this reference, in a signed, non-repudiatable message.

It should be noted that one simple use of the RSA algorithm (as proposed for use in the existing EDI Service) is that it can be used to encrypt data in such a way that nobody else could have done it. Use of the algorithm to ‘sign’ a message in this way means that the sender cannot claim later that they never sent it, that it is a forgery, or that some details have been changed. It is much more secure than a signed piece of paper.

In the scenario described above, the Client knows that the Insurer cannot deny making the offer of underwriting this business, and the Insurer knows the Client cannot deny accepting the offer and promising to pay the money.

One feature of doing business this way would be to bring insurance into line with most other products. A greengrocer says, I can offer you these carrots and those peas at this price. You choose whether or not to buy what is being offered. You may choose to ask questions - are they grown locally? - but in the end, you, the customer, decide whether or not to buy. The shopkeeper does not decide whether or not to sell.

Another feature of this scenario is that we do not have a quote request in the current sense. It is more of an invitation to respond to this set of information. The only thing which corresponds to our existing proposal form is sent from the Insurer to the Client.

Leaving aside the details, the key point is that EDI can be used in an interactive way to negotiate the best deal available to meet this Client’s needs in a rapidly changing marketplace. The static, rigid concepts of quote and proposal were only needed when everything was done by hand or on paper forms.

A related point is that the basis of calculation does not need to be tailored to the needs of a paper based system. If you need to look up premiums in a book, you have to work with age next birthday. A computer can calculate a different (but more correct and therefore more competitive) premium for every day of your life. Now there is an incentive for signing today rather than tomorrow. The premiums could be updated every hour, although it may not be helpful for the figures to be changing as you sit and chat to the Client.

Similarly, there is no good reason why any policy should not be able to cover you for 14 years, 1 month and 23 days. It should be no more expensive to quote for this than to quote for 14½ years, 14 years and 2 months, or for just 14 years, and if that is what the customer wants to buy, is there any reason not to supply it?

We can therefore use existing data far more precisely. We can also ask more precise details. For example, we can ask real questions about occupation. Entry of a potentially dangerous occupation can trigger supplementary questions to determine the extent of the risk. Each occupation with each set of supplementary details can then have a precise effect on the premiums of benefits. It is all just number crunching, and computers are good at that.

Not only is the rigid division into quote and proposal no longer required, we do not need to know what data is required at the start of the process: if something is missing, the Insurer can ask for it. We can even design the system so that we do not need to know at the outset what data may be asked for. Part of the conversation could go like this:

To the Insurer: please give me a quote on this data.

Reply: I’ll give you a quote if you tell me the size of feet.

To the Insurer: what is a ‘size of feet’?

Reply: it is a two digit integer, it belongs to the Life Assured, it has this reference for the internal storage and EDI, it has these characteristics, and uses this screen prompt and help text.

Providing this kind of mechanism would, perhaps, silence for good those who argue that the technology is driving the business. Insurers could change their questions and hence their calculations as often as they wanted. Nobody, apart from the strictly neutral Industry Database Administrator, would have any warning that the questions were about to change until the change was implemented. No only do you no longer have to print and distribute tens of thousands of copies of the revised form with the new question before it can be used, you no longer have to distribute thousands of copies of the revised software with your new question.

All insurance business can be summarised as follows. The Insurer makes the following offer which prospective Clients respond to.

•  some money (b) if an event (e) takes place within a period (f);
•  some money (c) if the event does not take place; and
•  some money (d) whatever happens;

•  the details described above, from (a) to (f);
•  who you are (g); and
•  anything (h) which would affect the probability of event (e).

Note that the factors (h) may affect the probability of event (e) either by making it more likely or less likely to occur. This is recognised by car and household insurance but not at present by life assurance where the assumption is made that any non standard factor will increase the risk. Hence, people are not encouraged to improve their chances of staying fit and healthy by the lure of lower premiums.

The money (a), (b), (c) and (d), can be single and/or regular amounts, the regular amounts can be all the same or vary according to a fixed calculation or depend upon an agreed external factor such as the RPI.

The event (e) can only be one the customer is legally entitled to insure against, so this can be broken down into a number of discrete options.

The period (f) is very straightforward.

The customer (g) again should be perfectly simple. If we do not all have a unique identity card number by then, we can establish a unique insurance industry reference number, so the personal details will only have to be supplied when you first apply for some insurance or when they change.

The only real complexity is in establishing what information may be required for (h) in determining the probability of the event. But we already have proposal forms for each existing type of policy, so there is a reasonable starting point, and because we need to use the precise details entered to calculate the risk, we ensure that information is captured, and a ‘catch-all’ question caters for anything we have not thought of.