In the United Kingdom the customary approach to non-life insurance has been to rely more on the practical experience and flair of the underwriter than on a statistical examination of the experience. In some branches of non-life insurance an office is able to make adequate adjustments to its rates on the basis of claim ratios (i.e. the amounts of claims divided by the premiums) for fairly broad groups. In the more competitive branches such as motor insurance the experience needs to be examined in much greater detail, not merely to satisfy the minimum requirement of guarding against selection of the better risks by other insurers, but more positively in order to find out the extent to which the risk varies with each of the existing rating factors and any possible alternatives. Also, because of the rapidity with which conditions are liable to change, the experience studied must be as up to date as possible if reasonably reliable estimates of the future are to be made.

The object of this paper is to look at some of the practical problems which have to be faced in an analysis of motor insurance statistics and to point towards some of the solutions. In order to cover the ground, it has not been possible to examine any of the issues in any great depth and only private car insurance is considered. Much of the theoretical background to the subject has already been very ably put to this Society—in particular in papers by R. E. Beard and P. D. Johnson—with the result that a certain element of repetition has been inevitable, but this paper is in no way intended to be a complete theoretical treatise. The intention has been to consider various problems, discuss some actual statistics and their implications and then to continue. It is hoped that this may lead to a better understanding of the practical issues involved and at the same time may encourage others to delve into a quite entrancing subject which offers considerable scope for useful research.

I use the expression ‘money market’ to describe those institutions in the City of London who are active in borrowing and lending money for short periods, from a single day to perhaps 5 years at the most. From being a tight circle centred on the Bank of England, the discount houses, and the clearing banks the money market has now grown to include, as active investors, public corporations, local authorities, merchant banks, overseas banks, hire-purchase companies, building societies, trustee savings banks and Lloyd's syndicates. In addition, institutions with long-term liabilities, such as insurance companies and pension funds, will occasionally have business to transact in short-dated securities, an example being the profitable employment of funds earmarked for the purchase of real property in a few months' time. In any event, it is worth while to remember that the long term consists of a series of short terms.

Last year the Manchester Actuarial Society asked me to speak to it on my experience before I entered the profession, with particular reference to investments. This invitation caused me to review my experience over about twenty years in the Trustee Department of a Bank in supervising a large number of miscellaneous portfolios ranging from quite small ones to substantial pensions funds. The general conclusion which I reached was that investment theories generally were surprisingly little help. This was a conclusion which was rather depressing and I invited the Society to do their best to prove me wrong, adding that for once in a way the speaker would be only too pleased if someone did just that.

Before committing myself to writing the present paper, I awaited that by Messrs Weaver and Hall given to the Institute earlier this year, and the discussion on it, as I hoped that I would find in either one or the other some of the answers to the fundamental questions which were troubling me. I regret to say that whilst I am not in any way criticizing the computer method, I find that these fundamental difficulties still persist.