Employers' Liability Insurance (1882-1931)

A series of Acts of Parliament came into effect during the final two decades of the nineteenth century that extended British workers’ rights to compensation in the event of workplace injury or death. These laws meant a step-change in the magnitude of legal liabilities that employers would be exposed to in the event of workplace accidents. In nineteenth-century industrial Britain, this could represent a substantial business risk, particularly in the accident- prone manufacturing and extraction industries. Employers’ liability insurance quickly emerged as a significant new line of general insurance business. Its complexity, potentially long-term nature and exposure to mortality risks naturally attracted strong actuarial interest—on of the face of it, it certainly appeared a more ‘actuarial’ type of general insurance than fire insurance.

In Great Britain, as in several countries across Europe, the common law accorded injured parties some general rights to compensation from those who caused the injury. These rights were vague and limited and were not specifically related to injuries caused at work. The Employers’ Liability Act of 1880 gave explicit rights to workmen who were injured in the workplace due to defective machinery or the negligence of superiors. The compensation was limited in value to three years’ wages and was payable in the form of a lump sum. William Whittall wrote the first British actuarial paper to tackle the specific challenges raised by insurance of this form of liability. It appeared in the Journal of the Institute of Actuaries in 1882.[1] Like much of early, and indeed later, actuarial work in general insurance, many analytical compromises were required as a result of the limited quality and quantity of historical claims data. Whittall set out to analyse the rates of both fatal and non-fatal accidents and how they varied by occupation, but a sparsity of data constrained him to the statistical analysis of fatal accidents only.

In a similar spirit to Samuel Brown’s fire insurance analysis, Whittall had to splice together data from disparate sources, neither of which arose directly from insurance claims data. He used the 1871 census to obtain estimates of the numbers employed in different occupations. For accident numbers, he made use of the records of fatal accidents (in 1870, 1871 and 1872) that were maintained at the government’s General Records Office. These records had the helpful feature of recording the occupation of the deceased (although they did not distinguish between those accidents that occurred in the workplace and those that did not). From these data sources Whittall compiled tables of fatal accident rates for a comprehensive array of occupations. These results showed, for example, that only 1.3 authors per 10,000 died annually from a fatal accident, whereas horse breakers experienced 47.9 such deaths. Clearly occupation was a material risk factor for fatal accident rates!

The actuarial audience at the Staple Inn discussion of the paper, however, proffered a great deal of scepticism towards the reliability and relevance of these statistics. The President, A.H. Bailey, cautioned against the usefulness of the statistics for setting liability insurance premiums given that fatal accidents made up a relatively small proportion of liability claims and were not necessarily strongly correlated with the rate of non-fatal accidents. Whittall himself acknowledged that the granularity of the occupations meant that the numbers of deaths in some of the estimates resulted in unreliably small sample sizes. He also acknowledged that the fast-changing nature of some occupations could quickly render some historical rates obsolete. F.G.P. Neison, who had some experience of working with liability claims data from the coalmines and railways industries, highlighted that the records of the General Records Office may not provide a reliable record of the occupation of the deceased, and ‘did not provide any satisfactory basis for the determination of the rates of fatal accidents’.[2] As in fire insurance, the application of actuarial techniques continued to be frustrated by the lack of availability of relevant, reliable historical claims experience data.

The Workmen’s Compensation Act of 1897 extended the rights of employees to compensation for workplace injuries. This act entitled the employee’s dependents, if he had any, to three years’ wages in the event of a fatal accident, and in the event of a non-fatal accident that rendered the employee unable to work, 50 % of his weekly earnings were payable during the period of incapacity after the second week. Whereas the 1880 act only entitled the employee to compensation in the event of a form of negligence on the part of his employer, the 1897 act entitled the employee to compensation for any workplace injury, providing it was not wilfully committed by the employee. The 1897 act was also notable for establishing injured employees’ right to weekly incapacity compensation rather than only a lump sum benefit. Further Workmen’s Compensation Acts of 1900 and 1906 extended the industries in which the act was applicable. The 1906 act made its application almost universal. Similar forms of legislation were introduced around continental Europe and Scandinavia during the same period (though not in the USA, Canada or Australia).

A behemoth, prize-winning paper by John Nicoll, published in 1902, represented the next notable British actuarial contribution on the subject of employers’ liability insurance.[3] Some 20 years of liability insurance business experience had been accumulated in Britain by this time, but no actuarial basis for premium-setting had been established, even in theory, beyond the early efforts of Neison and Whittall on accident rates. Standard industry practice was to charge a premium amount set as a percentage of the total wages of the employees of the business.

Nicoll derived formulae for the value of fatal and non-fatal accident compensation, including the new feature of weekly compensation during the period of incapacity, from actuarial first principles. For the valuation of fatal accident compensation, Nicoll’s formula was a function of the rate of fatal accident of a given occupation (which was specified as a function of age); and the probability of the employee having dependants (also specified as a function of age). For the valuation of non-fatal accident compensation, his formula used probabilities of different degrees of incapacitation (total and partial) occurring for each occupation as a function of age; mortality rates for the incapacitated as a function of age (treating partial and total incapacity separately); and the rate at which non-fatal accidents recover from their incapacity. Assumptions were also required about wage levels and how they varied as a function of age.

Whilst the actuarial community at the Staple Inn discussion of the paper voiced appreciation for the application of actuarial techniques to this complex form of insurance, a visiting non-actuarial practitioner in employee liability insurance was much more sceptical:

Actuaries could not tell, any more than they [general insurance underwriters] could, what the rate should be, because they had no data ... the time, he thought, was not ripe for actuarial knowledge to be called in.[4]

Thus, by the start of the twentieth century, some recurring themes were already well-established in the British actuarial relationship with general insurance: a lack of data rendered statistical and analytical methods of limited use, and the actuarial role in general insurance was therefore at best unclear.

William Penman’s Journal paper of1911 was the first to focus specifically on actuarial approaches to reserving for outstanding claims rather than premium setting.[5] Penman was another example, like Samuel Brown, of a life actuary who dabbled in general insurance thinking, as well as investment thinking, and who would go on to reach a leading position in the profession (Penman was President of the Institute of Actuaries during the years 1940-1942).

The reserves required in a general insurance business can be considered to have a few high-level components whose relative size will vary with the type of business that is written. At the highest level, reserves can be required for two basic categories of claims: claims that arise due to future events that may occur during the outstanding period of insurance covered beyond the valuation date by the premiums already received; and claims for events that have already occurred but have yet to be settled.

The reserve for the first of these categories of claims is generally referred to as an unexpired premium reserve. It reflects the portion of the insurance premium received that is charged to cover the period of insurance beyond the valuation date. Alternatively, an unexpired risk reserve may be set up that is intended to reflect the portion of exposed-to-risk that is still outstanding on the insurance cover (these two quantities may differ for reasons such as seasonality effects—for example, more motor accidents might occur in winter than summer).

Reserves for claims that have already occurred are required due to delays in the claim reporting and claim settlement processes. These two forms of delay each has its own form of reserve: incurred but not reported (IBNR) reserves are for claims that have already occurred but are not yet known to the insurer (that is, there is a delay between the occurrence of the event that causes the claim and the notification of the insurer of the claim); and outstanding claims reserves are for claims that the insurer has been notified of but which are yet to be settled. Settlement delays may arise due to the time taken to agree the claim size. This process may simply involve negotiation between the insurer and the policyholder, or may have to be settled by the courts.

Penman’s paper focused on a particular form of outstanding claim reserve that arose from the 1906 act: the reserve that arose not from settlement delays, but simply because the form of settlement for incapacity compensation was in the form of an annuity for the period of incapacity rather than an immediately payable lump sum. The standard industry practice was to reserve for these outstanding claims by making individual case estimates of the required reserve based on medical opinion on the duration of the injury. To the trained actuarial mind, this approach appeared arcane, expensive, unnecessarily subjective and open to deliberate manipulation. Penman contrasted the case estimate approach with life office practice and argued it was ‘surely contrary to all the principles of valuation to deal with individual cases on their merit’.[6]

Penman’s paper adopted a pattern that was similar to Nicoll’s a decade earlier. He developed an actuarial method for valuing the reserves based squarely on standard life office techniques, only for general insurance practitioners to tell him that there was no adequate data with which to apply his formula. Penman’s approach was intuitive enough as an actuarial approach: the outstanding claim was a weekly annuity that would remain payable as long as the employee was alive and remained incapacitated. It could be valued using a mortality table for incapacitated workers together with a table for the rate of recovery from incapacity (which could be set as a function of age) and a valuation interest rate. In referring to the calibration of the recovery rates, Penman conceded:

The only way in which these items of information can be obtained is by a very complete analysis of a large number of claims and such an analysis has not been possible for me to make.[7]

Penman did acquire some claims experience data from two companies and he used this to develop illustrative bases and valuations. He showed that the rate of recovery from incapacity tended to be lower at older ages and argued this should be explicitly allowed for in the reserving basis.

The Staple Inn discussion was broadly negative. The themes were familiar. Several speakers voiced strong doubts about the availability of sufficient relevant data to implement the actuarial approach set out by Penman. There were so many factors affecting the duration of incapacity that the separation of claims into homogenous groups would produce such small groups that the result would not differ significantly from the case estimation approach that the author had strongly criticised. The business was in a state of flux, with factors such as judicial interpretation rendering historical experience less relevant to the future. One speaker pointed out that valuing life business by case estimation may not be so absurd if the information was readily available to support it.

It was almost two decades before another paper appeared in either of the British actuarial journals on the topic of employee liability insurance (or any other form of general insurance). Then, in 1929 and 1931, two papers were published, partly motivated by the 1925 and 1926 Workmen’s Compensation Acts. Hugh Brown’s paper,[8] published in the Transactions of the Faculty of Actuaries in 1931, reviewed the implications of the 1925 and 1926 Acts of Parliament. A notable feature of these acts was that they allowed the weekly incapacity compensation payments to be commuted to an immediately payable lump sum at the behest of the employer, providing the payment was not less than 75 % of the prevailing market value of an immediate life annuity. This provided a natural benchmark for reserving for outstanding claims, though the actuary had discretion to attach a different value to the liability in the regulatory board of trade returns providing a justification was provided along with details of the basis used. However, an actuary was only required for the valuation of weekly incapacity payments that had been in duration for five years or more. For other outstanding claims, the practice remained to value those liabilities on an individual case basis using available medical opinion. Hence, 20 years after Penman’s paper and almost 30 years after Nicoll’s, liability insurance practices remained largely unaltered by actuarial input: outstanding claims were either valued on an individual case basis, or were essentially assumed to be life annuities. No actuarial method for the estimation of the duration of the claims based on statistical analysis of rates of recovery from incapacity was implemented in practice.

Brown advocated a pragmatic compromise. Instead of valuing the outstanding claims using intricate assumptions about incapacity recovery rates and how they varied as a function of age, together with incapacity mortality assumptions, he came up with something rather counter to actuarial convention: that for claims of less than five years’ duration, age could be completely disregarded from the annuity valuation. Instead, he suggested grouping the claims by type of employment and then applying an average duration factor to each employment group that could be based on the analysis of historical data. He did not provide any analysis of such data and how these factors might vary by employment category, however. His suggestion, whilst pragmatic in theory, remained somewhat enigmatic in practice.

The motivation for the second of this pair of papers, written by R.G. Maudling and published in the Journal in 1929,[9] was related to the author’s own experience of analysing a large amount of liability insurance claims data. The data was specifically related to the mining industry and was sourced from the experience of one particular Colliery Owners’ Mutual Indemnity Fund. Mutual indemnity funds were a popular approach to coinsuring liability exposure amongst the employers of some industries, and the mining industry was one of the biggest examples. He considered the data for the period from 1908 to 1923, and his data set included 11,000 claims. Maudling categorised his data by the type of injury or disease that was the cause of the claim, and by age at date of accident. He found that the rate of discontinuance of incapacity was a function of both. He used the data to provide annuity valuation tables as a function of age and type of injury.

This was, in spirit, the implementation of the actuarial frameworks advocated by Nicoll and Penman some 30 and 20 years earlier, finally brought to life with a meaningful dataset. However, Maudling also noted that the observed rates of incapacity discontinuance had varied over the data period. In particular, the discontinuance rates tended to be lower when wages were lower and vice versa. Data heterogeneity appeared inescapable in general insurance. Maudling’s papers represented the final notable contributions of British actuarial thought to general insurance business until after the Second World War.

  • [1] ; Whittall (1882).
  • [2] Neison, in Discussion, Whittall (1882), p. 212.
  • [3] Nicoll (1902).
  • [4] Green, in Discussion, Nicoll (1902), p. 559.
  • [5] Penman (1911).
  • [6] Penman (1911), p. 112.
  • [7] Penman (1911), p. 121.
  • [8] Brown (1931).
  • [9] Maudling (1929).
< Prev   CONTENTS   Source   Next >