It is, however, the nature of the losses under compensation policies which does most to render ratemaking a difficult problem. The complex schedule of payments of the various acts makes provision for compensation which may be continued for several years, in some cases even to the death of the recipient. All such payments are charged to losses under the policy which covered the risk at the time of the accident. Since the first policies were written in 1911, and the large majority since then, it follows that a large portion of these charges for deferred losses must be in the nature of estimates, subject to the inevitable error which accompanies estimates. The tendency of the error in this case was considered on pages 218-220.

The Machinery of Rate-making.—The efforts which are being made to solve the problem of securing correct premium rates are commensurate with its complexities and difficulties. Individual insuring organizations have created statistical and actuarial departments and are devoting much study to the subject. The stock insurance companies maintain the National Workmen's Compensation Service Bureau, an organization the purposes of which are the gathering of statistics of workmen's compensation insurance, the translating of these statistics into manual rates, and the development of systems of schedule and experience rating. This central bureau has branches in numerous states and its activities extend throughout the country. Certain states have created bureaus for the regulation of rates; manual, schedule, and experience;

1 California, New York, Massachusetts, and Pennsylvania.

and the approval of rates by the insurance commissioner is required in several instances. Some states determine the rates for their state funds by legislative action, a method which tends to maladjustment.2

Coöperation is absolutely essential to justice in the making of rates and it is particularly desirable that such coöperation embrace all of the parties in interest. Even though much had been accomplished in the various bureaus, it was evident that greater uniformity and accuracy could be attained by a more comprehensive group and, at the suggestion of Insurance Commissioner Hardison of Massachusetts, the Joint Conference on Workmen's Compensation Rates met in New York City during the last months of 1915. At this conference two state rating bureaus were represented by committees of their company members (both mutual and stock); the Workmen's Compensation Service Bureau by a company committee; and five state insurance departments by official delegates.3 Committees were chosen to consider specialized branches of rate-making, and opinion from every angle was brought to bear on rate problems. It is to be hoped that such conferences may become a permanent

2

This practice obtains in Montana, Nevada, Oregon, Washington, and Wyoming. In Wyoming the law provides that each employer shall contribute two per cent of his annual pay roll to the Industrial Accident Fund. This disregard of relative hazards is symptomatic of the generally unscientific nature of the Wyoming act.

The two state bureaus were those of Massachusetts and New York. The state departments represented were those of California, Maryland, Massachusetts, New York and Pennsylvania. The Industrial Commission of Wisconsin was also represented.

feature of compensation rate-making. The present rate manual owes much to the labors of this conference. To quote from the special report of the New York State Insurance Department, "Taken as a whole, the work of the Conference is a long step forward in the application of scientific rating methods to the business of workmen's compensation insurance."

METHOD OF CALCULATING MANUAL RATES*

The several factors which might properly be considered by an insurer in the calculation of a rate have been described; where practically possible all factors applicable to a given rate should be considered. present all of the factors mentioned in the preceding chapter have their influence in the calculation of rates except "underestimate of outstanding losses," "effect of merit rating," and "profit." The first of these factors represents a tendency which may disappear with improved methods of computing reserves for unpaid losses; the second, a tendency which more scientific rating will probably eliminate; but the third should be considered, at least by the stock companies. It is expected, of course, that rates as now calculated will yield a sufficiently liberal income to pay all losses and to leave a margin of profit, even though these three items are not specifically considered.

The influence which the several factors exert on the rate are expressed by differentials and loadings. A

The method here described is that used in the computation of rates for the major part of compensation insurance written in the United States. Departures from this method are in matters of detail rather than of principle.

differential is unity plus or minus a percentage figure, the percentage figure being a variable which is to be applied to a basic quantity in order to make it representative of new conditions. For example, if it were determined that the risk of damage from fire in all classes of buildings in community "A" was fifty per cent greater than in community "B" and the rate of loss were known in "B" for each class, the rate for a particular class in "A" would be found by multiplying the rate for the same class in "B" by the differential, 1.50. A loading is a flat amount or a percentage added to the expected annual loss cost to provide for contingencies, expenses, or profit.

Basic Pure Premiums.-The first step in securing a manual rate is the determination of the basic pure premium; a quantity representative of the loss experience under the original Massachusetts law, checked by statistics from other states where compensation has been in force for a considerable length of time and where the figures are reliable. This pure premium is used as a basis for calculating rates for all other states, since it is regarded as the most accurate expression of loss cost experience which it is possible to secure at this stage of development of the business.

Law Differential.-Since the provisions of the laws governing compensation payments vary from one state to another it is necessary to modify the basic pure premium in order to determine what the probable loss experience would have been under the act of each state. The degree of modification is expressed by law differentials which measure variations between the schedules of compensation.

Law differentials are calculated by the application of the compensation schedule of the various acts to the Standard Table of Accident Gravity. The Standard Table 5 was constructed by Dr. I. M. Rubinow and was the result of a careful study of American and European statistics of accident gravity which seemed to show that the physical results of accidents averaged much the same everywhere; that out of a very large number of accidents the same number would result in death, in dismemberments of various kinds, in total disability of one week, two weeks, three weeks, etc., and in partial disability of varying duration. On the basis of these statistics the table shows per 100,000 accidents:

I. Number of fatal cases.

2. Number of dismemberments, distributed according to their nature.

3. Number of cases of permanent total disability other than dismemberments.

4. Number of cases of permanent partial disability other than dismemberments, distributed ac-. cording to the percentage of disability.

5. Number of cases of temporary disability, distributed according to duration.

Another table furnishes data of the number and degree of dependents in fatal cases.

The probable cost of 100,000 accidents under a given compensation schedule may be found by computing the payments which would be made on account of each group of deaths or disablements in the table,

5 See p. 229.