A Standard Accident Table I. M. Rubinow, Ph. D. THE SPECTATOR COMPANY The Poetry of Earth is Newer Dead James S. Elston CORNELL UNIVERSITY LIBRARY Date Due Cornell University Library HD7101 .R89 A standard accident tobjg as a basis for olin 3 1924 032 467 536 Cornell University Library The original of this book is in the Cornell University Library. There are no known copyright restrictions in the United States on the use of the text. http://www.archive.org/details/cu31924032467536 A STANDARD ACCIDENT TABLE AS A BASIS FOR COMPENSATION RATES Distribution of 100,000 Accidents BY I. M. RUBINOW, Ph.D. PRICE, $1.50 THE SPECTATOR COMPANY CHICAGO OFFICE: 135 WILLIAM STREET, Insurance Exchange. NEW YORK. * a Copyright 1915, By THE SPECTATOR COMPANY, NEW YORK. f)^p¥/^ A STANDARD ACCIDENT TABLE AS A BASIS FOR COMPENSATION RATES. By I. M. Rubinow, Ph.D., Chief Statistician, Ocean Accident and Guar- antee Corporation, Ltd. The rapid growth of compensation legislation in this country- has created a sudden demand for accident statistics which the American statistical literature was utterly unable to meet. The majority of American compensation acts are based upon private compensation insurance which may be solvent only when based upon accurate premium computations. The Ger- man method of assessment insurance for industrial accidents can only be successful when built upon a compulsory organ- ization of the insurance carrier. This method has been fol- lowed in a few states only, notably Washington and West Virginia. The difficult problem of calculating proper rates for compensation insurance, rates which should not be excessively high and yet sufficient to meet all the slowly developing charges resulting from accidental injuries, found the business of casualty insurance somewhat unprepared. The develop- ment of compensation legislation was so precipitated and so unexpected that there was little time to prepare the necessary material. As compensation was a substitute for the discredited system of employer's liability, so insurance of the obligations created by compensation acts developed out of the methods of employ- er's liability insurance. Naturally enough the casualty com- panies (that is the insurance companies writing so-called mis- cellaneous lines, including almost all lines outside of life, fire and marine, liability insurance being the most important of these) were forced to rely upon the experience under lia- bility insurance for making hurried estimates as to the prob- able cost of accident compensation. It became evident very soon that the two systems were entirely different, not only from the point of view of social justice, but also from that of insurance practice. The experience under one was of little 3 4 A Standard Accident Table. service in shaping methods for the conduct of the other branch of insurance. For the benefit of statisticians not familiar with the conditions of employer's liability insurance these may be briefly stated here: An employer's liability insurance contract obligates the insurance company (for a certain consideration stipulated in advance and known as the premium) to reimburse the insured employer in case he is called upon to pay damages for accidental injuries sustained by his employees in accordance with the verdict of a court. In practice the insurance company, in order to reduce the possible losses through such court verdicts, undertakes to defend the suit and also endeavors to settle amicably out of court as many suits, or claims threatening to become suits, as possible. Naturally, the losses, while depend- ing upon the frequency and severity of the injuries sustained, do not at all bear any definite relation to them. Perhaps in an equal degree they depend upon the stringency of the lia- bility legislation of each particular state, upon the nature of the court decisions as to the \alidity of the "three defenses," upon the skill of the attorneys in defending suits and arguing appeals, and upon the efficiency of adjusters to settle a large number of claims before the injured employees are tempted to bring suit against the employer. In compensation insurance the cost in any one industry evidently depends entirely upon the number and character of the injuries occurring in that industry, as well as upon the amount of compensation granted by the particular law. Under employer's liability laws there frequently is no logical or equitable relation between the character of the injury and the cost of settlement. Very much more depends upon the existing legal evidence as to the cause and the manner of occurrence of the accidents. For this reason the legal features of the cases are much more carefully studied than their surgical or medical features. Every large casualty company had some sort of a statistical organization even before the advent of compensation legislation, but the statistical work done ap- proached the problem from a different point of view. The experience studied dealt largely with the relations between money losses paid and premiums collected, or the exposure in Distribution of 100,000 Accidents. 5 dollars of pay roll, so as to obtain either the loss ratio or the so-called pure premium.* Very seldom were the accidents reported studied, and hardly ever were their physical nature or their economic effect upon the earning capacity of the wage-earner considered. This experience for numerous minute subdivisions of indus- try was almost the only available source of information upon which to build compensation rates. No wonder then that the rates at first proposed were in the nature of guesses only, and many of them went far beyond the mark. An effort was made by the central rate-making organization of the casualty com- panies to utilize the experience of the so-called workmen's collective insurance, t This is a form of insurance more closely approaching workmen's compensation because it covers em- ployees of certain establishments for all accidental occupa- tional injuries irrespective of fault, and the benefits are subject to a definite schedule, though very much more limited than any schedule of benefits under compensation. The records of this form of accident insurance, had they been properly kept and had there been enough of them, would have presented very valuable information as to the number and nature of accidents in separate industries. Unfortunately the volume of this busi- ness available to the rate-making bureau was much too small, perhaps a paltry $100,000,000 of wage exposure, equal to some 200,000 employees working for one year. This amount, divided into many different industries, leaves very little to depend upon for most independent industries. Besides, the actual records were very far from satisfactory as to the nature of the injury and duration of disability. Instead of the acci- dent experience under these collective policies, their loss expe- rience was utilized as a starting point and the rates charged for *By Iobs ratio is meant the percentage proportion between losses and premiums. By pure premium is understood the coat of losses per hundred dollars of wages paid by the insured employer to his employees. This method of computation, as well as the reference of all computations to the exposure of wages paid rather than persons employed, is unsatisfactory from the point of view of theoretical accident statistics, but is explained by the conditions of the business. Ideal accuracy would undoubtedly require that the number of employees and the time spent in work be computed, but there are so many practical difficulties in the way of accurate determination of these facts by thousands of independent employers that the rough and ready measurement of the risk by the volume of pay roll in practice presents much more accu- rate results. tSee article, "Establishing Rates for Workmen's Compensation," by Theodore E. Gaty, Market World and Chronicle, January 11, 1913. 6 A Standard Accident Table. insurance under that form of contract. But as these rates were originally computed from very limited data, the build- ing could not be stronger than the foundation. There is no doubt that the dissatisfaction of employers with the conduct of compensation insurance during the first experimental years was largely caused by the absence of accurate data from which scientific rates could have been built. The question has often been asked why the voluminous and carefully studied accident statistics of European countries has not been utilized at this critical moment. Germany, for instance, having begun insurance against industrial accidents nearly thirty years earlier, possesses almost a library of statis- tical material on this problem. There were several reasons why that experience was not and could not have been utilized. To begin with, a widespread feeling of skepticism in regard to all European experience may often be found even among students of social conditions, and is especially strong among practical American business men. While in life insurance mathematically accurate data have been in use for many decades, casualty insurance is still managed very largely as a business and very little as a science. Knowledge of foreign statistical sources among casualty men, with a few notable exceptions, was rare. It was easy to argue that European statistics were utterly inapplicable because, as everyone would readily admit, our own industrial activity is so much more hazardous, accident prevention less known, factory inspection less efficient, and for these reasons the number of industrial accidents is very much greater. Moreover, the various European compensation acts are quite different from each other, especially in their provisions as to the prescribed scales of compensation, and the American acts still more different from the European acts, and unfortunately, from each other. Furthermore, European accident statistics are published and studied according to many different methods. Efforts to establish a uniform system of accident statistics in Europe, though advanced by the International Statistical Institute, have not as yet been successful. The definition of what con- stitutes an accident, the distribution of accidents into groups according to their surgical and economic results (that is as to Distribution of 100,000 Accidents. 7 the physical nature of the injury and as to its effects upon the earning capacity of the victim), the classification of industries used, etc., all this is different in every known statistical source. Finally, the statistics published are in such a form as to be much more useful for general social purposes than for the computa- tion of rates. It may be very instructive to know that the accident frequency in the metal working industry is 50 per cent, higher than in the woodworking industry, but the insur- ance carrier does not undertake to insure either the one or the other. Insurance must be written and rates quoted for cer- tain industrial establishments manufacturing definite articles or for contractors performing definite jobs, and no European source presents its data in sufficient detail to meet the demands of the practical casualty underwriter for specific information in regard to some 1,500 different classifications of industrial undertakings. All this explains why it is impossible to draw directly upon European sources for the necessary information. Does it follow, therefore, that all European experience is alto- gether worthless to help meet the difficulties arising out of the application of compensation acts in this country during the first few years? In an article published early in the history of compensation* the writer suggested a method for computing compensation rates by means of utilizing all possible American and European statistics. The method suggested may be outlined by means of the following quotations : All scientific rates must include the following three elements : first, agent's commissions; second, expenses of central admin- istration; third, the pure cost or pure premium. In the matter of workmen's compensation in the United States, the first two of these items depend entirely upon American conditions, and are easily controlled and computed. All the uncertainty, all that is still obscure, lies in the third item — the factor of pure cost or pure premium. Now, the pure premium itself may be resolved into three factors: first, the accident rate; second, the proportion of the various classes of accidents (that is as determined by degree of disability, duration of disability, etc.), to the whole number of accidents; third, the compensation scale to be used. •Arriving at the Cost of Workmen's Compensation in the United States, Market World and Chronicle, June 22, 1912. 8 A Standard Accident Table. Summarized, the plan calls for the following operations: first, determine the rates of accident in the various industries on the basis of American experience; second, determine the relative proportion of fatalities of permanent disability cases, partial or total, and the relative duration of temporary cases, on the basis of European experience; third, applj to the results thus obtained through the combination of American and European experience the various compensation scales of the various American compensation laws. The first suggestion outlined above, as to what may be termed accident frequency, had in view a possible utilization of the many thousands of accident reports transmitted to the casualty companies by their policy holders. For a careful study of accident statistics these reports are of little value because of their faulty character, but it did seem to the writer that the simple count of these reports for each separate indus- try might give a sufficient indication of relative accident frequency. The most important part of the plan is contained in the second suggestion that European experience might be utilized for the purpose of distributing accidents into groups according to gravity. This suggestion has found enthusiastic support among some actuaries and insurance statisticians, while it has been severely criticized by others. One may perhaps be ex- cused for stating this suggestion in the terms originally used: It is when we come to the proportion of the various classes of accidents that we meet with the greatest difficulties for here we have almost no available American experience at all . . . we must look for light elsewhere . . . until we have gone far enough with workmen's compensation in the United States to have our own experience. ... As regards this problem of the nature and effects of injuries . . . there can be no great differences between one country and another. An exam- ination of the statistical material of various countries shows that there is practically no variation from country to country. It appears that whatever its nationality, the human machine is a human machine, and that its average resistance to injury and its average speed of recovery from injury vary very little. As regards the first suggestion that the accident frequency of various branches of industry be ascertained from all records of casualty companies dealing with liability insurance, it has Distribution of 100,000 Accidents. 9 never been practically utilized. With the rapid accumulation of experience under compensation insurance, the necessity for making use of this method is passing away. While the failure of the casualty companies to make use of the plan proposed seems unfortunate because it might have prevented many errors, this at present is a matter of purely historic interest. The second suggestion as to the method of studying acci- dent gravity was destined to play a more important part in the development of compensation legislation and insurance. Though extremely simple in itself and almost obvious to any- one who has made any study at all of foreign accident statistics, it appeared like a new scientific clue to many actuaries who were groping for some practical method of computing the cost of compensation. It was impossible to make any direct com- parisons between the American compensation scales and those of Germany or France and determine in that way the relative cost of the different acts. The differences in the scales were so many that it was necessary to weight each difference in these provisions in accordance with the frequency with which it might come into play, and the comparative share of each group of accidents in the total cost of all. As far as the writer's information goes, Professor A. W. Whitney, at that time consulting actuary of the California Industrial Accident Commission, was the first to make use of the above suggestions, and apply a hypothetical distribution of accidents based upon European experience to this problem, to determine the probable cost of the proposed California Acts as compared with the then California Act and with other acts in existence at the time.* Another California student of the problem, Mr. A. H. Mowbray, a consulting actuary who was chairman of the Social Insurance Committee of the Com- monwealth Club of San Francisco, has also applied this method to a comparison of the proposed California Act with the exist- ing Illinois and Massachusetts Acts.f Mr. Mowbray made a careful analysis of the statistics of various countries as pub- *See "Memorandum concerning a proposed scale of Compensation Benefits to be paid to Workmen injured through industrial accident now under consideration by the Industrial Accident Board of Cali- fornia," 1912. tSee a typewritten memorandum, entitled "A Suggested Basis for the Determination of Comparative Costs of Different Compensation Schedules, and Some Comparisons thereof." 10 A Standard Accident Table. lished in the twenty-fourth Annual Report of the United States Commissioner of Labor, and concluded that "the close agreement" between these various statistical data "seems to strengthen the case in favor of Mr. Rubinow's hypothesis." Thus the plan outlined above was never used in its entirety for the purpose of computing compensation rates, but it proved to be quite adaptable to the specific purpose of ascertaining the difference in the cost of different compensation acts, or what may be called a differential between different acts. While the efforts of Professor A. W. Whitney and Mr. A. H. Mowbray were of theoretical importance only and did not serve as a basis for rate computation, the first practical applica- tion of the method occurred in connection with the New York Compensation Act, passed by the New York legislature in December, 1913, to go into effect July 1, 1914. The act con- ferred upon the New York Insurance Department the duty of passing upon the adequacy of the rates to be charged for com- pensation insurance. The declared purpose of this legislation was to protect the policy holders as well as the injured employ- ees against the possible danger of insolvency of the insurance carriers, which might result from inadequate rates. In order to act intelligently under the new power conferred upon it, the insurance department faced the problem of establishing some standard measure of rate adequacy at the time the New York Compensation Act, different from all other acts and more gen- erous than most of them, was going into effect. The problem appeared extremely difficult, almost insolvable. A compari- son with the rates in force in other states at the time did not seem helpful, since these rates were subject to many criticisms. Fortunately, the Insurance Department of Massachusetts, which had called upon the insurance companies early in 1914 to furnish results of compensation insurance in that state, was able to prepare the experience (that is, the relation between losses sustained, premiums received, and the wage exposure) on some $500,000,000 of wages by separate industrial classifica- tions. Thus some basis for calculation of pure premiums was available. The rate-making bureau of the casualty companies agreed to utilize this volume of experience in computing New York rates. Distribution of 100,000 Accidents. 11 The writer was consulted by the actuaries of the New York Insurance Department and was asked to suggest a method for comparing the cost 1 of the two acts. The method suggested embodied the principles above outlined, that is, a separate valuation of a given series of accidents, reflecting the normal distribution according to gravity, under both acts, and the comparison of the resulting cost. By a happy coincidence the preliminary results of the tabulation of industrial accidents in Massachusetts for the first year of operation of the compen- sation act (July 1, 1912, to June 30, 1913), made by the Indus- trial Accident Board of that state, also appeared at the time. The 90,000 accidents reported in Massachusetts were therefore used as a basis for the computation. The problem assumed the following form : how much would these 90,000 accidents cost according to the Massachusetts law and how much according to the New York law, a proportion between the two amounts being the true differential between the laws of the two states. But while the Massachusetts report was extremely useful, it did not analyze its material in sufficient detail to enable one to compute the cost accurately. Expecially was the report unsatisfactory in regard to the question of permanent partial disability, the number of such cases, and the degree of dis- ability resulting. As the New York act is extremely liberal in regard to such cases, they represent a very substantial part of the entire cost of compensation, and some conception as to their number and gravity was absolutely necessary. For this purpose very detailed figures of Austrian and German accident statistics (analyzed in the Twenty-Fourth Annual Report of the United States Commissioner of Labor) were recommended. This was perhaps the first formal application of European experience to American compensation insurance. The discussions accompanying the adoption of the New York compensation rates emphasized the great necessity of some scientific method for computing such rates for all com- pensation states. In most other states where compensation acts were in operation grumblings were heard against exorbi- tant and inequitable rates. Comparisons of rates in different states for identical industries were made by industrial accident 12 A Standard Accident Table. boards as well as private employers, and justificationb for the differences disclosed were urgently demanded. The difficul- ties created by the co-existence of so many different acts were obvious. Not only did this make the work of preparing rates for each state so much more difficult, but it also vitiated com- parisons of experience unless some method of eqjuating these differences were applied. To Professor A. W. Whitney (who became general manager of the Workmen's Compensation Service Bureau early in 1914) credit is due for the plan of bringing the insurance rates in all compensation states into harmony through a proper sys- tem of differentials instead of trying to prepare an independent schedule of rates in each state on the basis of its own experience. Systems of state differentials were not altogether new in casualty insurance. They had been in use for a time in con- nection with employer's liability insurance. But while the old system of state differentials was largely a result of under- writing judgment (as to the comparative stringency of liability laws and as to the general conditions in various states) Pro- fessor Whitney's plan presupposed a scientifically accurate objective method of computation. For the purpose of devising such a method a committee of casualty insurance experts was appointed by Professor Whit- ney.* It was agreed that the method of computing costs of a given number of accidents, as advocated by the writer, promised the most satisfactory results. Instead of trying to depend upon the unsatisfactory accident statistics of any one of the American states or foreign countries, it was decided to estab- lish one definite standard schedule of distribution of accidents according to gravity of the result, or in other words, to con- struct something akin to the standard mortality table in life insurance. This first step necessary for the computation of differentials was entrusted to the writer. The study which follows was prepared in response to this demand. As far as the writer is aware, it is the first effort of its kind ever made in *Thie committee consists of Mr. B. D. Flynn, Assistant Secretary of the Traveler's Insurance Com- pany of Hartford, Conn., Mr. Stanley L. Otis, Actuary of the Workmen's Compensation Service Bureau, Mr. C. E. Scattergood, Assistant Secretary of the Fidelity & Casualty Company, and the writer for the Ocean Accident & Guarantee Corporation as chairman. Distribution of 100,000 Accidents. 13 the domain of accident statistics or compensation insurance. It has been adopted by the committee mentioned above, whose work received the unanimous approval of the Workmen's Compensation Service Bureau, and it lies at the foundation of almost all compensation insurance rates at present in force in this country. Definition of an Accident. In endeavoring to use European statistics and especially to compare the data of one country with those of another, the first serious difficulty is that arising out of the differences in regard to the definition of what constitutes an industrial acci- dent. Reports on accident statistics have been strongly influenced by the provisions of compensation laws. In fact, there were scarcely any satisfactory accident statistics in any country previous to the enactment of compensation acts. There are, however, many differences between compensation acts as to the kind of accidents covered by their provisions. Several acts leave the care of minor accidents to other co-exist- ing institutions, especially where a well organized system of sickness insurance is found. Thus, to quote the best known illustrations, in Germany all accidents for the first thirteen weeks, and in Austria those for the first four weeks, are left to the care of sick benefit funds. Naturally, only those acci- dents come under the compensation system which result in losses extended beyond this minimum period. In other coun- tries injuries lasting less than a certain minimum length of time are left uncompensated altogether. In most cases the published accident statistics disregard accidents which do not extend beyond this so-called waiting period. Thus, German accident statistics cover only accidents leading to death, per- manent disability, and temporary disability of over thirteen weeks. In Austrian statistics, accidents leading to temporary disability under four weeks are excluded; in Italy, those under five days; in Russia, those under four days; in France, under three days, etc. Evidently, since the number of such minor accidents is very large, the percentage of all classes of serious accidents whether resulting in death or permanent disability, etc., will depend very much upon how many of those small 14 "A Standard Accident Table. accidents are included, or in other words, will depend entirely upon the accepted definition of an industrial accident. This difficulty does not necessarily make European data useless, but it requires an adjustment before comparisons are made, so that before the figures of any country are to be used, all accidents which do not come under the uniform classifica- tion must be eliminated, and on the other hand, for such coun- tries as Germany and Austria where the definition of an acci- dent is very narrow, the figures must be adjusted by means of some factor to compute the probable number of accidents under the uniform definition. Only after such adjustments are made does the principle hold true that the distribution of accidents according to grav- ity and consequences is fairly uniform in all countries. This is well illustrated by the following example showing the compara- tive distribution of 42,063 accidents treated by the Leipzig sick fund and 154,926 accidents occurring in Russia for the three- year period, 1904 to 1906: ClasseB of Accidents. Germany. Russia. Number. Per Cent. Number. Per Cent. 32,235 7,618 1,968 242 76.16 18.58 4.68 .58 113,353 32,896 7,596 1,081 73.13 21.28 4.90 Fatal .69 Total 42,063 100 154,926 100 A comparison of the figures in the above table seems to indicate a certain substantial difference between the distribu- tion of accidents in the two countries compared. As a matter of fact, the difference is that in the German sick funds a good many minor cases have been reported which have been elimi- nated from the Russian statistics because of their duration being under four days. This reduces the total number of accidents but increases the percentage for the classes over four weeks, over thirteen weeks, and fatal, at the same time reduc- ing the percentage of accidents under four weeks. These fig- ures are here computed on a basis of one fatal accident. We find the following series expressed in multiples of the number of fatal accidents: Distribution of 100,000 Accidents. 15 Leipzig. Russia. Fatal accidents 1 1 Over 13 weeks 8 7 Over 4 and under 13 weeks. . 31 30 Under 4 weeks 133 105 This clearly indicates the remarkable similarity between the statistics of the two countries except for the group of accidents of under four weeks' duration, where the material difference is explained by the considerations stated above. The definition decided upon is identical with that accepted by the Statistical Committee of the Workmen's Compensation Service Bureau for the study of accident statistics of the casu- alty companies; namely, only such accidents are to be counted and studied which disable the injured employee for any time other than the day of injury. The same definition was ac- cepted by several conferences of officials of bureaus of labor and by the American Association for Labor Legislation, and promises to become the standard definition of an industrial accident in the United States. It is practically equivalent to excluding accidents leading to disability of less than one day, because the injured person who does not return to work in the beginning of the day after the injury, is not likely to return at any time during that day. This definition of an industrial accident is very much broader than that used in most governmental studies of accident sta- tistics. However, a precedent may be found in the statistics of accidents to employees of the United States Government recently published.* There are practical reasons for selecting this definition. Even under liability insurance conditions, casualty companies were insistent that the minor injuries be carefully reported because the most trivial injury occasionally developed into a substantial claim or suit. The accident reports, or "notices" as they are technically designated, have been much more numerous than accidents of sufficient gravity. A narrow definition of "accident" would have entirely destroyed this basis of underwriting experience. •See Compensation for Injuria to Employees o] the United States Arising from Accidents Occurring between August 1, 1908, and June SO, 1911, published by the Department of Commerce and Labor, Wash- ington, 1913. 16 A Standard Accident Table. Certain difficulties are created by this definition because there is very little statistical material available to indicate the number of these minor accidents, and adjustments of almost all data become necessary. For the purpose of such adjust- ment, use was made of the statistics of injuries to United States employees, as explained below. Five Ghoitps of Accidents. Having accepted the standard definition, the problem is to ascertain the following relations : First — The relative number of fatal accidents. Second — The relative number of permanent total disability cases. Third — The relative number of dismemberments and their character. Fourth — The relative number of permanent partial dis- ability cases and their distribution by degree of disability. Fifth — The relative number of temporary disability cases and the distribution of such cases according to duration. While it would be quite practicable to deal in percentages extending the figures to any necessary degree of accuracy, it seemed preferable to assume a standard total sufficiently large to permit of all necessary sub-divisions and yet not so large as to appear quite fanciful. The number of 100,000 was de- cided upon since in many of the larger states that many in- dustrial accidents will occur in a year or two. The advantage of such a large base is that in the computations individual cases are dealt with and not fractions of cases. The above division into five groups is necessitated partly by the nature of the problem and partly by the peculiarities of the American compensation acts. There is a very fine line of demarcation between permanent partial disability and permanent total disability, but an independent determination of the latter cases becomes necessary because special provisions for them are made in most American acts. Still more impor- tant is the treatment of cases of dismemberment by specific benefits in American acts, instead of considering the propor- tion of earning capacity lost, which is the European method. Of these five sub-divisions, some are due to definite physical Distribution of 100,000 Accidents. 17 facts, so to speak, and not dependent upon differences in inter- pretation of the law. This includes fatal accidents, the num- ber and character of dismemberments, and perhaps the dis- tribution of temporary disability cases according to duration. On the other hand, when we deal with cases of permanent partial disability, and especially their distribution by degree of disability, we are dealing with facts which are largely influ- enced by differences in interpretation of law, and here the largest variation between one country and the other may be expected. In utilizing the statistics of foreign countries for the purpose of computing each one of these five groups of accidents, different checks and tests must therefore be intro- duced. Fatal Accidents. The experience of Massachusetts for the first year of the application of the Compensation act, as reported by the In- dustrial Accident Board, gives 474 fatal accidents out of the total of 90,168, or some 525 per 100,000. The Massachusetts report has no definition of accidents but counts all accident notices received. The following figures indicate the impor- tance of adjusting the accident definition for the purpose of obtaining the percentage of fatal injuries : Percentage of Fatal Accidents in Vabious Countries. Massachusetts (1912-1913) — 0.5 per cent, of all notices. Leipzig Fund (1887-1903) — 0.6 per cent, of accidents of over three days' duration. Russia (1900-1906) — 0.7 per cent, of accidents of over four days' duration. Italy (1902) — 0.75 per cent, of accidents of over five days' duration. Austria (1897-1901)— 4.1 per cent, of accidents of over four weeks' duration. Germany (1899-1908) — 8.4 per cent, of accidents of over thirteen weeks' duration. In not a single instance do we find a definition of "accident" which would correspond to the one adopted here. To make the Massachusetts data conform to this definition, it is neces- sary to eliminate from the 90,168 notices those accidents where 18 A Standard Accident Table. duration of disability did not last over one day. Their num- ber is not stated in the report, but it is indicated in a diagram that in 41 per cent, of all non-fatal accidents the disability did not last over one day: , * 90,168 — 474 fatal accidents = 98,694 non-fatal accidents. 89,694 X. 41 =36,774 accidents of not over one day's dura- tion. Deducting these, the total number of accidents in Massachu- setts in 1912-1913 is reduced from 90,168 to 53,394, and the 474 fatal accidents constitute 0.888 per cent., or 888 per 100,000 accidents. Before these figures are accepted, comparisons with a few other countries are necessary. In Austria from 1897 to 1901 there were 3,871 fatal accidents out of a total of 95,269 acci- dents of over four weeks' duration. To adjust this to our standard American definition of an "accident" the following method may be used: According to the report on accidents among United States employees for 1908 to 1911, the 20,835 accidents were dis- tributed as follows : Duration. 1 to 7 days 7 to 14 days 15 to 28 days All other Total 20,835 100.0 As accidents of over four weeks' duration constitute 25.8 per cent, of all accidents of over one day's duration, we may assume that the 95,269 accidents of over four weeks' duration in Aus- tria correspond to 95,269h-25.8 per cent. =369,260 accidents of over one day's duration on which basis Austria shows 1,048 fatal accidents per 100,000. For a period of ten years the German industrial accident associations reported 54,166 fatal ; accidents out of a total of 642,344 accidents of thirteen weeks' duration, the fatalities constituting 8.43 per cent. Again, the United States statis- tical report referred to above indicates that accidents of over Number. Per Cen 7,482 35.9 3,753 18.0 4,238 20.3 5,362 25.8 Distribution of 100,000 Accidents. 19 thirteen weeks' duration constitute only 8.89 per cent, of all accidents of over one day's duration. The number of accidents of over one day's duration in Germany for the ten years would be equal to 642,344^-8.89 per cent. = 7,225,460, and the num- ber of fatal accidents per 100,000 would be 750< In Italy a report for 1902 shows 430 fatal accidents out of a total of 57,617 accidents of over five days' duration, or 746 per 100,000. In Russia for a period of three years there were 2,345 fatal acci- dents out of 299,874 accidents of over four days' duration, or 782 per 100,000. In both these latter cases the true average would be somewhat smaller if all the petty accidents had been included. The United States report shows 670 fatal accidents out of a total of 20,835, or 3,216 per 100,000, which is evidently too high a figure explained by the very hazardous nature of the occupations covered by the Act of 1908. Averaging the fatal accident frequency in only four countries, Austria, Germany, Italy, and Russia, where a fairly general distribution of indus- tries and occupations obtains, the average is 837 per 100,000,- which is not very far from the Massachusetts figure of 888. In accepting the latter figure, it must be remembered that the Massachusetts report was prepared very soon after the com- pletion of the year. Experience in Germany has demonstrated that within four or five years the number of fatal accidents increases by about 5 per cent, because of the subsequent death of many injured persons whose injuries originally did not appear to be fatal. Loading, therefore, the accepted figure by 5 per cent., we arrive at 932 fatal accidents per hundred thou- sand injured. Dismemberments . Because most American acts contain specific schedules of benefits for cases of dismemberment, it is necessary to make an estimate of their number as well as the distribution among different kinds of dismemberments. As these cases are not treated in that fashion under the European compensation acts, comparatively little information is available. The best is that contained in the Austrian report for 1897 to 1901, which is the only source for exact data as to distribution of as large a num- ber as 8,686 cases of dismemberment according to the exact nature of the loss sustained in each case. 20 A Standard Accident Table. Some test is necessary to establish how far the proportion of dismemberments differs, if at all, from such American expe- rience as we have. The Massachusetts report shows 967 dis- memberments out of a total of 90,162 accidents. To make this comparable with the Austrian figures, all accidents of under four weeks' duration must be eliminated, leaving only 11,237 accidents of over four weeks' duration. Dismember- ments constitute 8.6 per cent, of that number. Of the Aus- trian 8,686 cases, 109 resulted fatally, leaving 8^577 cases, which in proportion to the 95,269 accidents of over four weeks' duration is 9 per cent. As Austrian figures are more complete because a certain number of cases must lead to amputation some time after the original injury has been sustained, it is felt that the Austrian and Massachusetts figures on that point are sufficiently comparable to permit the use of the Austrian figures. The experience from Michigan as reported in the National Compensation Journal of June, 1914, for six months, December 1, 1914, to May 31, 1914, gives 318 dismemberments and 6,046 cases of temporary disability, giving a proportion of about 5 per cent. But the Michigan report includes all acci- dents of over two weeks' duration. As the number of acci- dents of from two to four weeks' duration is approximately equal to the number of accidents of over four weeks' duration, it would indicate that cases of dismemberment in Michigan amount to about 10 per cent, of all the accidents of over four weeks' duration, which is even higher than the Austrian figures. Therefore, on the whole, the Austrian figures may be accepted. In the following statement is given a classified list of dismem- berments occurring in Austria within the period specified, and also the number per 100,000 on a basis of all accidents of one day's duration or over, for which purpose the 95,269 accidents of over four weeks' duration were assumed to represent 369,260 accidents according to our accepted definition. Distribution of 100,000 Accidents. 21 Kind of dismemberment. Number of cases in Austria, and whether resulting in permanent dis- ability or not, but exclusive of cases resulting fatally. Number per hun- dred thousands on an assumption of 369,260 cases. 1. Loss of left arm 2. Loss or right arm 3. Loss of left hand 4. Loss of right hand Loss of entire fingers, two phalanges or more: 5. Loss of left thumb 6. Loss of right thumb 7. Loss of left index 8. Loss of right index 9. Loss of left middle finger 10. Loss of right middle finger 11. Loss of left ring finger 12. Loss of right ring finger 13. Loss of left little finger 14. Loss of right little finger 15. Loss of thumb and one or more fingers, left hand 16. Loss of thumb and one or more fingers, right hand. . . . 17. Loss of two or more fingers, left hand 18. Loss of two or more fingers, right hand Loss of one phalange of finger: 19. Loss of left thumb 20. Loss of right thumb 21. Loss of left index , 22. Loss of right index 23. Loss of left middle finger 24. Loss of right middle finger 25. Loss of left ring finger 26. Loss of right ring finger 27. Loss of left little finger 28. Loss of right little finger 29. Loss of fingers accompanied b,y injuries of other fingers of the same hand, left 30- Loss of fingers accompanied by injuries of other fingers of the same hand, right 31. Loss of one leg 32. Loss of both legs 33. LoSBof toes 34. Lobb of one eye 35. Loss of one eye with injury to the other 36. Loss of both eyes Total dismemberments 8,577 2,323 Permanent Total Disability. Specific provisions for compensation of cases of permanent total disability are found in many compensation acts differing materially from provisions for permanent partial disability. In many states life pensions are granted for the permanent total cases where only temporary limited benefits are given for permanent partial disability. While cases of this nature are few, the cost must be a material factor as compared with the total cost of compensation. Therefore, some estimate as to the probable number of such cases becomes necessary. 22 A Standard Accident Table. Here foreign material must be used with extreme care because it is a condition largely depending upon the interpretation given to the term. In some cases there can be no difference of opinion as to the existence of permanent total disability. In others, however, it will largely depend upon the liberality with which compensation acts are administered. No Ameri- can data on the subject exist, and the European data show considerable fluctuations. The percentage of cases of per- manent total disability in official reports fluctuate between .06 per cent, in Italy and 1.51 per cent, in Austria. To a large extent this difference, as already explained, is due to the differ- ences in the definition of the word "accident," and the number of minor accidents excluded. To obtain a basis of comparison without the necessity of adjusting all figures to one uniform basis of an accident of over one day's duration, we have as- sumed as a measure the proportion of cases of total perma- nent disability to 100 fatal accidents with the following results: Number of Cases of Permanent Total Disability per 100 Fatal Accidents. Austria . 28 . 5 Germany 14.7 Russia. 12.8 France 8.4 Italy 7.5 These differences are not to be explained by physical conditions only. We find that the Austrian interpretation of the defini- tion of permanent total disability is the most liberal of all and almost twice as liberal even as that of Germany. Averaging the proportions in the five countries, we obtain 14.38 cases of permanent total disability per 100 fatal accidents, which is about one half of the proportion in Austria alone. However, in view of the excessive ratio shown in Austria as compared with other countries, the average for the five seems to be preferable, and this indicates 14.4 cases of permanent total disability per 100 fatal accidents. On an assumption of 932 fatalities per 100,000 accidents, we get 133 cases of per- manent total disability. As against this figure we have already included 14 cases of loss of both eyes and 3 cases of loss of both Distribution of 100,000 Accidents. 23 legs in the dismemberment schedule. We have also assumed 62 cases of loss of one eye with injury to the other, 10 per cent. of which, or 6 cases, constitute permanent total disability ac- cording to the Austrian experience, so that 23 cases of perma- nent total disability are already included in the dismemberment schedule, and cases of permanent total disability not dismem- berments are therefore reduced to 110. Permanent Disability Other than Dismemberment. This group of injuries presents perhaps the greatest difficulty in the effort to construct our standard accident table. Per- manent reduction in the earning capacity of the injured work- man, which does not necessarily destroy his economic use- fulness entirely, but puts him down in the economic scale, frequently follows the class of injuries described above as dismemberment. It is evident that a man with one arm is worth less in the labor market than he was while both arms were intact. But permanent partial disability is not at all limited to such self-evident cases. Unhealed fractures, badly reduced dislocations resulting in loose joints, stiff joints, partial paralysis, ruptures, and many similar cases may have the same economic effect as loss of part of body. Because dis- memberments have been made subjects of special regulation in most acts, it becomes necessary to separate those from other cases of permanent partial disability. Unfortunately, in a good many American acts, and still more in the administration of compensation laws in a good many American states, this condition of permanent partial disability due to other causes than dismemberment is not yet always sufficiently well rec- ognized.* Perhaps for this reason it is useless to expect any accurate reports of such cases in the early statistics of compensation in *An interesting illustration of thiB may be found in the report of the Michigan Industrial Accident Board aa to accidents occurring between December 1, 1913, and May 31, 1914, as published in the National Compensation Jam nal for June, 1914. The first table of this report is entitled: "Accidents causing per- manent partial disability classified by part of body affected," and in this table are listed 318 accidents, every one of which is a case of dismemberment. It seems that the Michigan Industrial Board was abso- lutely unaware of the existence of these cases of permanent partial disability which are not due to dismem- berment, and yet it is quite evident that a large number of such cases must have occurred. They are not self-evident at the time the accident has happened. In a great many of such cases a long time may elapse before the permanent nature will be realized. 24 A Standard Accident Table. this country, but it must not be assumed that because Amer- ican statistics have not recorded such cases, they do not occur in actual practice. Undoubtedly, they have been handled in each state as cases of temporary disability until the permanent character has been established. Undoubtedly, also, in a good many cases the reduction in earning capacity if not excessive has been and is being disregarded. It is known, however, that there have been such cases in the state of Massachusetts. The first report of the Industrial Accident Board indicates on page 324 that there have been 1,457 cases of permanent partial dis- ability though the number of dismemberments as stated on page 19 is only 967, leaving 490 cases of permanent partial dis- ability not dismemberments, or about 50 per cent, as many as dismemberments. Again, in the first report of the Industrial Accident Board of California covering the period from Septem- ber 1, 1911, to December 31, 1912, we find an analysis of 9,627 accidents, of which 412 were fatal and 534 resulted in perma- nent disability, and of the latter only 79 cases or about 15 per cent, are cases of permanent injury not due to dismemberment. In the state of Washington for the first year of operation of the act there have been 685 cases of permanent disability against 279 fatal accidents, the proportion being so high that it was quite probable that a large number of cases not dismen- berments have been included. But while their existence is being recognized, comparatively few of them are as yet recorded. It is necessary, therefore, to turn to European data to get a better conception of their importance and value. As already indicated, a comparison of statistics of various countries is somewhat vitiated by the differences in the definition of an accident, and since the reduc- tion of all the statistical data to one basis of accidents of one day's duration is a rather difficult undertaking, the method of comparing them with the fatal accidents offers a more con- venient shortcut. The following table shows a degree of variation in this pro- portion among various European countries: Distribution of 100,000 Accidents. 25 PERMANENT DISABILITY CASES IN PROPORTION TO FATAL ACCIDENTS. Countries. Years. Number of Fatal Accidents. Number of Acci- dents Resulting in Permanent Disability. Ratio to Fatal Accidents. 1897-1906 1905-19G8 1899-1906 1899-1908 1899-1908 1898-1902 1895-1905 1904-1906 8,349 1,838 389 18,708 59,893 2,224 832 2,345 82,446 8,204 4,192 140,877 313,219 9,101 4,496 34,981 9.9 4.6 10.8 7.5 5.2 Italy 4.4 5.4 14.9 94,578 598,116 6.3 The ratio of permanent disability cases to fatal accidents seems to vary from 4.4 in Italy to 14.9 in Russia, the average being 6.3. The ratio seems to be fairly uniform in four or five coun- tries such as Belgium, Italy, Germany, and Norway, being about 5 per cent., but rises to 10 per cent, or over in Austria, Denmark, and Russia. The variations are evidently sig- nificant of differences in the matter of judgment and decision in regard to individual cases rather than of bare physical facts. For all of these countries except Austria all cases of permanent partial disability, including those resulting from dismember- ment, are stated together. Since the dismemberment cases have already been treated in a different way, it is necessary to obtain information as to the number of other cases, and for this Austria alone offers statistical data. Because of this acci- dental advantage of Austrian figures over all others, they have been used very largely by the New York State Insurance Department in its work of determining a differential between the cost of compensation under the New York and the Massa- chusetts compensation acts, and the use of these Austrian figures was very violently contested. The difficulty cannot be denied that Austrian statistics indicate twice as large a number of cases of permanent disability as most other coun- tries. A careful inspection of the Austrian tables as re-pub- lished in the Twenty-Fourth Annual Report of the Commis- sioner of Labor does raise serious doubts as to whether such lib- erality of interpretation may be expected in any of the states for thefirst five or ten years, especially since it is twice as high 26 A Standard Accident Table. as in Germany. Even in Austria the proportion of such cases has been regularly increasing. Thus, in 1897 the proportion of permanent disability cases to fatal accidents was 8 to 1; in 1900, 9.8 to 1; and in 1906, 11.5 to 1. We feel safe, therefore, in assuming that for the next few years at least, five cases of permanent partial disability against one fatal accident will be nearer the actual conditions than the proportion obtaining in Austria now. In other words, we may safely reduce the number of permanent partial cases as indicated by Austrian figures by 50 per cent. In our table we have assumed 932 fatal accidents which would make the total number of permanent disability cases 4,660. The number of dismemberments we found to be 2,323 per 100,000. As a matter of fact, Austrian statistics indicate that out of a total of 8,579 cases of dismemberments 795, or 9.3 per cent., were so slight that even in Austria they were not considered as cases of permanent disability. Subtracting this 9.3 per cent, from the 2,323 cases, we have 2,108 cases of dismemberments per 100,000 leading to permanent partial dis- ability. Subtracting this number from the total assumed number of permanent disability cases, we get 2,552 cases of per- manent disability not due to dismemberments. Of these, as indicated earlier, 110 are cases of permanent total disability, leaving 2,442 cases of partial permanent disability not due to dismemberment. Decree of Permanent Partial Disability. The next question that requires investigation is the dis- tribution of these cases according to degree of disability, upon which compensation depends. Here also we must j draw entirely upon European data, and this again is a matter of judgment largely and everything depends upon the customary method of adjusting such cases. It is reasonable, therefore, to expect a very wide difference between results in different countries. When an effort is made to compare information on this point as given in the reports of different countries, an addi- tional difficulty arises from the fact that the sub-groupings in ■ different countries are not uniform. Thus, for instance, in Distribution of 100,000 Accidents. 27 German statistics the permanent disability cases are divided into only four groups: Under 25 per cent, disability, between 25 per cent, and 50 per cent., between 50 per cent, and 75 per cent., and between 75 per cent, and 100 per cent. Italian reports contain ten groups, one for each 10 per cent, of dis- ability, while Austria has a division into seven groups on an entirely different basis. Finally, the third difficulty consists in the fact that in all countries except Austria all cases of per- manent disability are lumped together whether due to dis- memberment or not. It is reasonable to assume that the degree of disability due to dismemberment is often more serious than that due to other causes. In view of the specific provi- sions for dismemberments, it is necessary to eliminate these from the cases of permanent disability, but the only country for which this is feasible is Austria. In the following table is shown the distribution of all cases of permanent disability in Austria according to the degree of disability. Similar data are given also for cases of dismem- berment as well as for all cases of permanent disability other than dismemberment. The first column indicates the rate of benefit in percentage of wages, which is the method of classification used in Austrian sources. The second column states the degree of disability covered by the rate of benefit, the Austrian law granting 60 per cent, of the loss of earning power. The actual number of cases in 1891 to 1901 is shown in column 3, and the percentage distribution in column 4. A computation of the average degree of disability is made in columns 5 and 6 on an assumption that the mean of the two limits is the average degree of disability for each group. The last column shows the results of this computation. As was to be expected, dismemberments lead to higher degrees of disability. This is well shown by the percentage column as well as by the average computed. The average degree of disability for dismemberments is 43.1, and for all other cases 29.1. But since there are numeri- cally a great many more cases of permanent disability in Austria not due to dismemberments, the average degree of disability for all cases (32.1) is only slightly affected by the higher degree of disability of the dismemberment cases. 28 A Standard Accident Table. AUSTRIA. All Cases op Permanent Disability. Rate of Ben- efit (Per Cent, of Wages). Degree of Disability. (PerCent.). Number of Cases. Per Cent. Average Degree of Disability. Total Amount of Disability (Average Degree X No. of Cases) . General Average. 5 or under 8 or under 3,629 9.8 4 14,516 6-11 9-18 11,707 31.7 13 152,191 12-19 19-32 7,126 19.4 25 178,150 20-29 33-48 5,760 15.6 40 230,400 30-39 49-65 3,487 9.4 50 195,272 40-50 66-83 4,098 11.1 74 303 252 60 100 1,104 3.0 100 110,400 36,911 100.0 1,184,181 32.1 Disbursements ■S)lk. : *!■', ' '. i?-