by company, cost of hiring new employees if the additional hiring expense is significant, cost of excess spoilage by new employees, and demurrage.

Space does not permit a review of the reasons for qualifying some of those items and for omitting some others frequently cited, such as overhead. In any event, the relative unimportance of this group may be seen in the tabulation above. It did not exceed 5 percent in any injury category.

Background of Estimating Costs

About 25 years ago, H. W. Heinrich of the Travelers Insurance Co. investigated a very large number of cases drawn from his company's files; he concluded that the uninsured ("indirect") costs resulting from industrial accidents amounted to about four times as much as the compensation and medical payments made by the insurance companies.2 (Under this formula total costs are five times the direct costs.) He intended that ratio only for a general average and recognized that it did not apply to individual companies. Originally the National Safety Council, and later the President's Conferences on Industrial Safety, noted the need for another means of calculating accident costs. Accordingly, in 1946 the National Safety Council authorized a study of industrial accident costs by the writer in order to develop a reliable method of estimating the cost to an individual firm of its industrial accidents. It is from this study that information summarized here is drawn.3

The first major step was to determine precisely what elements of uninsured costs clearly result from industrial accidents and are subject to reasonably satisfactory measurement. Previous studies were never suitable for comparison because they were not based on a standard frame of costs. Indeed, the results of one or two such studies that were made with stop-watch accuracy were largely vitiated by the inclusion of some invalid cost element that dominated the total cost estimate. The significance of the accompanying list is not in the inclusion of heretofore undiscovered elements of cost, but rather in the refinement of earlier lists, with the elimination of duplicating or invalid items.

Careful analysis showed, as executives and accountants had suspected, that use of the 4 to 1 ratio between uninsured costs and the so-called

'direct cost" (i. e., the amount of compensation and medical payments) is an unsatisfactory way of estimating the uninsured cost to an individual company. The fundamental reason is that no consistent correlation exists between the uninsured cost and the "direct" cost of individual accidents. Also, "direct" costs occur primarily in lost-time cases, but the ratios among numbers of the various types of cases are not constant among establishments, nor even through different periods in one company. Furthermore, under the method of arriving at the total accident cost by multiplying the "direct" cost by 5, one severe infection or an amputation with its attendant heavy direct cost might triple the estimate of the uninsured cost.

However, use of some ratio is sound, because it would be too costly to make a permanent practice of recording the particular costs of every accident. The following method was therefore developed and is now recommended by the National Safety Council in its Safe Practices Pamphlet on accident cost.

Cost Element Method

Average uninsured costs per case are found for each of the four types of accidents (i. e., first-aid, doctor's, lost-time, and no-injury) mentioned earlier. Since it has become good standard practice to keep records that show the number of each of these types of accidents (except the noinjury accidents, which may be estimated), it is easy to multiply the number in each category by the previously determined average uninsured cost per case. Adding these products together gives the total uninsured cost. This plus the insurance premiums (or the cost of self-insurance) is the total measurable accident cost for the ordinary run of accidents. Catastrophies, or extraordinarily costly accidents, should be investigated individually, not included in the averages. By catastrophes is meant, here, accidents so serious that comparable costs are not likely to be met more than once in 2 or 3 years, if ever. For many companies, deaths would be in that category.

Under this method the total cost estimate is no longer tied to the single factor of "direct cost" as it is in the well-known "4 to 1" formula. Chance variations in injury severity will not affect the estimate of uninsured cost, which will become

strictly a function of the actual number of cases recorded in each accident category.

The only difficult part of this procedure is securing the constants representing the average uninsured cost per case and a means of estimating the number of no-injury accidents. This may be done best by conducting a pilot study in the company for which figures are wanted.

Brief forms for supervisors to fill in and detailed instructions for the safety investigator have been prepared. These were used in gathering the data referred to earlier. Use of this method starts before the accident occurs, and estimates on certain items are made by the supervisor the day the accident happens.

The length of the pilot study necessary depends on the size of the concern and the nature of its accident experience. A satisfactory first-aid average can be computed in a few days, if normal records are kept in the medical department. Doctor's cases may require 2 or 3 months of investigation and lost-time and no-injury cases, 6 months or a year, perhaps. From the pilot study it is possible to determine a ratio between the number of lost-time cases and the number of no-injury accidents.

The four categories of cases may be more precisely described as follows, beginning with the lost-time injuries:

(1) Permanent-partial and temporary-total disabilities.

(2) Temporary-partial disabilities and medicaltreatment cases requiring the attention of a physician outside the plant.

(3) Medical-treatment cases requiring only firstaid or local dispensary treatment and resulting in property damage of less than $20.

(4) No-injury accidents or accidents causing very minor injury not requiring the attention of a doctor, but resulting in property damage of $20 or more or loss of eight or more man-hours.

As this method comes into common use, a body of information as to average costs for each category gradually will be built up. Many concerns will, no doubt, choose to sacrifice a little in accuracy and use average uninsured costs for their type of industry established elsewhere, in order to avoid the necessity of conducting a pilot study.

The only data available currently on average uninsured costs for the several categories of accidents or on the ratio of no-injury accidents to lost-time cases are those determined by the writer from his analysis of over 2,000 cases which occurred in 1947. These data, together with instruction sheets and sample forms for the development of similar figures, will be available shortly in book form.4

Estimates based upon those figures should be more complete and accurate than were developed by the former methods. It should be emphasized, however, that individual companies can further improve the accuracy of their estimates by conducting their own pilot studies to establish specific averages and ratios for their particular operations. Such pilot studies seldom will cost more than a few hundred dollars and may produce collateral savings far in excess of their cost. Participating supervisors frequently become very safety conscious when the elements of accident. costs are disclosed through their own investigations. Some remarkable reductions in accident frequency have resulted merely from the added interest in safety, thus created.

Accident prevention is important any time, but in a period of material and labor shortages, such as appears to lie ahead of us, it is particularly important to the owners of industry and to the welfare of the Nation.

Associate Professor, Michigan State College.

1 In safety literature there has been occasional confusion in the use of the terms, "accidents" and "injuries." Despite the opinion of some safety leaders that a satisfactory definition of an accident is impossible, the following definition appears to be satisfactory for purposes of cost analysis. Industria' accidents are those unintended occurrences arising out of employment that either cause personal injury or cause property damage or interference with production under such circumstances that personal injury might have resulted. This makes industrial accident synonymous with the kind of occurrences industrial safety directors are trying to prevent. Therefore, the accidents include the standard categories of industrial injuries plus no-injury accidents.

2 For lack of any other plan for estimating those uninsured costs, safety leaders and engineers made grateful use of the ratio. It has often been misapplied in such a way as to result in an actual ratio of 6 or 7 to 1. The fact that such discrepancies went unnoticed only bears out the fact that it was not taken very seriously by executives when applied to their own firms.

Forty-five industrial organizations cooperated in the study. As a consequence, the method developed is based on the experience of several hundred supervisors, and the recommendations of accountants, personnel men, and superintendents. The method has been tried out in the cooperating organizations, and detailed cost analysis of between 2,000 and 3,000 accidents has provided indications of typical cost relationships.

See Simonds, Rollin H., and Nuernberger, George: Industrial Accidents, Cost and Methods of Prevention, to be published within the year by Richard Irwin, Inc.

945927-51- -2

Trends in Wages in 1950

FREDERICK W. MUELLER*

BEHIND THE widespread wage increases during 1950 lay the gathering momentum of business recovery from the mid-1949 low and the change in economic climate induced by the Korean outbreak.

By June of 1950, the Nation was enjoying a high level of prosperity, having overcome the uncertainty of the early months of recovery. With the start of the Korean action in late June, a new set of elements was introduced which had deep significance for every sector of the economy. The economic implications of Korea were etched even more sharply by the Chinese Communist intervention in November.

In 1949, the level of activity in manufacturing was such as to keep average hourly earnings practically stable. Nonmanufacturing wages, to an extent catching up, fared somewhat better. On the whole, however, money wages were more stable in 1949 than in any other postwar year. Collective bargaining focused mainly on supplementary practices, particularly pension and insurance plans. These issues were not dropped in 1950, but the increasing concentration of collective bargaining on wage rates undoubtedly slowed down the developments in fringe items.

That a substantial wage movement was developing became reasonably clear during the first 6 months of 1950. Average hourly earnings in manufacturing rose 31⁄2 cents from January to June, although part of the rise was the result of a longer workweek at premium overtime rates. The number of workers receiving general increases was proportionately greater than in 1949. The amounts of the hourly adjustments in manufacturing, however, were not large compared with

previous years; 5 cents was probably the most common figure. Among nonmanufacturing industries, both the relative volume and the size of the increases appeared to be somewhat larger.

The Korean conflict led to sharp changes in these developments. Subsequent wage negotiations proceeded in a framework of an initial rush by consumers for goods, of sharply rising prices, of anticipated price and wage controls, and of anticipated diversion of production to military needs.

Under these conditions, post-Korean wage settlements moved ahead at an accelerating rate. Both the number and size of wage adjustments increased. Many companies gave their second increases of the year even where union agreements did not require a wage reopening. In some major wage situations, notably basic steel, aluminum, and northern cotton textiles increases averaging about 10 percent were negotiated. The cost-ofliving escalator and deferred increase provisions reincorporated in the General Motors-UAW contract in May were adopted by other companies employing large numbers of workers. These developments continued up to the initial wageprice stabilization orders in January 1951.

Various measures are available to indicate how much earnings rose in 1950 as a result of these forces. They include data on money and real wages, weekly and hourly earnings, wage rates and take-home pay. By far the most complete information refers to production workers in manufacturing industries.

Manufacturing Industries

Gross weekly earnings of production workers rose 13.3 percent from January 1950 to January 1951. Gross hourly earnings advanced somewhat less-9.7 percent. The greater rise in weekly pay is the result of a lengthening of the average workweek-from 39.7 hours in January 1950 to 41.0 in January 1951. Most of the rise in gross hourly earnings during the year was clearly traceable to higher wage rates. Part of the rise was due to increased premium overtime payments that accompanied the lengthened hours, and, to a smaller degree, shifts of employment to higher wage industries. It is estimated that, excluding the effect of premium pay for overtime,2 hourly earnings increased 8.5 percent. Excluding the effects of both

wages occurred before the Korean conflict. The Bureau's Consumers' Price Index was slightly more than 1 percent higher in June than in January 1950. From June 1950 to January 1951, the advance was 6.6 percent. Consequently, the rise of 4.5 percent in gross weekly earnings in the pre-Korean period was reduced only to 3.3 percent by application of the Consumers' Price Index. The 8.3 percent advance from June 1950 to January 1951 fell to 1.6 percent when deflated by the rising prices of cost-ofliving items.

Industry Variations. There was variation in wage movements among different components of manufacturing during the year. On the whole, though, similarity of change was more pronounced than dissimilarity. Average hourly earnings, excluding overtime and the effect of employment shifts among different industries, rose 8.2 and 7.8 percent in durable and nondurable goods production, respectively. This strong parallelism remains when the pre-Korean and post-Korean periods are considered separately. In other words, wage movements were widespread and by no means confined to a few sectors of the economy. The similarity in trends did not extend in the same degree to the industries comprising each of the two broad groups just discussed but the differences were generally not great. In the 11 durable-goods industry groups, the rise ranged from 6.8 to 11.2 percent. Variations among nondurables were somewhat wider. At the low end was the 4.0 percent increase in the printing and publishing industries. At the top was the 10-percent rise in the tobacco industry, which was to some extent affected by the 75-cent minimum wage put into effect under the Fair Labor Standards Act in January 1950. The parallelism persists when the wage-rate changes are regarded in cents-per-hour terms. Although the increases ranged from 7 to 14 cents, 17 of the 21 industry groups showed average increases between 9 and 13 cents.

Such differences in the size of wage changes as did occur were not sufficient to affect substantially interindustry relationships in hourly earnings levels. The ranking of the various industry groups by gross average hourly earnings remained almost unchanged between January 1950 and January 1951.

Changes in weekly earnings varied more among industries than did hourly increases. Gross weekly earnings in durable goods rose somewhat more rapidly than nondurables, 13.9 and 10.6 percent, respectively. In the pre-Korean period the rapid growth in durable-goods hours was the primary cause of the 5.8-percent rise in weekly earnings; this increase compared with a 1.9percent change in weekly earnings in nondurable production from January to June 1950. Subsequent to June 1950, the situation was reversed, with nondurable manfacturing hours and weekly earnings advancing faster.

TABLE 2.-Average hourly earnings of production_workers in manufacturing and in selected nonmanufacturing industries, by major industry groups, January and June 1950 and January 1951