nificantly more favorable than it was at the time the latest amendatory legislation was enacted (in 1961). In other words, the present program is adequately financed over the long run, according to the best actuarial estimates available, and so may be said to be actuarially sound within the limits of variation that are naturally inherent in such long-range cost estimates. It is recognized, however, that some additional financing is necessary for the DI portion of the program, but this can be obtained by reallocation of the total contribution income of the program without any increase in the overall contribution rates and still would leave the OASI portion of the system in a better financial position than it was at the time the 1961 amendments were approved. DECEMBER 2, 1963.

Memorandum from : Robert J. Myers. Subject: Actuarial cost estimates for hospital insurance bill under rising earnings and rising hospitalization-cost assumptions.

I have been asked to discuss certain assumptions concerning future earnings and hospitalization costs that were suggested during the hearings before the Committee on Ways and Means on the hospital insurance bill and to indicate why I used the assumptions that I did rather than the alternatives suggested. The nature of the alternatives discussed and their effect on the costs of the hospitalization benefits proposal and of the OASDI system

One alternative that has been discussed would assume the continuation into the long-range future of recent trends in the relationship between hospital costs and the general wage level, while at the same time assuming that on into the longrange future there would be no change in the maximum earnings base under the OASDI system.

In the recent past, the general earnings level has increased at a rate of about 3 percent a year, while hospital costs have risen about 6 or 7 percent a year, so that there is a differential of about 3 to 4 percent. Assuming the continuation of these trends into the indefinite future and assuming at the same time no change in the maximum earnings base would have the following effects:

(1) Eventually hospitalization costs would exceed 100 percent of the earnings of all workers in the country-let alone of taxable earnings.

(2) Virtually everyone entitled to cash benefits under the OASDI system would have the maximum benefit prescribed under the law, since they would have their benefits figured on the maximum creditable earnings. The earnings. The earnings of the lowest paid part-time worker would eventually rise to the present maximum earnings base.

(3) The cash benefits of the OASDI system would be only a very small proportion of a person's previous earnings.

(4) As a percentage of taxable payroll, the cost of the OASDI system would be considerably less than it is presently estimated to be-to the extent of about 14 percent of taxable payroll.

I did not use such an assumption in the cost estimates because I consider that it is completely unrealistic and is even an "impossible" one. It is inconceivable that hospital prices would rise indefinitely at a rate faster than earnings because eventually no one-a currently employed wage earner, let alone an older person— could afford to go to a hospital under such cost circumstances.

As a numerical example, let us consider a full-time male worker now earning the "typical" amount of $20 per day, or $5,200 per year (which is, by coincidence, the same as the earnings base under the bill). The average daily cost for hospitalization (including not only room and board, but also other charges) for persons of all ages is about $40 currently, or twice the average daily wage. If wages increase 3 percent per year, and if hospital costs increase 7 percent per yearindefinitely into the future then the following situation will occur:

Quite obviously then, it is an untenable assumption that there can be a sizable differential between the increase in hospitalization costs and the increase in earnings levels that will continue for a long period into the future.

One important reason for the fact that recent hospitalization costs have risen faster than the general earnings level is that the wages of hospital employees have risen at a faster rate than the general earnings level. Personnel costs are about 60 percent of all hospital costs. The fact that the wages of hospital employees have been rising at a faster rate than all earnings reflects a "catching up" from a situation where hospital workers were significantly underpaid in relation to other workers.

It is obvious that such a trend cannot continue and that a point will be reached after which wages paid to hospital workers would rise, on the average, at the same rate as the general earnings level. Nor can other elements in hospital costs be assumed to rise indefinitely at a faster rate than the general earnings level. It is not unlikely that the price of hospital services will for a considerable time rise faster than other prices, but if the price of any product continues to rise faster than earnings, it would eventually be priced out of the market. Actually, over the long run, hospital costs to the consumer are likely to show conflicting trends. On the one hand, improved technology is leading to more expensive hospital services and to the need for additional personnel. On the other hand, the duration of hospital stays is declining as a result of the improvement in care.

Another alternative that has been discussed is to assume a continuation of recent trends for a period of time, say 10 years, with hospitalization prices from then on rising at the same rate as earnings, but with no change in the maximum earnings base indefinitely into the future. This assumption has the same basic defects and weaknesses as did the previous assumption of continuously rising earnings and hospitalization costs rising continuously at a more rapid rate, except that the effects are somewhat deferred. The only major difference is that hospitalization costs would not exceed 100 percent of the earnings of all workers in the country, although they would exceed 100 percent of taxable earnings.

Still another alternative that could be considered is to assume a continuation of recent trends for a period of time, say 10 years, with both hospitalization costs and earnings leveling off thereafter, and with no change in the maximum earnings base from the presently proposed $5,200 at any time. This assumption has the following effects:

(1) The estimated cost of the hospital insurance proposal would be somewhat higher than presently estimated. But, on the other hand, the estimated level-cost of the OASDI cash-benefits program, using the same assumptions, would be lower than now by about 0.6 percent of taxable payroll. This reduction in estimated cost of the OASDI cash-benefits program would be somewhat greater than the increase in the hospitalization cost estimate of about 0.5 percent of taxable payroll that would occur under these assumptions.

(2) The highest earnings subject to contributions would be more than 30 percent below average earnings. As a result, much as under the previous alternative assumptions, a vast majority of those entitled to OASDI benefits would eventually be getting close to the maximum benefits, and these benefits would represent a relatively low proportion of past earnings.

I did not use this assumption in the official cost estimates because I consider it also completely unrealistic. It seems to me that it would be unwise to base the cost estimates on an assumption that, even though earnings may rise substantially in the future, no adjustments would be made in the cash benefits and the maximum earnings base. Such an assumption leads to estimates showing a significant reduction in the level-cost of the OASDI system and would indicate the system to be substantially overfinanced. This might then result either in a decision to increase cash benefits now or to reduce the statutory schedule of contribution rates.

Yet, as a matter of fact, as increases in earnings did occur, it would become necessary to increase the benefits under the program if the beneficiaries are not to be forced to live at a level which, with time, would become increasingly below that of other Americans. Thus, since these assumptions result in estimates showing lower costs for the cash benefits than will occur, because they are counting on gains to the program from assumptions that will almost certainly require the program to be liberalized, they are not conservative assumptions. In my opinion, they are dangerously lacking in conservatism.

27-166-64-pt. 1- 26

The nature of the assumptions used in the official estimates and why they were chosen

The assumptions concerning the relation of hospitalization costs and earnings used in the official estimates are described in Actuarial Study No. 57, dated July 1963. One of these assumptions is that, on the average, into the indefinite future there will be a continuation of the present ratio of hospital prices to wages.

For purposes of the estimated level-cost expressed as a percentage of taxable payroll, it is not significant whether hospitalization prices rise for a considerable period of time faster than earnings and then later rise at a rate somewhat lower; whether hospitalization prices remain level and earnings remain level; or whether they rise together from now on. From the standpoint of the cost estimates, the result is the same as long as, on the average and for the long run, hospitalization costs bear a constant relationship to earnings, or, in other words, in the event that earnings go up that hospital costs, on the average and over the long run, rise at the same rate.

In the event that earnings do rise as they have in the past, it is assumed that the maximum earnings base will be increased from time to time. This is a realistic assumption based on past performance. The maximum earnings base has been increased from $3,000 in 1939 in a series of steps until today it is $4,800. If, however, by any chance the earnings base were not increased for a few years in the future, even though earnings rose, then the system as a whole would still be actuarially sound since the savings to the OASDI cash-benefits portion, under any set of reasonable assumptions, would more than offset the loss to the hospitalization portion.

I believe that it can be seen from the previous discussion that this assumption, rather than those mentioned earlier, provides the most realistic basis for financing the program because:

(1) If increases in earnings are to be anticipated before they occuras proposed under the alternative assumptions-the cash-benefits portion of the system would appear to be so overfinanced currently that there would be great pressure to increase current benefits on the basis of an ability anticipated under the assumptions to finance them.

(2) On the contrary, in the assumption that I used, no allowance is made for savings which will occur as earnings rise. It is assumed instead that the OASDI system will be kept reasonably up to date, in that increases in earnings and increases in the maximum earnings base will occur in a parallel manner and will be offset by increases in benefit amounts and that, therefore, the contribution rates scheduled in present law are as necessary in the event of rising earnings as on the assumption of level earnings.

(3) In the event that hospitalization costs do rise, and if the maximum earnings base and the deductibles are kept up to date, the estimated hospitalization costs would be met fully by the presently proposed financing basis. If, for a time, adjustments in the earnings base fall somewhat behind earnings increases, there will be such savings to the overall program that sufficient funds will be generated to more than support the entire program without increasing the contribution rate, but then funds would have to be reallocated between the cash benefits and the hospitalization portions of the system.

The CHAIRMAN. Mr. Curtis.

Mr. CURTIS. First, I want to explore a little bit the problems of cost. I recognize that some of the data that I am going to ask for is available and yet I have found that I have had to interpolate in order to get at the point that I think is true, which is the problem of hospital cost as it affects the individual citizen.

What we are directly seeking is the impact of hospital costs on the individual.

Now, it is frequently pointed up, of course, that hospital costs are rising, and I think they will continue to rise in the consumer price index. Yet, I do not know the figures and I have seen these, that the average length of stay in a hospital for sickness has gone down. Am I correct?

Dr. WILSON. I believe, in my experience, it has leveled out for the time being.

Mr. CURTIS. Actually, it has gone down on an average nationally about 3 days in the past 10 years, something like that.

Dr. WILSON. Over the past 10 years. I was only giving you the last couple of years.

Mr. CURTIS. That is important when we compare costs in relation of one year to another and look at the trend.

I am going to ask if you will supply for the record the data which I direct attention to. This is one that I would like to get, the up-todate figures on the average length of stay in the hospital. This becomes important because if you go into a hospital for a particular illness or accident and, true, the cost per day has gone up but if the length of stay is 3 days less on an average, or whatever, then in order to understand the impact on the individual, it is the cost per day times the length of stay. So we need to get as best we can the cost of illness.

So, to any extent that there is information available on kinds of sickness and length of stay, I would like to have it. There are tremendous productivity increases in the developments in the hospital. In other words, fewer people can do more in meeting a particular illness. Actually, the cost per illness of a kind in many instances has dramatically dropped rather than increased.

Now, the other figure which I have never seen and yet I have been asking for it for years; first, we ought to have the figures on the rate of recovery per hospital stay in relation to morbidity. I have often pointed out that one reason hospital stays are shorter than 20 or 30 years ago is that the person would die in the first 2 or 3 days but the rate of recovery per hospital stay, as opposed to those who die, has increased I suspect quite dramatically; yet these figures I have never seen and I think this bears more importantly on this question of cost than anything else.

Of course, if the person dies as a result of the sickness he or she has no problem as far as costs are concerned.

I have often pointed out that the problem we have in the health care of the aged is the result of the success of our system, not its failure.

People are actually living, they go to the hospital, the costs are up and they live, with the problems of bearing that cost and a lot of other factors.

So, if you have figures, and I think you could develop them or maybe they are already available and I haven't seen them, of what the rate of recovery is per hospital stay and then, as best we can relate that again to this problem of the average length of stay in the hospital which relates again to the cost per day.

Do you follow what I am trying to seek?

Dr. WILSON. Yes, I think we can get that. (The information referred to follows:)

HOSPITAL COSTS, 1946-62

Average hospital expense per patient-day in the United States in 1962 reached a new high of $36.83. This was an increase of $1.85, or 5.3 percent, over 1961. Hospital expense per patient-day is an aggregate figure derived by dividing total expenses (including outpatient expenses) by the number of

adult inpatient days. Excluded are those expenses incurred by inpatients but not billed by the hospital; such as the cost of medical and surgical services rendered by the patient's physician. Data on average hospital expense per patient-day are computed by the American Hospital Association based on annual surveys of all hospitals accepted for listing by the association.

Between 1946 and 1962 total expense per patient-day in non-Federal shortterm general and special hospitals nearly quadrupled. The total expense per patient stay did not increase as much during this period, because the average length of stay was shortened by about 16 percent in the interim-from 9.1 days in 1946 to 7.6 days in 1962. Total expense per patient stay averaged $279.91 in 1962, an increase of 226 percent over the average cost of $85.57 in 1946.

Basically there are two trends in hospital economics that account for the rise in hospital expense per patient-day: (1) the expansion in the range and volume of services provided and (2) increased payroll expense. A patientday of care today includes more services and facilities and more intensive treatment than was the case in earlier years. As the quality and quantity of care provided hospital patients increases, hospital expenses will inevitably rise. But this factor alone does not account for the rise in costs. Payroll expense, which accounted for 62 percent of hospital expenses in 1962, went up sharply from $619 million in 1946 to $4.2 billion in 1962, or nearly sevenfold. Historically, the pay scales of hospital employees have been considerably below those of other workers. The continuing upward adjustment in hospital pay scales, including the substitution of cash payments for fringe benefits and reduction in the workweek, have been major elements in the sharp rise in payroll expense. Average payroll expense per employee rose from $1,226 in 1946 to $3,506 in 1962, an increase of 186 percent. This measure is not entirely accurate in that it assumes that the same types of employees were hired; actually the expansion of services required employees who were more highly skilled and hence could command higher pay.

Percentage increases in hospital costs for short-term general and other special hospitals

2 Total expense divided by the number of adult inpatient admissions.

3 Includes full-time personnel plus full-time equivalents of part-time personnel. Excludes residents, interns, and students.

The increase in hospital utilization and in the usage of complex technical services have necessitated increases in the number of employees, and particularly of higher salaried technical and professional employees. Since 1946, the number of full-time equivalent employees per hundred patients has risen approximately 60 percent, from 148 employees per hundred patients in 1946 to 237 per hundred in 1962.