Second Meeting,
Wednesday, July 26, 2000
Thursday, July 27, 2000
Attendance:
Members of the Technical Panel:
Dale Yamamoto, Chairman
Principal
Benefits Practice
Hewitt Associates
Lincolnshire, IL
Alice Rosenblatt
Senior Vice President of Merger and Acquisition Integration
Wellpoint Health Networks
Thousand Oaks, CA
James Robinson
Associate Scientist
Center for Health Systems Research and Analysis
University of Wisconsin-Madison
Madison, WI
Michael Chernew
Associate Professor
Department of Health Management and Policy
University of Michigan
Ann Arbor, MI
David Cutler
Professor of Economics
Department of Economics
Harvard University
Cambridge, MA
Len Nichols
Principal Research Associate
Health Policy Center
The Urban Institute
Washington, DC
Consultant to the Panel
Sam Gutterman
Director and Consulting Actuary
Price Waterhouse Coopers
Chicago, IL
Staff of the Health Care Financing Administration(now known as Centers for Medicare and Medicaid Services(CMS)), Office of the Actuary
Richard Foster, Chief Actuary
Sol Mussey, Director of the Medicare and Medicaid Cost Estimates Group
Mark Freeland, Deputy Director, National Health Statistics Group
John Shatto, Actuary
John Wandishin, Actuary
W. Kent Clemens, Actuary
Nate Singer, Actuary
Jackie Carroll, Executive Officer
Donna Holt, Secretary
Terry Peach, Secretary
Staff of the Office of the Assistant Secretary for Planning and Evaluation:
Ariel Winter, Executive Director of the Panel, ASPE, DHHS
Gene Moyer, Consultant to ASPE and to the Panel,
DHHS
Others Present:
Phil Ellis, Economist, Treasury Department
Linda Baker, General Accounting Office
Linda Greenberg, Health Care Financing Administration(now known as Centers for Medicare and Medicaid Services(CMS))
Noah Meyerson, Congressional Budget Office
Robert Nguyen, Congressional Budget Office
Kevin Hayes, Medicare Payment Advisory Commission
Proceedings of the Meeting:
July 26
The meeting was called to order at 9:09 A. M. by Ariel Winter, who welcomed the group.
He indicated that Christy Schmidt would not be at the meeting because of other commitments and that no one from the public had indicated that they wished time on the agenda, but that they might do so before the end of the day. He also indicated that the minutes from the first meeting would hopefully be ready by the next meeting. Then he turned the meeting over to Mr. Foster.
Mr. Foster discussed all the materials that had been distributed to the panel since the last meeting and materials available today.
He also asked Linda Greenberg to discuss materials being developed for HCFA(now known as CMS) by the RAND Corporation. Ms. Greenberg indicated that these are based on meetings of medical and technical panels. Two of these are completed on medical advances available for treating the aged in the areas of cancer and cardiovascular care. There will be two more panels before the end of the contract.
Mr. Foster promised to bring draft reports submitted by RAND on the two panels already completed to the afternoon session.
Mr. Winter then turned the meeting over to Mr Yamamoto.
Mr. Yamamoto also welcomed the group and indicated that before the end of the meeting, the group would decide to divide itself into sub-groups to study specific topics and to report on them to the entire group.
Then, he turned the meeting over to Mr. Robinson to discuss work he had done on the Actuaries’ spreadsheets on the Hospital Insurance Trust Fund. He has written memos on the information flows in the spreadsheets which will be sent to the rest of the panel. He discussed the way the spreadsheets worked and made some suggestions for making the flow of the spreadsheets more obvious and making documentation of the sheets easier. He also suggested that other members look over the spreadsheets and his memos on them to make sure that his memos really are more easily understood than the original.
Mr. Foster then indicated that he had felt for some time that the documentation for the spreadsheets needed to be improved and that Mr. Robinson’s work should make this task much easier.
Mr. Robinson also indicated that his work on the spreadsheets had not unearthed any errors. This should make the actuaries feel more comfortable in using the spreadsheets.
Mr. Yamamoto then recognized Mr. Mussey who introduced the remaining presentations for the morning.
Mr. Mussey indicated that Mr. Wandishin and Greg Savord, a senior actuary and special assistant to the Chief Actuary, would discuss the Hospital Insurance Trust Fund assumptions.
Mr. Wandishin then discussed the annual increase in admissions per person graphs over the last several years; these graphs showed extreme variability and were difficult to understand. This was especially true for SNF and home health utilization. A second graph showed utilization for the same services with managed care utilization removed. Several panel members commented that the two graphs seemed very similar. Mr. Wandishin agreed that they were.
There were many questions about how the assumptions of utilization for persons who went into managed care were formed. Further there were questions about the possibility of autocorrelation causing spiking in the resulting series. These questions could not be fully answered, but Mr. Foster felt that they were a fruitful avenue for future research. Mr. Cutler and Mr. Foster suggested that the large spike years be investigated at the DRG level.
Mr. Wandishin also suggested that some of the changes in the time path occurred because of the aging of the population. Mr. Gutterman asked about the decreases in expenditures from 2007 to 2021 and was told that in those years the population was getting younger during those years. There was considerable discussion of this as members of the panel tried to digest what they were seeing.
Mr. Wandishin then began a discussion of the Hospital Market Basket Index and the way it is adjusted to compute the SNF and Home Health Market Baskets which are about 0.2% above the Hospital Market Basket.
In response to questions by Mr. Cutler and Mr. Chernew, Mr. Foster made the point that the Hospital Market Basket is about 60% wage increase and 40% price increases and that in the long term, the actuaries assume that hospital prices and wages grow at the same rate as other prices and wages in the economy.
Mr. Wandishin also made the point that the market baskets shown were also based on Data Resources Incorporated assumptions for SNF and Home Health because the trustees have no separate assumptions for these market baskets. This seemed to satisfy the panel and Mr. Savord proceeded to discuss the case mix estimates.
Mr. Savord began with the definition of case mix as the admission-weighted average relative weight for all admissions. The process begins by trying to ascertain the case mix time path for recent periods using the most recent hospital data. The data are relatively spotty early in the fiscal year, but as the year progresses the data tend to stabilize into a much better estimate of the increase for that year and especially for the early months of the year. Mr. Chernew brought up the question of seasonality in the monthly estimates. Mr. Savord indicated that the estimates were adjusted for seasonality. Mr. Foster made the point that while they were looking at monthly numbers, the final objective was annual increases and those became better as the year progressed.
Ms. Rosenblatt asked about changes in the weights for individual DRGs over time and was told that DRG weights are recalibrated each year based on hospital charges for the previous year. Ms. Rosenblatt asked if this recalibration could by itself result in case mix increases. Mr. Savord said that the normalizing of the index should keep the average weight the same and so keep that from happening.
The panel also discussed the fact that recently the index actually decreased for the first time ever and there was speculation about why this occurred last year. Efforts to combat fraud and abuse were mentioned as possible reasons. Mr. Savord further mentioned that administrative policy changes such as the removal of age as a complicating condition made the index move in aberrant ways and made its movement more difficult to predict.
Mr. Cutler made a point about changes in technology and changes in treatment from inpatient to outpatient settings causing movements in the average case mix. Mr. Savord agreed and said it was very difficult to separate such changes from those produced by fraud and abuse. Mr. Cutler also made the point that changes in treatment setting might make internal consistency across trust funds more difficult.
During the discussion of technology change and case mix, Mr. Nichols asked if there were experiences in other countries which might help determine the effects of technology changes on case mix. Mr. Cutler mentioned that there were things resembling DRGs in Canada and that there might be studies based on these. Mr. Nichols indicated that he was looking for a country which had not changed its payment system so often so that one could isolate the results of technology changes from those of administrative and other payment changes and also from fraud and abuse.
Mr. Chernew and Mr. Cutler then had an exchange about whether longer averages would help to get rid of the trend lines which show upcoding and then downcoding. Mr. Savord suggested that upcoding was reaching a saturation point anyway and that if HCFA(now known as CMS) continued to exercise diligence, it might stop upcoding from happening again.
Mr. Foster suggested that one might look hospital by hospital for large changes in average case mix.
Mr. Robinson asked about staff time use studies on the SNF side which might throw some light on the problem. Mr. Savord replied that the closest they could come was the RAND study from 1988. He did not think there were any similar more recent studies. Mr. Robinson suggested that a subcommittee study the problem. Several members made more comments, but no real decisions were made.
Mr. Wandishin then changed the discussion to the remaining items in the “other” category in HI services. The first item in this category is the group of hospitals exempt from PPS. While there are payment limits, they are basically paid their reasonable costs. Since 1997, however, the limit has been the 75th percentile of such costs. Exempt hospitals whose costs are below the limit get 15 percent of the difference between their actual costs and a target which is last year’s average cost updated by a market basket. These hospitals have been growing much faster than PPS hospitals until they represent up to 2% of the total increase. On the other hand, the rehabilitation hospitals who get just over half the exempt dollars are, beginning in October of this year, going to be paid on a prospective basis. Similarly, the psychiatric hospitals which make up another 45% of the exempt dollars will be paid on a prospective basis beginning October 1, 2001. Long term care hospitals, another 3% of the exempt dollars will be paid prospectively after October 1, 2002. Because the remainder is so small, the actuaries assume a zero growth in all the exempt categories.
Mr. Wandishin then discussed other HI categories: direct and indirect medical education, disproportionate share payments, nursing and allied health education, kidney acquisition, and hospital bad debt.
Mr. Savord discussed the treatment of hospital capital related expenses. They had been using the AHA panel study data for this, but the panel study was discontinued. They are paying to reinstate it, but there will be a gap in the data. The problem is exacerbated by the fact that hospital capital cost report data tend to be delayed even more than non-capital cost data.
In response to a question from Mr. Chernew, Mr. Savord gave a thumbnail sketch of how the capital PPS works including the phase-in to complete PPS.
At this point there were several comments and questions from the Panel concerning background material about what they had heard. Mr. Yamamoto called a short break.
Upon resumption, Mr. Mussey summarized what the panel was to hear for the rest of the morning. Mr. Shatto was to discuss SMI assumptions about the physician sustainable growth rate and residual factors; Mr. Warfield was to discuss the outpatient hospital and other SMI providers.
Mr. Shatto began with a review of the basic methodology for the SMI projections. They look at annual increases in charges by quarter and break out enrollment increases, price increases, and a residual. They further assume a behavioral offset as physicians try to make up for price cuts through induced extra services or higher intensity of those services. The behavioral offset is about 30% of the price cut. The chart showed that a major increase in volume-intensity occurred in 1992, the year the Medicare Physician Fee Schedule came into use. Before 1992, the volume intensity per enrollee increased by about 5.5% per year. Since 1992, it has been about 1% per year. A volume limit was also imposed when the fee schedule began to be used. Unfortunately, the volume factor was set at a fairly high level and physicians were paid a bonus if they did not exceed the volume factor. Thus there were large bonuses paid in the years 1992 and 1993.
The panel then asked about other specific factors which might be isolated to reduce the residual. The actuaries agreed that a study of these might be worthwhile.
The previous discussion had been mainly about establishing the baseline. Mr. Shatto then turned to the actual projections. Basically this is a 1% increase graded up to 2.5% in three years and set at that level for the remainder of the ten year period. After 10 years, it is eased into the long term assumption, which is equal to the increase in GDP.
Fortunately the residual is small relative to the sustainable growth rate system. This was put into place by BBA and adjusted by BBRA. It replaced the Medicare Volume Performance System. The new system is cumulative, keeping track of expenditures since 1997, the year of passage of the BBA. The base period value is grown by four factors, the Medicare economic index, the CPI, fee for service enrollment, and changes due to new laws and regulations. Finally, expenditures are constrained not to grow faster than real GDP, so prices are decreased to make that happen.
Questions from the panel elicited the information that for the next 75 years, prices are assumed to be falling. Mr. Shatto also said that physicians have not yet felt the effects of the system and will not until next year. When they do, the system may get adjusted again.
At this point, Mr. Foster suggested that the group break for lunch and it did so at 1:00 P. M.
Mr. Yamamoto convened the meeting at 1:50 and asked the actuaries to go ahead with the discussion of outpatient costs.
Mr. Warfield said that most of the information will deal with the PPS system which was going to be covered later in the afternoon. Prior to PPS, reimbursement was total hospital costs less beneficiary payments for deductibles and coinsurance. For 1999, because hospital charges were so much higher than costs, coinsurance represented 49.1% of costs rather than the 20% the law envisions. One of the objectives of PPS is to lower this to 20% over time.
Mr. Warfield then moved on to other SMI providers. There are two categories of these: those reimbursed through the carriers and those reimbursed through the fiscal intermediaries. The first category includes durable medical equipment (prosthetics and orthotics), but does not include parental and enteral nutrition. It also includes lab services, ambulatory surgical center facility costs, and a few other services. Each of these is primarily updated for price by the CPI. These are initially projected by the 3-year or 5-year average rate of increase and tapered down in the case of DME to about 4% and for labs to about 2.5%. Mr. Cutler asked questions about the length of the terms and Mr. Nichols about the size of ambulatory service center costs. Mr. Chernew asked about age adjustment in DME costs and was assured that they were not age-adjusted.
Mr. Warfield then moved on to other SMI providers, specifically, lab costs for services performed in the outpatient department of a hospital. This was increased by the factor used for other labs.
Home health and other services reimbursed by intermediaries, including renal dialysis facilities, rural health clinics and some rehab facilities are adjusted for price changes and the residual is adjusted to what seems a more reasonable level.
Mr. Gutterman asked about dialysis done in the home as it switches categories. Mr. Chernew said that in most cases home dialysis is done through some facility. Mr. Warfield then indicated that these were in general not price updated during the 1990's or were updated by a fee schedule.
Mr. Chernew asked about age and mortality adjustment and was assured that data would not allow such adjustment except for aggregate enrollment adjustments.
Mr. Cutler and Mr. Chernew gave a presentation on lessons from economic literature on health care cost growth. First they made the point that within each age cohort, people are less likely to be disabled, i. e., they are healthier and that the methodology used by the actuaries should have a health term in it. This would require that the actuaries try to forecast how technology affects health and health costs. They further reviewed several approaches to the problem appearing in the literature. While one can bicker about the numbers, the overwhelming conclusion is that technology and technological progress has been the overwhelming cause of health care cost growth. They reviewed a study by Newhouse which showed that at least 60% of the growth in health care costs was the result of technology change and that technology was driving cost growth. Technology has changed in definition over the years, but defined properly it has been the driver of cost growth. Technology tends to bring people into a treatment path who would otherwise not have been since new technologies deliver less painful treatments.
Managed care organizations try to hold down the use of technology but are unsuccessful generally.
Ms. Rosenblatt also mentioned that technology may include items other than medical technology, such as the Internet. Mr. Gutterman suggested that the aged are not prone to use the Internet, but Ms. Rosenblatt said that they have their children helping them.
Mr. Cutler also suggested that while most young people are in managed care and most aged are not, the treatment standard used for the 85% of the overall population in managed care influences treatment standards for the 75% of the aged who are not in managed care.
He went on to discuss expenditure numbers. His calculations again showed that the bulk of spending above the rate of growth of GDP is in technology.
Mr. Foster said the Office of the Actuary’s assumptions team is looking into the same issues without coming to firm conclusions. He felt that the recommendation in this area will end up being the panel’s most difficult.
Mr. Chernew then brought up one further point about the size of Medicare spending and current law. If Medicare spending gets to 30% of GDP, it is very likely that the law will be changed. Mr. Gutterman pointed out that the 1991 panel had several scenarios in which they predicted 38% of GDP.
Mr. Yamamoto pointed out that private health insurance had changed over the years with much more cost sharing and that these might substitute for changes in current law.
Mr. Foster said that if Medicare beneficiaries were naked and starving, they will encourage a change in the law. The OACT assumptions team, however, concluded that this was only likely to happen to low income beneficiaries and not to the bulk of beneficiaries in the foreseeable future.
Of more importance, Mr. Foster asked about the ramifications of saying that health is going to improve and that this will cut Medicare spending in the future when there is no evidence that spending is declining. Further if we give heart bypasses to 75 year olds and those people then have lower health costs, we may be spending a lot to get those lower costs and it may not show up.
Mr. Gutterman also pointed out that much of Ken Manton’s work was concerned with long term care and that it might not apply to the general health care. Mr. Cutler said that a decline in one Activity of Daily Living is worth about $500 in health spending and one Instrumental Activity of Daily Living about $200 and that if one accounted for the decline in ADLs and IADLs the age effect disappears.
Mr. Yamamoto turned the meeting over to Mr. Mussey again. He introduced Elliot Weinstein to discuss the Home Health PPS and Mr. Warfield to further explain Outpatient hospital PPS.
Mr. Weinstein began by indicating that the first 100 home health visits after a hospitalization or a skilled nursing facility (SNF) visit is paid by Part A. All other visits are paid by SMI and will be paid by the new PPS. He went through the types of home health services, but indicated that 90% of all visits are by home health and SNF aides. Physical therapy makes up most of the rest.
The current interim payment system is inefficient and unpopular. The new system is supposed to be budget neutral with the current system and the limits under current law are to be decreased by 15% beginning October 1, 2001 and may never come into effect. In 2002 and 2003, the PPS amounts are to be increased by the market basket less 1.1% and after 2003, by the market basket. There will be no phase-in of the systems. Basically, when the new system comes in during a patient’s treatment, the visits before the change will be paid under the old system and those after will be paid under the new.
The new system is based on a 60 day episode since 60% of episodes are less than 60 days. The payments are wage index and case mix adjusted. The case mix consists of 80 groups depending on the kind and severity of the service. Variation in case mix is quite large–the highest case mix value is about 6 times the lowest value. Outlier payments are currently set at 5% of the total. Providers get 60% of their expected payment for the first episode per patient up front and 40% when the episode is over. For succeeding episodes with the same patient, the ratio is half and half. A physician must certify necessity at the end of each episode. There are also rules when the episode is shorter than 60 days and when the patient transfers to another provider. There are also important adjustments when the patient becomes much sicker after having some number of visits and when the patient is seen only once or twice in a 60 day period.
The basic payment is the PPS amount divided by 1.05 to take the outlier payments into account.
Mr. Nichols asked about the budget neutrality adjustment and was told that it was a .84 multiplicative factor and that this should result in a payment decline for fiscal year 2001equal to the $137 million in the President’s budget. The statute does not, however, provide for changing this number in subsequent years. If there is a considerable error, the statute will have to make provision for this.
There were many questions about where the $137 million number came from and about what would happen in future years. Mr . Weinstein mentioned a possible $79 per episode increase if it proved necessary. Mr. Nichols also asked about the division of visits or dollars between Part A and Part B. Mr. Wandishin said that about two thirds of the dollars will be in Part B.
Mr. Chernew asked about the rapid increase and decline in home health benefits. Mr. Wandishin said that they projected Part A and Part B expenditures for home health together and that they assumed large increases in the first three years of PPS as more and more people used the service. After the first three years, the increases are case mix and market basket driven. Also, of course, SMI expenditures rise rapidly as an increasing portion of Part A home health expenditures are transferred to Part B.
Mr. Warfield then discussed PPS for outpatient expenditures. The basic unit for reimbursement is the 450 Ambulatory Payment Classifications or APCs. The APC is composed of the unadjusted copayment amount and the Medicare reimbursement amount. For 1999, the total is budget neutral to the amount of revenue hospitals would have received under the old system. The copayment amount is equal to 20% of the national median charges for the services in each APC. Using the median reduces beneficiary payments by about 12% in 1999. It will still, however, be about 43.9% on average. One of the objectives of PPS is to bring this down to about 20% over time.
The APCs are to be updated by the hospital market basket less 1% during the first three years and by the market basket after that.
The coinsurance amount will be the larger of 20% of the 1999 rate or 20% of the current rate. If the rate is higher than that, it will remain constant until it becomes 20%. This is expected to occur in around 2020 for many services. There have been proposals to make the decline faster.
Mr. Warfield then discussed the modeling of this complicated structure. For the base year, they come up with an amount which is about 93% of total costs or about the amount of the total less hospital laboratory services. They further estimated an unadjusted copayment amount as about 49% of that number reduced by 12.2% leaving a 43-44% copay. Medicare reimbursement is the difference between the rate for the APC and the unadjusted copayment amount.
For future years, the total is increased by some multiple of the hospital market basket which turns out to be about what the increase in the per capita costs has been in recent years or perhaps a little less. The residual starts at 1% and increases over time to about 3%. They also included a case mix adjustment of about 2.5% for the first six years, 1.5% for some number of years after six, and 0.5% after that.
Mr. Chernew and Mr. Cutler asked several questions about the relationship between outpatient and home health. The actuaries did not have data available to answer those questions.
There was considerable discussion of the relationship between Part A and Part B, between managed care and total expenditures, and between home health and outpatient care.
After a brief recess, Mr. Clemens began a presentation on stochastic modeling. Stochastic modeling began from a notion that it would be good to have a risk measurement tool to help in setting the premium. He assumed a normal distribution of growth rates at any given time around the actual incurred rates over the years. The variance of the normal distributions was developed by examining the variance of growth rates over time. There were 45 of these variances: eight service types plus managed care for each of 5 enrollment types. These were also summed to get a total variance. For some of these covariances were also calculated. The home health variance was the same for all enrollment categories. Further, the extreme variance caused by the SGR calculation for physicians was reduced by a special adjustment. The summations were weighted by the proportion each service represented of the total. Finally managed care was included with a weight of 1 and an assumption of perfect positive correlation except for the period in which the rates were set. The means of these distributions were the intermediate projections. Mr. Clemens, in response to several questions, also indicated that each year is done with separate draws independent of each other. Mr. Foster indicated that the question about independent draws was hinting at the fact that it might be better to do this year’s draws conditioned on last year’s draws, but they haven’t figured out how to do that.
There were other questions about whether the means used were appropriate. As Mr. Gutterman said, the method captured the statistical or process variance, but not the parameter variance.
Many of the questions raised about the method had to do with long term projections. Mr. Clemens and Mr. Foster assured the panel that the method was used to estimate 10 year premiums and not for years after that. Further, they had compared the model results with a more traditional ARIMA (Auto-Regressive Integrated Moving Average) model in the aggregate. ARIMA models analyze the residuals of time series regressions. The results were quite similar.
Mr. Nichols, Mr. Gutterman, Mr. Robinson, and Mr. Chernew discussed the model and felt that the model might have a problem in that in the early years, the variance is low and in the out years too high because of compounding.
Mr. Nichols then suggested that it might have been more helpful to construct a longer term model on the status of the trust fund in spite of the problems.
Mr. Clemens moved on to the income model for SMI–general revenues, premiums, and interest from the trust funds. Expected benefits are taken to be the mean of the normal distribution growth rate times last year’s benefits. Having determined benefits, the model calculates the premium income after putting in a contingency margin to keep the premium from varying too much from year to year and a condition that premiums cannot be less than the previous year. Interest is calculated as the rate times the size of the trust fund. General revenues are determined as a residual.
The Panel then discussed the concept of uncertainty. Mr. Gutterman suggested that if one were making the decision to forecast for today in 1925, conditions are so different that the question of uncertainty makes little sense.
Mr. Cutler asked about the probability of the HI trust fund being broke in 25 years. Mr. Foster said that there have been times when the probability of the fund being broke soon was very high, but he did not have an answer to the specific question.
Mr. Gutterman suggested that one of the contributions the Panel might make is to suggest potential applications for stochastic methods.
Mr. Meyerson of the Congressional Budget Office suggested also that there might be a rule of thumb concerning an upper limit on the percent of income any person might spend on health care.
At 5.50 P. M., Mr. Yamamoto adjourned the meeting until the morning.
July 27
The meeting was called to order at 9:15 A. M. by Mr. Winter, who welcomed the group and turned the meeting over to Mr. Mussey who summarized the planned activities for the day.
Mr. Shatto talked about the alternative assumptions to the intermediate assumptions and about the measures of actuarial soundness. The alternative assumptions are not meant to be a confidence interval in a statistical sense. Rather they are set to take into account a reasonable range of possibilities.
In general, these assumptions for HI vary during the first 25 years by about 2% above and below the intermediate assumptions, grade down to zero over the next 25 years, and remain at zero during the last 25 years. For SMI, this affects general revenues, but does not affect the balance in the trust fund.
Mr. Nichols asked about the history of these alternative assumptions. Mr. Mussey said that when the program began, the actuaries were only doing 25 year projections. Not until the mid-1980's did the actuaries begin doing 75 year projections. Mr. Gutterman said it was actually in response to a recommendation of the 1991 panel. Mr. Mussey replied that the actuaries were doing 75 year projections for HI before 1991, but that the 1991 panel had recommended that they keep doing 75 year projections for HI and begin doing them for SMI.
Mr. Cutler asked what these two alternatives were supposed to represent and was told that they were basically to emphasize that there was considerable uncertainty in the estimates and projections and to keep people from focusing completely on the intermediate estimates.
Mr. Gutterman asked if there had been many challenges to the methodology. Mr. Foster said that there had not been, but that the 1991 panel had suggested that this panel focus more on the methodology.
Mr. Mussey reminded the group that there were at one time four alternatives. A worst case scenario was included, but it turned out to be within the range of the other three. The worst case scenario varied only the economic assumptions, which do not affect HI and SMI as much as they do the OASDI.
Mr. Chernew asked about the case mix assumption. Mr. Foster said that one could get an idea about that from the sensitivity analysis in the HI report which shows health care cost factors.
Mr. Gutterman asked about the mortality and other assumptions which were difficult to interpret.
Mr. Foster agreed that they were difficult to interpret, but felt that the fact that the stochastic and the traditional alternatives are fairly similar, while coincidental and arbitrary, helps. If they had gone for longer than 10 years, however, the stochastic interval would have been smaller than the traditional.
Mr. Yamamoto asked about whether the plus or minus 2% could be changed easily. Mr. Foster said that if the trustees adopted it, the change could be made.
Mr. Robinson asked if significant decisions had ever been made on the basis of the high and low cost alternatives. Mr. Foster could not think of any, but when the 1971 Advisory Council wanted to do away with the alternatives and the actuaries did, the benefit provisions in the 1972 amendments to the Social Security Act were stable only under the intermediate projections and had the alternative projections been available, this would have been obvious. Because the alternatives were not available, the amendments were passed. Thus doing away with the alternatives had contributed to a fiasco and the actuaries began computing them again.
Mr. Nichols asked if these alternatives had been used to influence the public debate. Mr. Foster assured him that they did not try to do that, but if there was a proposal which the trustees felt was bad for the system, they have been willing to say that. They have not, however, tried to influence policy directly or to present a preferred future to the public. This is one of the main reasons the Trustees Report is very careful to talk only about present law. Further, the technical panels help to keep the assumptions reasonable and not a set picked for some perhaps nefarious purpose. The Public Trustees also help assure that the assumptions are reasonable.
Mr. Chernew talked about the policy implications of the reports which are far-reaching and large. He asked how concerned the actuaries are about consistency over time and incremental changes over time which change the results. Mr. Foster said that it was definitely a concern and the changes since the BBA are a good example. If the results wobble around a lot, the reports may lose credibility. On the other hand, if the situation in the economy and the trust fund changes as it did after the passage of the BBA, it is important to take those changes into account rather quickly, at least for the short run. The long run projections may not require such fast reactions.
Mr. Chernew also asked about changes in the assumptions which might make a big difference in, say, the year of exhaustion. Would the actuaries try to grade those assumptions in over time? Mr. Foster said that they put 80% of the weight in trying to have the best assumptions possible and 20% into the concern for the abruptness of the changes.
Mr. Cutler pointed out that there was a positive reason for making changes slowly because that gives more time to see how new information is panning out. Mr. Robinson then asked about a reconciliation of projections from one year to the next. Mr. Foster pointed out the President’s budget projections do go category by category to indicate why it has changed. This is not as elaborate as what people like Ms. Rosenblatt go through, but it is a step in that direction.
Ms. Rosenblatt indicated that her group has a problem with such reconciliations because they also have to react quickly to changes.
This led to a discussion of presentation issues which may be the problem. Mr. Foster indicated that they had been working on these for some years and will continue to do so.
Mr. Gutterman asked about a table in the SMI report similar to the sensitivity table in the HI report. Mr. Foster said they were continuing to work on that. Several members suggested that the actuaries think about a time series table in the report.
Mr. Cutler asked about the effect of changes in the base estimates because of a good year. Mr. Chernew asked about whether growth rates are affected by changes in the base estimates. Mr. Foster said that there is some effort made to recognize where one is in the business cycle. What is done near the top of the cycle is different from what is done nearer the bottom of the cycle. Right now we are above projected GDP and an effort is made to gradually reduce that to the projection level.
Similarly, the home health reduction in 1998 and 1999 is assumed to catch up rapidly to its old level. There may be other examples. Basically, Mr. Chernew suggested that this means that they weight recent experience at a lower level to moderate the amount of the improved base they put into the trend. Mr. Foster agreed with this, but suggested that as additional years of data become available, the actuaries are able to adjust the assumptions so they are closer to reality.
Mr. Chernew further argued that it seemed an effort to keep the size of the increases at what appears to be a relatively small level. Mr. Foster agreed with this also, but gave an example of the case mix in HI in which it had fallen about half a percent a year for the last three years as hospitals cleaned up their upcoding practices. It does not really help, however, when one tries to determine what will happen in the future. They may stay at their current level, or they may revert to their not so clean practices. In the first case one would get a catching-up effect; in the second case one would get normal growth.
Mr. Shatto then talked about measures of actuarial soundness. For SMI it is pretty easy since the general revenue component is adjusted each year. For HI, however, it is more difficult. There are two separate tests for HI, a short run test and a long run test.
The short range test is met if the trust fund begins at 100 percent of expected benefits and stays at or above that level for the next ten years. If the trust fund is below that level at the beginning, grows so that it is at or above that level by the fifth year, and stays at or above that level for the remaining years, the short run test is also met.
The long range test is actually 66 different tests, all of which must be met to pass the test. In the first 10 years of the period, the fund must be in actuarial bounds. In the entire 75 year period, the cost rate must be within 5% of the bounds in each of the succeeding 66 years to meet the test. The HI program has not come close to meeting this test in recent memory. The panel asked some questions about these tests and Mr. Chernew suggested that the short run test was more important than the long run test because of its immediacy.
Mr. Yamamoto asked what the trustees could do if the tests were not met. Mr. Foster said that they have in the past called for action sooner rather than later.
Mr. Cutler suggested that the trust funds’ recent longer projected length of life makes the long run test more relevant.
Mr. Mussey then went through the 1991 panel recommendations. Ignoring the two which were already met or which were beyond the scope of the trustee’s report, there were 21. Most of these have been implemented.
Mr. Chernew asked whether most or all the responses to the recommendations had been data oriented instead of research oriented. Mr. Foster talked about the Medicare Current Beneficiary Surveys which the Office of the Actuary began and got the funding for, but which after the HCFA(now known as CMS) reorganization ended up somewhere else. Nevertheless, the MCBS is a wonderful data source which the Office would have a hard time doing without. He also mentioned the Shatto, et al., study on physician behavior offset and the Mark Freeland/Al Peters study of the impact of insurance on health care costs. In addition, they encourage HCFA(now known as CMS)’s Office of Strategic Planning to do some research helpful to the office when they can.
Mr. Chernew asked about Office-sponsored academic research. He and Mr. Nichols felt that there should be more funding of such research. Mr. Cutler felt that there should be ongoing research to monitor the many changes in payment methods. Mr. Foster indicated that there was a lot of such research going on throughout HCFA(now known as CMS) and that some of this was on topics which take several years to bring to fruition–Outpatient PPS is an example. Mr. Gutterman mentioned that there should be an ongoing project on sustainable growth for physicians. Mr. Nichols asked about whether some of these projects were being done by the Social Security Administration for OASDI. Mr. Foster said he was not sure.
Mr. Chernew and Mr. Nichols said that there was a danger that academics would take the research in the wrong direction. To avoid this, someone might be needed to help the office write the requests for proposals to assure that academic research projects provide information on desired topics. Mr. Foster felt that this whole area might be a reasonable subject for panel recommendations.
Ms. Rosenblatt also suggested that some research on turning points in the business cycle might be helpful, but was not sure.
Mr. Foster said that it was easier to forecast the future than to understand the past and that such research might be very helpful to the office.
Mr. Chernew suggested that a model based on estimating cost and utilization levels rather than growth rates might be helpful. Growth rates might then be backed out from the level estimates.
At this point Mr. Yamamoto called for a five minute break.
After the break, Mr. Yamamoto asked for ideas about how the group was going to wrote the report. After considerable discussion, the group decided on five sub-groupings to cover items listed on the blackboard. The five groups and their members are (Chair is listed first):
1. Medicare Assumptions Cutler and Rosenblatt
2. Projection Methodology Robinson and Chernew
3. Long Range Growth Assumptions Chernew, Cutler, Gutterman
4. Stochastic Forecasting and Uncertainty Gutterman, Robinson, Nichols
5. Presentation Issues, Research and Data Nichols and Rosenblatt
After the groups had been arranged, Mr. Winter asked Mr. Nichols to try to set up a presentation by CBO on their forecasting model and Mr. Foster to try to set up a presentation on the Maryland model. Both agreed.
Mr. Gutterman asked for a presentation on incurred versus cash estimates.
Mr. Winter asked the group to review and comment on the draft minutes from the first meeting. He further asked the group to consider dates for an October meeting.
October 11 and 12 were chosen if two days are needed.
The meeting was adjourned at 12:05 P. M.