Medicare+Choice: Payments Exceed Cost of Fee-for-Service	 
Benefits, Adding Billions to Spending (23-AUG-00, HEHS-00-161).  
                                                                 
Pursuant to a congressional request, GAO reviewed Medicare Choice
program payment issues, focusing on: (1) whether program spending
for Medicare Choice plan enrollees has exceeded what		 
Medicare-covered care for these beneficiaries would have cost in 
the fee-for-service (FFS) Medicare program; and (2) the extent to
which payments to individual plans differ from expected FFS	 
costs.								 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   HEHS-00-161					        
    ACCNO:   164042						        
  TITLE:     Medicare+Choice: Payments Exceed Cost of Fee-for-Service 
Benefits, Adding Billions to Spending				 
     DATE:   08/23/2000 
  SUBJECT:   Cost analysis					 
	     Health care costs					 
	     Health care programs				 
	     Health insurance					 
	     Health insurance cost control			 
	     Managed health care				 
	     Medical services rates				 
	     Payments						 
	     Medicare Choice Program				 
	     Medicare Fee-for-Service Program			 

                                                                 
Medicare+Choice: Payments Exceed Cost of Fee-for-Service Benefits, Adding Billions to Spending
(08-Aug-00, HEHS-00-161).                                         
                                                                 
Medicare+Choice has not yielded savings for Medicare because its plans attract a 
disproportionate selection of healthier and less-expensive         
beneficiaries relative to traditional fee-for-service (FFS) Medicare--a
phenomenon known as favorable selection. The program spent about $3.2
billion, or 13.2 percent, more on health plan enrollees than if enrollees had
received services through traditional FFS Medicare. Although the Health Care
Financing Administration (HCFA) introduced a new methodology to adjust
payments for beneficiary health status, it may ultimately remove less than half of
the excess payments caused by favorable selection. Spending forecast errors
built into plan payment rates and provisions in the Balanced Budget Act caused
an additional $2 billion, or eight percent, in excess payments to plans. Although
all of the 210 plans in the study received excess payments, the percentage of
estimated excess payments varied substantially among plans. The largest
estimated excess payment totaled $334 million, or 40 percent more than
Medicare would have paid in an FFS plan. Nine plans received payments below
their enrollees' expected FFS costs. When excess payments due to forecast
error are included, only two of the 210 plans were paid less--$1.7 million and
$175,000--than its enrollees' expected FFS costs.
-------------------------Indexing Terms------------------------- 
RPTNO HEHS-00-161                                                 
ACCNO A01313                                                     
TITLE Medicare+Choice: Payments Exceed Cost of Fee-for-Service Benefits, Adding Billions
to Spending.
DOCDTE 08/11/2000 
SUBJECT                                         
                                  
         Cost analysis
         Health care costs
         Health care programs
         Health insurance
         Health insurance cost control
         Managed health care
         Medical services rates
         Payments
         Medicare Choice Program
         Medicare Fee-for-Service Program
 
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
HEHS-00-161

A

Report to Congressional Requesters

August 2000 MEDICARE+ CHOICE Payments Exceed Cost of Fee- for- Service
Benefits, Adding Billions to Spending

GAO/ HEHS- 00- 161

Letter 3 Appendixes Appendix I: Methodology 22

Appendix II: Comments From the Health Care Financing Administration 34

Tabl es Table 1: Self- Reported Health Status of 1996 Medicare Beneficiaries
in Health Plans Generally Better Than That of Beneficiaries

in FFS 8 Table 2: Summary of Findings on Aggregate Medicare+ Choice

Payments, 1998 11 Table 3: Number of Plans That Received Payments Below and
Above

Expected Fee- for- Service Costs, 1998 17 Table 4: Mean FFS Costs for Aged
and Disabled Medicare

Survivors and Decedents, 1997 25 Table 5: Joiner Cost Ratio by Market
Penetration and Size and Age

of Plan 27 Figures Figure 1: Excess Payments Due to Inadequate Risk
Adjustment

Varied Widely Across Medicare+ Choice Plans, 1998 16 Figure 2: Illustration
of Regression- Toward- the- Mean for Enrollees in

Hypothetical Plan Z 30

Abbreviations

AAPCC Adjusted Average per Capita Cost BBA Balanced Budget Act of 1997 BBRA
Balanced Budget Refinement Act of 1999 CMHS Continuous Medicare History
Sample CRS Congressional Research Service ESRD end- stage renal disease FFS
fee- for- service HCFA Health Care Financing Administration HMO health
maintenance organization HHS Department of Health and Human Services NPCM+
CGP national per capita Medicare+ Choice growth percentage RTM regression-
toward- the- mean

Health, Education, and Human Services Division

Lett er

B- 285401 August 23, 2000 Congressional Requesters Recently, a number of
health plans that participate in the Medicare+ Choice program have announced
that they will reduce or terminate their participation beginning January 1,
2001. 1 These announcements follow the substantial plan withdrawals that
occurred in 1999 and 2000. Over the 3 years, more than 1. 6 million
beneficiaries will have had to switch to a different plan, or the fee- for-
service program, because the plan in which they were enrolled no longer
serves their geographic area.

These withdrawals are of special concern to policymakers not only because of
the disruptions they cause for beneficiaries, but because the Medicare+
Choice program, created by the Balanced Budget Act of 1997 (BBA), was
designed to expand beneficiaries' health plan options. The act included
payment changes and other provisions to encourage the wider availability of
health maintenance organizations (HMO) and permitted other types of health
plans, such as preferred provider organizations, to participate in Medicare.

The Medicare+ Choice program was also expected to improve Medicare's
financial posture by better controlling spending growth. Before the BBA's
enactment, a number of studies- conducted by us and other government and
independent researchers- concluded that Medicare spent more on beneficiaries
enrolled in managed care plans than it would have spent if these individuals
had received care through traditional, fee- for- service (FFS) Medicare. 2
The reason was that Medicare's payment method- which provides plans a fixed,
per- enrollee payment, regardless of the enrollee's actual health care
costs- did not adequately account for the fact that health plans tended to
attract a disproportionate number of healthier- than1

In this report a plan refers to a managed care organization's Medicare
operations in a defined geographic area.

2 Our studies include Medicare HMOs: HCFA Can Promptly Eliminate Hundreds of
Millions in Excess Payments (GAO/ HEHS- 97- 16, Apr. 25, 1997). Other
studies were published by the former Physician Payment Review Commission
(now the Medicare Payment Advisory Commission), the Health Care Financing
Administration (HCFA), and the New England Journal of Medicine.

average beneficiaries with lower- than- average health care costs. 3
Accordingly, the BBA contained provisions to adjust plan payments to better
reflect the expected health care utilization of Medicare beneficiaries
enrolled in plans.

Industry representatives believe that the earlier studies' findings are no
longer valid because the mix of beneficiaries enrolled in plans has changed.
They claim that as Medicare's managed care program matured and more
beneficiaries enrolled in plans, the health status differences between FFS
beneficiaries and plan enrollees that may have previously existed largely
disappeared- particularly in areas where a relatively high percentage of
beneficiaries are enrolled in plans. Industry representatives contend that
the BBA's payment rate changes were so severe that Medicare is no longer a
sufficiently profitable line of business for many plans. They also have
stated that the payment changes have caused many plans to reduce coverage
for non- Medicare services, such as outpatient prescription drugs, that help
attract beneficiaries to plans. Consequently, industry officials argue,
Medicare must increase payments to maintain plan participation in the
Medicare+ Choice program.

To assist congressional deliberations on Medicare+ Choice payment issues,
you asked us to determine (1) whether program spending for Medicare+ Choice
plan enrollees has exceeded what Medicare- covered care for these
beneficiaries would have cost in the FFS program and (2) the extent to which
payments to individual plans differ from expected FFS costs. To conduct our
study, we analyzed FFS and Medicare+ Choice payment data maintained by the
Health Care Financing Administration (HCFA), the agency that administers
Medicare. Specifically, we analyzed the 1997 FFS claims history of nearly
800,000 beneficiaries who enrolled in a plan for the first time in 1998- the
most recent years for which complete data were available. We adjusted these
costs to reflect the fact that over time enrollees' FFS costs likely
approach the average cost of demographically similar FFS beneficiaries.
Finally, we compared these estimated costs with the actual capitation
payment the plans received, and developed plan- specific and aggregate
measures of excess payments or underpayments. Our study included 210 of the
346 Medicare+ Choice plans

3 Medicare risk contract HMOs- plans that received fixed monthly capitation
payments- accounted for about 90 percent of Medicare managed care enrollment
in 1998. Prior to BBA, Medicare managed care plans also included cost-
contract HMOs and health care prepayment plans that were reimbursed for the
costs they incurred, less the estimated actuarial value of beneficiary cost-
sharing.

that were in operation in 1998. 4 These plans enrolled 87 percent of all
beneficiaries in Medicare+ Choice plans. Our work was done from September
1999 to July 2000 in accordance with generally accepted government auditing
standards. (For more detail on our methodology, see app. I.)

Results in Brief Medicare+ Choice, like its predecessor managed care
program, has not been successful in achieving Medicare savings. Medicare+
Choice plans

attracted a disproportionate selection of healthier and less- expensive
beneficiaries relative to traditional FFS Medicare (a phenomenon known as
favorable selection), while payment rates largely continued to reflect the
expected FFS costs of beneficiaries in average health. Consequently, in 1998
we estimate that the program spent about $3.2 billion, or 13. 2 percent,
more on health plan enrollees than if they had received services through
traditional FFS Medicare. This year HCFA implemented a new methodology to
adjust payments for beneficiary health status. However, our results suggest
that this new methodology, which will be phased in over several years, may
ultimately remove less than half of the excess payments caused by favorable
selection. In addition, the combination of spending forecast errors built
into plan payment rates and BBA payment provisions caused an additional $2.0
billion, or 8 percent, in excess payments to plans. Instead of paying less
for health plan enrollees, we estimate that aggregate payments to Medicare+
Choice plans in 1998 were about $5. 2 billion (21 percent), or approximately
$1, 000 per enrollee, more than if the plans' enrollees had received care in
the traditional FFS program. It is largely these excess payments, and not
managed care efficiencies, that enable plans to attract beneficiaries by
offering a benefit package that is more comprehensive than the one available
to FFS beneficiaries, while charging modest or no premiums.

Nearly all of the 210 plans in our study received payments in 1998 that
exceeded expected FFS costs because their enrollees were healthier than
average beneficiaries. However, the percentage of estimated excess payments
varied substantially among plans. About two- thirds of the plans received
payments that were at least 10 percent more than enrollees would have cost
Medicare in the traditional program, even without considering excess
payments due to forecast errors. The largest estimated excess

4 We excluded 136 plans because they had too few new enrollees to produce
statistically reliable results.

payment to an individual plan totaled $334 million, or 40 percent more than
Medicare would have spent if the plan's enrollees had been covered under
FFS. We also estimated that nine plans received payments below their
enrollees' expected FFS costs. However, when excess payments due to forecast
error are included, only 2 of the 210 plans were paid less ($ 1. 7 million
and $175,000) than its enrollees' expected FFS costs.

Background Inherent in Medicare's FFS program is an incentive for providers
to deliver more services than necessary because each additional service
generates

additional provider revenue. Policymakers have therefore looked to managed
care plans to curb unnecessary spending. Plans have a financial incentive to
provide care efficiently because they receive a fixed monthly amount for
each beneficiary, regardless of what the individual enrollee's care actually
costs. These potential efficiencies cannot result in actual savings for
Medicare if plan payments are higher than beneficiaries' expected FFS costs.
Before 1998, plan payment rates were based on average local FFS spending.
Although payments were adjusted for certain beneficiary characteristics such
as age and sex, this adjustment did not adequately account for differences
in enrollees' health and expected health care costs. The BBA changed how
plan payments were calculated beginning in 1998 and, beginning in 2000,
required adjustments to make payment rates better reflect differences in
beneficiary health status.

Medicare+ Choice Payment Before 1998, Medicare's HMO capitation rates,
established separately for

Method Built on Experience each county, were set at 95 percent of the
estimated average cost of care

With Risk Contract HMO for the FFS population. The rates were discounted 5
percent under the

Payments assumption that the managed care plans provided care more
efficiently

than the less- restrictive FFS program and that the government should share
in the savings. Under this methodology, county rates varied widely because
they reflected the substantial differences in FFS per- beneficiary spending
among counties. In 1997, for example, county rates ranged from a low of $221
per month in Arthur County, Nebraska, to a high of $767 in Richmond County
(Staten Island), New York. The rates paid to plans for enrollees were also
adjusted up or down in an attempt to account for estimated cost differences
among beneficiaries due to variations in health care needs- a refinement
known as risk adjustment.

Risk adjustment is necessary because individuals differ in the extent to
which they use various health care services and their health care costs vary
accordingly. Medicare's risk adjuster has relied on beneficiary age, sex,
and

other demographic factors to predict expected health care utilization. This
demographic- based risk adjuster could reflect only gross differences- such
as the lower expected health care costs of 65- year- olds relative to
80year- olds. If two individuals shared the same demographic
characteristics, Medicare paid plans the same amount for both of them, even
if one was in poor health and required extensive medical care while the
other was in excellent health and rarely needed to see a physician.

Consequently, if a plan attracted a disproportionate share of beneficiaries
in better- than- average health, an outcome known as favorable selection,
Medicare's payments to the plan would exceed the expected FFS cost of
providing Medicare- covered benefits to the plan's enrollees. 5 Our 1997
study of Medicare payments to California HMOs found that, on average,
payments exceeded the expected FFS costs of the plans' enrollees by
approximately 19 percent in 1995. 6 Other studies have also found
substantial excess plan payments. 7

A 1996 survey of Medicare beneficiaries also indicated that plans likely
experienced favorable selection (see table 1). Medicare beneficiaries
enrolled in plans were more likely than beneficiaries in the FFS program to
report that they were in good or excellent health. Moreover, these health
status differences translate into cost differences. For example, in 1996,
average per- person Medicare FFS spending for beneficiaries who reported
they were in excellent health was approximately $2,100, whereas for
beneficiaries reporting poor health the average was about $11,700. 8

5 Whether plans intentionally attract a healthier, or more favorable,
selection of Medicare beneficiaries is controversial. Favorable selection
may, in part, be the natural consequence of limits on provider choice.
Because enrolling in a particular plan may mean having to switch providers,
sicker individuals- who are more likely to have longstanding provider
relationships- may prefer to remain in FFS, where unrestricted provider
choice does not disrupt existing patient- provider relationships.

6 As reported in the 1997 study, this difference represented 16 percent of
total plan capitation payments. 7 See Physician Payment Review Commission,
“Risk Selection and Risk Adjustment in Medicare,” Annual Report
to Congress, ch. 15 (Washington, D. C.: Physician Payment Review Commission,
1996) for a summary of studies on favorable selection.

8 1996 Medicare Current Beneficiary Survey.

Table 1: Self- Reported Health Status of 1996 Medicare Beneficiaries in
Health Plans Generally Better Than That of Beneficiaries in FFS

Numbers in percent

Beneficiaries in health Beneficiaries in Self- reported health status plans
FFS

Excellent or good 81 70 Fair or poor 19 30 Limits on three or more daily
living

4.9 11. 7 activities a a Such as bathing, dressing, or eating.

Source: HCFA, “Medicare Current Beneficiary Survey, 1996,” A
Profile of Medicare: Chartbook 1998 (May 1998).

The BBA Changed Payment In establishing the Medicare+ Choice program, the
BBA also changed the

Methodology and Required methodology used to set plan payments. Although
1997 county rates

Improved Risk Adjustment formed the foundation for county rates in 1998 and
subsequent years,

annual rate increases were no longer tied to changes in local FFS spending.
Specifically, each county's rate was to be set at the highest of

a minimum amount, or “floor,” set at $367 in 1998 and increased
annually; an amount 2 percent higher than the previous year's county rate;
or an amount reflecting a blend or weighted average of the county rate and

a price- adjusted national rate. The floor was meant to encourage plans to
offer services in areas that historically had low payment rates and few if
any participating plans- primarily rural counties. The blended rate- which
is to be phased- in over 6 years- was designed to reduce the geographic
variation in rates that had resulted from tying county rates to local FFS.
The blending mechanism moves all county rates closer to the national average
by assigning aboveaverage payment rate increases to low- payment- rate
areas, and belowaverage payment rate increases to high- payment- rate areas.
9

9 For example, in 2000 the average county rate increased by about 5 percent.
However, some low- payment- rate counties experienced rate increases over 13
percent, while high- paymentrate counties were assigned the minimum 2-
percent increase.

The BBA also sought to reduce the excess in Medicare's managed care payments
by holding down per capita rate increases for 5 years. Specifically, the BBA
set the annual factor used to update floor and blended payment rates equal
to the projected national growth in per capita Medicare spending minus a
specified percent: 0.8 percent in 1998 and 0. 5 percent in each of the
following 4 years. 10

In 1998, 1999, and 2001 no county received a blended rate. 11 Instead,
counties were either assigned the floor payment rate or the 2- percent
minimum increase. This occurred because of a budget neutrality provision in
the BBA. The law specified that estimated total Medicare+ Choice payments
should equal the amount that would have been spent if 1997 county rates had
been trended forward by the national annual update factor. However, national
per capita spending grew relatively slowly and the guaranteed 2- percent
increase and floor payments pushed total estimated Medicare+ Choice spending
above the budget neutrality amount in these 3 years. The blending provision
could not be implemented because doing so would have further increased
spending. 12 If low- payment- rate counties received higher than average
payment rate increases, the additional spending could not be offset by
lowering the increases in the high- payment- rate counties because those
counties were guaranteed a 2percent minimum increase.

The BBA required HCFA to develop and implement a health- based risk adjuster
by January 1, 2000. 13 The law authorized HCFA initially to collect hospital
inpatient data from plans and later to collect more comprehensive

10 The annual update factor is known as the national per capita Medicare+
Choice growth percentage (NPCM+ CGP). The Balanced Budget Refinement Act of
1999 (BBRA) increased the update factor in 2001 to the national growth in
per capita Medicare spending minus 0. 3 percent.

11 In 2000, about 60 percent of counties received the blended rate. The
remaining counties received the minimum 2- percent increase or the floor
amount. 12 In 1998 and 1999 the guaranteed 2- percent update and floor
payment rate increased expected spending slightly above the budget
neutrality amount. In 2000, spending under the BBA's county rates was
projected to be budget neutral and the blending provision was implemented.
However, the Congressional Research Service (CRS) reports that in 2001 the
combination of the guaranteed 2- percent update and floor payment rate will
result in aggregate plan payments that exceed the budget neutrality amount
by about $1 billion.

13 Technically, the law directs its requirements at the Secretary of the
Department of Health and Human Services (HHS), but as a practical matter,
compliance is carried out by HCFA, an HHS agency.

medical encounter data to use in developing the risk adjuster. HCFA
announced a phase- in schedule that called for gradually applying an interim
risk adjuster, based on hospital admissions data only, first to a fraction
(10 percent) of a plan's Medicare payments and then to greater shares each
year.

HCFA proposed to apply an improved risk adjuster, based on a more
comprehensive set of medical services, to 100 percent of payments by 2004.
The BBRA slowed the proposed phase- in schedule of the interim risk adjuster
and called for additional studies by HCFA and the Medicare Payment Advisory
Commission on risk adjustment implementation issues. 14

1998 Medicare+ Choice Our analysis shows that Medicare+ Choice payments in
1998 were 21

Payments Exceeded percent, or $5.2 billion, higher than the amount Medicare
would have spent

if plan beneficiaries had received care in the traditional FFS program.
Estimated FFS Costs

While Medicare payments to plans on behalf of these enrollees totaled $29.8
billion, we estimate that total FFS payments for the same population would
have been $24.6 billion. Approximately $3. 2 billion of the $5.2 billion
difference was the result of inadequate risk adjustment. That is, Medicare
payments were not adjusted sufficiently to account for the generally better
health, and lower expected costs, of plan enrollees. 15 Excess payments due
to inadequate risk adjustment will persist as long as there is favorable
selection and plan payment rates do not sufficiently account for beneficiary
health status. The remaining $2 billion resulted from errors in the Medicare
spending forecasts used to establish the 1998 county payment rates. These
excess payments will diminish over time as future county rates are adjusted
to correct errors in prior spending forecasts. Table 2 summarizes our
findings on aggregate excess payments.

14 P. L. 106- 113, app. F. 15 In establishing minimum, or floor, county
payment rates, the BBA intentionally set payment rates above average FFS
spending in certain counties. If we had excluded the amount of plan payments
above FFS costs due to the floor rates, our estimate of excess payments
would have been approximately two- tenths of 1 percent lower.

Table 2: Summary of Findings on Aggregate Medicare+ Choice Payments, 1998
Measures Results

Number of Medicare+ Choice plans in study 210 Share of all Medicare+ Choice
enrollees 87 percent Aggregate payments to plans in study $29. 8 billion
Estimate of enrollees' expected costs in FFS $24. 6 billion Aggregate excess
payment amount $5. 2 billion Excess due to inadequate risk adjustment $3. 2
billion Excess due to forecast errors in payment rates $2. 0 billion Source:
GAO analysis of HCFA data.

Inadequate Risk Adjustment As a group, beneficiaries who join plans tend to
be healthier and use fewer

for Healthier- Than- Average Medicare services than beneficiaries who remain
in FFS. For example, in

Enrollees Resulted in 1997 Medicare spent about 30 percent less on those
beneficiaries who

Excess Plan Payments joined a plan in 1998 than it spent on demographically
similar beneficiaries

who remained in FFS. Our analysis suggests beneficiaries who have been plan
members for several years continue to use fewer health care services than
FFS beneficiaries, although the gap is much narrower.

Medicare's risk adjustment methodology, unchanged in 1998 from previous
years, did not adequately address differences in beneficiary health status.
While health plans tended to enroll less- expensive beneficiaries,
Medicare's payments were too generous because they were based on the
expected costs of enrollees in average health. 16 Consequently, we estimate
that in 1998 Medicare paid plans an average of 13. 2 percent more than it
would have spent if the plans' enrollees had received care under the
traditional FFS arrangement. In the aggregate, the lack of an adequate risk
adjustment

16 A favorable selection of beneficiaries does not mean that plans attracted
only the healthiest individuals, but rather that they served a smaller
proportion of the costliest Medicare beneficiaries. HCFA reports that 10
percent of Medicare beneficiaries account for 63 percent of Medicare
expenditures.

methodology in the face of favorable selection increased Medicare spending
by approximately $3.2 billion. 17

Our results suggest that HCFA's new health- based interim risk adjuster,
when fully implemented, may only eliminate half of the excess payments due
to favorable selection. The first year of plan- submitted data on
beneficiaries' service use indicates that full implementation of the interim
risk adjuster would have reduced plan payments by about 5. 9 percent in
2000. 18 Its failure to eliminate more of the excess payments may be due to
its reliance on limited data. The interim risk adjuster uses only certain
hospital inpatient data to measure beneficiary health status because those
are the only service- level data plans provide to HCFA. Although it
represents a significant improvement over the previous methodology based
strictly on demographic data, a risk adjuster that used more complete
beneficiary data could set payments that better reflect beneficiaries'
expected health care costs. In 2004, HCFA intends to implement a refined
risk adjuster that would use medical data from physicians' offices and
hospital outpatient departments, in addition to hospital inpatient data.

Errors in FFS Spending Payment rates for Medicare+ Choice plans are set
partly on the basis of

Forecasts and BBA Payment forecasted growth in national per capita spending
in Medicare's FFS

Provisions Add to Excess program. Consequently, inaccurate forecasts can
result in plan payment

Plan Payments rates that are overstated or understated compared to actual
FFS spending

per beneficiary in a particular year. The BBA's guaranteed 2- percent
minimum annual increase and other payment provisions can also cause payment
rates to deviate from expected per capita FFS spending. Our analysis
indicates that the combination of a spending forecast error and the BBA
payment provisions resulted in aggregate plan payments that exceeded
enrollees' estimated FFS costs by approximately 8 percent, or $2 billion, in
1998.

17 In conducting our analysis, we made conservative methodological choices
that tended to reduce our estimates of excess payments. For example, we
compared 1998 aggregate plan payments with full estimated FFS costs, even
though, historically, plan payments were based on only 95 percentofthese
costs.

18 To minimize disruptions for plans and beneficiaries, HCFA phased in the
new risk adjuster and computed 90 percent of each plan's payments using the
old methodology and 10 percent using the new methodology. Thus, on average,
actual plan payments were reduced by less than 1 percent relative to what
they would have otherwise been in 2000.

HCFA estimates that the 1997 county rates were set too high and overstated
average FFS spending by about 3 percent. Prior to the BBA, this error would
have had a limited effect on future county rates because rates in each
subsequent year were based on spending forecasts that were corrected for
previous errors. 19 In establishing the Medicare+ Choice program, however,
the BBA specified that 1997 county rates be used as the basis for all future
county rates and did not allow an adjustment for prior forecasting errors.

Under the prior methodology, HCFA would have increased average 1998 rates by
approximately 1 percent. Instead, nearly all Medicare+ Choice plans received
the BBA's 2- percent minimum payment rate increase in 1998. Plans that
served beneficiaries living in counties subject to the floor rate received
higher rate increases. In contrast to the payment rate increases received by
plans, per capita FFS spending on the aged actually declined by about 3
percent between 1997 and 1998. Thus, even without considering the effects of
favorable selection, the combination of the forecast error in the 1997 rates
and the minimum annual rate increase resulted in $2 billion in excess
payments to Medicare+ Choice plans in 1998.

Although it may take more than 5 years, the effect of the 1997 forecast
error will be largely mitigated by the BBA provision that slows Medicare+
Choice rate increases relative to the growth in FFS spending between 1998
and 2002. Furthermore, the BBA does allow HCFA to correct post- 1997
forecast errors when setting 1999 and later county rates. However, the
agency's actions are constrained by the law's minimum payment provisions.
For example, 2001 payment rates would be lowered by more than 1 percent if
HCFA fully corrected the 1998, 1999, and 2000 FFS spending estimates and
applied the reduced update factor. However, to fulfill the BBA's minimum

19 One might expect forecast errors to be random. FFS spending would be
overestimated and rates would be set too high in some years, but this would
be balanced by underestimates in other years when rates would be set too
low. In practice, however, HCFA has nearly always overestimated FFS spending
on aged beneficiaries. Between 1985 and 2000, the agency overestimated
spending in 13 years and underestimated spending in 3 years. In the 5 years
before the implementation of the BBA, HCFA's average forecast error was 3. 4
percent. As a result, instead of setting rates at 95 percent of FFS
spending, HCFA effectively set rates at 98. 2 percent (0.95 x 1. 034) of FFS
spending.

payment rate requirements all county rates will have to be increased by at
least 2 percent. 20

Excess Payments Helped Our current findings suggest that many of the
additional benefits enjoyed

Finance Additional Benefits by plan enrollees may have been the result of
Medicare's overly generous

Provided to Enrollees payment rates, not of efficiencies achieved under
managed care.

Medicare+ Choice plans must, at a minimum, provide all of the services
available to beneficiaries under the traditional FFS program. However, plans
have typically provided substantial additional benefits, such as reduced
beneficiary cost sharing, coverage for outpatient prescription drugs, dental
care, and routine vision care, while charging modest or no premiums. Plans
offered some of these additional benefits to fulfill Medicare requirements.
If a plan projects that its Medicare revenues will be higher than the
projected cost of providing basic Medicare services, the plan must contract
to deliver additional benefits at no additional cost to beneficiaries. Many
plans, however, exceeded these minimum requirements and voluntarily provided
extra benefits in an effort to retain existing members and attract new ones.
Based on data submitted by plans, HCFA estimates that Medicare+ Choice plans
are using 22 percent of their Medicare revenues to provide additional
benefits in 2000. This figure is consistent with our previously reported
analysis of plans' 1997 data. 21

Effect of Inadequate The effect of Medicare's inadequate risk adjustment
methodology varied

Risk Adjustment Varied among plans because of differences in the health
status of their enrollees.

The vast majority of plans in our study (201 out of 210) served
beneficiaries Substantially Among

who tended to be healthier than FFS beneficiaries and thus received
Individual Plans

payments that exceeded their enrollees' expected FFS costs. Nine plans,
however, served beneficiaries who were generally less healthy than FFS
beneficiaries. Medicare paid these plans less than it would have spent to
serve these plans' members in the FFS program. 22 Nevertheless, the
widespread finding of excess payments counters claims that favorable

20 As a result of a technicality in the rate setting process, payment rates
will be increased by about 3. 3 percent in counties that receive the floor
rate. 21 Medicare+ Choice: Reforms Have Reduced, but Likely Not Eliminated
Excess Plan Payments( GAO/ HEHS- 99- 144, June 18, 1999).

22 Excludes excess payments due to forecast errors.

selection is not a problem in markets with substantial Medicare+ Choice
enrollment.

A few plans had an excess payment rate above 30 percent, but the median
excess payment rate was approximately 12 percent. Among the nine plans in
our study that were disadvantaged because their enrollees tended to be in
below- average health, Medicare's payments were as much as 10. 6 percent
below enrollees' expected heath care costs (see fig. 1).

Figure 1: Excess Payments Due to Inadequate Risk Adjustment Varied Widely
Across Medicare+ Choice Plans, 1998

Note: Excess payments are defined as Medicare payments to a plan minus the
expected FFS costs of a plan's enrollees. Payments exclude the excess
amounts caused by forecast errors imbedded in the 1998 county rates. The
excess payment rate is computed as the ratio of a plan's excess payments to
its enrollees' expected FFS costs.

Source: GAO analysis of HCFA data.

The amount of excess payment that plans received depends upon both the
degree to which they experience favorable selection and the number of
Medicare beneficiaries they served. In 1998, 40 plans each received more
than $20 million in excess payments due to inadequate risk adjustment. The
largest estimated excess payment to an individual plan totaled $334 million,
or 40 percent more than Medicare would have spent if the plan's enrollees

had been covered under FFS. Among the nine plans that were paid less than
their enrollees' estimated FFS costs, the largest difference between
estimated costs and payments was $8.4 million (see table 3). However, when
excess payments due to forecast error are included, only 2 plans were paid
less ($ 1.7 million and $175,000) than their enrollees' expected costs.

Table 3: Number of Plans That Received Payments Below and Above Expected
Fee- - -- for- Service Costs, 1998

Amount by which plan payments differed from expected FFS costs because of
inadequate risk adjustment Number of plans

Up to $8. 4 million below FFS 9 Up to $2 million above FFS 35 Over $2
million to $5 million above FFS 57 Over $5 million to $10 million above FFS
44 Over $10 million to $20 million above FFS 25 Over $20 million to $50
million above FFS 31 Over $50 million above FFS 9

Tot al 210

Note: Dollar amounts reflect only those excess payments or underpayments due
to inadequate risk adjustment. They do not include excess payments
attributable to forecast errors.

Source: GAO analysis of HCFA data.

While the lack of an adequate risk adjuster can lead to excess payments, it
can also result in plans being penalized if they serve a disproportionate
share of Medicare beneficiaries with expensive health care needs. This can
discourage a plan from seeking to enroll those beneficiaries, or developing
programs that would foster a reputation for effectively treating certain
costly diseases.

Some analysts have suggested that favorable selection might diminish over
time. They hypothesize that health status differences between beneficiaries
in health plans and beneficiaries in FFS could disappear as plan enrollment
grows. However, our results indicate that this is not likely. Most plans
that were well- established, enrolled a substantial number of beneficiaries,
and were located in areas with relatively high rates of Medicare managed
care enrollment experienced some level of favorable selection and therefore
received excess payments. For example, a plan in our sample had an

estimated excess payment rate of 18. 5 percent and received more than $100
million in excess payments, even though it had been serving Medicare
beneficiaries for more than a decade, had well over 200,000 enrollees in
1998, and was a dominant plan located in an area where over 33 percent of
the Medicare beneficiaries were enrolled in Medicare+ Choice plans.

Excess payments persisted, in part, because total enrollment in managed care
plans continued to grow and because many of the new enrollees were in
relatively good health. In 1998, about 15 percent of all enrollees in our
sample had been in FFS the previous year. On average, these beneficiaries
used about 30 percent less Medicare services (in dollar terms) in the year
before they switched from FFS to managed care than demographically similar
beneficiaries who remained in FFS.

The difference between the health care needs of plan beneficiaries and FFS
beneficiaries is smaller for beneficiaries who have been plan members for
several years. However, our analysis indicates that for up to 8 years, plan
beneficiaries continue to have somewhat lower expected costs than their
demographic counterparts who remained in FFS. Almost three- quarters of the
enrollees in our 1998 sample were in managed care plans for 5 years or less,
so that the effect of enrollees' better- than- average health produced
excess payments for most plans.

Conclusions In creating the Medicare managed care program that preceded
Medicare+ Choice, the Congress sought a mechanism to better control

program spending. However, evidence indicated that the payment methods used
in the earlier program resulted in Medicare paying more, not less, for
beneficiaries enrolled in plans. To address this and other problems, the BBA
substantially changed the methodology used to set plan payments, beginning
in 1998. Some industry representatives have suggested that the BBA's payment
reforms were too severe. They point to the recent plan withdrawals to
support their claims that the Medicare+ Choice program is in danger.

Our current study indicates that, in 1998, Medicare paid plans $5.2 billion,
or 21 percent, more than the estimated FFS costs of plans' Medicare
beneficiaries. The seeming paradox between our findings and the industry's
position is resolvable. Medicare+ Choice plans are being paid too much for
what was originally intended- providing beneficiaries the package of
Medicare- covered services at less cost than the traditional FFS program.
However, Medicare+ Choice plans may not be paid enough for what they

have been offering to attract beneficiaries- a more comprehensive benefit
package beyond that covered for FFS beneficiaries for only modest or no
premiums.

The problem of excess payments can be addressed in part by better adjusting
payments for the actual health status of enrollees. Such a step would also
protect those plans that attract sicker- than- average enrollees. This year,
HCFA began to implement a new risk adjuster that does a better job of
calibrating payments to reflect differences in beneficiary health status.
The agency also is developing an even more refined risk adjustment
methodology. However, because of the slow and uncertain phase- in schedules,
it may be several years before excess payments caused by health status
differences are reduced substantially.

The decision on whether to adjust the county rates so that the remaining
excess payments are eliminated or to heed plans' calls for higher Medicare+
Choice payment rates will require balancing the concerns of beneficiaries,
taxpayers, and plans. Adjusting plan payments so that the program pays no
more for a Medicare+ Choice enrollee than for a traditional Medicare
beneficiary with equivalent health status is going to mean smaller payments
and most likely lower profits for plans, fewer plans participating, and
less- generous extra benefits for enrollees. These consequences raise for
the Congress the question of whether payments should be modified to protect
plans and the fraction of the Medicare beneficiary population enrolled- even
if that protection results in Medicare spending more on the Medicare+ Choice
enrollee than for the traditional FFS Medicare beneficiary and increases the
difficulty of financially sustaining the Medicare program.

Agency Comments In commenting on our report, HCFA agreed that Medicare
spends more on Medicare+ Choice enrollees than it would have spent if those
enrollees had

received services through the traditional FFS program. The agency said that
our findings were consistent with previous studies of favorable selection,
which also concluded that Medicare managed care plans tend to attract
beneficiaries with better- than- average health status. In addition, HCFA
agreed that the BBA payment provisions and a 1997 spending forecast error
had contributed to excess plan payments.

HCFA noted that it is phasing in a risk adjustment method based on hospital
diagnosis data to help reduce excess payments caused by favorable selection.
If the method were fully implemented, plan payments

would be reduced by an average of 5.9 percent. In 2004, HCFA will begin
phasing in a risk adjustment method based on more comprehensive information
that will more accurately measure health status and help avoid excess
payments. The agency commented that our report reinforces the importance of
proceeding with the implementation of risk adjustment.

Finally, HCFA agreed with our conclusion that Medicare+ Choice payments are
too high for providing the Medicare benefit package, but may not be enough
for the expanded benefit package that plans have offered to attract
beneficiaries. According to its own analysis, approximately 76 percent of
Medicare+ Choice payments support Medicare- covered services. Plans use the
remaining 24 percent to provide extra health care benefits (9 percent) and
to reduce beneficiaries' premiums and cost sharing (15 percent). It
acknowledged that some plans may not find Medicare payments adequate to
finance Medicare- covered services and the additional benefits they have
been offering. The full text of HCFA's comments appears in app. II.

We are sending copies of this report to the Honorable Donna E. Shalala,
Secretary of Health and Human Services; the Honorable Nancy- Ann Min
DeParle, Administrator of HCFA; and other interested parties. We will also
make copies available to others upon request.

If you or your staffs have any questions about this report, please call me
on (202) 512- 7114 or Laura A. Dummit on (202) 512- 7119. Other major
contributors to this report included James C. Cosgrove, Hannah F. Fein, Jim
S. Hahn, and Richard M. Lipinski.

William J. Scanlon Director, Health Financing

and Public Health Issues

List of Requesters The Honorable Charles E. Grassley Chairman The Honorable
John B. Breaux Ranking Minority Member Special Committee on Aging United
States Senate

The Honorable William V. Roth, Jr. Chairman The Honorable Daniel Patrick
Moynihan Ranking Minority Member Committee on Finance United States Senate

The Honorable John D. Dingell Ranking Minority Member Committee on Commerce
House of Representatives

The Honorable William M. Thomas Chairman The Honorable Pete Stark Ranking
Minority Member Subcommittee on Health Committee on Ways and Means House of
Representatives

Appendi Appendi xes xI

Methodology We developed a methodology to estimate how much Medicare would
have spent on each Medicare+ Choice plan enrollee if he or she received
services under the traditional fee- for- service (FFS) arrangement in 1998.
1 The difference between the estimated FFS costs of a plan's enrollees and
Medicare's payments to that plan is a measure of excess (or under-) payment.
The aggregate excess payments we report are the sum of the excess payments
made to the 210 plans in our sample. These plans served approximately 87
percent of all beneficiaries in Medicare+ Choice plans in 1998. We excluded
plans that had relatively few new enrollees (typically, small plans), so our
aggregate excess payment amount is not an estimate of national excess
payments. 2

To conduct our analysis, we obtained HCFA data from the following sources:
(1) the Standard Analytic File, a repository of claims information; (2) the
Continuous Medicare History Sample (CMHS), a database that contains
longitudinal cost and enrollment information on a sample of Medicare
beneficiaries from 1974 to the present; (3) the Group Health Plan file, a
database that contains health plan enrollment information; and (4) the 1997
Adjusted Average per Capita Cost (AAPCC) rate book and accompanying
worksheets, which contain summary demographic and cost information on
Medicare beneficiaries in each county, and the 1998 payment rate file.

Medicare+ Choice We classified Medicare+ Choice enrollees into two groups-
survivors

Enrollees' FFS Costs (beneficiaries who did not die in 1998) and decedents
(beneficiaries who

died in 1998)- and developed slightly different methodologies to estimate
and Excess Plan

FFS costs for each group. Although relatively few beneficiaries die each
Payments

year, their health care costs before death tend to be high and may follow a
different pattern than other beneficiaries' costs. We then aggregated the
estimated FFS costs by plan and compared them to Medicare+ Choice payments
to estimate excess plan payments.

1 In addition to the amount Medicare pays for claims, our estimate of FFS
costs includes other reimbursable provider expenses and Medicare's claims
processing costs. 2 To minimize the undue influence of outlier observations
on our estimates, we excluded 136 plans that had fewer than 500 new
enrollees in 1998 from our analysis. Our estimates apply to enrollees
entitled to Medicare benefits because they are aged or disabled.
Beneficiaries with end- stage renal disease (ESRD) were excluded because
Medicare uses a different formula to set plan payments rates for these
individuals.

Methodology to Estimate To estimate the FFS costs of plan enrollees who
survived 1998, we analyzed

FFS Costs of Survivors the 1997 FFS costs of beneficiaries who switched from
FFS to a plan

between 1997 and 1998. We then extrapolated from their cost patterns to
estimate the FFS costs of all plan enrollees. To implement this approach we
had to (1) estimate a base rate that reflected 1997 per capita spending on
FFS survivors in each county, (2) calculate a joiner cost ratio for each
plan that measured 1997 FFS spending on new plan enrollees relative to
spending on demographically similar beneficiaries who remained in FFS, (3)
derive a set of regression- toward- the- mean (RTM) factors for each plan
that estimate how much FFS costs of enrollees in the plan differ from
joiner's 1997 FFS costs, and (4) develop a national update factor to convert
1997 spending levels to equivalent 1998 spending. The following describes
the four steps.

1. Estimate a base rate for each county that reflects the cost of serving
the average FFS beneficiary who did not die during 1997. (Separate base
costs were estimated for the aged and disabled Medicare populations.) We
began with the 1997 county per capita costs, which represent HCFA's
estimated cost of providing Medicare- covered services to the average FFS
beneficiary in each county. 3 However, these rates reflect the cost
experience of a county's entire FFS population and thus include the
typically high costs associated with persons who died during the year. 4 To
estimate the cost of survivors only, we used the fact that each county's per
capita costs can be expressed as a weighted average of the per capita costs
of survivors and the per capita costs of decedents. That is,

3 This figure does not include the 5- percent discount used to set county
payment rates. 4 For a discussion of death- related costs, see J. Lubitz, J.
Beebe, and C. Baker, “Longevity and Medicare Expenditures,” New
England Journal of Medicine, Vol. 332, No. 15 (1995), pp. 9991003; J. Lubitz
and R. Prihoda, “Medicare Services in the Last 2 Years of Life,”
Health Care Financing Review, Vol. 5, No. 3 (1984), pp. 117- 31; and J.
Lubitz and G. Riley, “Trends in

Medicare Payments in the Last Year of Life,” New England Journal of
Medicine, Vol. 328, No. 15 (1993), pp. 1092- 96.

We then used the CMHS to estimate a national death factor that measures the
difference in costs between persons who die and those who survive during the
calendar year. 5 Separate death factors were estimated for the aged and
disabled Medicare populations. 6 As shown in table 4, the cost of aged
decedents is approximately 4 times the cost of aged survivors while the cost
of disabled decedents is nearly 5 times the cost of disabled survivors.

5 Because our analysis of death- related costs was restricted to the period
of January 1997 to December 1997, the death factor is not equivalent to the
costs associated with death over the last year of life. That is, for persons
who died in January we observe 1 month of FFS costs, and for those who died
in December, 12 months of costs. However, higher costs for personswhodied in
1998 raised theestimated 1997costofsurvivors. Weassumed thatthe relationship
between survivor and decedent costs was the same across counties. The CMHS
did not contain a sufficient number of cases to allow reliable estimates of
county- specific death factors. Our state- level analysis of decedent and
survivor FFS costs showed that the death factor was similar among states and
that, consequently, use of a national death factor was reasonable.

6 Beneficiaries, excluding plan enrollees and persons with ESRD, were
assigned to one of five age cohorts. Mean 1997 FFS costs were then computed
separately for survivors and decedents in each age cohort. We weighted these
mean costs by the proportion of survivors and decedents in the FFS
population within each age cohort and calculated an aggregate ratio of
decedent- to- survivor costs. Population weights were determined from the
1997 Denominator File, a HCFA database that contains enrollment information
on all Medicare beneficiaries.

Table 4: Mean FFS Costs for Aged and Disabled Medicare Survivors and
Decedents, 1997

Death factor Mean survivor

Mean decedent (survivor costs/

Medicare status costs costs decedent costs)

Aged 4,606 18,973 4.12 Disabled 5,291 25,673 4.85 Note: To control for age
differences in the CMHS and the FFS populations, group means are weighted by
the proportion of the FFS survivors and decedents in each age cohort.

Source: GAO analysis of the CMHS.

Average decedent costs could then be expressed as the product of the death
factor and average survivor costs in equation 1. We combined terms and
rearranged equation 1 to solve for average survivor costs in terms of the
county per capita costs, proportions of survivors and decedents, and the
national death factor. 7 This yielded equation 2:

7 Because decedents were enrolled, on average, for only 6 months of the
year, we annualized the death factors.

Because the demographic mix of survivors varies among counties, we then
adjusted each county's costs using the average demographic factor for FFS
survivors in that county. The result is our estimate of the base cost in
each county. 8

2. Compute a joiner cost ratio for each plan that measures prior FFS
spending on that plan's new enrollees relative to average FFS spending on
demographically similar beneficiaries in the same county. We computed 1997
Medicare spending on each beneficiary who left FFS and joined a plan in 1998
(these beneficiaries were termed joiners). 9 We then compared that spending
to the base rate for demographically similar beneficiaries in the joiner's
county of residence. 10 The ratio of aggregate FFS spending on each plan's
joiners to the spending of demographically similar beneficiaries is the
plan's joiner cost ratio. For example, a joiner cost ratio of 0. 75
indicates that a plan's joiners had costs that were three- fourths of those
of demographically comparable FFS beneficiaries in the year before

8 The average demographic factor for survivors must be imputed because
information on beneficiaries who are working- aged or Medicaid- eligible is
not available from HCFA enrollment files. To determine the average
demographic factor for survivors in each county, we multiplied the cell
frequencies for each risk group (from the 1997 AAPCC workbook) by the
corresponding age and gender survival rates of FFS beneficiaries in that
county. We then multiplied the adjusted cell frequencies by the 1997 risk-
score for each cell and calculated a separate average demographic factor for
aged part A, aged part B, disabled part A, and disabled part B. Age and
gender survival rates for a county were obtained from the 1997 denominator
file. Our approach assumes similar survival rates across the following risk
groups: working- aged, Medicaid eligible, non- Medicaid eligible, and
institutionalized.

9 We define joiners as enrollees who spent at least 6 months in FFS in 1997
and at least 7 months in a Medicare+ Choice plan in 1998. Beneficiaries who
enrolled in a plan immediately upon becoming eligible for Medicare were
excluded from our joiner group because they had no prior Medicare FFS costs.
However, our 1997 study (GAO/ HEHS- 97- 16) indicated that these
individuals- known as “age- ins”- tended, like other joiners, to
be healthier than demographically comparable beneficiaries in FFS. We
therefore estimated FFS costs for age- ins using the same methodology
applied to other enrollees. In our sample, age- ins composed about 28
percent of all new plan enrollees during 1998.

10 We applied Medicare's demographic- based risk adjustment factors to the
base rate to account for gross differences in beneficiary health status. For
example, in 1997, the part A risk factor was 0. 65 for a 65- year- old man
who was not institutionalized, not eligible for Medicaid, and not classified
as “working aged,” which means that his part A FFS costs were
expected to equal 65 percent of the average per capita cost. Thus, we
compared the part A cost of a joiner with these demographic characteristics
to an amount equal to 65 percent of the part A base rate in the joiner's
county of residence. We followed the same approach for part B costs.

enrollment. In our study, the weighted average joiner cost ratio was 0. 67,
and 93 percent of plans had a ratio of 0. 80 or lower. 11

The joiner cost ratio did not vary substantially with the number of years
since the start of a plan's Medicare contract or the size of a plan, but
tended to be slightly higher among plans operating in markets with
relatively high percentages of beneficiaries enrolled in plans (see table
5). 12 However, even in those areas where Medicare managed care has a
significant presence, beneficiaries who join plans generally have costs that
are substantially below those of comparable FFS beneficiaries.

Table 5: Joiner Cost Ratio by Market Penetration and Size and Age of Plan
Plan characteristics Mean joiner cost ratio Number of plans

Years since start of plan's Medicare contract 1- 5 years 0. 69 135 6- 10
years 0. 69 34 11+ years 0. 68 41 Medicare+ Choice market penetration
(percent) <10 0. 65 23 10 to 19 0. 67 60 20 to 29 0. 69 72 30 to 39 0. 72 24
40+ 0. 72 31 Plan size (number of enrollees) <10,000 0. 69 86 10, 000 to 49,
000 0. 70 97 50, 000+ 0. 65 27 Source: GAO analysis of HCFA data.

11 Weights were assigned according to the number of enrollees in a plan. The
distribution of joiner cost ratios is consistent with our 1997 study (GAO/
HEHS- 97- 16). In that study, the average joiner's costs were 64 percent of
FFS beneficiaries' costs.

12 Market penetration for each plan was calculated as an enrollee- weighted
average of the market penetration rates (percentage of Medicare
beneficiaries enrolled in any plan) in the counties it served.

3. Compute a set of regression- toward- the- mean (RTM) factors for each
plan that indicate how longer- term members' FFS costs compare to joiners'
costs. The RTM factors account for the tendency of individual enrollee costs
to converge toward the group mean over time. Beneficiaries who had lower-
than- average costs before enrollment in a plan may incur increased costs in
subsequent years. Conversely, beneficiaries who had higher- thanaverage
costs before enrollment may experience reduced costs. That is, enrollees'
costs tend to regress toward the mean the longer they remain plan members.
13 The RTM factors specify the degree to which enrollees' estimated FFS
costs differ from the pre- enrollment FFS costs of new enrollees.

We used CMHS data to analyze how FFS costs changed for a sample of
beneficiaries over a 5- year period. 14 We grouped beneficiaries into eight
cost cohorts based on their 1993 FFS costs relative to the overall sample
mean cost in that year ($ 2,575). 15 We then computed average FFS spending
for each cohort relative to the sample mean FFS costs in each of the
following 4 years. For example, among beneficiaries who had the lowest FFS
costs in 1993, the average cost was 3 percent of the sample mean. (Many
beneficiaries in this group had no Medicare costs.) In 1994, FFS costs for
the same individuals averaged 44 percent of that year's sample mean. By
1997, the group's FFS costs averaged 59 percent of the year's sample mean.
In contrast, average spending on the group of beneficiaries who were most
expensive in 1993 dropped from 1, 678 percent of the sample mean in 1993 to
317 percent of the sample mean in 1997. We then extrapolated the CMHS cost
trends to impute an additional 4 years of relative spending for each cost
cohort.

We then computed plan- specific joiner cost ratios from the CMHS cost trends
and estimated each enrollee's 1998 FFS costs relative to the average

13 Such cost increases are plausible for two reasons. First, beneficiaries
may postpone discretionary care in the months before joining a health plan
to avoid paying Medicare coinsurance and deductibles. Alternatively,
beneficiaries may be more likely to join a plan during a spell of unusually
good health and low service use but subsequently resume service use at a
level that reflects average use by a demographically comparable beneficiary.
In both cases, prior FFS spending may be atypically low.

14 We excluded plan enrollees and individuals who died during this period.
15 We adjusted beneficiaries' FFS costs using HCFA's average geographic
adjusters to account for geographic cost differences. The eight cost cohort
categories were defined by the ratio of beneficiaries' 1993 costs relative
to the overall sample mean using the following ranges: less than 0.10, 0.
11- 0. 25, 0. 26- 0.50, 0. 51- 1.0, 1.01- 2. 0, 2. 01- 4.0, 4.01- 10,
greater than 10.

cost of a demographically similar beneficiary in FFS. An average cost trend
for each plan was computed by using the distribution of its joiners' 1997
FFS costs to weight the eight CMHS cost trends. Each enrollee was then
assigned a plan- specific RTM factor based on the number of years he or she
had been a plan member. 16 These factors express the relationship between
estimated relative FFS costs and the plan's joiner cost ratio. 17 For
example, figure 2 illustrates the estimated RTM for beneficiaries enrolled
in a hypothetical plan Z, which has a joiner ratio and RTM factors typical
of an average plan in our sample. Joiners to the plan had 1997 FFS costs
that equaled 67 percent of the average FFS cost for demographically similar
beneficiaries who did not enroll in a plan. However, the CMHS cost trends
indicate that in 1998- during their first year of plan membership- plan
joiners' FFS costs would have increased to 85 percent of average FFS costs.
Thus the RTM factor for plan Z enrollees in their first year of membership
is 1.27 (0. 67 x 1.27 = 0. 85). We similarly estimated relative FFS costs
for enrollees with 8 or fewer years of plan membership. We assumed that
there was no difference between plan enrollees' costs and FFS beneficiaries'
costs for enrollees who had 9 or more years of plan membership. 18

16 To determine the amount of RTM an individual enrollee experienced, we
calculated the total number of years he or she had been enrolled in any
Medicare health plan that received a fixed monthly payment per enrollee.

17 RTM factors are plan- specific because they depend on the FFS cost
history of each plan's joiners. 18 Approximately 15 percent of enrollees in
our sample had more than 8 years of HMO enrollment.

Figure 2: Illustration of Regression- Toward- the- Mean for Enrollees in
Hypothetical Plan Z

Proportion of FFS Costs

1.00

0.96 0.97 0.95

0.94 0.92

0.90

0.90 0.88 0.86

0.85

0.84

0.80 0.75 0.70

0.67

0.65 0.60 0.55 0.50

Year 1 2345678

Years Enrolled

Before

in a Plan

Joining

RTM Factor 1.26 1.29 1.31 1.34 1.37 1.40 1.43 1.45

Note: Numbers have been rounded to the nearest hundredth.

4. Compute a national update factor that adjusts estimated 1997 spending to
1998 levels. As a result of BBA, HCFA no longer calculates AAPCC rates.
Therefore, we multiplied 1997 FFS spending by the ratio of 1997 national per
capita spending to 1998 national per capita spending to estimate 1998
spending. Because per capita FFS spending fell between 1997 and 1998, the
update factor was 0. 97 for aged beneficiaries and 0. 96 for disabled
beneficiaries.

After we completed the preceding subtasks, imputing the 1998 FFS cost of
each survivor enrolled in a plan involved multiplying a series of terms.
Essentially, for each month in which a surviving beneficiary was enrolled in
a plan, we began with the 1997 base rate multiplied by the beneficiary's
risk adjustment factor to obtain the expected average cost of a
demographically

comparable FFS beneficiary who was alive for the entire preceding year. We
then multiplied that amount by the plan's joiner cost ratio and the RTM
factor applicable for the beneficiary's tenure in a Medicare plan. Finally,
we updated the resulting estimate to account for the change in the base rate
between 1997 and 1998. This process is illustrated for one survivor by
equation 3.

Our approach assumes that when longer- term plan members first enrolled in a
plan, they had FFS spending that was similar to the spending for current
joiners. That is, we assume that a plan that experienced favorable or
adverse selection of enrollees (healthier or sicker than average) in 1998

experienced similar favorable or adverse selection in previous years. This
approach allows us to estimate costs for all plan enrollees based on the
subset of enrollees who had FFS costs in the prior year. 19 Some plan
representatives have indicated that new enrollees in recent years have been
less healthy and more costly compared to new enrollees in earlier years. To
the extent this is true, our joiner ratios will overestimate what was spent
on long- term enrollees when they were last in FFS and result in an
underestimate of excess payments. 20

Methodology to Estimate We assumed that decedents in plans have the same
costs as decedents in

FFS Costs of Enrollees Who FFS and that these costs can be expressed as a
fixed ratio, referred to as a

Died During 1998 death factor, of the average annual cost of FFS survivors.
Consequently, we

estimated the total costs associated with a decedent as the product of three
terms: the annualized base rate, the death factor, and the cost update

19 Our method assumes that the health status of a plan's joiners is similar
fromyear to year. In an earlier study, we tested this assumption by
examining the costs of joiners from different counties over a period of
several years. We found that the costs of joiners did not differ greatly
from year to year.

20 Analyses by GAO and other researchers suggest that favorable selection,
as measured by joiner cost ratios, may be slightly lower (indicating
relatively less- healthy joiners) in markets with relatively high managed
care enrollment. See Call, Dowd, Feldman, and Maciejewski, “Selection
Experiences in Medicare HMOs: Pre- Enrollment Expenditures,” HCFA
Review, Vol. 20, No. 4 (Summer 1999) , pp. 207- 208. Although this study
found that

favorable selection declines as the market share of Medicare HMOs increases,
favorable selection remained substantial in areas with high HMO penetration.

factor. We then aggregated the decedent- related costs experienced by each
plan. These steps are illustrated for a single decedent by equation 4.

Methodology to Estimate We used 1998 Medicare+ Choice payment rates, along
with the number and

Excess Payments or demographic characteristics of each plan's enrollees, to
calculate plans'

Underpayments monthly Medicare payments. We then subtracted from this amount
the

estimated FFS costs of the plan's enrollees (both survivors and decedents).
The difference between these amounts is our measure of excess payments (or
underpayments). The excess payment rate expresses excess payments as a
proportion of total estimated costs. This is illustrated in equation 5.

Comments From the Health Care Financing

Appendi xII Administration

Lett er (101894) Lett er

Ordering Information The first copy of each GAO report is free. Additional
copies of reports are $2 each. A check or money order should be made out to

the Superintendent of Documents. VISA and MasterCard credit cards are
accepted, also.

Orders for 100 or more copies to be mailed to a single address are
discounted 25 percent.

Orders by mail: U. S. General Accounting Office P. O. Box 37050 Washington,
DC 20013

Orders by visiting: Room 1100 700 4th St. NW (corner of 4th and G Sts. NW)
U. S. General Accounting Office Washington, DC

Orders by phone: (202) 512- 6000 fax: (202) 512- 6061 TDD (202) 512- 2537

Each day, GAO issues a list of newly available reports and testimony. To
receive facsimile copies of the daily list or any list from the past 30
days, please call (202) 512- 6000 using a touchtone phone. A recorded menu
will provide information on how to obtain these lists.

Orders by Internet: For information on how to access GAO reports on the
Internet, send an e- mail message with “info” in the body to:
info@ www. gao. gov or visit GAO's World Wide Web home page at: http:// www.
gao. gov

To Report Fraud,

Contact one:

Waste, or Abuse in

Web site: http:// www. gao. gov/ fraudnet/ fraudnet. htm

Federal Programs

e- mail: fraudnet@ gao. gov 1- 800- 424- 5454 (automated answering system)

GAO United States General Accounting Office

Page 1 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Contents

Page 2 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Page 3 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments United States
General Accounting Office

Washington, D. C. 20548 Page 3 GAO/ HEHS- 00- 161 Medicare+ Choice Plan
Payments

B- 285401 Page 4 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 5 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 6 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 7 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 8 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 9 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 10 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 11 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 12 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 13 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 14 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 15 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 16 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 17 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 18 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 19 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 20 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

B- 285401 Page 21 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Page 22 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I

Appendix I Methodology

Page 23 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 24 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 25 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 26 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 27 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 28 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 29 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 30 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 31 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 32 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix I Methodology

Page 33 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Page 34 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix II

Appendix II Comments From the Health Care Financing Administration

Page 35 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix II Comments From the Health Care Financing Administration

Page 36 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix II Comments From the Health Care Financing Administration

Page 37 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

Appendix II Comments From the Health Care Financing Administration

Page 38 GAO/ HEHS- 00- 161 Medicare+ Choice Plan Payments

United States General Accounting Office Washington, D. C. 20548- 0001

Official Business Penalty for Private Use $300

Address Correction Requested Bulk Rate

Postage & Fees Paid GAO Permit No. GI00
*** End of document ***