Financial Experience of Managed Care Organizations Participating in Medicare+Choice

Transcription

1 Contract No.: (6) MPR Reference No.: Financial Experience of Managed Care Organizations Participating in Medicare+Choice Final Report January 18, 2001 Robert Schmitz Thomas Kornfield Submitted to: Health Care Financing Administration Office of Strategic Planning 7500 Security Boulevard, C Baltimore, MD Project Officer: Brigid Goody Submitted by: Mathematica Policy Research, Inc. 600 Maryland Avenue, S.W. Suite 550 Washington, DC (202) Project Director: Anna Cook

2

3 CONTENTS Chapter Page I INTRODUCTION... 1 II DATA... 3 A. MEDICARE+CHOICE CONTRACTS AND ENROLLMENT... 3 B. ADJUSTED COMMUNITY RATE PROPOSALS... 4 III EMPIRICAL APPROACH... 7 IV RESULTS... 9 V DISCUSSION REFERENCES APPENDIX ii

4 TABLES Table Page II.1 FINANCIAL INDICATORS... 6 IV.1 INDICATORS OF MCOPROFITABILITY, 1998 AND 1999, BY MARKET AREA IV.2 INDICATORS OF MCO PROFITABILITY, 1998 AND 1999 BY HCFA REGION AND M+C MARKET PENETRATION RATE IV.3 INDICATORS OF MCO PROFITABILITY, 1998 AND 1999, BY HCFA REGION AND M+C MARKET PENETRATION RATE: PLANS OPERATING BOTH IN 1998 AND IV.4 INDICATORS OF MCO LIQUIDITY BY MARKET, 1998 AND IV.5 INDICATORS OF LIQUIDITY, 1998 AND 1999, BY HCFA REGION AND M+C MARKET PENETRATION RATE IV.6 MEAN DAYS IN UNPAID CLAIMS, 1998 AND 1999 BY HCFA REGINO AND M+C MARKET PENETRATION RATE iii

5 I. INTRODUCTION Although the Balanced Budget Act (BBA) of 1997 changed several elements of Medicare managed care, probably no change has received more attention or press coverage than the provisions governing payments to Managed Care Organizations (MCOs). Under the BBA s Medicare+Choice program, MCOs are now paid a monthly capitation rate equal to the highest of the following three amounts: a blend of local and national rates stabilizing by 2003 at 50 percent local and 50 percent national; a so-called floor rate set at $367 in 1998 and updated each year; and a minimum update equal to a 2 percent increase over the previous year s rate. Because many Medicare risk contracts had previously operated in counties in which the monthly Medicare capitation payment exceeded the U.S. average, most 1998 capitation payments were set by the third method- a 2 percent increase over 1997 rates, with the prospect of continued slow growth in payments for 1999 and beyond. Administrators from the Health Care Financing Administration (HCFA) have asserted that the new rates remain above the actuarial value of services provided to beneficiaries who enroll in Medicare+Choice plans, a position also taken by the General Accounting Office (Hash 1999, GAO 1999). Nevertheless, many MCOs have left the program, complaining of low payment rates and excessive regulation and paperwork. The departure of some (or perhaps even all) MCOs from Medicare is arguably a desirable outcome of the new payment system. The rationale for risk contracting under Medicare had always been that such plans could provide care comparable in quality to that delivered under fee-for-service arrangements, though at lower cost to the Medicare Trust Fund. If, however, the risk premium required to induce MCO participation 1

6 in the Medicare+Choice program is so high that capitation rates must exceed the expected value of Medicare outlays under fee-for-service arrangements, then the benefit of managed care to the Medicare program is in serious doubt. At the same time, many analysts have noted that Medicare+Choice plans have been especially attractive to low-income beneficiaries who might otherwise have great difficulty affording copayments for Part B services and, most particularly, prescription drugs. A 1996 survey of beneficiaries enrolled in Medicare risk HMOs found that 60 percent had incomes of $20,000 or less and 20 percent incomes under $10,000 (Nelson et al., 1996). An earlier survey by Brown et al. (1993) found that the average income of enrollees in Medicare risk plans was 13 percent lower than that of beneficiaries who remained under fee-for-service. For low-income beneficiaries who do not qualify for Medicaid, loss of Medicare HMO coverage may impose financial burdens they are ill equipped to bear. Whether the availability of Medicare risk plans is the best means for addressing the needs of such beneficiaries, there is a legitimate public policy interest in monitoring the availability and performance of Medicare+Choice plans over time. This report is part of a larger project to monitor the Medicare+Choice program over time. Specifically, it presents results that describe the financial performance of Medicare+Choice plans during 1998 and It relies on measures of profitability, liquidity, and efficiency drawn from Adjusted Community Rate (ACR) proposals submitted by MCOs for plan years 2000 and The plans selected for analysis are risk contracts that operate in 69 metropolitan statistical areas (MSAs). The 69 areas include all MSAs with a population of 1.5 million or more and all other MSAs in which 30 percent or more of the Medicare population is enrolled in managed care plans. Chapter II of the report describes the data sources used in the analysis, the measures selected, and the limitations of the data. Chapter III outlines the analytic methods and weighting of area-level means. Chapter IV presents the results, which are discussed briefly in Chapter V. 2

7 II. DATA Assessing the financial performance of organizations participating in the Medicare+Choice program would best be accomplished through an analysis of data describing the relationship between total revenue and the cost of providing services to individuals enrolled in the health plans sponsored by these organizations. The effect of Medicare + Choice participation on financial performance could be gauged by examination of the incremental contributions of both Medicare capitation payments to total revenue and of the cost of providing care to Medicare + Choice enrollees to total cost. Lacking specialized data on the costs MCOs incur in furnishing care to Medicare beneficiaries, we are unable to carry out this latter task. We instead exploit information gathered through the process of rate setting and regulation. This report relies heavily on financial information drawn from Adjusted Community Rate proposals for calendar years 2000 and The information is combined with data describing the number of Medicare+Choice contracts and total Medicare enrollment under these contracts for those same years. A later report will draw on information contained in the Physician Incentive Plan (PIP) disclosure form to characterize financial arrangements between MCOs and providers and to approximate the changes in the size of physician networks. This chapter describes all data sets, including the PIP disclosures. A. MEDICARE+CHOICE CONTRACTS AND ENROLLMENT Two data files maintained by the Health Care Financing Administration were used to ascertain the number of plans operating in each market and the number of Medicare enrollees by plan. The Geographic Service Area (GSA) File identifies the counties in each contract s service area. We used the GSA File to identify managed care plans operating in each of the 69 markets under study. 3

8 The Medicare Managed Care Market Penetration State/County/Plan Files report the number of beneficiaries and the number of Medicare+Choice enrollees, by contract, in each county in the United States. After identifying the contracts serving each of the 69 market areas selected for this report, we determined the number of beneficiaries enrolled in each contract, by market, by using the Market Penetration File. Nelson et al. (2000) provide details. B. ADJUSTED COMMUNITY RATE PROPOSALS Managed care organizations wishing to participate in the Medicare+Choice program are required to submit an Adjusted Community Rate (ACR) proposal by July 1 of the preceding year. Calculations carried out in the ACR submissions produce cost ratios that in turn are used to estimate Medicare + Choice plan costs. In developing these ratios, organizations may account for differences in historical utilization of Medicare and non-medicare populations. Plans may use other (non-historical) estimates to establish an ACR value for new services. If the weighted average of payment rates over the counties served by the MCO (termed the Annual Payment Rate or APR) exceeds the cost of benefits, then the MCO may place the excess in a stabilization fund or use it to fund additional benefits. MCOs are not allowed to earn a higher profit on their Medicare+Choice plans than on their non Medicare plans. ACR submissions for 2000 and later required plans to report limited financial data for the full calendar year before the submission. Thus, ACR proposals for 2000, submitted in July 1999, contained financial data for ACR proposals for 2001 contained financial data for Reported items include net worth, total and operating revenue, overall profit margin, operating profit margin, overall expense ratio, the ratio of assets and long-term bonds to current liabilities, and days in unpaid claims. Data from ACR proposals suffer from several limitations and, accordingly, do not provide as complete or accurate a picture of the Medicare+Choice market as we might wish. 4

9 Financial information reported in the ACR reflects the experience of the managed care entity as a whole, including those who are and those who are not Medicare beneficiaries. Moreover, we do not know how many people not covered by Medicare are enrolled in each plan. 1 Even if we did know the number of Medicare and non- Medicare enrollees, our ability to allocate total cost to Medicare and non-medicare components would be limited. Financial indicators drawn from ACR proposals and summarized at the market (MSA) level do not necessarily reflect the financial circumstances of that market or the adequacy of Medicare+Choice payment rates relative to Medicare cost. Financial information reported in the ACR reflects the plan s experience for all areas in which the plan operates. Financial data are reported only for those plans applying to continue their participation in the Medicare+Choice program. We use financial indicators for 1998 that are reported in mid-1999 by plans that wish to participate in the Medicare+Choice program in Data for 1999 are reported in mid-2000 by plans that wish to participate in ACR data accurately characterize financial behavior only if the number of plans failing to continue in the program (and thus not reporting data on past financial experience) is relatively small. Financial indicators used for this report are drawn from Worksheet B1 of the ACR proposal. Indicators appear in three categories: performance, liquidity, and efficiency. We selected six indicators for analysis in this report. Table II.1 presents the indicators and their definitions. The second measure of liquidity listed in the table is reported on the ACR proposal in order to provide a more comprehensive description of the relationship between liquid assets and liabilities than is suggested by the current ratio. The ratio of assets and long term bonds to current liabilities takes into account the fact that many organizations move a good deal of their spare cash to longer-term assets, such as Treasury and blue-chip corporate bonds (HCFA 2000). 1 Although Worksheet A of the ACR requires MCOs to report projected Medicare and non Medicare enrollment for the coming rate year, we do not at this point believe it reasonable to assume that the ratio of Medicare to total enrollment was the same during the base year. 5

10 TABLE II.1 FINANCIAL INDICATORS Indicator PERFORMANCE Definition Operating profit margin Overall profit margin Overall expense ratio Ratio of operating profit to operating revenue. Higher values indicate higher performance. Ratio of total profit to total revenue. Higher values indicate higher performance. Ratio of direct medical cost plus administrative cost to operating revenue. Higher values indicate lower performance. LIQUIDITY Current ratio Current assets and long-term bonds divided by current liabilities Ratio of current assets to current liabilities. Higher values indicate higher liquidity. Ratio of the sum of current assets and long-term bonds to current liabilities. Higher values indicate higher liquidity. EFFICIENCY Days in unpaid claims Ratio of claims payable to hospital and medical expenses divided by 365. Higher values indicate lower efficiency. SOURCE: Health Care Financing Administration. Instructions for Completing the Adjusted Community Rate Proposal for Contract Year The ACR data file does not consistently report the value of certain variables reported as ratios. The pattern of values strongly suggests that some MCOs multiplied the ratios by 100 to express the results as percentages while others did not. To ensure consistency, we recalculated the values for operating profit margin and overall profit margin by using appropriate profit and revenue data from the ACR file. Recalculation was impossible in other cases because the necessary quantities are not reported on the ACR data file. After examining the distribution of values for the overall expense ratio, however, we decided to divide the reported value by 100 whenever the reported value exceeded This produced a set of values that we believe is most likely to reflect the true quantities, all expressed in pure ratio form. Appendix A details certain other decisions regarding the ACR data. 6

11 III. EMPIRICAL APPROACH We selected for analysis those MCOs holding risk-bearing contracts under the Medicare+Choice program in any of 69 Metropolitan Statistical Areas (MSAs) in the United States with a population of 1.5 million or more or with a Medicare managed care penetration rate of 30 percent or higher in Nelson et al. (2000) present details regarding the selection of market areas. Within each market, weighted means for 1998 and 1999 were computed for each of the variables listed in Table II.1 by using ACR proposals for 2000 and Weights for each contract were defined as the ratio of that contract s number of Medicare+Choice enrollees in the specified market to the total number of Medicare+Choice enrollees in the market. All weights were scaled to sum to one for each market area. Weighted means were also computed for the same two years by HCFA region and by a combination of the Medicare managed care penetration and local Medicare+Choice payment rates. For this analysis, markets with managed care penetration rates of 25 percent or more were distinguished from markets with managed care penetration rates below 25 percent. The level of 1998 Medicare+Choice payment rates was defined by the rates relationship to the 1998 United States Per Capita Cost (USPCC). The ratio of each contract s 1998 Medicare+Choice payment to the USPCC was placed into one of the following categories: less than 1.00, 1.00 to 1.15, and greater than To assess the extent to which observed changes between 1998 and 1999 were caused by changes experienced by plans reporting data for both years (and thus participating in Medicare+Choice for 2000 and 2001) and by the entry and exit of plans from the market, weighted means for profitability were repeated by using only those plans participating in the market in both 2000 and

12 We made no attempts to conduct statistical tests on observed differences over time or across markets, regions, or other strata. Such tests would be misleading for two reasons. First, the markets examined here are not meant to constitute a sample representing a larger population about which we seek to draw inferences. The markets and years selected are the population of interest. Second, all financial information is reported at the level of the contract and not the market area. Because some contracts operate in more than one market, identical information is repeated for each market under study. Thus, the observations for each market cannot be assumed to be independent of each other. Statistical tests that assume such independence will ordinarily provide erroneous results. 8

13 IV. RESULTS In most of the 69 market areas, 1998 s mean operating profit rate and overall profit rate of MCOs operating under Medicare+Choice contracts were negative. Table IV.1 displays weighted mean values for operating profit, overall profit, and overall expense ratio for 1998 and 1999 in each of the 69 selected markets. Medicare risk contracts operated in 64 of the 69 market areas in Mean operating profit was positive in that year for only 26 markets; mean overall profit was positive in 22 markets. 2 Although both mean operating profit and mean overall profit were more likely to increase rather than decrease between 1998 and 1999, the change was not dramatic. In 1999, mean operating profit remained negative in half of the market areas that operated with risk contracts. The pattern of growth or decline in profit was not geographically uniform. Plans operating in markets in the Northeast, such as Boston, New Haven, New York, and Newark, tended to exhibit increases in operating and overall profit between 1998 and 1999, though in many cases mean profit remained negative in Plans operating in the South, especially in Texas and Louisiana, and plans operating in Colorado were more likely to have experienced reductions in profit. Table IV.2 shows weighted mean profit and expense ratios by HCFA region. Aside from the decreases in Regions 6, 8, and 10, represented in our sample by 20 markets in Louisiana, Texas, New Mexico, Colorado, Oregon, and Washington, mean profitability remained the same or increased slightly over the period. 2 Because the data are not market-specific as previously noted, it would be more precise to say that in 26 markets, the weighted mean of operating profit among all MCOs that compete in those markets was positive. 9

14 TABLE IV.1 INDICATORS OF MCO PROFITABILITY, 1998 AND 1999, BY MARKET AREA (Weighted by market share in contract total) No. of Plans Operating Profit Margin Overall Profit Margin Overall Expense Ratio Albuquerque, NM Atlanta, GA Bakersfield, CA Baltimore, MD Baton Rouge, LA Boston, MA Boulder, CO Chicago, IL Cincinnati, OH Cleveland, OH Colorado Springs CO Dallas, TX Daytona Beach, FL Denver CO Detroit MI Dubuque IA Eugene, OR Fort Lauderdale, FL Fort Worth, TX Grand Junction, CO Honolulu, HI Houma, LA Houston TX Jacksonville, FL Kansas City, MO Killeen, TX Las Vegas, NV Los Angeles, CA Medford, OR Miami, FL Minneapolis, MN Modesto, CA Nassau, NY New Haven, CT New York, NY Newark, NJ

15 TABLE IV.1 (continued) No. of Plans Operating Profit Margin Overall Profit Margin Overall Expense Ratio Norfolk, VA Oakland, CA Olympia, WA Orange County, CA Philadelphia, PA Phoenix, AZ Pittsburgh, PA Portland, OR Pueblo, CO Riverside, CA Rochester, NY Sacramento, CA St. Louis, MO Salem, OR San Antonio, TX San Diego, CA San Francisco, CA San Jose, CA San Luis Obispo, CA Santa Barbara, CA Santa Rosa, CA Seattle, WA Spokane, WA State College, PA Stockton, CA Tampa, FL Tucson, AZ Vallejo, CA Ventura, CA Washington DC West Palm Beach FL Williamsport, PA Yolo, CA

16 TABLE IV.2 INDICATORS OF MCO PROFITABILITY, 1998 AND 1999, BY HCFA REGION AND M+C MARKET PENETRATION RATE (Weighted by market share in contract total) Total 69 Number of Operating profit margin Overall profit margin Overall expense ratio markets HCFA Region Region 1 (CT, MA) Region 2 (NJ, NY) Region 3 (DC, MD, PA, VA) Region 4 (FL, GA) Region 5 (IL, MI, MN, OH) Region 6 (LA, NM, TX Region 7 (MO) Region 8 (CO) Region 9 (AZ, CA, HI, NV) Region 10 (OR, WA) Ratio of 1998 M+C payment rate to 1998 USPCC < M+C Market Penetration <25% M+C Market Penetration 25% Above to M+C Market Penetration <25% M+C Market Penetration 25% Above > M+C Market Penetration <25% M+C Market Penetration 25% Above NOTE: Markets examined for this study are not necessarily representative of all markets with Medicare+Choice plans. States in the 69-market sample are listed by HCFA region. 12

17 Table IV.2 also displays mean profit margins and overall expense ratios by level of Medicare+Choice payment and degree of market penetration. Although one might reasonably have expected profitability to be higher in markets with higher payment rates, no such relationship is evident in the table. Again, one must bear in mind that profits are reported for MCOs as a whole and are not market specific. There is, however, a barely discernible tendency for mean profit margins to increase in markets with Medicare+Choice penetration rates below 25 percent and to decline or remain the same in markets with penetration rates above 25 percent. In each of the three payment strata, the mean operating profit margin increased between 1998 and 1999 in markets in which less than 25 percent of beneficiaries were enrolled in Medicare+Choice plans. In those markets where 25 percent or more of beneficiaries were enrolled in such plans, operating profit rates declined in two cases and remained unchanged in the other. The change in mean profit between 1998 and 1999 reflects both changes in the composition of the sample of MCOs that participated in the Medicare+Choice program from one year to the next and changes in profit by MCOs that participated in each of the two years. Tables IV.1 and IV.2 cannot distinguish whether the slight rise in operating profit between 1998 and 1999 occurred because individual Medicare+Choice contracts became more profitable or because the least profitable MCOs left the Medicare market after The latter explanation is potentially important. While 311 contracts filed ACR proposals for 2000 (reporting base-year data for 1998) for the 69 markets under study here, only 213 did so for Table IV.3 repeats the information provided by Table IV.2 using only those contracts that participated in 3 Because contracts may operate in two or more markets, the numbers of contracts cited here are not necessarily unique. There were 311 contract/market combinations reported for 2000 and 213 for

18 TABLE IV.3 INDICATORS OF CONTRACT PROFITABILITY, 1998 AND 1999, BY HCFA REGION AND M+C MARKET PENETRATION RATE: PLANS OPERATING IN BOTH 1998 AND 1999 (Weighted by market share in contract total) Total 69 Number Operating profit margin Overall profit margin Overall expense ratio of markets HCFA Region Region 1 (CT, MA) Region 2 (NJ, NY) Region 3 (DC, MD, PA, VA) Region 4 (FL, GA) Region 5 (IL, MI, MN, OH) Region 6 (LA, NM, TX) Region 7 (MO) Region 8 (CO) Region 9 (AZ, CA, HI, NV) Region 10 (OR, WA) Ratio of 1998 M+C payment rate to 1998 USPCC < M+C Market Penetration < 25% M+C Market Penetration 25% or Above to M+C Market Penetration < 25% M+C Market Penetration 25% or Above > M+C Market Penetration < 25% M+C Market Penetration 25% or Above NOTE: Markets examined for this study are not necessarily representative of all markets with Medicare+Choice plans. States in the 69-market sample are listed by HCFA region. 14

19 Medicare+Choice in both 2000 and Because a large number of contracts left the Medicare+Choice program at the end of 2000 and only a few new contracts entered for 2001, nearly all of the entries for 1999 are identical in the two tables. That most of the entries for 1998 are roughly the same as well indicates that 1998 financial characteristics were broadly similar for contracts that left the program at the end of 2000 and those that remained. Whether or not those programs that left also had similar financial characteristics in 1999 cannot be determined since their 2001 ACR proposals, which would have reported the data, were never submitted. While MCOs participating in the Medicare+Choice program may have been slightly more profitable in 1999 than in 1998, their liquidity, on average, was about the same in 1999 as in Table IV.4 displays weighted means, by market, for two measures of plan liquidity: the current ratio (current assets divided by current liabilities) and the ratio of current assets and longterm bonds to current liabilities. Table IV.5 shows the same measures by HCFA region and by payment rate and Medicare+Choice market penetration. Although the weighted means across all contracts were about the same for the two years as shown in Table IV.5, the means by market for both measures were more likely to decline than to increase between 1998 and In 63 markets for which means were available in both 1998 and 1999, the mean of the ratio of cash plus long-term bonds to current liabilities decreased in 37 cases, increased in 24 cases, and remained unchanged in two cases. If we restrict attention to instances in which the change in the ratio was substantial, the difference is even greater. Of the 24 markets in which the mean of the ratio changed by more than 0.2, 16 showed a decline in the ratio and eight showed an increase. Results for the current ratio are similar. Measures of liquidity change by HCFA region are difficult to interpret because the two measures themselves do not move in parallel. As gauged by the current ratio, liquidity improved 15

20 TABLE IV.4 INDICATORS OF MCO LIQUIDITY BY MARKET, 1998 AND 1999 (Weighted by market share in contract total) No. of plans Current ratio Cash + LT bonds divided by current liabilities Albuquerque, NM Atlanta, GA Bakersfield, CA Baltimore, MD Baton Rouge, LA Boston, MA Boulder, CO Chicago, IL Cincinnati, OH Cleveland, OH Colorado Springs CO Dallas, TX Daytona Beach, FL Denver CO Detroit MI Dubuque IA Eugene, OR Fort Lauderdale, FL Fort Worth, TX Grand Junction, CO Honolulu, HI Houma, LA Houston TX Jacksonville, FL Kansas City, MO Killeen, TX Las Vegas, NV Los Angeles, CA Medford, OR Miami, FL Minneapolis, MN Modesto, CA Nassau, NY New Haven, CT New York, NY Newark, NJ Norfolk, VA Oakland, CA Olympia, WA Orange County, CA Philadelphia, PA Phoenix, AZ Pittsburgh, PA Portland, OR Pueblo, CO

21 TABLE IV.4 (continued) No. of plans Current ratio Cash + LT bonds divided by current liabilities Riverside, CA Rochester, NY Sacramento, CA St. Louis, MO Salem, OR San Antonio, TX San Diego, CA San Francisco, CA San Jose, CA San Luis Obispo, CA Santa Barbara, CA Santa Rosa, CA Seattle, WA Spokane, WA State College, PA Stockton, CA Tampa, FL Tucson, AZ Vallejo, CA Ventura, CA Washington DC West Palm Beach FL Williamsport, PA Yolo, CA

22 TABLE IV.5 INDICATORS OF LIQUIDITY, 1998 and 1999, BY HCFA REGION AND M+C MARKET PENETRATION RATE (Weighted by market share in contract total) Number of Current ratio Cash + LT bonds divided by current liabilities markets Total HCFA Region Region 1 (CT, MA) Region 2 (NJ, NY) Region 3 (DC, MD, PA, VA) Region 4 (FL, GA) Region 5 (IL, MI, MN, OH) Region 6 (LA, NM, TX) Region 7 (MO) Region 8 (CO) Region 9 (AZ, CA, HI, NV) Region 10 (OR, WA) Ratio of 1998 M+C payment rate to 1998 USPCC < M+C Market Penetration < 25% Market Penetration 25% of above to M+C Market Penetration < 25% Market Penetration 25% of above > M+C Market Penetration < 25% Market Penetration 25% of above NOTE: Markets examined for this study are not necessarily representative of all markets with Medicare+Choice plans. States in the 69-market sample are listed by HCFA region. 18

23 in Regions 4, 5, 9, and 10. As gauged by the ratio of cash plus long-term bonds to current liabilities, however, liquidity improved in Regions 2 and 7. Both measures indicate a sharp fall in liquidity in Region 8, which is represented in our sample by contracts operating in Boulder, Denver, Pueblo, and Colorado Springs. Table IV.6 shows mean days in unpaid claims for 1998 and 1999 both by HCFA region and by level of Medicare+Choice payment and degree of managed care penetration. Higher values are often considered to be signs of inefficiency or financial distress. The biggest relative increases occurred in Regions 2, 7, and 9. Whether coincidence or not, Regions 2 and 7 were the only HCFA regions to show an increase in the ratio of cash plus long-term bonds to current liabilities, and Region 9 showed the smallest decline (among regions exhibiting declines). A natural hypothesis is that some plans may have increased their liquidity by making providers wait longer for payment, but this is no more than speculation at this point. As in the case of operating profit, there is the suggestion in Table IV.6 of a relationship between Medicare+Choice market penetration and the change in mean days in unpaid claims. In each case, mean days in unpaid claims increased by a greater percentage in markets in which market penetration exceeded 25 percent compared with those in which market penetration was less than 25 percent. 19

24 TABLE IV.6 MEAN DAYS IN UNPAID CLAIMS, 1998 AND 1999, BY HCFA REGION AND M+C MARKET PENETRATION RATE (Weighted by market share in contract total) Number of markets Percent change Total HCFA Region Region 1 (CT, MA) Region 2 (NJ, NY) Region 3 (DC, MD, PA, VA) Region 4 (FL, GA) Region 5 (IL, MI, MN, OH) Region 6 (LA, NM, TX) Region 7 (MO) Region 8 (CO) Region 9 (AZ, GA, HI, NV) Region 10 (OR, WA) Ratio of 1998 M+C payment rate to 1998 USPCC < M+C Market Penetration < 25% M+C Market Penetration 25% of Above to M+C Market Penetration < 25% M+C Market Penetration 25% of Above > M+C Market Penetration < 25% M+C Market Penetration 25% of Above NOTE: Markets examined for this study are not necessarily representative of all markets with Medicare + Choice plans. States in the 69-market sample are listed by HCFA region. 20

25 V. DISCUSSION While the present project seeks to monitor the financial performance and behavior of managed care plans operating under the Medicare+Choice program, there is reason to question whether the currently available data permit such monitoring. Financial data reported on the ACR proposals have three important shortcomings when used for monitoring purposes. First, the data are specific to neither the individual markets in which plans may operate nor the Medicare population served. All data apply to both Medicare and non Medicare enrollees in the plans covered by the contract. Moreover, there is no reliable information on the number of non Medicare enrollees. Therefore, it is impossible to assess whether changes in profitability, net worth, liquidity, or other measures of financial health are the result of changes in Medicare payment or rules or the result of other changes unrelated to Medicare policy. Second, ACR proposals are filed only by MCOs that seek to continue their participation in the Medicare+Choice program. An MCO that wished to participate in the program in 2000 filed an ACR proposal in 1999 and reported financial information for Financial information from plans that participated in 2000 but did not apply for the program for 2001 will never be available since these organizations did not submit an ACR proposal for that year. The accuracy of our depiction of financial performance in 1999 hinges on the representativeness of plans that remained in the program. In the markets studied for this report, nearly one-third of the plans providing ACR data for 1998 did not provide data for 1999; accordingly, the extent of possible bias is severe. Third, the accuracy of reported ACR data is itself open to question. As noted in Chapter II, the structure of the data allowed us to detect certain errors. Because profit and revenue information were reported, we could check the accuracy of operating profit margin and overall profit 21

26 margin as it appeared on the file. The file also repeated MCO-level information for each plan operating in a given market area. This allowed us to identify instances in which certain information, such as operating revenue or profit, should be identical on two separate records but proved dissimilar. In certain cases, for example, those in which two numbers differ by a factor of 10, it was possible to make reasonable guesses about which was more accurate. In other cases, there was no basis for a decision. Bearing in mind the serious limitations of the data, are there any lessons to be drawn from a comparison of financial indicators for 1998 and 1999 as reported on ACR proposals? Tables IV.1 through IV.6 contain no striking results pointing to early effects of Medicare+Choice on profits, liquidity, or plan efficiency. The key finding of this paper, if any, is the absence of dramatic change. Nevertheless, several empirical regularities deserve further study in the future. The average operating and overall profit margin of Medicare+Choice contracts operating in the 69 markets areas rose slightly between 1998 and Nevertheless, contracts in Louisiana, Texas, and Colorado were more likely to experience reductions in profit over the period. Profits increased more rapidly in the Northeast and East. Measures of profitability increased by larger amounts on average in markets where the penetration rate for Medicare managed care was below 25 percent and remained constant or fell in markets where the rate exceeded 25 percent. It is tempting to speculate that greater plan competition (in the form of richer benefits or increased marketing cost) may have exerted upward pressure on contract costs in areas of high market penetration. This explanation, however, is highly speculative. Additional evidence could be secured through an analysis of year-to-year changes in plan benefits or in the ratio of marketing costs to total cost. Increases in either would tend to suggest that competitive pressure accounted for the decrease in profit. 22

27 On average, the reported liquidity of MCOs remained about the same in 1998 and The distribution of changes in liquidity was, however, skewed; substantial decreases in liquidity were more common than substantial increases. Because those HCFA regions that showed increases in one measure of liquidity (the ratio of cash plus long-term bonds to current liabilities) also exhibited the largest increase in mean days in unpaid claims, it is possible that some contracts raised their liquid assets by delaying payments to providers, though such deferments cannot be verified with the data available. There are two steps one might take at this time to improve our ability to monitor the financial performance of MCOs under Medicare+Choice. The first is to conduct a more thorough examination of the accuracy and reliability of the data than that so far attempted. As noted, some data elements appear suspect. Errors in the data have much greater effect when aggregating to the market level, which, in many cases, involves only a handful of contracts. The second step is to try to secure some information about the relative magnitudes of Medicare and non Medicare enrollment in plans represented in the ACR data. 4 Paying special attention to plans in which the Medicare share of enrollment is high would furnish better estimates of the likely effect of Medicare policy on financial characteristics of plans. 4 In principle, data maintained by Interstudy could be used for this purpose. The process of linking Interstudy data and ACR data is, however, complex and time-consuming because the two data files are not now cross-referenced. 23

28 REFERENCES Brown, R., J.W. Bergeron, D.G. Clement, et al. The Medicare Risk Program for HMOs Final Summary Report on Findings from the Evaluation. Princeton, NJ: Mathematica Policy Research, February Hash, Mike. Testimony on Medicare+Choice. Senate Finance Committee. Washington, DC: June 9, Nelson, Lyle, Amanda Cassidy, Thomas Kornfield, et al. Early Experience Under Medicare+Choice: First Interim Report. Washington, DC: Mathematica Policy Research, January 4, Nelson, Lyle, Marsha Gold, Randall Brown, et al. Access to Care in Medicare Managed Care: Results from a 1996 Survey of Enrollees and Disenrollees. Washington, DC: Mathematica Policy Research, November 7, U. S. Department of Health and Human Services. Health Care Financing Administration. Instructions for Completing the Adjusted Community Rate Proposal for Contract Year 2001, Baltimore, MD: DHHS. U. S. General Accounting Office. Medicare+Choice: Reforms Have Reduced but Likely Not Eliminated Excess Plan Payments. Letter Report, June 10,

29 APPENDIX Our initial examination of the ACR data identified anomalous values for certain variables appearing in the file. This brief appendix describes changes made to the data prior to analysis. Changes were made in two categories: replacing percentages with decimals and selecting one of two or more inconsistent values. We consider each in turn. A. PERCENTAGE AND DECIMAL VALUES In computing ratio variables such as the medical expense ratio (MER), administrative expense ratio (AER), and overall expense ratio (OER), most MCOs reported decimal fractions between 0.00 and This is consistent with the directions given on page 33 of the instructions for completing the 2001 ACR proposals. Other plans appear to have multiplied these ratios by 100 to express the value as a percentage between 0 and 100. We adopted the following rule, which appears to furnish the most reasonable results: IF OERxx > 3.75 then divide OERxx, AERxx, and MERxx by 100, where xx = 98 or 99 as appropriate. In one instance, a negative reported value of OER was changed to a positive value before applying the rule. B. RESOLVING CONTRADICTORY VALUES Duplicate values appear frequently in the file and are expected because contract-level information is submitted for each plan identifier within a specific contract (H number). On occasion, entries that appear to be (legitimate) duplicates differ in the reported values of a single variable. Often the two values differ by a factor of 10, strongly suggesting that a simple keystroke error 25

30 accounts for the discrepancy. In other cases, we selected a particular value as most likely to be correct based on its relationship to other variables from the same record. 26