Establishing an Analytical Framework for Measuring the Role of Reinsurance in the Health Insurance Market

This is a limited analytic study of the role of reinsurance (and similar coverages) in three insurance markets: primary health insurance coverage offered by indemnity insurance companies; health insurance coverage offered by health maintenance organizations; and health benefits coverage offered by employer-sponsored self-insured plans. The purpose of the study is to develop a better understanding of the roles that reinsurance plays in the market for health benefit coverage by looking at the amount of risk borne by reinsurance in the three primary conduits for health coverage: indemnity health insurance, HMO coverage and self-funded health plans.

Prepared for The Office of the Assistant Secretary for Planning and Evaluation,

Department of Health and Human Services by The Lewin Group,

March 20, 1997

"

A. Introduction

Under a delivery order from the Assistant Secretary for Planning and Evaluation (ASPE) at the Department of Health and Human Services, the Lewin Group has undertaken a limited analytic study of the role of reinsurance (and similar coverages) in three insurance markets: primary health insurance coverage offered by indemnity insurance companies;¹ health insurance coverage offered by health maintenance organizations; and health benefits coverage offered by employer-sponsored self-insured plans.² The purpose of the study is to develop a better understanding of the roles that reinsurance plays in the market for health benefit coverage by looking at the amount of risk borne by reinsurance in the three primary conduits for health coverage: indemnity health insurance, HMO coverage and self-funded health plans.

The primary tasks of the study were:

• To identify and to develop a better understanding of the sources of information about the amount or level of risk borne by reinsurance in the three specified markets.
• To collect information where possible from the identified sources.
• Based upon the available information, to develop a profile of the role that reinsurance plays in the three markets.
• To develop a baseline for measuring the use of reinsurance in the three markets and to present options for obtaining more detailed information on the role of reinsurance in the markets.

Although reinsurance is widely used by the underwriters of health benefits (i.e., insured and self-funded plans), its role is little understood by policymakers. Basically, reinsurance is insurance bought by insurers,³ and is used by insurers to limit their risk exposure. Reinsurance contracts can be structured in many ways, and insurers can use reinsurance to limit their risk for an individual contract or exposure, for a group of contracts or exposures, or for a whole portion of the insurer's business.

In the health insurance and health benefits contexts, the most common forms of reinsurance coverage are individual and aggregate stoploss insurance. Insurers purchase individual stoploss insurance to limit or cap their claims exposure for any covered individual over a specified period (such as a year). For example, an insurer may purchase stoploss coverage that would reimburse the insurer for 90 percent of the claims costs incurred by a covered individual in excess of a threshold (or deductible) of $50,000 in a calendar year. Aggregate stoploss is used by insurers to limit their aggregate risk exposure over a number or group of specific risks. For example, an insurer may purchase stoploss coverage that would reimburse it for 90 percent of the claims costs in excess of a specified amount, such as 120 percent of the expected total claims costs for a group of risks (e.g., a large employer or an association of small employers). Other forms of reinsurance also may exist for health benefits. For example, a smaller insurer may reinsure a portion of the risk for an entire line (e.g., small group policies) or book (e.g., association group) of business. The reinsurer could receive a portion of the premiums and be responsible for a portion of the losses, either specifically or in aggregate. ⁴

Reinsurance plays important roles in these three markets. The availability of reinsurance permits primary insurers (including self-funded plans) to reduce their risk exposure in ways that can affect the availability and affordability of coverage. For example, the availability of reinsurance encourages more employers to create self-funded benefit arrangements, because through reinsurance they can protect themselves from the catastrophic costs associated with one or more very large claims. Self-funded employers also may be more reluctant to cover certain high-cost events, such as organ transplants, without the ability to reinsure. Reinsurers, by selling stoploss coverage to a number of health benefit programs, can create a relatively large pool over which to spread the risk of these catastrophic claim losses.

Reinsurance also plays an important role in protecting the financial solvency of insurance entities. Insurers generally can use bona fide reinsurance to reduce their minimum capital requirements. This can be particularly important for smaller and start-up entities. For example, a start-up HMO with $1 million in surplus could be ruined financially by one or two large claims during its early months of operations if it could not limit its claims exposure through reinsurance. Once the HMO has grown larger (e.g., 50,000 or more lives), reinsurance may be a less important aspect to its operations, because the HMO would be able to absorb the risk of catastrophic losses much more easily.

The potential role of reinsurance has been an important topic in several recent policy debates. For example:

• There have been questions about whether smaller employers who are claiming to be self-funded (to escape the effects of state insurance laws) actually are shifting a substantial portion of the risk of loss to the insurance market through reinsurance. The National Association of Insurance Commissioners (NAIC) recently developed a model regulation that attempts to define a minimum threshold for stoploss coverage (coverage with lower thresholds would be considered to be traditional health insurance), but the model has been controversial and at least one state court has found that a state law based on the model is preempted by ERISA.
• Reinsurance has been suggested to have a potentially important role in determining the appropriate minimum solvency standards for new or alternative insurance arrangements, such as provider sponsored organizations (PSOs) and multiple employer welfare arrangements.

Understanding better the roles that reinsurance currently plays in the market might assist policymakers as they address these and other important policy issues.

---

¹ The term "indemnity health insurer" here refers to organizations that are licensed as health insurers rather than as HMOs. Organizations licensed as health insurers often offer managed care products, including PPO coverage.

² Although it is often referred to as reinsurance, stoploss coverage purchased by self-funded employers is not reinsurance. Reinsurance technically refers to an insurance arrangement between a licensed insurance entity and another insurer. Employer stoploss coverage is more accurately referred to as excess of loss coverage.

³ Unless stated otherwise, we will consider self-funded health plans as insurers for the purposes of this discussion. We also will refer to excess of loss coverage purchased by such plans as reinsurance unless otherwise noted.

⁴ It should be noted that the existence of reinsurance typically does not eliminate the obligation of the primary insurer to pay a claim, even if its reinsurer fails to meet its obligations. Reinsurance generally is a contract between a primary risk taker and a reinsurer.

B. Methodology

The purpose of this study was to undertake a limited examination of the role of reinsurance in the three target markets. The study was essentially exploratory. In conducting the study, we attempted to determine what data was available about reinsurance in these markets, whether it was useful in measuring the level of risk borne by reinsurance, whether it was available for analysis, and, if the necessary data was not available, how it could be collected.

Our approach to the study included the following steps:

• Conducting a literature review to find information on the targeted reinsurance markets and potential sources of information on the amount of reinsurance in those markets.
• Polling researchers and people associated with the reinsurance and excess (i.e., stoploss) industries for assistance in identifying sources of information on the amount of reinsurance in the target markets.
• Acquiring and analyzing data from the sources we identified.
• Identifying key variables for assessing the level of risk borne by reinsurance in the three markets.⁵
• Interviewing participants in each of the three target markets to develop profiles of the role of reinsurance in those markets.
• Interviewing individuals associated with public and private surveys to determine whether additional information about reinsurance would be collected in future surveys.
• Developing potential strategies for collecting information about the level of reinsurance in each of the three markets.

We have conducted several literature reviews seeking information about the role of reinsurance in the health insurance market, both for this study and for a previous ASPE project. Our reviews included both academic and research journals and the insurance and employee benefits trade press. We also obtained information from private firms (such as Foster Higgins) that conduct their own surveys related to health insurance and health benefits.

In addition to the literature review, we polled individuals from associations and other groups that we believed would be likely to know about any existing research on the reinsurance industry. Most of these are well known, and others were identified by other organizations we contacted. We contacted the following organizations:

• A.M. Best
• American Academy of Actuaries
• American Association of Health Plans
• Association of Private Pensions and Welfare Plans
• Health Insurance Association of America
• National Association of Insurance Commissioners
• Reinsurance Association of America
• Society of Actuaries
• Self Insurance Institute of America

Several public and private sources of data on health insurance and health benefits were also contacted. When the project began, we anticipated that state insurance filings by insurers and HMOs would be a potential source of information for assessing the risk borne by reinsurance in at least the indemnity health insurance and HMO reinsurance markets. Our initial plan had been to visit three states to obtain information from a sample of the annual financial statements filed by insurers (including insurers offering reinsurance) and HMOs in the state. However, after discussions with representatives of the NAIC, we determined that we would be able to obtain from the NAIC electronic versions of the Annual Statement filings for indemnity (non-HMO) insurers in the entire U.S., obviating the need to collect the information from individual states. We obtained data from the reinsurance section (Schedule S) of the Life, Accident, and Health Annual Statements of all insurers who filed. Schedule S contains detailed information about reinsurance ceded and assumed by the filing insurer.⁶ The information in Schedule S for ceded insurance includes such things as the name and location of each carrier to whom an insurer has ceded risk, the paid and unpaid losses related to each contract, the premiums paid for each contract, and the type of the reinsurance contract (delineated along several variables, including whether the contract is group or individual, whether there is a coinsurance requirement and the type of coinsurance, and whether it is a yearly renewable term contract). Information for assumed insurance includes the name and location of each insurer that has ceded risk to the assuming insurer, the amount of reinsurance in force, the premium involved, and the type of reinsurance contract. Transactions between affiliated insurers are shown separately in the exhibits.⁷ The Schedule S instructions and a sample of each of the exhibits is attached as Attachment A.

Unlike data for indemnity insurers, HMO Annual Statement filings are not currently available through the NAIC.⁸ We identified two states that would locate and copy the relevant reinsurance information from the HMO annual statements: Wisconsin and Pennsylvania. HMOs also file quarterly financial statements, but the annual statement information is more complete. We collected information about all of the HMOs operating in those two states.

In addition to state regulators, we contacted several additional public and private sources, including the Bureau of Labor Statistics of the Department of Labor (regarding their Employee Benefits Survey); the Agency for Health Care Planning and Research (about the Medical Expenditure Panel Survey); the National Center for Health Statistics (about the National Employer Health Insurance Survey); the Rand Corporation (about surveys they have conducted and are conducting for the Robert Wood Johnson Foundation); KPMG and Gail Jensen (funded in part by the Kaiser Family Foundation); A. Foster Higgins (about their annual survey of employers); Hay Huggins (about periodic surveys of employers); and Tillinghast (about periodic surveys of employers and about reports that they had conducted a survey of reinsurers).

As the results that follow suggest, only a limited amount of the information we sought was available from the sources we contacted, which meant that we were unable to prepare an analysis about the amount of risk borne by reinsurance in the three target markets. For two of the markets, the indemnity insurance and self-funded employer markets, there is really no reliable data currently available that would provide an understanding of the terms of the reinsurance arrangements used by market participants. For the HMO reinsurance market, useful information is available in the annual financial statements filed by HMOs in the states in which they operate, but currently there is no central source for the information. Beginning in 1997, information from some 1996 state filings will be available from the NAIC. However, because states are currently not required to file with the NAIC, this information will most likely be incomplete. In place of an analysis of the amount of risk borne by reinsurance, we have attached samples of the information that we were able to collect and have included a discussion of its usefulness and limitations in the results section below.

The limited amount of available data also hampered our ability to identify key variables for either present or future analysis. Without good existing information, or a good sense of the type of information that might become available to researchers in the future (particularly for the indemnity insurance and employer stoploss markets), it is difficult to shape appropriate and useful analyses of the three target markets. However, we did identify some variables that we believe would be key to future data collection efforts if such efforts were attempted.

Due to the lack of information, we also were unable to develop profiles of the role of reinsurance in the three markets. Rather, in order to gain a conceptual understanding of the three reinsurance markets, we conducted interviews with reinsurers, brokers, and third party administrators. We identified reinsurers in the indemnity market by using the NAIC Annual Statement data to identify the reinsurers with the highest premiums. For the HMO market we selected the reinsurers with the highest frequency among the Annual Statement filings in Wisconsin and Pennsylvania. We gathered data on employer stoploss by interviewing the carriers identified in the other two markets, or others whom these respondents recommended. We used the information from these interviews to construct profiles of the three reinsurance markets.

In the next section, we discuss the important variables that would assist researchers, policymakers and others to understand the role played and the level of risk borne by reinsurance in each of the markets. The results of our data identification and collection efforts are discussed in the following section, separated by target market. We then present the profiles of the role of reinsurance in the three markets developed from interviews with market participants, and conclude with a discussion of potential strategies for collecting information about the level of reinsurance in each of the three markets.

---

⁵ We limited our study to a description of the structure and parameters of reinsurance arrangements (e.g., the level of coverage, whether the coverage is specific or aggregate). Because data is lacking, we do not explore how the variability of claims costs interacts with these structures and parameters.

⁶ Ceding is transferring risk to another insurer; assuming is accepting risk from another insurer. Ceding insurers are the buyers of reinsurance and assuming insurers are the sellers of reinsurance. Sometimes, an assuming insurer will transfer a portion of the assumed risk to another insurer. This is referred to as a retrocession.

⁷ Affiliated insurers are insurers that share a certain percentage of common ownership or control and include subsidiary arrangements.

⁸ They will be available electronically from the NAIC beginning in 1997 for calendar year 1996.

C. Variables for Understanding the Role of Reinsurance

In this section, we look at the types of information about reinsurance arrangements in the three markets that would help researchers and others better understand the roles that reinsurance plays and the level of risk that is transferred to reinsurers in each of the markets.

As discussed above, because of the lack of reliable information about reinsurance arrangements, particularly for the indemnity insurance and self-funded employer markets, our discussion of key variables for analysis must generally be conducted on a theoretical level. For the HMO reinsurance market, where some useful information is available, we can better relate our discussion of key variables to available data.

The type of information about reinsurance arrangements that we may wish to know for each of these markets will vary to some extent with the policy issues we are trying to address. For example, the scope of work for this project developed by ASPE staff suggested that the issues of interest for the indemnity market relate to market concentration and whether smaller insurers were acting principally as retailers or as primary risk takers. One of the issues of interest for the self-funded employer market relates to the use of stoploss coverage by smaller employers and the potential that stoploss is being used to evade state insurance laws.

Another factor that will be relevant to identifying key variables for understanding reinsurance roles in a market is the number of different types of reinsurance arrangements used in a market. If reinsurance is used for a variety of purposes in a market, identifying variables that elucidate and differentiate each of those roles could be problematic. This is a potential issue for the indemnity insurance market, where the use of reinsurance may be influenced by a number of different strategic or financial factors.

Below we identify key analytic variables for each of the three markets.

1. Indemnity Insurance

Specifying variables for the indemnity insurance market is the most difficult of the three markets we are considering for several reasons. First, how and why reinsurance is used by indemnity insurers is not well understood in the policy community, and the policy questions about its potential roles have not been fully formulated. Second, the potential roles for reinsurance in the indemnity insurance market are numerous, which makes identifying information that relates to the issues that policymakers may wish to understand more difficult. Reinsurance may potentially play a role in such things as market entrance strategies (e.g., limiting risk while developing experience in a new market), market exit strategies (e.g., leaving a market by transferring the risk of existing business to another carrier through assumption reinsurance⁹), financial management (e.g., purchase of specific or aggregate stoploss), marketing strategies, or for other strategic or financial reasons unique to the insurer.

The policy questions posed by ASPE relate to potential issues of market concentration; more specifically, whether there may be a pattern of a larger number of primary insurers writing coverage but using reinsurance to transfer a substantial portion of the risk of loss to a more limited number of insurers.¹⁰ We believe that the following variables would be the most important to understanding if these types of arrangements are prevalent in the market:

• Cessations by primary health insurers. The most basic information that is needed is the presence of a reinsurance arrangement. As discussed below, current information about health reinsurance does not necessarily distinguish reinsurance transactions by type of health insurance. Since the identified policy issues relate to basic health insurance (e.g., hospital and surgical coverage; major medical coverage) and not to supplemental or other types of health risks, information about reinsurance arrangements (including the information described in the variables below) should be limited to transactions related to primary health insurance.

• Structure of Transaction. Another factor that is important to understanding the role of reinsurance in this market is the structure of the transaction. Identifying if the reinsurance is structured as proportional (e.g., reinsurer receives a set proportion of the premiums, expenses and losses for each reinsured risk)¹¹ or excess of loss may be useful for the purposes of the stated policy issues. For example, an employer interested in truly catastrophic coverage is likely to purchase excess of loss coverage. However, an employer interested in passing most of the risk on to a reinsurer could use either an excess of loss arrangement with a very low threshold, or a proportional arrangement with the reinsurer receiving a large share of the premium and loss exposure for each covered group or individual. Given only information about structure, it may be reasonable to assume that proportional contracts transfer most of the risk, whereas excess of loss contracts may either transfer most of the risk or insure against catastrophic loss.

• Risk of Loss Transferred. In addition to the structure of the reinsurance transaction, it is important to know the magnitude of the risk that is being transferred. This can be a difficult issue for health insurance, because the insurance risk associated with a covered individual is concentrated at higher dollar levels and is not proportional to the percentage of premium.¹² Useful indicators would be the proportion of losses transferred in proportional arrangements and the specified thresholds for excess of loss arrangements (dollar thresholds for arrangements related to single lives and percentage thresholds for aggregate arrangements). In addition, it would be useful to know if the primary insurer retained any share of the losses (e.g., coinsurance) above any excess of loss thresholds.
• Unit of Reinsurance. Another important factor is the subject unit of the reinsurance transaction -- is the reinsurance relevant to the risk associated with a particular individual or some larger aggregation of lives. Relevant distinctions might include transactions related to: individual lives (and families); single employer groups; multiple employer groups; discrete blocks of business (e.g., all health insurance sold to groups with fewer than 50 lives or all health insurance sold through the contract with the local business coalition); and all health insurance of the primary insurer.
• Identification of Risk Assumers. To determine if there is significant market concentration occurring through reinsurance, it is important to know the identity of the insurers assuming the risk being transferred through reinsurance.

---

⁹ Assumption reinsurance is discussed in the next section.

¹⁰ We should note, as ASPE staff did in their scope of work, that there is nothing illegal or necessarily wrong with this type of arrangement. Policymakers, however, may want to know where risk is concentrated in the market as they assess policy options for potential market reforms.

¹¹ See "Glossary of Terms," Reinsurance Association of America (1994).

¹² See Robinson, J. "Stoploss Analysis." Memorandum. Coopers and Lybrand L.L.P. October 18, 1994.

2. HMO Reinsurance

HMO reinsurance raises a more discreet and clear cut set of policy issues: to what extent are HMOs using reinsurance to minimize risk, and how does reinsurance use vary across HMOs and by size of plan? As mentioned above, some useful information about HMO reinsurance arrangements already is available from state insurance filings, although, as discussed below, there appears to be some variability in how arrangements are reported. We believe that the following variables would be the most important to understanding if these types of arrangements are prevalent in the market:

• Cessations by HMOs. The most basic information that is needed is the presence of a reinsurance arrangement. Information about the size of the HMO (e.g., total HMO premiums, probably including premiums of affiliates) also should be available so that the reinsurance arrangements can be related to the size of the ceding HMOs.
• Scope of Risk. An important factor in understanding HMO reinsurance arrangements is understanding what type of risk is being covered. Our interviews with reinsurers indicate that most HMO reinsurance is limited to hospital risk, although some arrangements are broader. Differentiating the types of services reinsured, including hospital, and professional services, would be useful.
• Structure of Transaction. Another important factor is the structure of the transaction. Our interviews with reinsurers indicate that most HMO reinsurance appears to be excess of loss coverage (typically called stoploss coverage in this context), so the most important distinctions are between specific and aggregate stoploss arrangements. Public purchasers also may wish to know if the risk for public programs is transferred on the same bases as the risk for commercial coverages.
• Risk of Loss Transferred. In addition to distinguishing between specific and aggregate stoploss arrangements, it is important to know the magnitude of the risk that is being transferred. This could be determined by the excess of loss threshold levels (dollar thresholds for specific arrangements and percentage thresholds for aggregate arrangements). In addition, it would be useful to know if the HMO retained any share of the losses (e.g., coinsurance) above the stoploss thresholds.

3. Self-funded Employers

The issues associated with reinsurance¹³ by self-funded employers actually are quite similar to those for HMO reinsurance -- understanding how self-funded employers use reinsurance to minimize risk volatility and how that use varies by size of plan. Of particular interest in this market is the use of "excess of loss" coverage (typically called stoploss coverage in this context) by small employers. This interest is spurred in part by suggestions that small employers are being offered stoploss coverage with low threshold levels as a method of evading state insurance laws. With low threshold levels, a substantial portion of the insurance risk appears to be transferred to the stoploss insurer (making the insurance resemble traditional indemnity insurance), but the employer is able to claim that its plan is self-funded. Self-funding enables the employer to avoid state premium taxes and state-mandated benefits requirements, because ERISA exempts self-insured plans from such requirements.

We believe that the following variables would be the most important to understanding if these types of arrangements are prevalent in the market:

• Cessations by self-funded employers. The most basic information that is needed is the presence of a reinsurance arrangement. Information about the size of the employer (e.g., number of employees or number of covered lives) also should be available so that arrangements can be analyzed by size of employer.
• Structure of Transaction. An important factor is the structure of the coverage. Our interviews with brokers and reinsurers indicate that most employer reinsurance appears to be stoploss coverage, so the most important distinctions are between specific and aggregate stoploss arrangements. If possible, it also would be useful to understand the extent to which employers purchase "minimum premium plans." Minimum premium plans are arrangements in which employers retain the risk for a substantial portion, but not all, of the expected losses for a defined period. The risk above the retention point is insured by an insurer, which typically also provides the administrative functions to the employer. In some arrangements, the risk transfers back to the employer after a certain level of losses; in other arrangements all of the risk over the retention point is transferred to the insurer.
• Risk of Loss Transferred. In addition to distinguishing between specific and aggregate arrangements, it is important to know the magnitude of the risk that is being transferred. The relevant information is the stoploss threshold levels (dollar thresholds for arrangements for specific arrangements and percentage thresholds for aggregate arrangements). In addition, it would be useful to know if the employer retained any share of the losses (e.g., coinsurance) above any stoploss thresholds. Similar information would be useful in the case of minimum premium plans.

---

¹³ As discussed above, the term reinsurance is not technically correct as it relates to self-funded employers.

D. Results

In this section, we review the results of our efforts to locate information on the reinsurance arrangements in the three markets.

We generally were unable to identify either data sources or existing research that would enable us to quantify the risk being ceded to reinsurers in the three markets. The problems we encountered differed across the three markets. While a substantial amount of information is available about the reinsurance transactions of indemnity (non-HMO) health insurers, the data is aggregated in ways that make it difficult to specifically identify transactions related to primary health insurance offered to groups and individuals. The information available from state insurance filings about reinsurance purchased by HMOs is more useful in understanding the level of risk that HMOs cede to reinsurers, but lack of detail and inconsistencies in the way that HMOs report the information in their annual statements reduces its analytic potential. And, despite the substantial amount of information available about the prevalence of self-funding and the use of reinsurance by self-funded employers, there is virtually no reliable data about the level of risk being transferred to reinsurers in these arrangements.

These findings are presented in more detail below, separated by the target markets.

1. Indemnity Insurance

We were able to identify only one source of information on the amount of reinsurance purchased by indemnity insurers, which was the annual statements filed by licensed insurers with state insurance departments and with the NAIC. The annual statements are detailed financial disclosure documents that present a large amount of information about the financial and insurance transactions of licensed insurers. As discussed above, the reinsurance section of the annual statement (Schedule S) contains information that allows us to identify both companies that cede risk and companies that assume risk. The primary purpose of the annual statement is to permit state insurance regulators to assess the financial strength of insurers doing business in their states.

Although Schedule S contains a significant amount of data about the reinsurance transactions of indemnity insurers, the information is not collected or presented in a way that allows the types of transactions that we are interested in for this project to be isolated from other health-related reinsurance transactions. Conversations with NAIC staff confirmed that the transactions listed on the schedule could relate to many types of health coverages, including supplemental coverages (such as Medicare supplemental insurance and hospital indemnity coverage) or accident policies. In addition, our conversations confirmed that the schedule also includes transactions that relate to the sale of blocks of insurance coverage from one insurer to another. Often when an insurer transfers a piece of insurance business to another insurer, the purchasing insurers assumes the risk for the business through a reinsurance arrangement, called assumption reinsurance. The arrangement might be temporary (i.e., until new policies can be issued by the purchasing insurer to the policyholders) or it might last for a period of years.¹⁴ The potential inclusion of these other types of risks makes the Schedule S information unsuitable for our purposes.

If the targeting problem could be addressed, the reinsurance information shown in Schedule S could be useful in increasing our understanding of the role of reinsurance in this market. The schedule contains information on the amounts of reinsurance claims paid (both by ceding and assuming carriers) and on the premiums paid for reinsurance. However, information related to the structure of the reinsurance transactions is somewhat limited: coinsurance levels and deductible amounts that must be paid by the ceding insurer are not shown (although the presence of a coinsurance requirement is noted). An example of the type of information available in Schedule S is shown in Attachment A. To illustrate the magnitudes of reinsurance transactions shown on Schedule S, we compiled a list of the 50 reinsurers who reinsure the largest number of indemnity carriers, and their total premiums. This information is shown in Attachment B.

Suggested methods of collecting the information that would be needed to assess the role of reinsurance in the indemnity insurance market are discussed below in the section on Future Enhancements to Data Collection.

---

¹⁴ For example, under federal law related to Medicare supplemental policies, it is unlawful for an insurer to issue a policy that does not meet one of the ten standard policy designs developed by the NAIC. However, there are many Medicare supplemental policies in force that were issued prior to the effective date of the new standards. If an insurer wished to sell its old policies to a new insurer, or if a group wanted to move existing group policies to a new insurer, the new insurer could not reissue the old policy forms under its name, because it could not issue policies that do not conform to the new standards. However, the business could be transferred to a new insurer through an assumption reinsurance arrangement, under which the new insurer would assume 100 percent of the liability associated with the policies issued by the original insurer. The new insurer could assume the administration of the policies as well (although the original insurer would retain the ultimate financial responsibility if the new insurer was unwilling or unable to meet its responsibilities under the reinsurance arrangement).

2. HMO Reinsurance

The HMO Annual Statements provide some information about the range of types of reinsurance arrangements into which HMOs enter. Tables C-1 and C-2 in Attachment C present selected information from these statements for the HMOs in Pennsylvania and Wisconsin, respectively. Unfortunately, these HMOs do no report the details of their reinsurance arrangements consistently. Some report only a deductible amount, others only a coinsurance percentage, and still others only an annual or lifetime maximum. Many have a combination of these. While these inconsistencies make it difficult to interpret the information that is provided, we can make several observations.

Generally, the range of arrangements in Pennsylvania and Wisconsin are similar. Specific deductibles (sometimes called stoploss thresholds) range from $50,000 to $250,000in Pennsylvania and from $25,000 to $200,000 in Wisconsin.¹⁵ (One HMO in Pennsylvania reports a deductible of $1 million). A clear relationship between total earned premiums and deductible size does not appear to exist in Pennsylvania, although the two reinsuring HMOs with the largest premiums (above $400 million) do have the largest deductibles ($250,000). In Wisconsin, deductibles and premiums appear to follow a more clear pattern. HMOs with higher total earned premiums (above $60 million) tend to have higher deductibles ($100,000 to $200,000), while HMOs with lower earned premiums tend to have lower deductibles ($50,000 to $85,000).

Fewer than one third of Pennsylvania HMOs and one quarter of Wisconsin HMOs report copayment percentages in their Annual Statements. Among reporting HMOs in both states, the amount the reinsurer pays after the deductible is reached ranges from 50 to 90 percent, usually closer to the latter. It is unclear whether the HMOs that do not report copayment percentages have 100 percent of their claims paid after reaching their deductible, or merely omit this information.

Just under half of the HMOs in both states report either annual or lifetime limits on coverage. Annual limits are more common than lifetime limits in Wisconsin, but both occur with equal frequency in Pennsylvania. The annual or lifetime limit is most often set at $2 million, and less frequently at $1 million.

Several different arrangements exist in Wisconsin. One medium-sized HMO has aggregate coverage in addition to its specific coverage, with an aggregate deductible of $750,000. One reports coverage for physician services as well as hospitalization. In this case, the deductible is lower for the physician coverage ($15,000, versus $35,000 for hospitalization). One reports coverage only for certain transplant procedures, while another reports a separate (lower) coverage percentage for transplants in non-approved hospitals. Finally, one has a lower deductible for Medicare risk utilization than for commercial and Medicaid utilization.

All HMOs in Wisconsin have some type of reinsurance, while the two largest HMOs in Pennsylvania (U.S. Health Care and the Keystone Health Plan East) are self-insured. These two HMOs have earned premiums of more than $1 million.

We located one analysis looking at HMO reinsurance arrangements.¹⁶ The analysis suggested that the size of the deductible is related to the HMO's financial strength, and generally correlates with the size of the HMO. Figure 1 shows the relationship between the deductible level and the number of members in an HMO. Deductibles under $50,000 are most common among HMOs with fewer than 10,000 members, while deductibles in excess of $75,000 are most common among HMOs with more than 50,000 members.

HMO by Size of Specific Deductible

Source: Jack LaMar, Pricing and Underwriting Stoploss: "HMO Reinsurance." Presented at the Society of Actuaries Spring Meeting, 1996.

---

¹⁵ Based on our interviews with reinsurers, we assume that the deductibles reported are specific deductibles unless otherwise labeled, and that they apply only to hospitalization unless otherwise noted.

¹⁶ Jack LaMar, Pricing and Underwriting Stoploss: "HMO Reinsurance." Presented at the Society of Actuaries Spring Meeting, 1996.

3. Self-funded Employers

There are a number of surveys that collect basic information about self-funded health plans, including the proportion that purchase some form of insurance to cover a portion of plan losses. Arrangements that use insurance to limit plan losses are sometimes referred to as partially self-insured plans or minimum premium plans depending on the survey (we will use the term stoploss insurance in this section). We contacted a number of agencies and firms that conduct annual or periodic surveys of employee benefits, including:

• Bureau of Labor Statistics, Department of Labor
• Tillinghast
• Foster Higgins
• Hay Huggins
• KPMG

We found that while several of these surveys collect (or used to collect) information about the presence of individual or aggregate stoploss arrangements, none of the surveys collected additional detail about the terms of the insurance, such as the stoploss threshold or coinsurance amounts. Representatives of the Bureau of Labor Statistics told us that they previously attempted to collect information on whether or not an arrangement was a minimum premium plan (which they defined as the presence of some insurance), but they ceased doing so because they believed that the information that they were collecting was not reliable, particularly from smaller employers. Information about the presence of individual and aggregate stoploss coverage from the surveys conducted by Foster Higgins and KPMG is displayed in Attachment D.

Although none of the annual or periodic employer surveys collected detailed information on employer-purchased stoploss arrangements, three data sources we identified collected (or are collecting) some additional details about such arrangements. The first is a survey conducted in 1996 by KPMG and Gail Jensen, which collected information about the stoploss thresholds of stoploss insurance purchased by a sample of employers with self-funded health benefit plans. The survey also collected information about the presence of a minimum premium plan, but no specific information about the risk allocation under the plan. The second is the National Employer Health Insurance Survey (NEHIS), conducted by three agencies in the Department of Health and Human Services (the Agency for Health Care Policy and Research, the National Center for Health Statistics, and the Health Care Financing Administration), which collected information on the premiums paid by self-funded employers for stoploss coverage. The third is a survey that is currently in the field, sponsored by the Robert Wood Johnson Foundation and being conducted by the Rand Corporation, which, like the KPMG/Jensen survey, is asking about the stoploss threshold of stoploss insurance purchased by large and small employers with self-funded health benefit plans.

Information compiled for the Lewin Group by KPMG and Jensen on the stoploss arrangements and similar arrangements reported by employers in their survey is shown in Tables E1 to E6 in Attachment E. These tables show the percentage of employers (by firm size) reporting to have partial self-insurance or other risk arrangements, the distribution of those other risk arrangements across three categories (minimum premium plans, reinsurance or stoploss, and other), the percentage of smaller firms (less than 200 employees) with stoploss arrangements, and the distribution of those stoploss arrangements by threshold level.

While the type of information contained in the KPMG/Jensen survey would help policymakers begin to understand the role of reinsurance, particularly in the small employer market, the extremely small percentage of the small employers in the sample reporting the presence of stoploss coverage (less than 1 percent of employers covering less than 4 percent of workers) means that these numbers should be considered with a great deal of caution.¹⁷

The second employer survey that collects some detailed information on employer-purchased stoploss coverage is NEHIS. As part of its efforts to collect information on health insurance costs, NEHIS asked employers to report the premiums paid for stoploss coverage, although no other information was collected about the terms of the stoploss coverage.

Although the lack of information on stoploss coverage terms will limit the scope of the analysis of reinsurance that can be made with the NEHIS data, the stoploss premium information still may permit some general inferences to be made about the level of risk being transferred to stoploss insurers by employers. Such an inference would rest on assumptions about the general types of stoploss coverage available in different markets and its costs (varied to the extent possible by employer size, geography and other relevant characteristics). The exercise would involve estimating the amount of stoploss coverage employers could purchase, given the premiums that they reported paying and assuming that they purchased the types of coverage typically available to similar employers. While the resulting estimates of the amount of reinsurance purchased would be subject to a great many caveats, they would mark an improvement over what is known today and may serve as a starting point for other research efforts.

Data from NEHIS should be available sometime in the Spring of 1997. One concern about the data is that conversations with representatives of the National Center for Health Statistics (one of the agencies compiling NEHIS results) indicated that the response rate for the questions related to stoploss premiums may be relatively poor. This would diminish the survey's usefulness for analyzing the level of stoploss coverage purchased by employers and may make any analysis based on the responses suspect.

The second employer survey that collects some detailed information on employer-purchased stoploss coverage is being conducted by the Rand Corporation and the Research Triangle Institute for the Robert Wood Johnson Foundation. The Robert Wood Johnson Foundation Employer Health Insurance Survey asks employers with self-funded plans whether they purchase specific or aggregate stoploss coverage and, if so, what the coverage thresholds for the stoploss coverage are. The survey was still in the field at the time this report was completed, so we do not know what types of responses the survey is getting and what the response rate is. The principal advantage of this survey is that it has a large sample of employers, including small employers, in each of the states in which the survey is being conducted. Another advantage is that the wording of the questions does not assume much knowledge about stoploss on the part of the employer. While some questions about the structure of stoploss arrangements would remain, such as the extent to which employers retain any share of the risk above aggregate or specific stoploss thresholds (e.g., retain a coinsurance obligation), if the survey succeeds in getting good responses, it has the potential of providing substantially more information than we have today about the way that small and large employers use stoploss coverage and the level of risk that is being assumed in the reinsurance market.¹⁸

---

¹⁷ The large percentage of small employers with conventional plans that did not know their threshold (almost one-third) also raises a note of caution.

¹⁸ The survey also asks whether the employer has a minimum premium plan, but does not ask any detailed questions about such arrangements (as it does of those with stoploss policies), limiting the usefulness of the data somewhat.

E. Market Profiles

Because information on all three markets we investigated is limited, we supplemented our data collection with interviews with reinsurers, brokers, and third party administrators (TPAs) familiar with these markets. These interviews provide a conceptual understanding of the three markets.

1. The Indemnity Market

Insurers purchase reinsurance for other insurers for a number of reasons, so this market is more difficult to characterize than the employer or HMO stoploss markets. Informants suggested that health insurers that purchased reinsurance tended to be smaller, regional carriers. The business reinsured tends to be individual or small group health insurance. Typical arrangements include specific stoploss (one informant estimated that 75 to 80 percent of this business was specific stoploss), aggregate stoploss for a book of business (e.g., association small group business), and quota share arrangements. In a quota share arrangement, the primary insurer retains a set percentage of the losses and the premium and transfers the remainder to the reinsurer. The primary insurer uses reinsurance to find a level of risk exposure that is commensurate with its size, business strategy and available resources. In some cases, reinsurers provide expertise or services to the primary insurer, including case management on severe claims. One informant indicated that they ceded a portion of the reinsurance risk they assumed to other reinsurers (called a retrocession). He suggested that a portion of reinsured risk is ultimately ceded to a few large reinsurers.

The market is currently very active. A survey by CIGNA estimated 13 or 14 major reinsurers in the market, and one reinsurer estimated roughly 300 to 400 insurers ceding risk. Competition in the market is high, and one reinsurer noted that no reinsurers have achieved much profit. Premiums, however, have increased recently, up to five or six percent this year from two or three percent last year.

2. The HMO Reinsurance Markets

The existence of HMOs has produced two separate markets for reinsurance, one for reinsurance sold to HMOs, and another for reinsurance sold to the provider groups capitated by HMOs. We refer to the former as "HMO reinsurance," and the latter as "provider excess coverage." We discuss these two markets separately.

a. HMO Reinsurance

Our informants suggested that HMO reinsurance is a product that is primarily used by smaller, less well capitalized HMOs, including start-up plans. HMO reinsurance is usually written on a specific stoploss basis and primarily is for hospital claims.¹⁹ The stoploss retentions for start-up and smaller HMOs may be as low as $35,000 to $50,000 per enrollee, but rise as HMOs increase in size. HMOs with 10,000 to 50,000 members are likely to have stoploss thresholds of around $100,000 per enrollee; very large HMOs are likely to have very high thresholds (e.g., $250,000 and up per enrollee) or no coverage at all. Informants estimated the average stoploss threshold to be roughly $125,000.

The stoploss coverage usually is subject to a coinsurance requirement, with the HMO responsible for 10 to 20 percent of the per enrollee costs above the threshold. One large HMO reinsurer reported that the stoploss coverage payments to HMOs for high cost cases are based on the actual payment scheme used by the HMO to reimburse hospital costs. Out-of-area events are subject to a maximum daily average cost.

The price for stoploss coverage varies by a number of factors, including the demographics of the members, whether members are covered by Medicare, Medicaid, or commercial insurance, geographic location, members' benefits, hospital contracts, network coverage for tertiary care, and the target market of the HMO. The stoploss premium for Medicare enrollees may be twice as much as the premium for commercial enrollees.

The market for HMO reinsurance had remained stable for a many years, until two to three years ago, when a number of new firms entered the market and pricing became very competitive. Our informants characterized the current market as driven largely by reinsurers who were "buying business," that is, selling reinsurance at a loss in order to gain market share. The market is stabilizing again, and informants estimated that six to eight reinsurers will remain.

Informants stated that HMO reinsurance is a difficult market to make a profit in because HMOs have become very sophisticated purchasers of reinsurance. Large HMOs with a good understanding of their claims risk are able to play the market, and only choose to purchase reinsurance when they calculate it to be cheaper than self-insuring. As capitation increases, HMOs are less likely to purchase reinsurance coverage (although the capitated provider groups purchase provider excess coverage, discussed below).

---

¹⁹ State required coverages for continuation benefits and insolvency coverage usually are wrapped into the stoploss arrangements.

b. Provider Excess

Many HMOs capitate the hospitals and physician groups with whom they contract, effectively transferring the risk onto these providers. These provider groups in turn seek reinsurance, or "provider excess" coverage, to guard against inordinately high losses. This coverage is almost always specific coverage; aggregate coverage is extremely rare. Providers purchase this coverage in one of two ways. Some provider groups purchase provider excess coverage through the HMOs with whom they contract. In this case, a reinsurer may price the coverage for the HMO, which includes it as a package in its contract with the provider group. Other provider groups purchase the coverage directly from a reinsurer. A provider group that contracts with several HMOs may find that purchasing reinsurance from a single reinsurer is more efficient than purchasing it from each HMO separately.

Informants stated that provider groups tend to be thinly capitalized and want to purchase a significant level of protection. Informants suggested that a number of provider groups are still relatively unsophisticated and sometimes reinsure too much of their risk. On the physician side, one informant reported stoploss thresholds in the range of $7,500 to $12,000 per enrollee, although he had seen thresholds as low as $5,000. The informant reported a broader range for hospitals -- from a low of $35,000 per enrollee to as much as $100,000 to $125,000 for commercial risk. The stoploss threshold for Medicare risks might be much higher.

The market has been marked by extreme competition over the past five to seven years. An influx of new competitors offering large discounts over existing provider excess premiums brought prices down to artificially low levels. Informants estimated that prices may be as much as 50 percent too low, caused by a number of factors, including aggressive pricing to build market share, a large number of competitors, and lack of sophistication by reinsurers of appropriate prices for a new market. Informants reported that the market is stabilizing and that prices are rising. One informant suggested that provider groups were unaware of the low prices and would be unwilling to pay a greater percentage of their capitations for stoploss. He indicated that prices were unlikely to rise dramatically, but that the terms of coverage would become less favorable to providers (i.e., they would have to retain much more risk).

Informants estimated that there were about 20 sellers of provider excess coverage. One informant estimated that only one company was profitable in the prior year in this business. The largest carrier left the market; its competitors estimated that it was running at a loss ratio of over 200 percent (i.e., claims were twice premiums).

3. Self-funded Employers

Employers offering self-funded health benefit plans to their employees often purchase stoploss coverage to limit their loss exposure. Informants stated that the employer stoploss market was primarily aimed at employers with more than 50 employees, although some carriers were selling stoploss arrangements to employers smaller than that. As employers get larger, their need for stoploss decreases; informants indicated that self-funded employers with over 5,000 employees rarely purchased stoploss coverage. The decision to buy stoploss is also related to the cash flow of the company and the length of time the employer has been self-insured. Well financed employers and employers who have experience with self-insurance are less likely to purchase stoploss.

Employer stoploss coverage is provided on both a specific and aggregate basis. Specific stoploss limits the employer's exposure for the costs associated with each covered life. Aggregate stoploss limits the employer's total exposure as a percentage of some expected amount of claims. Smaller self-funded firms are likely to purchase both specific and aggregate coverage, whereas larger self-funded firms are more likely to just have specific coverage. One informant estimated that among employers with 1,000 lives, all have specific coverage but only about half buy aggregate coverage. Firms with more than 1,000 lives tend to purchase only specific coverage. Aggregate coverage is sometimes "thrown-in" for larger employers, for whom aggregate coverage is not as important because their claims are more predictable. Aggregate-only policies (without specific coverage) appear not to exist.

Stoploss thresholds, or deductibles, (both aggregate and specific) are computed based on a number of factors, including the demographics of the group and the type of coverage. Informants suggested that most employers absorb for every covered life approximately 10 percent of expected annualized aggregate claims. In other words, the specific deductible is set at roughly 10 percent of aggregate expected claims.²⁰ For example, an employer covering 50 lives with an expected annual cost per life of $2,000 would have $100,000 in expected claims over the year. The specific deductible for this employer would be set at approximately 10 percent of $100,000, or $10,000 per employee.²¹

Aggregate stoploss thresholds also vary by the size of the firm. Smaller companies (e.g.., 50-100 lives) usually have aggregate stoploss limits of 120 to 125 percent of expected claims. Larger companies have lower stoploss limits (because their claims are more predictable), more often between 115 to 120 percent of expected claims. Very large firms, if they purchase aggregate coverage, may have a threshold of 110 percent of expected claims.

Informants estimated that there were about 150 to 200 providers of employer stoploss coverage.²² Employers usually purchase stoploss either from third party administrators (TPAs) or from brokers. One informant noted that in California and Chicago, brokers tend to arrange coverage, while in the majority of the country, TPAs drive the market. Like a broker, a TPA will contact several insurers in an effort to get the best price. This is because a TPA that processes an employer's claims has a fiduciary relationship with the employer, and must demonstrate due diligence as well as compete with other TPAs.

---

²⁰ One TPA that deals exclusively with small employers provided the rule of thumb that the specific deductible is set at 10 percent of the aggregate stoploss threshold, which is somewhat higher than expected claims.

²¹ This is a general rule of thumb, and is not as applicable to large employers, for which the specific deductible would almost certainly be smaller than ten percent of expected claims. One informant suggested that the specific deductible usually ranges from two to twelve percent of aggregate expected claims.

²² One informant suggested a much higher number (over 300), but several others estimated between 150 and 200.

F. Future Enhancements to Data Collection

Given the lack of data about reinsurance arrangements in two of the markets (i.e., indemnity and self-funded employer markets) and some of the limits on data availability in the HMO reinsurance market, the question becomes what options are available to researchers and policymakers for developing better data about the role of reinsurance in the three markets. In this section, we discuss how additional information could be obtained through existing data sources. We do not address all of the issues associated with developing a new set of surveys of insurers or employers to gather this information because the costs of creating a separate survey for this type of information appear large in relation to the potential benefit.

In considering the options for collecting reinsurance information discussed below, several criteria are important:

• Reliability. Can the type of information that is being sought be reliably collected from the data source. Reinsurance transactions can be complex, and it is important that the person providing the information have sufficient familiarity with the reinsurance transactions of the surveyed entity to be able to address detailed questions related to thresholds, retentions and unit of risk. For example, individuals at BLS told us that they had ceased collecting information on minimum premium plans because the individuals that they collected information from at employer establishments were not familiar with these arrangements.
• Cost. A second important consideration is whether seeking additional information about reinsurance as part of existing data collection efforts would substantially increase the costs of those efforts. Costs could be increased in several ways, such as lengthening interview times and increasing data cleaning and processing costs.
• Response burden. A third important consideration is the effect that seeking additional information on reinsurance would have on the burden of respondents, both in terms of time and costs. Increasing response burden can discourage participation in data collection efforts and lower response rates. It also imposes financial costs on respondents.
• Purpose of the existing data collection effort. A fourth consideration is whether attempting to gather additional data on reinsurance is consistent with the purpose of the existing data collection activity. Attempting to gather descriptive information through a data collection effort that has another primary purpose (e.g., as a regulatory tool) may be objectionable to the regulated industry and to the regulators (unless the information furthers some regulatory purpose).

1. Indemnity Market

The most logical data source through which to gather additional data on reinsurance arrangements by indemnity insurers is the NAIC annual statement, which is discussed above. The annual statement collects detailed information annually from every licensed insurer, so it could be a source of information from ceding insurers and insurers assuming reinsured risk. As discussed above, detailed information about individual insurance transactions of ceding and assuming insurers already is collected on Schedule S of the annual statement, so it would be a logical place to request some of the additional information identified above in the section on key variables. In fact, it is difficult to imagine a more efficient method of collecting this information from ceding and assuming insurers.

In terms of reliability, cost, and even response burden, collecting the information through the NAIC annual statement makes a great deal of sense. The statement is completed by and submitted to individuals with expertise in the area, so reliability is likely to be quite good. And given the amount of information currently collected in the annual statement on reinsurance, the marginal effects on costs and response burden of requesting additional reinsurance information are relatively small. The response rate will not be effected because the filing of the annual statement is mandatory.

A potential objection to the use of the annual statement to collect this information, however, is that it does not serve a regulatory purpose. The NAIC or the insurance industry could object to the use of the annual statement to collect purely descriptive information, in part because this might be seen as the first in a long line of general or specific questions about the insurance industry that could be answered by adding a question to the statement. Unless regulators could be convinced that knowing more about the reinsurance arrangements of indemnity insurers was relevant to the insurers solvency or to their oversight of the health insurance market, they may be reluctant to add additional reinsurance questions to the statement.

There are no other good options for collecting detailed information on the reinsurance arrangements of indemnity insurers. There are a number of private companies that collect and distribute information about insurers to the general public or the investment community, including A.M. Best, Moodys, and Weiss Research. However, the size and purpose of these data collection efforts are not directed at answering questions of public policy, making the data largely inadequate for our purposes.

2. HMO Reinsurance

Most of the discussion above related to gathering additional reinsurance data for the indemnity market is equally applicable to securing additional data about HMO reinsurance. The NAIC HMO annual statement is the best current source of detailed information on HMO reinsurance arrangements and the most logical source of additional information on the topic. The additional details that we described in the section above on variables for understanding reinsurance arrangements are actually only minor additions and clarifications to the type of information that many HMOs provide now in the annual statement. In fact, the clarifications related to the type of services being reinsured and the presence of any retention of risk about the stoploss thresholds would actually improvethe usefulness of the information for regulators, who would be in a better position to evaluate the arrangements' effects on solvency. This may reduce potential opposition from the NAIC or the HMO industry to making such changes.

If additional information cannot be gathered through the NAIC HMO annual statement, the most likely other sources of information are the private companies and associations that collect and disseminate information on HMOs. Interstudy is one such company; the American Association of Health Plans also collects and disseminates information about HMOs gathered through an annual survey of its member companies. As with indemnity insurance, however, the rather narrow demand for this type of information may discourage these sources from collecting and disseminating the information.

3. Self-funded Employers

Unlike the information about reinsurance arrangements by indemnity insures and HMOs, which is best collected from insurers, employers are the most logical source of information about self-funded employer stoploss arrangements and minimum premium plans, although the insurance industry is a potential source of information if reliable data cannot be collected from employers.

One advantage of collecting this information through employers is that comparisons can be made between employers that use stoploss and those that do not. There also are a number of existing surveys that gather information from employers about their health benefit arrangements (including if they are self-funded), providing a number of potential avenues for securing additional details about excess of loss arrangements in this market.

One option for collecting the additional information is the Employee Benefits Survey (EBS) conducted by the Bureau of Labor Statistics. The EBS is designed mainly to collect information on the benefits that employers provide to their employees. In the past, BLS has attempted to collect some information on stoploss arrangements (they use the term minimum premium plans for any split risk arrangement), but they ceased doing so because of concerns about data reliability. BLS is currently in the process of a series of major survey redesigns, in which three major surveys (the EBS, along with the Employment Cost Index and the Occupational Compensation Survey Program) will be combined into a single survey known as Comp2000. BLS staff are in the process of selecting items to field test for inclusion in Comp2000, and have indicated they may be willing to test items related to stoploss.

Reliability and response burdens are likely to be the chief concerns with any attempt to collect detailed information on stoploss coverage or minimum premium plans through employers. To collect the information, the respondent, or the data collection analyst conducting the survey in the field, would likely need to consult records of the stoploss or minimum premium transaction. For reliability, the surveyor may need to confirm the information with additional staff at the respondent firm or with an insurer or insurance broker. BLS officials raised the concern that this could put a substantial burden on survey respondents and also could significantly increase survey costs.

If information cannot be collected from employers, another option would be to attempt to collect some of it from insurers. The NAIC annual statement could be amended to have insurers provide information on the aggregate level of stoploss and minimum premium plan coverage and the number of employers covered. Collecting details through the NAIC reporting form about the risk allocation in these arrangements would not seem to make sense because there are far too many of them, and presenting the information in this fashion would be a large burden on insurers. All of the potential objections discussed above to using the NAIC annual statements to collect this type of information are applicable here.

G. Conclusion

Although reinsurance is widely used by the underwriters of health benefits (i.e., insured and self-funded plans), its role is little understood by policymakers. Reinsurance plays important roles in the health care markets that we considered. Primary insurers (including self-funded plans) reduce their risk exposure in ways that can effect the availability and affordability of coverage. Reinsurance also plays an important role in how primary insurers and HMOs assure their financial stability and meet minimum capital requirements imposed by states.

The potential role of reinsurance has been an important topic in several recent policy debates. State insurance regulators have raised concerns that certain stoploss arrangements are being used to help small employer health plans avoid state insurance reforms. Reinsurance also has been suggested as a potential vehicle to assist provider sponsored health plans and multiple employer benefit arrangements demonstrate financial adequacy.

We undertook a limited analytic study of the role of reinsurance in three insurance markets: primary health insurance coverage offered by indemnity insurance companies; health insurance coverage offered by health maintenance organizations; and health benefits coverage offered by employer-sponsored self-insured plans. Our review of the potential sources of information concluded that there are no existing data sources or existing research that would enable us to quantify the risk being ceded to reinsurers in the three markets. The problems we encountered differed across the three markets. While a substantial amount of information is available about the reinsurance transactions of indemnity (non-HMO) health insurers, the data is aggregated in ways that make it difficult to specifically identify transactions related to primary health insurance offered to groups and individuals. The information available from state insurance filings about reinsurance purchased by HMOs is more useful in understanding the level of risk that HMOs cede to reinsurers, but lack of detail and inconsistencies in the way that HMOs report the information in their annual statements reduces its analytic potential. In addition, the information currently must be collected separately from each state, but it will be available from a central source (the NAIC) in the near future. Finally, despite the substantial amount of information available about the prevalence of self-funding and the use of reinsurance by self-funded employers, there is virtually no reliable data about the level of risk being transferred to reinsurers in these arrangements.

If new efforts were to be made to collect information about reinsurance, the most logical method of collecting information in the indemnity insurance and HMO reinsurance markets would be through the annual statements collected from insurers and HMOs by the NAIC. These statements include detailed annual information from every licensed insurer and HMO, and have substantial benefits in terms of reliability, cost, and response burden. However, concerns may be raised by regulators or insurers about using the annual statement to collect information that does not serve a regulatory purpose. The annual statements are regulatory tools and using them for research-related data collection could be seen as undermining their primary purpose over time.

Employer surveys would be the most appropriate source for collecting additional information about self-funded employer stoploss arrangements and minimum premium plans. There are a number of existing surveys that could be expanded to collect this information, including the EBS. BLS is currently in the process of redesigning their employer surveys and are in the process of selecting items to field test for inclusion. However, adding detailed questions about stoploss insurance to employer surveys could raise significant problems in the areas of reliability and response burden. Collecting this information may require surveyors to review records of the stoploss or minimum premium transaction or to confirm information with additional staff at the respondent firm or with an insurer or insurance broker. This could add substantially to survey costs. At the current time, a survey being conducted by the Rand Corporation is attempting to collect some detailed information about stoploss arrangements and the presence of minimum premium plans. The experience of that survey effort should provide an indication of the difficulties of collecting this type of information through employer surveys.

Attachments

Attachment A

Schedule S Instructions and Sample Exhibits

(This part of the report consisted of a photocopies of forms and instructions which were not suitable for conversion to html.

Attachment B

Reinsurers Reinsuring the Largest Number of Indemnity Carriers

Table B-1

Source: The Lewin Group tabulations of Annual Statement data from NAIC.

Attachment C

HMO Annual Statement Data

Table C -1 Selected Reinsurance Data from Pennsylvania HMO Filings, Sorted by Total Premiums

No.	Company	Reinsurance Expense Net Recoveries	Total Premiums Earned	Name of Reinsurer	Limits of Coverage	Exp Date	Comments
1	U.S. Health Care Systems of PA	0	1,312,067,811	None (self-insured)
2	The Keystone Health Plan East	0	1,045,082,645	Self Insured
3	The Keystone Health Plan West	(588,398)	455,833,612	Trans-General Life Insurance Co	$250,000 deductible	07/01/96
4	Health America Pennsylvania	2,176,846	414,800,909	American Service Life	80% of inpatient services in excess of $250,000	12/31/95
5	The Geisinger Health Plan	83,951	216,185,913	Allianz	$2 million maximum lifetime	06/30/96
6	Qualmed Plans for Health of PA	(3,228)	215,065,648	Allianz Life Ins Co. of North Am	90% in excess of $100,000 up to $2 million lifetime	06/30/96
7	The Healthcare Management Alternatives Inc.	1,830,000	212,229,990	Continental Assurance Co	$23 million in aggregate	01/31/97
8	Health Partners of Philadelphia, Inc.	498,366	207,422,231	John Alden Life Ins Co.	Reinsurance for 90% of defined costs in excess of $150K and $125K, respectively in '95 and '94, per MA member per policy year. Hospitalization costs subject to recovery were $1,500.8K and $2,231.2K in '95 and '94, respectively.	(Not Specified)
9	The Keystone Health Plan Central	523,051	173,171,138	Trans-General Life Insurance Co	Recoveries limited to 70% of amounts in excess of $250,000 stop loss	12/31/96
10	The HMO of Northeastern PA	326,009	142,615,982	Allianz Life Insurance Co	$2 million	Continuous until cancelled
11	Aetna Health Plans of Central and Eastern PA, Inc.	(196,698)	114,363,354	Aetna Life Ins Co.	Unlimited	Indefinite
12	The Oxford Health Plans (PA)	(219,169)	99,847,599	Allianz Life Ins Co. of North Am	$150,000 stoploss limit, per member per year, lifetime limit of $2 million	07/31/95
13	The Prudential Health Care Plan (dba Prucare)	0	56,155,292	The Prudential Insurance Co of America	$1 million per person deductible	None	Unclear why no reinsurance expense
14	Amerihealth HMO, Inc.	280,104	53,656,601	Allianz Life Ins Co. of North Am	$75,000 deductible, $2 million per member lifetime	07/01/96
15	HealthGaurd of Lancaster, Inc.	(156,414)	46,249,965	Allianz Life Ins Co	$100,000 deductible, $2 million lifetime	12/31/96
16	Aetna Health Plans of Western PA, Inc.	54,149	27,659,054	Aetna Life Ins Co.	Unlimited	03/31/96
17	Central Medical Health Plan, Inc.	155,967	21,936,814	AON Managed Care/Combined Insurance	$125,000 deductible and [$99,000 retension estimate]	09/30/96
18	Alliance Health Network	(65,279)	21,720,860	Allianz Life Ins Co.	$50,000 deductible per occurrence on hosp I/P claims, & 85% of excess	04/30/96
19	Cigna Health Care of PA	23,854	4,891,421	Connecticut General Life	$150,000 deductible; amounts thereafter reimbursed at 80%	Continuous
20	Greater Atlantic Health Plans, Inc.	0	0	QualMed Health and Life	$1 million per contract year, $2 million per lifetime	06/30/96	"No membership"
21	HIP of PA, Inc (d/b/a HIP Health Plan)	0	0	Allianz Life Insurance Co of North America	2 policies starting with a $50K deductible to $180K and $185K deductible to a maximum individual coverage of $2M. There are varying percentages of coinsurance.	07/01/96
22	The Medigroup HMO, Inc.	0	0	BCS Insurance Co	$2 million per member	06/01/96	(Apparently no membership)
23	The Riverside Health Plan, Inc.	0	0	American Service Life	50%-90% dependent on contracted services	12/31/95	Apparently no membership)
24	The Vista Health Plan	0	0	None			"No subscribers"

Source: The Lewin Group tabulations of HMO Annual Statement data Pennsylvania.

Table C-2 Selected Reinsurance Data from Wisconsin HMO Filings, Sorted by Premiums

No.	Company	Reinsurance Expense Net Recoveries	Total Premiums Earned	Name of Reinsurer	Limits of Coverage	Exp Date	Comments
1	U.S. Health Care Systems of PA	0	1,312,067,811	None (self-insured)
2	The Keystone Health Plan East	0	1,045,082,645	Self Insured
3	The Keystone Health Plan West	(588,398)	455,833,612	Trans-General Life Insurance Co	$250,000 deductible	07/01/96
4	Health America Pennsylvania	2,176,846	414,800,909	American Service Life	80% of inpatient services in excess of $250,000	12/31/95
5	The Geisinger Health Plan	83,951	216,185,913	Allianz	$2 million maximum lifetime	06/30/96
6	Qualmed Plans for Health of PA	(3,228)	215,065,648	Allianz Life Ins Co. of North Am	90% in excess of $100,000 up to $2 million lifetime	06/30/96
7	The Healthcare Management Alternatives Inc.	1,830,000	212,229,990	Continental Assurance Co	$23 million in aggregate	01/31/97
8	Health Partners of Philadelphia, Inc.	498,366	207,422,231	John Alden Life Ins Co.	Reinsurance for 90% of defined costs in excess of $150K and $125K, respectively in '95 and '94, per MA member per policy year. Hospitalization costs subject to recovery were $1,500.8K and $2,231.2K in '95 and '94, respectively.	(Not Specified)
9	The Keystone Health Plan Central	523,051	173,171,138	Trans-General Life Insurance Co	Recoveries limited to 70% of amounts in excess of $250,000 stop loss	12/31/96
10	The HMO of Northeastern PA	326,009	142,615,982	Allianz Life Insurance Co	$2 million	Continuous until cancelled
11	Aetna Health Plans of Central and Eastern PA, Inc.	(196,698)	114,363,354	Aetna Life Ins Co.	Unlimited	Indefinite
12	The Oxford Health Plans (PA)	(219,169)	99,847,599	Allianz Life Ins Co. of North Am	$150,000 stoploss limit, per member per year, lifetime limit of $2 million	07/31/95
13	The Prudential Health Care Plan (dba Prucare)	0	56,155,292	The Prudential Insurance Co of America	$1 million per person deductible	None	Unclear why no reinsurance expense
14	Amerihealth HMO, Inc.	280,104	53,656,601	Allianz Life Ins Co. of North Am	$75,000 deductible, $2 million per member lifetime	07/01/96
15	HealthGaurd of Lancaster, Inc.	(156,414)	46,249,965	Allianz Life Ins Co	$100,000 deductible, $2 million lifetime	12/31/96
16	Aetna Health Plans of Western PA, Inc.	54,149	27,659,054	Aetna Life Ins Co.	Unlimited	03/31/96
17	Central Medical Health Plan, Inc.	155,967	21,936,814	AON Managed Care/Combined Insurance	$125,000 deductible and [$99,000 retension estimate]	09/30/96
18	Alliance Health Network	(65,279)	21,720,860	Allianz Life Ins Co.	$50,000 deductible per occurrence on hosp I/P claims, & 85% of excess	04/30/96
19	Cigna Health Care of PA	23,854	4,891,421	Connecticut General Life	$150,000 deductible; amounts thereafter reimbursed at 80%	Continuous
20	Greater Atlantic Health Plans, Inc.	0	0	QualMed Health and Life	$1 million per contract year, $2 million per lifetime	06/30/96	"No membership"
21	HIP of PA, Inc (d/b/a HIP Health Plan)	0	0	Allianz Life Insurance Co of North America	2 policies starting with a $50K deductible to $180K and $185K deductible to a maximum individual coverage of $2M. There are varying percentages of coinsurance.	07/01/96
22	The Medigroup HMO, Inc.	0	0	BCS Insurance Co	$2 million per member	06/01/96	(Apparently no membership)
23	The Riverside Health Plan, Inc.	0	0	American Service Life	50%-90% dependent on contracted services	12/31/95	Apparently no membership)
24	The Vista Health Plan	0	0	None			"No subscribers"

Source: The Lewin Group tabulations of HMO Annual Statement data Pennsylvania.

Attachment D

Published Reinsurance Tabulations

Table D-1

Percentage of Employees in Plans with Various Funding and Risk Sharing Arrangements, by Firm Size, 1996

Source: Lewin computations from data from KPMG Peat Marwick Survey of Employer-Sponsored Health Benefits, 1996.