Beyond the Water's Edge: Charting the Course of Managed Care for People with Disabilities - Conference Resource Book. Costs and Incentives in a Mental Health and Substance Abuse Carve Out


Ching-to Albert Ma and Thomas G. McGuire
Draft; preliminary and unfinished


Financial support from National Institute of Drug Abuse Cooperative Agreement 1-P50-DA10233-01 and grant K05-MH01263 from the National Institute of Mental Health is gratefully acknowledged. We thank our research assistant Didem Bernard for her excellent help. We are grateful to the Group Insurance Commission of the Commonwealth of Massachusetts for allowing us access to the data.



This paper examines the overall change in costs of mental health and substance abuse services in a carve out program initiated in 1993 by the General Insurance Commission (GIC) of the Commonwealth of Massachusetts. Claims data for two years before (July 1991-June 1993) and two years after (July 1993-June 1995) the carve out were obtained from the GIC. These data were accompanied by an eligibility file for the four-year sample period. The exact financial arrangements in the vendor-payer contract are examined and described. The paper provides a full description of incentives, including multi-year contract renewals, and the payments and incentives associated with the administrative portion of the payments. We use those incentives to generate hypotheses about the effects of managed care on patterns of service use and cost.

The paper quantifies the changes in costs between the two years before the carve out and the two years after. By examining patterns of services in a population of continuously enrolled individuals, we eliminate selection-related changes in characteristics of the population.

The paper's main contribution is to describe and decompose the effect of managed care for mental health and substance abuse and to relate the observed effects to the incentives in the contract. We show the total plan and employee payments by month of service date for all major categories of expenditures, such as inpatient, other residential, and office visits. Trends in medical care prices will be used to adjust the data. Our basic decomposition therefore show impact by type of services, and show this separately for plan and employee-paid costs.

Our findings indicated significant savings after the carve out. Total and plan costs reduced by 50% to 70% over the four-year period. The pattern of cost reductions are similar with respect to outpatient and inpatient services, as well as to mental health and substance abuse services. The estimated average price of a mental health outpatient visit increased over time in the sample period, whereas that of a substance abuse outpatient visit decreased slightly.



Many big employers and payers have contracted with specialty management firms to administer the delivery of mental health and substance abuse (MHSA) benefits to their enrollees. This so-called MHSA "carve out" appears to be a most significant recent development, and has led to a new "behavioral healthcare" industry consisting of firms specializing in this service. Oss (1994) estimates that in 1994 over 50 million people in the U.S. are in some carve out program. "Risk-based" contracts, in which the specialty vendor (usually a for-profit corporation) bears some or all of the financial risk associated with MHSA services, are used in about half of all carve out programs. The rapidly growing use of separate carve-out contracts has been stimulated by reports of very favorable cost experience for many payers, with some savings reported to be in the range of 40 percent or more (Frank, McGuire and Newhouse 1995).

From an employer's or a payer's point of view, a carve out contract addresses the longstanding issues of moral hazard and adverse selection associated with insurance for mental health services (McGuire 1981; Frank, Huskamp, McGuire and Newhouse forthcoming; Frank, Glazer and McGuire 1996). Moral hazard is contained by the techniques associated with managed care--price negotiations, provider network selection and monitoring, prior authorization and utilization review. Adverse selection can be addressed by unification of the financial risks associated with mental health within a single contract; by pooling all persons in the same contract, no plans compete to avoid costly MHSA users.

Although the carve out approach offers these potential advantages in principle, the practical importance of this new form of insurance contract remains to be established. Favorable experience of innovative firms need not be a good predictor of what happens to the typical employer. First, if payers who first adopted carve out methods for MHSA services management are those with below average management efficiency in their previous existing plan ("low-hanging fruit" in the language of the industry), then the effectiveness of carve outs may be much less for payers with well-run plans (Frank, McGuire and Newhouse 1995). Second, the experience of a particular payer and population is often influenced by many specific factors, some of which may not apply to other payers. Therefore, it appears important to study the diversity of payer and population characteristics, vendors' management techniques, and the actual contracts between them carefully, before generalizations are made.

We contribute to the accumulating evidence on carve outs and managed care by reporting on the experience of the MHSA carve out of a major employer in Massachusetts--the Commonwealth itself. In this first paper in a continuing project on this case, we relate the incentives in the contract to the aggregate experience. First, we describe the MHSA carve out contract between the Commonwealth of Massachusetts and the vendor, and identify its incentive implications. Second, we analyze insurance claims data for two-year periods before and after the carve out. We examine the association between the contract incentives and the actual cost outcomes, and use the period before the carve out as a benchmark for comparison. Before-and-after comparisons can be problematic because the underlying population can change. We have therefore selected for detailed analysis a group of enrollees who are continuously covered for the entire four-year data period, and examine the actual use and cost experience for them before and after the carve out.



A behavioral health carve-out program was initiated in 1993 by the GIC of the Commonwealth of Massachusetts. The largest private payer in the state with an enrollment base of about 120,000, the GIC is responsible for providing health insurance to state and some local employees and their dependents. The GIC contracted with a combination of traditional indemnity insurers as well as HMOs since the middle of the 1970s. Between fiscal years 1989 and 1992, the State Hancock Plan, administered by John Hancock Mutual Life Insurance Company, was the indemnity plan for GIC enrollees. This managed fee-for-service plan included preadmission certification, utilization and concurrent reviews, second opinions and discharge planning, as well as pharmacy provider networks as managed-care features. These provisions applied to all areas of medical care, including MHSA services. In addition, GIC contracted with 14 HMOs (staff/group and network models) and offered them as enrollment options to employees.

The GIC voted to change its health benefit plans in late 1991. The stated goal was to improve the value of services to employees given the overall expenditure level, increase enrollment in managed care, and reduce risk fragmentation and adverse selection problems (Group Insurance Commission, Request for Proposal, 1992, p.1-3). To achieve this, the GIC retained services of a management consulting firm to assist with the evaluation of its existing benefits program, and the search for alternative benefit designs. One of the consultant's recommendations adopted by the GIC was the development of a separate MHSA carve-out progarm for enrollees. By a proposal request and subsequent biddings and negotiations, GIC selected a behavioral health care firm, Options Mental Health, Inc., from among five applicants, to set up a managed care mental health network of physicians and providers, and to manage mental health and substance abuse care on a partially at-risk basis.

The trade press contains many favorable reports of the experience of employers with carve out plans. Battagliola (1994) summarizes the experience of IBM which implemented a behavioral health carve out in 1991. In 1989, IBM was spending $106 million on MHSA benefits for its employees and dependents; this was going up at 10 percent per year, and consuming 15 percent of all health benefit costs. The carve out (with Value Behavioral Health [VBH]) consisted of a PPO with differential in-network and out-of-network cost sharing, expansion of alternative treatments, strengthening of an Employee Assistance Plan (EAP), and utilization review. By 1993, IBM's mental health costs had fallen to $59.2 million and only 10 percent of health benefit costs. Clearly, something happened here! The article provides some information about enrollment changes (the number of employees was falling by 3-4 percent per year in the later years of the data), prices (inpatient cost per day fell by 40% between the pre and post periods), and benefit changes, but understanding what happened is difficult because no information is provided on the nature of the contract between IBM and VBH, or on the composition of the expenditure changes. Finally, it is worth mentioning that IBM began the initiative with a very generous plan and very high rates of spending per employee, approximately $660 per employee per year on MHSA, more than double the national average for the period. Reducing costs by 30 percent (in real terms) still leaves IBM far above average rates of spending.1

The formal research literature on carve outs is just emerging. Grazier et al. (1993) examine outpatient utilization data one year before and one year after implementation of a PPO point-of-service plan with a benefit change for 4,220 continuously enrolled, active employees. Overall the rate of outpatient use went up slightly, but the visits per user fell slightly. The employer/vendor contract was "administrative services only" or ASO, so the vendor bore no explicit financial risk associated with utilization.

Frank and McGuire (1996) describe the experience of a carve out plan for MHSA in Massachusetts Medicaid with aggregate data from one-year pre and three and a half years post institution of a behavioral healthcare carve out. Price reductions for inpatient care and the virtual elimination of inpatient treatment for substance abuse appear to have been the main mechanism generating savings of approximately 25 percent per enrollee in real terms. The reduction in services was experienced virtually entirely by the disabled Medicaid beneficiaries. AFDC enrollees saw their costs (adjusted for medical price inflation) go up slightly over the course of the contract. The one-year contracts between the state and the vendor, Mental Health Management of America (MHMA) were almost entirely ASO contracts, and gave the vendor small incentives to reduce costs. Massachusetts began the period ranking third among the state in terms of overall health care spending for per Medicaid beneficiary. [ref]



Data for this project come from eligibility and health claims files, covering the period July 1991 through June 1995, and provided to us by MEDSTAT. Identifying information about the contract holder was scrambled so that claims data could be merged with eligibility information without identifying contract holders. The eligibility data allow us to calculate the average number of Primary Insured Participations, or PIPs for each month. A PIP is essentially a contract holder.2 Family contracts may cover more than one individual. We use relation, sex and date of birth information to identify individuals.

For some analyses we use a subsample of PIPs consisting of those covered by the GIC for the entire four-year sample period.3 The purpose of identifying this "continuously covered" population was for a better control of sample characteristics. All of these individuals have been covered by the GIC before and after the carve out. Cost outcomes of the continuously covered subsample will be compared to those of the entire sample. About 40,000 individuals are in our continuously covered population.

In the post carve out period after July 1993 we sought information about any claim for MHSA that would be covered by the carve out contract. Inpatient and other residential care was included in the sample. For outpatient care, we extracted any claim with a mental health procedure. A comparable selection criteria was used for the pre period as well, to make utilization in the pre period comparable to utilization in the post period.4

The claims data contain several cost related fields. The contract between GIC and Options is driven by the amount that the GIC has to pay, so some of our analysis will be based on the payments by GIC reported on the claim. Claims also contain information about payment amounts that are the responsibility of the beneficiary such as copayments and deductibles. Finally, covered charges represent the total negotiated price that Options has arrived at with the provider. Normally, the sum of GIC payments, beneficiary payments, and other payer obligations (if any) will be covered charges. Providers also report charges, but we will not use this information in this paper.

Units of services such as length of stay (LOS) and visits on some outpatient claims are also reported on claims. Price per unit will be calculated by dividing covered charges by the appropriate units.

Claims data for the last two months in the sample period appear to be incomplete, apparently because of delays in the submission and processing of claims. We requested data as of November 1995, allowing three months past the final service date, but this was not long enough to accumulate almost all claims for the last quarter of data. For this reason, we discarded the claims data for the last three months in the sample period, and instead base the last year's figures on seasonally adjusted nine-month data.

By any standard the data show a very significant cost reduction after the carve out. Table 1 summarizes the findings for the entire enrolled population; all prices and costs are in current year dollars. For this population, the total net payment from GIC for all MHSA services was about $9.32 million for fiscal year 94 (July 1993-June 1994), and $7.29 million for fiscal 95. These compare to $16.93 million in fiscal 92 and $14.87 million in fiscal 93, the two years before the carve out. The average GIC payment per PIP per month for the four years between 1992 and 1995 were, respectively, $20.32, $17.84, $9.52, and $7.49. The average GIC payment per enrollee per month for these years were, respectively, $13.91, $12.22, $6.04, and $4.76.

Table 2 presents similar cost figures for the continuously covered population, and all price and cost figures are in constant 1995 dollars, with medical price index adjustment. Here the total GIC payment between 1992 and 1995 were, respectively, $10.45, $8.47, $4.60, and $3.89 millions. The average payments per PIP per month were, respectively, $32.41, $26.24, $14.26, and $12.08; per enrollee per month figures were, respectively, $22.03, $17.84, $9.70, and $8.22. Overall various indicators of "costs" have decreased between 50% and 70% in four years. We also find that total costs of MHSA services--the total paid by GIC, enrollees, as well as any third parties--show a similar pattern. Thus, the savings were not simply achieved by shifting costs from the GIC to enrollees or another payer.

Table 3 categorizes the plan and total costs of the continuously enrollees according to inpatient versus outpatient services: inpatient costs declined by about 50% while outpatient costs by more than 60%. The breakdown of these changes according to MHSA care are illustrated in Table 4.



To understand the contract between Options and the GIC, it is useful to provide some background about the proposal request and negotiation processes. In the Request for Proposal (RFP), each potential bidder was provided with a summary of the plan enrollment, costs, and utilization pattern data for two years before the RFP was released. For each of the two years, the data included hospital admission and outpatient visit rates per 1,000 enrollees, number of hospital days per 1,000 enrollees, costs per hospital admission and outpatient visit, and costs per employee. These data were given for MHSA services, both separate and combined, for all employee groups.5 Utilization pattern data, such as distribution of admissions by diagnosis and outpatient visits, readmission rates, patterns of large claims, were also provided.

The GIC and its consultants first used the data to establish a set of benchmark projections of costs and savings. Each potential vendor was asked to provide its own set of projections, and the two sets of projections were compared and evaluated after the bids were submitted.6 Finally, Options was selected as the winner, and the details of the final MHSA contract were decided.

We now describe the contracts between the GIC and Options. The initial contract was for a one-year duration, and began in July 1993. It was expected at the time that the contract renewal for a second year would happen when the initial contract expired. We will briefly describe the benefit and coverage design. Detailed descriptions of the financial arrangements between GIC and Options will then be provided.

Important dimensions of the new benefit plan for MHSA were dictated by the GIC in the RFP. The MHSA carve out would be a managed care plan, nominally similar to the "managed care" in the previous Hancock plan, but expected to be more aggressive. The GIC specified the in-network and out-of-network benefits, goals for provider networks, and even the utilization levels (10, 20, 30 visits) at which the vendor should be intervening in the care process. Implementation of these features were of course to be left to the vendor. Benefits to enrollees choosing in-network care in the point-of-service plan were expanded from coverage before the carve out. Providers were to be precertified by Options before being admitted to the network. Whether an enrollee receives care from a network provider or not, precertification must be obtained from Options by calling a toll-free telephone number before care began (except for emergencies). A Clinical Case Manager was responsible for precertification. Options must be notified within 24 hours of any hospitalization, whether emergency (life-threatening), urgent, or routine. Complaints and grievances were reviewed by Options representatives, as are disagreements with clinical determinations.7

Financial aspects of the carve out that are relevant to enrollees are as follows.8 Generally, in-network coverage for inpatient services is complete with no deductibles; out-of-network inpatient coverage is 80% of allowed charges, with a 60 days limit per year and with a two-admission or two-episode lifetime limit on substance abuse treatments. In-network outpatient visits are free for the first four, subject to a $20 copayment for the fifth to twenty-fifth, and subject to a $40 copayment thereafter. Out-of-network outpatient coverage is 50% of allowed charges, and subject to a maximum of 15 visits per year. In-network out-of-pocket expenses are limited to $1,000 per individual and $2,000 per family. Finally, the lifetime benefit maximum is $1 million.

Benefits and cost sharing in the MHSA carve out program were substantially better for the enrollees than their previous plan. Before the carve out, mental health inpatient coverage at a general hospital was complete for 120 days (after a $150 deductible), then 96% after annual deductible. But mental health coverage at a psychiatric hospital was complete for only 60 days, and at 80% thereafter with a limit of 300 days. Perhaps, the most striking difference was that before the carve out, substance abuse coverage at a substance abuse facility was at 80% and only up to $10,000 a year after deductible. Outpatient MHSA coverages were respectively at 50% and 80%, with respective limits of $1,500 and $2,500 per year after deductible. The annual benefit limit was $500,000; lifetime, $1,000,000. The benefits after the MHSA carve out represented significant improvements, especially for in-network care.

The financial contract between the GIC and Options consisted of two main parts. First, for the fiscal year beginning July 1993, each month Options received from the GIC a fee (the ASO fee), which was calculated by multiplying the number of PIPs by $3.43. Second, this rate would be adjusted upward by 5% in the second year unless otherwise agreed upon by the GIC and Options.9 The contract for fiscal 1993-4 also specified a target claims cost of $20.72 per month per PIP. Besides serving as a benchmark to evaluate cost effectiveness of the contract, it would be used to adjust the ASO fee. In the actual implementation, for the fiscal year beginning July 1994, the ASO was revised to $3.17 per month per PIP, and the target level lowered to $15.39 per month per PIP.

For the fiscal year beginning July 1993, the target was established at $20.72 per month per PIP. The $20.72 refers to the portion of costs paid by the GIC, and does not include enrollee cost sharing. At the end of the fiscal year, the actual claims costs would be compared with the aggregate claim target (aggregate, because the rate was stated in terms of per month per PIP), and the ASO fee would be reduced by an amount equal to 20% of the excess of actual claims over the target, but this reduction would not be more than 20% of the ASO fee for the contract year. For example, if the claims cost turned out to be $21.72 per month per PIP, then the ASO fee would be reduced by $0.2 (20% of $21.72-$20.72) per month per PIP. The maximum cost overrun for which Options's ASO was reduced was $(20.72+3.43)=$24.15. For fiscal year 1995-6, the target was reduced to $11.19 per month per enrollee, but the ASO fee was raised to $5.18 per enrollee per month. The adjustment of the ASO fee according to the excess of claims costs over target remained unchanged.

Besides the adjustment of the ASO fee according to the discrepancy between actual claims cost and the target, Options was required to satisfy performance targets. During the first year, the set of performance guarantees consisted of five items, but expanded to sixteen in the second. The following is a sample from those in both years:10

  • At least 90% of enrollees surveyed by an independent contractor should be satisfied with the services they received.
  • Options should deliver reports by due dates.
  • Options should guarantee claims financial accuracy to be no less than 99%; payment accuracy, 97%; procedural and coding accuracy, 95%.
  • In the event that any of these guarantees was not met, Options must pay a penalty to GIC equal to 2% of the ASO fee for each guarantee violation, but the maximum of such penalty payment could not exceed 23% of the ASO fee.

It is important to keep in mind that the overall benefit package was expanded substantially after the carve out. In particular, coverage for in-network outpatient care was greatly improved. If enrollees' copayment and deductible remain unchanged, this coverage improvement must tend to increase use. Furthermore, even if use did not increase due to the benefit expansion, the improvement in coverage for in network care would tend to shift costs to the GIC from other payers whohave contracts with the GIC enrollees. Thus, Options would have to implement some cost savings measure simply to be able to maintain costs to the GIC at existing levels. Indeed, the initial claims target of $20.72 per PIP was such a level that savings by Options would just offset any cost increasing effects of the benefit expansion.



First consider the explicit incentives in the first year of the contract, and focus on the financial penalty and rewards associated with the claims target. Up to 20% of the ASO fee could be refunded to the GIC in case the actual cost was higher than the target level. The ASO fee to Options was the result to negotiations, and was paid regardless of the costs actually incurred in administration; thus, it was a type of prospective payment. The ASO fee included a profit allowance, but the actual profit or loss might be higher or lower depending on the costs actually incurred by Options.

Clearly, Options would attempt to economize on its own administrative expenses. If controlling MHSA costs requires Options' resources, such resources would only be provided if Options is properly motivated. Indeed, the carve-out contract does contain explicit incentives for Options to control MHSA service costs. The most explicit of such incentives is associated with the claims target. The ASO fee could be reduced by up to 20% in response to costs accumulating above the claims target. To such a small company, this probably represented a significant amount of potential earnings. Nevertheless, most of the financial risks remain with the GIC. In spite of the fact that the contract is written in terms of a per PIP per month payment, the contract is very far from being a "capitation" contract in which risk is shifted largely to the vendor.

These points are extremely important and illustrated in different ways in Figures 1 and 2. Figure 1 shows how the ASO fee to Options, and costs to the GIC vary with the actual level of claims costs per PIP in the contract's first year. Options faces some risk, but this is quite small in comparison with the possible cost variations faced by the GIC. Given the different sizes of Options and the Commonwealth of Massachusetts, the risk sharing arrangement appears to be sensible. Although GIC does bear most of the MHSA service costs, the remaining cost responsibility assumed by Options still seems significant for providing incentives for Options to meet the cost target. Figure 2 depicts the same risk sharing arrangement in a "proportional" way. Here, it is clear that the carve-out contract does not shift all cost responsibilities to the vendor.

As we noted above, the contract between GIC and Options was subject to renewal after the first year. The initial contract did specify an automatic adjustment on the ASO fee by 5% but other details of the contracts were open to revisions. In fact, in the second year, the same type of contract was signed by the GIC and Options, but the cost target was lowered from $20.72 per month per enrollee to $15.39 (or about 25%).



The ASO fee arrangements for Options contained a number of very interesting features. First, the contract did not allow the ASO fee to increase when Options was able to lower costs below the target level, but was subject to the risk of up to 20% of the fee for cost overruns. For a company of the size of Options, the total risk does not appear to be totally insignificant. This perhaps contrasts with the Massachusetts Medicaid behavioral health carve out (see Frank and McGuire, 1996), where the "at-risk" contract imposed a maximum penalty of $300,000 in the first year of the contract. In contrast, if the ASO fee was $3.43 per month per enrollee, for a population of 70,000 PIPs, Option's potential penalty in a year could be more than $560,000. If the use of aggressive managed care to reduce claims costs meant higher administrative expenses, the incentives established by the ASO fee mechanism would imply that costs should not be expected to fall significantly below the target level. But in actual fact, the first-year claims costs did fall significantly below the target. This brings us to the second point.

Options might have correctly anticipated that significant cost savings in the first year could have two effects. First, its superior performance might prompt the GIC to raise its expectation about cost saving potentials. A likely consequence was that GIC would lower the target rate. This phenomenon of superior contract performance resulting in more demanding terms in the future is called the "ratchet effect" in the contracting literature (see Laffont and Tirole, 1986, for example). Second, Options might think that it could convince the GIC that its value to the behavioral mental health carve out was high by demonstrating excellent performance in the first fiscal year. This could enhance Options's bargaining power in the contract renewal for the third year. In addition, it might also be a good signal to the market, so that Options's prospect of winning new contracts would be improved. We will call this the "reputation effect."

Clearly, the ratchet and reputation effects act against each other: the former induces Options to lower its performance, but the latter provides the opposite inducement. We can argue that Options in fact chose a performance level that traded off these two opposing effects. It was interesting to observe that the target rate was lowered in the second year by about 25% (in normal terms), and further reduced in the third year, but the administrative fee was reduced by a little in the second year, and then raised significantly in the third year.

From the perspective of incentives, the existence of a penalty for cost levels that are above the target does not necessarily imply that the target level will be achieved. In fact, Options might optimally choose to violate the target, incurring some penalty while saving administrative expenses. Nevertheless, the contract did not provide any incentive for Options to lower costs below the target level, since Options was unable to keep any savings. Therefore, it seems to us that what needs to be explained was the fact that Options achieved much more: in each of the years after the carve out, the actual costs were lower than the target level by a significant amount. Here, our hypothesis is that the reputation effect initially dominated the ratchet effect: for small cost reductions beyond the target level, Options's reputation began to build up, but the ratchet effect did not become important until significant savings beyond the target level was attained.

To understand the impact of the carve out, it is important to distinguish different two sets of relationship changes. First, Options was brought in to implement the provision of MHSA services by managed care. Whereas before the carve out, only those enrollees with the HMOs had their care delivered via managed care, all enrollees were under the management of Options since the carve out. This is a form of demand-side management. Second, Options set up a network of providers for enrollees. Before the carve out, providers negotiated individually with the GIC. After the carve out, Options, on behalf of the GIC, centralized all negotiations with providers. This affects the supply side. The first change may have the effect of reducing inappropriate use of MHSA services, since preadmission authorization, utilization review, and other monitoring may deter or screen out some demands for services. The centralization of bargaining makes Options a "monopsonist" buyer with market power, and enables it to use the size of the GIC population to secure a lower price from providers.

The above arguments suggest the following decomposition analysis. Consider any single type of service, say an outpatient visit. By definition, the total cost of this service in a given period of time is equal to the total number of times this service is used multiplied by the average price of each service. A reduction in total cost of this service can come about through a reduction in the quantity, the price, or both. From the claims data, we calculate the total number of outpatient visits for the periods before and after the carve out. Using the data of outpatient costs, we can estimate the average price per visit. For inpatient services, we calculate LOS of each episode and obtain the average LOS by dividing by the number of inpatient episodes. Using the inpatient costs data, we then estimate the average price per inpatient day. As it turns out, after the carve out, the claims data separated out from all inpatient services an additional class: inpatient service at an alternative setting. These are inpatient services performed at a less intensive setting such as residential facility, partial facility, intensive outpatient, residential professional and partial professional settings.

Table 5 and 6 present the decomposition of MHSA outpatient services and costs. For the continuously enrolled population, we estimate the prices per MHSA outpatient visit by dividing the total outpatient plan costs (after discarding outliers that may simply reflect adjustments to previous claims) by the total number of visits. We express the estimates both in terms of current year dollars and constant 1995 dollar. Table 5 shows an upward trend for MH outpatient prices, but a downward trend for SA. Nevertheless, we should note that the outpatient coverage of MH was significantly improved after the carve out; before the carve out, MH outpatient coverage was at 50% while SA at 80%. Table 6 presents the our own analysis and that from Options on number of admissions and average LOS per admission. While the data we received from MEDSTAT gave us numbers of admissions that were comparable to those Options reported, the total number of inpatient days were higher from our own analysis. Furthermore, we were unsuccessful in decomposing from our data total inpatient days into "conventional" and "alternative setting" inpatient services. Nevertheless, there is a slight decrease in the total of admissions as well as the ALOS in both analyses. From Table 5 suggests that the dramatic decrease in outpatient costs could be due to reduction in quantities, since "prices" either increased or remained relatively stable. On the other hand, Table 6 suggests that the reduction in inpatient plan costs mainly could be a result of price reduction, since numbers of admissions as well as ALOS did not decreaseas much as the total plan costs.



  • The anticipated cost shifting from enrollees to the plan is offset by decrease in prices. Because of the improved MHSA coverage and benefits for enrollees, expenses for the plan should tend to increase. But in the GIC experience, this increase was more than compensated by the decrease in prices that GIC had to pay providers as well as by the effect of managed care quantities.
  • Both outpatient and inpatient costs decrease. Despite the general view that managed care will tend to shift the demand for MHSA services from inpatient to outpatient, the GIC experience shows a mixed result. For mental health services, the decrease in outpatient costs between fiscal 92 and 95 was significantly less than inpatient, while these costs decreases were almost in the same percentage for substance abuse.
  • Both MH and SA services decrease in quantity uniformly.
  • The target level in the contract must be understood in relation to the penalty. That is, the entire schedule of ASO fees must be analyzed. Although there are penalties for failing to maintain the target, there is no a priori reason to expect that the target will be maintained. The vendor may optimally fail to maintain costs below target, incurring the penalty while avoiding administrative costs.
  • In the case of the GIC MHSA carve out, the target level is related to the ratchet and reputation effects. We find that even when Options faces no financial gains from reducing costs below the target, in fact that was what happened. Meeting a target is insignificant when a contract is viewed in isolation, but may have repercussions when contract renewals and bidding for new contracts are considered part of a firm's incentive. As in many other industries, a good reputation is a very valuable asset to a firm. Our finding is consistent with the "long term" perspective of contracting. Whenever a carve out program requires the contracting out of the administrative and management duties of the deliveries of medical services, the long term effects of contracting must be considered.



  1. For other examples, see Alexander Consulting Group (1990) on McDonnell Douglas; Altman and Price (1993) on Alcan; and Umland (1995).
  2. According to the contract, a PIP is a covered person who is an employee, a retired employee (of various classes), a covered student age 24 or over, an individual not part of a family unit covered under some continuation provision (see Appendix D of the Agreement for Managed Mental Health Services by and between Commonwealth of Massachusetts Group Insurance Commission and Options Incs. 1993). Thus, the total number of PIPs does not correspond to the total of all enrollees; rather each PIP roughly corresponds to a unique employee identification number in the enrollment records. In particular, spouses and most dependents are not PIPs.
  3. We actually selected these enrollees by identifying contracts with months of enrollment of 46 or greater of a possible 48.
  4. Some cost shifting between MHSA and general medical care is possible. For instance, inpatient treatment for alcohol abuse could be reclassified by a clinician as treatment for gastrointestinal problems and paid for under the general health insurance benefit. We are not in a position to evaluate how much of such cost shifting has occurred. For study of this in another context, see Norton et al. (1996).
  5. Active employees, retiree and survivors, and all groups.
  6. In many instances, potential vendors were asked to justify their projections, or to provide information on the basis of which those calculations were obtained.
  7. It is unclear whether any outside arbitration would be allowed.
  8. The benefit and enrollees' out-of-pocket payment designs for fiscal years 93-94 and 94-95 are identical.
  9. An implementation fee was also paid by the GIC in the first few months. This was calculated at $.35 per PIP per month.
  10. See Merrick (1996) for more discussion of the performance targets.



Altman, L. And W. Price. "Alcan Aluminum: Development of a Mental Health 'Carve Out'." New Directions for Mental Health Services, Fall 1993, pp.55-65.

Alexander Consulting Group. "The Impact of the ASSIST Program During 1989 at the McDonnell Douglas Helicopter Company." Health Straties Group, May 1990, Westport, CT.

Battagliola, Monica. "Breaking with Tradition." Business and Health, June 1994, pp.53-56.

Frank, Richard G., Haiden A. Huskamp, Thomas G. McGuire, and Joseph P. Newhouse. "Some Economics of Mental Health Carve Outs." Archives of General Psychiatry, forthcoming, 1996.

Frank, Richard G. And Thomas G. McGuire. "Massachusetts Medicaid..." Psychiatric Services, forthcoming, 1996.

Frank, Richard G., Thomas G. McGuire, and Joseph P. Newhouse. "Risk Contracts in Managed Mental Health Care." Health Affairs, 1995, 14:3, pp.50-64.

General Insurance Commission, Commonwealth of Massachusetts. "Request for Proposal for Integrated Employee Assistance and Mental Health and Substance Abuse Program." October 16, 1992.

Grazier, K.L., R.M. Scheffler, S. Bender-Kitz, and P. Chase. "The Effect of Managed Mental Health Care on Use of Outpatient Mental Health Services in and Employed Population." In R.M. Scheffler and L.F. Rossiter, Advances in Health Economics and Health Services Research, 1993, Volume 14, pp.71-78, Greenwich, CT: JAI Press.

McGuire, Thomas G. Financing Psychotherapy: Costs Effects and Public Policy, 1981, Ballinger Publishing, Cambridge.

Merrick, Elizabeth. "Dissertation draft." 1996, Brandeis University.

Oss, Monica. "Managed Behavioral Health Market Shares in the United States." Open Minds, 1994, Gettysburg, PA.

Umland, Beth. "Behavioral Healthcare Benefit Strategies of Self-Insured Employers." Behavioral Healthcare Tomorrow, November/December 1995, pp.65-70.

TABLE 1. Mental Health and Substance Abuse Costs
Entire Set of Enrollees FY 92 FY 93 FY 94 FY 95
Average monthly PIPs (92 is est)
Average monthly enrolled
Total cost
Total plan cost
Total cost per PIP per month
Plan cost per PIP per month
Total cost per enrollee per month
Plan cost per enrollee per month


TABLE 2. Mental Health and Substance Abuse Costs (adjusted for inflation, in 1995 $)
Continuously Covered Enrollees FY 92 FY 93 FY 94 FY 95
Average monthly PIPs
Average monthly enrolled
Total cost
Total plan cost
Total cost per PIP per month
Plan cost per PIP per month
Total cost per enrollee per month
Plan cost per enrollee per month


TABLE 3. Inpatient and Outpatient Costs
Continuously Covered Enrollees FY 92 FY 93 FY 94 FY 95 % Change
Total outpatient cost
Plan outpatient cost
Total inpatient cost
Plan inpatient cost


TABLE 4. Breakdown of Mental Health and Substance Abuse Costs
Continuously Covered Enrollees FY 92 FY 93 FY 94 FY 95 % Change
Plan total outpatient MH cost
Plan total outpatient SA cost
Plan total inpatient MH cost
Conventional inpatient MH
Alternative level MH
$4,689,307 $3,246,222 $1,605,253
Plan total inpatient SA cost
Conventional inpatient SA
Alernative level SA
$1,188,883 $826,180 $332,671
NOTE: Inpatient MH, inpatient SA cost figures are from service claim file.


TABLE 5. Price Estimates of Outpatient Mental Health and Substance Abuse
Current Year Dollar FY 92 FY 93 FY 94 FY 95
Continuous set: MH outpatient
Continuous set: SA outpatient
Contant 1995 Dollar FY 92 FY 93 FY 94 FY 95
Continuous set: MH outpatient
Continuous set: SA outpatient


TABLE 6. Inpatient Quantity of Mental Health and Substance Abuse
Data from MEDSTAT FY 92 FY 93 FY 94 FY 95
Number of admissions
Total number of days
Data from OPTIONS Annual Report FY 92 FY 93 FY 94 FY 95
Number of admissions
Total number of inpatient days
Total number of alternative setting days
Total number of days
ALOS (counting only inpatient days)
ALOS (counting all days)






View full report


"96cfpk01.pdf" (pdf, 147.17Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report


"96cfpk02.pdf" (pdf, 45.07Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report


"96cfpk04.pdf" (pdf, 50.53Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report


"96cfpk05.pdf" (pdf, 52.03Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report


"96cfpk06.pdf" (pdf, 1.02Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®