The Costs and Effects of Parity for Mental Health
and Substance Abuse Insurance Benefits

Chapter 3 - Actuarial Assumptions

This chapter begins with a discussion of some of the key components of the studies conducted during 1996 that estimated the increase in premiums due to mandates requiring parity in mental health and substance abuse (MH/SA) insurance coverage. Previous actuarial predictions of premium increases due to MH/SA parity varied widely, primarily due to differences in their assumptions. We also discuss the assumptions underlying the updated model we used to make our cost estimates, which are discussed in chapter 4.

Previous Actuarial Studies of Parity

In 1996, four actuarial studies predicted the increase in health insurance premiums that would have resulted had the Domenici-Wellstone amendment to S.1028 passed. (The Domenici-Wellstone amendment would have required full parity for mental health benefits.) These predictions varied widely. They ranged from 3.2 percent by Coopers and Lybrand (Bachman, 1996b) to 8.7 percent by Price Waterhouse (Rodgers, 1996). Milliman and Robertson, Inc. predicted a 3.9 percent premium increase (Melek and Pyenson, 1996a). The Congressional Budget Office (CBO, 1996) predicted a 4.0 percent increase.

Four other actuarial studies were conducted in 1996. Two studies estimated the increase in premiums for S.298, which would have mandated parity for serious mental illnesses (SMI) only. This bill also did not pass. These estimates ranged from 2.5 percent (Melek and Pyenson, 1996b) to 11.4 percent (Watson Wyatt Worldwide, 1996). Two other studies estimated the increase in premiums due to the Mental Health Parity Act of 1996. These estimates were similar--0.4 percent (Lemieux, 1996) and 0.3 percent (Bachman, 1996a).

Assumptions Underlying the Previous Studies

Differences in the estimates of the costs of different parity proposals are largely the result of differences in the assumptions underlying the models used in the studies (Sing and Hill, 1998). These actuarial models include the components discussed below. The studies are summarized in Table 3.1.

TABLE 3.1
ACTUARIAL COST STUDIES ESTIMATING THE EFFECTS OF EXPANDED MH/SA INSURANCE BENEFITS

Legislation Analyzed	Study	Organization for Which the Analysis was Conducted	Estimated Increase in Premiums Due to Mandate
S.1028	Congressional Budget Office (May 13, 1996)	U.S. Congress	5.3 percent for indemnity plans 4.0 percent for a composite of indemnity and managed care plans
	Coopers and Lybrand LLP (April 1996)	American Psychological Association	3.2 percent
	Milliman and Robertson, Inc. (April 12, 1996)	The Coalition for Fairness in Mental Illness Coveragea	3.9 percent for mental illness and substance abuse 3.2 percent for mental illness (excluding substance abuse)
	Price Waterhouse (May 1996)	Association of Private Pension and Welfare Plans Business Roundtable ERISA Industry Committee National Association of Manufacturers	8.7 percent for composite of fee-for-service, PPO, POS, and HMO plans
S.298	Watson Wyatt Worldwide (March 1996)b	Association of Private Pension and Welfare Plans	8.4 percent for lower demand response 11.4 percent for prudent demand response
	Milliman and Robertson, Inc. (April 11, 1996)	Coalition for Fairness in Mental Illness Coveragea	2.5 percent
S.2031	Congressional Budget Office (September 1996)	U.S. Congress	0.4 percent initially 0.16 percent after employer responses
	Coopers and Lybrand LLP (September 1996)	American Psychological Association	0.3 percent for indemnity plans 0.12 percent for composite of fee-for-service, PPO, POS, and HMO plans

^aThe Coalition for Fairness in Mental Illness Coverage represents The National Alliance on Mental Illness, the National Mental Health Association, the American Managed Behavioral Healthcare Association, the American Psychiatric Association, and the National Association of Psychiatric Health Systems.
^bThe Watson Wyatt Worldwide estimates were made for all mental health diagnoses, not just SMI diagnoses.

Diagnoses. The text of the mental health parity provision to S.1028 (the mental health parity proposal) did not indicate whether "mental health" referred to mental health services only, or to both mental health and substance abuse services. Consequently, some studies included all MH/SA diagnoses in the parity mandate, and some included only mental health, excluding substance abuse. S.298, the SMI parity proposal, applied only to SMI diagnoses such as schizophrenia, manic depressive disorder, and major depression. However, one study estimating the costs of S.298 included all mental health diagnoses instead of only SMI diagnoses.³

³The Watson Wyatt Worldwide estimates include all mental illness treatment (Hay/Huggins Company, Inc. 1997c).

Health care delivery system. This component indicates the distribution of enrollees in fee-for-service (FFS), preferred provider organization (PPO), point-of-service (POS), and health maintenance organization (HMO) plans. Most studies estimated the premium increase for a composite of these plans. These composite studies specified a separate set of assumptions for, and computed the impact in, each delivery system. The impact in each delivery system was then weighted and summed with the weighted impacts of the other delivery systems to estimate an aggregate impact. Two studies estimated the premium increase for a PPO plan.

The impact of managed care. The impact of managed care indicates the effect of utilization management activities on health expenses. The studies generally characterized FFS plans as loosely managed delivery systems (with little control of health expenses), PPO/POS plans as moderately managed systems (with moderate control of expenses), and HMOs as tightly-managed systems (with firm control of expenses). However, one study assumed that POS plans were also tightly managed.

Pre-parity MH/SA benefit package. Some studies estimated the increase in premiums for a "typical" pre-parity MH/SA benefit package, and others estimated the increase for a "leaner" pre-parity MH/SA benefit package that required larger out-of-pocket costs for enrollees using the same level of services. Some of the leaner packages specified a 50 percent inpatient MH/SA coinsurance, instead of the typical 80 percent coinsurance.

Induced demand. This component incorporates the models' assumption about consumer response to a change in the price of MH/SA services. When MH/SA insurance coverage is expanded, the price of these services to plan enrollees declines, since there may be higher covered service limits and lower enrollee cost sharing. The induced demand effect indicates the degree to which consumers increase their use of MH/SA services in response to a decline in the price of these services. The assumptions for the increase in health plan expenses due to induced demand ranged from 5 percent to 69 percent.

Reduced utilization management for MH/SA services. Parity could require plans to manage medical/surgical and MH/SA treatment to the same extent. One study assumed that most insurance plans more stringently manage MH/SA treatment than medical/surgical care. That study also assumed that these plans would, therefore, reduce their management of MH/SA treatment to comply with the parity legislation. The other studies did not make these assumptions. None of the federal or state MH/SA parity laws that were passed on or before 1995 mandated and enforced parity with respect to utilization management.

Shift from public-sector to private-sector provision. State and local governments finance a substantial portion of MH/SA services. Most of the studies assumed that expanding private coverage of MH/SA services would have no effect on the amount of services provided in the public sector, or that these effects would not be relevant to their premium estimates. Two studies, however, assumed that a substantial shift would occur.

Different Assumptions in Two Key Areas

The actuarial studies of the Domenici-Wellstone amendment (S.1028) had widely different assumptions in two important areas. These areas are (1) the impact of managed care, and (2) whether parity results in a shift from public-sector to private-sector service provision.

The impact of managed care. Utilization management is key to constraining health care costs. The highest estimate of premium increases due to parity, 8.7 percent by Price Waterhouse (Rodgers, 1996), and the lowest estimate, 3.2 percent by Coopers and Lybrand (Bachman, 1996a), illustrate the different assumptions about managed care.

Coopers and Lybrand assumed that 50 percent of consumers would be in an HMO or POS plan, and that both of these plans are tightly managed. However, POS plans are typically classified as being moderately managed because they cover services provided by non-network providers, while HMOs do not. This suggests that instead of 50 percent, less than 30 percent of enrollees in 1996 would have been in tightly managed plans (Jensen, et al., 1997; Miller and Luft, 1994).

Unlike the other models, the Price Waterhouse model assumes that parity applies to utilization management, which will result in less management of MH/SA services. The model assumes that before parity, MH/SA services are more tightly managed than medical surgical services. Consequently, the model assumes that the amount of utilization management for MH/SA services in FFS, PPO, and POS plans would drop after parity, leading to an increase in expenditures. Although a law could require parity with respect to utilization management, to date no state or federal law has mandated and enforced such a requirement. Therefore, this assumption currently is not very realistic.

A shift from the public to the private sector. Some of the models assume that parity in employment-based insurance will cause a shift in the provision of MH/SA services from the public sector to the private sector. In other words, they assume that if the private sector provides more MH/SA coverage, consumers may use less publicly financed services and more MH/SA services paid for by private insurance.

In particular, Price Waterhouse assumed that this shift would raise FFS mental health expenditures by 50 percent and PPO/POS mental health expenditures by 21 percent (Rodgers, 1996). This is the second reason their predictions are higher than the other studies. No quantitative research has examined such shifts as a result of parity mandates, and we did not find any evidence of such a shift in the case study findings presented in chapter 2.

Actuarial Study for This Report

Based on this analysis of previous studies, our own case study findings, a review of the research to date, and input from this project's expert panel of actuaries and economists (see Appendix D), we made the actuarial assumptions described below. A majority of the expert panel members who commented on the assumptions agreed that the model's assumptions were reasonable.⁴ The panel includes representatives from three of the five organizations that produced actuarial estimates for federal parity legislation in 1996.

⁴In October 1997, we asked all expert panel members for their comments on the initial assumptions. Only one panel member did not respond. In November 1997, a draft of Sing and Hill (1998a) was sent to panel members to see if they had any additional comments on any of the model assumptions.

Assumptions Underlying the Updated Study

Baseline benefit packages. The baseline benefit packages for FFS, PPO, POS, and HMO plans represent typical packages that reflect the Mental Health Parity Act of 1996. In other words, for each type of plan, the baseline benefit package is the one that has the highest percentage of enrollees (the statistical "mode"). It is also the one in which the benefit maximum for mental health services equals the benefit maximum for medical/surgical services.

Induced demand. As noted earlier, the induced demand effect indicates the degree to which consumers increase their use of MH/SA services in response to a decline in the price of these services. Our assumptions about induced demand are based on findings from the RAND Health Insurance Experiment, which examined the effects of various cost-sharing arrangements on the use of health services and on the health status of individuals (Manning, et al., 1989). We assume a slightly lower response from people enrolled in HMOs. This assumption was suggested by two expert panel members.

Amount of utilization management. For medical/surgical services, the model assumes that FFS plans are lightly managed, PPO and POS plans are moderately managed, and HMOs are tightly managed. For MH/SA services, we assume more aggressive management. For FFS, PPO, and POS plans, we assume that, on average, the management of MH/SA benefits leads to a 25 percent reduction in costs compared to no management. For HMOs, we assume that MH/SA services are provided by behavioral carve-out plans that aggressively manage care and yield large cost savings.

Administrative costs. Our assumptions about administrative costs are based on a 1994 HayGroup study for the CRS (Hay/Huggins Company, Inc., 1997a). We assume that HMOs have higher administrative costs, especially for their MH/SA services.

Shift of MH/SA service delivery from the public to the private sector. We assume that parity does not shift the provision of MH/SA services from the public sector to the private sector. This assumption is based on case study findings described in the previous chapter.

Family demographics. Based on data from the Current Population Survey, the model assumes that 84 percent of employees with family coverage have a covered spouse and that there are 1.22 children per family. The model also assumes that spouses have expenditures that are 1.08 times higher than those of employees.

Table of Contents | Previous | Next