Skip to main content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.


The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Growth in Premiums in the FEHBP from Mental Health Parity

In January 2001, the Federal Employees Health Benefits (FEHB) Program, the largest employer-sponsored health insurance program in the Nation, instituted a mental health and substance abuse (MH/SA) parity policy in compliance with an earlier Presidential directive. This policy mandated that MH/SA services would be covered to the same extent as general medical care with respect to benefit design features, such as deductibles, copayments, and limits on visits and inpatient days.

In the fall of 2000, the Department of Health and Human Services awarded a contract to evaluate the implementation and impact of MH/SA parity benefits in terms of access, utilization, cost, and quality of care. The findings of this evaluation are available in the report "Evaluation of Parity in the Federal Employees Health Benefits (FEHB) Program: Final Report" at

As the report was being finalized for publication, ASPE commissioned an independent actuarial analysis of the impact of MH/SA parity on premiums. The results of this analysis are reported in the attached memorandum.



FROM: Sarah Beavers and Jim Mays, ARC
DATE: 20 May 2005
RE: Growth in Premiums in the FEHBP from Mental Health Parity


Mental health parity was implemented by the Federal Employees Health Benefits Program (FEHBP) in 2001. This analysis explores the impact of mental health parity on FEHBP premiums, describing the magnitude of the effect and the reasons for the increase in premiums. The analysis is based on benefit costs, and FEHBP premiums are assumed to change proportionally with benefit costs. Therefore, an analysis of benefit costs would produce similar results to an analysis of premiums.

Mental Health and Substance Abuse (MHSA) as a percent of total health benefits grew an extra one percentage point from 2000 to 2001. Sixty percent of this was attributable to higher benefit rates for mental health, while 30% was attributable to beneficiaries purchasing more services and 10% to beneficiaries purchasing more expensive services. This confirms an analysis by the Office of Personnel Management that most of the increase in spending was from improved benefit levels.

MHSA as a % of Total Benefits Absent Parity 1.8%
MHSA as a % of Total Benefits With Parity 2.8%
Excess MHSA as a % of Total Benefits From Parity   1.0% 
Reasons for Growth in MHSA as a % of Total Benefits
   Higher Benefit Rates 60%
   More Services 30%
   More Expensive Services 10%

The conclusion that part of the increase in plan costs following the mental health parity implementation was attributable to increased total spending per enrollee was not corroborated by the "Evaluation of Parity in the Federal Employees Health Benefits (FEHB) Program: Final Report," December 31, 2004 (prime contractor Northrup Grumman Information Technology, Inc.). While modest increases in benefit payments were observed, the evaluation did not find increased total spending per enrollee across the subset of FEHBP plans studied. If that is a more accurate assessment of the expected results of the parity implementation, then the premium impact would be reduced by almost half relative to the aggregate analysis' results.


This analysis of the FEHBP used data from the Federal Employees Plan Blue Cross/Blue Shield plans for 1998-2003. This data is for non-Medicare, non-DOD individuals. Data used were either the sum of the data for actives, children, and annuitants, or data for all. The data shows total MHSA covered expenses and benefits paid, as well as inpatient days and non-inpatient charge lines and members using the services.

MHSA as a percent of total health benefits was 0.94 percentage points higher in 2001 than would have been expected had parity not been enacted. Absent mental health parity, MHSA as a percent of total benefits was declining. The average decline from 1998 to 2000 and 2001 to 2002 was 0.02 percentage points per year. Assuming this decline would have happened from 2000 to 2001 absent parity, it is possible to calculate the excess in MHSA as a percent of total attributed to parity by subtracting the expected percent (1.82%) from the true percent (2.76%). This results in MHSA as a percent of total health benefits being 0.94 percentage points higher in 2001 than it would have been without parity.

Three reasons for the extra percentage point of benefits going to MHSA in 2001 were considered. The effect of parity was to lower cost-sharing for mental health services. Lower cost-sharing means the plan pays more for services when beneficiaries buy the same services they would have bought absent parity. Lower cost sharing also causes beneficiaries to buy more services. The third effect is that beneficiaries buy more expensive, higher quality services. There are also interaction terms such as the fact that some of the additional services would be more expensive. These three reasons (and interactions) are assumed to partition the entire excess costs from mental health parity.

In order to partition the excess 0.94 percentage point of MHSA benefits, it is necessary to calculate the dollars associated with it. Multiplying the 2001 total benefits ($5,632,153,321) for actives, children, and annuitants by the 2001 value for MHSA as a percent of total (2.76%) results in MHSA benefits of $155,313,447. Only 1.82% (2.76%-0.94%) was expected absent parity. The extra 0.94% multiplied by total mental health benefits results in $52,965,823 of excess mental health benefits as a result of parity in 2001.

Data to estimate the effects of beneficiaries buying the same services but the plan having to pay more for them were the inpatient and non-inpatient benefit rates for mental health. For beneficiaries buying more services, the calculation was based on inpatient days and non-inpatient charge lines per 1,000 beneficiaries. To estimate the effects of purchases of more expensive services, the data were inpatient costs per day and outpatient costs per charge line.

In each case, year to year growth in the measures was calculated, and then excess growth as a result of parity was calculated. For benefit rates, the method was identical to the method used to calculate excess MHSA as a percent of total benefits. For days and charge lines per 1,000 members, there was no data for 1998 and 1999, so growth for “normal” years was based on 2001 to 2003 data. The average growth for 2001 to 2003 was subtracted from the growth in 2000 to 2001 to calculate excess growth. For cost per service, it appeared that part of the effect of parity happened from 2001 to 2002, so this year was excluded when calculating the “normal” growth rate absent parity. Inflation is captured as part of the “normal” growth rate.

The effects in 2001 of the excess growth from 2000 to 2001 for inpatient and non-inpatient for each of the three possible causes cited above were estimated. Of the $102 million ($155-53 million) total that mental health benefits would have been absent parity, 34% was for inpatient and 66% for non-inpatient, using the average share of inpatient and non-inpatient across all years in the data.

Therefore, $35 million would have been spent on inpatient and $67 million on non-inpatient mental health in 2001 absent parity. To directly estimate the effects of parity, these dollars were multiplied by excess growth in each of the three factors. For instance, direct growth in the benefit rate for inpatient mental health was calculated as the excess growth (.2390) multiplied by the $35 million that would have been spent on inpatient for a direct effect of $8.4 million in additional spending on inpatient attributed to the plan having to pay more for the same inpatient services. Following this method for the plan spending more on non-inpatient, beneficiaries buying more inpatient and non-inpatient services, and beneficiaries buying more expensive inpatient and non-inpatient services, $38 million, or 72% of the excess spending was explained, and $14.7 million was not.

Part of the unexplained excess spending is a result of interactions between the factors. For instance, the plan pays more per service on additional services. The plan pays more per service on the more expensive services, and some of the additional services are more expensive. It is possible to estimate the interactions by multiplying the excess dollars from one cause by the excess growth from another cause. In order to estimate the effect of the plan paying more per service for additional inpatient services, $8.4 million is multiplied by the excess growth in inpatient days (0.1237) for a total of $1 million. Following this method for each possible combination would cause double counting of the interactions, so each interaction was counted only once. It also produces a question as to how to split a particular interaction term among the two causes that are interacting.

One solution is to split the interaction terms pro rata according to the share of the direct effects for each cause. For the interaction of the plan paying more for more services, this was done as follows. The plan paying more per service accounted for 60.9% of the direct effects for more money being spent on inpatient. Utilization of more services accounted for 31.5%. When those two causes interact, 65.9% (60.9/[60.9+31.5]) is attributed to plans paying more per service. Therefore, $685,374 of the slightly more than $1 million interaction term is from the effects of the higher insurance costs, and $354,724 is from the fact that more services are purchased. This method was followed for each of the interactions for inpatient and non-inpatient services, and the interaction effects from each cause were added to the direct estimates for that cause.

The interaction effects explained $3.9 million of the $14.7 million that was not explained by direct effects. The remaining residual of $10.8 million was spread pro rata across the 6 categories according to the fraction attributable to direct plus interaction terms for each category. Spreading the residual this way implicitly assumes that the excess change in MHSA percent can be fully explained by plans paying more for the same services, beneficiaries buying more services, and beneficiaries buying more expensive services. It is also implicit that the error term for each one of these causes is not out of proportion to its explained share.

After adding the residuals to each category and summing inpatient and non-inpatient, the excess spending from parity was partitioned across the three causes. Plans paying more per service accounted for 60.0% of the additional spending, increased utilization accounted for 28.9%, and more expensive services being purchased accounted for 11.1%.

The impact of mental health parity on FEHBP premiums should be very close to the estimated change in benefits. Total premiums in an FEHBP plan are benefits plus administrative costs and profit. Parity moves MHSA services to managed behavioral health organizations. Although administrative costs may be slightly higher because of management, it is unlikely that the administrative loading overall would be much different. Therefore, the 1% increase in benefits converts to a 1% increase in total premiums. Because the enrollee contribution formula is tied to overall plan costs, both enrollee premiums and government costs should also rise by 1%.