A Report on the Actuarial, Marketing, and Legal Analyses of the CLASS Program. Attachment 1: Adverse Selection Memo

10/14/2011

Memo: Adverse Selection
From: Bob Yee
To: Kathy Greenlee
Date: April 27, 2011

In order to use the model developed under contract by ARC as the tool to develop premiums for CLASS, I have conducted a review of the model. It takes a portion of the general working population and determines the premiums that will be sufficient to pay future benefits and expenses. I find the model performs the projection calculations correctly given a set of specific assumptions. These assumptions of future events (claim rates, claim severity, mortality rates, lapse rates, expenses and investment yields) are derived mostly from population data and related experience in private long-term care insurance.

A major departure from the experience of private insurance is the effect of adverse selection arising from the lack of underwriting in CLASS. Private insurance provides little help as it is almost always underwritten. The ARC’s model expresses the adverse selection effect as a load to the normal expected claim costs without adverse selection. This load represents the anticipated higher proportion of unhealthy enrollees than in a normal mix of healthy and unhealthy workers.

The initial load is assumed to be a function of the enrollment rate and the prevalence rate of severely disabled (i.e. benefit qualifying) enrollees. The load decreases monotonically in time to no less than 10% (see attached exhibit for a more detailed description). While this formula is plausible, the resulting load to the normal expected claim costs is subjective and not supported by actual experience. Moreover, this is only one possible approach and may not even be correct. One can argue that the load may be increasing in the future when all unhealthy enrollees, initially qualified for benefits or not, are taken into account.

In order to illustrate the possible range of the adverse selection effect, I calculated the two end points of the range using the model. The monthly premium for an enrollee age 50 at time of enrollment is $60 with no adverse selection51. The corresponding monthly premium with full adverse selection (i.e. all enrollees are qualified for benefits at time of enrollment) is $564, or 9 times. It would even be much higher if claim severity is appropriately adjusted for these enrollees. The model generates a monthly premium of $93 using its adverse selection formula. This wide range bracketing the model premium suggests that the extreme sensitivity of the adverse selection assumption needs to be considered when developing the CLASS premiums.

Our task at hand is not merely to produce ‘best-guess’ premiums from a set of assumptions. The proposed plan(s) must be actuarially sound. In conjunction with other provisions, the CLASS Act imposes a rather severe condition for actuarial soundness. That is, the CMS actuary must certify that the assumptions associated with the premiums are reasonable in order to ensure that the CLASS program is sustainable over 75 years.

In practice, this is perhaps a more formidable task than the corresponding financial oversight in private insurance. From a solvency perspective, private long-term care insurers are typically multi-lined insurers and their solvency is spread over other lines of business, such as life insurance, annuities, disability insurance, etc. CLASS has only one line; it has no other apparent sources for support. Besides the various underwriting techniques to mitigate adverse selection, insurers control their risk exposure by phased roll-outs over a period of time and by adjusting premiums for newly issued policies. The CLASS program will be made available with limited restrictions through both individual and employer-based enrollment. It will be exposed to an unknown degree of adverse selection at the onset. Premium revision for future enrollees may be too little and too late to temper a larger premium deficiency already in the program. Quick premium revision may be seen as a sign of program instability, thus affecting subsequent enrollment results. CLASS essentially has only one shot to get it right.

As we have discussed previously, if the premiums are not set properly, the required premium rate increase may be substantial. In private long-term care insurance, large rate increases have often led to disastrous rate spirals. This is so because the increases are driving out the healthy insureds as premiums become increasingly unaffordable. This is an unacceptable scenario for CLASS.

The impact of adverse selection on claim experience is driven by the proportion of unhealthy enrollees in the program. Low enrollment generally means greater likelihood of an unfavorable proportion of unhealthy enrollees. It is a basic insurance principle that any sound insurance program has a good spread of risk. That is, there is an appropriate mixture of healthy and unhealthy enrollees to keep the premiums reasonable and stable.

Success in CLASS enrollment will largely depend on the price point and actions from private insurance. CLASS premiums must be attractive relative to the perceived value of the benefits. They will be compared to the respective premiums of private plans. We have previously shown that the premiums from the model are approximately twice that of private group plans with similar benefits. Moreover, agents will be inclined to sell against CLASS to the healthy workers. The proportion of unhealthy enrollees will be dependent on the mixture of individual and worksite enrollment. This proportion is subject to systematic encouragement for enrollment by certain organizations. Because the CLASS program allows individuals to enroll directly, it cannot effectively control the influx of unhealthy workers. The actual results for enrollment and proportion of unhealthy enrollees are highly unpredictable.

To illustrate this another way, there are approximately 200,000 workers with 2+ ADLS or cognitive impairment out of approximately 100,000,000 workers52. The normal annual claim rate is less than 2% but it is 100% for these severely disabled workers. Thus these workers are more than 50 times more likely to claim than the average workers in a given year. Suppose we decided that a 20% load53 for adverse selection is marketable against private insurance that has no such load. It would then take 22 healthy enrollees over a 10 year period to support one such disabled enrollee who can readily claim54. In 2010, there were approximately 187,000 policies issued under group long-term care insurance primarily to workers55. Even if we are wildly successful by enrolling half of this figure in one year, we can only allow up to 4,300 workers (less than 3% of the total) who have 2+ ADLs or cognitively impaired to enroll before the loaded premiums are inadequate.

Without a valid value proposition to the healthy workers, empirical evidence or any risk mitigating measures, there is great uncertainty in quantifying the adverse selection effect. With its theoretical formula, the ARC model is useful in demonstrating its impact on premiums. However, the model, by itself, should not be relied upon for prudent rate setting.

Uncertainty calls for conservatism. My current professional opinion is that the actuarially sound premiums for the basic CLASS plan in the statute, as well as the so called ‘Modified’ CLASS plan, are that of a pre-paid plan56.

It is not a coincidence that many experts have maintained that adverse selection is the major obstacle for the CLASS program. Any workable design must address it in order to receive certification as an actuarially sound plan.

View full report

Preview
Download

"index.pdf" (pdf, 350.5Kb)

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

Preview
Download

"appA.pdf" (pdf, 150.08Kb)

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

Preview
Download

"appB.pdf" (pdf, 548.14Kb)

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

Preview
Download

"appC.pdf" (pdf, 150.56Kb)

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

Preview
Download

"appD.pdf" (pdf, 318.45Kb)

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

Preview
Download

"appE.pdf" (pdf, 319.19Kb)

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

Preview
Download

"appF.pdf" (pdf, 155.03Kb)

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

Preview
Download

"appG.pdf" (pdf, 181.05Kb)

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

Preview
Download

"appGa.pdf" (pdf, 151.02Kb)

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

Preview
Download

"appGb.pdf" (pdf, 123.12Kb)

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

Preview
Download

"appH.pdf" (pdf, 256.35Kb)

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

Preview
Download

"appI.pdf" (pdf, 358.34Kb)

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

Preview
Download

"appJ.pdf" (pdf, 1.21Mb)

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

Preview
Download

"appJa.pdf" (pdf, 396.06Kb)

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

Preview
Download

"appJb.pdf" (pdf, 313.17Kb)

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

Preview
Download

"appJc.pdf" (pdf, 252.71Kb)

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

Preview
Download

"appJd.pdf" (pdf, 261.97Kb)

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

Preview
Download

"appK.pdf" (pdf, 186.22Kb)

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

Preview
Download

"appL.pdf" (pdf, 788.88Kb)

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

Preview
Download

"appM.pdf" (pdf, 249.45Kb)

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

Preview
Download

"appN.pdf" (pdf, 7.72Mb)

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

Preview
Download

"appNa.pdf" (pdf, 208.04Kb)

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

Preview
Download

"appNb.pdf" (pdf, 6.96Mb)

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

Preview
Download

"appNc.pdf" (pdf, 622.02Kb)

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

Preview
Download

"appNd.pdf" (pdf, 211.36Kb)

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

Preview
Download

"appO.pdf" (pdf, 2.01Mb)

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

Preview
Download

"appP.pdf" (pdf, 9.15Mb)

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

Preview
Download

"appPa.pdf" (pdf, 233.71Kb)

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

Preview
Download

"appPb.pdf" (pdf, 253Kb)

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

Preview
Download

"appPc.pdf" (pdf, 546.97Kb)

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

Preview
Download

"appPd.pdf" (pdf, 505.4Kb)

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

Preview
Download

"appPe.pdf" (pdf, 462.21Kb)

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

Preview
Download

"appQ.pdf" (pdf, 285.57Kb)

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

Preview
Download

"appR.pdf" (pdf, 797.76Kb)

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®