In an essay format: Compare and contrast adverse vs. favorable selection.  What is the evidence of favorable selection in HMO plans and Medicare programs?

Essay format: double-spaced, 12- font, in MLA or APA format and placed through SafeAssign

Supporting references required, one of which must be the textbook.

Underwriting & Rate Making
Chapter 6

Overview
Premium computations
Medical Loss Ratio under the ACA
Underwriting
Rate Making
Community rating
Experience rating
Self-insuring
Combining dissimilar groups
Underwriting and the ACA

Premium Computations & Loading Fees
Gross premium = Pure premium/(1-loading percentage)
Pure premium is the expected loss
Loading percentage is the markup to cover objective risk, profit and costs of marketing, adjudicating and processing claims, coordinating benefits and providing access to networks.
Pauly & Percy (2000) estimated that the loading percentage was ~ 10% for group coverage and 50% for non-group coverage
Karaca-Mandic et al. (2011) used data from the MEPS-IC and concluded that loading fees were:
Similar for firms with <100 workers at ~ 34% Firms with 100 to 10,000 workers had loading fees of about 15% Firms with >10,000 workers had loading fees of about 4%

Loading Fees & Loss Ratios
Traditionally insurers have used their own approaches to defining which costs were claims and which belonged in the loading fees.
Unless there were state insurance regulations to the contrary, it made little difference, the market was usually concerned with the gross premium
The ACA changed this with a requirement for a minimum loss ratio

Medical Loss Ratios in the ACA

MLR =
(Medical Claims + Quality Enhancing Efforts)
(Premiums – Taxes)
ACA requires individual plans and group plans offered to groups up to 100 workers have an MLR of at least 80%
Larger group plans are to have an MLR of at least 85%
If these thresholds are not met, insurers must rebate premiums
Abraham & Karaca-Mandic (2011) estimate that in nine states at least half of individual insurers would likely fail to meet the 80% threshold if the ACA were then in effect.

Underwriting
Identifying the determinants of claims experience
Establishing risk pools with known expected losses and minimum variance
Matching new members to the appropriate risk pool

Objective Risk
Objective Risk = σ/(μ √N)
= variance/(expected loss times square root of covered lives)
o Risk increases with variance
o Risk decreases with the size of the expected loss
o Risk decreases with the number of covered lives

Rate Making
Community Rating
Manual Rating
Adjusted Community Rating
Community Rating by Class
Experience Rating
Prospective Rating
Retrospective Rating

Community and Manual Rating
Community Rating
All covered lives are in the same risk pool
Manual Rating
Individuals are placed in risk pools based upon common characteristics
Age, gender, location, occupation, industry, etc.
Health status

Prospective Experience Rating
Based upon prior claims of the group
Credibility factor
Weight of pool vs. group claims experience

Risk borne by the insurer
Rate is quoted for the forthcoming period

Retrospective Experience Rating
Basic premium covers administration, claims adjudication, and any stoploss features.
The firm is charged for each dollar of actual claims.
Stoploss features limit a firm’s risk:
Aggregate stoploss—If total claims exceed a negotiated dollar maximum, the firm pays no additional cost.
Specific stoploss—If a specific claim exceeds a negotiated dollar maximum, the firm pays no additional cost.
Typically, the firm makes quarterly payments and pays a “retro” fee at year-end to reconcile quarterly payments with actual experience.

Self-Insured Status and the ACA
It is relatively rare for smaller firms to be self-insured due to the objective risk issues we discussed earlier.
One can buy stoploss coverage to reduce the risk of unexpectedly high claims costs
Many have suggested that smaller firms with low claims costs may buy stoploss coverage as a means of avoiding being mixed in with higher cost firms in the exchanges.

Manual Methods Used by HMOs

Adjusted Community Rating
Community Rating by Class

Adjusted Community Rating
Use the claims experience of the entire pool.
Use “contract mix” (i.e., proportion of single, couple, family) within each group.
Define “contract size” (i.e., people in family) based upon group’s data.
Define “charging ratios” (i.e., ratio of couple and family to single rate) based upon group’s data.

Community Rating by Class
Premium typically based upon adjusted community rating factors.
Plus:
Age and gender mix of the group
Industry classification of the group

Effect of State Policies on the Market for Private Nongroup Health Insurance

Objective:
To identify the effects of state nongroup insurance reforms of the 1990s on the purchase of insurance coverage by “more healthy” and “less healthy” individuals.

Source: LoSasso and Lurie (2009)

Background
Between 1993 and 1996, eight states enacted legislation that limited the factors that insurers could use in setting premiums in the nongroup (i.e., individual) insurance market.
These states also had “guaranteed issue” provisions.
Kentucky, Massachusetts, Maine, New Hampshire, New York, New Jersey, Vermont, and Washington

Enacted “Community Rating”
The states all disallowed the use of health status or medical underwriting.
New Jersey: Pure community rating
New York: Geographic differentials allowed
Vermont: +/– 20 percent for demographics (except Blue Cross and HMOs)
New Hampshire: 3 to 1 ratio for age
Others: age, geography, family composition, gender

Data and Methods
Survey of Income and Program Participation (SIPP)
Nationally representative panel surveys of individuals
1990 to 2000
Those age 18 to 64 in 41 identifiable states
Approximately 35,750 observations per year
Probit regression:
Non-group Coverage = f(law, individual characteristics, state, and year fixed effects)

Results for “Healthy”
“Healthy” are men age 22 to 35 with self-reported health status of excellent or very good.
Adoption of community rating resulted in:
2.0 percent decline in nongroup coverage
37 percent decline relative to the mean
3.9 percent increase in probability of being uninsured
13 percent increase relative to the mean

Results for “Unhealthy”
“Unhealthy” are all those age 40 to 64 with self-reported health status of poor
Adoption of community rating resulted in:
4.5 percent increase in nongroup coverage
50 percent increase relative to the mean
7.4 percent decline in probability of being uninsured
50 percent decrease relative to the mean

Conclusion
Community rating had the effect of lumping dissimilar risks into the same pool
Raised the premium for low-risk folks relative to what they used to pay
Some of whom now dropped out of the health plan
Lowered the premium for high-risk folks relative to what they used to pay
Some of whom now joined the health plan

Underwriting in the ACA
Under the ACA insurers may not use health status in establishing rates
The may not use gender
Age rating is limited to a 1 to 3 band

Society of Actuaries Study
MEPS data on use and cost of services

Suppose there are an equal number of men and women in each age group.

Moving the ACA from 1 to 3 rating band to 1 to 5
Suppose the age range of health care spending is closer to 1 to 5 rather than 1 to 3
The ACA 1 to 3 rule raises premiums to young adults and encourages them to forgo coverage
What would happen to premiums if the rule was changed to allow a 1 to 5 band?

Discussion Questions
Suppose your insurance competitor begins to use information from genetic testing to set insurance rates in the nongroup market. Suppose further that these tests do identify meaningful differences in claims experience. You choose not to implement such a model. Discuss what is likely to happen to your enrollee mix and to the premiums you charge. Suppose federal law prevents the use of generic data to set rates (as the Genetic Information Nondisclosure Act of 2008 does). Does this mean that as an insurer you can ignore the effects of generics data?

Discussion Questions
Those who frequently watch late-night TV have undoubtedly seen commercials that say “no health questions.” In what way is this phrase relevant to the purchase of insurance? Who is likely to be attracted to the policies being offered? What do you expect about the premiums of these policies relative to policies that do ask health questions?

Discussion Questions
Suppose your small, general medicine physician group is offered a capitated managed care contract. Through this contract, the group will be paid on a capitated basis. This means that the group will be responsible for all of the costs of each patient’s care. What does the concept of objective risk tell you about the desirability of this contract?

Discussion Questions
Terhune (2002) described “reunderwriting” in the nongroup market. In this process, individuals are reclassified into a higher-risk group once they have a significant illness or claim. Discuss the effects of such a model on premiums and enrollment. If healthcare claims were essentially random over time in each of an insurer’s established risk pools, how would this affect your conclusions?

Discussion Questions
Given the discussion of self-insured health insurance plans, under what conditions would you expect a small employer to become self-insured?

Discussion Questions
The available data suggest that experience rating (together with self-insurance) is by far the most common underwriting method used for large employers. The avoidance of state insurance regulations and premium taxes may explain why firms tend to self-insure, but why do you think experience-rated approaches are more common than manual rating?

Discussion Questions
The ACA prohibits medical underwriting in the non-group market. Other things equal, what effect will this prohibition have on the premiums of healthy and unhealthy enrollees? Who are likely to be the healthy? The unhealthy? The ACA also prohibits gender differences in premiums in the exchanges. What effect is this likely to have on premiums?