On October 12 America’s Health Insurance Plans (AHIP) released a commissioned report by Price Waterhouse Cooper (PWC), “Potential Impact of Health Reform on the Cost of Private Health Insurance Coverage.”   The study reported that health care reform as envisioned by the Senate Finance Committee would raise the cost of private health insurance by 23 percent above the costs projected for the current system from 2009-2016. 

The timing of the report–released one day before the vote of the Senate Finance Committee of its bill–brought an immediate and sharp response by the mainstream media, congressional Democrats, and the Obama administration.  Besides its timing, other issues impaired the study’s credibility.  First, the report did not include potential savings brought about by reform legislation in estimating the cost of coverage.  Second, the report provided limited information as to how cost estimates were derived.  Third, when methods were described, the assumptions at times seemed implausible.

Beneath the opaque and questionable numbers, however, AHIP calls to our attention an issue that should be of great concern to all supporters of reform and universal coverage: the destabilizing selection effects of insurance reforms without an effective individual mandate.  This blog examines evidence about reformed markets operating with and without mandates.

Democrats and reform supporters remember the “Harry and Louise” ads that effectively contributed to the demise of the Clinton health care reform in 1994, so there is reluctance among some Democrats and reformers to take arguments of the insurance industry at face value.   The insurance industry makes a convenient punching bag.  Unlike physicians, health plans don’t normally have a personal relationship with patients.  Insurers play a necessary role as the “bad cop,” sometimes denying coverage for services deemed medically unnecessary.  On occasion, the services are necessary but denied. To make this worse, private insurers occasionally appear to be “rogue cops.”  Health plans routinely deny coverage to individuals who need protection the most, and have been caught rescinding coverage when persons who passed their underwriting screens incurred major medical bills.

Before continuing, let me disclose my relationship with AHIP and the health insurance lobby.  I was the director of research at AHIP from 1994 to 1996. I also was associate director of research at the Health Insurance Association of America from 1986 to 1990.

Market Rules

Insurance reforms would change market rules, forbidding the use of pre-existing condition clauses and medical underwriting by health status and medical history.   Reforms would require guaranteed issuance and guaranteed renewal of individual and small-group policies.  Both Republican and Democratic reform proposals would either limit or prohibit different forms of underwriting. 

Actuaries are fond of the aphorism, “You can’t buy fire insurance when your house is on fire.”  It is not possible to have an effective market that prohibits medical underwriting for individual and small-group insurance without an effective mandate to purchase insurance.  Otherwise, healthy people might wait until they are sick to purchase coverage, and insurance markets would experience a serious and debilitating case of unfavorable selection.  The Senate Finance Committee proposal imposes no financial penalty in 2013, the first year of operation.  In subsequent years, the penalty increases to $200 per year in 2014, $400 in $2015, and $750 by 2017.   In contrast, the House bill released today would impose  a tax capped at 2.5% of adjusted gross income above a specified level.

Once an effective mandate is operational, it is also necessary to have an income-related schedule of subsidies so the uninsured can afford their required coverage.  About two-thirds of the uninsured are from households with less than 200 percent of poverty income.  To paraphrase Uwe Reinhardt, insurance market reform, coupled with an individual mandate, and a schedule of subsidies, are each legs of a three-legged stool.  All legs are necessary or the stool will topple.

How potentially destabilizing could selection issues turn out to be?  Among the uninsured, spending is more concentrated than among the Medicare and employer-based populations.   The top 10 percent of spenders account for 76 percent of spending among the uninsured, as compared to 49 percent for Medicare and 62 percent for those with private insurance, according to Sam Zuvekas and Joel Cohen

So any analysis of reform must understand a few technical relationships:

1.  What is the impact of the financial penalty on the take-up rate of the uninsured?

2.  Under an individual mandate, how does the level of financial penalties affect adverse selection for people buying from the individual market?

3.  How does the schedule of subsidies interact with the above two questions?

Penalties should strongly increase the take-up rate among the uninsured.  When the formerly uninsured pay premiums, they purchase risk protection and “moral hazard.”   When paying a penalty, the uninsured receive no benefit for their payment.  If the uninsured were rational and had perfect knowledge (which they don’t), the decision might be as follows:

Remain uninsured if (the sum of expected out-of-pocket payments for medical services + financial penalty) is less than ((expected out-of-pocket expenses for medical services with insurance + out-of-pocket expenses for premiums) – (value of additional services paid by insurance + value of risk protection)).

People in poorer health will place a greater value on the additional services paid for by insurance and the value of risk protection.  Additionally, subsidies will remove some of the incentive not to participate by reducing premium contributions for lower-income people purchasing health insurance.  The ease of signing up will greatly determine the take-up rate also, as will the effectiveness of the marketing campaign to reach young adults.

Searching for Evidence

For this critical issue, I began my search believing that the many think-tanks and universities providing timely research about health care reform would have already modeled the relationship between the financial penalty and take-up and selection effects.   I was wrong.   PWC and the Congressional Budget Office did not want to put their models into the public domain.  PWC officially declined to speak with me, while CBO officials did not return my phone calls. 

I called colleagues and the usual suspects.  The list includes research centers at UCLA, and UC Berkeley, Jon Gruber at MIT, Jack Hadley at George Mason, and the Urban Institute.   I also contacted a few actuarial firms.  The Urban Institute, UCLA, and Berkeley are now working on the issue.  But no one had an estimate at this time depicting the relationship between the size of the penalty and the take-up rate.  Additionally, no one currently has a model showing the changing risk selection under different financial penalties. 

Therefore, I present a series of descriptive statistics that provide an initial glance at the effect of a financial penalty on take-up rates in conjunction with a mandate in the individual market.  These statistics are only the starting point of analysis.

Employer-Based Insurance: Comparing People Who Decline and Take Up Coverage

COBRA allows those who lose their job to pay for coverage from their former employer for the next 18 months at 102 percent of the cost of coverage.   It is not popular with employee benefit managers for two reasons.  First, it requires considerable administrative resources.  Second, sicker people are more likely to take up COBRA than the overall population.  Claims expenses for COBRA enrollees are roughly 50 percent greater than the average for the employer group.

Analyzing data from the 1997 Health Interview Survey, Linda Blumberg and Len Nichols compared people taking up and declining employer-based coverage.  Decliners were much younger.  Twenty-two percent of decliners were ages 18-24 compared to 7 percent of takers.  Takers were far more likely to be married and have children.  Decliners had lower incomes and far less education (22 percent of decliners had not graduated from high school versus 7 percent of takers.)  Surprisingly, statistically greater numbers of takers saw themselves as having “excellent” or “very good” health status.  Mental health was also worse among decliners.  Decliners were less likely to have chronic conditions such as diabetes, cancer, and joint pain. As expected, takers were nearly 50 percent more likely to have seen a doctor in the past 12 months.

The preceding analysis suggests that socio-economic status plays as great a role in take-up decisions of employees as the perceived need for health services.  With an individual mandate and subsidies, decliners are likely to become takers and become part of the employer-based health insurance system.

The Massachusetts Experience

In 2007 Massachusetts became the first state in the country with an individual mandate. Today the penalty for not having insurance is roughly half the cost of the most affordable plan available to an individual. About 440,000 fewer people are uninsured today than at the time of the passage of the reform plan, and the uninsurance rate has fallen to a national-low figure of 2.7%.  The number of people with individual coverage has expanded by 45,000.

Massachusetts requires community rating in the individual insurance market with adjustments for age and geographic location.  It has merged the small-group and individual markets, but currently, most enrollees insured through the Connector have individual insurance.  Other northeastern states–New York, Vermont, New Jersey, and Maine–also require community rating and guarantee issue for individual insurance.  Hence, it is possible to examine a crude natural experiment analyzing the impact of an individual mandate on selection effects. Admittedly, other factors should be held constant, but given the proximity of the states, and the fact that Massachusetts is one of the highest-cost states in the nation, we would expect premiums to be as high or higher in Massachusetts as in its neighboring states. 

Using the Website ehealthinsurance, my son Brad, an actuary, compared the cost of individual policies.  In New York City, only “Gold” policies are available, and premiums for a 29-year-old cost about $1,100 a month.  On the Massachusetts Connector, a similar policy costs about $550 per month.   In New Jersey for $420 per month, a 29-year-old male can purchase a policy with a $2,500 deductible.  For $420 a month, a 29-year-old male in Massachusetts can purchase a policy with a $250 deductible.  In Vermont, a 29-year-old can purchase a policy for $520 per month with a $3,500 deductible.  In summary, in states with community rating and without an individual mandate, the cost of individual coverage was substantially higher than in Massachusetts in all cases.

The Urban Institute conducted extensive household surveys in the fall of 2006 (before health reform) and 2008 (after reform) in Massachusetts.  Over 3,000 adults ages 18-64 were included in the sample in 2006, and over 4,000 in 2008.  To test how the composition of the uninsured changed during the years, and to determine if healthy or young people were more likely to pay the penalty and remain uninsured, I compared the uninsured populations in the two years (Exhibit 1, at the bottom of this post). 

At a minimum, there is no clear case that the young and healthy disproportionately chose to remain uninsured.  People ages 18-25 were 42 percent of the uninsured in 2006 and 31 percent in 2008, whereas the percentage of people ages 50-64 increased from 14 percent to 20 percent.  One intervening factor may be that young adults were allowed to go on their parents’ employer-based plan up to age 26.  People with chronic conditions constituted 30 percent of the uninsured in 2006 and 39 percent in 2009.  College graduates’ share of the uninsured increased from 19 percent to 25 percent.  In the case of Massachusetts, an individual mandate plus subsidies for low- and moderate-income workers appears to have prevented adverse selection among the newly insured.

A Quick Comment on Cost Sharing

One analysis that diminishes the credibility of the AHIP report is the issue of cost shifting, the proposition that when Medicare pays less for services, providers are able to make up the lost revenue by increasing charges to private payers.  PWC sees every dollar reduced by Medicare shifted to private insurers, which in turn increases the cost of private insurance.  If the world worked this way, then few hospitals would have financial problems.  I have heard this simple theory of cost sharing expressed at many conferences and meetings of benefit consultants and hospital administrators.  Health economists are almost evenly split as to whether cost shifting, as opposed to price discrimination, takes place, but I have never met anyone who believes in 100 percent shifting. 

Cost shifting presumes that hospitals always have in reserve unused market power, and implies that competition among buyers for discounts does not work. Private insurers compete by obtaining discounts from providers.  Does that mean when Blue Cross negotiates a sizable discount, that Humana ends up paying more?   If Medicare payment levels were raised significantly, would hospitals reduce or freeze their charges to private insurers?  I believe that if Medicare raised its payment levels, hospitals would likely purchase more technology, and within a year or so this would lead to higher operating expenses and then increased charges. 

Final Thoughts

The timing of and unclear and questionable methods used in the AHIP/PWC report should not lead policymakers to dismiss the valid concerns it raises.  Insurance reforms without an effective mandate and schedule of subsidies will destabilize the individual and small-group system.  I would speculate that the Senate Finance Committee voted to reduce penalties to appease “moderates” with libertarian leanings.  When fewer uninsured take up coverage, the aggregate cost of reform declines, but the cost per insured person increases due to unfavorable selection.

Health reform is a system of inter-linked relationships. Removing one leg of the stool is not an option.  I suggest at the end of the health care reform debate, CBO analysts contribute their spreadsheets to the National Archives so future generations can marvel at how many interconnected relationships the CBO had to link.

Exhibit 1

Comparing Percent of Massachusetts Uninsured in 2006 with 2008







Fair or Poor Health



Chronic Condition













































US Citizen












less than High School



High School/some college



College graduate






Work status



Full Time



Part Time












300-499 percent of poverty