In the pending health care bills, low-income individuals and families who buy health insurance outside employment will get large government subsidies. Those subsidies vary by locale. This represents a significant implicit policy decision with enormous distributional and political consequences. I don’t think most Members or their constituents have focused on this. I think they should.

Let’s start with the Knights, a family of four with adults age 40. The family has $44,000 of income, putting them at twice the poverty line for a family of that size. The Knights do not get health insurance through their employer. The Knight family lives in Las Vegas.

Their friends the Ford family are identical in every way, except they live in Portland, Maine. They, too, make $44,000 of income, but it doesn’t go quite as far as the Knights’ $44K, because it costs almost 6% more to live in Portland than in Las Vegas, according to CNN.com’s cost-of-living calculator. Utility costs are much higher in Portland, and food prices are 2% higher in Portland.

Suppose we are designing a new national program to subsidize food for modest-income families. We have a range of choices.

1. At one extreme, both families get the same subsidy amount.
2. At the other extreme, both families pay the same net amount for groceries, meaning the Ford family gets a bigger subsidy, since groceries cost more in Portland.

Which is fair? In a system of locally-elected representatives, your answer probably depends on where you live. I can construct legitimate arguments for either extreme, or for a midpoint policy such as the Ford family getting a 2% bigger subsidy than the Knights.

Now what if people in Portland eat 5% more than people in Las Vegas? If we go with approach (2), should the Ford family get a subsidy that accounts for the higher prices in Maine and their greater food consumption? Should the two families pay the same amount for groceries if they’re eating different amounts? Again, there’s no objectively right answer. My personal preference is to favor approach (1).

Different federal spending programs take each extreme approach and many variants in between. Some adjust for regional variations; others do not.

Now let’s look at what the Baucus bill does for the new low-income subsidies to purchase health insurance outside of employment. Here is the key sentence from the conceptual description of the Senate Finance Committee-reported bill (labeled as page 27, it’s page 30 of the PDF).

The premium credit amount would be tied to the second lowest-cost silver plan in the area where the individual resides.

This is approach (2) (and it becomes clear it’s the extreme when you study the details). If you live in an area with relatively inexpensive health plans, low- and moderate-income people in your area will receive smaller government subsidies than their similarly-situated identical twins who live in relatively high-cost areas.

A “health plan” is not a commodity like “a gallon of milk.” A health plan in Las Vegas is quite different from one in Portland. While the overall cost-of-living in Portland is higher, health care spending is much higher in Las Vegas. This higher health spending is a function of different prices and different usage of medical care.

Atul Gawande wrote a much-discussed article on this topic in The New Yorker, “The Cost Conundrum:” What a Texas town can teach us about health care. There are wide geographic dispersions in medical care spending, and it cannot all be explained by different prices. While I disagree with Gawande’s policy conclusions, I recommend the article.

Since insurance premiums ultimately reflect the cost of medical care used, insurance premiums too will vary widely from one area to the next. This brings us to the effects of the policy decision in the new health care bills.

We are fortunate have a calculator created by the (liberal) Kaiser Family Foundation to do most of the hard work for us.

The Kaiser calculator makes a simplifying assumption that premiums in high-cost areas will be 20% higher than in average areas, and premiums in low-cost area will be 20% lower than in average areas. That seems like a reasonable assumption to illustrate the conceptual point.

I have chosen Las Vegas and Portland because they represent high-cost and low-cost areas respectively. Using Kaiser’s assumption, I will assume that a typical health insurance premium costs one-third less in low-cost Portland than in high-cost Las Vegas (1 – (80%/120%)) = 1/3. Note that while the overall cost-of-living in Portland is higher than in Las Vegas, per-person health spending is much lower in Portland in part because of differences in medical care usage.

Under the Baucus/Senate Finance Committee bill, both the Ford family and the Knight family will pay only $3,070 for family health insurance after netting out their new government subsidy. That’s an incredible deal for either family.

It also represents approach (2) described above. Both families pay the same amount, post-subsidy, for health insurance. Since (using Kaiser’s assumption) health insurance costs 1/3 less in Portland than in Las Vegas, under the Baucus bill the Knights in Vegas will get a $6,365 subsidy, while the Fords in Portland will get a $3,220 subsidy, 49% less than the Knights. (The Kaiser calculator gives me these subsidy amounts.)

Is this fair? One family, living in a higher-cost area, gets a subsidy twice as large as the other because of differences in medical care usage in their local markets?

The President has made a big deal about regional differences in health spending as an opportunity for making American health care more efficient while retaining high health outcomes. That’s one of the points of the Gawande article, that greater usage of medical care does not result in similar improvements in health. Minnesota is the usual example, where per person medical spending is low, while health outcomes are quite high.

Academics have done a lot of research on this. If the New Yorker article excites you, go learn about the famous RAND Health Insurance Experiment and the Dartmouth Atlas of Health Care.

The President’s push to address this source of inefficiency would lead one to design a new subsidy program favoring approach (1), equal subsidy amounts that are independent of regional differences in health spending. Approach (2) has the downside of directing greater subsidies toward areas of greater usage. When you subsidize something you get more of it, so if you subsidize areas with greater usage, you should expect even more usage. Whatever your view on the equity question, the Baucus/SFC choice of approach (2) will likely increase disparities in health spending and exacerbate the problem the President has correctly identified. This is inefficient and counterproductive.

Efficiency is not, of course, the only goal of subsidy design. If my experience is any guide, most Members of Congress (and many citizens, and all local press) will care first about whether their modest-income families are being treated fairly relative to other similarly-situated families in other areas. Again this is a matter of perspective, but I think the approach chosen in both the Baucus bill and the House Energy & Commerce bill looks terribly unfair by creating such enormous disparities in subsidy amounts. To me it looks like if you’re a modest-income family in a low-health-spending area, you’re getting shafted relative to those in higher-spending areas.

Let’s look at a few more examples, using the same example family (4 people with adults age 40, $44K income, no health insurance through their job). The Baucus bill gives this family the same after-subsidy cost for health insurance. This means:

• If the family lives in the high-cost Bronx, Chicago, Baton Rouge, Detroit, or Las Vegas, they will get a government subsidy of $6,365 to buy health insurance.
• A family with the same income living in average-cost St. Louis, Reno, or Delaware will get a government subsidy of $4,792. That’s 25% less than the Bronx or Chicago family.
• A family with the same income living in low-cost Little Rock, Indianapolis, Portland, or Nebraska, will get a government subsidy of $3,220. That’s 49% less than the Baton Rouge or Detroit family.

I ran similar numbers for the House Energy & Commerce Committee bill, the one most-discussed before the August summer recess, and got similar results. I wanted to see both the relative subsidy levels in both bills, and which parts of the country might qualify for different subsidy amounts.

I needed a way to divide the country up into low, average, and high-cost areas. The nice people at Dartmouth have done the Dartmouth Atlas of Health Care, which extensively examines regional differences in medical care usage and prices. The Dartmouth folks break down per-capita Medicare spending by geographic area. It’s certainly not a perfect proxy for private health plan premiums, but we’re only trying to divide places up into high-average-low, so I think it works fairly well for a back-of-the-envelope exercise like this one. Fee-for-service Medicare spending tends to be highly correlated with non-Medicare spending in the same area. I end up with 134 “low cost” areas, 114 “average cost” areas, and 58 “high cost” areas.

I’m sure a team of researchers could do a slightly better job, but I would bet their final list would look a lot like mine. To be careful, though, I think of this as an illustrative list of regional differences for this thought experiment. I do not claim these are the actual subsidy amounts for each region, because I have had to make and use some simplifying assumptions. The actual different amounts and regions could be determined only after a drawn-out and incomprehensible regulatory process months after enacting a new law. So while the following tables are necessarily educated guesses, I hope they illustrate the rough impacts that will result from this critical policy choice that no one is discussing.

Here are the subsidy amounts for our example family for the two different bills:

Comparison of government subsidies by geographic area

The above methodology produces the following areas:

Illustrative geographic areas for varying low-income health insurance subsidies

Do Members of Congress understand the massive distributional policy choice they are making by supporting these bills?

I’ll bet most of them don’t.

(photo credit: Christopher Chan)

The health insurance lobby, known as AHIP: America’s Health Insurance Plans, released a study last night showing that elements of the Baucus bill would make health insurance more expensive than under current law. The study is by PriceWaterhouseCoopers, and Karen Ignani, the head of AHIP, wrote a two-page memo summarizing it.

Here’s the most damaging part of Ms. Ignani’s memo (but take it with a huge grain of salt, for reasons I will explain):

The report makes clear that several major provisions in the current legislative proposal will cause health care costs to increase far faster and higher than they would under the current system. The report finds that the proposal “will increase premiums above what they would increase under the current system for both individual and family coverage in all four market segments for every year from 2010-2019.”

For example, the analysis shows that the cost of the average family policy is approximately $12,300 today and will rise to:

• $15,500 in 2013 under current law and to $17,200 if these provisions are implemented.
• $18,400 in 2016 under current law and to $21,300 if these provisions are implemented.
• $21,900 in 2019 under current law and to $25,900 if these provisions are implemented.

In fact, between 2010 and 2019 the cumulative increases in the cost of a typical family policy under this reform proposal will be approximately $20,700 more than it would be under the current system.

What the study says

The PWC study purports to analyze “several major provisions” of the Baucus bill, not the whole bill. Specifically, PWC looks at:

1. Combining mandates for guaranteed issue and community rating with a leaky individual mandate.
2. Taxes on health insurers and health providers such as drug and medical device manufacturers.
3. Whether cuts in Medicare reimbursement rates to medical care providers will be “shifted” to those buying private health insurance.
4. The “Kerry” tax on high-cost health insurance plans.

The study argues that the combined effects of these provisions would increase health insurance premiums across the board, and in some cases quite significantly.

Reaction to the study

The first two parts look decent. I lack the data to check them, but the basic analysis seems right, or at least it confirms my view of things. The Leavitt/Hubbard/Hennessey op-ed warned that insurance “reforms” that would benefit the predictably sick would also raise premiums for younger and healthier workers, and would create an incentive for you to wait to buy insurance until you get sick. PWC believes both these things would happen under the Baucus bill, based on the presumption that the individual mandate is soft and “leaky,” allowing an increasing number of people each year to avoid the mandate.

It’s also solid to assume that taxes on insurers and medical care providers will be passed through to consumers as higher health insurance premiums.

While doctors and hospital administrators swear by it, I have always been skeptical of the cost-shifting argument. If you believe that a hospital will raise the prices it charges privately insured patients in reaction to cuts in reimbursement rates from government programs, you must believe (1) the hospital has pricing power and (2) it has until now charged less than it could. (1) is quite plausible in some circumstances. I find (2) incredible. If someone has pricing power, I generally believe they will exert it. Are we to believe that providers of medical care were charging privately insured patients less than they could have before the cuts in government payment rates? I am happy to hear arguments on the other side.

The PWC study assumes that medical care providers will pass through every dollar of reduced Medicare provider reimbursement rates as a dollar of higher costs to privately insured patients. That’s absurd.

PWC assumes the Kerry tax on insurers selling high-cost health plans will be passed through to consumers. That’s a safe assumption. But they also assume that those higher costs will be distributed to those who purchase plans of any cost. That’s just silly. PWC should assume that the costs will be passed through only to those who buy high-cost plans. They also acknowledge that purchasers will change the benefits and structures of health plans to avoid the new tax, but ignore these adjustments in their calculations. Bogus.

The study’s biggest flaw

PWC, AHIP, and Ms. Ignani are careful to write that they are studying the effects on insurance premiums of four elements of the Baucus bill, rather than the effects of the entire Baucus bill. This gets watered down or even lost in the press coverage, and I imagine the political discussion will center around “Baucus bill makes health insurance more expensive.” Not coincidentally, AHIP opposes the four elements studied by PWC.

I believe the Baucus bill would make health insurance more expensive, but we can’t tell this from the partial PWC study. The PWC analysis ignores three important effects of the Baucus bill:

1. More insured people means greater demand for medical care, raising prices for both medical care and health insurance.
2. CBO thinks competition in the exchanges will somewhat reduce premiums for those who buy health insurance outside of employment. I doubt this factor is large.
3. The Baucus bill would subsidize the purchase of health insurance for those lower- and middle-income people who buy health insurance outside of employment.

I believe the first factor is the most significant source of premium increase in the Baucus bill. But AHIP likes this factor, so they left it out of the study they requested of PWC.

Chairman Baucus’ response

Chairman Baucus’ staff is emphasizing the subsidies. This is a weak response that should make other Democratic members nervous. They are, in effect, conceding that their bill makes health insurance more expensive. Sure, it’s more expensive, but don’t worry, we subsidize a lot of people so it ends up costing them less on net.

Once again, this confuses gross and net costs, and conflates reducing health insurance costs with shifting those costs onto others.

This argument fails on policy and political fronts:

• Policy: The President’s primary goal has been to slow the growth of health insurance costs. The PWC study is flawed, but its qualitative conclusions are correct: the Baucus bill would make private health insurance more expensive, not less. The bill therefore fails to achieve the President’s core policy goal.
• Policy: Subsidies are available only to those who buy health insurance outside of employment. If the Baucus bill makes health insurance more expensive, then everyone who gets health insurance through their job loses: higher costs, lower wages, and no subsidies.
• Politics: The last time I checked, more than 100 million people get their health insurance through their job or the job of a family member. Question for Chairman Baucus: How does your bill help a relatively young and relatively healthy worker who gets health insurance through work? Doesn’t your bill make this worker’s health insurance more expensive, and therefore cut his wages?
• Politics: Some of those who buy health insurance outside their job would get government subsidies larger than their wage cuts, and some Congressional Democrats think this makes these people winners. I think most Americans would say they would rather not have a policy that cuts their wages by $1 and in exchange promises them a government subsidy worth slightly more than $1. I would rather keep $1 in wages than exchange them for $1.05 of government subsidies.

The politics of this study

The politics of this study cut both ways. The headline numbers make it harder for Democrats to vote for the bill, even though the study is weak and incomplete. The timing makes it look like AHIP is trying to kill the bill by releasing the study and new ads the day before the Finance Committee is supposed to vote.

At the same time, nobody likes the health insurers, and Democrats may hope that Congressional Republicans “align” with AHIP so Democrats can have an easy-to-attack enemy alliance. Health insurers helped kill the Clinton Health Plan in 1994, but they are unpopular so nobody wants to be seen as their ally.

Foolish AHIP

AHIP’s strategy is inscrutable. If your goal is to kill the bill, fine, release a study like this the day before markup ends, and come out guns a-blazin’. But this outcome has been foreseeable for months. If AHIP’s goal was to kill this bill, they should have done this months ago.

A much better explanation is that AHIP is trying to use this study to generate support for modifying the bill. This would be consistent with AHIP’s and Ms. Ignani’s rhetoric, and with their apparent strategy to work with the White House and Democratic Congressional majorities to support legislation and try to modify it to their liking. Ms. Ignani is a Democrat and former union official inclined to work with a Democratic President and Congress. She may also be playing strategic defense, hoping that by not directly opposing legislation she can avoid an all-out war with a White House and Congress that can hurt her members in countless ways.

If this is still AHIP’s strategy, they still got the timing wrong. Washington Democrats are inclined to pick fights with the health insurers, and the timing of this release gives them an excuse to do so. Left-wing Democrats can use this move to justify shifting more of the policy pain to insurers, not less. We already saw an absurd “windfall profits tax” on health insurers floated last week.

And these provisions in the Baucus bill have been telegraphed for months. Why wait so long to go public opposing them?

Health insurers win financially if and only if final legislation includes a strong individual mandate and does not take too much directly out of health plan hides. That requires threading a tiny needle. If Ms. Ignani’s strategy backfires, she could destroy private health insurance in America.

What Republicans should do

Congressional Republicans should not embrace the AHIP study, but instead focus on the critical policy questions raised by it. Does the Baucus bill make health insurance more expensive? Does it cut wages for most Americans who today have employer-provided health insurance? If so, by how much?

Republicans should ask CBO to answer these questions about the Baucus bill, and quickly. The AHIP study opens the door to this debate by framing the questions, but CBO is the only trusted source of information to answer them.

It is important that CBO be asked the right questions. Important details can skew the answer. For instance, the Baucus bill would cause about 3 million people to lose their employer-provided health insurance. These people would end up with higher wages. The vast majority, in contrast, would see premium increases and lower wages. It is important that CBO analyze these two populations separately rather than net out the effects as they have done in their previous publications.

The PWC study is flawed in its details, but qualitatively correct in its conclusion: the Baucus bill would make health insurance more expensive for most Americans, and in doing so would mean a wage cut for most. If CBO confirms this, the bill will die.

(photo credit: Lift-off by aussiegall. No, it has nothing to do with the post. I just think it looks good.)