Hypothetical Mean

Commentary from an Actuarial and Economic Perspective

Archive for the ‘Health Insurance’ Category

The AV Calculator’s Magic $297.06

leave a comment »


The first AV Calculator was published prior to the 2014 benefit year.  Deep in the bowels of the calculator were a series of continuance tables.  In these tables, you could add up all the parts that make up health insurance claims — in-patient claims, ER claims, drug claims — and compare that to the total expected cost per member.  If you do that comparison, you will find that the pieces are $297.06 short of the total, for all “combined” continuance tables.

That’s true for the platinum tables that have higher expected costs.  It’s true for the bronze tables that have lower expected costs.  That’s true of the silver and gold tables that are somewhere in between.  To be clear, it’s not just true that all have a gap.  All have the exact same size of gap, $297.06.

Then along came 2015, when CMS introduced something called an “effective coinsurance rate” calculation its draft calculator.  All of the continuance tables were identical to the 2014 calculator, so the magic $297.06 appeared again, repeatedly.  More interestingly, somewhere around line 3200 of the code inside Excel, you will find the following:

    eff_coins = (eff_coins + 297.06 * coins) / (Worksheets(Sheetstr).Cells(88, “C”).Value)

Suddenly, the magic $297.06 that was fixed for any benefit tier in 2014 becomes a function of the underlying coinsurance for the benefit plan (the variable “coins”).  The draft 2015 calculator was never approved for use, so no harm, no foul, but set the seeds for …

… the 2016 Calculator , the odd code that appeared in the draft 2015 calculator appears again, this time around line 3300 (search for “+ 297.06”).  This time, the calculator goes into production.

Even worse, for 2016 ER costs in the continuance tables were increased by more than 13% (6.5% per year for two years).  IP costs went up similarly.  Physician costs went up similarly.  In fact, every component of health spending for every benefit tier went up by about that same amount … except for the poor, lonely, uncategorized $297.06.  This is why when you measure annualized trend by category from the continuance tables, you will see roughly a 6.5% rate applied to each category … but the total increases by something lower than that, 6.2% and change, depending (see below).

And now, at long last, we have a draft 2017 Calculator .  Again, each component of the continuance tables went up 6.5%.  So, naturally, the first thing I check in the new calculator is the magic $297.06.  Would it survive another wave of trend?  Would it remain constant across benefit plans?  Would they continue to apply coinsurance to it in the macro, despite not allowing the underlying richness of the benefit tier to impact it?

Yes, YES, and YESSS!

This, my friends, is why you don’t set your premiums based upon the AV calculator.  Because outside of Washington DC bureaucracy, there is nothing magic about $297.06 that makes it work for every benefit, every year, where you can simultaneously *apply* coinsurance (within the effective coinsurance rate calculation in the macro) and *not apply* coinsurance (the same gap exists in each metallic tier).

And in case anyone is interested, there’s a parallel issue within the separate medical/rx portion of the calculator.  Perhaps that will be fun in a post I’ll do for 2018.

A summary of all of the relevant continuance tables, trend values, and the calculation of this magic $297.06 are below the fold.

P.S.  Despite the sarcastic tone in this post, in all seriousness if there’s a rational explanation for what this $297.06 represents, I would like to hear about it.  If there is such a rationale, it should be a prominent part of the documentation.


Read the rest of this entry »


Written by Victor

December 9, 2015 at 3:31 am

The CBO and Large Group Premiums

leave a comment »

Lower premiums, per person, frequently mean higher premiums paid by families.  This counter-intuitive fact leads commentators like Ezra Klein to a common misinterpretation of the CBO scoring of the Senate healthcare proposal:

When the Congressional Budget Office looked at this question (pdf), they found that for Americans in the large-group market (134 million of us), premiums would go down by 1 to 3 percent.

That is not a fair description of what they found.  The CBO found that the premiums, per member, would go down.  They did not find that the premium rates people pay will go down.  There is a large and substantive difference between the two.  The primary reason for the CBO finding is higher enrollment rates as employees and their dependents attempt to avoid mandates and the Exchanges.  Both of those actions, however, are likely to increase premium *rates* and the cost of health insurance for families, a point that is conveniently overlooked.  Here’s how this happens.

Scenario 1: A young employee, currently uninsured, buys from his company because of healthcare reform.  Assume that average premiums, per person, drop from $300 to $295 for the company (like the CBO finding).  The company’s overall healthcare burden, however, has increased because they are now paying for an additional person.  Therefore, if the company keeps its total contributions the same, there is less money per employee, meaning that the amounts paid by workers has to increase to compensate.  Only if the company increases its total contribution by more than the increased cost of the additional insureds will each family’s burden decrease.  But then, of course, the company is paying disproportionately more in health benefits, putting downward pressure on wages.  The CBO models and commentary quit at the per person per month costs within each segment, and don’t tell us how companies are expected to adjust their contributions in response.  But either way, families that currently purchase insurance are clearly squeezed when you work through the ramifications.

Scenario 2:  The spouse of a worker signs up for the plan to avoid the mandate.  The number of enrollees per employee clearly goes up in this scenario.  Therefore, even though the per person cost is lower, the per employee costs goes up.  The resulting increase in paid costs per worker is now an obvious and inescapable conclusion.  Will the employer pass this higher cost per worker on in terms of reduced wages or higher required healthcare premium contributions?  Either way the families that are currently insured will be hurt.

The core logical problem is that the CBO is reporting average premiums per person.  But no one actually pays health insurance on that basis.  Premiums are paid on a per contract basis (per employee in the group market).  The amount you are charged, a premium “rate”, is determined based on the type of contract you have, and after the employer determines an amount they will contribute toward that rate.  Higher or lower per person per month costs influence this amount, but you aren’t done with the calculation at that point.  You still need to calculate premium *rates*, after employer contributions, to determine the impact on actual people.  And, as noted in this post, the rates can move counterintuitively relative to the premiums per person.

Written by Victor

March 10, 2010 at 5:05 am

A Public Challenge

leave a comment »

I hereby publicly challenge defenders of the CBO healthcare model to address the following:

a) Produce 2010 and 2016 state-specific non-group rates, before and after reform, so that an apples-to-apples comparison on premiums after reform can be demonstrated against premiums for sale in the marketplace today.

States like Arkansas today have rates 50% of the national averages being modeled by the CBO.  After reform, the regulatory changes will push all states closer to the subsequent national averages.  The maximum geographic variation in the CBO models appears to be 0.8 to 1.2, meaning that after reform, Arkansas will probably have rates equal to 80% of the national average.  That represents a 60% increase in premiums in the state of Arkansas.  This is BEFORE any effect modeled by the CBO with respect to reform’s impact on the national average itself.

b) Provide 2016 non-group premiums divided between the grandfathered and Exchange blocks.  Currently, the CBO is advertising that the national average COMBINATION of policies issued under current rating rules (grandfathering) and future rating rules (Exchange) will only increase by 10-13%.  Obviously, it is likely that the grandfathered plans will only see part of this increase (parts due to taxes and similar provisions), while the Exchange will be subject to higher rates and adverse selection.

c) Provide 2016 subsidies by state, by FPL category.  States like Arkansas, even after reform, may have premiums 33% lower than other states, simply because of cost of care differences.  That means that low-cost states like Arkansas will get fewer subsidy dollars per enrollee than states that have out-of-control costs.

d) Provide justification for the assumption that non-Exchange individual policies will remain steady or grow between 2014 and 2019 under reform.  It will be illegal to sell grandfathered policies.  Why are their models producing sales outside of the Exchange into 2019?  Most policies issued today terminate when you move across state lines; termination of grandfathered products will result from migration, if nothing else.  The concern here is that their Technical Documentation does not seem to allow for the actual features of individual products, namely their lifetime duration and exit provisions.  One concern is that their model may not be incorporating lapse rates within that block, meaning they are understating enrollment in the Exchange, thereby understating the cost of the bill.

None of these challenges should be interpreted as a professional statement that I am ceding that the CBO projections are otherwise reasonable.  Rather, I offer these challenges so that if they are ever taken up, all of us will either see the limitations of their work more clearly, or so that these fairly obvious concerns can be abated.  It is stunning to me that they did not provide greater specificity in response to Evan Bayh’s requests for greater information.

Written by Victor

December 3, 2009 at 12:17 pm

Rate Reductions Can Increase Your Costs

leave a comment »

The CBO’s 11/30 premium analysis of the Reid healthcare bill estimates a 0% to 3% reduction in premiums per employee for large firms.  This has been interpreted as positively impacting large employers and their employees.  This interpretation is likely false.

A primary driver of this reduction in per-employee premiums is the addition of healthy employees and dependents driven to coverage because of the mandate.  This means that more employees and dependents will participate in the health plan.  Average costs may fall, but total costs will increase.  That will force employers to make budget decisions which, in turn, will likely increase the premium costs to individuals.

My discussion will focus on an illustrative example: a firm with 100 employees, 80 of who take up coverage.  The premium rates charged this groups are $200 per employee per month, and the employer pays 75% or $150.  The employer pays $12,000 per month ($150 * 80 employees).  Each employee pays the remaining $50.

Under reform, let’s assume that all 100 employees participate in the plan, driving the average cost downward by 2%, so the actuarially-fair rate is $196 per employee per month.  The employer continues to pay 75%, which is now $147.  The employer now pays $14,700, a 22.5% increase in their healthcare bill!

One response worth discussing would be if the firm keeps their $12,000 total contribution frozen.  This would translate into $120 per month off of each employee’s premiums.  This would result in the employee premiums increasing from $50 to $76, a whopping 52% increase.

Significant care must be taken to interpret the CBO numbers correctly.  Unfortunately, the CBO paper itself isn’t sufficiently detailed for us to discern the size of the impact I discussed above, nor how they model the employer contributions to health plans.

For the economics geeks out there, if you are still reading, you’ll recognize that the problem here is that the implicit wage reduction funding the healthplan is paid by all employees, yet the benefit is limited to a subset of people who sign up for the plan.  This results in cross-subsidization between workers.  The CBO is telling us that this cross-subsidization may fall after reform, with a resulting dynamic change in the implicit wage reduction.

(Note: they also tell us that the worst risk may bail on the employer healthplan, and go to the exchange.  This may happen for low-paid workers eligible for cost-sharing subsidies.  The fact that this may happen for some employers does not mitigate the fact that the effect I describe above will also happen for some employers.  They also tell us that large groups with better-than-average costs may be grandfathered, leaving the worse-than-average groups to participate in the Exchange-based pooling; somehow this does not impact large group premiums or cause an increase in the pooled rates insurance companies must charge).

Written by Victor

December 2, 2009 at 10:48 am

Playing “What If” with the CBO numbers

leave a comment »

What if the CBO was more than $90 billion off its cost estimate for HR3962 (blog entry here)?  I think it may be, and it is easy to explain why.

Yesterday, I pointed out that the projected non-group enrollments outside of the Exchange were highly implausible.  Specifically, since individual new-sales outside of the Exchange would be made illegal, it seems illogical to project a static or increasing population of individual policy holders outside of the Exchange.

In today’s marketplace, people tend to cycle through individual coverage; as many as 35%-40% of policies terminate before they reach their first anniversary.  We could take all of the CBO assumptions, but simply assume that only 25% of individual policies lapse annually.  The result would be more than $90b in additional costs between 2015-2019, as illustrated in the table below.

The first three lines in the following table are straight from the various CBO reports.  The fourth line simply assumes that 25% of the policies that existed in 2015 lapse in 2016, and so forth through 2019.  I then assumed that those who lapse join the exchange at an average cost equivalent to those already on the Exchange.  The result is a whopping $91 billion difference.

This illustrates two things.  First, we have very little idea what is really going on inside the CBO models, and results this bizarre suggest that more peer review is warranted.  Second, even if the initial CBO model is perfectly fine, tiny tweaks in assumptions can result in massive swings in cost for this sort of program.  I would argue that this program design is inherently risky; the only way to keep the costs even remotely under control is to build a series of firewalls between employer-based coverage and individual coverage (less than 20% of people are even eligible for the Exchanges let alone the subsidies).  If any of these firewalls leak or if any of these assumptions are slightly wrong, we could be in serious fiscal difficulty.

2015 2016 2017 2018 2019 TOTAL
Cost of Subsidy ($b) 82 96 103 111 120 512
Baseline Non-Group Enrollment (m) 14 14 14 14 15 71
CBO Non-Group/Other Enrollment (m) 8 8 8 8 9 41
Reform Mkt w/ 75% Retention (m) 8 6 5 3 3 24
Add’t’l Subsidy Cost ($b) $0 $10 $18 $26 $38 $91
Avg. Subsidy/Subsidized Enrollee 5500 5800 6100 6500 6800

Written by Victor

October 31, 2009 at 2:35 am

CBO weirdness with HR3962

with one comment

Compare and contrast Table 2 from the CBO analysis of HR3962, the new House healthcare bill.  Specifically note that non-group /other net enrollment increases from 24 million to 30 million between 2015 and 2019.  Now compare this to the bill itself, Title II, Section 202, Page 94:

Individual health insurance coverage that is not grandfathered health insurance coverage under subsection (a) may only be offered on or after the first day of Y1 as an Exchange-participating health benefits plan.

How can enrollment go up if new enrollment is disallowed outside the Exchange?

The problem is further exacerbated when you realize the new dependents born after “Y1” (A-1-B, page 91) are not allowed onto their parents’ plans and that individual business generally has a 30-40% lapse rate annually.  Granted, people may not be willing to part with the relatively cheaper grandfathered plans, so the lapse rate is likely to fall.  But it can’t go negative.

Perhaps the circle is squared by the ephemeral “other” category in the analysis.  But I doubt it.  Compare the HR3962 score with the Baucus bill score, which separated “other” from “non-group”.  In that score, those two categories moved in tandem under “current” law, and it seems reasonable to presume that the “other” category is mostly unaffected by reform (I presume it is Medicare disability, etc.).

There are many such examples like this, where you just want to shake the CBO and ask “What are you smoking?  And … Can I have some of that?”  I think I’ll need it.

Written by Victor

October 30, 2009 at 1:50 am

Once Covered, Always Covered

leave a comment »

Tom Maguire at “JustOneMinute” came up with that catchy phrase — Once Covered, Always Covered — to discuss the main component of my desired healthcare reform approach that I began to outline last year. With a catchy phrase in hand and with no hope of having any impact on the current healthcare deform bills being discussed today, here’s a talking-point listing of some of the components of my reform ideas.


Portability of one’s insurance is the most easily addressed component of reform.  That’s where the Once Covered, Always Covered approach kicks in.  By law, everyone who is currently covered could switch coverage if they move, if they leave their job, or whatever.

All plans would be assigned an “actuarial value” and you could seamlessly turn in a group coverage plan for an individual-market plan as long as the newer plan was at the same actuarial value or lower.  Currently, the transition from group coverage to individual coverage is non-standard across states.  The transition to group coverage is covered by HIPAA, but has some holes.  Lastly, a Once Covered, Always Covered approach would eliminate administratively costly COBRA provisions.

All reform proposals carry with them tradeoffs; any pretention otherwise is dishonest.  The downside of this approach is higher “new business” individual rates.  State insurance departments would have to be vigilant to prevent carriers from capriciously entering a market, buying business, exiting the market, and then raising rates rapidly.  I have solutions in mind to minimize these problems, but they would get technical and wouldn’t entirely solve the problem.  But this is a tradeoff that I think many would willingly make.


If people are denied coverage today, they have guaranteed-access to state-run high risk pool insurance.  I would build on this in one of two ways.  First, we could federally subsidize the pools and streamline access.  Alternatively, we could work through the private insurance system entirely and have the federal government reinsure/subsidize high risk individuals.  Expected costs for initially sick individuals lower over time just as expected costs for initially healthy individuals rise over time.  I would taper the subsidies to mirror this reverse “underwriting curve”.  In either case, the federal subsidy should not fully pay for the increased cost of these individuals; there has to be some penalty for remaining outside the system until you become sick.  The penalty in today’s world is the administrative hassle to get into a high risk pool as well as the relatively high premiums in them.


No one has a good idea how to rein in cost in the current system.  Therefore, the only way we can be certain to lower costs is to reduce what is covered.  I would divide claims into four categories.  1) Claims that would be continued to be paid under private insurance, 2) claims that are not socializable and therefore will not be allowed to be paid under private insurance, 3) claims that fulfill a well-defined public objective, 4) non-essential medical services.  Examples of the latter two cases are as follows.  Examples of “non-socializable” claims are those that aren’t clinically more effective than cheaper and available alternatives, drugs advertised direct to consumer, claims that carry a disproportionate personal benefit and therefore don’t need to be socialized, like Chantix for smokers (Chantix is cheaper than cigarettes).  Examples of claims that fulfill a public objective include vaccinations, preventive health measures, etc.  I would also argue that maternity claims (which aren’t “insurable”) should fit into the category, including the extremely high cost and ethically electric issues of premature baby costs.  Tax revenue would have to raised to pay for this last category of benefits.  Non-essential medical services would include chiropractic services, mental health benefits, etc.  These services could be included in healthcare plans, but plans without these services must be sold to make sure coverage for essential services is affordable.

By focusing private insurance on just those sorts of claims that are medically necessary and that can be insured and socialized across individuals, we can significantly reduce the cost of coverage.  We need to begin a wave of “reverse insurance mandates” because we need to make truly necessary care affordable.  And if you want to tack federal subsidies onto these leaner.

I could go on, but what are the odds that anyone is actually reading this, let alone someone who could make this healthcare deform movement make some sense?

Written by Victor

October 26, 2009 at 11:45 pm