Dear colleagues:

This set of commentaries has been very interesting, and I'd like to think
it's what such listserves are for.  So I reorganized them
somewhat-chronologically (basically in terms of who was responding to whom)
and put them into a Word document, which I have attached.  In case anyone
wants a coherent record.

A few maybe concluding comments (last I'll say, at least):

* There are lots of reasons to object to the role of employers in U.S.
health insurance.  Uwe is essentially confirming that, at least for himself,
one of the attractions of the excise tax is that it might do what I say it
risks doing.  

The question is: what are the practical alternatives?  In the U.S. debate, a
master political assumption has been that reformers should not seem to
threaten to take away the insurance that people already have, and like.  Uwe
would like to do so; I wouldn't at all mind doing so if we could replace
that coverage with something better.  But it doesn't appear as if we can.

The excise tax in the Senate bill is designed to reduce the government
subsidy that currently exists, if an employer has a disproportionately
expensive group of employees.  Analysts who object to the tax preference
believe it is unfair because it subsidizes higher-income people more than
lower-income people.  But this particular "fix" in general reduces subsidies
for sicker people more than for healthier people.  So, as Shirley explained,
the "compromise" doesn't fit the intent of the original criticism.  It also,
logically, puts pressure on employers to reduce coverage.  To Uwe, this is a
good thing.  And, if the bills were written in a way that made it easy for
larger employers to get into the exchange system, that might be fine.  But
they're not written that way.  So the net effect will be to force employers
who have expensive groups to provide lesser-than-average benefits to those
groups.  Or, phrased differently, to force employees who are members of
expensive groups to purchase less than the average amount of coverage.

** Uwe is of course right about the reformed German system.  But if you had
asked either Uwe or me in 1990, "would you take the current German system?"
we both, I believe, would have taken it in a heartbeat.  And in that system
there was not the same choice among funds.  The reformed system appears to
be an improvement, and makes the employer less of a focus of risk-pooling.
I'll confess to not quite understanding why any employer still has its own
fund, however, unless there are some risk-pooling advantages and/or it
enables the employer to claim to be providing slightly better benefits to
its employees.  I do think Uwe and I agree that the whole idea that
employers should manage health insurance for their employees -- define
benefits that vary significantly among employers, manage networks, negotiate
prices, etc. -- is daft.  Where we might disagree is, I'd be happy to throw
in the risk-adjustments and all-payer systems and a lot of benefit
standardization on top of the current "employer-based" system and call that
a huge improvement, which would still enable reformers to claim they've
"preserved the current system."  I don't know if he would settle for that or
not.

***The question about regressivity is a matter of perspective.  The usual
definition of whether a tax is "progressive" or "regressive" is in
comparison to a flat proportion of income.  If everyone pays the same
percentage, that is a neutral tax.  If richer people pay a higher
percentage, that is progressive.  If lower incomes pay a higher percentage,
that is regressive.  What, then, is the effect of a change?  

Lets say taxes were cut, across the board, by 10%.  A person paying $40,000
before would now be paying $36,000.  A person paying $10,000 before would
now be paying $1,000.  The progressivity of the overall tax code has not
changed at all: payments for individuals remain in the same proportions as
before.  But, if  you just look at the size of the cut, higher-income people
are getting a larger cut.  The traditional analysis by many health
economists treats the tax preference as a cut to a pre-existing revenue
system, and because the cut is larger for higher-income people, Uwe and
others call that a "regressive" policy.  Karen's point is, the right
question is how any tax provision changes the overall distribution
before-and-after the change.  From that perspective, in my hypothetical
example, I think Karen would call the change neutral (which I think is
correct, in normal public finance language).

Karen's analysis concludes that eliminating the exemption "would increase
the tax burden on lower-income households, as a percentage of income and as
a percentage of current tax liability, more than it would for higher-income
households."
http://www.commonwealthfund.org/~/media/Files/Publications/Issue%20Brief/200
9/May/Progressive%20or%20Regressive%20A%20Second%20Look%20at%20the%20Tax%20E
xemption/PDF_1269_Schoen_progressive_or_regressive_ESI.pdf  This seems more
similar to the standard definitions of progressivity/regressivity, as I
understand them, than does the usual standard used by health economists,
which focuses on the dollar value of a change rather than how the effect of
the change on the underlying distribution of taxes as a percentage of
income. In political practice, people throw both definitions around.  When
Reagan cut taxes across-the-board, for example, liberals condemned that
because the tax cut was bigger for high-income individuals.  But Davis'
definition makes as much or more sense, in traditional public finance terms
as I understand them, as does the one used in the standard criticism of the
tax preference.

This does not address the more basic question of whether "regressive" is all
that bad.  A payroll tax that does not apply above a cap is "regressive."
That, of course, is the norm in sickness fund systems.  Since higher-income
people save a larger share of their incomes, a VAT is regressive.  In
practice, most scholars of social insurance, I think, would take a neutral
or even mildly "regressive" tax to pay for medical care (as in the sickness
fund systems), because the benefits are a much larger share of income for
lower income people.  There's a political science literature, for example,
on how neutral or regressive taxes (VATs, payroll contributions) are the
basis for European welfare states; and some conservatives oppose a VAT
precisely because they think it would be used to create "bigger government."
Tim Jost's point can be rephrased as follows: that, even if the tax
preference is mildly regressive, it is the political cost of providing equal
benefits to workers with very different incomes; and that is a far better
deal for the low-income workers than for the high-income workers.

Anyway, I (and, perhaps, Uwe given his long support for the German model and
his proposal of a VAT) would take mildly regressive financing to support
reform anyway.  

****The final questions to me are: IF you have a system in which employers
do provide medical care coverage; SHOULD there be any government subsidies
to those employers?  And, IF there are subsidies to employer pools, how
should they be administered and how should they vary?

I think that, if there is any insurance outside of the exchanges, it should
certainly be subsidized.  Uwe points out that the tax break should be
available for individuals if it exists for employees.  But, equally,
subsidies should exist for employees within a company that provides
insurance just as they are available for individuals who buy insurance
through the exchange.

There are two kinds of subsidies in the exchange.  The first is by risk,
because exchange premiums are supposed to be set by modified community
rating.  The second is by income, with larger subsidies at lower incomes,
phasing out to zero at an income of 4 times the poverty level.

Employer-purchased insurance partly subsidized by a tax break provides some
rating subsidy because the employer pool transfers funds from the less-risky
employees in the pool to the more-risky employees.  Because the tax subsidy
is greater for more expensive groups, there is also a rating subsidy across
employers.  Tim and I are arguing that there is also a hidden income
transfer, from high-income employees to low-income employees within a group.
As Uwe says, this is an empirical question.  It would be cool if Jon Gabel
or someone would do a survey and ask employers if they would provide the
same salary increase to everyone if they did not pay for insurance.  To the
extent we have evidence on this, I think it tends to support Tim and my
concerns.  The clearest evidence is the fact that Congress felt a need to
pass a law requiring equal treatment.  Other evidence includes unions' view
of the bargaining situation,  

So the difference here in part involves risk-aversion.  I think there is a
good chance that what many U.S. economists believe (is that an OK
formulation, Adam?) is not accurate in practice, so I'm loath to make the
test and find out that I'm right.  Now, in practice, if we eliminated the
tax exemption, and everyone bought through the exchange system, the
lower-income employees might end up with a sufficient subsidy, in the
exchange, to make up for what they lose in income.  But, if that were true,
the net budgetary effects for the federal government might be quite nasty.
The desire to avoid paying direct subsidies is, of course, one reason why
the exchange system is limited to smaller employers.   

If we're going to have some version of the current employer-based system,
however; and if the subsidy is going to take the form of the tax preference
as currently structured; then we're back to the fact that the excise tax has
exactly the wrong effect.  In the current system, the tax subsidy is a
perverse, puzzling, but moderately effective way of reducing the effects of
varied risk.  Groups with greater risk get a greater subsidy.  Cutting it
back mainly for groups with the greatest financial risk of medical costs
would mainly make the system worse.  As Uwe says, however, we're in this box
because of the system we've inherited.  

Cheers,
Joe