When
Governor Mark R. Warner introduced his proposed tax
plan three weeks ago, he closed his presentation
with a powerful statement: Despite raising new tax
revenues of roughly $500 million a year, 65 percent
of all Virginia taxpayers would wind up paying less
in taxes. The political calculus was brilliant. Not
only would the governor funnel an extra $715 million
into K-12 schools and many millions more into
protecting
the state's coveted AAA bond rating, the fact that
nearly two-thirds of all taxpayers actually stood to
benefit would nullify the inevitable
objections of tax-phobic Republicans.
The
numbers struck some observers as improbable,
however. Columnists Barton Hinkle and Ross McKenzie
at the Richmond Times-Dispatch questioned whether the governor was engaging in some
budgetary prestidigitation. Writing in Bacon's Rebellion (See
"White
Men Can't Add," December 1, 2003), Paul
Goldman suggested that the only way to make the
numbers add up was to claim credit for phasing out
the car tax -- a tax cut already promised voters. Then
reporter Christina Nuckols at the Virginian-Pilot
extracted some crucial concessions from the
governor himself. First, the 65-percent number
did not include the impact of a proposed 22.5-cent
hike in the cigarette tax. Second, the governor used
a number of educated guesses and statistical
assumptions. But lacking access to detailed numbers,
these observers could take their analysis only so far.
Remarkably,
no one else seemed terribly interested in how the
governor determined he could raise taxes by $500
million a year yet have 65 percent of Virginia
taxpayers come out ahead. The tax
plan was devilishly complex. The governor proposed
raising the sales taxes on some goods and lowering
it on others, restructuring the income tax with multiple
changes to thresholds, exemptions and income brackets, and finishing the
phase-out of the car tax, which involved reimbursing
taxpayers at different rates around the state. How was it possible,
I wondered, to calculate the
impact on millions of taxpayers with any degree of
precision? Was the governor using hard numbers, or
was he throwing out a back-of-the-envelope
calculation and hoping no one would dig too deep?
Friday,
I sat down with Finance Secretary John Bennett for
some answers. For more than an hour, Bennett
patiently walked me through his calculations and
shared extensive financial data that no one else has
reported. He was totally
forthcoming, answered every question I posed to him and
candidly laid bare
the limitations to his methodology. Here are the key
conclusions I drew from that discussion:
-
Bennett
is making reasonable and defensible assumptions
regarding the impact of the governor's proposed
changes to the income tax and sales tax.
Furthermore, he has built in
a $50-per-taxpayer margin for error. If, by his
calculations, a given taxpayer gained less
than $50, he or she was not included among the
beneficiaries.
The
bottom line: The 65-percent number accurately reflects those
portions of the tax plan that can be measured but
omits significant portions -- totaling $175 million,
equivalent to 36 percent of total net impact --
whose effects cannot be measured. Change the
assumptions and the "65 percent" number
evaporates.
State
Income Taxes
Bennett's
method for calculating who benefits from the tax
plan is quite clever. He starts with a database,
maintained by the state Department of Taxation,
which contains the files of 2.9 million residents in
2001, the latest year for which the files are
available.
Bennett
readily concedes that taxpayers in 2001 are not
identical to taxpayers in 2005, but most of them
will be the same. For calculating the effects of the
Warner Tax Plan, he says, the database provides a
very close approximation of Virginia's taxpayer
profile.
Bennett cranked Warner's
proposed tax changes -- new thresholds, exemptions,
deductions and tax brackets -- through the tax filer
database and calculated a
precise impact literally taxpayer by taxpayer. Using
this technique, not only can he figure out the gain
or loss in tax revenues, but he can count an exact
number of winners and losers.
Very
cool. So far, I'm very impressed by Bennett's logic.
Sales
and Use Tax
There
is no way to know exactly how much every individual
pays in sales taxes. The data simply aren't
collected. However, the federal Bureau of Labor
Statistics conducts an annual household expenditure
survey, which breaks down spending patterns in
households by 11 income classes ranging from $5,000
to over $90,000. The survey breaks down spending on
14 different categories including groceries.
From
that data, Bennett developed a regression equation
that relates income to spending on taxable items. He
used the equation to estimate food and non-food
expenditures for every household filing a tax
return. Then he imputed a sales tax to those
expenditures, and calculated what the payments would
have been before and after the Warner tax plan. In
the final step, he applied the expected gain/loss to
each individual tax file.
No
problems here. The methodology isn't perfect, but
it's based on real-world data. For the most part,
Bennett's estimate probably reflects reality pretty
closely.
Car
Tax
The
issues are trickier when it comes to estimating the
impact of phasing out the car tax. The state
reimburses taxpayers based on the value of their
cars -- capped at $20,000 but ranging down to less
than $1,000. Not all households own the same number
of cars. Furthermore, tax rates vary from locality
to locality.
Due
to the complexities, Bennett made a series of what
he contends are conservative assumptions. First, he
assumed a maximum of two cars per household. Because
many households in actuality own more than two cars,
he figures he under-
estimated
the tax benefit of the phase-out for many
households. Secondly, he based the benefit on only
two years of phase-out by 2006 -- equivalent to 15
percent of the tax -- rather than the full 30
percent when the phase-out is complete.
On
the other hand, Bennett knew that applying an
"average" car tax benefit based on a
statewide average would be meaningless because
reimbursements vary so widely across the state.
Therefore, he calculated an average tax payment locality
by locality. To Fairfax County taxpayers
residing in his data files, he applied the average
Fairfax County reimbursement. To Bath County
taxpayers, he applied the average Bath County
reimbursement.
Here's
where I start having misgivings. Bennett clearly was
doing the best he could with the data he had
available, and he tried to apply cautious
assumptions. But was that good enough? The fact is,
even in an affluent locality like Fairfax County --
home to one million people -- there is tremendous
disparity in incomes. There are a lot of affluent
telecom executives tooling around in BMWs worth way
more than $20,000. And there are loads of Salvadoran
construction workers driving $3,000 jalopies. By
applying a county-wide average, Bennett's
methodology overstates the benefits of the car-tax
phase-out to the Salvadorans and understates the
benefits to the Beamer-driving yuppies. This bias
inflates the benefits of the tax plan to low- to
middle-income taxpayers.
Other
Taxes
Bennett
readily concedes that his car-tax assumptions are
imperfect. He just defends them as the best he can
devise. I'll give him credit: I didn't raise a
single issue that he hadn't thought about already.
Furthermore, he's very open about acknowledging the
limits of the data. His defense is that he's built
multiple layers of conservative assumptions into his
methodology. Even if you're squeamish about his
car-tax assumptions, he says, he's still built a $50
cushion into his tally of winners. If, according to
his methodology, a taxpayer would save only $49, he
got numbered among those who would not
benefit.
I
can respect that. But look what Bennett left out.
Increasing the cigarette tax to 25 cents per pack
would raise $145.8 million in fiscal 2006. But the
finance secretary did not crank those numbers into
his calculations. Why not? Because only 25 percent
of adult Virginians smoke, he says, and he couldn't
find any data that would allow him to correlate
those smokers with taxpayer status. So, he left them
out of the equation. If someone could present him
with decent data, Bennett told me, he would
willingly include the tobacco tax in his number
crunching.
I'm
sorry, but I don't buy his reasons for omitting the
cigarette tax. According to the
American Lung Association, 36.7 percent of people
with high school educations (or less) are smokers,
versus 11.9 percent with B.A. degrees or more.
Similarly, according to the National
Center for Chronic Disease Prevention and Health
Promotion, the least educated education people
are twice as likely to be heavy smokers -- 25
cigarettes per day or more than the best educated.
If
we assume that roughly 750,000 of Virginia's
taxpayers are smokers, they share the burden of
paying that $145.8 million in cigarette taxes -- or
close to $150 per person. Of course, heavy smokers
will pay even more. These heavy smoking taxpayers
will come disproportionately from the lower income
groups that Bennett has counted among the
beneficiaries of the tax plan. The heavy burden of
the cigarette tax will overwhelm his $50 safety
margin -- pushing tens, maybe hundreds, of thousands
of taxpayers into the loser category.
Finally,
there's the issue of eliminating the $6,000 age
deduction -- a tax break worth more than $300 -- for
taxpayers who turn 62. In fiscal 2006, closing that
loophole will deprive near-elderly taxpayers out of
nearly $29 million. The number will grow as new
cohorts of roughly 50,000 to 60,000 Virginians reach
62 in each successive year. Now, I don't see how
someone turning 62 is any more deserving of a
special age-related tax break than someone turning,
say, 50 (my age), so I have no problem with the
repeal. But it is disingenuous not to count this
measure in the calculation of winners and losers. If Bennett had
included these people, he'd have to chalk up several
tens of thousands more taxpayers in the loser
column.
In
sum, I believe that Bennett got close to the
mark with most of his calculations. But his decision
to omit key elements -- cigarette taxes and
eliminating the turning-62 deduction, worth a
combined $175 million -- was indispensable to
attaining that magical 65-percent number. There may
have been methodological problems, as Bennett
argues, with including those tax burdens in his
taxpayer database. But one is equally entitled to
dispute the methodological soundness of a
calculation that leaves them out.
--
December 1, 2003
|