The revenue projections
were generated by an economic model developed by the author for the Mackinac
Center for Public Policy. The model differs from the standard, "static" models
commonly used in government agencies, in
that it attempts to estimate the actual dynamic behavior of individuals when
confronted by changes in tax rates.
The baseline assumptions  for current law include:
SEV growing 9.7% in 1993
(coming off the "freeze" during 1992), 5.8% in 1994, and 6.5% annually in the
accounting for 2% of SEV growth annually;
Millage rates which
continue to grow under current law; and
An assumed state
reduction in "Homestead" or "Circuit Breaker" Income Tax Credits of 14% of the
gross property tax reduction.
Legislative and Local Government
The model does not assume
that any other companion tax reduction legislation will pass, beyond that
necessary to implement "A." In particular, it
does not include Senate Bill 146, which would create a one-year lag in
assessments in 1994. This is a key difference between this independent study and
the "consensus" revenue estimates adopted by the staffs of the Department of
Treasury, Senate Fiscal Agency, and House Tax Committee.
School millage rates have
been increasing under current law and indeed are a major justification for the
proposal. As discussed in the text. the proposal would probably result in more
taxpayer control of school finances, and slower-growing millage rates. Nonetheless, to
make the tax-reduction projections more conservative, school millage rates were
assumed to grow at 3/10 a mill each year through 1995 under the proposal, 50%
faster than the 2/10 a year rate assumed under current law.
There are three explicit
dynamic effects in the model. The first is tax capitalization, discussed at
length in Appendix II. A provisional tax cut is calculated, using a simple
static analysis. The provisional tax cut is then capitalized, by
multiplying it by the capitalization ratio. This capitalizedtax cut is then added to the baseline True Cash Value (TCV), and,
multiplied by the assessment ratio, becomes the (unlimited) SEV for the "A"
projection. This SEV will be larger than for current law, so the gross tax cut – the difference
between current law and projected tax revenue – will be smaller than a static
analysis would indicate.
second area of dynamic influence is the projection of additional tax revenue
from other taxes, on additional disposable income left from the property tax
cut. A fraction of this is assumed to be taxed, after incorporating the
avoidance factor for the sales and use taxes. The third dynamic effect is the
tax avoidance behavior of consumers.
Assumptions for dynamic effects in the model include:
Turnover (sales of property) of 10% a year.
School millage rates which grow faster under "A" than under current law for the
first few years, both because of fewer "Headlee" rollbacks, and the possibility
that school districts would attempt to regain some of the reduction in millage.
avoidance by Michigan taxpayers, who are assumed to reduce their tax payments by 1/10th the proportional increase in the sales and use tax rates. This
"elasticity" is assumed to the 2/10 in 1993, when consumers could easily
accelerate purchases of big-ticket items to avoid the sales and use tax
Capitalization Ratio of 8, equivalent to about an 12% cap rate.
capitalization ratio, tax elasticity, and other dynamic factors used in the
model are conservative. A full dynamic model would attempt to include changes in
employment, investment, and other factors, and would likely generate larger
dynamic changes than those indicated here.
assessment growth cap presents a difficult modeling problem. The algorithm used
here to project SEV under the assessment cap works as follows:
unlimited SEV series is created, based on the True Cash Value,
as augmented by tax
share of the current year SEV, equal to 1/T where T = the turnover period,
is calculated as
equal to the same share of unlimited SEV. This accounts for the newly sold
remaining (T-1)/T share is projected from last year's SEV, by adding new construction and increasing it at the limited rate.
and all models, suffers from the fact that averages and totals were used. rather
than truly aggregating all the various local units of government. This will
always introduce some error.