National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
SYMPOSIUM ON THE DEVOLUTION REVOLUTION
ISSUES RAISED BY THE
NEW FEDERALISM
STEVEN D. GOLD
*
THE CONTENT OF THE DEVOLUTION
REVOLUTION1
INTRODUCTION
The November 1994 elections reignited
interest in many fundamental issues.
The assignment of functional roles to
local, state, and federal governments,
the appropriate balance between the
public and private sectors, and the types
of grants provided by the federal
government all could be significantly
altered. As this is written in early 1996,
it is not clear whether major policy
changes will be enacted before the
November elections, but the probability
appears to be high that federal policies
will be significantly altered no later than
1997.
Devolution refers to passing responsibility down to another level of government. It is difficult to discuss for several
reasons. First, it is inherently complicated, with numerous important effects
on states and citizens and major
differences in how various states will
be affected. Second, as the cliche says,
the devil is in the details. Some approaches to reducing federal aid and
increasing state flexibility have very
different effects than others. Third,
the welfare and Medicaid systems
would be fundamentally altered,
confronting state governments with
new choices. There is enormous
uncertainty about how they would
respond. Fourth, the time horizon
matters: the short-run effects of
devolution are likely to differ considerably from the long-run effects.
This paper addresses the implications of
these developments for the country and
for public finance research. In terms of
research needs, it is concerned with two
primary questions: How can research be
improved so that it is better able to
explain the policy choices that state and
local governments are making? What
can be done to make research more
useful to state policy makers?
*
The New Federalism embodied in the
Republican revolution has three major
aspects affecting state and local
governments. (1) They would receive
less federal aid. (2) Some of the most
important aid programs would be
changed from matching to
nonmatching grants. (3) States would
have more flexibility in operating
Urban Institue, Washington, D.C. 20037.
273
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
NATIONAL TAX JOURNAL VOL. XLIX NO. 2
programs. A major uncertainty is the
size of the aid reductions in relation to
the benefits from increased flexibility.
state revenues will be about five percent
lower than they would have been in
that year. Spreading this reduction over
seven years, we can say that total
revenues will grow about 0.7 percent
less per year than they otherwise would
have.
Aid Reductions
The driving force in reducing aid is the
effort to balance the federal budget by
the year 2002. Aid to states will bear a
disproportionate share of the spending
reductions needed to balance the
budget because the three largest
spending categories—Social Security,
defense, and interest payments—are
likely to be cut relatively little if at all.
In the reconciliation bill vetoed by the
president in late 1995, aid would have
been reduced approximately 26 percent
in 2002.2 If there is a budget agreement, the eventual aid reduction will
probably be less than 26 percent
because the federal tax reduction will be
smaller and more optimistic economic
assumptions have reduced the prospective deficit in 2002. But the aid reduction would still be substantial, probably
on the order of 15–20 percent.
Whether states can absorb this reduction without having to reduce services
or increase revenues depends on how
fast current services spending and
revenue are growing. If the growth rate
of revenue were at least 0.7 percent
higher than the growth rate of spending, states could adjust to the loss of
federal aid without having to cut back
spending or raise taxes and fees.
However, it appears that most states are
not so fortunate. Indeed, many seem to
have structural deficits. (A common
strategy for states has been to shift their
fiscal problems, such as by relying
increasingly on tuition to fund higher
education, cutting back on aid to
schools and other local governments,
and pushing costs into the future. Each
of these shifts represents a failure to
maintain current services).
This projected reduction is from what
aid would have been under current
federal law. In nominal dollars, the total
amount of aid would increase. It is
important, however, to disaggregate
aid. The entire nominal increase
between 1995–2002 is attributable to
Medicaid, which accounted for about
40 percent of total aid in 1995.
Excluding Medicaid, aid would fall in
nominal dollars. It should also be
recognized that the baseline from which
the aid reductions are calculated does
not reflect the impact of inflation on
discretionary programs in some years.
This does not imply that states face an
immediate fiscal catastrophe. In fact, aid
reductions will be gradually phased in,
so they will not be particularly dramatic
at first but rather will grow steadily over
time. Besides, many states have accumulated substantial reserve balances
that can provide a cushion to help them
absorb aid cutbacks at first. But those
balances represent nonrecurring
revenue, and they would disappear fast
if a recession occurred.
Some observers may discount long-term
projections in the belief that policies
would probably be changed before
2002, resulting in smaller cutbacks.
Whether that assumption is accurate
depends, in part, on whether a constitu-
Federal aid represents between a
quarter and a third of state government
resources.3 A 15–20 percent reduction
from what aid would have been in 2002
implies that, other things being equal,
274
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
SYMPOSIUM ON THE DEVOLUTION REVOLUTION
tional amendment to require a balanced
federal budget is adopted. If it were, it
would be much more difficult to avoid
large aid reductions in the out-years.
needed to stimulate the production of
services that produce external benefits.
It appears that the externality argument
for welfare and Medicaid spending is
being rejected. People seem not to
accept the idea any longer that there
are production externalities—for
example, that poverty breeds crime and
that welfare tends to reduce it. At the
same time, belief in consumption
externalities appears to be waning—the
idea that one person’s welfare is
affected adversely if others are in
poverty.6
Another imponderable is how well the
economy performs. The federal
government’s projections assume that
eliminating the deficit would raise the
growth rate of the economy. If that
occurred, it would help states in two
ways—federal aid reductions would not
have to be as large because the federal
deficit would be smaller, and state tax
revenue would grow more. Assumptions
about stronger economic growth rate
(and the positive effect of deficit
reduction on the level of interest rates)
are, however, extremely uncertain.
Another important implication of
adopting block grants is that the
amount of financial aid provided by the
2
federal
government to states will not
automatically respond during a recession. States usually experience serious
2
fiscal
problems when the economy
contracts, because tax revenue is
depressed and the caseloads of programs such as Medicaid and welfare
grow substantially. During the most
recent recession these developments
were partially offset by a large increase
in federal aid. Between 1990–1992,
state spending rose 20.4 percent while
state tax revenue was growing only 9.1
percent. A 34.3 percent federal aid
increase made it much easier for states
to avoid deficits or painful spending
reductions and tax increases (Gold,
1995a). Although some Congressional
proposals include contingency funds to
augment certain block grants if unemployment rises sharply, states would not
enjoy a large increase in federal aid
during the next recession as they have in
the past.
Shift from Matching to Nonmatching
Grants
At least as important as the amount of
grants is their form. By proposing a shift
from open-ended matching grants to
block grants for Medicaid, Aid to
Families with Dependent Children
(AFDC), and some smaller programs, the
reforms incorporated in the budget
reconciliation act would have significantly decreased the incentive for states
to spend their own money on the
1
affected
programs. Since 1965, the
federal government has paid from 50–
83 percent of the cost of Medicaid and
AFDC, so it cost states only 17–50 cents
1 increase their spending for those
to
programs by a dollar. Under block
grants, it would cost a dollar for states
to spend a dollar.4 Thus, the devolution
revolution would reduce not only
federal spending but also state spending. It is really a prescription for lower
government spending.
The automatic growth of federal aid
caused by the recession of the early
1990s suggests that the full implications
of switching to block grants will really
not be felt until the next recession
occurs. Eliminating open-ended match-
The virtual abandonment of openended matching grants5 is inconsistent
with the view that such grants are
275
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
NATIONAL TAX JOURNAL VOL. XLIX NO. 2
ing grants means that the state budget
crunch will be much more severe during
a recession than it has been during the
downturns of the past several decades.
spending, to cut off beneficiaries of
programs, to reduce service standards,
or to shift costs to others (e.g., by
repealing the Boren Amendment that
has forced states to pay higher rates to
hospitals and nursing homes than they
3
otherwise
would have). Such provisions
help states fiscally but do not necessarily
make programs more efficient.
On the other hand, the federal deficit
will not increase as much in recessions
as it has done in the past. There is,
however, not perfect symmetry. States
cannot run up big deficits as the federal
government can. Their efforts to stave
off deficits will tend to destabilize the
national economy.7
Historical Perspective
The New Federalism rolls back the clock
in some ways to before the Great
Society and in other respects to before
the New Deal. By fundamentally
changing Medicaid, it affects one of the
major legacies of the Great Society. By
abolishing AFDC, it undoes a guarantee
that was established as part of the New
Deal.
Enhanced Flexibility
The New Federalism would give states
the freedom to redesign programs in
more efficient and effective ways, but
the quantitative magnitude of the
benefits from increased flexibility is
uncertain. The advantages gained by
reducing paperwork are probably not
worth a great deal in comparison with
the loss of funding that is associated
with the block grants,8 but the relaxation of regulations could be more
significant. Freedom from the voluminous rules governing how programs
must be operated could yield substantial
efficiency gains, particularly if the
federal government has been prescribing standards that are higher than local
citizens desire. Advocates of devolution
point to the results of welfare and
Medicaid waivers as indications of the
kind of changes that will occur. For
example, Oregon and Tennessee used
waivers to substantially expand Medicaid coverage, and Wisconsin was able to
significantly reduce welfare caseloads in
two counties.
But while the New Federalism would
fundamentally change the form of those
programs, it does not come anywhere
near returning us to where we used to
be in terms of the amount of aid
provided. Federal aid is currently 3.2
percent of GDP. That is less than the
peak level of 3.6 percent in the late
1970s, but it is higher than in any year
before 1973. The comparable figure in
1965 was 1.9 percent, and in 1940, 0.9
percent.
If federal aid were cut 20 percent from
its 1995 level, it would be no lower than
it was in 1992. After falling in the early
1980s and growing relatively slowly
through most of the remainder of that
decade, the amount of aid began to rise
rapidly in 1989. Just four years later it
had grown from $124 billion to $194
billion. Most of this increase occurred
automatically, without explicit initiatives
from the federal government. As
explained above, much of it reflected
the recession that led to surging
demand for social services (defined
The gains from enhanced flexibility
depend on how creative the states are
and how sweeping the block grant
legislation turns out to be. Some of the
freedom that state officials seek involves
allowing them to reduce their own
276
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
SYMPOSIUM ON THE DEVOLUTION REVOLUTION
broadly to include health, income
security, and direct services). Another
part was caused by spending increases
related to new federal mandates. But
part also was the result of state actions
that shifted costs to the federal government, particularly for Medicaid.9
WHAT IS NEEDED TO IMPROVE
RESEARCH
If the New Federalism is actually
enacted, it will eventually stimulate
considerable research by specialists in
public finance and public policy. A
thorough discussion of the pitfalls and
priorities for such research would
consider such issues as how welfare and
Medicaid reform affected work effort,
family structure, the nonmarital birth
rate, and access to health care. This
review will focus on a narrower set of
issues, those dealing with state tax and
spending policies.
De Facto Devolution
As this is written in early 1996, the
stalemate between the president and
Congress has brought a halt to what
once seemed an irresistible march
toward block grants and adoption of a
balanced budget plan. Nevertheless, it
appears very likely that much of the
devolution agenda will continue to
advance.
Unfortunately, unless major changes are
made in how public finance research is
conducted, it is likely that a large
portion of that research will fail to
recognize critical features of the state
fiscal landscape. This will significantly
impair its usefulness.
Two points can be made. First, many
states already have waivers permitting
them to escape from a large number of
the federal regulations that have
governed welfare and Medicaid
programs, and the administration
appears ready to approve more waivers
liberally. The end of “welfare as we
know it” is already occurring, with 37
states having welfare waivers of some
kind (varying from far reaching to
minor). The same thing is happening to
Medicaid.
It would exceed the scope of this article
to discuss all of the improvements that
are needed to improve the relevance of
research for the real world of state
governments. Rather, two important
issues will be addressed here: using
appropriate data and focusing on some
of the key subjects that have received
inadequate scrutiny.
Second, momentum toward reducing the
federal budget deficit appears likely to
continue. If it does, aid to states will be
curtailed, probably resulting in a reduction of the proportion of state revenue
coming from the federal government.
This amounts to de facto devolution,
although it will occur more gradually than if
it were compressed into seven years.
Data
One of the main limitations of much
existing research is the use of inappropriate data. In some cases, this is due to
a mistake on the part of the researcher.
But, in many situations, the relevant
data are somewhat difficult to obtain.
Most studies of state spending and tax
policies use data published by the U.S.
Census Bureau. Those data are adequate and appropriate for testing many
theories, and sometimes they are the
only option available. Census data,
Whether a big budget agreement is
reached in 1996 or 1997, the federal
system appears likely to change fundamentally. The issues are how fast it
changes and how it changes.
277
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
NATIONAL TAX JOURNAL VOL. XLIX NO. 2
however, are often inappropriate and
too dated to be useful for understanding recent developments.
general fund and school aid fund with
others related to independent or quasiindependent authorities that were
established by the state government.
The difference is substantial. In 1992,
for example, general fund spending was
$296 billion, while the Census reported
that general state spending was $612
billion. (Note that the Census refers to
its most important measure of spending
as general spending, which should not
be confused with general fund spending; the two have nothing in common.)
Timeliness is one serious problem. As of
March 1996, the most recent information on state finances from the Census
Bureau was for fiscal year 1993, which
ended in June 1993 for most states. 10
For local finances, the most recent data
were for fiscal year 1992. Such old
information is of limited value in
analyzing state and local policies in
1996 because major changes have
occurred in the interim. For example,
since 1993, the rate of growth of
Medicaid spending and federal aid have
slowed sharply, states have switched
from raising taxes to cutting them, and
corrections spending increases have
escalated.
The two ways of viewing spending yield
very different pictures of the nature of
state activity. For example, for 1992,
elementary and secondary education
represented 34.0 percent of general
fund spending but only 19.8 percent of
Census general spending. Corrections
was 5.6 percent of general fund
spending but only 3.3 percent of Census
general spending. Interest payments on
general debt were 4.0 percent of the
Census measure but much less of
General Fund spending.11
Census data have a number of other
important shortcomings. Two of the
most important are the following:
(1) they aggregate all spending and
revenue, failing to distinguish
between what is in or out of the
general fund; and
(2) they do not report Medicaid as a
distinct category of expenditures
even though Medicaid has been at
the heart of state budget policy for
a number of years.
Both measures of spending are valuable,
depending on the issue one is concerned with analyzing. If the focus is on
the total scope of state government
activity, the Census measure is preferable. But if one wants to understand
what is going on in state budget
deliberations, the general fund is
better.
The Importance of the General Fund
For example, in the 1970s and 1980s,
there was a major expansion of activity
financed by bonding, such as authorities
that underwrite subsidized mortgages
or help to finance environmental
projects or medical facilities. This
expansion received little if any state tax
support, so it was largely independent
of annual budget deliberations. It was
captured by the Census measure but
was ignored by the general fund.12
When governors and legislators battle
about the state budget, they usually
focus primarily on the general fund,
which normally receives the great
majority of non-highway-related state
tax revenue. In some but not most
states, there are also some other
important tax-supported funds, particularly for school aid. The Census Bureau
aggregates all funds, combining the
278
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
SYMPOSIUM ON THE DEVOLUTION REVOLUTION
This emphasis on the general fund
needs to be modified in two respects.
First, spending earmarked for schools in
a special fund is a very close substitute
for the general fund. It should be
aggregated with general fund spending
to provide the best measure of how
state spending patterns are changing.
Second, there is not uniformity about
how the general fund is defined across
states or for a particular state over time.
General fund data are more useful for
comparing developments in particular
states in various years than for making
interstate comparisons. If a state has
made a material change in what it
includes in the general fund, analysts
should adjust for this in their research.
Fortunately, states do not often make
major changes in the definition of their
general funds.
different from 1988, but by 1995, it had
significantly changed once again.13 To
understand the real story of state
finances, there is no substitute for
incorporating such institutional information.
Using More Appropriate Data
If the data collected by the Census
Bureau are inadequate, what should be
done? 14 There are two alternatives—
information from individual states
themselves and reports published by
national organizations of state governments. As mentioned above, little
reliable analysis is available on whether
state budgets are structurally balanced,
which is a very important issue for
understanding how state fiscal policies
are changing. To resolve this issue
obviously requires data that are available
only at the state level. It will be discussed in the next section.
Medicaid
Medicaid is the second largest program
in most state budgets, exceeded only by
aid to elementary and secondary
education. It is also an extremely
complicated program. Although
approximately half of the participants
are children, they account for only
about one-sixth of spending. Rather,
about two-thirds of the spending is for
the elderly and disabled.
The National Conference of State
Legislatures (NCSL) and the National
Association of State Budget Officers
(NASBO) publish numerous reports that
fill in some of the gaps left by the
Census Bureau.15 Studies by Poterba
(1994, 1995) and Blackley and DeBoer
(1993) are examples of recent analyses
that creatively used NCSL and NASBO
reports to analyze state fiscal developments in the early 1990s.
The Medicaid program has been
changing rapidly in several dimensions.
For one thing, its costs have risen
sharply. In 1991 and 1992, for example,
total Medicaid spending rose approximately 28 percent per year. Its financing
has also been in a state of flux as states
first used creative financing to shift costs
to the federal government, and then the
federal government gradually tightened
regulations to prevent or restrict many
of the stratagems employed by states.
The way states paid for Medicaid in
1991 and 1992 was considerably
Despite their great value, the NCSL and
NASBO reports have some shortcomings
that researchers should consider. For
example, as already mentioned, states
do not employ the same accounting and
budgeting systems. A low year-end
balance is a sign of unusual fiscal stress
in most states, but some states make
budget adjustments so that they always
report a low balance. In addition, the
projections in the reports are often
misleading, tending to underestimate
279
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
NATIONAL TAX JOURNAL VOL. XLIX NO. 2
spending and revenue increases and
year-end balances.16 The information in
the reports about periods that have
already ended is more useful than the
projections. The reporting methodology
used by particular states has not always
been consistent from year to year, making
comparisons across years sometimes
unreliable. Finally, the information
submitted by states is occasionally
influenced by political considerations. In
1993, for example, it was widely
believed in Sacramento that California
ended the fiscal year with a large deficit,
and that is what NCSL reported.
Governor Pete Wilson’s administration,
however, reported to NASBO that there
was a budget surplus, and that is what
it said. Despite such problems, the
NCSL and NASBO reports are extremely
useful documents on the whole.
will revenue from the existing tax system
grow? How much will the spending
necessary to maintain the current level
of services increase?
In many (but not all) states, the income
elasticity their state tax systems is lower
than it was in the late 1970s, and tax
revenue now tends to grow more slowly
than total economic activity. Among the
reasons for the lower responsiveness of
tax revenue to economic growth are the
trend toward flatter income tax rates,
the adoption of indexing in 17 of the
41 states with income taxes, the
movement of more households into the
highest bracket of state income taxes,
and the increasing economic importance
of services, which are lightly taxed by
state sales taxes. Unfortunately, there is
little good research to document the
extent of the reduction of elasticity.17
Well-grounded estimates of elasticity
using time-series data need to make
adjustments for the effects of legislated
changes; such information can be
obtained only at the state level.
Vital Research Subjects
Five subjects are essential for understanding state fiscal policies: whether
state budgets are structurally balanced,
the factors affecting discretionary tax
changes, the determinants of state
spending, the interactions between
state and local governments, and the
effects of federal aid. The first four
subjects have received relatively little
attention. While the fifth has been studied
extensively, it requires more disaggregated research, particularly in relation to
Medicaid and block grants in general.
These issues are largely the same ones
that would need to be studied even if
no major devolution occurs. Before one
can trace the effects of new federal
policies, it is necessary to understand
what is already occurring in state policy.
There are numerous important conceptual issues in analyzing the spending
side of the state budget, such as the
following.
(1) Is the concern with maintaining
current services or complying with
current laws? Many recent studies
do not really assume current
services. Rather, they often
extrapolate past trends or assume
that policies follow existing
legislation, which may permit
erosion of service levels or improvements in services.
(2) What is assumed about productivity? Traditionally, productivity
improvements in the public sector
have lagged, but with the emphasis on “reinventing government”
in many state capitals, it may be
Structural Balance
The analysis of whether state budgets
have structural surpluses or deficits can
be divided into two parts: How rapidly
280
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
SYMPOSIUM ON THE DEVOLUTION REVOLUTION
possible to maintain services even
if spending is squeezed.
(3) What does “current services”
mean in relation to local aid—
maintaining real spending per
person or continuing the traditional relationship between state
aid and local spending?
income tax rates (Gold, 1991). But firm
conclusions about the effect of the tax
system’s elasticity on revenue and
spending cannot be obtained until good
estimates of elasticity are available.
Determinants of Spending
An important area for research would
focus on the simultaneous determination of spending for the major parts of
the state budget. It is widely believed
but not well documented in public
finance literature that Medicaid and
corrections spending have, to some
extent, crowded out spending for
education and welfare programs.
Moffitt (1990) concluded that Medicaid
has crowded out welfare spending, and
Wyckoff and Fossett (1996) found an
impact of Medicaid on school aid, but
much further research is needed along
these lines. Neither of those studies
took a general approach that considered
all of the major budget categories,
recognizing that, if more is spent on one
program, less can be devoted to others.
Discretionary Tax Changes
The articles by Poterba (1994) and
Blackley and DeBoer (1994) mentioned
above both made important contributions to understanding why states raised
their taxes during the early 1990s, but
they merely represent a beginning
toward filling an important information
gap. Additional research is needed on
state tax changes during periods when
fiscal conditions are healthy as well as
when they are stressed.
This issue is closely related to the
question of whether states with more
elastic tax systems tend to have faster
growing revenue and spending. From a
political point of view, it would seem
considerably easier to increase revenue
automatically from the nominal growth
of income than to increase it by
explicitly raising tax rates or expanding
the tax base. In fact, however, most
studies have found that the proportion
of tax revenue from the personal
income tax (the proxy usually used to
indicate differences in elasticity) has a
weak effect on revenue/spending
increases or no effect at all. (Oates,
1975; Feenberg and Rosen, 1987) If
that is true, it would imply that discretionary tax changes offset differences in
the natural growth of revenue in
response to economic changes. That
would be consistent with the wellestablished fact that states with
relatively high personal income tax rates
were much more likely to reduce them
in the 1980s than states with low
Most research has focused on determinants of state and local spending.
While that is important from some
perspectives, it is also important to
analyze state and local spending
separately because decisions are often
made independently at the two levels.
Interactions between State and Local
Governments
Aside from elementary and secondary
education, relatively little attention has
recently been devoted to the shifting
balance of spending and revenue
between state and local governments.
But important changes are occurring.
The U.S. Census Bureau began reporting
on state aid in 1957. From that year
until 1988, aid fluctuated between 34–
37 percent of general state spending. It
281
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
NATIONAL TAX JOURNAL VOL. XLIX NO. 2
fell below that level in 1989, and by
1992, the last year for which figures are
available, aid was only 32.3 percent of
general spending. Ladd (1991) provides
one of the few recent analyses of local
aid.
literature agrees that block grants would
tend to depress welfare spending to
some degree, the so-called “race to the
bottom” would be much faster according to some researchers than others.
The Importance of Disaggregation
An important priority for future research
should be the effects of devolving state
programs to local governments. In view
of the substantial variation that already
exists among states, it should be
possible to provide useful guidance to
state policy makers about how to
devolve programs without creating
more problems than are solved.
One of the underlying themes of this
section is the importance of disaggregation. For example, various states must
be considered separately because states
are not monolithic in their policies, what
the Census Bureau calls welfare spending must be divided between Medicaid
and cash assistance, individual spending
programs must be considered to
understand the effects of federal
matching provisions, revenue from
charges must be considered separately
from miscellaneous revenue, and states
must be analyzed distinctly from local
governments.
Effects of Federal Aid
A vast amount of research has been
done on the effects of federal aid on
state and local spending. Little of it
extends into the period from 1989–93
when federal aid soared. Little of it deals
with Medicaid, which, as mentioned
above, now accounts for approximately
40 percent of the total amount of aid
provided. Much of the research is
relatively highly aggregated, making it
impossible to measure with precision
the effect of differing matching ratios.
The reference to local governments
requires elaboration. Many of the
points made here about the need for
research on states apply to local
governments as well. There is far less
research on the fiscal behavior of
counties—the fastest growing general
governments in the country—than on
states. From some perspectives, it is
important to consider state and local
policies rather than those of states or
localities alone. We need both kinds of
analyses—those that examine state and
local governments separately and those
that study them together.
Welfare is one program that has been
analyzed extensively. According to
Chernick and Reschovsky’s review of the
literature (1995), it has been difficult if
not impossible to separate the price
effects of aid from the income effects
because a state’s matching rate depends
on its per capita income. Considerable
difference exists between two bodies of
research, with one group of studies
finding much larger price effects than
the other. This issue would be extremely
important if the federal government
converted welfare to a block grant,
which would sharply increase the price
of spending on welfare from the state
budget point of view. While the
Issues Raised Specifically by Devolution
Much of the research on devolution can
be divided into four important areas.
(1) Responsiveness to price changes.
How much will spending decrease
if block grants substitute for
matching grants?
282
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
SYMPOSIUM ON THE DEVOLUTION REVOLUTION
(2) Diversity of demands. How much
will the variation in spending
among states increase? If the
federal government has caused
spending to rise to a much higher
level than citizens of some states
desire, differences in the level of
spending could grow considerably.
Among the factors influencing
these changes will be the effect of
relaxing mandates on service
levels, differences in matching
rates, and possible differences in
the elasticity of demand for
services.
(3) Diversity of production technology.
If the federal government allows
greater freedom to design
programs, there should also be
increased variation in how states
structure and operate the delivery
of services.
(4) Distributional effects. If the
federal government eliminates the
national safety net, there could be
a significant increase in inequality
in terms of income and services
received.
long-term trends and intermediate-term
trends, and one’s sense of the direction
of change depends on the time period
one considers. For example (Gold et al.,
1996), consider the following.
(1) State taxes rose much faster than
local taxes in the quarter century
from 1970–95, but that is entirely
because of what happened in the
1970s. Since 1985, local taxes
have risen faster.
(2) Likewise, states now pay a
considerably higher share of the
cost of elementary and secondary
education than they did in 1970,
but since 1987, the state share has
been trending downward.
(3) States are much more dependent
on personal income taxes than
they were in 1970, but in the past
decade, the average effective
income tax rate has not risen.
(4) Reliance on the property tax fell
sharply in the 1970s, but since
1982, the property tax has
rebounded, growing faster than
personal income (although it still
claims a much lower proportion of
income than it did 25 years ago).
In analyzing these effects, analysts
should be careful to avoid the pitfall of
placing too much emphasis on policies
adopted in the one or two years after
devolution goes into effect. The initial
state response may be muted because
of inertia or by reserves that can be
tapped during a transition period. The
long-run effects of devolution are likely
to be much greater than its short-run
results.
If one is predicting what will happen in
the next five or ten years, the trends
since the mid-1980s seem more
significant than the longer-term trends
going back to 1970. What, then has
been happening to spending and tax
policy? Let us consider spending first.
THE CONTEXT OF THE DEVOLUTION
REVOLUTION
Table 1 shows that some important
changes occurred in the composition of
state spending between 1990–94:
Expenditures
Research about the impact of the New
Federalism has to consider the context
in which it is occurring. State policies are
already changing. One of the problems
with studying states is that there are
(1) Medicaid rose most, growing from
9.1 percent to 12.8 percent of the
budget. Among the causes of this
increase were health cost inflation,
283
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
NATIONAL TAX JOURNAL VOL. XLIX NO. 2
enrollment growth, the main
reason was that states relied
heavily on tuition increases to
substitute for state tax support.
(5) Aid to Families with Dependent
Children rose slightly, from 2.5
percent to 2.8 percent of the
budget. This is not much of an
increase in view of the 27.5
percent caseload increase that
occurred. Real benefits were
reduced in most states.
(6) The remaining programs received a
smaller share of the pie, as states
focused on trimming bureaucracies, shifted some health programs
into Medicaid, and downsized or
provided meager increases for
other programs.
TABLE 1
COMPOSITION OF STATE SPENDING, 1990 AND 1994
1990
1994
Elementary and secondary
education
36.3%
36.2%
Higher education
14.0%
12.5%
AFDC
2.5%
2.8%
Medicaid
9.1%
12.8%
Corrections
5.2%
5.9%
32.8%
29.8%
Miscellaneous
Note: Spending consists of general fund expenditures
plus other expenditures from state revenue for
elementary and secondary education.
Source: See Gold and Ritchie, 1996.
the impact of the recession in
raising caseloads, and the shifting
of other health programs into
Medicaid.18
(2) Corrections also rose sharply, from
5.2 percent to 5.9 percent of total
spending, reflecting popular
support for “get tough on
criminals” policies such as “three
strikes and you’re in.”
(3) School aid, the largest share of the
budget, was stable at about 36
percent of the total. This was not a
strong performance considering
that schools had to contend with
large enrollment increases during
this period.19 Competition from
Medicaid and corrections is
probably an important reason why
school aid did not increase more.
After the recession ended, real
spending per pupil, including
federal and local funds, rose
much less than in other nonrecession periods during the past 50
years.
(4) Higher education was the big loser
in the battle for state support,
falling from 14.0 percent to 12.5
percent of spending. While part of
this slippage was due to slow
What would one expect under the New
Federalism? If federal aid cutbacks
cause state fiscal stress, the programs
that fared badly in the early 1990s
would probably again be losers, but
block grants would cause Medicaid and
welfare spending to grow less than
before. Higher education and miscellaneous programs would continue to
fall as a proportion of the budget,
and welfare spending would also go
down. Corrections spending would
surely continue to grow rapidly. School
aid would have to battle hard for
funds, and Medicaid would rise much
more slowly than it did in the early
1990s.
On another level, spending may be
undergoing changes in response to the
“reinventing government” movement,
which can be viewed as attempting to
change the production function for
government services, or at least moving
up to the production function from an
inefficient position away from it. To the
extent that this movement succeeds,
lower spending may not imply reduced
services.
284
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
SYMPOSIUM ON THE DEVOLUTION REVOLUTION
An important issue to monitor and
analyze will be how widely state
responses differ. Will many states
sharply reduce their spending on
poverty-related programs if given the
chance, as might be predicted by those
who remember how backward some
states were before federal programs and
mandates began to expand in the
1960s? Or have states changed so much
that they will maintain most services
even though they are not required to?
Conclusions
If the New Federalism is adopted in
something like the form proposed by
Congressional Republicans, the country
will be in for a grand experiment. In
some important ways, state governments have changed fundamentally.
Starting in the 1960s, they became
more representative, more competent,
and much larger. If the federal government pulls back sharply, there is no
precedent for telling how the states will
respond.
Taxes
The problem of predicting their response is complicated by some other
factors. Twenty-one states now have
term limits for legislators. While the
effects of these newly enacted limitations are still uncertain, they are likely to
reduce expertise, increase the shortterm orientation of policy making, and
enhance prospects for adoption of new,
untested policies. In addition, nearly half
of the states have constitutional or
statutory spending or revenue limits that
did not exist 20 years ago. Since most of
those limitations make no allowance for
changes in federal aid, they could
present formidable barriers to state
efforts to offset reduced reliance on
federal aid.
State tax policy has been dominated
during the past several years by small
net tax reductions. This followed a large
number of tax increases at the start of
the decade. It is precisely what normally
has happened for the past 20 years,
with increases during recessions and
decreases in their aftermath (Gold, 1996).
It may be, however, that state tax policy
is in the process of becoming more
conservative. At least, that is the
impression one gains from shifts in the
political ideologies of many recently
elected governors and legislators.
This issue is unsettled because not many
proponents of tax cuts also explicitly call
for large reductions in services. They
usually claim that services can be
maintained by increasing efficiency, or
they gloss over the issue. But attitudinal
surveys usually find that, while citizens
like tax cuts, they do not want to lose
services.
The New Federalism promises to present
great opportunities for public finance
researchers. But if they are to take
advantage of it, they will have to be
more creative about obtaining data and
pay more attention to institutional
details than they have in the past.
Recent experience suggests that tax cuts
will tend to occur in states with relatively high tax rates and relatively elastic
tax systems. The cut in high-tax states
reflects the intense competition for
economic development that is an
important part of the environment for
making state fiscal policy.
ENDNOTES
The views expressed in this article are solely those
of the author and should not be attributed to the
Urban Institute or the New Federalism Project and
its sponsors. The author appreciates helpful
comments on an earlier draft by Howard Chernick,
George Peterson, Len Nichols, and the editor.
285
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
NATIONAL TAX JOURNAL VOL. XLIX NO. 2
1
2
3
4
5
6
7
8
9
The term “devolution revolution” was coined by
Richard P. Nathan.
This estimate was prepared by the Center on
Budget and Policy Priorities. For a discussion of
this issue developed before the reconciliation bill
was vetoed, see Lav and St. George (1996). The
26 percent figure is the decrease in the year when
the budget is supposed to be balanced. Many
other analyses discuss the cumulative decrease
over seven years, which is a smaller percentage
reduction because the aid decreases grow over
time.
According to the U.S. Census Bureau, federal aid
was approximately 26 percent of state general
revenue in 1992. One component of general
revenue is miscellaneous revenue, a large portion
of which is dedicated to programs that are outside
the scope of annual budget deliberations, such as
interest received in return for loans to companies
(for economic development subsidies) or to home
owners (for mortgage subsidies). Excluding
miscellaneous revenue raises the proportion of
state resources accounted for by federal aid.
This is not necessarily true in all cases. Under one
Medicaid block grant proposal, there would still be
a price effect for low-spending states. That is, they
would receive the maximum possible amount of
federal aid only if they spent more than they were
already spending.
As of March 1996, it appeared likely that Congress
intended to retain an open-ended matching grant
for foster care.
These points were suggested by George Peterson.
Numerous states do have deficits during recessions.
In many states, the balanced budget requirement
requires that a balanced budget be enacted, but if
the budget falls into deficit during the course of
the year, that is acceptable. All states do, however,
attempt to eliminate deficits quickly.
That was the conclusion about the block grants
created in 1981 according to Peterson et al.
(1986).
At first, states used donations from hospitals to
provide their matching funds and then returned
the funds to the hospitals as Medicaid reimbursements. A variation was to impose a tax on health
care providers, which sometimes needed to be
paid only if the provider agreed. Some of these
taxes guaranteed that the provider would receive
at least as much back from the state as it paid in
tax. When donations were outlawed and the
federal government required that hospital taxes
not have “hold harmless” provisions, some states
turned to intergovernmental transfers, which were
like donations, except that they came from local
government hospitals or university medical
facilities. All of these devices allowed states to
painlessly obtain more federal aid without
spending any regular tax money. Another strategy
was to convert a health program that was formerly
funded completely by state revenue into part of
the Medicaid program, thereby obtaining federal
support for at least half of the cost.
10
The fiscal year ends March 31 in New York, August
31 in Texas, and October 31 in Alabama and
Michigan.
11
There are no national figures for interest spending
in general funds, but Maryland’s situation is
illustrative. Its general fund interest outlays are less
than 1 percent of the total, while interest accounts
for more than 4 percent of general spending
reported by the Census Bureau.
12
During the 1980s, such activities led to rapid
increases in both interest outlays and interest
earnings, making it appear as if the reach of
government was growing more rapidly than was
true for core activities. Much of the growth of
interest revenue, for example, was not available for
funding regular programs because it was balanced
by higher interest expenses.
13
Between 1990–92, for example, the proportion of
state (i.e., nonfederal) Medicaid dollars obtained
outside of the general fund rose from 5–19.9 percent.
14
The National Income and Product Accounts do not
provide a satisfactory alternative. Although they
are much more timely than Census data, they are
even more highly aggregated, both in terms of
spending functions and in that they combine state
and local governments.
15
Both NCSL and NASBO publish reports on forecasts
of general fund spending and revenue and yearend balances as well as estimates of those
indicators for the most recently completed year.
Their coverage otherwise differs, providing
complementary features. NCSL publishes
projections of spending increases for five major
spending categories, and its compilation of
legislated tax changes is more detailed and
accurate. NASBO reports spending in the two
most recently completed years, enabling it to
incorporate later revisions than NCSL. In addition,
NASBO publishes a report on state spending that
shows outlays for the three past years for the
major categories of spending, dividing it between
general fund, other state funds, federal aid, and
bonding.
16
These biases tend to exist when the economy is
expanding. When a recession occurs, revenue and
balances generally are lower than projected.
Spending is higher for some programs but lower
for others.
17
The first two reasons why elasticity has fallen relate
to one-time changes, while the second two refer
to continuing developments. For a review of what
state revenue officials believe about the elasticity
of their tax systems and a review of the literature,
see Gold (1995b).
18
No reliable data are available to measure how
much health spending was shifted into Medicaid.
The spending considered here is the sum of
286
National Tax Journal
Vol 49 no. 2 (June 1996) pp. 273-87
SYMPOSIUM ON THE DEVOLUTION REVOLUTION
general fund spending plus earmarked state
spending for elementary and secondary schools.
This is a rough approximation of how tax dollars
are spent, except that it excludes most highwayrelated spending. It excludes spending paid for
with federal aid, user charges, or the special
devices used to help finance Medicaid.
19
Enrollment rose 6.9 percent during this period, the
largest increase since school enrollment peaked in
1973. This was, of course, much less than the
increases in welfare caseloads, Medicaid rolls, and
prison populations.
Gold, Steven D., and Sarah Ritchie.
“State
Spending Patterns Have Been Changing.” State
Tax Notes (January 15, 1996): 187–91.
Gold, Steven D., Elizabeth I. Davis, Deborah
A. Ellwood, David S. Liebschutz, and Sarah
Ritchie. How Funding of Programs for Children
Varies among the 50 States. Albany: Center for
the Study of the States, 1996.
Ladd, Helen F. “The State Aid Decision:
Changes in State Aid to Local Governments,
1982 to 1987.” National Tax Journal 44 No. 4
(December, 1991): 477–96.
Lav, Iris J., and James R. St. George.
“Can
States Assume Federal Roles?” In Proceedings of
the Eighty-Seventh Annual Conference on
Taxation. Columbus: National Tax Association–
Tax Institute of America, 1996.
REFERENCES
Blackley, Paul R., and Larry DeBoer.
“Explaining State Discretionary Revenue
Increases in Fiscal Years 1991 and 1992.”
National Tax Journal 46 No. (March, 1993):
1–12.
Moffittt, Robert. “Has State Redistribution
Policy Grown More Conservative?” National Tax
Journal 43 No. 2 (June, 1990): 123–42.
Chernick, Howard, and Andrew Reschovsky.
“Entitlements and Block Grants.” Presented at a
conference on devolution sponsored by the
Economic Policy Institute (December, 1995).
Oates, Wallace E. “Automatic Increases in Tax
Revenues—The Effect on the Size of the Public
Budget.” In Financing the New Federalism,
edited by Wallace E. Oates. Baltimore: Johns
Hopkins, 1975.
Feenberg, Daniel R., and Harvey S. Rosen.
“Tax Structure and Public Sector Growth.”
Journal of Public Economics 32 (1987): 185–
201.
Peterson, George E., Randall R. Bovbjerg,
Barbara A. Davis, Walter G. Davis, Eugene C.
Durman, and Theresa A. Gullo.
Reagan and
the States. Washington, D.C.: The Urban
Institute, 1986.
Gold, Steven D. “Interstate Competition and
State Personal Income Tax Policy in the 1980s.”
In Competition Among States and Local
Governments, edited by Daphne A. Kenyon and
John Kincaid. Washington, D.C.: The Urban
Institute, 1991.
Poterba, James M. “State Responses to Fiscal
Crises: The Effects of Budgetary Institutions and
Politics.” Journal of Political Economy 102
(August, 1994): 799–821.
Gold, Steven D., ed. The Fiscal Crisis of the
States: Lessons for the Future. Washington,
D.C.: Georgetown University Press, 1995a.
Poterba, James M. “Balanced Budget Rules and
Fiscal Policy: Evidence from the States.”
National Tax Journal 48 No. 3 (September, 1995):
329–36.
Gold, Steven D. “The Income Elasticity of State
Tax Systems: New Evidence.” State Tax Notes 8
No. 18 (May, 1995b): 1849–56.
Wyckoff, James, and James Fossett.
“Has
Medicaid Growth Crowded Out State Spending
on Education?” Journal of Health Politics, Policy
and Law (forthcoming, 1996).
Gold, Steven D. “State Tax Cuts in 1995: Is
Something New Afoot?” Public Budgeting and
Finance (forthcoming, 1996).
287