Content uploaded by Steven C Deller
Author content
All content in this area was uploaded by Steven C Deller on Feb 19, 2015
Content may be subject to copyright.
Content uploaded by Steven C Deller
Author content
All content in this area was uploaded by Steven C Deller on Oct 23, 2014
Content may be subject to copyright.
A Model of Asymmetries in the
Flypaper Effect
Steven C. Deller* and Craig S. Maher
†
In this applied research study we examine the changing fiscal relationship between state
and local governments. Our research question is simple: Do local governments treat state
aid during periods of stability and instability in a systematic manner? Using data on
Wisconsin’s unconditional shared revenues program from 1990 to 2000, we find evidence
of a flypaper effect and that the relationship tends to be asymmetrical. The manner in which
local governments treat intergovernmental aid is different between periods of increases and
decreases in aid. Specifically, using a model that allows for the identification of structure
shifts we find evidence of fiscal replacement. In addition, we find that changes in aid impact
types of spending differently. When aid is reduced, policymakers appear to be less inclined
to cut police and fire services than they are to cut services such as parks and recreation.
Introduction
Perhaps one of the most frequently cited phenomena in the empirical fiscal federalism
literature is the ‘‘flypaper effect,’’ the phenomenon where increases in unconditional
aid have a stimulatory effect on recipient government expenditures greater than
equivalent changes in personal income.
1
Given the simple median voter model of
taxpayer (i.e., consumer) behavior, a dollar of grant money should have the same
effect as a dollar change in personal income. The empirical literature, however, has
consistently found that intergovernmental aid has a much greater impact than changes
in income, or that aid has a greater stimulative effect than predicted by median voter
theory. As outlined by Bailey and Connelly, several hypotheses have been advanced for
why the use of transfers does not follow the predictions of economic theory, ranging
from the complexity of matching grants to fiscal illusion to sloppy empirical
specification and estimation methods to poorly thought-out theoretical foundations.
2
The empirical literature has accepted the presence of the flypaper effect, and more
recent research has sought to determine whether the flypaper effect is symmetrical by
focusing on the effects of grant reductions or the introduction of instability into grant
programs.
3
To date, the results of this latter literature have been mixed, but evidence
*University of Wisconsin-Madison
†University of Wisconsin-Oshkosh
Publius: The Journal of Federalism volume 36 number 2, pp. 213–229
doi:10.1093/publius/pjj005
Advance Access publication December 14, 2005
ªThe Author 2005. Published by Oxford University Press on behalf of CSF Associates: Publius, Inc.
All rights reserved. For permissions, please email: journals.permissions@oxfordjournals.org.
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
has tended to support the notion of asymmetry. In other words, governments tend to
treat increases and decreases in aid differently. This question is particularly relevant
today as fiscal pressures at the federal and state levels have led to both a decline in
intergovernmental aid to the lowest levels of government and the introduction of a
great deal of uncertainty.
In this study we add to this body of research by empirically testing for asymmetries
in Wisconsin municipalities’ responses to unconditional aid. As previously described,
Wisconsin’s shared revenues program is modeled after the now defunct federal
revenue sharing program.
4
The aid does not have any ‘‘strings’’ attached, and
municipalities can spend the monies on whatever they choose, including reduction of
local taxes, fees, and charges. The principal contribution of this case study is that more
than fifteen years of audited fiscal data are available at the municipal level by program
expenditure. We can, therefore, address the question of asymmetry in the flypaper
effect on total expenditures and expenditures by program. This point is important
because our earlier work demonstrated different flypaper effects depending on the type
of expenditure.
5
The current fiscal crisis facing state government in Wisconsin has introduced a
great degree of uncertainty into the local fiscal planning process. The shared revenue
program at the beginning of the decade was stable and expanding, but fiscal
uncertainty and changing political philosophies at the state level have called into
question the robustness of the program. Flat funding has caused the program to
stagnate in real terms over the past several years, particularly since 1995 (figure 1). This
0
50
100
150
200
250
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
All Cities All Towns All Villages
0
50
100
150
200
250
1987
1988
1989
1990
1991
1992
1994
1995
1996
1997
1998
1999
2000
All Cities All Towns All Villages
Figure 1 Per capita shared revenues (2002 dollars).
Source: Wisconsin Department of Revenue (calculation by the authors).
214 Steven C. Deller and Craig S. Maher
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
study thus provides a unique opportunity to test for symmetry in the flypaper effect.
We use a model of structural change to uncover how local officials treat aids during
times of aid stability and times of aid uncertainty.
This article proceeds as follows. We review the literature on the flypaper effect, with
special emphasis on asymmetries, and then briefly outline the Wisconsin shared
revenues program and the current political environment. We then present our
structural change model, empirical model, and our empirical results. We close with
some concluding comments.
The Flypaper Effect and Types of Asymmetries
Empirical tests of the flypaper effect have largely been based on median voter theory.
6
Median voter theory predicts that government spending will reflect the demands of the
median voter and that the introduction of unconditional aid will simply substitute
local money, with no impact on spending.
7
From a purely theoretical perspective,
unconditional aid will have the same effects as an identical increase in local income.
Despite these predictions, empirical studies have consistently shown that intergovern-
mental aid does, in fact, stimulate local expenditures more than local income.
Research in this area has also demonstrated that different types of aid and sources
of aid funding have different effects on local spending behavior. For instance, Chubb’s
X
G
A
F
C
x’’
y’
x’
y
x
DB G
inducement
Fiscal restraint Fiscal replacement
Figure 2 Modeling fiscal asymmetry.
A Model of Asymmetries in the Flypaper Effect 215
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
explanation of the flypaper effect demonstrates that unconditional aid should
stimulate spending less than conditional aid.
8
Additionally, Grossman’s analysis of
local spending found that federal intergovernmental aid had a greater effect on local
spending than state aid, suggesting that the greater the separation between revenue
source and spending source, the greater the impact on local spending.
9
What these
studies suggest is somewhat of a hierarchy of grant effects: federal conditional aid
should have the greatest effect on local spending, followed by state conditional aid and
state unconditional aid.
10
Asymmetries Research
Although we appear to know a great deal about the positive effect of increases in
intergovernmental aid, it has only been recently that researchers have empirically
examined the impact of aid reductions. From a purely economic perspective, Heyndels
describes the flypaper effect within the context of a community’s choices between
public (G) and private (X) expenditures (figure 2).
11
Government revenues comprise
taxes that are determined by levels of private income and unconditional aid.
The flypaper effect is described as the difference in the impact on government
spending based on changes in personal income compared with changes in aid. Again,
using Heyndels’s (2001) framework, assume that starting from budget line AB, the
budget curve shifts to CD. If the shift is induced by growth in private income, the new
optimum point will be x0. If, however, the shift is caused by an increase in aid, the
flypaper effect suggests the new optimum point will be x00 . With point x00 lying to the
right of point x0, public expenditures are said to be more responsive to grant effects
than to changes in personal income. More specifically, the flypaper effect contends
that an increase in unconditional grants will produce a higher rate of expenditures
while also lowering taxes.
The issue at the heart of this study is the impact on spending when grant payments
are reduced. From a purely economic perspective, a reduction in grants should shift
the budget line back to AB. Yet, there is empirical evidence suggesting that for a
number of reasons governments react differently to reductions in grants than they do
to increases. Gramlich suggests that once programs are created they produce advocates
who make it difficult to eliminate the affected programs.
12
Heyndels describes such
an effect as fiscal replacement. Fiscal replacement hypothesizes that expenditure
reductions will be less than grant reductions and that the difference will be
compensated by increases in local taxes.
13
Using Heyndels’s diagram, if, starting
from x00, communities move to a point such as y, expenditures are reduced less and
taxes are increased more than predicted by the symmetry case. Another possible
outcome is fiscal restraint, which, as described by Heyndels, suggests that a
community would move from x00 to y0, which lies to the left of x, implying that
public expenditures are lowered more than in the symmetric case and that taxes are
enhanced less.
216 Steven C. Deller and Craig S. Maher
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
A second approach that has not seen as much discussion in the fiscal literature
hinges on the notion of uncertainty in the aid programs.
14
During periods of
increasing aid, local officials may be more likely to treat the aid as a permanent source
of revenue and hence build the aid into base budgeting. Here one might reasonably
expect increases in aid to substitute for local revenues. But if aid programs are
declining and the political future of aid programs is challenged, these aids become less
permanent and more transitory in nature. As aids decline or become more uncertain
local officials will be less likely to build the aid into base budgets and will be more likely
to treat the aid as special one-time funds to leverage local revenues. In a study of
federal aids Benton finds evidence supporting the transition in how state officials treat
the permanency of the aids.
15
He finds that in the short run state and local
governments will request additional money to fund programs in the wake of lost
federal funds. In the long run, however, once the reality of the decline in federal funds
is realized, state and local funding begins to decline. For our purposes, if aids move
from a permanent to a transitory type of income, one would expect to find what
Heyndels calls fiscal replacement.
Empirical Evidence of Asymmetries
Heyndels’s recent analysis of the impact of unconditional grants on Flemish
municipalities finds support for the asymmetries argument in that when grants
grow, spending also increases. The same, however, cannot be said about grant
reductions. Heyndels’s research demonstrates that spending is, in fact, unaffected by
grant reductions, with taxes apparently compensating for the lost revenues. In other
words, rather than reducing local spending in some proportion to the lost aids, local
municipalities substitute local dollars. Similar results are revealed in Gramlich’s
examination of the impact of federal grants on U.S. state and local government
expenditures.
16
Gramlich found that when federal grants were increased, state and
local expenditures also rose; however, when grants were reduced, state and local
government expenditures remained constant.
In an examination of health care expenditures in Italy, Levaggi and Zanola find that
local governments respond to declines in grants-in-aid from the central government
by increasing local deficits rather than reducing expenditures.
17
They conclude that
this result provides evidence of asymmetric responses to intergovernmental grants.
Melo finds evidence of asymmetries in response to grants-in-aid in Colombia where
subnational authorities try to cover the reduction in transfers.
18
In an examination of how local governments in the United States treat highway aids
from the federal and state governments, Deller and Walzer find evidence of structural
shifts in how grants-in-aid are treated.
19
During the 1980s, under the Reagan/Bush
policies of fiscal federalism, federal grants had a much greater stimulative effect at the
end of the period than at the beginning. No such shift was found for state aids. Deller
and Walzer conclude that the differences are attributable to a perception that federal
A Model of Asymmetries in the Flypaper Effect 217
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
aid is less certain and more transitory in nature. For our purposes in this study, this
shift in the treatment of grants-in-aid is evidence of asymmetry.
Stine’s research on the impact of state and federal grants on Pennsylvania counties
reveals somewhat different results.
20
Apparently, when the federal grants were
reduced, counties cut expenditures by more than the amount of grant reduction. Yet,
when the state cut grants to the counties, the reductions were partially compensated by
increases in own-source revenues. Gamkhar and Oates’s empirical examination of the
impact of changes in federal grants on state and local government expenditures during
1953–1991, however, reveals no asymmetry effects.
21
In summary, in the eight published empirical studies of asymmetry effects, we have
three different sets of findings: symmetry (Gamkhar and Oates), fiscal replacement
(Gramlich; Benton; Deller and Walzer; Heyndels; Melo; Levaggi and Zanola), and
inducement (Stine).
Wisconsin’s Shared Revenue Program
In Wisconsin, intergovernmental aid plays a central role in understanding the state’s
fiscal picture. This aid serves three primary purposes: (1) supporting the operation of
state and federal programs, (2) reducing spending disparities between local
jurisdictions (i.e., school aids and shared revenues), and (3) reducing local property
tax burdens. In FY 2000, Wisconsin state government collected 64.5 percent of total
state and local general revenues yet was responsible for only 39.9 percent of total state
and local general expenditures.
22
Two programs in particular are at the heart of this
discrepancy between where revenues are collected and expenditures incurred: state
support of K-12 education and shared revenues to municipalities and counties.
Wisconsin’s shared revenue program is most similar to those found in Florida,
Massachusetts, Michigan, Minnesota, and New Jersey in that a large portion of the
revenue is distributed on the basis of tax-raising capacity.
23
Where Wisconsin differs
from most states is that the funding source for the program is not tied to a specific tax
such as sales or income but is general-purpose revenue support. For the past several
years Wisconsin has passed down to local governments nearly $900 million annually in
the form of shared revenues. The combination of shared revenues and transportation
aids, in particular, has made local governments in Wisconsin highly dependent upon
these transfers from the state for their general operating budgets. In FY 2000,
approximately 35 percent of local government revenues in Wisconsin came in the
form of state aid including highway aids, health and human services aid, and shared
revenues.
As in the federal revenue sharing program of the 1970s and 1980s and those found
in all but a handful of states, shared revenues are transferred to local governments with
no strings attached. Local governments are free to use these funds in any way they
deem appropriate. These funds can be used to replace locally generated property tax
218 Steven C. Deller and Craig S. Maher
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
dollars, fund programs at higher levels, or some combination of the two. For some
municipalities, shared revenues account for a large percentage of operating dollars. For
the City of Beloit, for example, over half of the annual operating budget comes from
shared revenues. In FY 2000 the average city in Wisconsin was dependent on shared
revenues for slightly more then one in every five dollars of revenue. Approximately
15 percent of village and almost 14 percent of town total revenues are from shared
revenues.
The fact that Wisconsin operates such a large program with no strings attached
makes it somewhat unique in the United States. According to a Michigan Senate Fiscal
Agency issue paper, only Nevada distributed a larger share of total state general
expenditures through unrestricted payments to local governments (11.5 percent
compared with Wisconsin’s 9.3 percent).
24
Furthermore, only eleven states allocated
more than 4 percent of total general expenditures to unrestricted shared revenue
programs. Although the Wisconsin shared revenue program is atypical (hence,
perhaps, calling into question how transferable our results may be), it provides a clean
test of the asymmetry of the flypaper effect. Before we can directly test for patterns in
how shared revenues are used by Wisconsin local governments, it is important to gain
an appreciation of the current political environment in the state. The shared revenues
program has been in place for almost three decades and has been a stable and growing
source of revenues for local governments in Wisconsin. The structural deficit that has
faced Wisconsin state government over the past decade directly threatens the shared
revenues program.
In an effort to reduce property taxes, beginning in FY 1997, the state committed
itself to directly fund two-thirds of K-12 school districts’ costs.
25
Largely as a result of
this commitment, state support of K-12 school aids currently accounts for 44 percent
of total state general-purpose expenditures. These school aids, combined with shared
revenues, accounted for more than half of the state’s general-purpose budget in FY
2003. As seen in figure 1, the state began to curtail its commitment to shared revenues
in 1995–1996 in anticipation of the new commitment to K-12 education.
The amount of state support for local programs in Wisconsin, coupled with its
historically high tax rates, has led to debate about the merits of the fiscal relationship
between the state and its local governments. Within the past few years governors have
created a commission (Wisconsin Blue Ribbon Commission on State and Local
Partnerships for the 21st Century, known as the Kettl Commission) and task force
(State of Wisconsin Task Force on State and Local Government, known as the Sheehey
Task Force) to provide recommendations for ‘‘repairing’’ this relationship. The
language used in the Kettl Commission report is consistent with the thinking reported
in fiscal illusion literature:
the Commission has sought to strengthen accountability in the system by
aligning, to the greatest extent possible, the level of government that sets policy
A Model of Asymmetries in the Flypaper Effect 219
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
with the level that raises the revenue. ... This is a critical issue because of the
complicated patterns by which Wisconsin’s governments raise and spend
money.
Former governor McCallum went so far as to propose the elimination of the shared
revenues program, accusing local governments of being ‘‘big spenders.’’ In the
subsequent debates the state’s shared revenue program was described as ‘‘enabling’’ the
‘‘out of control’’ spending of local officials. The argument used for such a proposal is
that local spending, and hence local taxes, are stimulated by unconditional aid, such as
shared revenues.
More directly, when one considers the major categories of expenditures by state
government in Wisconsin, the top four categories are K-12 education, Medicaid, the
university system, and shared revenues. Most other categories of expenditures are
relatively small in comparison with the ‘‘big four.’’ The shared revenues program, in
essence, has a bull’s-eye squarely on it. Many within Wisconsin have argued that the
state-local relationship debate is a cover for the elimination of the shared revenues
program.
A Model of Structural Change
To test for structural shifts we follow Deller and Walzer and specify a relationship
between spending (E), state shared revenues (S), and income (I) over two time periods
(tand t– 1):
26
Et1¼bt1It1þat1St1þdXþet1, ð1Þ
Et¼btItþatStþdXþet:ð2Þ
Here Xis a set of control variables whose relationship is hypothesized to have
remained constant over time and eis a well-behaved error term. Combining the two
equations to obtain change over time yields
EtEt1¼btItþatStþdXbt1It1at1St1dXþetet1:ð3Þ
Rearrange terms and we have
EtEt1
ðÞ¼btItbt1It1
ðÞþatStat1St1
ðÞþetþet1
ðÞ:ð4Þ
Note that the set of control variables (X) drops out of the analysis. Given our
framework there is no change in the influence of these control variables over time;
hence they are removed from the analysis. Now add and subtract b
t
I
t–1
and a
t
S
t–1
to
yield:
EtEt1
ðÞ¼btItbt1It1
ðÞþbtIt1btIt1
ðÞþatStat1St1
ðÞ
þatSt1atSt1
ðÞþetþet1
ðÞ:ð5Þ
220 Steven C. Deller and Craig S. Maher
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
Rearrange terms and simplify:
EtEt1
ðÞ¼btbt1
ðÞIt1þbtItIt1
ðÞþatat1
ðÞSt1þatStSt1
ðÞ
þetþet1
ðÞ:ð6Þ
Define DE(E
t
E
t1
), Db(b
t
b
t1
), DI(I
t
I
t1
), Da(a
t
a
t1
),
DS(S
t
S
t1
), and «(e
t
e
t1
), and the equation to be estimated can be
stated as
DE¼DbIt1þbtDIþDaSt1þatDSþ«:ð7Þ
Our empirical model then focuses on per capita expenditures for two time periods, per
capita income for two time periods, and finally shared revenues per capita for two time
periods.
This specification provides us with two parameters to focus attention on. First, Da
allows us to tell whether there is a structural change in how aids are treated and,
second, a
t
allows us to focus on asymmetries. Based on the discussion of asymmetries
in intergovernmental aid advanced by Gramlich, Deller and Walzer, Heyndels,
Gamkhar and Oates, Melo, and Levaggi and Zanola, we suggest that
at¼0!symmetry; at>0!fical replacement; at<0!inducement:27
This formulation provides us with three separate tests. First, the standard flypaper
result can be examined by directly comparing the coefficients on the change variables
(b
t
versus a
t
). Second, we can see whether the influence of aids has changed over time
(Da). Finally, we have the indirect test of symmetry as outlined above. We could also
think of the symmetry question in light of the change in aids coefficient (Da). If aids
become more stimulative over a period in which aids are declining, one could
conclude that fiscal replacement is occurring. Specifically, if a dollar in aids has a
stronger impact after a period of decline, the only way to explain the increased
stimulative effect is through fiscal replacement. We could extend the logic to aids
having a less stimulative effect being evidence of inducement.
To test how local officials treat unconditional aid from the state we construct and
estimate a series of linear equations using a cross section of data for 1,778 Wisconsin
cities, villages, and towns. Given the audited fiscal data from the Wisconsin
Department of Revenue and the socioeconomic data from the 1990 and 2000
censuses, we are able to directly estimate equation (7). To minimize the potential
presence of ‘‘spikes’’ in annual revenue and expenditure data we use an average
between 1989 and 1990 for the beginning of the period and an average between 1999
and 2000 for the end of the period.
A potential problem with using the two census years (1990 and 2000) is that aid
levels in 2000 are consistently higher in 2000 than in 1990. This may lead to the
statistical problem widely referred to as ‘‘regression to the mean.’’ As described above,
A Model of Asymmetries in the Flypaper Effect 221
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
the Wisconsin shared revenues program did not enter its period of uncertainty and
decline until 1995, after the beginning of the study period. If one places a trend line
between 1990 aid levels and 2000 levels, the slope is positive, perhaps resulting in the
regression to the mean problem. Unfortunately, the richness of data provided by the
decennial census is not available on an annual basis. We would argue that we are
attempting to model behavioral changes over time and that 1990 represents a period of
aid growth and certainty whereas 2000 represents a period of aid decline and
uncertainty. Although regression to the mean may be a problem, we believe that the
behavioral differences should be sufficient to enable them to be drawn out from
the data.
28
Dependent Variables
Our detailed audited data allow a closer examination of the impact of unconditional
aid on specific types of expenditures. Earlier work by the authors demonstrated
different flypaper effects depending on the type of expenditure.
29
In general, the
flypaper effect was greatest on ‘‘luxury’’ goods (e.g., culture and education, parks and
recreation) and had the weakest impact on ‘‘essential’’ goods (e.g., police and fire
protection). This raises the question of whether similar asymmetry effects will be
found based on the type of expenditure. Thus in addition to looking at total
expenditures per capita, we also look at per capita expenditures on protective services
(police, fire, and ambulatory), roads (road maintenance exclusive of construction),
waste disposal (solid waste collection and disposal and wastewater treatment), and
what we deem ‘‘quality of life’’ services (culture and education including public
libraries, parks and recreation, and conservation and development efforts).
Independent Variables
The direct specification of equation (7) requires data on per capita income and per
capita shared revenues. In addition, because of the great heterogeneity of Wisconsin
municipalities in terms of size (ranging from large urban places such as Milwaukee,
Madison, and Green Bay to numerous small rural places), we have also included a
number of households to capture scale effects. Thus, the base equation has per capita
income, per capita shared revenues, and number of households in 1990 (t– 1) and
change in per capita income, per capita shared revenues, and number of households.
In addition to these base variables, we estimate a second specification drawing on
variables we hypothesize would be contained in the vector of control variables in
equations (1) and (2), which are removed from the final specification of equation (7).
We include these as lagged variables to test the sensitivity of the results for the core
model. We include the share of employment in more professional industries (finance,
insurance, and real estate, professional related services, and public administration)
based on the work of Hendrick.
30
Other demographic factors previously found to
222 Steven C. Deller and Craig S. Maher
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
affect spending behavior include the percentage of the population eighteen years of
age or younger, the percentage of the population aged twenty-five or older with at
least a bachelor’s degree, the median value of owner-occupied homes, the percentage
of the employment base in manufacturing, and property taxes per capita.
31
We offer
no specific hypotheses on how we might expect these control variables to influence
changes in expenditures but rather introduce them to test for the stability of the
parameter estimates of the core model. All these control variables are for 1990 (t– 1).
Findings
The results of our analysis are reported in tables 1 and 2. Consider first the core
equation for total expenditures (table 1, Model A). Given the specification of the
structural change model we see evidence of the flypaper effect (b¼0.0462 <
a¼5.8379) and some evidence of economies of size in production given the negative
coefficient associated with the change in number of households. Looking at the change
coefficients that are associated with the base variables, we see that there is no structural
shift with respect to per capita income and number of households, but there does
Table 1 Change in per capita total expenditures
Model A Model B
Shared revenues per capita 1990 (Da) 3.4179 (24.43) 3.3448 (23.10)
Change in shared revenues per capita 1990–2000 (a) 5.8379 (39.64) 5.7753 (38.09)
Per capita income 1990 (Db)0.0050 (0.67) 0.0223 (1.50)
Change in per capita income 1990–2000 (b) 0.0462 (5.39) 0.0420 (4.27)
Number of households 1990 0.0419 (1.00) 0.0751 (1.74)
Change in number of households 1990–2000 0.1602 (1.76) 0.1541 (1.72)
Percentage of population with college degree 1990 — 419.7394 (0.67)
Percentage of population under age twenty 1990 — 893.2513 (1.44)
Percentage of employment in manufacturing 1990 — 491.1551 (1.66)
Percentage of employment in professional services 1990 — 986.6156 (2.06)
Percentage of households with income
less than $15,000 1990 — 1,120.7206 (3.34)
Median house value 1990 — 0.0043 (1.68)
Property taxes per capita 1990 — 2.4099 (8.07)
Intercept 464.2062 (4.62) 179.1481 (0.52)
Adjusted R
2
0.7150 0.7280
F-statistic 744.50 367.10
Sample size 1,778 1,778
Condition index 10.85 47.02
Heteroskedasicity chi-square (marginal significance) 67.69 (0.0001) 179.68 (0.0001)
A Model of Asymmetries in the Flypaper Effect 223
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
Table 2 Change in per capita expenditures—selected services
Protective services Road work Solid waste Quality of life
Shared revenues per capita 1990 (Da) 0.3202 (10.21) 0.0127 (4.39) 0.6450 (11.71) 0.4702 (18.77)
Change in shared revenues per capita 1990–2000 (a) 0.8763 (26.53) 0.0604 (19.75) 0.7025 (12.12) 1.0408 (39.48)
Per capita income 1990 (Db)0.0003 (0.16) 0.0001 (0.46) 0.0005 (0.16) 0.0042 (3.15)
Change in per capita income 1990–2000 (b) 0.0060 (3.12) 0.0001 (0.83) 0.0105 (3.10) 0.0057 (3.68)
Number of households 1990 0.0265 (2.81) 0.0001 (0.07) 0.0129 (0.78) 0.0010 (0.14)
Change in number of households 1990–2000 0.0835 (4.10) 0.0009 (0.48) 0.0019 (0.05) 0.0059 (0.37)
Percentage of population with college degree 1990 — — — —
Percentage of population under age twenty 1990 — — — —
Percentage of employment in manufacturing 1990 — — — —
Percentage of employment in professional services 1990 — — — —
Percentage of households with income less than $15,000 1990 — — — —
Median house value 1990 — — — —
Property taxes per capita 1990 — — — —
Intercept 49.9392 (2.21) 2.6288 (1.26) 106.4090 (2.69) 120.6233 (6.70)
Adjusted R
2
0.7074 0.6648 0.0777 0.7975
F-statistic 717.36 588.69 25.96 1,168.02
Sample size 1,778 1,778 1,778 1,778
Condition index 10.85 10.85 10.85 10.85
Heteroskedasicity chi-square (marginal significance) 34.37 (0.1556) 22.69 (0.7018) 27.43 (0.4408) 49.01 (0.0059)
224 Steven C. Deller and Craig S. Maher
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
appear to be a structural shift with respect to the treatment of shared revenues
(Da¼3.4179). This result supports our central hypothesis that local officials treat
shared revenues differently at the end of the period than at the beginning of the period.
Turning attention to the issue of asymmetry, the positive and statistically significant
coefficient on change in shared revenues per capita suggests that local governments in
Wisconsin are likely to replace lost shared revenues with own-source revenues. If we
expand the model to include a set of control variables (table 1, Model B) we see that
the results of the core model are consistent in terms of coefficient signs and
magnitude.
32
If we turn attention to the subcategory of expenditures we see a similar pattern.
Consider protective services expenditures (table 2). Again, given the specification of
the structural change model the flypaper effect is supported (b¼0.0060 <
a¼0.8763). The negative and statistically significant coefficient on change in
number of households suggests economies of scale in the production of protective
services. We also see that there is a structural shift in how shared revenues are treated
in funding protective services (Da¼0.3202). As with total expenditures per capita,
the positive and significant coefficient on change in shared revenue per capita suggests
that local officials in Wisconsin will replace lost shared revenues with own-source
revenues.
For road maintenance expenditures we again see evidence of the flypaper effect, but
the magnitudes of the coefficients are consistently smaller than those for total
expenditures and other categories of expenditures. The relatively small size of the
structural coefficient on road maintenance compared with total expenditures is
expected given the presence of a large road aid program in Wisconsin. In other words,
because there is a large dedicated road aid program local municipalities tend to use
only a small portion of shared revenues on road maintenance expenditures. There is
also evidence that fiscal replacement is present, but again the relatively small
coefficient suggests that the road aid fund in Wisconsin is sufficiently large to afford
municipalities the flexibility to use shared revenues on other functions.
A similar pattern is uncovered for both waste management services and what we
call quality of life services. In the case of both service areas the impact of shared
revenues per capita is larger than that of per capita income, supporting the notion of
the flypaper effect. There is no evidence of a structural shift in how income influences
waste management services, but a positive and statistically significant coefficient on
per capita income for quality of life services suggests that as income increases the
demand for things such as libraries and cultural and educational services increases at a
faster rate. In both waste and quality of life services, the positive and statistically
significant coefficient on shared revenues per capita (Da) suggests that how local
officials treated shared revenues changed during the 1990s. The only conclusion that
can be drawn is that shared revenues became more stimulative over time or there is
evidence of fiscal replacement.
A Model of Asymmetries in the Flypaper Effect 225
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
What these findings suggest is that our core model provides us with stable results
and evidence supporting the notion of the flypaper effect; that the magnitude of the
flypaper effect has changed over time—specifically shared revenues have become more
stimulative over time; and finally, that there is strong evidence of fiscal replacement.
We suggest that the logic outlined by Deller and Walzer is worth additional
consideration.
33
Specifically, in periods of stability and growth in intergovernmental
aids, local officials are likely to treat the transfers as a permanent source of income and
build it into their base budgets (i.e., the aides are less stimulative). But as these aids
become uncertain, local officials will treat them as transitory and they will have a more
stimulative effect. In the logic of Heyndels, aids are asymmetric.
34
Conclusion
This study adds to recent research on asymmetries in the flypaper effect relating to
whether recipient units of government respond in the same fashion to cuts in
unconditional aid as they do to increases. The existing research has concluded that
recipient governments do not act the same when aid is cut. The most common
explanation of this asymmetry in the flypaper effect is that once programs are added,
they become entrenched and are thus difficult to cut when aid is reduced.
35
This study is an important contribution to this line of research because it examines
the impact of one of the largest unconditional aid programs in the United States. In
addition, Wisconsin’s shared revenue program has experienced very little growth over
the past decade, which resulted in over half of the state’s municipalities losing aid
during the 1990s. Our findings add to the growing evidence that recipients of
unconditional aid do not act the same when funds are cut. In general, Wisconsin
communities that received more shared revenues also spend more than similar
communities that received less shared revenues, controlling for personal income
(flypaper effect). In addition, Wisconsin communities that experienced reductions in
shared revenues over the decade generally did not cut spending in proportion to those
cuts (fiscal replacement).
Moreover, although every other ‘‘asymmetries’’ study has looked at the impact on
total expenditures, our earlier research suggests that the impact of aid varies by
expenditure type.
36
This study once again demonstrates the importance of examining
expenditures at the subtotal level. Although spending on protective services follows a
pattern consistent with total expenditures, the same cannot be said about spending on
transportation, waste management, and quality of life services. The findings seem to
make sense in that despite reductions in intergovernmental aid, policymakers will be
less inclined to cut police and fire services than ‘‘quality of life’’ services such as parks
and recreation services.
In more general terms, this study helps us understand the impact of recent state
cuts in intergovernmental aid due to poor fiscal conditions. In the short term, local
226 Steven C. Deller and Craig S. Maher
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
governments will be inclined to replace the reductions with local funds—at least for
protective services. Other quality of life services will not fare quite as favorably.
By employing a model of structural change we offer more than one way to think
about and measure asymmetry. In addition to examining the nature of the effect—
fiscal replacement versus inducement versus symmetry—we suggest that the logic
outlined by Deller and Walzer is worth additional consideration.
37
Specifically, in
periods of stability and growth in intergovernmental aids, local officials are likely to
treat the transfers as a permanent source of income and build it into their base budgets
(i.e., the aid is less stimulative). But as these aids become uncertain, local officials will
treat them as transitory and they will have a more stimulative effect.
Although this study moves the literature one step forward, it falls far short of
providing a definitive answer. Issues related to regression to the mean and potential
endogeneity problems must be further examined. Perhaps more important, our
theoretical foundations remain weak. Although the work by Heyndels provides a
framework for thinking about the problem in a descriptive way and the permanent-
transitory income approach provides an alternative view, the theoretical approaches
are still rooted in median voter theory.
38
Clearly, additional theoretical work is
required.
Notes
1. David Bradford and Wallace Oates, ‘‘Toward a Predictive Theory of Intergovernmental
Grants,’’ American Economics Review 61 (August 1971); Edward M. Gramlich and Harvey
Galper, ‘‘State and Local Fiscal Policy Behavior and Federal Grant Policy,’’ Brooking Papers
on Economic Activity 1 (1973); Edward M. Gramlich, ‘‘Intergovernmental Grants, a Review
of the Empirical Literature,’’ The Political Economy of Fiscal Federalism, ed. Wallace Oates
(Lexington, MA, 1977), pp. 219–239; Paul N. Cournat, Edward M. Gramlich, and Daniel
L. Rubinfeld, ‘‘The Stimulative Effect of Intergovernmental Grants, or Why Money Sticks
Where It Hits,’’ Fiscal Federalism and Grants-In-Aid, ed. Peter Meiszkowski and William
H. Oakland (Urban Institute, 1979).
2. Stephen Bailey and Stephen Connolly, ‘‘The Flypaper Effect: Identifying Areas for Further
Research,’’ Public Choice (June 1998): 335–361.
3. Edward M. Gramlich, ‘‘Federalism and Federal Deficit Reduction,’’ National Tax Journal
(September 1987): 299–313; Edwin J. Benton, ‘‘The Effects of Changes in Federal Aid on
State and Local Government Spending,’’ Publius 22 (Winter 1992): 71–82; William F. Stine,
‘‘Is Local Government Revenue Responsive to Federal Aid Symmetrically? Evidence from
Pennsylvania County Governments in an Era of Retrenchment,’’ National Tax Journal 57
(December 1994): 779–816; Steven C. Deller and Norman Walzer, ‘‘Structural Shifts in the
Treatment of Intergovernmental Aid: The Case of Rural Roads,’’ Journal of Agricultural and
Applied Economics 27 (1995): 522–535; Shama Gamkhar and Wallace Oates, ‘‘Asymmetries
in the Response to Increases and Decreases in Intergovernmental Grants: Some Empirical
Findings,’’ National Tax Journal 49 (December 1996): 501–512; Bruno Heyndels,
A Model of Asymmetries in the Flypaper Effect 227
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
‘‘Asymmetries in the Flypaper Effect: Empirical Evidence for the Flemish Municipalities,’’
Applied Economics 33 (2001): 1329–1334; Ligia Melo, ‘‘The Flypaper Effect under Different
Institutional Contexts: The Colombian Case,’’ Public Choice 111 (April 2002): 317–345;
Rosella Levaggi and Roberto Zanola, ‘‘Flypaper Effect and Sluggishness: Evidence from
Regional Health Expenditures in Italy,’’ International Tax and Public Finance 10 (September
2003): 535–547.
4. For an excellent description of the rise and fall of the federal revenue program see Bruce
Wallin, From Revenue Sharing to Deficit Sharing: General Revenue Sharing and Cities
(Washington, DC: Georgetown University Press, 1998). See also Steven C. Deller, Craig
S. Maher, and Victor Illudo, ‘‘Wisconsin Local Government, State Shared Revenues and
the Illusive Flypaper Effect,’’ Public Budgeting, Accounting and Financial Management
(forthcoming).
5. Steven C. Deller and Craig S. Maher, ‘‘Categorical Municipal Expenditures with a Focus on
the Flypaper Effect,’’ Public Budgeting and Finance 25 (September 2005): 73–90.
6. In a democracy, theory suggests that the voter in the middle of the political spectrum, the
median voter, will be the swing voter and that political parties will tend to migrate to
the middle, reflecting the preferences of the median voter. The key then to understanding
the demands for public policy is to understand the preferences of the median voter.
Attempts to provide an alternative to the median voter approach have suggested that the
median voter approach is too simplistic. For example, Roemer and Silvestre, ‘‘The Flypaper
Effect Is Not an Anomaly,’’ Journal of Public Economic Theory 4 (January 2002): 1–17,
provide a game-theoretic approach in which theoretical results are consistent and agree with
the empirical literature. Unfortunately, their theoretical model does not lend itself to easy
empirical testing. Clearly, additional theoretical work is required. See John Chubb, ‘‘The
Political Economy of Federalism,’’ American Political Science Review 79 (1985): 994–1015;
Mala Lalvani, ‘‘The Flypaper Effect: Evidence from India,’’ Public Budgeting and Finance 22
(September 2002): 67–88.
7. Bradford and Oates, ‘‘Toward a Predictive Theory’’; Gramlich and Galper, ‘‘State and Local
Fiscal Policy Behavior’’; Gramlich, ‘‘Intergovernmental Grants.’’
8. Chubb, ‘‘The Political Economy of Federalism.’’
9. Philip J. Grossman, ‘‘The Impact of Federal and State Grants on Local Government
Spending: A Test of the Fiscal Illusion Hypothesis,’’ Public Finance Quarterly 18 (1990):
313–327.
10. Federal unconditional, or block, grants were excluded since they rarely exist today: Lalvani,
‘‘The Flypaper Effect.’’ For more comprehensive reviews of the flypaper literature see
Elizabeth Becker, ‘‘The Illusion of Fiscal Illusion: Unsticking the Flypaper Effect’’ Public
Choice (1996): 85–102; Brian E. Dollary and Andrew C. Worthington, ‘‘The Empirical
Analysis of Fiscal Illusion,’’ Journal of Economic Surveys 10 (September 1996): 261–297;
Bailey and Connolly, ‘‘The Flypaper Effect.’’
11. Heyndels, ‘‘Asymmetries in the Flypaper Effect.’’
12. Gramlich, ‘‘Intergovernmental Grants.’’
228 Steven C. Deller and Craig S. Maher
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
13. Heyndels, ‘‘Asymmetries in the Flypaper Effect.’’
14. Deller and Walzer, ‘‘Structural Shifts.’’
15. Benton, ‘‘Effects of Changes in Federal Aid.’’
16. Gramlich, ‘‘Federalism and Federal Deficit Reduction.’’
17. Levaggi and Zanola, ‘‘Flypaper Effect and Sluggishness.’’
18. Melo, ‘‘Flypaper Effect.’’
19. Deller and Walzer, ‘‘Structural Shifts.’’
20. Stine, ‘‘Is Local Government Revenue Responsive?’’
21. Gamkhar and Oates, ‘‘Asymmetries in the Response to Increases and Decreases in
Intergovernmental Grants.’’
22. U.S. Bureau of Census.
23. Michigan Senate Fiscal Agency, ‘‘Michigan’s Revenue Sharing Program: History, Descrip-
tion, and A Brief Comparison to Other States’’ (1996).
24. Ibid.
25. Wisconsin Legislative Fiscal Bureau, ‘‘Shared Revenue Program: Informational Paper #18’’
(2003).
26. Deller and Walzer, ‘‘Structural Shifts.’’
27. Gramlich, ‘‘Federalism and Federal Deficit Reduction’’; Deller and Walzer, ‘‘Structural
Shifts’’; Gamkhar and Oates, ‘‘Asymmetries in the Response to Increases and Decreases in
Intergovernmental Grants’’; Heyndels, ‘‘Asymmetries in the Flypaper Effect’’; Melo,
‘‘Flypaper Effect’’; Levaggi and Zanola, ‘‘Flypaper Effect and Sluggishness.’’
28. A second potential empirical problem is that of endogeneity with the variables of equation
(7), specifically the change variables DE,DI, and DS. If the Tiebout hypothesis that levels of
public services influence economic performance is correct, then we should expect feedback
that is not captured in our simple linear regression model. We acknowledge this potential
problem and leave it for future work.
29. Deller and Maher, ‘‘Categorical Municipal Expenditures.’’
30. Rebecca Hendrick, ‘‘Revenue Diversification: Fiscal Illusion or Flexible Financial Manage-
ment,’’ Public Budgeting and Finance 22 (December 2002): 52–72.
31. Deller and Maher, ‘‘Categorical Municipal Expenditures.’’
32. Similar results were found when we examined the subcategory expenditures.
33. Deller and Walzer, ‘‘Structural Shifts.’’
34. Heyndels, ‘‘Asymmetries in the Flypaper Effect.’’
35. Gramlich, ‘‘Federalism and Federal Deficit Reduction.’’
36. Deller and Maher, ‘‘Categorical Municipal Expenditures.’’
37. Deller and Walzer, ‘‘Structural Shifts.’’
38. Heyndels, ‘‘Asymmetries in the Flypaper Effect.’’
A Model of Asymmetries in the Flypaper Effect 229
at University of Wisconsin-Madison Libraries on February 19, 2015http://publius.oxfordjournals.org/Downloaded from
