General Interest

State and Local Government Review

2018, Vol. 50(2) 119-131
Fiscal Decentralization and ª The Author(s) 2018
Article reuse guidelines:

Financial Condition: The sagepub.com/journals-permissions

DOI: 10.1177/0160323X18794014
journals.sagepub.com/home/slg
Effects of Revenue and
Expenditure Decentralization
on State Financial Health

Akheil Singla1 and Samuel B. Stone2

Abstract
Although there are strong theoretical arguments about both the benefits and costs of decen-
tralization in the federalism literature, there is little on how second-order fiscal decentralization
effects the financial health of state governments. This study examines this question and adds to the
understanding of state-and-local fiscal relations. Using financial indicators that measure several
dimensions of financial condition, the research estimates the effect of revenue and expenditure
decentralization on state fiscal health. It finds that while state financial condition is unaffected by
revenue decentralization, there is a curvilinear relationship between expenditure decentralization
and long-term state financial condition.

Keywords
fiscal federalism, financial condition, decentralization, devolution, financial health

Recent political gridlock at the federal level has devolution, the empirical work on the subject
had an interesting effect on American federal- does not demonstrate conclusive evidence
ism. States have stepped into the void created about what other effects these trends might
by the absence of federal policy-making, taking have on state or local governments. Devolution
the lead in a variety of policy arenas (Gamkhar can improve programmatic outcomes but doing
and Pickerill 2012; Goelzhauser and Rose so frequently requires moving resources from
2017). This first-order devolution from federal one level of government to another, which can
to state government has dominated most con-
versations about federalism. A critical but less 1
School of Public Affairs, Arizona State University, Phoenix,
covered development coming out of the Great
AZ, USA
Recession has been the increased importance 2
Division of Politics, Administration & Justice, California
of local governments in the policy process. State University, Fullerton, Fullerton, CA, USA
States have tasked local governments with
“a greater role as the true laboratories of Corresponding Author:
Akheil Singla, School of Public Affairs, Arizona State
democracy” (Daigneau 2011). University, 411 N. Central Ave., Suite #400, Phoenix, AZ
Although there are a large number of theore- 85004, USA.
tical arguments about the costs and benefits of Email: akheil.singla@asu.edu
120 State and Local Government Review 50(2)

erode the resource base of one level of govern- relationship between states and their local gov-
ment. This is an important area of study ernments. This relationship is known as
because the reassignment of responsibilities second-order federalism and the shifting of
between one level of government and another responsibility from states to local governments
can represent a massive transfer of obligations is usually called second-order devolution.
and resources. For example, states experiencing Studying state and local interactions poses dif-
fiscal crises respond by devolving responsibil- ferent challenges from the federal–state inter-
ities down to local governments, which, in turn, action, as local governments are the political
experience fiscal pressure of their own in order creations of the states; for this reason, state–
to meet their new obligations (Reschovsky local relations might be best described as pater-
2004). This study addresses this issue by exam- nalistic (Berman 2003; Zimmerman 1995).
ining what effect fiscal decentralization (the Recent literature on American federalism tends
fiscal result of devolution) to local govern- to more closely address second-order federal-
ments within a state has on the financial health ism and is correspondingly less certain about
of state governments. the benefits of decentralization (Oates 2008;
Weingast 2009).
An example of this is the problem of the soft
Background & Theory budget constraint. That is, in times of fiscal cri-
sis, lower-level governments can seemingly fall
Second-order Federalism back on a higher level government to bail them
Fiscal federalism is “the vertical structure of out (Kornai, Maskin, and Roland 2003). Rather
the public sector. It explores, both in norma- than facing a hard ceiling on their budgets that
tive and positive terms, the roles of the differ- would encourage more moderation in spending
ent levels of government and the ways in or forecasting, local governments have a “soft”
which they relate to one another . . . ” (Oates cap that might be lifted by the state via emer-
1999, 1120). American federalism, in particu- gency assistance like extra intergovernmental
lar, has a more decentralized structure of aid (Qian and Roland 1998). This system of
authority, as the Constitution gives all powers intergovernmental grants in aid can result in
that are not specifically enumerated to the fed- dependency by local governments on higher-
eral government to the states. The earlier liter- level governments, which creates an incentive
ature on federalism contends that though to defy fiscal discipline (Qian and Roland
centralized and decentralized systems can 1998). Empirical studies (Saxton et al. 2002;
both adequately provide public goods and Kloha, Weissert, and Kleine 2005) note that
social welfare functions through redistribu- diminishing intergovernmental aid from states
tion, decentralized systems are more efficient to local governments often results in fiscal
at matching preferences to local public goods stress for local governments, which, in turn, can
(Musgrave 1959, 1965; Olson 1969; Oates require state intervention.
1972, 1977, 1991). Additionally, there are Centralization, on the other hand, can force
strong historical, political, and cultural norms the internalization of local jurisdictional
in favor of local self-government in the United externalities, which may more equitably distri-
States (Ostrom, Bish, and Ostrom 1988; Bru- bute welfare, but may be costlier for local gov-
nori 2007). Although there are some excep- ernments (Goodspeed 2002; Kornai, Maskin,
tions (Mclure 1967), there seemed to be and Roland 2003). Additionally, because the
general agreement that the American system central government pays for grants with taxes
of fiscal decentralization more closely meets levied across all lower level governments, the
these ideals (in theory if not always in cost to a single local government using the
practice). grants to pay back debt is less than the cost of
However, it is not clear whether this empha- the higher-level tax on its own residents. Sub-
sis on decentralization always extends to the national governments thus have a strong
Singla and Stone 121

incentive to rely on intergovernmental grants. which mandated that state and local govern-
The proposed solution to this problem is a sys- ments produce government-wide (in addition
tem of independent local revenues. to fund-based) financial statements using
There are a number of empirical studies that accrual accounting (Government Accounting
deal directly with the centralization or decen- Standards Board 1999). These statements have
tralization of the fiscal relationships between led to a new body of research that describes dif-
states and their local governments. One contri- ferent measures that capture financial condition.
buting factor to centralization within a state Most of the current research uses ratios that
was the trend of state governments creating measure financial condition as solvency (i.e.,
departments and agencies as corollaries to fed- ability to meet obligations) in four areas: cash,
eral departments and agencies in order to man- budget, long-term, and service (Wang, Dennis,
age federal aid (Stephens 1974; Fisher 2003). and Tu 2007). Cash solvency captures the abil-
There is little evidence of regional trends in ity to meet obligations in the immediate future;
centralization; in fact, centralization is highly budgetary solvency focuses on the ability to
variable between states (Bowman and Kearny meet obligations over a budgetary cycle of
2011). But there is a strong negative associa- one-to-two years; long-term solvency focuses
tion between population and centralization on the ability to meet long-range obligations
(Stephens 1974). The decades of the 1930s and like pensions or municipal bonds that extend
1960s witnessed the most centralization from out over multiple decades. Service solvency
local to state government, plateauing in the focuses on a government’s ability to provide
1980s (Bowman and Kearny 2011). States that essential services to citizens. These different
are geographically larger, more populous, dimensions of solvency are independent, mean-
more urban, have lower per capita income, and ing a government may be solvent in the near
are not located in the south will be less fiscally term (e.g., be able to pay its employees) while
centralized (Wallis and Oates 1988). In an being insolvent in the long term (e.g., be unable
exploration of the determinants of debt decen- to make payments to bond holders). Others
tralization, Greer and Denison (2016) find have built upon this framework, adding to or
political and institutional factors, rather than subdividing these four areas, or suggesting
financial or economic ones, to be the biggest additional and alternative metrics by which one
drivers of decentralization. might capture solvency (Rivenbark, Roenigk,
In sum, the theoretical discussions about the and Allison 2010; Johnson, Kioko, and Hildreth
relationship between higher- and lower-level 2012; Ross, Yan, and Johnson 2015). The indi-
governments tend to emphasize that their fiscal cators articulated by the literature have been
resources and responsibilities are closely inter- found to be individually reliable predictors of
twined and are dynamic. The next section dis- financial distress, validating the metrics as
cusses the concept of financial condition measures of financial health (Stone et al. 2015).
before considering the relationship between it
and devolution.
The Relationship between Fiscal
Decentralization and Financial Condition
Financial Condition Although the topics of second-order devolution
Financial condition analysis refers to the mea- and financial condition have each received con-
surement of the fiscal health of a government. siderable attention on their own, there is very
The construct attempts to capture a govern- little literature that attempts to tie the two
ment’s ability to meet its obligations—to citi- together. What work there is tends to focus on
zens and creditors alike—over a variety of the effects of devolution on the lower level of
time horizons. The more recent scholarship government. The present study is distinctive
draws on the advent of Government Account- because it explores the effects of decentraliza-
ing Standards Board statement number 34, tion on the financial condition of state
122 State and Local Government Review 50(2)

governments. Nevertheless, it draws on the However, there is a line of research that sug-
prior work in this area to derive hypotheses gests that devolution is a net positive for gov-
about the effects of second-order devolution ernments. This evidence exists in a variety of
on the financial condition of state govern- settings, ranging from improved performance
ments. At this point, it is also important to in individual programs (Kim and Fording
clarify the usage of the terms “devolution” and 2010) to broader indicators like economic
“decentralization.” Although the literature is growth (Akai and Sakata 2002). These studies
not quite of a single mind on this, and there harken to the potential benefits of devolution
is considerable overlap in their meanings, the and decentralization: increased efficiency of
present research defines the transfer of the allocation of public goods by more closely
functions, programs, and the administration aligning the provision of services with those
thereof from higher to lower level govern- who are receiving the benefits of those services
ments as devolution; the degree that fiscal pol- (Ostrom, Tiebout, and Warren 1961). Given
icies and activity carried out by lower rather these results, fiscal decentralization ought to
than higher-level governments is decentraliza- improve the financial condition of state govern-
tion. Fiscal decentralization represents a par- ments across the board.
ticular aspect of devolution that is the focus How, then, does one reconcile the diverging
of this research. expectations and findings on the effects of fis-
In the broadest sense, there is an intuitive cal decentralization on financial condition? A
connection between devolution and financial similar issue faced scholars of innovation, who
condition: the former deals with who controls suggested that while additional resources are
resources and is responsible for service deliv- required to implement many innovations, hav-
ery while the latter deals with the ability of a ing additional resources may reduce the incen-
government to use its resources to meet obliga- tive to innovate in the first place (Nohria and
tions. There should be a link between the two, Gulati 1996, 1997). A theoretical and empiri-
but the direction is perhaps more complex than cally supported solution has been that the rela-
one might think. On the one hand, the level of tionship between resources and innovation was
government with more resources and/or less nonlinear; some resources are necessary to
burdens ought to have a stronger financial con- innovate, but too many and organizations lack
dition. On the other hand, there is evidence the will to take the risks inherent to innovative
that devolution can improve the performance activity (Singla, Stritch, and Feeney, forthcom-
of individual programs (Kim and Fording ing; Tan 2003; Nohria and Gulati 1997). It may
2010), which may suggest that an “optimal be the case that the relationship between fiscal
devolution” strategy could improve the finan- decentralization and financial condition is sim-
cial condition of all governments. ilar in that some decentralization is beneficial,
There are a number of studies to support the but too much is harmful. On the revenue side,
idea that the level of government with more some decentralization allows local govern-
access to resources or fewer service obligations ments to be self-sufficient and reduces the need
will have a stronger financial health (Jimenez for state-level resource transfers, but too much
2009; Stone 2015; Reschovsky 2004; Baicker decentralization eats into the state’s own access
and Gordon 2006). The concurrent themes in to resources. On the expenditure side, some
these studies are that access to additional decentralization may generate efficiencies by
resources are beneficial to fiscal health and more closely linking public goods provision
larger demands placed on existing resource to citizens, but too much may overburden local
pools via increased service obligations tend to governments and increase the need for state
harm fiscal health. By this logic, the financial intervention, oversight, or financial assistance.
condition of state governments should improve In both cases, one might expect to observe a
when expenditures are decentralized and reven- curvilinear relationship (e.g., an inverted “U”
ues are not. shape) between fiscal decentralization and state
Singla and Stone 123

financial condition. There is mixed evidence of 2015). This allows empirical work on financial
such a relationship with respect to effects of condition to select a single measure from a bin
law enforcement department size on client out- rather than requiring the construction of a mul-
comes (Ostrom 1976). Based on this, one might tiple of measures within each one to capture the
expect a nonlinear relationship between finan- same construct. In addition, recent literature
cial condition and devolution. finds flaws with methods that attempt to cap-
It is possible to summarize the literature as ture the multidimensional nature of financial
follows: (1) there is substantial theoretical and condition via a single measure like an index
empirical evidence for a link between financial (Singla, Comeaux, and Kirschner 2014; Stone
condition and fiscal decentralization, (2) the et al. 2015; Clark 2015). This research opera-
direction of this relationship is unclear, and tionalizes financial condition with four mea-
(3) the nature (i.e., linear versus nonlinear) of sures: the current ratio, the operating ratio, the
this relationship is unclear. The principal goal long-term liability ratio, and the net asset ratio.
of the present research is to explore these issues It omits a measure of service solvency because
and shed light on the direction and nature of the recent work has found these measures to be
relationship between fiscal decentralization and inconsistent and poor predictors of financial
financial condition. condition (Stone et al. 2015). Instead, it
employs two measures of long-run solvency:
the net asset ratio1 and the long-term liability
Data and Method ratio. While each measures a similar construct,
one is focused on adequacy of assets and the
Data
other on the demands of liabilities. These mea-
This article explores the intersection of sures have also been shown to be externally
financial condition and fiscal decentralization. valid predictors of fiscal distress (Stone et al.
Specifically, it estimates the effect of decentra- 2015; Gorina, Maher, and Joffee 2018).
lization on the financial condition of the fifty The data to construct these measures come
states from 2005 to 2013. Addressing this ques- from state government Comprehensive Annual
tion requires operationalizing the two concepts, Financial Reports (CAFRs), specifically from
but as illustrated in the prior section, there are the statement of net assets and statement of
multiple ways to do so. This section describes activities for each state over the eight-year
how to measure the constructs and the data used period. These variables are constructed from
to do so. More detail on these variables and government-wide (i.e., governmental activities
their construction is available in the Online plus business-type activities) financial data pre-
Supplemental Table S1. sented on a full-accrual accounting basis.

Dependent variable: State financial condition. The

literature on the measurement of financial con- Independent variables: Fiscal decentralization.
dition has coalesced in recent years around a Studies of fiscal centralization or decentraliza-
more or less common set of ratios first tion typically use ratios of state activity to local
employed by Wang et al. (2007). Some of these activity. Stephens (1974) divides state spending
are also used in a number of other similar stud- by combined state and local spending. Wallis
ies (Rivenbark, Roenigk, and Allison 2010; and Oates (1988) use a similar measure, substi-
Johnson, Kioko, and Hildreth 2012; Ross, Yan, tuting expenses for revenues. As the purpose of
and Johnson 2015). Most financial condition this study is to explore the effects of decentra-
ratios attempt to measure one of the following lization rather than centralization, the local
constructs: short-term solvency, budgetary sol- share moves to the numerator to capture the
vency, long-run solvency, and service sol- decentralization of both revenues and expendi-
vency. When placed into these bins, ratios are tures. The decentralization measures come
highly correlated with each other (Stone et al. from the census of governments finance
124 State and Local Government Review 50(2)

Table 1. Linear Models, Fiscal Decentralization on State Financial Condition.

Long-term
Current Ratio Operating Ratio Net Asset Ratio Liability Ratio

Variables b SE b SE b SE b SE

Independent variables
(Lag) revenue 2.336** 0.69 0.34 0.30 0.10 0.12 0.228*** 0.04
decentralization
(Lag) expenditure 9.861** 3.94 0.65 0.43 0.60 0.33 0.08 0.23
decentralization
Economic controls
Unemployment 0.134* 0.06 0.0174** 0.01 0.0173*** 0.00 0.00 0.00
Real-state gross product 0.00 0.00 3.27e07* 0.00 6.64e07* 0.00 0.00 0.00
Log (total expenditures) 1.29 0.89 0.21 0.20 0.03 0.08 0.117*** 0.03
Median household income 0.00 0.00 0.00 0.00 1.19e05*** 0.00 1.39e05*** 0.00
Gini coefficient 19.95 13.64 1.452* 0.72 0.22 0.74 0.12 0.30
Demographic controls
Population 9.96e07** 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Black population 0.00 0.00 0.00 0.00 5.29e07** 0.00 7.44e07*** 0.00
Population density 0.0161** 0.01 0.00 0.00 0.0103*** 0.00 0.0112*** 0.00
Population (%) under 7.41 11.41 0.99 2.06 5.996*** 1.71 4.031*** 0.58
eighteen
Population (%) over 91.40*** 25.17 2.871* 1.51 8.216** 2.62 5.839*** 1.00
sixty-five
Population (%) w/ bachelor’s 14.27 19.32 2.22 2.27 3.495** 1.21 1.212*** 0.30
or grad. degree
Political controls
Citizen ideology 0.00 0.00 0.00270*** 0.00 0.00 0.00 0.00 0.00
State government ideology 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Constant 36.42* 18.80 5.88 4.13 2.09 1.98 1.988** 0.66
Observations 400 400 400 400
Number of groups 50 50 50 50
Within R2 0.1417 0.2107 0.4106 0.5939
*p < .1.
**p < .05.
***p < .01.

section, specifically the summary table for each As financial condition is in some respects a
year from 2005 to 2013. function of the economy, the models include
variables for the unemployment rate, state
Control variables. In addition to the variables of gross domestic product, median household
interest, the present study controls for a vari- income, a Gini coefficient to capture the distri-
ety of time-variant factors including political bution of income, and the log of total expendi-
ideology, economic conditions, and the tures. 2 These variables come from the state
demographics of the population. The empiri- CAFRs, the Federal Reserve, and the American
cal approach, detailed in the next section, Community Survey.
captures the effects of any time-invariant In addition to economic factors, there is a
characteristics of the states; correspondingly, robust body of literature demonstrating that the
the controls capture factors that may affect demographic characteristics of a jurisdiction
financial condition that are subject to change can have an effect on financial outcomes. Com-
over time. munities with larger black populations, for
Singla and Stone 125

instance, tend to be have lower household Greer and Denison 2016). A Pesaran test sug-
income, higher unemployment, and lower gests that there is cross-sectional dependence
homeownership rates (Shapiro, Meschede, and in the present study’s data as well, necessitating
Osoro 2013). Jurisdictions with a larger elderly a correction. Rather than traditional panel-
population may have different service prefer- corrected standard errors, these models use
ences and consequently may choose to disin- Driscoll–Kraay standard errors, which are
vest in future human capital (Gradstein and robust to cross-sectional dependence, autocor-
Kaganovich 2004). And states that are more relation, and heteroskedascity. In addition,
densely populated may need to engage with the decentralization measures are lagged to
capital markets more frequently to meet addi- account for potential endogeneity between
tional infrastructure needs (Glover and Simon financial condition and decentralization.3
1975). Moreover, there is evidence that demo-
graphic changes can affect both financial con-
dition and fiscal decentralization (Stone et al. Results
2015; Stephens 1974; Akai and Sakata 2002).
To prevent these factors from confounding the Fiscal Decentralization in the States
analysis, the models include variables captur- A descriptive analysis of the dependent vari-
ing the total size of the population, as well as ables reveals that there is more between varia-
the total black population, population density, tion than within variation, indicating that
child population percentage, population per- financial condition within a state tends to be
centage over the age of sixty-five, and the edu- relative stable over time. This is more the case
cation level of the state population. These with the longer-term financial condition
controls come from the U.S. Census Bureau variables (e.g., net asset ratio and long-term
and the American Community Survey. liability ratio) than it is with the shorter-term
Finally, to address concerns that financial financial condition variables. There is a similar
condition may be affected by the political lean- pattern with the decentralization variables, with
ings of a state government or its citizens, the more between variation than within variation.
models include a measure of citizen ideology In general, the data show that nearly all states
within a state and a measure of the ideological in all years are more inclined to decentralize
makeup of state government. The former comes expenditures than revenues. More detail,
from Berry et al. (1998) and the latter from a including a figure showing average expenditure
similar group’s follow-up work (Berry et al. decentralization against average revenue
2010). These variables have been updated decentralization for each state over time, is
annually by Fording (2017) and therefore cap- available in Online Supplemental Table S2 and
ture changes in citizen and government ideol- Supplemental Figure S1.
ogy over time.
Estimating the Effect of Fiscal
Method Decentralization on Financial Condition
To determine the effect of fiscal decentraliza- Table 1 shows the results of the linear empirical
tion on state financial condition, the present models. When interpreting these results, recall
research estimates eight OLS models—a linear that the literature suggested a clear link
and nonlinear model for each measure of finan- between fiscal decentralization and financial
cial condition—with state and year fixed condition but was unclear on the nature and
effects. The state fixed effects control for the direction of that relationship. Statistically
impact of any time-invariant state characteris- significant results for the decentralization
tics. Other research using panel data for states variables in these models imply a linear rela-
has often found cross-sectional dependence tionship. A positive b coefficient on both the
(De Hoyos and Sarafidis 2006; Hoechle 2009; revenue and expenditure decentralization
126 State and Local Government Review 50(2)

that revenue decentralization is unlikely to

have a meaningful effect on medium or long-
term state financial condition. For expenditure
decentralization, the linear results are similar:
the current ratio has a positive and statistically
significant b coefficient, but the other models
do not. In the current ratio model, the effect size
is moderate, as a 1 standard deviation increase
in expenditure decentralization is associated
with a .39 standard deviation increase in the
current ratio.

A Curvilinear Relationship?
Models that include both the revenue and
expenditure decentralization variables as well
as squared terms (e.g., revenue decentraliza-
tion  revenue decentralization) further
explore the nature of the relationship between
fiscal decentralization and state financial con-
dition. If both the unaltered variable and its
squared term are statistically significant and
have b coefficients with opposing signs, there is
evidence of a curvilinear relationship. Table 2
Figure 1. Relationship between expenditure
decentralization and long-term financial condition. shows truncated results of these models,
though the full results are available in the
Online Supplemental Table S3.
variables would suggest that fiscal decentrali-
Before turning to the results, it is important
zation is generally positive for state financial
to note that the R2 for each of the nonlinear
condition. Alternatively, a negative b coeffi- models is slightly stronger than it is for the cor-
cient on the revenue decentralization and a pos- responding linear model, indicating that the
itive sign on the expenditure decentralization inclusion of the square terms provides some
variable would imply that states with more additional explanatory power. F-tests for the
access to resources and/or reduced service obli- joint significance of the square terms, however,
gations are more financially healthy. suggest that they are only meaningful additions
As the table shows, the current ratio model to the models for the net asset ratio and
shows statistically significant results and a neg- long-term liability ratio. This implies that the
ative b coefficient for the revenue decentraliza- nonlinear models are stronger for long-term
tion variable, indicating that as revenues are financial condition, while the linear models are
more decentralized, state short-term financial perhaps better for short- and medium-term
condition worsens. The effect size is small, but financial condition.
substantive, as a 1 standard deviation increase In the nonlinear models, there is no evidence
in revenue decentralization corresponds to a that revenue decentralization affects state
.1 standard deviation decrease in the current financial condition, as none of the variables are
ratio. For the other three models—operating statistically significant. With respect to expen-
ratio, net asset ratio, and long-term liability diture decentralization, however, the results
ratio—there is no evidence of a relationship show a clear curvilinear relationship between
between revenue decentralization and financial long-term financial condition and expenditure
condition. Based on these results, one can see decentralization. In both the net asset ratio and
Singla and Stone 127

Table 2. Nonlinear Models, Fiscal Decentralization on Long-term Financial Condition.

Net Asset Ratio Long-term Liability Ratio

Variables b SE b SE

Independent variables
(Lag) revenue decentralization 0.63 0.46 0.37 0.21
(Lag) revenue decentralization2 0.63 0.46 0.18 0.21
(Lag) expenditure decentralization 3.200*** 0.52 2.986*** 0.61
(Lag) expenditure decentralization2 2.577*** 0.67 2.892*** 0.51
Observations 400 400
Number of groups 50 50
Within R2 0.4146 0.5994
*p < .1.
**p < .05.
***p < .01.

long-term liability ratio models, both the unal- variable. Moreover, the results from these mod-
tered variable and the square term are statisti- els are broadly consistent with models where
cally significant at the p < .01 level, with outliers with a Cook’s distance of 4/n or .01
coefficients that are in the opposite direction. have been removed. 4 In these robustness
This suggests a curvilinear relationship between checks, the models for both the net asset ratio
expenditure decentralization and state long-term and long-term liability ratio have the same sign
financial condition. and statistical significance for the expenditure
The precise nature of the curvilinear rela- decentralization variables.
tionship is depicted in Figure 1. The figure
plots the marginal effects of a change in expen-
diture decentralization against each measure of Conclusion
long-term financial condition based on the The results have important policy implica-
regression output. The y-axis for the long- tions. First, revenue decentralization has a
term liability ratio is reversed to make interpre- very limited effect on state financial condi-
tation of each chart the same: moving up on the tion, meaning that providing local govern-
y-axis shows an increase in financial condition, ments with access to revenue streams may
while moving to the right shows an increase in not materially harm state governments. How-
expenditure decentralization. As the figure ever, states are making proactive choices; it
shows, the relationship is an inverted “U” is unlikely that they would choose a level of
shape, where increasing decentralization ini- revenue decentralization that would materially
tially improves financial condition up to an harm their ability to meet obligations.
inflection point, after which greater decentrali- On the other hand, expenditure decentraliza-
zation harms long-term financial condition. tion is both more common and larger in magni-
The figure also includes a scatter plot of the tude than revenue decentralization. It is also
observed data. Although visual interpretation associated with changes in state financial con-
of the figure may suggest that the results are dition, meaning that the link between decentra-
being influenced by outliers at the tails of the lization and financial health is not simply a
distribution of expenditure decentralization, the function of access to resources or service obli-
marginal effect of a change in expenditure gations. This is consistent with the theoretical
decentralization on financial condition is statis- justifications of decentralization, which argue
tically significant across nearly all observed that public goods provision ought to be pro-
levels of expenditure decentralization for each vided by the lowest level of government
128 State and Local Government Review 50(2)

possible because doing so increases the effi- financial condition by either centralizing or
ciency of that provision. The curvilinear results decentralizing, but those in the middle are
provide context: some amount of decentraliza- unlikely to see meaningful effects. And based
tion may act as a rising tide, improving the on the revenue decentralization finding, states
financial condition of states even if that decen- could make these choices without altering the
tralization reduces access to resources as a revenues currently available to their local gov-
result. But too much decentralization may harm ernments. Or, resources could be decentralized
state financial condition. Given the potential to local governments without causing harm to
economies of scale associated with service pro- the financial condition of state governments.
vision, this indicates an “optimal” level of Such a development would be contrary to the
expenditure decentralization. The curvilinear prevailing trends of the last seventy years.
results allow for a point estimate of this level
by calculating the local maximum of each func- Declaration of Conflicting Interests
tion graphed in Figure 1. The implied long-term The authors declared no potential conflicts of inter-
financial condition maximization point is an est with respect to the research, authorship, and/or
expenditure decentralization ratio of .625 for publication of this article.
the net asset ratio and .525 for the long-term lia-
bility ratio.5 These values are approximately Funding
one standard deviation apart. The authors received no financial support for the
To put this threshold in context, recall that research, authorship, and/or publication of this
the mean value of expenditure decentralization article.
is .49, with a low of .20 (Hawaii in 2006) and a
high of .67 (Nevada in 2008). In total, thirty- ORCID iD
one states are below the threshold for the Akheil Singla http://orcid.org/0000-0002-3334-
long-term liability ratio, with fifteen of those 0728
more than one standard deviation below the
optimal and five of those more than two stan- Supplemental Material
dard deviations away. The further below the Supplemental material for this article is available
optimization point a state is, the more it stands online.
to gain by some expenditure centralization.
Moving from about 2 standard deviations Notes
below the mean to about one standard deviation 1. Another construction of the net asset ratio is:
below the mean (from about .31, or Delaware in (restricted net assets þ unrestricted net assets) /
2013 to .4, or Arkansas in 2012), for instance, total liabilities. That specification has a .7 corre-
yields a predicted improvement in long-term lation with the traditional one and an empirical
liability ratio of .23 standard deviations. The model substituting the alternative net asset ratio
jump from one standard deviation below the specification yielded results that are substantively
mean to the mean (from .4, or Arkansas in similar to those presented in the article.
2012, to .5, or Pennsylvania in 2005), however, 2. There may be an endogeneity problem with the
yields a negligible change in long-term liability total expenditures variable because it is also the
ratio. On the other side of the optimization denominator of the operating ratio. However,
threshold, the effects are similar: going from omitting the total expenditures control variable
the mean to 1 standard deviation above the from the operating ratio models does not mean-
mean yields a negligible change in financial ingfully alter the results, indicating that any endo-
condition, but going from the 1 standard devia- geneity is not driving the substantive conclusions
tion to 2 yields a .2 standard deviation decrease of the article.
in long-term financial condition. 3. Although the time-lag approach in the models
Thus, states toward the tails of the distribu- addresses the possibility that decentralization is
tion may be able to materially improve their a response to financial condition change, it is
Singla and Stone 129

possible that the models are not correctly captur- local governments. Public Budgeting & Finance
ing the direction of the effect (i.e., the relation- 35:66–88.
ship may be endogenous). Thus, the results Daigneau, Elizabeth. 2011, April 2011. Fend-for-
cannot be taken as truly causal in nature. yourself localism. Governing. http://www.govern
4. In most cases, the dropped observations were ing.com/topics/mgmt/fend-for-yourself-local
state-years in which a particular state had an ism.html (accessed October 8, 2017).
abnormally high or low value on the financial De Hoyos, Rafael E., and Vasilis Sarafidis. 2006.
condition indicator serving as the dependent Testing for cross-sectional dependence in panel-
variable. data models. Stata Journal 6 (4): 482.
5. This estimate must be taken with caution, as Fisher, Ronald C. 2003. The changing state-local fis-
states may have reasons to prefer more centrali- cal environment: A 25 year retrospective. In State
zation at the cost of impaired financial condition. and local finances under pressure, ed. David L.
A state with a lower cost of capital than its local Sjoquist, 9–29. Northampton, MA: Edward
governments, for instance, may prefer to centra- Elgar.
lize borrowing and increase intergovernmental Fording, Richard C. 2017. State ideology data.
transfers despite harming its own financial https://rcfording.wordpress.com/state-ideology-
condition. data/ (accessed July 30, 2017).
Gamkhar, Shama, and J. Mitchell Pickerill. 2012.
The state of American federalism 2011–2012:
References A fend for yourself and activist form of bottom-
Akai, Nobuo, and Masayo Sakata. 2002. Fiscal up federalism. Publius: The Journal of Federal-
decentralization contributes to economic growth: ism 42:357–86.
Evidence from state-level cross-section data for Glover, Donald R., and Julian L. Simon. 1975. The
the United States. Journal of Urban Economics effect of population density on infrastructure:
52:93–108. The case of road building. Economic Develop-
Baicker, Katherine, and Nora Gordon. 2006. The ment and Cultural Change 23:453–68.
effect of state education finance reform on total Goelzhauser, Greg, and Shanna Rose. 2017. The
local resources. Journal of Public Economics state of American federalism 2016–2017: Policy
90:1519–35. reversals and partisan perspectives on intergo-
Berman, David R. 2003. Local governments and the vernmental relations. Publius: The Journal of
states. Armonk, NY: M. E. Sharpe. Federalism 47:285–313.
Berry, William D., Richard C. Fording, Evan J. Goodspeed, Timothy J. 2002. Bailouts in a federa-
Ringquist, Russell L. Hanson, and Carl E. Klar- tion. International Tax and Public Finance 9:
ner. 2010. Measuring citizen and government 409–21.
ideology in the US states: A re-appraisal. State Gorina, Evgenia, Craig Maher, and Marc Joffe.
Politics & Policy Quarterly 10:117–35. 2018. Local fiscal distress: Measurement and pre-
Berry, William D., Evan J. Ringquist, Richard C. diction. Public Budgeting & Finance 38:72–94.
Fording, and Russell L. Hanson. 1998. Measuring Government Accounting Standards Board. 1999.
citizen and government ideology in the American Statement no. 34 of the Government Accounting
states, 1960-93. American Journal of Political Standards Board, basic financial statements—
Science 42:327–48. And management’s discussion and analysis. Nor-
Bowman, Ann O’M., and Richard C. Kearney. 2011. walk: CT.
Second-order devolution: Data and doubt. Pub- Gradstein, Mark, and Michael Kaganovich. 2004.
lius: The Journal of Federalism 41:563–85. Aging population and education finance. Journal
Brunori, David. 2007. Local tax policy: A federalist of Public Economics 88:2469–85.
perspective, 2nd ed. Washington, DC: Urban Greer, Robert A., and Dwight V. Denison. 2016.
Institute Press. Determinants of debt concentration at the state
Clark, Benjamin Y. 2015. Evaluating the validity level. Public Budgeting & Finance 36 (4):
and reliability of the financial condition index for 111–30.
130 State and Local Government Review 50(2)

Hoechle, Daniel. 2007. Robust standard errors for Olson, Mancur. 1969. The principle of fiscal equiva-
panel regressions with cross-sectional depen- lence: The division of responsibilities among dif-
dence. Stata Journal 7 (3): 281. ferent levels of government. The American
Jimenez, Benedict Salazar. 2009. Fiscal stress and Economic Review:479–87.
the allocation of expenditure responsibilities Ostrom, Elinor. 1976. Size and performance in a fed-
between state and local governments: An eral system. Publius 6:33–73.
exploratory study. State and Local Government Ostrom, Vincent, Robert Bish, and Elinor Ostrom.
Review 41:81–94. 1988. Local government in the United States. San
Johnson, Craig L., Sharon N. Kioko, and W. Bartley Francisco, CA: ICS Press.
Hildreth. 2012. Government-wide financial state- Ostrom, Vincent, Charles M. Tiebout, and Robert
ments and credit risk. Public Budgeting & Warren. 1961. The organization of government
Finance 32:80–104. in metropolitan areas: A theoretical inquiry.
Kim, Byungkyu, and Richard C. Fording. 2010. American Political Science Review 55:831–42.
Second-order devolution and the implementation Qian, Yingyi, and Gerard Roland. 1998. Federalism
of TANF in the US States. State Politics & Policy and the soft budget constraint. American Eco-
Quarterly 10:341–67. nomic Review:1143–62.
Kloha, Philip, Carol S. Weissert, and Robert Kleine. Reschovsky, Andrew. 2004. The impact of state gov-
2005. Developing and testing a composite model ernment fiscal crises on local governments and
to predict local fiscal distress. Public Administra- schools. State and Local Government Review
tion Review 65:313–23. 36:86–102.
Kornai, Janos, Eric Maskin, and Géard Roland. Rivenbark, William C., Dale J. Roenigk, and Gre-
2003. Understanding the soft budget constraint. gory S. Allison. 2010. Conceptualizing financial
Journal of Economic Literature 41:1095–136. condition in local government. Journal of Public
McLure, Charles E. 1967. The interstate exporting of Budgeting, Accounting & Financial Management
state and local taxes: Estimates for 1962. 22:149–77.
National Tax Journal 20:49–77. Ross, Justin, Wenli Yan, and Craig Johnson. 2015.
Musgrave, Richard. A. 1959. The theory of public The public financing of America’s largest cities:
finance. New York, NY: McGraw-Hill. A study of city financial records in the wake of
Musgrave, Richard A., ed. 1965. Essays in fiscal fed- the Great Recession. Journal of Regional Science
eralism. Washington, DC: Brookings Institution. 55:113–38.
Nohria, Nitin, and Ranjay Gulati. 1996. Is slack good Saxton, Gregory D., Christopher W. Hoene, and Ste-
or bad for innovation? Academy of Management ven P. Erie. 2002. Fiscal constraints and the loss
Journal 39:1245–64. of home rule: The long-term impacts of Califor-
Nohria, Nitin, and Ranjay Gulati. 1997. What is the nia’s post-proposition 13 fiscal regime. The
optimum amount of organizational slack? A American Review of Public Administration 32
study of the relationship between slack and inno- (4): 423–54.
vation in multinational firms. European Manage- Shapiro, Thomas, Tatjana Meschede, and Sam
ment Journal 15:603–11. Osoro. 2013, February. The roots of the widening
Oates, Wallace E. 1972. Fiscal federalism. New racial wealth gap: Explaining the black-white
York, NY: Harcourt Brace Jovanovich. economic divide. Research and Policy Brief.
Oates, Wallace E. 1977. The political economy of fis- Waltham, MA: Institute on Assets & Social
cal federalism. Lexington, MA: Lexington Books. Policy.
Oates, Wallace E., ed. 1991. Studies in fiscal feder- Singla, Akheil, James Comeaux, and Charlotte L.
alism. Aldershot, UK: Edward Elgar. Kirschner. 2014. Blind, broke, and bedlam: Dif-
Oates, Wallace E. 1999. An essay on fiscal federal- ferentiating fiscal stress from bankruptcy in Cali-
ism. Journal of Economic Literature 37:1120–49. fornia. Public Finance and Management 14:306.
Oates, Wallace E. 2008. On the evolution of fiscal Singla, Akheil, Justin Stritch, and Mary K. Feeney.
federalism: Theory and institutions. National Tax forthcoming. Constrained or creative? Changes
Journal 61:313–34. in financial condition and entrepreneurial
Singla and Stone 131

orientation in public organizations. Public Wang, Xiaohu, Lynda Dennis, and Yuan Sen Tu.
Administration. 2007. Measuring financial condition: A study of
Stone, Samuel B. 2015. The effect of fiscal decentra- US states. Public Budgeting & Finance 2:1–21.
lization on the financial condition of municipal Weingast, Barry R. 2009. Second generation fiscal
Government. International Journal of Public federalism: The implications of fiscal incentives.
Administration 38:453–60. Journal of Urban Economics 65:279–93.
Stone, Samuel B., Akheil Singla, James Comeaux, Zimmerman, Joseph F. 1995. State-local relations,
and Charlotte Kirschner. 2015. A comparison of 2nd ed. Westport, CT: Praeger.
financial indicators: The case of Detroit. Public
Budgeting & Finance 35:90–111.
Stephens, G. Ross. 1974. State centralization and the Author Biographies
erosion of local autonomy. The Journal of Poli- Akheil Singla is an assistant professor at the School
tics 36:44–76. of Public Affairs at Arizona State University. His
Tan, Justin. 2003. Curvilinear relationship between research focuses on public financial management at
organizational slack and firm performance: Evi- the state and local level, with a specific focus on
dence from Chinese state enterprises. European municipal debt markets as well as local government
Management Journal 21:740–49. financial health.
Wallis, John Joseph, and Wallace E. Oates. 1988.
Samuel B. Stone is an associate professor in the
Decentralization in the public sector: An empiri- Division of Politics, Administration & Justice at
cal study of state and local government. In Fiscal California State University, Fullerton. His research
federalism: Quantitative studies, ed. Harvey S. focuses on public financial management at the state
Rosen , 5–32. Chicago, IL: University of Chicago and local level, with a specific focus on fiscal feder-
Press. alism and municipal home rule.