Economic Modelling 59 (2016) 520–528

Contents lists available at ScienceDirect

Economic Modelling

journal homepage: www.elsevier.com/locate/ecmod

Tax reform, fiscal decentralization, and regional economic growth: New

evidence from China☆
Zhou Yang
Department of Economics and Legal Studies, Robert Morris University, 6001 University Blvd., Moon Township, PA 15108, USA

a r t i c l e i n f o a b s t r a c t

Article history: This paper is the first to examine heterogeneous responses to changes in fiscal decentralization policy associated
Received 2 November 2015 with the 1994 tax reform in China and the nonlinearity of these responses. Using panel data for 29 Chinese prov-
Received in revised form 24 July 2016 inces over the period 1990–2012, the paper shows that the effects of revenue and expenditure decentralization
Accepted 25 July 2016
on economic growth vary across the three main sectors, with the largest impact on the secondary sector. The
Available online xxxx
results also suggest there is an inverted U-shaped relationship between the degree of revenue (expenditure)
JEL classification:
decentralization and the growth of the secondary sector. In addition, slow-growing and fast-growing provinces
H72 respond differently to changes in both revenue and expenditure decentralization with greater responses from
H77 fast-growing provinces. The responses from fast-growing provinces appear to be nonlinear, indicating that
O23 fast-growing provinces benefit from appropriate revenue (expenditure) decentralization, but the effect turns
negative as the degree of decentralization becomes excessively high. The study provides a way to reconcile the
Keywords: contradicting evidence found in the literature. In addition, the finding on heterogeneous responses to fiscal
Tax reform decentralization has important policy implications.
Fiscal decentralization
© 2016 Elsevier B.V. All rights reserved.
Economic growth in China

1. Introduction
Over the past several decades, the devolution of fiscal powers to sub-
national governments has taken place in many countries. As countries
continue to decentralize fiscal powers to lower levels of government,
it becomes increasingly important to understand various impacts of fis-
cal decentralization policy. During the transition to a market economy,
China has undergone several major reforms. As an important fiscal
reform, the 1994 tax reform replaced the previous fiscal contracting sys-
tem with a tax sharing system. This tax reform has generated significant
variation in fiscal decentralization policy, which offers an excellent
opportunity to investigate the influences of fiscal decentralization poli-
cy. China has made great efforts to decentralize its fiscal system since
the tax reform. Despite these significant changes, the effects of fiscal de-
centralization policy in China are still debatable.
Previous theoretical studies have shown that fiscal decentralization
helps improve the efficiency of the public sector and promote economic
growth because local governments are better positioned than the cen-
tral government to provide public services that match local preferences
and needs. The theorem depends on key assumptions that the govern-
ment aims to maximize social welfare and public goods are provided
uniformly. Interestingly, empirical studies on the effect of fiscal
☆ The author is indebted to three anonymous referees and the editor for very helpful
comments on an earlier draft. The author is, of course, responsible for remaining errors.
E-mail address: yangz@rmu.edu.
decentralization offer conflicting evidence. Some studies find that fiscal
decentralization has been conducive to China's economic development.
Qi (1992) and Qian (1999) concluded that the fiscal contracting system
(1980–1993) provided incentives for subnational governments to
promote local economic growth. In contrast, other research shows
that fiscal decentralization has been detrimental to China's economic
growth. Jin and Zou (2005) demonstrated that expenditure decentrali-
zation is negatively associated with economic growth and the finding
may imply that local governments are not responsive to local prefer-
ences and needs in practice. The lack of consensus in existing studies
calls for further investigation into its various potential influences.
This paper offers a new perspective on fiscal decentralization by
exploring the nonlinearity of its impact as well as heterogeneous re-
sponses across sectors and economic strata. The study contributes to
the literature on fiscal decentralization in two important ways. First, the
examination of a nonlinear effect provides us a chance to reconcile the
contradicting evidence found in the existing studies. The intuition behind
the nonlinear impact is that a higher degree of fiscal decentralization
promotes economic growth when the degree of fiscal decentralization
is relatively low; however, as the degree of fiscal decentralization
becomes excessively high, a further increase in the degree of fiscal decen-
tralization may impede economic growth. The rational for the negative
impact comes from the possibility that inefficient resource allocation
arises as the size of local government becomes unnecessarily large.
Second, none of the existing studies explore differential effects across sec-
tors and economic strata in China. This study makes the first attempt to

http://dx.doi.org/10.1016/j.econmod.2016.07.020
0264-9993/© 2016 Elsevier B.V. All rights reserved.
 Z. Yang / Economic Modelling 59 (2016) 520–528
investigate the differential impacts of fiscal decentralization across three
main sectors. In addition, it also delves into heterogeneous responses to
fiscal decentralization policy between fast-growing and slow-growing
provinces. The findings on these differential responses enable us to better
understand the impacts of fiscal decentralization policy and have impor-
tant policy implications that help inform future decision making.
Using province-level panel data for 29 Chinese provinces over the
period 1990–2012, the paper finds that the effect of fiscal decentraliza-
tion (measured by revenue decentralization or expenditure decentrali-
zation) on provincial economic growth varies across the three main
industry sectors in China, with the largest impact on the secondary sec-
tor. Interestingly, the impact on the secondary sector is positive when
the degree of revenue (expenditure) decentralization is relatively low,
but it turns negative as decentralization becomes overly aggressive. In
addition, slow-growing and fast-growing provinces respond differently
to changes in both revenue and expenditure decentralization with
greater responses from fast-growing provinces. Furthermore, the
impacts on fast-growing provinces appear to be nonlinear, implying
that fast-growing provinces benefit from initial revenue (expenditure)
decentralization, but the benefit diminishes with the degree of revenue
(expenditure) decentralization and the effect turns negative when the
degree of decentralization becomes excessively high.
The remainder of the paper is organized as follows. Section 2 offers a
brief literature review summarizing both theoretical and empirical
studies on fiscal decentralization. Section 3 highlights important institu-
tional background on the 1994 tax reform. Section 4 introduces the em-
pirical specifications, followed by a brief data description in Section 5.
Section 6 presents the results, and Section 7 concludes with policy
discussions.
2. Literature review
There have been rich discussions on fiscal decentralization from
both theoretical and empirical perspectives. Earlier theoretical work
has shown that decentralization, whether in the form of fiscal federal-
ism or in the form of reduced government intervention, can help im-
prove economic performance. Existing empirical studies on the
influence of fiscal decentralization do not always support this argument.
This section provides a brief overview of existing theoretical and empir-
ical research on fiscal decentralization, and discusses how this paper
helps reconcile the contradicting evidence.
2.1. Theoretical literature
Oates (1972) developed a theoretical framework for the analysis of
decentralization policy and considered the trade-off between central-
ized and decentralized provision of public goods. The study shows
that the decentralized provision improves social welfare given
heterogeneous preferences across regions and no spillover effects. The
theorem depends on key assumptions that the objective of the govern-
ment is to maximize social welfare and public goods are provided uni-
formly. In the absence of externalities, local governments perform
better than the central government when delivering public services
and matching local preferences and needs (Tiebout, 1956). In contrast,
Brennan and Buchanan (1980) examined fiscal federalism from a public
choice perspective and proposed the Leviathan hypothesis stating
“Total government intrusion into the economy should be smaller,
ceteris paribus, the greater the extent to which taxes and expenditures
are decentralized”. They argue that Leviathan governments operate like
monopolists to increase their control over local resources instead of
maximizing social welfare, and thus a decentralized fiscal system may
not produce better outcomes.
More recent theoretical work takes the political process and asym-
metric information between political agents into account. Some studies
consider the trade-off between centralized and decentralized provision
in a principal-agent model of electoral accountability. The presence of
521
asymmetric information results in inefficient government performance.
In this framework, decentralization encourages yardstick and tax
competition among local governments; therefore, it can improve
accountability and lessen information asymmetry (Besley and Case,
1995; Bordignon et al., 2004). In addition, increased electoral account-
ability ultimately translates into more efficient government perfor-
mance. However, Tanzi (1996) argued the assumption that local
governments are more likely to respond to local preferences and pro-
vide local public goods more efficiently than the central government
does not necessarily apply in a non-democratic political system. Conse-
quently, fiscal decentralization may lead to misallocation of resources.
Other studies reconsider decentralization theory in a political economy.
Lockwood (2005) developed a model where the decision is made
through majority voting over alternative levels of public goods provi-
sion. The benefit of decentralization comes from rising efficiency in
the decision-making process. Neyapti and Bulut-Cevik (2014) investi-
gated the welfare effects of institutional mechanisms and concluded
that welfare and the effectiveness of redistributive policy rise with the
degree of revenue centralization. To sum up, existing theory suggests
that the effect of fiscal decentralization depends on assumptions about
government's objectives and institutions.
2.2. Empirical literature
Empirical studies that examine the impact of fiscal decentralization
on economic growth in China also show conflicting results.1 A number
of studies argue that fiscal decentralization promotes China's economic
development. Qian and Weingast (1997) showed that the fiscal
contracting system protected local governments from revenue
predation by the central government, and thus helped retain necessary
resources for economic growth. Lin and Liu (2000) claimed that fiscal
decentralization contributes to China's rapid economic growth and
this effect comes from increased efficiency rather than rising invest-
ment. Shah (2004) noted that the merit of fiscal decentralization de-
pends on both revenue and expenditure assignments as well as the
specific stages of economic development. Jin et al. (2005) argued that
fiscal decentralization policy has substantially strengthened the fiscal
incentives of provincial governments which are associated with faster
provincial economic development and more reforms. He and Sun
(2014) examined the influence of fiscal decentralization on foreign
direct investment in China. Using provincial panel data over the period
1995–2002, they show that fiscal decentralization encourages the
inflow of foreign direct investment.
In contrast, other research suggests that fiscal decentralization im-
pedes China's economic growth. Davoodi and Zou (1998) reported a
negative relationship between fiscal decentralization and economic
growth in developing countries. Young (2000) demonstrated that fiscal
decentralization has led to market fragmentation and hindered
economic growth. Jin and Zou (2005) showed that expenditure
decentralization was negatively associated with economic growth
under the previous fiscal contracting system. The result may indicate
that local governments are not responsive to local preferences and
needs in practice. This is especially true if local officials are not selected
by local voters or “voting by feet” is not practical.
To sum up, the lack of consensus in existing empirical studies can
arise from differences in empirical specifications as well as different
sample periods. To my knowledge, these existing studies focus on esti-
mating an average effect on the economy. None of the previous studies
have investigated differential effects across sectors and economic strata.
It is important to explore heterogeneity in the responses to fiscal decen-
tralization policy. As Zhang (2006) stated, the impact of fiscal decentral-
ization is likely to vary across regions given differences in economic
structures and fiscal burdens. Furthermore, fiscal policy has often been
1
See Shen et al. (2012) for a more detailed literature review on fiscal decentralization in
China.
522 Z. Yang / Economic Modelling 59 (2016) 520–528
implemented to promote the development of certain industries during
China's transition to a market economy; therefore, it is expected that
fiscal decentralization policy would have varying impacts on different
sectors of the economy. For instance, industries in the secondary sector
are likely to be more sensitive to fiscal decentralization policy. This
paper is the first to explore heterogeneous responses to fiscal decentral-
ization policy in China.
In addition, previous studies on fiscal decentralization policy in
China mostly assume that the effect is linear, but the effect may be non-
linear. This study argues that the effect of fiscal decentralization on eco-
nomic growth also depends on the existing degree of fiscal
decentralization. In particular, if the degree of fiscal decentralization is
relatively low, a higher degree of fiscal decentralization promotes eco-
nomic growth. As subnational governments gain more fiscal power,
the size of governments continues to rise. Once it exceeds the optimal
size, the allocation of resources becomes inefficient and thus hinders
economic growth. This suggests that the first-order effect is positive
but the second-order effect may be negative. Note that the existence
of a nonlinear effect provides a possible way to reconcile the contradic-
ting evidence found in the existing studies: if most provinces have a
high degree of revenue or expenditure decentralization during the sam-
ple period examined in a given study, the estimated average effect tends
to be negative; otherwise, the estimated average effect tends to be
positive.
3. Background on the 1994 tax reform
The 1994 tax reform generated significant changes in China's fiscal
system (Shen et al., 2012). Consequently, the degrees of revenue decen-
tralization and expenditure decentralization both vary significantly due
to the tax reform. This variation provides a great chance to examine the
impact of fiscal decentralization. This section summarizes key changes
that are relevant to revenue and expenditure decentralization.
China's fiscal decentralization is characterized by the devolution of
fiscal power through the fiscal contracting system (1980–1993) and
the recentralization of revenues coupled with expenditure decentraliza-
tion under the tax sharing system (1994–present). Before 1994, China
had a fiscal contracting system that dampened the fiscal power of the
central government. Local governments served as agents of the central
government for tax collection purposes. Nevertheless, tax collections
were subject to frequent tax exemptions without authorization from
the central government. In addition, local firms and governments col-
luded to hide profits and shift deficits to the central government. This
contracting fiscal system enabled local governments to generate excess
revenues and let the central government cover deficits (Ahmad et al.,
2002). Local abuses of taxing power led to intense interjurisdictional
tax competition.
In 1994, China switched to the tax assignment (tax sharing) system
that enabled the central government to regain its fiscal power by cap-
turing major tax revenues and establishing the central tax administra-
tion. It lowered the corporate income tax rate substantially (from 55%
to 33%), and imposed a value-added tax (VAT) on most goods and
services. The National Tax Bureau was established by the central
government, collecting VAT, excise taxes, and income taxes from enter-
prises controlled by the central government. Local governments mainly
collected the following taxes: business taxes (turnover taxes on services
and sales of intangible assets and real estate properties); income taxes
from individuals and local firms; agricultural taxes; and taxes on land
use. The new tax system assigns taxes that are harder to collect to
local governments which may have more local information than the
central government.2
Since the 1994 tax reform, the central government has significantly
centralized revenue sources, such as the VAT, resource tax, personal
2
For example, urban maintenance and development taxes and taxes on the use of ara-
ble land are collected by local governments.
and corporate income taxes. It also adjusted the revenue-sharing for-
mula between the central and subnational governments. The clearer as-
signment of certain taxes helped remove some perverse incentives
created under the old fiscal contracting system. For example, excise
taxes were assigned to the central government after the reform, which
eliminated the incentive for local governments to overly expand
businesses with higher tax returns, such as distilleries and tobacco
companies (Zhang and Martinez-Vazquez, 2003).
In contrast to more centralized revenues, expenditures have become
more decentralized since the tax reform. The central government is re-
sponsible for nationwide services whereas subnational governments
are responsible for providing most local public goods and services. The
exclusive subnational responsibilities include urban maintenance and
construction, environmental protection, water supply, and community
services. In addition, subnational governments provide public services,
such as education, health care, social welfare, public safety, and other
local and urban services.
One important implication from the above discussion is that the tax
reform has greatly influenced the degree of fiscal decentralization from
both revenue and expenditure perspectives. This paper utilizes such im-
portant policy variation associated with the tax reform to explore differ-
ential impacts of fiscal decentralization. That being said, this paper does
not attempt to directly examine the impacts of the tax reform per se.
Another important implication is that the reassignment of tax collection
and expenditure responsibilities to subnational governments potential-
ly has varying influences on the development of different industries. In
particular, one would expect industries that are directly affected (e.g.
construction, education, and health services) by the reassignment to
be more responsive to changes in both revenue and expenditure decen-
tralization than other industries.
4. Empirical specifications
This paper examines the effect of fiscal decentralization policy asso-
ciated with the tax reform on provincial economic growth in China. The
study focuses on differential effects of fiscal decentralization policy and
allows for nonlinear impacts. First, it decomposes the economy into
three main sectors and investigates the differential effects across
sectors. Second, the paper further explores heterogeneous responses
to fiscal decentralization policy across different economic strata in
China using panel quantile regressions. For the first part of the analysis,
the following panel data model with fixed effects is employed:
Growthit ¼ α o þ β1 FDit−1 þ β2 FDit−1  FDit−1 þ φ Taxit−1
þ δ0 X it−1 þ μ i þ Y t þ εit; ð1Þ
where Growthit denotes the growth rate of real GDP in province i over
the course of year t. The study investigates real GDP growth in three
main sectors discussed below. FDit−1 denotes the degree of fiscal decen-
tralization in the previous year, and two measures are considered given
data availability. The model also includes the square of the decentraliza-
tion variable to allow for a nonlinear impact. As previously discussed,
the impact of fiscal decentralization may depend on the existing degree
of fiscal decentralization. Taxit − 1 is the lagged average tax burden as a
share of provincial GDP. Xit−1 consists of a vector of lagged provincial
control variables affecting economic growth in a given province. μ i
and Yt denote province and year fixed effects, respectively. Note that
province fixed effects account for unobserved heterogeneity across
provinces, and year fixed effects control for time-varying common mac-
roeconomic shocks as well as policy changes at the national level.3 εit is
the disturbance term. Specific definitions of all variables are discussed
below.
3
Year fixed effects help account for other changes associated with the 1994 tax reform.
It is possible that there are still some unobserved province-specific time-varying factors
that could influence economic growth.
 Z. Yang / Economic Modelling 59 (2016) 520–528
The dependent variable measures provincial economic growth in a
given year. For the purpose of this analysis, the aggregate economy is
decomposed into three main industry sectors in China: primary,
secondary, and tertiary industries.4 The growth of economic activity
(measured by real GDP) in each main sector is estimated as a function
of variables of interest and control variables.
The main variables of interest include the fiscal decentralization var-
iable and the square of this variable. In the literature, the degree of fiscal
decentralization can be measured in different ways. This paper con-
siders two common measures used in the previous studies: provincial
budgetary revenue per capita as a share of total budgetary revenue
per capita in the nation; and provincial budgetary expenditure per
capita as a share of total budgetary expenditure per capita in the
nation.5 As Jin and Zou (2005) mentioned, revenues collected at the
provincial level as a share of total revenue in the nation is a proxy for
the degree of fiscal decentralization, which captures the revenue collec-
tion process. Note that this paper uses the population-adjusted decen-
tralization measures suggested by Zhang and Zou (1998) to avoid
overestimating the degree of decentralization for large provinces. In
particular, the population-adjusted decentralization measures scale
revenues (expenditures) by population, that is, the degree of fiscal de-
centralization is measured by provincial share of revenue (expenditure)
per capita.
Before 1994, subnational governments acted as agents of the central
government for tax collection purposes. The 1994 tax reform split the
tax collection responsibility between the central tax administration
authority and subnational tax bureaus. One potential concern with the
revenue decentralization measure is that the relative size of local reve-
nue collection may not be a good indicator of fiscal decentralization in
China because most taxes were levied by the central government even
though they were mainly collected by local governments in early
years. If locally collected revenues are not spent locally, this measure
does not necessarily reflect local tax autonomy. To address this concern,
the second measure, provincial budgetary spending per capita as a share
of total budgetary spending per capita in the nation, is used as the alter-
native fiscal decentralization measure.
As discussed earlier, the empirical model allows the impact of fiscal
decentralization to be nonlinear, that is, the effect depends on the
existing degree of fiscal decentralization: appropriate fiscal decentrali-
zation can encourage economic growth, but pursuing decentralization
in an aggressive manner may harm economic growth. Correspondingly,
one would expect the coefficient on the fiscal decentralization variable
to be positive (β1 N 0) and the coefficient on the square term to be
negative (β2 b 0). If β1 + 2 β2FD b 0, the impact of fiscal decentralization
is negative. If β1 + 2 β2FD N 0, the impact of fiscal decentralization is
positive.
Another concern with the fiscal decentralization variable is the
potential endogeneity problem. One may argue that there could be
simultaneity between fiscal decentralization policy and economic
growth because more fiscal power may be granted to provinces with
greater economic growth. To avoid this potential problem, the fiscal de-
centralization variable is lagged one year. The Hausman test for
endogeneity is performed to confirm that the lagged decentralization
variable is not endogenous.6 In addition, it makes sense to lag the policy
4
Primary industry sector mainly refers to agriculture (including farming, forestry, ani-
mal husbandry and fishery). Secondary industry sector includes mining and quarrying,
manufacturing, production and supply of electricity, water and gas, and construction. Ter-
tiary sector consists of all other industries not included in primary or secondary industry
with services as the main part.
5
For a given province, degree of revenue decentralization
budgetary revenue in province i=population in province i
total budgetary revenue=total population
, where total budgetary revenue is the sum of
provincial and central budgetary revenues. Similarly, one can measure the degree of ex-
penditure decentralization for a given province.
6
Higher-order lags of the fiscal decentralization variable are used as instruments to per-
form the Hausman test.
523
variable since it takes time for the policy to have an impact on the
economy.
To account for other factors that influence economic growth, a set of
control variables are included. All control variables are lagged one year
to avoid potential endogeneity problem.7 Previous studies have showed
that the level of taxation affects economic activity; therefore, a tax var-
iable measured by provincial total tax revenue as a share of provincial
GDP is employed. This tax variable captures the overall tax burden in a
given province.8 In China, major tax rates are set by the central govern-
ment; however, tax efforts and exemptions vary across provinces. The
variation in the overall tax burden as a share of provincial GDP suggests
differences in the effective tax rate across provinces.
Additional factors influencing economic growth are also incorporat-
ed in the empirical model. Based on previous studies in the literature,
this paper considers the following provincial characteristics: structure
of labor force, total amount of investment on fixed assets, technology
advancement, growth of international trade, and size of the economy.
Four variables are employed to capture provincial labor markets:
urban employment in state-owned enterprises (SOEs) as a share of
total urban employment; growth rate of graduates with higher educa-
tion; growth rate of total urban employment; and urban unemployment
rate. The first variable accounts for the relative size of employed labor
force in state-owned businesses/entities whereas the second and third
control for the growth rates of potential skilled labor force and the
actual employed labor force. Urban unemployment rate accounts for
provincial business cycles. To take into consideration the influence of
technology, the growth rate of total patents granted in a given province
is included. In addition, total population is employed to control for the
size of provincial economy. Both total population and total amount of
investment on fixed assets enter the equation in log form. The model
also includes province and year fixed effects to control for unobserved
heterogeneity and common shocks.
For the second part of the analysis, the study further investigates dif-
ferential responses to fiscal decentralization policy across economic
strata. Panel quantile regressions with fixed effects are used to explore
heterogeneous responses to policy changes across quantiles of the
conditional distribution of provincial economic growth. Suppose that
the conditional quantile functions of the response of the tth observation
on the ith province, yit, take the following form:
Q yit ðτjxit−1 Þ ¼ α i þ x0 it−1 βðτÞ: ð2Þ
In the equation above, α implies a pure location shift effect on the
conditional quantiles of the response and is independent of quantiles.
The effects of the covariates, x, depend on the specific quantile, τ. To
estimate the model for several quantiles simultaneously, the study
employs the method proposed by Koenker (2004). In particular, the
method obtains the penalized quantile regression estimators
introduced by Koenker (2004).9 In general, this method overcomes
the difficulties associated with the infinite dimension of α by obtaining
an estimate of the model purged of the fixed effects and focusing on the
slope β(τ).
5. Data
Panel data for 29 provinces over the period 1990–2012 are obtained
from China Statistical Yearbooks and other data sources published by
7
There is still some chance that lagging the variables does not entirely eliminate the
endogeneity concern.
8
Changes in the tax policy at the national level are captured by year fixed effects.
= q n
T
9
The penalized estimators solve the following: Min ∑ ∑ ∑t¼1 wk ρτk ðyit −∝i −x0it−1 β
α;β k¼1 i¼1
n
^ towards a common
ðτ k ÞÞ þ λ ∑ j∝i j; where λ is a penalization parameter that shrinks ∝
i¼1
value. Example R code for the procedure is available from Koenker's webpage: http://
www.econ.uiuc.edu/~roger/research/panel/long.html.
524 Z. Yang / Economic Modelling 59 (2016) 520–528

National Bureau of Statistics of China.10 Tibet is not included in the anal- Table 1
ysis due to the significant amount of missing data.11 Table 1 reports the Summary statistics.

summary statistics of the data sample. There have been noticeable Variable Obs Mean SD
differences in the growth of economic activity among the three main in- Primary sector GDP growth rate 598 0.012 0.095
dustry sectors in China. The primary sector experienced a much slower Secondary sector GDP growth rate 598 0.070 0.080
growth compared to the secondary and tertiary sectors. This is not Tertiary sector GDP growth rate 598 0.070 0.058
surprising given that China has moved from agriculture-based economy Revenue decentralization 598 0.627 0.640
Expenditure decentralization 598 0.921 0.587
to manufacturing-focused economy. Another recent trend (not directly
Tax burden as share of GDP (tax burden/GDP) 598 0.067 0.028
shown in the table) is the rapid growth of service industries. Tertiary Share of urban employees in state-owned-enterprises 598 0.519 0.168
sector (primarily consists of service industries) becomes the sector Urban employment growth rate 598 0.017 0.052
with the highest growth rate near the end of my sample period. Growth rate of higher education graduates 598 0.127 0.154
As previously discussed, the degree of revenue (expenditure) decen- Patent growth rate 598 0.200 0.311
Foreign trade growth rate 598 0.199 0.217
tralization has varied significantly since the 1994 tax reform. Fig. 1 in Urban unemployment rate 598 0.034 0.010
Appendix A depicts the significant changes in the degrees of revenue Ln population (104 people) 598 8.113 0.805
and expenditure decentralization. A sudden drop in the degree of reve- Ln total investment (108 RMB) 598 7.077 1.387
nue decentralization occurred around 1994. It reflects changes in the
fiscal system: the previous fiscal contracting system has been replaced
by the tax assignment system since 1994. The tax assignment system impacts on the tertiary sector and the primary sector are smaller. Note
has an important feature, that is, the central government keeps a large that the coefficient on the square of the lagged revenue decentralization
fraction of major profitable taxes, collecting 74.5% of value-added tax variable is significantly negative in Model 2, suggesting a nonlinear
revenue, 61.5% of corporate income tax and 60% of personal income effect on the secondary sector. This implies that the growth of the
tax revenues.12 In addition, total tax revenue collected by the central secondary sector rises with the degree of revenue decentralization,
government has accounted for more than half of national tax revenue but it increases at a decreasing rate as revenue becomes more
since the tax reform. Further, the variation in revenue decentralization decentralized. When the degree of revenue decentralization exceeds
across provinces has declined since the reform. 3.93, the effect becomes negative, indicating that overly aggressive rev-
In contrast, expenditures have become more decentralized since the enue decentralization hurts the growth of the secondary sector.13 The
1994 tax reform, partially reflecting increased transfers from the central results from the specifications using the expenditure decentralization
government to provinces which support rising expenditure responsibil- measure (Models 4–6) show a similar pattern that expenditure decen-
ities. As noted in previous studies, revenues were generally more tralization has a nonlinear impact on the growth of the secondary sec-
decentralized than expenditures before the tax reform, which suggests tor, a positive first-order but negative second-order effect.14
there were greater provincial remittances to the central government be- Nevertheless, the impact of expenditure decentralization is insignificant
fore the tax reform (Jin and Zou, 2005). All things considered, the 1994 for the primary and tertiary sectors. To sum up, the results from the
tax reform has greatly altered the degree of fiscal decentralization, cre- models using both decentralization measures suggest that fiscal decen-
ating an opportunity to delve into differential responses to changes in tralization contributes the most to the growth of the secondary sector
fiscal decentralization policy as well as its nonlinear impact. among all three industry sectors over the sample period. This is consis-
tent with the previous discussion on the evolution of fiscal decentraliza-
tion policy, that is, industries that are directly affected (e.g. construction
6. Results
and electricity industries) under the reformed tax assignment system
are more responsive to changes in fiscal decentralization policy than
The empirical analysis focuses on differential effects of fiscal decen-
other industries. Further, the impact on the secondary sector is nonlin-
tralization policy in China. First, the study examines varying effects of
ear, indicating there exists an inverted U-shaped relationship between
fiscal decentralization on provincial economic growth across three
the growth of the secondary sector and revenue (expenditure) decen-
main sectors (primary, secondary, and tertiary sectors) of the Chinese
tralization. This finding provides a possible way to reconcile the contra-
economy. Second, it explores heterogeneous impacts across quantiles
dicting evidence discussed before. Prior to 1994, the degree of revenue
of the conditional distribution of provincial economic growth in China
decentralization was relatively high in a number of provinces, so the es-
using panel quantile regressions with fixed effects.
timated average effect during the period prior to the tax reform is likely
to be negative. After the 1994 tax reform, revenues became more
6.1. Differential effects across sectors centralized (a lower degree of revenue decentralization); therefore,
the estimated average impact is likely to be positive.
Table 2 presents the estimated effects across three sectors of a pro- As expected, increased tax burden slows down economic develop-
vincial economy. Models 1–3 employ the revenue decentralization ment in all sectors. Further, higher tax burden hurts the secondary
measure whereas Models 4–6 utilize the expenditure decentralization and tertiary sectors more than the primary sector. Interestingly, prov-
measure. The estimated effects on the economic growth of the primary inces with faster-growing urban employment experience a slower
sector using both decentralization measures are reported in Columns 2 growth for the primary sector but more rapid growth for the tertiary
and 5 (Models 1 and 4). Similarly, Columns 3 and 6 (Models 2 and 5) sector. This result may reflect the structural change in the labor market
demonstrate the estimated impacts on the secondary sector whereas as more workers have been diverged from the primary sector into the
Columns 4 and 7 (Models 3 and 6) display the estimated influences tertiary sector where earnings are generally higher. Rapidly rising num-
on the tertiary sector. Comparing the results across the three main sec- ber of graduates with higher education do not have a significant impact
tors of the economy (Models 1–3), revenue decentralization has the on any of the sectors. This result may partially reflect that students grad-
largest and positive first-order impact on the secondary sector. The uated from universities and institutions in a given province may not

10
Chongqing became a municipality seceded from Sichuan province in 1997. To keep
13
the data series consistent, Chongqing is included in Sichuan province over the sample pe- Based on the sample data, this indicates that revenue decentralization was negatively
riod in the study. associated with the growth of secondary sector in Shanghai in early 1990s prior to the tax
11
The results do not change much when available observations in Tibet are included in reform.
14
the regressions. The impact on the secondary sector becomes negative when expenditure decentrali-
12
Calculations are based on the 2012 data. zation measure is above 3.23.
 Z. Yang / Economic Modelling 59 (2016) 520–528 525

Table 2
Effects by sector of the Chinese economy.

Revenue decentralization Expenditure decentralization

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

Variable (lagged one year) Primary Secondary Tertiary Primary Secondary Tertiary

L. fiscal decentralization (FD) 0.0747⁎ 0.0974⁎ 0.0782⁎⁎⁎ 0.0073 0.1617⁎⁎⁎ 0.0290

(0.0436) (0.0526) (0.0266) (0.0408) (0.0502) (0.0398)
L. FD × L. FD −0.0083 −0.0124⁎ −0.0024 0.0077 −0.0251⁎⁎ 0.0005
(0.0061) (0.0068) (0.0034) (0.0071) (0.0096) (0.0072)
L. tax burden/GDP −0.5070⁎ −1.0270⁎⁎ −0.7782⁎⁎⁎ −0.2080 −0.9735⁎⁎⁎ −0.4388⁎⁎
(0.2678) (0.3792) (0.1996) (0.2667) (0.3359) (0.1928)
L. share of urban employees in SOEs 0.0012 0.1266 0.0847 0.0135 0.1338 0.1069⁎
(0.0773) (0.0810) (0.0566) (0.0767) (0.0795) (0.0608)
L. urban employment growth rate −0.0797⁎⁎⁎ −0.0050 0.0475⁎ −0.0788⁎⁎⁎ 0.0003 0.0540⁎⁎
(0.0181) (0.0287) (0.0273) (0.0172) (0.0244) (0.0226)
L. higher education graduate growth rate 0.0140 −0.0049 0.0123 0.0161 −0.0065 0.0126
(0.0372) (0.0197) (0.0169) (0.0370) (0.0203) (0.0170)
L. patent growth rate −0.0338 0.0209⁎ 0.0118 −0.0342⁎ 0.0179 0.0112
(0.0199) (0.0107) (0.0078) (0.0196) (0.0106) (0.0080)
L. foreign trade growth rate −0.0025 0.0297⁎ 0.0163 −0.0070 0.0220 0.0100
(0.0131) (0.0172) (0.0123) (0.0127) (0.0152) (0.0118)
L. Ln total investment −0.0303⁎ −0.0386 −0.0508⁎⁎ −0.0258⁎⁎ −0.0431 −0.0383⁎
(0.0150) (0.0289) (0.0220) (0.0121) (0.0265) (0.0225)
L. Ln population −0.0111 −0.0914 −0.0543 −0.0155 −0.0356 −0.0836
(0.0536) (0.0742) (0.0546) (0.0550) (0.0732) (0.0581)
L. urban unemployment rate −0.0065 −0.0069 −0.0080⁎⁎ −0.0062 −0.0039 −0.0087⁎⁎
(0.0052) (0.0060) (0.0037) (0.0048) (0.0059) (0.0036)
Constant 0.4688 1.1975 0.9984⁎ 0.4161 0.6034 1.0865⁎
(0.4784) (0.7240) (0.5109) (0.5094) (0.7229) (0.5535)
Province fixed effects Yes Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes Yes
Observations 598 598 598 598 598 598
Adjusted R-squared 0.5635 0.5460 0.4864 0.5644 0.5555 0.4686

Notes: Entries are coefficient estimates with robust standard errors in parentheses.
⁎ Denotes statistical significance at the 10% level.
⁎⁎ Denotes statistical significance at the 5% level.
⁎⁎⁎ Denotes statistical significance at the 1% level.

stay in the province after graduation; therefore, the province is unable
to obtain the benefits associated with a fast-growing pool of educated
job candidates. Growing number of patents granted promote the
growth of the secondary sector but slow down the growth of the prima-
ry industry. This may again reveal structural changes over the past de-
cades as China moves away from agriculture-based economy to
manufacturing-focused economy.
Unexpectedly, the level of investment (in the previous year) does
not have a significant impact on the growth of the secondary sector,
but it impedes the growth of both primary and tertiary sectors. One pos-
sible explanation is that investment in infrastructures driven by promo-
tion concerns of government officials resulted in rampant redundant
investment, therefore, inefficient investment in infrastructures and
buildings in the previous year led to misallocation of resources, hurting
economic growth. Higher urban unemployment rate hinders the
growth of the tertiary sector, but has no significant impacts on the
growth of the secondary and primary sectors.
In general, the results for the secondary sector based on the model
that employs the expenditure decentralization measure are largely con-
sistent with those from the model that uses the revenue decentraliza-
tion measure. The findings suggest that fiscal decentralization policy
differentially influences the three main sectors in China. The nonlinear
impact on the secondary sector suggests that there is an inverted
U-shaped relationship between fiscal decentralization and the growth
of the secondary sector. Nevertheless, this nonlinear effect does not
apply to the primary and tertiary sectors. The findings have important
policy implications. For instance, the use of expenditure decentraliza-
tion policy as a way to encourage the growth of the primary and tertiary
sectors is unlikely to be successful, but further decentralization of reve-
nues can promote the growth of these two sectors. On the other hand,
both revenue and expenditure decentralization can promote the growth
of the secondary sector, although the effect turns negative with hasty
and excessive decentralization. In addition, responses to changes in
other social and economic conditions also vary across sectors. These dif-
ferential effects across sectors may somehow suggest that resources
have been diverted from the primary sector to the other two sectors
due to structural changes in the Chinese economy.
6.2. Differential effects across economic strata
To further investigate heterogeneous responses to changes in fiscal
decentralization policy across economic strata, the paper utilizes panel
quantile regressions with fixed effects. The analysis addresses the ques-
tion of whether fast-growing provinces respond differently to policy
changes than slow-growing provinces. Table 3 demonstrates the results
from the specifications that explore the effect of revenue decentraliza-
tion at various quantiles (0.1, 0.25, 0.5, 0.75 and 0.9) of the conditional
distribution of provincial GDP growth. Similarly, Table 4 reports the re-
sults from the ones that examine heterogeneous responses to expendi-
ture decentralization across quantiles.
As shown in Table 3, revenue decentralization has a positive and
significant first-order impact on economic growth in provinces at the
middle and upper quantiles. Further, the impact of revenue decentrali-
zation is nonlinear at the highest quantile (τ = 0.9). The results suggest
that provinces with rapid economic growth respond differently than
those growing at lower rates. Higher tax burden hurts economic growth
across all five quantiles with the largest impact at the middle quantile.
Interestingly, a higher share of urban employees in state-owned-
enterprises promotes provincial economic growth at the middle and
upper quantiles. The rapid growth of patents granted speed up
economic growth across all five quantiles with the largest impact at
the highest quantile (90%), indicating the importance of technology
and innovation to economic growth. A higher growth of international
trade contributes to provincial economic growth at the 25% and 90%
526 Z. Yang / Economic Modelling 59 (2016) 520–528

Table 3
Panel quantile regression results (revenue decentralization).

Revenue decentralization

Variable (lagged one year) τ = 0.1 τ = 0.25 τ = 0.5 τ = 0.75 τ = 0.9

L. fiscal decentralization (FD) 0.0063 0.0285 0.0431⁎⁎ 0.0351⁎ 0.0590⁎⁎

(0.0220) (0.0204) (0.0190) (0.0205) (0.0244)
L. FD × L. FD 0.0032 −0.0016 −0.0065 −0.0054 −0.0085⁎
(0.0055) (0.0043) (0.0042) (0.0052) (0.0051)
L. tax burden/GDP −0.5113⁎⁎ −0.6044⁎⁎⁎ −0.6273⁎⁎⁎ −0.6039⁎⁎⁎ −0.4893⁎
(0.2082) (0.2304) (0.1919) (0.1758) (0.2569)
L. share of urban employees in SOEs 0.0049 0.0255 0.0799⁎⁎ 0.0748⁎ 0.0530
(0.0413) (0.0374) (0.0319) (0.0411) (0.0449)
L. urban employment growth rate −0.0331 −0.0151 −0.0010 −0.0486 −0.1354
(0.0438) (0.0312) (0.0288) (0.0541) (0.0826)
L. higher education graduate growth rate 0.0086 0.0142 0.0257 0.0387⁎⁎ 0.0356⁎
(0.0225) (0.0246) (0.0228) (0.0178) (0.0199)
L. patent growth rate 0.0405⁎⁎⁎ 0.0419⁎⁎⁎ 0.0380⁎⁎⁎ 0.0302⁎⁎⁎ 0.0429⁎⁎⁎
(0.0085) (0.0089) (0.0077) (0.0089) (0.0160)
L. foreign trade growth rate 0.0213 0.0424⁎⁎ 0.0152 0.0198 0.0271⁎
(0.0179) (0.0171) (0.0116) (0.0126) (0.0163)
L. Ln total investment −0.0074 −0.0070 −0.0018 −0.0033 −0.0171⁎⁎
(0.0047) (0.0052) (0.0045) (0.0061) (0.0071)
L. Ln population 0.0064 0.0031 −0.0025 −0.0046 0.0080
(0.0052) (0.0057) (0.0042) (0.0053) (0.0061)
L. urban unemployment rate −0.0075⁎⁎⁎ −0.0051⁎ −0.0033 −0.0084⁎⁎ −0.0031
(0.0026) (0.0028) (0.0025) (0.0033) (0.0056)
Constant 0.0736 0.0987⁎ 0.1111⁎⁎⁎ 0.1952⁎⁎⁎ 0.1868⁎⁎⁎
(0.0769) (0.0586) (0.0429) (0.0531) (0.0660)
Province fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
Observations 598 598 598 598 598
Pseudo R-squared 0.4706 0.4801 0.5163 0.5387 0.5625

Notes: Entries are coefficient estimates with standard errors in parentheses.

⁎ Denotes statistical significance at the 10% level.
⁎⁎ Denotes statistical significance at the 5% level.
⁎⁎⁎ Denotes statistical significance at the 1% level.

Table 4
Panel quantile regression results (expenditure decentralization).

Expenditure decentralization

Variable (lagged one year) τ = 0.1 τ = 0.25 τ = 0.5 τ = 0.75 τ = 0.9

L. fiscal decentralization (FD) 0.0413⁎ 0.0196 0.0550⁎⁎⁎ 0.0737⁎⁎⁎ 0.1090⁎⁎⁎

(0.0246) (0.0263) (0.0211) (0.0257) (0.0348)
L. FD × L. FD −0.0045 0.0007 −0.0081⁎ −0.0135⁎⁎ −0.0187⁎⁎
(0.0059) (0.0056) (0.0048) (0.0058) (0.0081)
L. tax burden/GDP −0.5532⁎⁎⁎ −0.5521⁎⁎⁎ −0.5542⁎⁎⁎ −0.5744⁎⁎⁎ −0.6837⁎⁎
(0.1464) (0.1664) (0.1384) (0.1642) (0.2653)
L. share of urban employees in SOEs 0.0246 0.0115 0.0702⁎⁎ 0.0851⁎⁎ 0.0435
(0.0344) (0.0398) (0.0353) (0.0426) (0.0588)
L. urban employment growth rate −0.0326 −0.0140 0.0000 −0.0557 −0.1506⁎
(0.0508) (0.0360) (0.0294) (0.0522) (0.0873)
L. higher education graduate growth rate 0.0187 0.0054 0.0286 0.0358⁎⁎ 0.0314
(0.0194) (0.0204) (0.0205) (0.0157) (0.0214)
L. patent growth rate 0.0410⁎⁎⁎ 0.0400⁎⁎⁎ 0.0376⁎⁎⁎ 0.0298⁎⁎⁎ 0.0455⁎⁎⁎
(0.0093) (0.0079) (0.0074) (0.0085) (0.0165)
L. foreign trade growth rate 0.0289⁎ 0.0397⁎⁎⁎ 0.0087 0.0161 0.0354⁎⁎⁎
(0.0161) (0.0149) (0.0098) (0.0127) (0.0132)
L. Ln total investment −0.0080 −0.0095 −0.0056 −0.0060 −0.0235⁎⁎
(0.0049) (0.0065) (0.0058) (0.0081) (0.0099)
L. Ln population 0.0158⁎⁎ 0.0082 0.0086 0.0095 0.0250⁎⁎⁎
(0.0077) (0.0086) (0.0062) (0.0082) (0.0082)
L. urban unemployment rate −0.0079⁎⁎⁎ −0.0068⁎⁎ −0.0042 −0.0077⁎⁎ −0.0068
(0.0025) (0.0030) (0.0026) (0.0031) (0.0050)
Constant −0.0307 0.0860 0.0378 0.0577 0.0769
(0.0709) (0.0625) (0.0461) (0.0636) (0.0896)
Province fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
Observations 598 598 598 598 598
Pseudo R-squared 0.4684 0.4829 0.5174 0.5403 0.5574

Notes: Entries are coefficient estimates with standard errors in parentheses.

⁎ Denotes statistical significance at the 10% level.
⁎⁎ Denotes statistical significance at the 5% level.
⁎⁎⁎ Denotes statistical significance at the 1% level.
 Z. Yang / Economic Modelling 59 (2016) 520–528
quantiles. Raising investment tends to slow down provincial economic
growth at the highest quantile. This again may reflect the adverse effect
of rampant redundant investment. Provinces with a higher urban un-
employment rate experience a lower growth, and the impact is more
pronounced at the 10% and 75% quantiles.
Table 4 displays the estimated impacts of expenditure decentraliza-
tion across quantiles. The first-order effect of expenditure decentraliza-
tion is positive and significant across all quantiles except in one case
(τ = 0.25). The second-order effect of expenditure decentralization is
negative and significant at the middle and upper quantiles, suggesting
a nonlinear effect for provinces with faster economic growth. The re-
sults imply that expenditure decentralization has a positive impact on
economic growth in slow-growing provinces; as provincial economy
grows much faster, further decentralization of expenditures raises pro-
vincial economic activity but at a lower rate. For the fastest growing
provinces, the impact turns negative when the degree of expenditure
decentralization exceeds 2.91. Similar to the results above, higher tax
burden impedes economic growth across all five quantiles. Results on
all other control variables are consistent with those shown in Table 3.
To summarize, the findings suggest that the effect of fiscal decentral-
ization on economic growth varies at different points in the conditional
distribution of provincial economic growth in China. In particular, as ex-
penditures become more decentralized, the first-order impact for fast-
growing provinces is positive and larger than that for slow-growing
provinces; however, the second-order impact for fast-growing prov-
inces is significantly negative whereas there is no second-order impact
for slow-growing provinces. In addition, the results indicate that expen-
diture decentralization is more effective than revenue decentralization
in promoting economic growth in slow-growing provinces. Fast-
growing provinces are responsive to both revenue and expenditure de-
centralization. Furthermore, responses to changes in other economic
factors also vary across provinces growing at different rates.
7. Conclusion
Despite the growing literature that explores the effect of fiscal de-
centralization on economic growth in China, little is known about the
differential effects of fiscal decentralization policy across sectors and
economic strata. The 1994 tax reform has resulted in significant changes
in the degree of fiscal decentralization from both revenue and expendi-
ture perspectives, offering a great opportunity to examine these differ-
ential impacts.
This paper is the first to empirically investigate the differential ef-
fects of fiscal decentralization policy associated with the tax reform on
economic performance across main sectors in China. In addition, the
study explores heterogeneous responses to policy changes across prov-
inces growing at different rates. More importantly, the study identifies
the nonlinear effect of fiscal decentralization, providing a chance to rec-
oncile the contradicting evidence in the literature.
Using province-level panel data for 29 Chinese provinces over the
period 1990–2012, the paper finds that the effect of fiscal decentraliza-
tion on economic growth differs across the three main sectors, with the
largest impact on the secondary sector. Further, the effect on the sec-
ondary sector is nonlinear, suggesting there is an inverted U-shaped re-
lationship between the growth of the secondary sector and revenue
(expenditure) decentralization. The nonlinear impact found in this
paper implies that the effect of fiscal decentralization on economic
growth (the secondary sector) also depends on the existing degree of
fiscal decentralization. An appropriate degree of decentralization helps
promote economic growth; however, fiscal decentralization can hinder
economic growth when it becomes overly aggressive. The differential
impacts across sectors have important policy implications. Both revenue
and expenditure decentralization can help promote the growth of the
secondary sector, but the effect turns negative with hasty and excessive
decentralization. A higher degree of revenue decentralization encour-
ages the growth of primary and tertiary sectors; nevertheless,
527
expenditure decentralization has no significant impact on these two
sectors. In addition, marginal responses to fiscal decentralization policy
vary at different points in the conditional distribution of provincial eco-
nomic growth. Slowly growing provinces respond positively to changes
in the degree of expenditure decentralization whereas rapidly growing
provinces act differently with a positive first-order but negative second-
order response. This paper offers new empirical evidence on the effect
of fiscal decentralization policy in China. The study opens new avenues
for further research that helps us understand the impacts of decentrali-
zation policy in developing countries.
Appendix A
Degree of fiscal decentralization
1
.9
.8
.7
.6
1990 1995 2000 2005 2010
94
19
year
revenue decentralization (provincial average)
expenditure decentralization (provincial average)
Fig. 1. Revenue and expenditure decentralization over time. Notes: previous fiscal
contracting system was replaced by the tax assignment system in 1994. Fig. 1 indicates
that the degree of revenue (expenditure) decentralization has varied significantly since
the 1994 tax reform.
References
Ahmad, E., Li, K., Richardson, T., Singh, R., 2002. Recentralization in China? IMF Working
Paper WP/02/168. Washington, D.C, International Monetary Fund.
Besley, T., Case, A., 1995. Incumbent behavior: vote-seeking, tax-setting, and yardstick
competition. Am. Econ. Rev. 85 (1), 25–45.
Bordignon, M., Cerniglia, F., Revelli, F., 2004. Yardstick competition in intergovernmental
relationships: theory and empirical predictions. Econ. Lett. 83 (3), 325–333.
Brennan, G., Buchanan, J., 1980. The Power to Tax: Analytical Foundations of a Fiscal Con-
stitution. Cambridge University Press, Cambridge.
Davoodi, H., Zou, H.F., 1998. Fiscal decentralization and economic growth: a cross-country
study. J. Urban Econ. 43 (2), 244–257.
He, Q., Sun, M., 2014. Does fiscal decentralization promote the inflow of FDI in China?
Econ. Model. 43, 361–371.
Jin, H., Qian, Y., Weingast, B.R., 2005. Regional decentralization and fiscal incentives: fed-
eralism, Chinese style. J. Public Econ. 89 (9–10), 1719–1742.
Jin, J., Zou, H.F., 2005. Fiscal decentralization, revenue and expenditure assignments, and
growth in China. J. Asian Econ. 16 (6), 1047–1064.
Koenker, R., 2004. Quantile regression for longitudinal data. J. Multivar. Anal. 91 (1),
74–89.
Lin, J.Y., Liu, Z., 2000. Fiscal decentralization and economic growth in China. Econ. Dev.
Cult. Chang. 49 (1), 1–21.
Lockwood, B., 2005. Fiscal decentralization: a political economy perspective. Warwick
Economic Research Papers No. 721. University of Warwick.
Neyapti, B., Bulut-Cevik, Z.B., 2014. Fiscal efficiency, redistribution and welfare. Econ.
Model. 41, 375–382.
Oates, W.E., 1972. Fiscal Federalism. Harcourt Brace Jovanovich, New York.
Qi, J.C., 1992. Fiscal reform and the economic foundations of local state corporatism in
China. World Politics 45 (1), 99–126.
Qian, Y., 1999. The institutional foundations of China's market transition. Annual World
Bank Conference on Development Economics.
Qian, Y., Weingast, B.R., 1997. Federalism as a commitment to preserving market incen-
tives. J. Econ. Perspect. 11 (4), 83–92.
Shah, A., 2004. Fiscal decentralization in developing and transition economies: progress,
problems, and the promise. Policy Research Working Paper 3282. Washington, D.C,
The World Bank.
528 Z. Yang / Economic Modelling 59 (2016) 520–528

Shen, C., Jin, J., Zou, H.F., 2012. Fiscal decentralization in China: history, impact, challenges
and next steps. Ann. Econ. Financ. 13 (1), 1–51.
Tanzi, V., 1996. Fiscal federalism and decentralization: a review of some efficiency and
macroeconomic aspects. Annual World Bank Conference on Development
Economics.
Tiebout, C.M., 1956. A pure theory of local expenditures. J. Polit. Econ. 64 (5), 416–424.
Young, A., 2000. The razor's edge: distortions and incremental reform in the People's Re-
public of China. Q. J. Econ. 115 (4), 1091–1135.
Zhang, X., 2006. Fiscal decentralization and political centralization in China: implications
for growth and inequality. J. Comp. Econ. 34 (4), 713–726.
Zhang, Z., Martinez-Vazquez, J., 2003. The system of equalization transfers in China, Inter-
national Studies Program Working Paper 03-12. Georgia State University, Andrew
Young School of Policy Studies.
Zhang, T., Zou, H.F., 1998. Fiscal decentralization, public spending, and economic growth
in China. J. Public Econ. 67 (2), 221–240.