new political reform in china: elections, public goods and income … · 2014. 3. 3. · but they...
TRANSCRIPT
Political Reform in China:
Elections, Public Goods and Income Distribution∗
Monica Martinez-Bravo†, Gerard Padró i Miquel‡, Nancy Qian§and Yang Yao¶
March 3, 2014
Abstract
We examine the effects of introducing village elections on public good expenditures,income distribution and land use in rural China. We construct a large panel data setby combining village administrative records with contemporaneously collected village-and household-level economic data and we use these data to document the history ofpolitical reforms and economic policies for over two hundred villages. We exploit thestaggered timing of the introduction of village elections to find that elections significantlyincreased public goods expenditure and reduced the amount of village land that thevillage government leases to enterprises, which is an indicator of corruption. Consistentwith a reduction in cronyism, we also find a reduction in the relative income of richhouseholds. These results suggest that local officials are better controlled by electionsrather than by bureaucratic monitoring.
Keywords: Institutions, Local Governance, Elections, Democracy; JEL: H4, H7,O1, P16
∗We are grateful to the editors, Larry Katz and Elhanan Helpman, and four anonymous referees fordetailed suggestions; Daron Acemoglu, Abhijit Banerjee, Doug Miller, Scott Rozelle and Lily Tsai for theirinsights; workshop participants at University of California at Berkeley, University of British Columbia, Cal-ifornia Institute of Technology, University of Chicago Booth GSB, Stockholm University (IIES), ColumbiaUniversity, Cornell University, Harvard University China Politics Workshop, New York University, North-western University Kellogg SOM, Princeton University Development/Labor Seminar, University of SouthernCalifornia, University of Toronto, Warwick University and the World Bank for their helpful comments. Wethank Yunnan Guo, Ting Han, Samuel Marden, Emily Nix, Yiqing Xu, and Linyi Zhang for excellent researchassistance. We thank the Chinese Ministry of Agriculture and their team of surveyors and field workers, andin particular, Wu Zhigang for his crucial role in our field work and data collection. We acknowledge financialsupport from Brown University PSTC, Stanford GSB Center for Global Business and the Economy, HarvardAcademy Scholars Research Grant, Yale University EGC Faculty Grant, the National Science FoundationGrant 0922087 and the European Union’s Seventh Framework Programme (FP/2007-2013) / ERC StartingGrant Agreement no. 283837.
†CEMFI, [email protected]‡London School of Economics, NBER, CEPR, BREAD, [email protected]§Yale University, NBER, CEPR, BREAD, [email protected]¶Peking University CCER, [email protected]
1 Introduction
The control of large bureaucracies, as the extensive literature on bureaucratic corruption
shows, is a difficult task.1 Lack of information and appropriate oversight often results in the
misbehavior of local officials.2 In autocratic countries, the control of local officials is further
complicated by the weakness of established channels to receive feedback from citizens.3 To
address this agency problem, several autocratic governments have introduced local elections
in recent years.4
China is a prominent example. The agency problem between the state and local officials
at the village level can take several forms. For instance, village officials are responsible
for raising funds from villagers in order to provide local public goods such as schooling.
Difficulties in bureaucratic monitoring allow them to shirk from the substantial effort re-
quired by this activity. Lack of effective oversight also allows the local officials to use their
control over collectively owned means of production, such as land or village enterprises, to
favor themselves and their cronies. During the 1980s and 1990s, village-level elections were
introduced to mitigate these problems. Policymakers intended elections to resolve agency
problems by giving local officials incentives to implement policies that appeal to a majority
of their constituents in order to obtain re-election.
The goal of this paper is to provide rigorous empirical analysis of the effects of the
introduction of local elections on local public goods, land use and income, the results of
which can help shed light on the effectiveness of elections in changing incentives for local
officials.5 We motivate these outcomes later in the introduction.1An classic example in this literature is Wilson(1989).2For recent overviews of this literature see Banerjee et al.(2012) and Olken and Pande(2012).3Autocracies typically limit the rights to associate, freedom of expression and freedom of the press, which
in democratic countries are important for the transmission of information on local scandals and demands.For instance, Besley and Burgess (2002) show that a free press is important for government responsiveness.
4For example, local elections have occurred in Indonesia under Suharto (1968-1998), Brazil during themilitary dictatorship (1964-1985), and Mexico under the PRI (1929-2000). Recently, local elections werealso introduced in Vietnam in 1998, in Yemen in 2001, and in Saudi Arabia in 2005. For a literature reviewof the nascent political science research on elections in dictatorships see Gandhi and Lust-Okar (2009).
5The theoretical basis for these claims comes from the rich literature on political accountability. In broadterms, this literature examines how elections induce politicians to provide more common interest policies –more public goods, less rent-seeking– or to redistribute in favor of an electoral majority. Redistribution and
1
Our study faces two notable difficulties. The first is the lack of detailed data on political
change and economic policies in rural China. To address this we construct the Village
Democracy Survey (VDS), a panel of over too hundred nearly representative villages from
29 provinces for the years 1982-2005. The survey documents the history of economic policies
and political reforms during this period, and contains detailed economic data on public
goods expenditure and financing. This is the longest and broadest panel ever constructed
to describe Chinese villages and is the first data that systematically document the changes
in the fiscal and political structure of village governments. We supplement the VDS with
economic data from the the National Fixed-Point Survey (NFS), which is collected yearly
from the same villages as the VDS by the Ministry of Agriculture.
The second difficulty lies in establishing the causal effect of the introduction of elections,
which were staggered in timing across villages. The first concern is reverse causality. Do
elections change economic conditions or do economic conditions generate demand for elec-
tions at the village level? The second concern is joint determination. Are the introduction
of elections and economic change jointly determined by a third factor such as broader re-
forms of rural policies? To address these concerns, we take advantage of two features of
the Chinese context. First, the timing of the introduction of elections is believed to have
been unrelated to village characteristics. Second, electoral reforms were supposed to have
been isolated to the village-level and did not affect higher levels of rural government. In the
paper, we provide a large body of anecdotal and quantitative evidence in support of these
two points.
The main empirical analysis proceeds in three steps. First, we document that the timing
of elections across villages within provinces is uncorrelated with a large number of observable
characteristics at the village level. This is consistent with the anecdotal evidence that the
timing of reforms was imposed top-down with little regard for village-specific characteristics.
common interest goods are therefore two complementary approaches to the majoritarian incentives inducedby elections. See the discussion in Persson and Tabellini, (2000, chapter 1). The recent theoretical literatureon democratization is also based on these majoritarian incentives. These theories predict that democracies,relative to autocracies, provide more public goods (e.g., Bueno de Mesquita et al., 2003)) and engage inmore redistribution (e.g., Acemoglu and Robinson, 2000).
2
Second, we implement a difference-in-differences (DD) strategy to estimate the causal effects
of the introduction of elections: we compare outcomes before and after the first election in
each village, between villages that had already introduced elections and those that have not.
The baseline specification includes village fixed effects that control for all time-invariant
differences across villages, year fixed effects that control for all changes over time that are
similar across villages, as well as province-specific time trends to control for the economic and
cultural divergence across China during our period of study. As with any DD strategy, causal
interpretation relies on the assumption that in the absence of electoral reforms, the evolution
of outcomes would be “parallel” across villages regardless of when they implemented the first
election. Finally, we support our interpretation with a large number of robustness exercises.
For example, we show that there is no evidence of pre-trends and that our estimates are
robust to controlling for pre-election characteristics as well as the introduction of elections
at the province level, which is the main source of endogeneity.
The first main outcome of our analysis is village government expenditure on public goods.
In rural China, local public goods are mostly financed by village contributions, which local
officials need to exert substantial effort to collect. In the early post-reform (post 1978) era,
it was widely believed that officials shirked in this task, leading to severe under-provision
of local public goods. Our results show that the introduction of elections increased total
local government expenditure on public goods by 154,500 RMB (22,071 USD), which ap-
proximately 100% of the sample mean.6 This is driven exclusively by expenditures financed
by villagers. In contrast, public goods financed by the upper levels of government are un-
changed by elections, which supports our interpretation that the results are not confounded
by simultaneous changes at the upper levels of government.
The second outcome is land use. All village land is collectively owned and most of it
is allocated by the village leaders to households in long-term leases. The village leadership
can retain some land under its direct control and lease it to village enterprises. The profits
from this collectively held land are supposed to benefit all villagers equally, but the lack of6Throughout the paper, monetary values are measured in constant terms. We use the conversion rate of
1 RMB = 7 USD, which was the historical average for the period we study.
3
transparency in managing land leases and enterprises provide local bureaucrats opportunities
for rent seeking.7 For this reason, villagers typically prefer land to be fully allocated to
households. We find that amongst villages that ever leased any land to enterprises, the
introduction of elections reduced the amount of land leased to enterprises by approximately
58 mu (3.8 hectares), which is approximately 50% of the sample mean.8 Amongst the
policies that village leaders are known to use to benefit elites, land use is the only one that
we observe directly.
The third outcome we examine is household income. If elections caused a systematic
reduction in the pro-elite bias in policy, then elections should reduce the income of rich
households relative to poorer ones. We find that the introduction of elections has little
effect on the income of the poorest half of households. However, the income of the 90th
percentile household is reduced by 5,056 RMB (722 USD), which is approximately 20% of
the sample mean. Consequently, the ratio between the income of the median household
and the 90th percentile household increased by 1.7 percentage-points. This reduction in
inequality is consistent with the existing elites’ position deteriorating with the introduction
of elections.9
The main results are consistent with the hypothesis that the introduction of elections
mitigates the agency problem at the village level: elected village officials provide more public
goods. They also allocate land to households and away from enterprises, which presumably
favors a majority of the villagers. Consistent with the new policies being less favorable to
the elite, elections reduce income for households at the top of the income distribution.
To shed light on the underlying mechanisms, we present several supplementary findings.
First, we find that the introduction of elections has a strong effect on public goods in villages
that experienced no leader turnover. This suggests that the introduction of re-election7See the Background section.81 mu = 1/15th hectare. Chinese households have very little farmland. Thus, our study uses the
indigenous measure.9Note that there are no positive effects for any income decile. It might well be that changes in land use
and other policies are inefficient, possibly because village and private enterprises were actually productive.Alternatively, it may be that the additional land that households receive increased their welfare in non-measurable ways, such as home production for own consumption, or that there are long-run gains thatcannot be measured with our data.
4
incentives, which is the only change that took place in villages that retained incumbent
leaders, plays a role in the effect of elections. Next, we examine the policies used to achieve
the reduction in inequality that followed elections. This is especially interesting in the
rural Chinese context since the village government does not have the legal authority to
impose recurring taxes and transfers. One possibility would be that elected officials increased
fees and levies for richer households (for example, to finance the increase in public goods
provision). However, using a subsample of villages for which we have detailed household-
level data, we find no evidence of such progressivity. In contrast, we find that the changes
in total household income ratios are paralleled by changes in agricultural income and other
income sources easily affected by the village government through its control over village land
and enterprises. This suggests that the village government partly addressed its inability to
redistribute income through taxes by reallocating the productive assets and employment
opportunities that were under its control.
Our study adds to the growing body of evidence on the effects of elections on public goods
and inequality in rural China.10 We improve on existing work by systematically documenting
the history of electoral reforms and the political and economic structure of Chinese villages
in detail for a relatively long panel. The larger sample size and richness of our data allow
us to be more empirically rigorous than past studies and examine a broader set of outcomes
that can help shed light on the mechanisms underlying the effects of elections.11 This study
complements a companion paper, Martinez-Bravo et al. (2011), which provides a large body
of evidence that the introduction of elections is successful in making the village government
at least partially accountable to villagers, and more generally, adds to the nascent literature
on governance in autocracies and, in particular, in China.12
10Past studies have used either panel data of relatively few villages or a cross-section of many villages toprovide evidence on the effect of elections in rural China. For example, Shen and Yao (2008) examines theeffect of elections on inequality and infers redistribution from changes in inequality. Several studies haverelated elections to public goods (e.g., Zhang et al., 2004; Luo et al., 2007, 2010; Mu and Zhang, 2011).Also, Gan et al. (2006) examines the relationship between elections and villager health shocks.
11For example, earlier studies have not examined the sources of public goods financing, land allocation,the different sources of income or the contribution of re-election incentives.
12For example, studies such as Lorentzen (2010), Martinez-Bravo et al. (2011), and Persson and Zhu-ravskaya (2011).
5
This study is closely related to within-country studies that have focused on various as-
pects of elections.13 Since elections are an essential element of democracy, these results
can also speak to the broader literature on democracy and economic policy.14 The existing
empirical evidence relating democratic transition to public goods and redistribution, which
mostly comes from cross-country studies, is inconclusive.15 Relative to cross-country com-
parisons, Chinese villages are much more comparable with each other and the introduction
of elections was not the result of social turmoil and other confounding factors. Our results
add to cross-country studies by showing that even a marginal move towards democratic
elections in the highly restricted setting of an authoritarian central regime can substantially
change village policy in public good provision and asset allocation. Our focus on elections
complements recent studies that emphasize the importance of constraints on the executive
in determining the effect of democracy on economic outcomes (e.g. Acemoglu and Robinson,
2001; Besley and Persson, 2011).16
This paper is organized as follows. Section 2 discusses the background. Section 3 presents
the conceptual framework and empirical strategy. Section 4 briefly describes the data.
Section 5 presents the main results, including the robustness checks. Section 6 presents
evidence on the importance of re-election incentives and Section 7 presents some additional13For example, Besley and Case (1995), Ferraz and Finan (2011) and Tyrefors and Pettersson-Lidbom
(2012) provide evidence for the role of several election characteristics in the United States, Brazil and Sweden,but they do not identify the effects of elections per se. Foster and Rosenzweig (2005) examines the effectof the introduction of rural elections on public goods provision in India, but focus on party competitionmechanisms. There is also a related literature examining the differences between elected and appointedjudges. For a recent example see Lim (2012).
14For instance see Acemoglu and Robinson(2001, 2006) or Boix(2003), both of which relate democratizationor extensions of the franchise to either increased public goods provision or redistribution. Husted andKenny(1997) and Miller(2008) provide empirical evidence for increased welfare spending in the context offranchise extension in the US.
15In the cross-section, democracy has been found to be positively associated with government size (Tavaresand Wacziarg, 2001), higher wages (Rodrik, 1999), lower inequality (e.g., Li et al., 1998), higher humancapital (Tavares and Wacziarg, 2001), and better health indicators (e.g.,Besley and Kudamatsu, 2006).However, in a large study looking at several socioeconomic policy dimensions, Gil et al. (2004) find thatdemocracy is associated with no difference in the outcomes they examine.
16Democracy is typically viewed as having two constituting elements: the presence of elections to determinewho will be the executive authority and the presence of institutionalized constraints, such as an independentjudiciary or media, on what the executive authority can do. The cited studies mainly focus on democracyas providing institutional constraints. In the Chinese context, such constraints are absent both before andafter the introduction of electoral reforms. Therefore our paper isolates the effect of democracy defined asthe presence of elections.
6
results on local fees, sources of income, and household income evolution. Section 8 offers a
conclusion.
2 Background
2.1 Villages and Village Governance in Rural China
A majority of the rural population in China lives in villages, which comprise the lowest
level of government administration. Above the village government, there are the semi-
equivalent levels of county and township governments, the prefecture governments, the
province governments, and ultimately, the central government in Beijing. The main eco-
nomic activity in a village is agriculture and households farm plots of land that are assigned
to them in long-term leases. The average village comprises approximately 400 households.
Village governments were first organized by the communist government during the early
1950s, with two groups of leaders in each village. First, there is the village committee. It
typically consists of three to five members and is led by the village chairman, henceforth
VC. Second, there is the Chinese Communist Party branch in the village. It is similar in
size to the village committee and is led by the village Party Secretary, henceforth PS. Before
elections were introduced, all of these positions were filled by appointment by the county
government with input from the village Party branch. There was no clear distribution of
authority between the two bodies and anecdotal evidence suggests that the party branch
had more power.17
The village government is responsible for two important village-level policies that di-
rectly affect the well-being of its citizens. First, village governments are responsible for the
provision of local public goods such as irrigation and primary schools. They are supposed
to decide which public goods to provide and to raise funds from villagers to finance them.
Second, local officials are also responsible for the use and allocation of collectively owned17In an earlier version of the paper, Martinez-Bravo et al. (2012), we explore the relative powers of the
VC and PS by documenting the signatures of each leader on important village decisions. We documentthat there is a shift in power from the PS to the VS after elections are introduced such that the number ofvillages where most policies are decided unilaterally by the VC increases, and the number of villages wheremost policies are decided unilaterally by the PS decreases.
7
means of production. The most important asset is arable land, all of which is publicly
owned in China. Most land is allocated to households for farming in long-term contracts.18
A fraction of land, no more than fifteen percent according to national law, may be retained
under the direct control of the village government so that it can make small adjustments to
household allocations without implementing a large-scale reallocation for the entire village.
During the 1980s and 1990s the land retained by the village government was often leased to
highly profitable village enterprises.19
Village governments do not have legal authority to impose any regular or recurring
taxes. Instead, to fund the activities of the village government, including local officials’
salaries, they can use proceeds from collectively owned sources of income such as village
enterprises or leasing land to enterprises (that can be from inside or outside the village), or
raise revenues by imposing ad hoc fees and levies, which we will henceforth refer to as local
fees for simplicity.
The obscurity of village enterprise and village government accounts typically means that
upper-level bureaucrats encounter enormous difficulties monitoring the activities of local
officials. As a consequence of this informational advantage, local officials who shirked in
providing public goods or who engaged in rent-seeking were able to maintain their positions.
There is an abundance of examples of corrupt village officials who neglected public good
provision and systematically extracted personal rents from land and enterprises controlled by
the village governments (e.g. Brandt and Turner, 2007, Kennedy et al., 2004, Oi and Rozelle,
2000, Rozelle, 1994; and Rozelle and Li, 1998). In response, villagers often resist paying local
fees, which in turn, starves local governments of funds and limits its ability to provide public
goods (e.g., Oi and Rozelle, 2000; Rozelle, 1994; Whiting, 1996). This negative feedback loop
further complicated the monitoring problem of the upper-level bureaucrats since they could
not distinguish whether low levels of public goods provision were an outcome of corruption,18Rural households cannot sell their land rights in China, and during the period of our study were also
prohibited from renting out their land. In most cases rural households were also restricted from hiringlaborers because households that did not farm their own land would lose land rights. See Jacoby et al.(2002) for a related study about tenure security in rural China.
19Additional responsibilities are the maintenance of law and order, the collection of grain taxes on behalfof the central government and the implementation of centrally mandated policies.
8
lack of effort by the local officials, the refusal of villagers to provide the necessary funds, or
the lack of demand from villagers.
2.2 Electoral Reforms
Motivation The first local elections were introduced in the early 1980s as collectives
were being dismantled. The difficulties in controlling local officials were paramount in the
discussions leading to the introduction of elections, as shown by this quote from the official
debate.
“Who supervises rural cadres? Can we supervise them? No, not even if we had
48 hours a day....” – Peng Zhen, vice-chairman of the NPC Standing Committee,
said at the chairmanship meeting of the Standing Committee of the Sixth NPC,
April 6, 1987 (O’Brien and Li, 1999).
Election proponents argued that village elections could fix the agency problems that were
plaguing local administration and generating discontent towards the regime at large. More
specifically, elections were expected to reduce the need for the central government to monitor
local officials by shifting monitoring responsibilities onto villagers. The idea was that making
local officials accountable to villagers would impose checks on the VC’s behavior and would
also allow villagers to select the most competent candidates.20
The Reform The initial introduction of elections changed the positions of all village
committee members from being appointed by the party-led county-level government to being
elected by villagers. The main legal requirements were that: i) the number of candidates
must exceed the number of positions; ii) term lengths were to be three years; and iii)
the winner must obtain 50% of votes in the last round of voting.21 The village committee
member who obtained the highest number of votes in the last round automatically became
the VC. All adult villagers had the right to vote and could abstain from voting. The village
Party Branch was unaffected by the reforms and remained appointed by the upper-levels20See Kelliher (1997) and O’Brien (1994, 1999) for descriptions of the policy debates that led to the official
introduction of local elections.21Elections with multiple candidates could thus undergo many rounds of voting.
9
of government. There was no change either to the size of the village committee or party
branch (e.g., the number of positions).
The law did not clarify the power relationship between the village committee and the
Party Branch, which remained ambiguous.22 Anecdotal evidence suggests that the power
arrangements between these two bodies were very heterogeneous across villages. Indeed, in
many areas the Party maintained control over villages by allowing the local Party branch to
nominate the candidates. For this reason, we refer to village leaders, which comprise both
bodies, as the subject of village decision-making. Rather than wholesale democratization,
this reform is thus better understood as a marginal change that made the local government
more accountable to villagers. Ultimately, the main change of the reform was to give villagers
the power to vote unsatisfactory VCs out of office.
In these elections there are no political parties and no slates of candidates with common
platforms. Candidates are drawn from the village and are thus typically well-known by
the villagers. As a consequence, candidates typically run on well-understood issues and are
probably selected for qualities that have been long observed by their fellow villagers.23
Timing Innovative provincial governments began experimenting with elections in the early
1980s. Elections were formally codified by the central government in the Organizational Law
on Village Committees (OLVC) in 1987. From this point onwards, all provinces were pushed
to introduce elections in all rural areas. A revision of the OLVC in 1998 required candidate
nominations to be open to all villagers.
The decision to introduce elections at the province-level was the result of political pres-
sure and bargaining between the central government and the provincial leaders. However,
implementation within provinces was mainly imposed top-down by bureaucratic fiat. Each
level of government would pilot the reform in a few select villages, and the reform would be
widely implemented once the procedures and logistics were tested (O’Brien and Li, 1999).22As Kelliher (1997) discusses, according to the law, the village committee operates under the leadership
(lingdao) of the Party.23There are very few accounts of actual electoral campaigning. In many cases, elections were set up with
only a few days’ notice (Unger, 2002: p. 221).
10
Anecdotal evidence from interviews that the authors conducted with county- and province-
level officials suggests that the pattern of the roll-out was mostly orthogonal to village char-
acteristics. This is consistent with the speed of roll-out within provinces. By all accounts,
villages had no discretion over the timing of introduction of elections, which is characteristic
of reforms in rural China.24
The interviews that the authors conducted with village-, county- and province-level
officials as well as the evidence from qualitative studies only point to two exceptions to
the quasi-random timing of the within province introduction of village elections. First, the
model villages that piloted the reform obviously received elections earlier. Second, there
are a few accounts of elections being delayed in “problematic” villages that had a history
of non-compliance with unpopular central government policies (e.g., One Child Policy or
the permanent expropriation of village land by the upper-levels of government) or had a
large kinship clan that could dominate other villagers in a majoritarian regime (e.g., Han
and O’Brien, 2009; Oi and Rozelle, 2000). To examine the quantitative importance of these
factors for determining the timing of elections, we collected data on the allowance of One
Child Policy exemptions and the incidence of upper-government land expropriations in the
VDS. Later, we will examine the correlation between these variables and the introduction
of elections. Afterwards, in the robustness section, we control for them explicitly to check
that they do not confound our main results. Finally, we also check that our estimates are
not driven by pilot or straggler villages in each province.24In his detailed study of elections, Sinologist Unger (2002, p. 222) writes that “These [elections] should
not be interpreted as bottom-up initiatives by the villagers themselves; they are not in a position to playany precedent-setting part in the initiation of new electoral reforms. There is a mistaken belief amongsome people outside China regarding this... elections are quietly being instituted at levels above the village,engineered first in selected districts at a distance from Beijing, through the connivance of the [central]Ministry of Civil Affairs and middle-ranking officials out in the regions”. Unger (2002) also notes the generalpassivity of villages in implementing rural reforms such as land reforms and the adoption of the HouseholdResponsibility Reform earlier in the reform era.
11
3 Conceptual Framework
3.1 Accountability
The anecdotal evidence described above suggests that prior to the introduction of elec-
tions, local officials could benefit from the agency problem in two distinct ways. First, they
could shirk in their efforts to provide and maintain public goods. Second, they could engage
in rent seeking and pursue policies that benefitted them and other village elites. Elections
were introduced in order to mitigate these concerns. Hence, if the reforms were effective
in making local officials accountable to villagers to some extent, we would expect the in-
troduction of elections to affect several outcomes. Public goods provision should increase
since re-election concerns give village officials incentives to exert the necessary effort. Due
to this increase in accountability villagers should also be willing to contribute more funds
to the government, further improving the provision of public goods. Land use and other
policies should change in favor of the majority preferences in the village, which would result
in changes in income distribution.25 The purpose of the rest of this paper is to document
the effect of the introduction of elections on these outcomes.
3.2 Empirical Strategy
The empirical strategy used in our main analysis assumes that the introduction of elec-
tions was quasi-random once province-specific time trends and village and year fixed effects
are included. Specifically, we use a differences-in-differences (DD) strategy, where we com-
pare the evolution of outcomes of villages that have had their first election to villages that
have not yet implemented their first election. Our baseline estimates control for village and
year fixed effects. Village fixed effects control for all time-invariant or slow-moving differences
between villages, such as geographic characteristics (e.g., hilliness or distance from a city)
or culture. Year fixed effects control for changes over time that affect all villages similarly
(e.g., national policy changes, macroeconomic growth). In addition, we add province-time
trends, which control for the widening differences across regions brought about by unequal25The effects of an increase in leader accountability on village policies are formalized in a companion
paper, Martinez-Bravo et al. (2011).
12
economic growth during the long time horizon of our study.26 Since we believe that the tim-
ing of elections is endogenously determined at the province level, but quasi-random within
provinces, these trends have the additional advantage of capturing a significant amount of
the cross-province variation. The baseline specification also controls for the second wave of
reforms that opened the nomination of candidates to villagers. This allows us to control for
potential heterogeneity in the effect of elections.27 The baseline equation that characterizes
the effect of elections is
Yvpt = βEvpt +λOvpt + γpt +δv +ρt + εvpt , (1)
where the policy outcome of village v in province p during calendar year t, Yvpt , is a function
of: a dummy variable, Evpt , that takes the value of one after the first election in village v
has taken place; a dummy variable, Ovpt , that takes the value of one after the first open
nomination in village v has taken place; province-year trends, γpt; village fixed effects, δv;
and calendar-year fixed effects, ρt . Since the timing of elections was largely decided at the
province level, we cluster the standard errors at the province-level. As we only have 29
provinces, we address the possibility of small sample bias in the clustered standard errors
by also presenting p-values derived from wild bootstraps as recommended by Cameron et
al. (2008).28 The main coefficient of interest is β . It is statistically different from zero if
elections affect a particular policy outcome.
Interpreting β as the causal effect of introducing elections does not require us to assume
that election timing within provinces was random. Instead, it requires the weaker assumption
that conditional on the baseline controls, the introduction of elections is not correlated with
time-varying village characteristics that affect the outcomes of interest through channels26In the robustness section we show that the magnitude of our estimates are similar when we control for
province-year fixed effects (although the estimates are less precise than when we control for province trends).27This improves the precision of our estimates, but does not affect the magnitude of estimated effects of
the introduction of elections. For brevity, we only report results where we control for the introduction ofopen nominations. Results without these controls are available upon request. Note that we do not controlfor other procedural differences in elections because they are much more likely to be endogenous.
28The bootstraps are estimated using 500-999 repetitions.
13
other than elections. We do not take this identification assumption as given and provide a
large body of evidence supporting its validity later in the paper. In particular, before we
present the main results, we present evidence that the timing of elections within provinces
was uncorrelated with a large number of village-level characteristics. Then, after we present
the main results, we show that there is no evidence of pre-trends and conduct a large number
of additional robustness and sensitivity checks. In particular, we show that our estimates
are similar when we control for the timing of the first introduction of village elections in
each province, which is the main source of endogeneity.
4 Data
4.1 The VDS and NFS Surveys
The primary data used in this paper for elections and public goods expenditure are from
the The Village Democracy Survey (VDS), a unique retrospective village-level survey con-
ducted by the authors of this paper. The first wave, conducted in 2006, records the history
of electoral reforms, de facto leader power, public goods expenditures, and the enforcement
of central government policies. The second wave, conducted in 2011, records the names and
characteristics of all village leaders since 1982. To ensure accuracy of the historical data,
the retrospective VDS relies on administrative records for each village when possible. When
village records are not available we relied on the recall of survey respondents, which include
all current and former living village leaders and elders (e.g., teachers) in each village. This
applies to very few of our variables and we will note it when they are relevant. The VDS
forms a balanced panel of 217 villages for the years 1982-2005. The villages we survey are
the same villages surveyed by the National Fixed-Point Survey (NFS), which we discuss
next.
Our measures of income and land are reported by the NFS, a detailed village- and
household-level economic survey collected and maintained by a research center of the Min-
istry of Agriculture of China. It is collected each year beginning in 1986, with the exception
of 1992 and 1994 due to administrative issues. The NFS villages were chosen in 1986 to
14
be nationally representative for rural China. Within each village, approximately 25% of
households were randomly selected in 1986 and followed over time; new households were
introduced over time to maintain representativeness.29 From the NFS, we were able to ob-
tain village-level data for variables such as the amount of village land that is dedicated to
farming and the amount of village land that is leased out to enterprises. These variables
will be used in several supplementary exercises examining public goods. We also obtained
the total income of households at the 10th, 25th, 50th, 75th and 90th percentiles of the
within-village-year total income distribution.30 For a subset of villages in ten provinces, we
were also able to obtain more detailed household-level data. Because this subsample is small
and non-representative, we only use it to supplement the main analysis. This subsample is
discussed further in Section 7.
The main analysis uses the VDS panel data with the addition of the variables for land and
income percentiles discussed above from the NFS. The main sample comprises a balanced
panel of 217 villages from 29 provinces.31
Our data have several advantages. First, to the best of our knowledge, the VDS data are
the most comprehensive data on village-level reforms ever constructed. They cover a period
of time starting in the early 1980s. In addition to recording the history of electoral reforms,
we also record the timing of other major rural reforms, the occurrence of village mergers, and
numerous other village-level characteristics. This allows us to control for heterogeneity across
villages more comprehensively than past studies, which is particularly important given the
natural diversity across China. The richness of the data also allows us to provide a detailed
analysis of the effect of elections on a range of policies and to assess the mechanisms driving
the reduced-form effects. Second, the NFS economic data and the village administrative29According to the Ministry of Agriculture, there is very little attrition and households and villages are
mainly added to adjust for gradual demographic changes.30We are grateful to Wu Zhigang for computing these statistics for us for each village-year.31There are 31 provinces in China at the end of our sample period. The two excluded provinces are
Tibet and Chongqing. Tibet is excluded because it is subject to different political and economic policies.Chongqing is a city-municipality that is excluded because it did not achieve provincial status until 1997.The three other city-municipalities with provincial status (Beijing, Shanghai and Tianjin) are included inour data. Each contain a substantial rural population (30% or higher). We will control for whether a villageis a suburb of a city later in the section on robustness and show that our results are not influenced by theirinclusion.
15
records that we surveyed in the VDS were collected contemporaneously. Hence, we avoid
recall bias. Third, the panel structure of the survey allows us to control for village fixed
effects and province-year trends. Finally, the fact that the NFS samples a large number
of households in each village means that we are able to examine the effects of elections on
income distribution within villages.
The main drawback is that the variables included in the NFS change over time to meet
the needs of the Ministry of Agriculture. Thus, we have a small number of observations
for many interesting variables (e.g., school enrollment) and cannot examine them in the
regression analysis.32 All observations in the empirical analysis are at the village-year level.
We will describe additional variables as they become relevant.
4.2 Descriptive Statistics
Several pieces of descriptive evidence are consistent with the anecdotal evidence on the
timing of electoral reforms discussed in Section 2. First, the data show that there is sub-
stantial within province variation in the timing of the first election in each village. In
a village-level cross-sectional regression, when we regress the year of the first election on
province fixed effects we find that the R-squared is 0.33. Thus, approximately 67% of the
variation in the timing of elections is within province. This is important for our strategy,
which largely relies on this variation.
Second, the timing of the rollout is consistent with rapid top-down implementation within
provinces and counties. Our data indicate that 16% of villages held their first elections
prior to the official introduction of elections by the county government, 66% held their first
elections the year that the county introduced elections, and 18% held their first election
afterwards.33 Table 1 shows that the average village implemented its first election within
the same year as the official introduction of elections in its county and five years after the first32There are many other interesting variables that are inconsistently collected and therefore not used in
our analysis (e.g., obligated working days, roads). Another drawback of the NFS is that it did not collectdetailed demographic data.
33Note that the timing of the official introduction of elections in each county is based on respondent recall.To maximize accuracy, our surveyors only record a date if all respondents surveyed in a given village agree.If there is no consensus, this variable is recorded as missing. Since provinces are large and respondents couldnot confidently recall the year of the first election within a province, the date of province-level introductionis inferred as the year of the first election in each province according to our survey.
16
election in the same province. Since the 29 provinces of our sample include approximately
2,885 counties and 623,669 rural villages (as defined by the number of village governments,
cunming weiyuanhui), these statistics imply that the average county was able to introduce
elections in 143 villages within one year.
Third, the fact that a small number of villages implemented elections before and after
the official introduction in each county is consistent with the anecdotal evidence that each
administrative division typically piloted the reform before officially introducing it and also
delayed elections in a few villages. Hence, given our identification assumption, it is important
to check that our baseline estimates are not driven by the early movers or the stragglers.
In addition, we can provide direct evidence that the timing of the first election is un-
correlated to most pre-reform village characteristics. We examine the pair-wise correlation
between election timing and a large number of village characteristics such village size, prox-
imity to an urban area, proxies for social and economic structure, measures of the pre-existing
level of public goods provision and other outcomes of interest. We measure all these village
characteristics in the first year that data are available.34 The sample for this estimation is
therefore one cross section of villages. We regress the year of the introduction of elections
on each village characteristic. Since we are interested in within-province variation, we con-
trol for province fixed effects. For brevity, we present the results for village demographic
and physical characteristics, the main outcome variables of our regression analysis, and the
unpopular upper government policies that we discussed in Section 2.35 Since it is difficult
to compare magnitudes across different regressors, Table 2 presents the standardized coef-
ficients for each regression. Only one of the correlates is statistically significant. Given the
large number of correlates that we examine (we examine over eighty additional insignificant
correlates that are not presented in the paper), the significance of one correlate is not nec-34Most variables reported by the NFS are available first in 1986. Land variables are available starting
in 1987. Measures of the One Child Policy and upper-government land expropriation from the VDS areavailable starting in 1982. The results are similar if we measure the latter two variables in 1986. The resultsare also similar if we measure all variables as the average of the first two years for which they are available.These alternative results are available upon request.
35See the Online Data Appendix for a description of the variables for upper-government land expropriationand One Child Policy exemptions, which we collect in the VDS.
17
essarily meaningful. Nevertheless, in the robustness exercises later in the paper, we will
control for this and other potentially confounding variables.
The data also provide several pieces of descriptive evidence that suggest that elections
were effectively implemented. We find that 79% of elections had more candidates than
positions, as the law required. Most of the elections with too few candidates were the
first elections in their villages, and were all immediately followed by fresh elections in the
subsequent year. This is consistent with the view that opponents to the electoral reform
were unable to fully derail the introduction of elections, and with qualitative accounts of
dissatisfied villagers demanding and obtaining recalls (O’Brien and Li, 2006). Table 1 shows
that, as legally required, elections occur every three years on average.36 Interestingly, there
was a 38% VC turnover rate for the first election, which is more than twice as high as the
average turnover rate in the sample (17%).
5 Main Results
5.1 Public Goods
We first examine public goods expenditure at the village level, which villages record as the
sum of expenditures on seven categories that are defined by the Ministry of Agriculture: ir-
rigation, primary schools, sanitation, within-village roads, electricity, the environment (e.g.,
planting trees), and “other”.37 These village administrative records are recorded in the VDS
and are available for all years and villages during 1986-2005. Table 3 presents the effect of
elections on public expenditures from estimating equation (1). Our data allow us to sepa-
rately examine expenditures according to the source of the funds, which we categorize into
funds from villagers and non-village sources.36Note that there is variation in this variable (the standard deviation is approximately one year), which
mitigates the concern that village records report elections as they are supposed to occur and rather thanwhen they actually occur.
37The villages in our sample began recording public goods expenditures in 1986 at the request of theMinistry of Agriculture. The accounting methods, the categories for public goods, and the sources of fi-nancing are all determined by the ministry. In addition to public goods expenditures, village governmentexpenditures also cover other items such as salaries of local cadres and expenditure on festivals and celebra-tions. In our data, public goods expenditures account for approximately 27% of total village governmentexpenditures.
18
Panel A of Table 3 lists the average amount for each type of expenditure and Panel
B presents the regression estimates. Note that for all estimates we report standard errors
that are clustered at the province level in parentheses and wild bootstrapped p-values in
square brackets immediately below. Column (1) shows that elections increase total public
good expenditures from all sources by approximately 154,500 RMB (22,071 USD), where
the sample mean expenditure is 142,800 RMB (20,400 USD). The estimate is significant
at the 5% level. The result is consistent with our hypothesis that elections changed the
incentives of village leaders and led them to exert more effort and be more responsive to
villagers’ demands for more public goods. The large magnitude of the increase is probably
a consequence of pent-up demand, since there was extreme low provision of public goods
prior to the elections (e.g., Luo et al., 2007; Luo et al., 2010).
To further support this interpretation, we analyze expenditure by source of funding in
columns (2)-(3) of Panel B. The top row shows that, consistent with the anecdotal literature,
village leaders are responsible for raising most of the funds required for village public goods
–approximately 70% of total funding for village public goods comes from village sources.
The regression results show that the aggregate increase in public goods expenditure found
in column (1) is entirely driven by an increase in funding from villagers. The estimate for
village financing in column (2) is similar in magnitude to the estimate for total financing and
statistically significant at the 5% level. In contrast, the estimate for non-village financing in
column (3) is near zero in magnitude and statistically insignificant. This result is important
for three reasons. First, it demonstrates that elections affect policy at the village level,
since the responsive funds are under the control of village leaders. Second, it contradicts
the notion that democratically elected leaders cannot raise revenues for public goods in our
context.38 Third, it contradicts the alternative interpretation that the estimated effects of
the introduction of elections are driven by an increased willingness of the upper-government
to fund local public goods, since direct transfers are the most immediate policy tool that
upper levels of governments would use to affect local public good provision.38For example, see the classic work of Huntington (1968) for a discussion of why democracy hinders the
government’s ability to raise taxes.
19
In addition to the results presented here, in the Online Appendix we also provide sug-
gestive evidence that the increase in public goods was likely to correspond to demands from
villagers.
5.2 Land Use and Household Income
Recall from Section 2 that village leaders are responsible for the allocation of village
land to households, and that some land is often leased out to village enterprises. Villagers
typically dislike this practice, because they suspect it is a source of rents for the village
leadership and its cronies, and they would rather have direct control over more land.39
Data for the use of village land is reported by the NFS for all villages for the years
1987-2005 (excluding 1992 and 1994). The villages in our sample dedicate approximately
96% of arable land (approximately 51% of total village land) to households for farming.
Approximately 75% of the remaining arable land is leased out to “enterprises”, a term which
we use for firms run by collectives or villagers (see Table 1). Since elections can only reduce
the amount of land leased out to firms if such land existed prior to the first election, we
restrict our analysis to villages that ever used any arable land for non-household farming
prior to the first election. This reduces the sample to 108 villages from 28 provinces.40
The estimates are displayed in column (1) of Table 4, where we use the same specification
as in equation (1). It shows that the introduction of elections reduced the amount of land
that is leased out to enterprises by approximately 57.7 mu (3.85 hectares), where the sample
average is approximately 108.5 mu (7.23 hectares). The estimate is statistically significant
at the 10% level.
Since the village leadership and other village elites likely captured most of the rents from
running village enterprises, reducing the amount of land allocated to enterprises and giving
it back to village households is a change that reduces pro-elite bias in village policies. Land
allocation to enterprises is just a particular example of the many levers the village leadership39Consistent with this view, in a cross-sectional study Brandt and Turner (2007) find that redistributing
collective land to villagers is positively correlated with re-election probabilities. We present evidence inSection 7 that income from enterprises is not shared equitably across households.
40Our results are similar if we alternatively restrict the sample to villages that ever leased any land toenterprises during any time in the sample.
20
can use to affect rent allocation across households.41 If elections cause a reduction in the
pro-elite bias of these policies, they should result in income losses for elite households. To
examine this possibility, we look at the effect of elections on household income at several
percentiles of the village distribution.
We have data on total household income for households on the 10th, 25th, 50th, 75th
and 90th percentiles of the within village and year income distribution for the full sample
of villages for 1986-2005 (excluding 1992 and 1994).42 In the top row of Table 4, we present
descriptive statistics on the distribution of household income to illustrate the extent of in-
come inequality in villages. In our sample, the richest households (90th percentile) typically
earn twice as much as the median household (50th percentile), who again earns about twice
as much as the 10th percentile. Hence, despite the fact that de-collectivization is recent,
there is already substantial inequality within villages. This can be the result of cronyism if
the richest households are benefitting from their connections to the village leadership. At
the same time, it could also be a consequence of some households being entrepreneurial and
successful and thereby experiencing more rapid income growth during the early reform era
than less productive households. These two causes of inequality are not mutually exclusive.
To examine the effect of the introduction of elections on income distribution, we estimate
the baseline equation (1) with household income at each percentile as the dependent variable.
Columns (2)-(6) of Table 4 show that elections do not statistically change incomes for
most households, but significantly reduce incomes for the 90th percentile households by
5,056 RMB (722 USD), where the sample mean is approximately 24,476 MRB (3497 USD).
This reduction represents approximately 20% of the sample mean. We also note that the
coefficients become progressively more negative for higher income deciles.
In column (7), we examine the ratio of the median household income to the 90th per-
centile household. The estimates show that elections increased the income of the median41For instance, the leadership is responsible for choosing who obtains a salaried job in the village gov-
ernment or in village enterprises, and as we discussed earlier, it also decides household land allocation. InSection 7, we examine other channels using a subsample of villages for which we have richer household data.
42For example, the income of the 10th percentile household refers to the income of the household that ison the 10th percentile of the income distribution within a village during a given year. These were computedby the Ministry of Agriculture from the NFS household level data.
21
household relative to the 90th percentile household by 1.7 percentage points. The estimate
is statistically significant at the 5% level. Relative to the mean income ratio of 0.53, this
increase is moderate in magnitude.
These results suggest that upon the introduction of elections, the village leadership
changed economic policies such as land distribution and the management of village en-
terprises in such a way that the former economic elites were less favored. The welfare
implications of this reduction in inequality are unclear and partly depend on the origins of
pre-reform inequality. Note that we do not observe a significant increase in income at any
level of the income distribution, so it might well be that these changes introduced economic
inefficiencies. We discuss this further in the conclusion. In any case, the relative drops of
income only affect the richest households, which is consistent with the interpretation that
they are a result of policy changes designed to please the majority of villagers.
5.3 Robustness
There are three main concerns for our empirical strategy. The first concern is that despite
controlling for province time trends, our baseline results are partly driven by cross-province
variation in timing, which is determined by provincial leaders for potentially endogenous
reasons. We address this issue in three ways. First, to control for the province-level timing
of the decision to introduce village elections we add a dummy that indicates whether any
village in a given province has introduced elections. The estimates of this robustness check
on our key outcomes are shown in Table 5 column (2). They are similar to the baseline
estimates displayed in column (1) for ease of comparison. Hence, our results are not an
outcome of province-level variation in the timing of the introduction of the reform, which is
the main source of endogeneity concerns.
A second way of accounting for province-level factors is to introduce province-year fixed
effects instead of province-year trends. This allows the influence of province characteristics
to vary flexibly over time. Column (3) shows that the coefficients are similar to the baseline,
although the estimates are less precise.
Finally, we control for province-level variables such as per capita GDP, per capita agri-
22
cultural GDP, and per capita government expenditure in public goods.43 The estimates
with these controls, which do not differ from our baseline, are shown in column (4).
The second main concern is that our estimates could be driven pilot or straggler villages
in the reform implementation (see Section 2), which may have been chosen endogenously. In
column (5), we repeat our estimates on a restricted sample where we drop pilot and straggler
villages.44 The estimates are nearly identical to the baseline.
The remaining concern is that there may be village-specific and time-varying determi-
nants of the introduction of elections that are not controlled for by the baseline controls and
that affect the outcomes of interest through channels other than elections. In addition to the
correlational evidence presented earlier in Table 2, we provide several pieces of additional
evidence against the possibility of endogenous timing.
First, we provide evidence for the parallel trends assumption. For our DD estimate to
be consistent, in the absence of elections, the outcomes of villages that introduced elections
earlier should have evolved along parallel trends with villages that introduced elections
later. Since it is impossible to observe the counterfactual trend, we follow the literature in
conducting a pre-trend analysis in the main outcomes of interest. The presence of a trend
in the years leading up to the introduction of elections would be a sign that there was a
different evolution in villages that were about to implement elections.
To investigate the presence of pre-trends, we estimate the coefficients of the dummy
variables of each year since the first election (i.e. dummies for one year before the first
election, two years before, etc.). We can simultaneously estimate these election-year dum-
mies together with the year fixed effects because the timing of elections is staggered. To
avoid compositional effects in estimating these election year dummy variables, we restrict
the sample to villages that are observed for the same number of years before and after the
election – i.e., we form a panel of villages that is balanced in election years. If there are43These data are reported by China Statistical Yearbooks.44A pilot village is defined as a village that implemented elections more than two standard deviations before
the average timing of villages in the same province. A straggler village is defined as one that implementedelections more than two standard deviations after the average timing of villages in the same province. If allof the villages in a province implement elections in the same year, then that province will have no pilot orstraggler villages.
23
no pre-trends, then the estimated dummies for the years prior to the first election should
be close to zero. Moreover, if the main results reflect changes that begin when elections
are introduced, the estimated dummies should begin to differ starting the first year of the
election.
The coefficients and the standard errors for the estimated effects of the number of years
since the first election are presented in Online Appendix Table A.4. The coefficients are
plotted in Figure 1. For visual reference, we plot the value of zero for the reference years.
To minimize noise, our reference group, indicated as -4 on the x-axis, actually comprises
of -5 and -4 years since the first election.45 The year when the first election takes place
is indicated as election-year 0 on the x-axis. There is no evidence of pre-trends for any of
the three outcomes, which is consistent with the parallel trends assumption. Moreover, the
trend breaks in the coefficients at year zero show that the effect of elections begin the year
of the first election, which is consistent with the interpretation that the main DD estimates
capture the influence of the electoral reforms rather than spurious trends.46
One may still be concerned that election timing is correlated with pre-conditions that af-
fect our outcomes of interest through channels other than elections. We address this concern
by directly controlling for the interaction terms of the characteristic that was significantly
correlated with the timing of the introduction of elections – the occurrence of upper govern-
ment land expropriation (see Table 2) – and the full vector of year fixed effects. Given the
anecdotal evidence on the delay of elections for villages with a history of non-compliance to
unpopular central government policies, we also control for the interaction of whether there
were exemptions to the One Child Policy and year fixed effects. These characteristics are45The main constraint to how many pre-reform years we can observe is data availability. Online Appendix
Table A.1 shows that extending the balanced panel to include more pre-election years would cause us to losetoo many observations.
46We also investigate the presence of pre-trends by regressing each of the three main outcomes againstall the explanatory variables in the baseline plus a linear trend in the years since the first election andits interaction with the post-election dummy variable. This estimates a years-since-the-first-election trend,where the trend is allowed to differ between the pre- and post-reform periods. The main advantage of thismethod is that we can use the full sample of villages. Since we do not dummy out each year since the firstelection and most villages are observed for at least one year during the pre- and post-first-election periods,this method is much less likely to be confounded by composition effects. The coefficient for the uninteractedtrend variable (the pre-trend) is always small in magnitude and statistically insignificant. We conclude thereis no pre-trend. These results are available upon request.
24
measured in the first year that data are available (1982). The interaction with year fixed
effects controls for the influence of these variables over time in a fully flexible manner, and
to some extent, it also controls for the influences of all of its correlates over time. Column
(6) of Table 5 shows that the coefficients with these additional controls are very similar to
our main estimates.
We consider the possibility that several other village-level factors could potentially con-
found the effect of elections on our outcomes of interest. These factors include whether a
village is a suburb of a city (a dummy variable interacted with year fixed effects), whether
the Tax and Fee Reform had been implemented, and the level of village social capital (in-
teracted with year fixed effects). To proxy for the latter we follow Tsai’s (2007) work in
using the presence of a lineage group (i.e., an ancestral hall, family tree) , village temple, or
a large kinship group to proxy for social capital.47 To maximize the statistical precision of
our estimates, we use the principal component of these three measures as our social capital
proxy. Column (7) shows that the resulting coefficients are similar to the baseline. In column
(8), we re-estimate the baseline on a sample restricted to villages that never experienced an
administrative merger with another village since 1982. The coefficients are also similar to
the baseline.
A final remaining concern is that the timing of electoral reforms could have coincided
with the timing of changes in the Communist Party. Since we interpret our results as
the consequence of increased accountability of the newly elected leader in the village, such
simultaneous change would be a confounding factor. To address this possibility, we conduct a
placebo test comparing VCs, who were affected by the electoral reform and Party Secretaries
(PS), who were not supposed to be affected. In particular, the PS continued to be appointed
by upper-level authorities after elections were introduced in the villages. If there is no
change in the Communist Party, then PS turnover and characteristics should not change
with the introduction of elections. Table 6 Panel B shows the estimates from regressing PS
turnover rate, age and years of educational attainment on the baseline explanatory variables.47To measure the size of the kinship groups, the VDS recorded names from the village roster.
25
Elections have no effect on any of these outcomes. In contrast, Panel A shows that elections
increased turnover rate, reduced age, and increased the years of educational attainment for
the VC. The results of the placebo test support our interpretation and are consistent with
the anecdotal literature which emphasizes that the introduction of village elections was not
accompanied by other changes at higher levels of government.
In addition to the robustness tests presented in the paper, we conduct many others that
are not presented here for brevity.48 For example, we address the fact that several villages
introduced elections prior to when our outcome data begin. To check that our estimates are
not driven by villages for which there is no variation in elections, we re-estimate the effect
of elections on each outcome where each regression is restricted to a sample of villages for
which we observe public goods before and after the first election. Alternatively, we check that
our estimates are not driven by especially motivated villages that implemented their first
election before the national law by re-estimating our main equation using a sample restricted
to villages that held their first elections after 1987. We also check that our estimates are
not driven by selection within counties by instrumenting for the introduction of elections at
the village level with the introduction at the county level or with the introduction at the
province level. All of these robustness exercises produce estimates that are very similar in
magnitude to our baseline results.
The results in this section show that the baseline results are very robust and unlikely
to be confounded by other factors. They support the interpretation of the estimates as
capturing the causal effects of the introduction of village elections.
6 Re-Election Incentives vs. Leader Selection
In the literature in political agency, citizens use elections to control politicians through
two mechanisms.49 First, elections can help voters address moral hazard problems by re-
warding good performance with re-election – i.e, elections serve as means to provide the48They are available upon request.49This literature is large, starting with the seminal contribution of Barro (1973). For textbook treatments,
see Besley (2006), and Persson and Tabellini (2000) .
26
correct incentives to office holders. Second, voters can use elections to select the politicians
that are more competent or whose preferences are better aligned with citizens’ preferences.
In this section, we explore whether these forces are relevant in our context by estimating
the interaction effect of the introduction of elections and whether there was turnover for the
first election.
The impact of elections in villages where there was no turnover is interesting because it
isolates the effect of re-election incentives. This is because the presence of the same official
before and after elections rules out any effect due to villagers having selected a better official.
Table 7 presents the results. For comparison purposes, Panel A restates the average effect
of the introduction of elections on our main outcomes: public goods expenditure funded
by villagers, amount of land leased out to enterprises and the incomes of the ninetieth
percentile. In Panel B, we introduce the interaction of the introduction of elections with a
dummy variable that equals one for villages where there is VC turnover in the first election.50
In this specification, the effect of elections on villages that experienced no turnover (64%
of the villages in our sample) is captured by the uninteracted post first election coefficient.
For all of the outcomes examined, the sign of this coefficient has the same sign as the
average effect displayed in Panel A. The effect for public goods is statistically significant at
the 1% level and the coefficient is larger in magnitude than the average effect in Panel A.
The uninteracted coefficient for land and income are reduced in magnitude and no longer
statistically significant. These results suggest the re-election incentives contributed to the
average effect of introducing elections, at least for raising public goods expenditure.
The effect of elections on villages that experience turnover is the sum of the uninteracted
post-first-election coefficient and its interaction coefficient. The sums of the estimates in
Table 7 panel B are approximately zero for public goods, negative for the amount of land
leased to enterprises and negative for income of the ninetieth percentile households. However,
the joint estimates are very imprecise (they are not presented). The interaction coefficients,
which capture the differential effect of the introduction of elections between villages with and50This equation is similar to equation (2), with the addition of the interaction term.
27
without turnover, are also statistically insignificant for all outcomes. It is difficult to infer
the relative contribution of the re-election incentives and selection mechanisms from these
coefficients. The interaction effect cannot be interpreted as the effect of elections through
leader selection because leader turnover is endogenous. For example, elections could cause
leader turnover where villagers have particularly high demand for reducing the amount of
land that is leased out or the income of rich households.
In summary, the results suggest that re-election incentives contributed to the average
effect of the introduction of elections. At the same time, we cannot rule out selection effects,
which would be consistent with our finding that elections reduced the average age and
increased the average education of VCs (see Table 6).
7 Policies for Redistribution
In Section 5, we showed that the introduction of elections causes a fall in top incomes
and a modest increase of the ratio of median to top incomes. The goal of this section is to
understand how the village government brought about the reductions in income inequality
given that it was unable to impose recurrent taxes on income. We first examine whether
the reduction in inequality is achieved by an increase in the progressivity of local levies.
Then, we examine whether the introduction of elections reduced the ratio of median to top
incomes for the sources of income that can be influenced by the village government.
To explore these issues, we use household-level data that breaks down household income
and expenditure by type. We were able to obtain these data for a subsample of villages of
the NFS survey. This subsample comprises 78 villages in eighteen provinces. The observable
characteristics of these villages do not differ significantly from the full sample.51
The disaggregated expenditure data report the fees and levies paid by households. Un-
fortunately, the data only report the sum of local fees paid to the village and county govern-
ments and does not distinguish payments to the village government. Therefore, interpreting
the following results requires the assumption that elections did not change the taxes paid51See Online Appendix Table A.2 for a comparison of observables and Table A.6 which shows that the
effect of elections on income for the full sample are replicated in this subsample.
28
to county governments. To the best of our knowledge, this is true.52
On average, households in the subsample pay 198 RMB per year in local fees on average,
which is approximately two percent of total household income. According to our data,
this amounts to 64% of the fees that households pay to levels of government above the
county (excluding the value of grain taxes).53 To examine the distribution of fees across
village households, we examine household payments of local fees according to households’
positions in the income distribution. In Table 8 Panel A, we show the mean and standard
deviation of the local fees rate (i.e., annual household local fee payment as a fraction of
total household annual income) paid at different points of the village income distribution.
In Panel B, we show the same statistics for the base year of 1986. The rate appears to
be constant across deciles. If anything, the average local fee rates are weakly decreasing
in income, although the change is not statistically significant. In Panel C, we present the
effect of the introduction of elections on the fee rate. The coefficients for post first election
are all positive and statistically significant at the 10% or higher level. They are similar in
magnitude for households in different quintiles of the village income distribution. If anything,
the coefficient, and thus the increase is slightly smaller for higher income households. Thus,
there is no evidence that elections increased the progressivity of local fees and levies. This,
together with the very low overall rate suggest that local fee collection is not a relevant
redistributive channel.
Next, we examine the effect of elections on the different sources of household income.54
The data show that for the average household, income from agriculture is about 67% of
total income. Wage income, which in rural villages is mostly earned from village enterprises,
accounts for about an additional fifteen percent of income, and dividends and other enterprise
income makes up seven percent.55 Village enterprises and household access to land, which is52The main reform affecting local fees was the Tax and Fee Reform. The VDS documents the introduction
of this reform which occurred towards the very end of our sample. Controlling for the introduction of thisreform has no effect on our results.
53The largest proportion of taxes that agricultural households pay to the central government is in the formof grain, which the government buys at a below-market price that it sets. The value of this tax payment isnot included in the NFS category of taxes to the central government.
54See Online Appendix Table A.5 for a balance sheet for average household income and expenditure.55Village enterprises were officially called different names at different stages of the early reform era due
29
the main input leading to household agricultural income, are both under the control of the
village government. It follows that about 90% of average household income can be influenced
by the village government. If the VC desires to engage in limited redistribution, it thus can
reallocate high quality land, salaried jobs or dividend income across households, since all of
these sources of income are under his control. However, as land reallocations will be strongly
resisted by the households that lose land, we would expect these tools to be used sparingly.
In Table 9, we present the estimates for the effect of the introduction of elections on
the median to top income ratios for all sources income that our data reports, as well as
for household land allocation. Column (1) presents the result for total income, which is
comparable with the result for the full sample in Table 4 and statistically significant at
the 5% level. Columns (2)-(4) show that the introduction of elections had similarly sized
effects on each source of income; the ratios for agricultural income, wage income, enterprise
income and land all increase by around two percentage points. The estimate for agricultural
income is significant at the 1% level, and it is particularly relevant given the importance of
agriculture in household income. Enterprise income is also statistically significant at the 5%
level. The estimate for the median to top ratio of household land is positive, but slightly
smaller in magnitude than the other estimates and insignificant.
Taken together, the results on household level income and expenditure are consistent
with our understanding of the limited tools available to the village government for income
redistribution.56
8 Conclusion
In order to better control local officials and prevent shirking and rent-seeking, the Chinese
government introduced village-level elections starting in the 1980s. Our results suggest that
the reforms were successful in changing the incentives of local officials to favor the interests
to regulatory changes regarding the ownership of former collective assets. Hence, our category of incomefrom village enterprises is the sum of all of these types of incomes reported in the NFS (e.g., income fromcollectives, partnerships or cooperatives, and enterprise dividends). Income from collectives and enterprisedividends make up over 97% of this category.
56To check the robustness of our estimates, we conduct the same set of sensitivity checks as for the fullsample estimates. They are not discussed for brevity and are shown in Online Appendix Table A.7.
30
of the village majority over that of the village elites. In particular, we find that public
goods expenditure made by the village government increased, and that the increase was
entirely financed by funds collected from villagers. This is consistent with the notion that
the introduction of elections increased villagers’ willingness to contribute funds to the local
government when the latter became more accountable – i.e., when the village government
was more likely to provide the public goods desired by villagers. We also find that elections
induce village governments to lease less land to enterprises. Consistent with the view that
the rents created by these enterprises were being captured by the village elite, we find
that elections caused income reductions among the richest households, which resulted in
a moderate reduction in inequality. Finally, we show suggestive results that the limited
redistribution brought about by elections is happening through the manipulation of income
streams that the village government can influence through their control over land allocation
and village enterprises.
To the extent that this institutional reform can be understood as a marginal increase in
democracy in rural China, our results are consistent with recent theories that characterize
democracies as regimes that are likely to engage in redistribution (e.g. Acemoglu and Robin-
son, 2001, 2006; Boix, 2003) and to provide more public goods (e.g., Besley and Kudamatsu,
2008; Bueno de Mesquita et al., 2003).
It is beyond the scope of our study to provide a welfare assessment of the policy changes
induced by elections. Given the extremely low prevailing levels of public good provision,
the increase in public good expenditure is likely to enhance welfare. However, it is very
difficult to assess the welfare effect of the observed patterns of redistribution. On the one
hand, if village enterprises were productive and salaried jobs were allocated efficiently in
the pre-election regime, the scaling down of enterprises and the change in the allocation of
salaried jobs might have reduced aggregate welfare and disincentivized effort by productive
members of the community. Redistribution would in this case be very inefficient. On the
other hand, the reductions in income among the richest households could reflect a reduction
in rent-seeking if the households benefitting from village enterprises prior to elections were
31
the cronies of the village government rather than the most productive members of the
community.57 The truth probably lies somewhere between these two extremes and may vary
across villages. Assessing the welfare effects of the introduction of elections is obviously an
important question for future research.
This study opens several other interesting avenues of future research. One important
question regards the heterogeneous effects of elections, which can potentially provide in-
sights into the pre-conditions of successful elections. For example, in subsequent research,
we investigate the roles of social fragmentation and social capital in determining the effects
of elections on public goods. Also, the motivation for autocratic regimes to introduce local
democracy is an important subject for future research. In particular, it is important to
understand when local elections are a step towards wider regime change versus when they
are used as an instrument of control by the central regime.58 We hope that the detailed in-
stitutional and economic data we have constructed will help future researchers in addressing
these and other important questions.57This is certainly the interpretation of village government conduct in Brandt and Turner (2007), Kennedy
et al. (2004), Oi and Rozelle (2000), Rozelle (1994) and Rozelle and Li (1998)58The latter hypothesis is consistent with the qualitative evidence discussed in Section 2. It is also con-
sistent with Lorentzen’s (2010) thesis that the Chinese government creates mechanisms for citizens to voicetheir preferences as a way to monitor cadres and improve governance, empirical evidence on the difficulty ofcentral planning in the context of the Chinese Great Famine (Meng et al., 2010) and current province-leveleconomic performance (Persson and Zhuravskaya, 2011). In particular, Persson and Zhuravskaya (2011)argues that Beijing’s recent strategy of rewarding regional leaders for measurable targets causes regionalgovernments to under-invest in objects that are difficult to observe and objects that only yield returns inthe long run.
32
ReferencesAcemoglu, Daron and James A. Robinson, “A Theory of Political Transitions,” Amer-ican Economic Review, September 2001, 91 (4), 938–963.
and James Robinson, “Why Did the West Extend the Franchise? Democracy, In-equality, and Growth in Historical Perspective,” The Quarterly Journal of Economics,November 2000, 115 (4), 1167–1199.
Banerjee, Abhijit, Rema Hanna, and Sendhil Mullainathan, “Corruption,” in RovertGibbons and John Roberts, eds., Handbook of Organizational Economics, Princeton Uni-versity Press, 2012.
Barro, Robert, “The Control of Politicians: An Economic Model,” Public Choice, 1973,14, 19–42.
Besley, T. and T. Persson, Pillars of Prosperity: The Political Economics of DevelopmentClusters The Yrjo Jahnsson Lectures, Princeton University Press, 2011.
Besley, Timothy, Principled Agents: Motivation and Incentives in Politics, Oxford: Ox-ford University Press, 2006.
and Anne Case, “Does Electoral Accountability Affect Economic Policy Choices? Evi-dence from Gubernatorial Term Limits,” Quarterly Journal of Economics, 1995, 110 (3),769–798.and Masayuki Kudamatsu, “Health and Democracy,” American Economic Review,May 2006, 96 (2), 313–318.
and , “Making Autocracy Work,” in Elhanan Helpman, ed., Institutions and EconomicPerformance, Cambridge: Harvard University Press, 2008.
and Robin Burgess, “The Political Economy of Government Responsiveness: Theoryand Evidence from India,” Quarterly Journal of Economics, 2002, 117 (4), 1415–1451.
Boix, Carles, Democracy and Redistribution, New York: Cambridge University Press, 2003.
Brandt, Loren and Matthew A. Turner, “The usefulness of imperfect elections: Thecase of village elections in rural China,” Economics & Politics, 2007, 19 (3), 453–480.
Cameron, A. Colin, Jonah B. Gelbach, and Douglas L. Miller, “Bootstrap-BasedImprovements for Inference with Clustered Errors,” The Review of Economics and Statis-tics, August 2008, 90 (3), 414–427.
de Mesquita, Bruce Bueno, Alastair Smith, Randolph Siverson, and James Mor-row, The Logic of Political Survival, Cambridge, MA: MIT Press, 2003.
Ferraz, Claudio and Frederico Finan, “Electoral Accountability and Corruption: Evi-dence from the Audits of Local Governments,” American Economic Review, 2011, Forth-coming.
Foster, Andrew D. and Mark R. Rosenzweig, “Democratization and the Distributionof Local Public Goods in a Poor Rural Economy,” Brown University Working Paper,Brown University 2005.
Gan, Li, Lixin Colin Xu, and Yang Yao, “Health Shocks, Village Elections, and Long-Term Income: Evidence from Rural China,” NBER Working Papers 12686, NationalBureau of Economic Research November 2006.
Gandhi, Jennifer and Ellen Lust-Okar, “Elections Under Authoritarianism,” AnnualReview of Political Science, 2009, 12 (1), 403–422.
33
Gil, Ricard, Mulligan Casey, and Xavier SalaiMartin, “Do democracies have differentpublic policies than nondemocracies?,” Journal of Economic Perspectives, 2004.
Han, Rongbin and Kevin O’Brien, “Path to Democracy? Assessing Village Electionsin China,” Journal of Contemporary China, 2009, 18 (60), 359–378.
Huntington, Samuel P., Political order in changing societies, New Haven, CT: Yale Uni-versity Press, 1968.
Husted, Thomas A. and Lawrence W. Kenny, “The Effect of the Expansion of theVoting Franchise on the Size of Government,” Journal of P, 1997, 105, 54–82.
Jacoby, Hanan G., Guo Li, and Scott Rozelle, “Hazards of Expropriation: TenureInsecurity and Investment in Rural China,” American Economic Review, 2002, 93 (5),1420–1447.
Kelliher, Daniel, “The Chinese debate over village self-government,” The China Journal,1997, pp. 63–86.
Kennedy, John J., Scott Rozelle, and Yaojiang Shi, “Elected leaders and collectiveland: Farmers’ evaluation of village leaders’ performance in rural China,” Journal ofChinese Political Science, 2004.
Li, Hongyi, Lyn Squire, and Heng fu Zou, “Explaining International and IntertemporalVariations in Income Inequality,” Economic Journal, January 1998, 108 (446), 26–43.
Lim, Claire, “Preferences and Incentives of Appointed and Elected Public Officials: Evi-dence from State Trial Court Judges,” Working Paper, Cornell University, 2012.
Lorentzen, Peter L., “Regularizing Rioting: Permitting Public Protest in an AuthoritarianRegime,” Working Paper Series, University of California at Berkeley 2010.
Luo, Renfu, Linxiu Zhang, Jikun Huang, and Scott Rozelle, “Elections, FiscalReform and Public Goods Provision in Rural China,” Working Paper, Freeman SpogliInstitute for International Studies, Stanford University 2007.
, , , and , “Village Elections, Public Goods Investments and Pork Barrel Politics,Chinese-style,” Journal of Development Studies, 2010, 46 (4), 662–684.
Martinez-Bravo, Monica, Gerard PadróiMiquel, Nancy Qian, and Yang Yao,“Do Local Elections in Non-Democracies Increase Accountability? Evidence from RuralChina,” NBER Working Papers 16948, National Bureau of Economic Research April 2011.
, , , and , “The Effects of Democracy on Economic Policy: Evidence from China,”Yale University Working Papers 2012.
Meng, Xin, Nancy Qian, and Pierre Yared, “The Institutional Causes of China’s GreatFamine, 1959-61,” NBER Working Paper 16361, 2010.
Miller, Grant, “Women’s suffrage, Political Responsiveness, and Child Survival in Ameri-can History,” Quarterly Journal of Economics, 2008, 123, 1287–1327.
Mu, Ren and Xiaobo Zhang, “The Role of Elected and Appointed Village Leaders inthe Allocation of Public Resources: Evidence from a Low-Income Region in China,” IF-PRI Working Paper http://www.ifpri.org/publication/role-elected-and-appointed-village-leaders-allocation-public-resources, International Food and Policy Research Institute 2011.
O’Brien, Kevin and Lianjiang Li, Rightful Resistance in Rural China, New York, NY:Cambridge University Press, 2006.
34
O’Brien, Kevin J., “Implementing Political Reform in China’s Villages,” Australian Jour-nal of Chinese Affairs, 1994, (32), 33 – 59.
and Lianjiang Li, “The Struggle over Village Elections,” in Merle Goldman and RoderickMacFarquhar, eds., The paradox of China’s post-Mao reforms, Cambridge, MA: HarvardUniversity Press, 1999.
Oi, Jean and Scott Rozelle, “Elections and power: The locus of decision-making inChinese villages,” The China Quarterly, 2000, 162, 513–539.
Olken, Benjamin A. and Rohini Pande, “Corruption in Developing Countries,” AnnualReview of Economics, 2012, 4, 479–505.
Persson, Petra and Ekaterina V. Zhuravskaya, “Elite Capture in the Absence ofDemocracy: Evidence from Backgrounds of Chinese Provincial Leaders,” SSRN eLibrary,2011.
Persson, Torsten and Guido Tabellini, Political Economics- Explaining Economic Pol-icy, Cambridge, MA: MIT Press, 2000.
Rodrik, Dani, “Democracies Pay Higher Wages,” The Quarterly Journal of Economics,August 1999, 114 (3), 707–738.
Rozelle, Scott, “Decision-Making in China’s Rural Economy: The Linkages between Vil-lage Leaders and Farm Households,” The China Quarterly, 1994, (137), 99–124.
and Guo Li, “Village leaders and land-rights formation in China,” American EconomicReview, 1998, 88 (2), 433–438.
Shen, Yan and Yang Yao, “Does grassroots democracy reduce income inequality inChina?,” Journal of Public Economics, 2008, (10-11), 1–17.
Tavares, Jose and Romain Wacziarg, “How democracy affects growth,” European Eco-nomic Review, August 2001, 45 (8), 1341–1378.
Tsai, Lily Lee, Accountability without Democracy: Solidary Groups and Public Goods Pro-vision in Rural China, New York, NY: Cambridge University Press, 2007.
Tyrefors, Bjorn and Per Pettersson-Lidbom, “Democracy, Redistribution, andPolitical Participation: Evidence from Sweden 1919-1938,” SSE Working Paperhttp://people.su.se/ pepet/democracyFeb2012.pdf, Stockholm School of Economics 2012.
Unger, Jonathan, The Nature of Chinese Politics: From Mao to Jiang, Cambridge, MA:M.E. Sharpe, 2002.
Whiting, Susan, “Contract Incentives and Market Discipline in China’s Rural IndustrialSector,” in John McMillan and Barry Naughton, eds., Reforming Asian Socialism, TheGrowth of Market Institutions, Ann Arbor: University of Michigan Press, 1996.
Wilson, James Q., Bureaucracy, Basic Books, 1989.
Zhang, Xiaobo, Shenggen Fan, Linxiu Zhang, and Jikun Huang, “Local governanceand public goods provision in rural China,” Journal of Public Economics, 2004, 88, 2857–2871.
35
Tab
le1:
Descriptive
Statistics
Varia
ble
Sour
ceO
bsM
ean
Std.
Dev
.#
of H
H in
Vill
age
NFS
3,64
142
1.81
280.
81N
ear C
ityN
FS5,
208
0.30
0.46
Tota
l Vill
age
Land
NFS
3,61
29,
245
14,7
19A
rabl
e La
nd (M
u)N
FS3,
612
2,29
52,
329
Sha
re o
f Vill
age
Land
that
is A
rabl
eN
FS3,
612
0.51
0.32
Use
d fo
r HH
Far
min
g (M
u)N
FS3,
612
2,21
52,
312
Not
Use
d fo
r HH
Far
min
g (M
u)N
FS3,
612
79.7
236
7.26
L
ease
d O
ut to
Ent
erpr
ises
(Mu)
NFS
3,61
260
.46
347.
61
Med
ian
HH
Ann
ual G
ross
Inco
me
Gro
wth
NFS
3,08
40.
069
0.18
9
The
Num
ber o
f Vill
age
Com
mitt
ee M
embe
rs
NFS
2,28
74.
362.
36Th
e N
umbe
r of V
illag
e Pa
rty C
adre
sN
FS2,
295
6.70
3.82
Party
Sec
reta
ry T
enur
eV
DS
5,20
810
.03
8.13
Vill
age
Chi
ef: T
enur
eV
DS
5,20
86.
696.
24
Has
Ele
ctio
nV
DS
5208
0.73
0.44
Has
Ope
n N
omin
atio
nV
DS
5208
0.20
0.40
Year
s bet
wee
n El
ectio
n In
trodu
ctio
ns in
Vill
age
and
Prov
ince
VD
S21
75.
025.
07Ye
ars b
etw
een
Elec
tion
Intro
duct
ions
in C
ount
y an
d Pr
ovin
ce**
VD
S21
74.
284.
67Ye
ars b
etw
een
Ele
ctio
n In
trodu
ctio
ns in
Vill
age
and
Cou
nty*
*V
DS
217
0.74
2.28
Year
s sin
ce la
st e
lect
ion
VD
S1,
084
3.16
1.02
VC
diff
eren
t fro
m p
revi
ous t
erm
*V
DS
4,31
20.
160.
361s
t Ele
ctio
n C
hang
ed V
C*
VD
S18
20.
380.
49
Not
es: E
ach
obse
rvat
ion
is a
t the
vill
age-
year
leve
l. V
DS
indi
cate
s tha
t the
var
iabl
e is
repo
rted
by th
e Vi
llage
Dem
ocra
cy
Surv
ey. N
FS in
dica
tes t
hat t
he v
aria
bles
are
repo
rted
by th
e N
atio
nal F
ixed
Poi
nt S
urve
y. *
Not
all
villa
ges r
etai
ned
reco
rds o
f V
C's
nam
es fr
om p
rior t
o th
e fir
st e
lect
ion.
**T
he y
ear o
f the
firs
t ele
ctio
n in
a c
ount
y is
bas
ed o
n re
spon
dent
reca
ll.
36
Table 2: Correlates of the Timing of the Introduction of Elections(1) (2) (3)
Dependent Variable: Year of 1st Election Coef. Obs R-Square
Near City -0.097 217 0.009Presence of Family with a Family Tree (jiapu) -0.109 217 0.012Presence of Family with Ancestral Temple (citang) -0.140 217 0.019Population Share of largest Clan 0.198 217 0.039Distance to High School 0.155 217 0.024Total Number of Households -0.324 217 0.105
Total Public Goods Exp (10,000 RMB) -0.015 217 0.000
Income 90th Percentile (RMB) -0.126 217 0.016Income 50th Percentile (RMB) -0.083 217 0.007Income 10th Percentile (RMB) -0.16 217 0.026
Income Growth 90th Percentile -0.115 217 0.013Income Growth 50th Percentile -0.075 217 0.006Income Growth 10th Percentile -0.119 217 0.014
Land Leased Out to Enterprises (Mu=1/15 Hectare) -0.145 217 0.021Land used for Household Farming 0.136 217 0.018Total Village Arable Land -0.156 217 0.024
One Child Exemptions -0.222 217 0.049Upper Government Land Expropriation -0.26** 217 0.068
Notes: Standardized coefficients are presented in column (1). These estimates use a cross section of villages. The control variables are measured in the base year (defined as the first year that data are available). All variables are demeaned by province fixed effects. Each coefficient is estimated from a bivariate regression of the residuals of the year of the first election on the residuals of the stated control variable. * and ** indicates statistical significance at the 10% and 5% levels.
37
Table 3: The Effect of the Introduction of Elections on Village Government Public GoodsExpenditure
(1) (2) (3)Obs. Mean Std. Dev.
Total 4,340 14.28 135.47 Irrigation 4,340 3.43 66.13 Schooling 4,340 0.02 0.35 Roads & Sanitation 4,340 5.07 88.21 Electricity 4,340 0.73 7.85 Environment 4,340 0.32 12.43 Other 4,340 2.72 49.70
From All Sources From Villagers From Other SourcesDependent Variable Mean 14.28 9.769 4.422Dependent Variable Mean 1986 6.248 3.649 2.599
Post 1st Election 15.45 16.37 -0.782(7.144) (7.705) (1.615)
Wild Bootstrap [0.018] [0.036] [0.734]
Observations 4,340 4,340 4,340R-squared 0.115 0.105 0.072Clusters 29 29 29
Dependent Variable: Public Goods Expenditures (10,000 RMB)
Notes: All regressions in Panel B control for the introduction of open nominations, province trends, village and year fixed effects. Standard errors, clustered at the province level, are presented in parentheses. Wild bootstrap p-values are presented in square brackets. The number of clusters is stated at the bottom of the table. The sample is a balanced village-level panel for the years 1986-2005.
B. Regression Results
A. Public Expenditure Categories (10,000 RMB)
38
Tab
le4:
The
Effe
ctof
theIntrod
uction
ofElections
onVillag
eLa
ndan
dtheHou
seho
ldIncome
Vill
age
Land
Le
ased
to
Ente
rpris
es
(Mu=
1/15
Hec
tare
)10
th25
th50
th75
th90
th
Inco
me
Rat
io
50th/9
0th
(1)
(2)
(3)
(4)
(5)
(6)
(7)
Dep
. Var
. Mea
n 10
8.5
5,08
37,
378
10,5
2715
,337
24,4
760.
529
Post
1st
Ele
ctio
n-5
7.72
724
4.7
130.
9-1
52.1
-1,0
21-5
,056
0.01
67(3
2.47
2)(1
96.4
)(3
18.4
)(5
87.9
)(1
,230
)(2
,996
)(0
.007
)W
ild B
oots
trap
p-va
lue
[0.0
70]
[0.2
12]
[0.6
94]
[0.8
54]
[0.4
18]
[0.0
88]
[0.0
24]
Obs
erva
tions
2,03
43,
763
3,76
33,
763
3,76
33,
763
3,76
3R
20.
558
0.77
20.
822
0.82
90.
813
0.71
00.
597
Clu
ster
s28
2929
2929
2929
Perc
entil
es o
f the
With
in-V
illag
e D
istri
butio
n of
Tot
al
Hou
seho
ld In
com
e (R
MB
)
Dep
ende
nt V
aria
ble
Not
es: A
ll re
gres
sion
s con
trol f
or th
e in
trodu
ctio
n of
ope
n no
min
atio
ns, p
rovi
nce
trend
s, vi
llage
and
yea
r fix
ed e
ffect
s. St
anda
rd e
rror
s, cl
uste
red
at th
e pr
ovin
ce le
vel,
are
pres
ente
d in
par
enth
eses
. Wild
boo
tstra
p p-
valu
es a
re p
rese
nted
in
squa
re b
rack
ets.
The
num
ber o
f clu
ster
s are
stat
ed a
t the
bot
tom
of t
he ta
ble.
The
sam
ples
are
vill
age
leve
l pan
els.
The
sam
ple
in c
olum
n (1
) is r
estri
cted
to v
illag
es th
at e
ver l
ease
d la
nd to
ent
erpr
ises
prio
r to
the
first
ele
ctio
n an
d in
clud
es th
e ye
ars 1
987-
2005
, exc
ept f
or 1
992
and
1994
. The
dep
ende
nt v
aria
ble
is th
e am
ount
of l
and
leas
ed o
ut in
yea
r t+1
(i.e
., a
lead
va
riabl
e). C
olum
ns (2
)-(7
) inc
lude
all
villa
ges a
nd th
e ye
ars 1
986-
2005
, exc
ept f
or 1
992
and
1994
. The
pan
els a
re n
ot fu
lly
bala
nced
bec
ause
of m
issi
ng v
alue
s in
the
depe
nden
t var
iabe
s.
39
Table 5: The Effect of the Introduction of Elections on Village Government Public GoodsExpenditure, Village Land and Household Income – Robustness to Controls
(1) (2) (3) (4) (5) (6) (7) (8)
Baseline
Control for Prov
Introduction of 1st Election
Control for Prov-Year
FE
Control for Prov Per
Capita GDP and Growth
Omit Early and Late
Introducers
Control for Year FE x Base Year Vars (One
Child Policy and Upper Government Land Expropriation)
Control for Year FE x Near City, Social Capital, Post Tax & Fee
Omit if Ever Merged with
Another Village
Post 1st Election 15.455 13.555 14.438 17.942 14.567 15.836 16.020 15.406(7.144) (6.059) (9.049) (7.853) (6.880) (7.016) (7.537) (7.858)
Wild Bootstrap [0.018] [0.018] [0.108] [0.014] [0.012] [0.048] [0.028] [0.078]
Observations 4,340 4,340 4,340 4,018 4,100 4,340 4,340 3,500R-squared 0.115 0.116 0.215 0.136 0.128 0.213 0.130 0.115Clusters 29 29 29 29 29 29 29 29
Post 1st Election -57.727 -55.563 -49.482 -54.178 -57.211 -50.189 -58.631 -52.162(32.472) (32.898) (30.816) (30.785) (33.405) (39.526) (31.474) (37.345)
Wild Bootstrap [0.070] [0.076] [0.082] [0.088] [0.058] [0.140] [0.084] [0.190]
Observations 2,034 2,034 2,034 1,919 1,940 2,034 2,034 1,740R-squared 0.558 0.559 0.744 0.591 0.559 0.564 0.578 0.614Clusters 28 28 28 28 28 28 28 27
Post 1st Election -5,056 -5,087 -4,738 -4,424 -3,152 -5,032 -4,764 -5,667(2,996) (3,051) (3,258) (3,047) (2,447) (2,817) (2,728) (3,587)
Wild Bootstrap [0.088] [0.088] [0.114] [0.154] [0.208] [0.178] [0.070] [0.096]
Observations 3,763 3,763 3,763 3,477 3,553 3,763 3,763 3,035R-squared 0.710 0.710 0.757 0.715 0.719 0.711 0.718 0.696Clusters 29 29 29 29 29 29 29 29
Notes: All regressions control for the introduction of open nominations, province trends, village and year fixed effects, and the additional controls stated in the column headings. Standard errors, clustered at the province level, are presented in parentheses. Wild bootstrap p-values are presented in square brackets. The number of clusters are stated at the bottom of the table. The samples are village level panels. Panel A includes the years 1986-2005. Panel B is restricted to villages that ever leased land to enterprises prior to the first election and includes the years 1987-2004, except for 1992 and 1994. The dependent variable is the amount of land leased out in year t+1 (i.e., a lead variable). Panel C includes all villages and the years 1986-2005, except for 1992 and 1994. Additional sample exclusions are stated in column headings. The panels are not fully balanced because of missing values in the dependent variabes.
A. Public Goods Expenditure from Villagers (10,000 RMB)
B. Village Land Leased to Enterprises (Mu=1/15 Hectare)
C. Household Income at the 90th Percentile (RMB)
Dependent Variable
40
Table 6: The Effect of the Introduction of Elections on the Characteristics of the VC andPS
Turnover Age Years of Edu Party Member(1) (2) (3) (4)
Dep. Var Mean 0.158 42.15 7.883 0.765
Post 1st Election 0.045 -2.442 0.791 -0.034(0.020) (1.037) (0.224) (0.043)
Wild Bootstrap p-value [0.001] [0.040] [0.000] [0.468]
Obs 4,312 4,188 4,194 4,274R2 0.065 0.430 0.611 0.484
Dep. Var Mean 0.151 44.35 8.417
Post 1st Election 0.011 0.312 -0.121(0.021) (0.632) (0.139)
Wild Bootstrap p-value [0.199] [0.607] [0.478]
Obs 4,365 3,546 4,356R2 0.072 0.677 0.572
Dependent VariablesA. Village Chairman (VC)
B. Party Secretary (PS)
Notes: All regressions control for post-open nomination, province trends, and village and year fixed effects. Standard errors, clustered at the province level, are reported in parentheses. Wild bootstrap p-values, clustered at the province level are presented in square brackets. The sample is a panel of villages for the years 1982-2005. The number of observations vary across columns due to missing values in the dependent variable.
41
Tab
le7:
The
Effe
ctof
theIntrod
uction
ofElections
forVillages
whe
rethePre-E
lectionIncu
mbe
ntRem
ainedin
Office
(1)
(2)
(3)
Publ
ic G
oods
Exp
endi
ture
fro
m V
illag
ers (
10,0
00
RMB)
Villa
ge L
and
Leas
ed
to E
nter
prise
s (M
u=1/
15 H
ecta
re)
Hou
seho
ld In
com
e 90
th P
erce
ntile
(RM
B)
Post
1st E
lect
ion
16.3
7-5
7.73
-5,0
56(7
.705
)(3
2.47
)(2
,996
)
Wild
Boo
tstra
p p-
valu
e[0
.036
][0
.070
][0
.088
]
Obs
erva
tions
4,34
02,
034
3,76
3R-
squa
red
0.10
50.
558
0.71
0Cl
uste
rs29
2829
Post
1st E
lect
ion
55.9
7-1
9.33
-2,7
24(2
9.44
)(3
6.75
)(2
,484
)
Wild
Boo
tstra
p p-
valu
e[0
.043
][0
.626
][0
.240
]
Post
1st E
lect
ion
x 1s
t Ele
ctio
n V
C Ch
ange
-4
2.08
-39.
85-2
,484
(29.
16)
(20.
05)
(2,6
64)
Wild
Boo
tstra
p p-
valu
e[0
.396
][0
.434
][0
.506
]
Obs
erva
tions
4,34
02,
034
3,76
3R2
0.11
20.
558
0.71
0Cl
uste
rs29
2829
Not
es:A
llre
gres
sions
cont
rolf
orth
ein
trodu
ctio
nof
open
nom
inat
ions
,pro
vinc
etre
nds,
villa
gean
dye
arfix
edef
fect
s.Th
ose
inPa
nelB
addi
tiona
llyco
ntro
lfor
the
inte
ract
ion
ofop
enno
min
atio
nsw
ithth
efir
stel
ectio
nV
Ctu
rnov
erdu
mm
yva
riabl
e.St
anda
rder
rors
,cl
uste
red
atth
epr
ovin
cele
vel,
are
pres
ente
din
pare
nthe
ses.
Wild
boot
strap
p-va
lues
are
pres
ente
din
squa
rebr
acke
ts.Th
enu
mbe
rofc
luste
rsar
esta
ted
atth
ebo
ttom
ofth
eta
ble.
The
sam
ples
are
villa
ge-le
vel
pane
ls.Co
lum
n(1
)inc
lude
sth
eye
ars
1986
-200
5.Co
lum
n(2
)is
restr
icte
dto
villa
ges
that
ever
leas
edla
ndto
ente
rpris
espr
ior
toth
efir
stel
ectio
nan
din
clud
esth
eye
ars
1987
-200
5,ex
cept
for
1992
and
1994
.The
depe
nden
tvar
iabl
eis
the
amou
ntof
land
leas
edou
tin
year
t+1
(i.e.
,ale
adva
riabl
e).C
olum
n(3
)inc
lude
sth
eye
ars
1986
-200
5,ex
cept
for1
992
and
1994
.The
pane
lsus
edin
colu
mns
(2)
and
(3)
are
notf
ully
bala
nced
beca
use
ofm
issin
gva
lues
inth
ede
pend
ent
varia
bles
.
A. B
asel
ine
B. In
tera
cted
with
Whe
ther
ther
e w
as V
C Tu
rnov
er in
the
1st E
lect
ion
Dep
ende
nt V
aria
ble
42
Tab
le8:
The
Effe
ctof
theIntrod
uction
ofElections
onHou
seho
ldPay
mentof
LocalF
ees
0 - 2
0th
20th
- 40
th40
th -
60th
60th
- 80
th80
th -
100t
h(1
)(2
)(3
)(4
)(4
)
Mea
n3.
141
2.74
62.
476
2.17
31.
749
St. D
ev.
3.34
02.
782
2.53
02.
255
1.96
6
Mea
n2.
675
2.44
12.
247
2.14
91.
931
St. D
ev.
2.32
92.
137
1.96
91.
917
1.79
8
Post
1st
Ele
ctio
n0.
522
0.39
10.
406
0.38
80.
315
(0.2
46)
(0.2
14)
(0.2
06)
(0.1
90)
(0.1
31)
Wild
Boo
tstra
ps[0
.072
][0
.094
][0
.088
][0
.076
][0
.036
]
Obs
erva
tions
2,79
22,
792
2,79
22,
792
2,79
1R
20.
542
0.59
80.
609
0.62
50.
583
Clu
ster
s18
1818
1818
Not
es: A
ll re
gres
sion
s con
trol f
or th
e in
trodu
ctio
n of
ope
n no
min
atio
ns, p
rovi
nce
trend
s, vi
llage
and
yea
r fix
ed e
ffect
s. St
anda
rd e
rror
s, cl
uste
red
at th
e pr
ovin
ce le
vel,
are
pres
ente
d in
par
enth
eses
. The
num
ber o
f clu
ster
s are
stat
ed a
t the
bot
tom
of t
he ta
ble.
The
sam
ples
are
vi
llage
-leve
l pan
els a
nd in
clud
e 78
vill
age
for t
he y
ears
198
7-20
05. D
ata
for 1
992
and
1994
are
inte
rpol
ated
. The
pan
el is
not
fully
ba
lanc
ed b
ecau
se o
f mis
sing
val
ues i
n th
e de
pend
ent v
aria
bles
.
Pane
l A. M
ean
Hou
seho
ld A
nnua
l Loc
al T
ax P
aym
ents
(% o
f Ann
ual H
ouse
hold
Inco
me)
-- A
ll Ye
ars
Pane
l B. M
ean
Hou
seho
ld L
ocal
Tax
Pay
men
ts (
% o
f Hou
seho
ld In
com
e in
198
6) --
In B
asel
ine
Year
(198
6)
Pane
l C. T
he E
ffect
of t
he In
trodu
ctio
n of
Ele
ctio
ns (A
ll Ye
ars)
Dep
ende
nt V
aria
ble:
Hou
seho
ld A
nnua
l Loc
al T
ax P
aym
ents
(% o
f Hou
seho
ld A
nnua
l Inc
ome)
Tota
l Hou
seho
ld In
com
e Q
uint
iles
43
Tab
le9:
The
Effe
ctof
theIntrod
uction
ofElections
onHou
seho
ldIncomean
dFa
rmland
According
toPre-electionPositions
ontheVillag
eDistribution
Tota
lA
gric
ultu
reW
age
Ente
rpris
es(1
)(2
)(3
)(4
)(5
)
Dep
. Var
. Mea
n0.
527
0.46
30.
145
0.08
410.
591
Dep
. Var
. Mea
n 19
860.
538
0.49
80.
0325
0.09
860.
608
Post
1st
Ele
ctio
n0.
016
0.02
10.
019
0.02
60.
013
(0.0
08)
(0.0
07)
(0.0
17)
(0.0
13)
(0.0
17)
Wild
Boo
tstra
p p-
valu
e[0
.090
][0
.018
][0
.284
][0
.016
][0
.526
]
Obs
2,79
22,
792
2,56
42,
257
2,76
0R
20.
568
0.72
30.
651
0.65
80.
633
Clu
ster
s18
1818
1818
Not
es: A
ll re
gres
sion
s con
trol f
or th
e in
trodu
ctio
n of
ope
n no
min
atio
ns, p
rovi
nce
trend
s, vi
llage
and
ye
ar fi
xed
effe
cts.
Stan
dard
err
ors,
clus
tere
d at
the
prov
ince
leve
l, ar
e pr
esen
ted
in p
aren
thes
es.
Wild
boo
tstra
p p-
valu
es a
re p
rese
nted
in sq
uare
bra
cket
s. Th
e nu
mbe
r of c
lust
ers a
re st
ated
at t
he
botto
m o
f the
tabl
e. T
he sa
mpl
es a
re v
illag
e-le
vel p
anel
s and
incl
ude
78 v
illag
e fo
r the
yea
rs 1
987-
2005
. Dat
a fo
r 199
2 an
d 19
94 a
re in
terp
olat
ed. T
he p
anel
is n
ot fu
lly b
alan
ced
beca
use
of m
issi
ng
valu
es in
the
depe
nden
t var
iabl
es. T
he n
umbe
r of o
bser
vatio
ns v
ary
acro
ss c
olum
ns d
ue to
mis
sing
va
lues
repo
rted
for t
he d
epen
dent
var
iabl
es.
Dep
ende
nt V
aria
ble
50th
/90t
h In
com
e R
atio
50th
/90t
h La
nd R
atio
44
Figure 1: The the Effect of the Introduction of Elections for Each Year Since the FirstElection
(a) Public Goods Expenditure (10,000RMB = 1,428 USD)
-20
020
4060
Coe
f of D
umm
y Va
rs fo
r Yea
rs S
ince
1st
Ele
ctio
n
-5 0 5Years Since 1st Election
(b) Village Arable Land that is Leased Outto Enterprises (Mu = 1/15 Hectare)
-80
-60
-40
-20
020
Coe
f of D
umm
y Va
rs fo
r Yea
rs S
ince
1st
Ele
ctio
n
-5 0 5Years Since 1st Election
(c) 90th Percentile Household Income(RMB = 0.14 USD)
-600
0-4
000
-200
00
2000
Coe
f of D
umm
y Va
rs fo
r Yea
rs S
ince
1st
Ele
ctio
n
-5 0 5Years Since 1st Election
Notes: These figures plot the coefficients for dummy variables for the number of years sincethe first election. The estimates control for the introduction of open nominations, province-year trends and village fixed effects. The samples are balanced panels of villages from fiveyears prior until five years after the first election. The reference group, indicated as -4 yearson the x-axis, comprises of five and four years prior to the first election; we plot a y-axisvalue of zero for this group. The coefficients and standard errors are reported in OnlineAppendix Table A.4.
45
A ONLINE APPENDIX
A.1 Data Appendix
The variables for upper-government land expropriation and One Child Policy exemptions
are from the VDS. To ensure truthful reporting, we ask the respondents to check the years
in which the village experienced any permanent loss of village land due to upper-government
expropriation, and to check the years in which any household in the village had a second or
higher parity birth. Respondents’ answers are based on village records and rosters. In a com-
panion paper, we check that changes in expropriation correspond to changes in total village
land (reported in the NFS) and changes in One Child Policy exemptions correspond to the
number of children age 7-13 seven years later (reported in the NFS). The questionnaire for the
VDS is available online at http://www.econ.yale.edu/~nq3/NANCYS_Yale_Website/styled-
4/styled-5/index.html.
A.2 Demand for Public Goods
According to our interpretation of the effect of elections, the public goods funded by
the increase in expenditures should correspond to the demands of a majority of villagers.
To determine whether this is the case, we perform an additional exercise where we proxy
for villager demand. Within the constraints of our data, we are able to examine irrigation
expenditure, which is the single largest item of public goods provision comprising 24% of
average expenditure. We proxy for the demand for irrigation expenditure with the share of
total village land that is arable in 1987 (the first year that land data are available) with the
logic that villagers with very little arable land will have high demand for irrigation, where
as villagers with access to a lot of arable land will have less demand for irrigation. In the
extreme case where a village is fully irrigated, villagers may even want to reduce the amount
of irrigation in order to have land for other purposes. Following this logic, we estimate
an equation that is similar to the baseline specification, equation (1) with the following
modifications. First, the dependent variable is a dummy variable for whether any public
expenditure on irrigation is made in a given year. Second, we create a dummy variable for
1
whether the share of arable land of a village in 1987 is below or above the sample median
and interact this dummy variable with the introduction of elections (and open nominations).
To ensure that differences in arable land share in 1987 do not influence the outcome through
channels other than the introduction of elections, we also control for the 1987 arable land
share interacted with year fixed effects.59
Column (1) in Table A.3 shows the effect of elections on irrigation investment. The
positive uninteracted coefficient shows that elections increased expenditure on irrigation for
villages that had arable share below the sample median in 1987. The estimate is statisti-
cally significant at the 1% level. The negative interaction coefficient suggests that elections
reduced expenditure in irrigation for villages with above median arable share land in 1987.
The estimate is statistically insignificant.
In column (3), we investigate the effect on schooling, which mostly refers to primary
schools. We proxy for the demand for schools with the number of children age zero to six
in 1986 (the first year that are available for children). Our logic is that villages with many
young children will demand more public schools than villages with few children. Using a
similar equation as in column (1), except that we substitute the demand proxies, we find
that the uninteracted coefficient is negative. This suggests that elections reduced public
expenditure on schooling in villages where the number of young children was below the
sample median in 1986. The interaction effect is positive, which suggests that elections
increased expenditures in public schooling in villages where there were many young children
in 1986. However, neither estimate is statistically significant.
Finally, we investigate the effect on public goods provision. Since our outcome variable
is obtained from village accounts, it is, in principle, possible that village leaders changed the
numbers they recorded without actually raising funds or increasing provision. Alternatively,59The variables for village land used for household farming is reported by the NFS. The estimating equation
can be written as
Yvpt = θEvpt +ζ Ovpt +β (Evpt ×Xvp)+λ (Ovpt ×Xvp)+θtXvp + γpt +δv +ρt + εvpt , (2)
where Xvp is a measure of the share of arable village land in 1987. The main effect of this time-invariantvariable is absorbed by village fixed effects. θ̂ is the effect of elections on villages with no arable land in1987. β̂ + θ̂x is the effect of elections for villages where the share of arable land in 1987 equals x.
2
village leaders may have raised funds but stolen them and then fabricated the numbers for
expenditure in the village accounts. The ideal way to check the relevance of these concerns
for our context is to investigate whether elections actually increase the provision of public
goods. We cannot do this due to data limitations. Instead, we are able to proxy for the
provision of irrigation with the share of arable land in a village. This follows the premise that
increased irrigation will likely increase the share of land that is arable. We can also proxy
for expenditure on schooling with the presence of a primary school, the data for which we
collected in the VDS. Columns (2) and (4) show that the signs of the estimates are consistent
with those for expenditures. However, they are imprecise. The estimates for arable land in
columns (2) are only significant at the 20% and 15% levels. The estimates for the primary
school dummy is significant at the 5% and 1% levels, but becomes insignificant with the
wild bootstraps.
3
Table A.1: The Timing of Electoral Reforms
Number of Villages Introducing
Cumulative % of Villages
Number of Villages Introducing
Cumulative % of Villages
Year (1) (2) (3) (4)
1982 13 5.99 1 0.721983 13 11.98 1 1.441984 42 31.34 7 6.471985 3 32.72 0 6.471986 35 48.85 4 9.351987 12 54.38 1 10.071988 7 57.6 1 10.791989 15 64.52 1 11.511990 25 76.04 1 12.231991 1 76.5 0 12.231992 3 77.88 1 12.951993 6 80.65 3 15.111994 2 81.57 3 17.271995 9 85.71 3 19.421996 4 87.56 18 32.371997 3 88.94 0 32.371998 6 91.71 6 36.691999 9 95.85 42 66.912000 7 99.08 12 75.542001 2 100 12 84.172002 0 100 11 92.092003 0 100 3 94.242004 0 100 1 94.962005 0 100 7 100
Total 217 139
First Open Nominations (Haixuan)First Election
Notes: Each observation is a village. The sample comprises a balanced panel of 217 villages for 1982 to 2005.
4
Tab
leA.2:Descriptive
Statistics
forHou
seho
ldIncomes
intheFu
llSa
mplean
dtheHou
seho
ldSu
bsam
pleof
Villag
es
Varia
ble
Obs
Mea
nSt
d. D
ev.
Obs
Mea
nSt
d. D
ev.
# of
HH
in V
illag
e3,
641
421.
8128
0.81
2,50
742
0.12
288.
48N
ear C
ity5,
208
0.30
0.46
2,79
80.
270.
44
Tota
l Vill
age
Land
3,61
29,
245
14,7
192,
495
7,22
410
,119
Ara
ble
Land
(Mu)
3,61
22,
295
2,32
92,
495
1,78
21,
188
Sha
re o
f Vill
age
Land
that
is A
rabl
e3,
612
0.51
0.32
2,49
50.
520.
31 U
sed
for H
H F
arm
ing
(Mu)
3,61
22,
215
2,31
22,
495
1,70
51,
173
Not
Use
d fo
r HH
Far
min
g (M
u)3,
612
79.7
236
7.26
2,49
577
.25
351.
67
Lea
sed
Out
to E
nter
pris
es (M
u)3,
612
60.4
634
7.61
2,65
561
.85
327.
37
Med
ian
HH
Ann
ual G
ross
Inco
me
Gro
wth
3,08
40.
069
0.18
92,
145
0.06
80.
185
The
Num
ber o
f Vill
age
Com
mitt
ee M
embe
rs
2,28
74.
362.
361,
583
4.22
2.08
The
Num
ber o
f Vill
age
Party
Cad
res
2,29
56.
703.
821,
587
6.98
3.84
Party
Sec
reta
ry T
enur
e5,
208
10.0
38.
132,
798
10.7
17.
82V
illag
e C
hief
: Ten
ure
5,20
86.
696.
242,
798
7.38
5.89
Has
Ele
ctio
n52
080.
730.
4427
980.
840.
36H
as O
pen
Nom
inat
ion
5208
0.20
0.40
2798
0.24
0.43
Year
s bet
wee
n El
ectio
n In
trodu
ctio
ns in
Vill
age
and
Prov
ince
217
5.02
5.07
2798
5.10
4.98
Year
s bet
wee
n El
ectio
n In
trodu
ctio
ns in
Cou
nty
and
Prov
ince
**21
74.
284.
6727
984.
404.
63Ye
ars b
etw
een
Ele
ctio
n In
trodu
ctio
ns in
Vill
age
and
Cou
nty*
*21
70.
742.
2827
980.
702.
25Ye
ars s
ince
last
ele
ctio
n1,
084
3.16
1.02
717
3.19
1.07
VC
diff
eren
t fro
m p
revi
ous t
erm
*4,
312
0.16
0.36
2,50
50.
160.
361s
t Ele
ctio
n C
hang
ed V
C*
182
0.38
0.49
2,50
50.
160.
36
Full
Sam
ple
Hou
seho
ld S
ampl
e
Not
es: E
ach
obse
rvat
ion
is a
t the
vill
age-
year
leve
l. V
DS
indi
cate
s tha
t the
var
iabl
e is
repo
rted
by th
e Vi
llage
Dem
ocra
cy S
urve
y. N
FS in
dica
tes t
hat t
he
varia
bles
are
repo
rted
by th
e N
atio
nal F
ixed
Poi
nt S
urve
y. *
Not
all
villa
ges r
etai
ned
reco
rds o
f VC
's na
mes
from
prio
r to
the
first
ele
ctio
n. *
*The
yea
r of
the
first
ele
ctio
n in
a c
ount
y is
bas
ed o
n re
spon
dent
reca
ll.
5
Table A.3: The Effect of Elections on Public Expenditure on Irrigation and Schooling
(1) (2) (3) (4)
Any Expenditure
(Dummy Var)
Arable Land Share
(Fraction)
Any Expenditure
(Dummy Var)
The Village has a Primary
School (Dummy Var)
Dependent Variable Mean 0.0504 0.513 0.0541 0.885
Post 1st Election 0.0661 0.0433 -0.0281 -0.0415(0.0182) (0.0308) (0.0185) (0.0196)
Wild Bootstrap [0.002] [0.208] [0.948] [0.940]Post 1st Election x High Base Year Arable Land Share -0.0435
(0.0338)
Wild Bootstrap [0.218]
Post 1st Election x Base Year Arable Land Share -0.118(0.0734)
Wild Bootstrap [0.118]Post 1st Election x High Base Year # Kids 0.0357
(0.0259)Wild Bootstrap [0.958]
Post 1st Election x Base Year # Kids 0.000105(0.000006)
Wild Bootstrap [0.986]
Observations 3316 3316 4340 4279R-squared 0.139 0.920 0.122 0.796Clusters 28 28 29 27
Dependent Variable: Public Expenditure onIrrigation Schools
Notes: The controls for arable land share and the number of kids are measured in the base year (the first year that data are available, 1986 for kids, 1987 for land). The explanatory interaction terms in columns (1) and (3) use dummy variables that indicate whether a village is above the median in terms of arable land share and the number of children, respectively. All regressions control for the introduction of open nominations, province trends, village and year fixed effects, as well as the interaction terms of post open nominations and the relevant base year variable. In addition, columns (1) and (2) control for the interaction of year fixed effects and the relevant base year variable. Standard errors, clustered at the province level, are presented in parentheses. Wild bootstrap p-values are presented in square brackets. The number of clusters are stated at the bottom of the table. The samples are village-level panels. Columns (3)-(4) include the years 1986-2005. Columns (1)-(2) include the years 1986-2005, except for the years 1992 and 1994. The panels are not fully balanced because of missing values in the dependent variabes.
6
Tab
leA.4:The
Effe
ctof
theIntrod
uction
ofElections
forEachYearSinc
etheFirst
Election
(1)
(2)
(3)
(4)
(5)
(6)
coef
seco
efse
coef
seD
umm
y Va
r for
# o
f Yea
rs si
nce
Firs
t Ele
ctio
n-4
0.
0.
0.
-33.
106
(10.
43)
238.
5(5
87.6
)1.
017
(47.
03)
-22.
182
(13.
99)
-298
.9(1
,592
)9.
961
(69.
48)
-1-3
.034
(17.
97)
-63.
46(2
,124
)-1
9.18
(85.
18)
0 (1
st E
lect
ion)
4.35
4(2
4.19
)1,
632
(3,5
53)
14.2
1(1
14.8
)1
65.0
6(7
2.51
)53
5.4
(3,5
91)
-59.
18(1
45.5
)2
-19.
04(3
8.28
)-2
,130
(4,5
63)
-72.
32(1
82.8
)3
15.6
4(2
4.93
)-2
,229
(4,9
76)
-47.
50(2
05.6
)4
-18.
64(7
2.22
)-3
11.6
(6,6
05)
-52.
90(2
22.4
)5
7.78
1(7
5.90
)-5
,885
(8,0
06)
-20.
24(2
62.8
)
Obs R2
Clu
ster
sJo
int p
-val
ue (P
re-E
lect
ion=
0)
Publ
ic G
oods
Exp
(R
MB
)
550
0.14
7
Not
es: A
ll re
gres
sion
s con
trol v
illag
e an
d ca
lend
ar y
ear f
ixed
effe
cts.T
he re
fere
nce
year
is fo
ur y
ears
prio
r to
the
first
ele
ctio
n. S
tand
ard
erro
rs a
re c
lust
ered
at t
he p
rovi
nce
leve
l. Th
e nu
mbe
r of c
lust
ers a
nd th
e p-
valu
e fo
r the
jo
int s
igni
fican
ce o
f the
pre
-ref
orm
yea
rs a
re p
rese
nted
at t
he b
otto
m o
f the
tabl
e. C
olum
ns (3
)-(4
) add
ition
ally
re
stric
t the
sam
ple
to v
illag
es th
at e
ver l
ease
d la
nd to
ent
erpr
ises
prio
r to
the
first
ele
ctio
n. T
he sa
mpl
es c
ompr
ise
of b
alan
ced
pane
ls o
f vill
ages
that
incl
ude
10 y
ears
: fro
m fi
ve y
ears
bef
ore
until
five
yea
rs a
fter t
he fi
rst
elec
tion.
The
refe
renc
e ye
ar c
ompr
ises
of f
ive
(-5)
and
four
(-4)
yea
rs b
efor
e th
e fir
st e
lect
ion.
Land
Lea
sed
out t
o En
terp
rises
(Mu)
90th
Per
cent
ile H
ouse
hold
In
com
e (R
MB
)
Dep
ende
nt V
aria
ble
0.48
20.
435
0.18
4
0.83550
466
0.82
516
1816
7
Table A.5: Balance Sheet of Household Income and Expenditure
Obs Mean Std. Dev.Total Income 2,798 13,161 15,338 from agriculture (and home production) 2,798 8,853 11,140 from wages 2,798 2,090 3,280 village enterprise income 2,798 902 2,245 other 2,798 1,316 3,081
Total Expenditures 2,798 6,176 13,629 Household management expenditures 2,798 1,963 6,639 Taxes to central government 2,798 128 386 Levies and fees to local governments 2,798 82 162 Total Consumption 2,798 3,424 6,645
Notes: Village enterprise income is the sum of income from collectives, dividends and parternship/ cooperatives. The sample comprises of an unbalanced panel of 160 villages in 80 provinces for the years for the years 1986-2005. Data for 1992 and 1994 are interpolated as the average of data from one year prior and one year afterwards.
8
Table A.6: The Effect of the Introduction of Elections on Household Income – Comparisonof Full Sample and Village Subsample Estimates
Percentiles of the Village Income Distribution: 10th 25th 50th 75th 90th
(1) (2) (3) (4) (5)
Dependent Variable Mean 5,083 7,378 10,527 15,337 24,476
Post 1st Election 245 131 -152 -1,021 -5,056(196) (318) (588) (1,230) (2,996)
Wild Bootstrap [0.212] [0.694] [0.854] [0.418] [0.088]
Observations 3,763 3,763 3,763 3,763 3,763R-squared 0.772 0.822 0.829 0.813 0.71Clusters (# Provinces) 29 29 29 29 29# Villages 217 217 217 217 217
Dependent Variable Mean 4,454 6,451 9,162 13,250 20,843
Post 1st Election -121 -262 -655 -1,537 -4,631(268) (465) (895) (1,718) (4,060)
Wild Bootstrap [0.680] [0.620] [0.456] [0.440] [0.320]
Observations 1,347 1,347 1,347 1,347 1,347R-squared 0.764 0.826 0.820 0.793 0.733Clusters (# Provinces) 17 17 17 17 17# Villages 78 78 78 78 78
B. Household Level Subsample
A. Full Sample
Dependent Variable: Household Income (RMB)
Notes: All regressions control for the introduction of open nominations, province trends, village and year fixed effects. Standard errors, clustered at the province level, are presented in parentheses. Wild bootstrap p-values are presented in square brackets. The number of clusters are stated at the bottom of the table. The samples are village-level panels and include the years 1987-2005.In Panel B, data for 1992 and 1994 are interpolated. The panels are not fully balanced because of missing values in the dependent variabes. The number of villages are stated at the bottom of each panel.
9
Tab
leA.7:The
Effe
ctof
theIntrod
uction
ofElections
onHou
seho
ldIncomean
dFa
rmland
Ratios–Rob
ustnessto
Con
trols
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Bas
elin
eC
ontro
l for
Pro
v In
trodu
ctio
n of
1s
t Ele
ctio
n
Con
trol f
or P
rov-
Year
FE
Con
trol f
or P
rov
Per C
apita
GD
P an
d G
row
th
Om
it Ea
rly a
nd
Late
Intro
duce
rs
Con
trol f
or Y
ear
FE x
Bas
e Ye
ar
Vars
(One
Chi
ld
Polic
y an
d U
pper
G
over
nmen
t La
nd
Expr
opria
tion)
Con
trol f
or Y
ear
FE x
Nea
r City
, So
cial
Cap
ital,
Post
Tax
& F
ee
Om
it if
Ever
M
erge
d w
ith
Ano
ther
Vill
age
Post
1st
Ele
ctio
n0.
016
0.01
50.
015
0.01
80.
010
0.01
70.
017
0.01
0(0
.008
)(0
.008
)(0
.011
)(0
.009
)(0
.007
)(0
.008
)(0
.008
)(0
.006
)
Obs
erva
tions
2,79
22,
792
2,79
22,
582
2,61
62,
792
2,79
22,
215
R2
0.56
90.
569
0.61
30.
583
0.57
60.
577
0.57
90.
567
Clu
ster
s18
1818
1818
1818
18
Post
1st
Ele
ctio
n0.
021
0.02
00.
021
0.02
30.
016
0.02
40.
020
0.02
2(0
.007
)(0
.007
)(0
.010
)(0
.009
)(0
.007
)(0
.006
)(0
.008
)(0
.007
)
Obs
erva
tions
2,79
22,
792
2,79
22,
582
2,61
62,
792
2,79
22,
215
R2
0.72
40.
724
0.74
60.
730
0.72
00.
731
0.72
80.
741
Clu
ster
s18
1818
1818
1818
18
Post
1st
Ele
ctio
n0.
019
0.02
00.
029
0.01
30.
013
0.01
90.
019
-0.0
00(0
.017
)(0
.017
)(0
.022
)(0
.019
)(0
.019
)(0
.016
)(0
.015
)(0
.015
)
Obs
erva
tions
2,56
42,
564
2,56
42,
367
2,39
42,
564
2,56
42,
065
R2
0.65
10.
651
0.68
60.
679
0.64
80.
657
0.65
60.
657
Clu
ster
s18
1818
1818
1818
18
Post
1st
Ele
ctio
n0.
026
0.02
50.
020
0.02
40.
024
0.02
40.
022
0.02
2(0
.013
)(0
.013
)(0
.016
)(0
.013
)(0
.017
)(0
.014
)(0
.015
)(0
.012
)
Obs
erva
tions
2,25
72,
257
2,25
72,
065
2,08
92,
257
2,25
71,
811
R2
0.65
80.
660
0.69
90.
675
0.64
40.
664
0.66
80.
610
Clu
ster
s18
1818
1818
1818
18
Post
1st
Ele
ctio
n0.
013
0.01
30.
015
0.01
60.
008
0.01
10.
008
0.01
7(0
.017
)(0
.017
)(0
.018
)(0
.016
)(0
.018
)(0
.017
)(0
.014
)(0
.019
)
Obs
erva
tions
2,76
02,
760
2,76
02,
550
2,59
62,
760
2,76
02,
183
R2
0.63
30.
633
0.66
30.
641
0.64
50.
639
0.64
30.
642
Clu
ster
s18
1818
1818
1818
18
Notes: A
ll re
gres
sion
s con
trol f
or th
e in
trodu
ctio
n of
ope
n no
min
atio
ns, p
rovi
nce
trend
s, vi
llage
and
yea
r fix
ed e
ffect
s, an
d th
e ad
ditio
nal c
ontro
ls st
ated
in th
e co
lum
n he
adin
gs. S
tand
ard
erro
rs, c
lust
ered
at t
he p
rovi
nce
leve
l, ar
e pr
esen
ted
in p
aren
thes
es. W
ild b
oots
trap
p-va
lues
are
pre
sent
ed in
squa
re b
rack
ets.
The
nu
mbe
r of c
lust
ers a
re st
ated
at t
he b
otto
m o
f the
tabl
e. T
he sa
mpl
es a
re v
illag
e le
vel p
anel
s and
incl
ude
the
year
s 198
6 to
200
5, e
xcep
t for
199
2 an
d 19
94.
Add
ition
al sa
mpl
e ex
clus
ions
are
stat
ed in
col
umn
head
ings
. The
pan
els a
re n
ot fu
lly b
alan
ced
beca
use
of m
issi
ng v
alue
s in
the
depe
nden
t var
iabe
s.
A. D
epen
dent
Var
iabl
e: 5
0th/
90th
Tot
al In
com
e R
atio
B. D
epen
dent
Var
iabl
e: 5
0th/
90th
Agr
icul
tura
l Inc
ome
Rat
io
C. D
epen
dent
Var
iabl
e: 5
0th/
90th
Wag
e In
com
e R
atio
D. D
epen
dent
Var
iabl
e: 5
0th/
90th
Ent
erpr
ise
Inco
me
Rat
io
E. D
epen
dent
Var
iabl
e: 5
0th/
90th
Lan
d
10