Machine Politics for the Middle Class: Distributing Benefits
and Building Party Support in an Electoral Autocracy
Stan Hok-Wui Wong
Department of Applied Social Sciences
Hong Kong Polytechnic University
email: [email protected]
phone: +852 67737216
Karl Ho
School of Economic, Political and Policy Sciences
University of Texas, Dallas
email: [email protected]
Harold D. Clarke
School of Economic, Political and Policy Sciences
University of Texas, Dallas
email: [email protected]
Abstract
Many authoritarian regimes build political organizations to distribute spoils, buy votes,
and win elections. Existing studies argue that these "political machines" target poor
and illiterate voters because their votes are cheap to purchase with non-programmatic
benefits. Focusing on the case of Hong Kong we investigate how political machines
survive and thrive in a society where votes are frequently too expensive to purchase.
Using interviews with District Councilors and survey data gathered in the 2015 Hong
Kong Election Study, we find that: (1) political machines tend to specialize in the
provision of highly personalized services; (2) demand for these services frequently
comes from middle-class citizens; and (3) unable to monitor individual votes, political
machines use services and benefits to buy turnout among groups of likely supporters.
The growing strength of pro-Beijing parties in Hong Kong reflects their successful
efforts to respond to constituents' demands for a wide variety of non-programmatic
benefits.
Key Words: machine politics, non-programmatic benefits, Hong Kong elections,
buying turnout, pro-Beijing parties
1. Introduction
Scholars studying electoral autocracies generally agree that these regimes emphasize
the distribution of non-programmatic benefits to build to electoral support. Existing
research is largely based on evidence from developing countries where the votes of
poor citizens are inexpensive to purchase. However, machine politics exists in
developed countries as well, some of which are electoral autocracies with high levels
of per capita income. Monopolizing state resources, these regimes can purchase
political support by dispensing various non-programmatic benefits. This paper
provides a quantitative case study of political machines in one rich authoritarian
regime—post-1997 Hong Kong. Three key findings emerge: (1) basic consumable
goods do not impress middle class voters, many of whom are employed in the private
sector. For this reason, political machines become adept at providing highly
personalized services; (2) demand for personalized services tends to come from
relatively well-to-do citizens because government welfare programs are designed to
help poor people who have institutional channels that address their needs; (3) because
consumption of personalized services is not indispensable and ballots are secret, it is
difficult for political machines to make benefits contingent on electoral support.
Consequently, the distribution of non-programmatic benefits is geared less toward
changing voting preferences and more toward increasing voter turnout among likely
supporters.
Hong Kong is a special case because it is not a sovereign state. On the
other hand, Hong Kong is autonomous except in areas of diplomacy and national
security. The city maintains its own currency, judiciary, legislature, and even
constitution. Unlike other parts of the People's Republic of China, Hong Kong has
resilient pro-democracy opposition parties, which regularly compete for
legislative seats through popular elections. Although the elections lack a level
playing field, which is a common characteristic of electoral autocracies,
opposition parties have managed to win a sizable number of seats in every
election. In this regard, Hong Kong is no different from a typical electoral
authoritarian regime that allows for multi-party elections, while striving to keep
the opposition at bay. It is exactly this special but not uncommon context that
makes Hong Kong comparable with other electoral autocracies.
Research on Hong Kong politics demonstrates how semi-competitive elections
have created an uneven playing field which enables pro-Beijing parties to foster
1
clientelistic local networks to expand their political influence while undermining
support for pro-democracy competitors. Wong (2014, 2015) examines the “supply
side” of the clientelistic relationship; namely, pro-Beijing parties’ motivation and
strategies to develop political machines to consolidate electoral support. How voters
respond to these parties’ grassroots strategies (i.e., the “demand side”) remains poorly
understood. Here, we address this topic using data gathered in interviews with more
than thirty District Councilors, 1 numerous personal communications and a large
representative survey (the Hong Kong Election Study) that provides data on voting
behavior in the 2015 District Council election. The 2015 District Council election is
the first such contest held after the unprecedented, albeit unsuccessful, Occupy
Movement, and many commentators consider it a litmus test of the popularity of the
HKSAR government.
The paper is divided into four sections. We first provide an overview of
theories on machine politics and then discuss the Hong Kong case. Next, we present
empirical evidence based on analyses of the election survey data. The final section
summarizes key findings and discusses their implications.
2. Political Machines and Electoral Support
Studies of electoral authoritarianism observe that an important function of holding
semi-competitive elections is to provide an authoritarian regime with an institutional
mechanism to buy political support through the non-programmatic distribution of
private benefits (Geddes 2006; Magaloni 2006; Lust-Okar 2008; Blaydes 2010;
Svolik 2012). If they receive private benefits from the regime, party members and
citizens alike are apt to conclude that they have a stake in its' survival. Consequently,
they will be likely to support the regime in subsequent elections and come to its
defense in times of crisis. Also, once voters develop dependence on government
largess, they become vulnerable to threats of withdrawing it (Diaz-Cayeros, Magaloni,
and Weingast 2003).
To ensure these strategies are effectively implemented, authoritarian regimes
require an organizational vehicle, a "political machine." The use of political machines
is particularly relevant in today's electoral autocracies because almost all authoritarian
regimes that hold elections have adopted the secret ballot. When vote choice is not
individually observable, rulers have difficulty in securing voters’ compliance—they
are unable to identify and punish those who obtained private benefits but breached the
implicit contract between the consumers and the supplier of the perks and privileges.
Stokes (2005) argues that a political machine can help parties solve this compliance
problem through “deep insertion into voters’ social networks” at the neighborhood
level. Nichter (2008) and Hidalgo and Nichter (2016) concur that political machines
help parties deal with the problem of verifying compliance, but use a different
mechanism. These authors argue that political machines dispense electoral rewards
not only to change voters’ preferences but also to mobilize supporters who would
otherwise abstain.2
1
Most interviews were conducted in 2012 and 2013, and seven more were done at the time of
the 2015 District Council election.
2
Blaydes (2010) highlights a different function of political machines in authoritarian regimes;
providing a mechanism to regulate intra-elite competition. Using the example of Mubarak’s
2
These studies of machine politics in electoral autocracies are mainly based on
evidence from less developed countries containing large numbers of poor people. As
Blaydes (2006) notes, in these socio-economic contexts electoral support is a
relatively inexpensive good. Thus, it is not surprising that existing studies say little
about the relationships between authoritarian political machines and voters who
receive machines' largess. As long as such a regime enjoys a monopoly over state
resources, it can purchase support because of a copious supply of poor people who are
willing to sell their votes for basic consumer goods.
The dynamics of machine politics in more affluent settings is likely different.
This is because socio-economic development conditions what regimes can offer and
how political machines behave. At low levels of economic development, citizens have
difficulty making ends meet and political machines can easily purchase their support
with basic consumer goods such as clothes, food, and entertainment. However, the
function of a political machine is not confined to the distribution of such quotidian
items. As noted, Stokes (2005) argues that a crucial function of a political machine is
to develop “deep insertion into voters’ social networks.” This is necessary when the
regime attempts to gain individual-level information about people so that it can design
a flexible reward-and-punishment system to micro-manage voter behavior. If the
regime provides only basic items, it does not need to build a complex organizational
structure.
Regarding benefit distribution, existing studies typically focus on three
questions: (1) is receipt of benefits contingent on political loyalty? (2) are benefits
directed at individuals or groups? (3) are benefits targeted at voters generally or party
members in particular? Using these questions as a guide, Stokes et al. (2013) provide
a classification of four types of non-programmatic benefits: patronage, pork-barrel,
vote buying/turnout buying, and unconditional benefits to individuals. It is
noteworthy that benefit distribution is unlikely exogenous to socioeconomic
development. In less developed settings, local economies are highly self-sufficient
such that most residents are subject to clientelistic linkages based on patrimonial
relationships (Kitschelt and Wilkinson, 2007). Local bosses can more easily impose
political support as a selection criterion for private benefits than is the case in settings
with more developed economies.
In fact, machine politics is not confined to poor countries. Although some
mature democracies such as the United Kingdom and the United States have
witnessed a decline of electoral clientelism (Camp, Dixit, and Stokes 2014), machine
politics remains very much alive in developed countries such as Japan and Italy.
When the power of political machines is such that one party dominates the provision
of patronage, machine politics likely will persist at high levels of per capita income.
Indeed, such power asymmetries aptly characterize the situation in many authoritarian
regimes where ruling parties exercise monopoly control over state resources.
Many electoral autocracies have high income levels. For instance, in 1999, the
year before the Kuomintang lost control of Taiwan's government, the country's per
capita income already was $20,000 (US). Similarly, circa 2014, both Malaysia and
Egypt, she describes a political reward system based on the number of votes party members
can garner for the ruling party.
3
Russia had per capita incomes approximately $24,000 (US) and Hong Kong and
Singapore rank eighteenth and seventh, respectively, among all states in per capita
income. Not surprisingly, literacy rates are high in these countries, ranging from 94
percent to 99 percent of the adult population. Another notable feature of these
regimes is that they all have records of deploying political machines to strengthen the
electoral advantages associated with being the incumbent government (Hwee 2002;
Wang and Kurzman 2007; Hale 2007; Ufen 2008; Kwong 2009; Wong 2015).
Even if one focuses on less developed authoritarian regimes, it is still
problematic to assume that machine politics are confined to distributing items such as
alcohol and free lunches because voters in such countries have varying incomes and
heterogeneous consumption preferences. Voter heterogeneity may not be a significant
issue if an authoritarian regime intends to gain support only from a minimum winning
coalition of impoverished people. However, recent studies show that many
authoritarian regimes have strategic reasons to build more extensive coalitions. In
particular, they often conduct expensive election campaigns, even though opposition
parties are too weak to mount substantial challenges. The reason is that extravagant
campaigns, together with the supermajority outcomes they frequently produce, signal
the incumbent’s invincibility which, in turn, reduces the probability of future
challenges or conflict within the ruling party (Magaloni 2008; Weeden 2008; Simpser
2013).
If it is strategically important to maintain a broad support coalition, an
authoritarian government cannot avoid co-opting voters with diverse backgrounds.
Although a small item may be sufficient to purchase the support of the most
economically distressed, it is unclear whether a political machine can win affluent
voters’ hearts and minds by increasing the quantity of giveaways. The nature of
interactions between political machines and non-poor voters thus merits further
investigation. In the next section, we report the results of a Hong Kong case study
designed to illuminate how political machines work in an economically developed
society.
3. Hong Kong’s Political Machines—What Parties Supply
Anticipating retrocession in 1997, Beijing promised to grant Hong Kong a high
degree of autonomy based on a principle known as “one country, two systems.” After
the transfer of sovereignty Hong Kong would be ruled by its citizens rather than by
officials from the mainland. 3 Hong Kong also would be allowed to preserve free
market capitalism, civil liberties and limited political freedom manifested in the
popular election of representatives to seats in the legislature (the Legislative
Council—LegCo) and the District Councils.
To contest these elections effectively, mainland China needed to cultivate proBeijing elements in Hong Kong. The first pro-Beijing party, the Democratic Alliance
for the Betterment and Progress of Hong Kong (DAB), was established in 1992. The
electoral performance of pro-Beijing parties was unimpressive in the 1990s, largely
because the Chinese government’s reputation was severely tarnished by the
Tiananmen Square incident (Ma and Choy 2003). Operating in an adverse political
environment, pro-Beijing parties (collectively known as the pro-establishment camp)
3
As Fong (2014) notes, ruling elites in Hong Kong, especially the capitalists, have often
turned to mainland authorities to protect their interests.
4
adopted a more pragmatic strategy; they downplayed grand ideological narratives and
focused on socio-economic appeals involving bespoke non-programmatic benefits
(Wong 2015).
Many pro-democracy opposition parties (collectively known as the pandemocrats) also distribute goods and benefits to constituents. However, the pandemocrats’ political machines pale in comparison with those controlled by proBeijing parties, partly because the latter enjoys resource advantages and partly
because opposition parties increasingly rely on new social media to target voters. The
strategy of pro-Beijing parties has proved successful, as evidenced by their growing
strength in the District Councils and increasing vote and seat shares in the LegCo.
What accounts for the success of the pro-establishment camp? Hong Kong
has a vibrant market economy. Unlike many developing countries where the state is a
main job provider, Hong Kong citizens are predominantly employed in the private
sector. Pro-Beijing parties are unable to make economic security contingent on
political loyalty. Nevertheless, they have superior resource advantages over their pandemocratic competitor—Pro-Beijing parties can swamp constituents with material
benefits. Conventional wisdom suggests that their electoral success is all about small
giveaways. However, in interviews with District Councilors, we have often heard that
small giveaways are insufficient to gain voter support for the following reasons:
(1) Voters are no longer impressed by small gifts: consumption of small
giveaways is no different from the consumption of other goods dictated by the
law of diminishing marginal returns.4 Moreover, unsolicited offers of goods,
especially in large quantities, may be perceived as harassment rather than
gestures of goodwill;
(2) Loyalties are uncertain: voters actually seek material benefits from both
political camps. Also, Councilors may find it difficult to evaluate the
effectiveness of material gifts;
(3) Distribution methods may have unintended consequences: unless a
Councilor can supply a popular good to every interested constituent, s/he may
upset those who fail to get the item.
A more efficacious way to gain support is to help people with problems they
are unable to solve independently. To do so, some Councilors endeavor to develop a
“niche” assistance market. For instance, a pro-establishment District Councilor we
interviewed regularly offered tutorial courses to teenagers. “Are they eligible voters?”
we asked. “Nope, but their moms are,” he replied.5 In another case, a rising star in the
Democratic Party established a reputation as a fighter against collusive tendering
(weibiao) in the renovation of residential buildings. His image is so well-known that
residents from other districts seek his assistance (Ming Pao Daily News 2015).
The literature on electoral choice features two major alternative theories of
party competition. Downsian theories (Downs, 1957) hold that parties maximize
support by positioning themselves at the “ideal point” of the largest segment of the
4
A pan-democratic District Councilor used these economic terms to describe the plight of the
resource-rich pro-establishment parties.
5
Interview March 1, 2013.
5
population. However, Stokes (1963; 1992; see also Clarke et al., 2004) argues that
parties typically compete primarily on valence issues where large majorities of voters
favor a particular end of public policy. The canonical example is a healthy economy.
Other important policy areas also are heavily valenced, with substantial majorities
demanding easily accessible, high quality health and educational services and high
levels of national and personal security. Debate on valence issues focuses not on the
"what to do," but rather on "how to do it" and, especially, "who can do it best."
A two-time Hong Kong District Councilor we interviewed argues that neither
of these models aptly describes how he approaches his constituents.6 He does not
target a specific ideological segment of his constituency. Nor does he focus on a few
valence issues. Instead, he provides bespoke services to multiple groups. For instance,
he organizes Chinese opera shows for the elderly and Japanese language courses for
people fond of Japanese culture. He even introduced a new “zorb" football game for
young people in his district.
Oftentimes, Councilors offer personal help to constituents. One interviewee
recounted how he helped a widow pursue compensation for her husband’s death. “I
also helped her hold the funeral,” he added.7 Such personal contacts are analogous to
the “deep insertion into voters’ social networks” that Stokes (2005) describes. Those
who have received the Councilor’s help likely become loyal supporters. They vote
for him and volunteer to help during election campaigns.
However, District Councilors cannot establish strong bonds with every
constituent and the secret ballot means that Councilors cannot identify who actually
votes for them. The latter is not a serious problem. Although individual support is
unknown, aggregate vote shares can be gauged. A DAB District Councilor illustrates
how this works. 8 The number of votes he can bring for the LegCo candidate he
supports signals constituents’ approval of his work: “If I could transfer 80 percent of
those who voted for me to the LegCo candidate, I am in a good shape. If I can only
get 50 percent, I need to worry about my re-election,” he explains. These percentages
can be estimated with considerable accuracy because the ballot-counting results of
Legislative Council elections are available at the polling station level. Kitschelt and
Wilkinson (2007) note that ballot data disaggregated at low levels make collective
monitoring feasible. With aggregate polling station results, political machines can
make fairly accurate assessments of the efficacy of their resource allocations.
Levels of personal interaction between District Councilors and constituents
also are correlated with the latter's social class backgrounds. Middle-class people
tend to demand more attention from District Councilors. There are two reasons: first,
middle-class persons are more capable of articulating their needs and discontents than
are poor, uneducated people. Second, requests from relatively well-off residents
usually involve more than help to make ends meet. Problems they encounter can be
challenging. Indeed, they frequently try to solve problems by themselves first
because they value their privacy and do not want the Councilors to know about their
affairs. In fact, many of our interviewees agreed that it is not always pleasant to take
care of middle-class districts. One junior DAB District Councilor confided that initial
6
Personal communication October 30, 2015.
Interview March 7, 2013.
8
Interview March 1, 2013.
7
6
interactions with her constituents were stressful because many of them were engineers,
doctors, accountants, and lawyers. She said these people gave her little respect, partly
due to her young age.9
Another young pan-democratic District Councilor had similarly unpleasant
experiences. He has been queried more than once by residents living in his middleclass district about his education qualifications. Undeterred, the Councilor found that
well-off residents have a weak link: time. When bombarded with condescending
questions he would first humbly admit his ineptitude. Then he would raise a question,
“I know you are smart and capable, but do you have time? I have plenty, and I am
here to help you.”10
Indeed, time is often the most important asset of a District Councilor. A proestablishment Councilor shared an experience with us. 11 Homeowners of two
buildings disputed which side should be held responsible for removing stagnant water
between their buildings. The Councilor was invited to solve the issue. Because both
sides were her constituents, she needed to be careful. She spent two weeks pounding
the doors of various government departments to determine ownership of the alley at
issue. Middle-class residents of the buildings likely had neither the interest nor the
time to engage with the problem.
The Councilor's time spent on resolving the issue pales in comparison with
another story we heard from a two-time pro-establishment winner. Some residents
living in his district were affected by the construction of a new, high-rise hotel. It
took him two years to negotiate with different authorities to reduce the height of the
hotel from 16 to six stories.12 These examples illustrate how the supply of Councilors’
personal assistance meshes with constituent demand.
4. The Demand Side - How Voters React
The preceding section examined how Hong Kong parties formulate strategies to
attract voters through gifts and services—the “supply side” of machine politics. In
this section, we examine the “demand side”— how voters react to non-programmatic
benefits District Councilors provide. For this purpose, we employ the results of a
large representative post-election internet survey conducted after the 2015 District
Council election. The survey data enable us to study relationships between machine
politics, political attitudes and voting behavior.
4.1 Key Questions
We focus on three theoretically and substantively significant questions:
(1) Does the distribution of non-programmatic benefits influence voting behavior?
By voting behavior, we refer to both vote choice and turnout. Conventional
wisdom suggests that Hong Kong's pro-establishment parties use non-programmatic
benefits to garner political support, although in our interviews we found that parties
from both political camps delivered non-programmatic benefits in exchange for votes.
9
Personal communication February 3, 2016.
Personal communication February 3, 2016.
11
Interview February 23, 2016.
12
Interview March 10, 2016.
10
7
Pro-establishment parties clearly have a resource advantage over their pan-democratic
counterpart (Wong 2014, 2015). Thus, we hypothesize that pro-establishment parties
should benefit more from distribution of non-programmatic benefits than pandemocratic parties.
Hypothesis 1. Voters who receive non-programmatic benefits are more likely to vote
for pro-establishment parties than pan-democratic parties.
Nichter (2008) (see also Aidt and Jensen, 2016; Hidalgo and Nichter, 2016;
Kam 2016; Lawson and Greene, 2014) argues that machine politics often are intended
to buy turnout rather than votes because the secret ballot means that political
machines are unable to monitor how recipients vote. We found echoes of this
argument in our research; numerous Councilors mentioned the impossibility of
ascertaining the effect of their gifts and services on individual vote choices. Yet, as
noted, parties can estimate the aggregate effect of their efforts by examining turnout
rates of likely supporters. This provides important information parties use to adjust
the level of non-programmatic benefits and suggests the following hypothesis:
Hypothesis 2. Those who received non-programmatic benefits are more likely
to vote than those who did not get benefits.
(2) Who are recipients of non-programmatic benefits?
No matter how many non-programmatic benefits political parties provide, the
benefits would not affect voting behavior if there were no demand for them.
Therefore, it is important to determine the origins of demand. As discussed, existing
studies in developing countries demonstrate that poor and illiterate voters are the main
targets of machine politics because they value non-programmatic benefits more than
wealthier citizens. In contrast, in Hong Kong the demand for District Councilors’
services is not confined to poor people. As discussed above, relatively affluent
citizens also seek help from Councilors when existing government services fail to
address their needs. This observation suggests two sets of alternative hypotheses:
Hypothesis 3a. Income level is negatively correlated with demand for nonprogrammatic benefits.
Hypothesis 3b. Income level is positively correlated with demand for nonprogrammatic benefits.
Hypothesis 4a. Education level is negatively correlated with demand for nonprogrammatic benefits.
Hypothesis 4b. Education level is positively correlated with demand for nonprogrammatic benefits.
(3) How do recipients of non-programmatic benefits view policies/politics?
Previous research on machine politics assumes that recipients of nonprogrammatic benefits exchange their votes for material benefits. According to
Stokes (2005), machine politics involve an “implicit deal whereby the machine
distributes goods and the recipient votes for the machine.” This observation implies
that recipients value the goods received, at least more than policy or ideological
8
concerns; otherwise, such an exchange would not occur.13 In turn, this suggests the
following hypotheses:
Hypothesis 5. Recipients of non-programmatic benefits are more likely to
answer “don’t know” than non-recipients when invited to comment on current
socio-political issues.
Hypothesis 6. Recipients of non-programmatic benefits tend to have more
moderate opinions than non-recipients, if they have opinions at all.
4.2 Survey Data
Our survey of the Hong Kong electorate was conducted immediately after the
November 2015 District Council election.14 The survey data include information on
2,160 respondents whose socio-demographic characteristics are displayed in the
Appendix, alongside population distributions. A post-stratification weight based on
gender, age, and region is used in the multivariate analyses reported below to ensure
the representative nature of the sample.
Measurement
Non-programmatic benefits are key variables in our analysis. To measure
consumption of these benefits, we focus on three classes of goods: (1) social activities,
(2) material gifts, and (3) personalized assistance. Social activities include domestic
and foreign travel, banquets, health services, concerts and other performances.
Material gifts consist of items such as food and beverages, cash coupons, calendars
and detergents. Personalized assistance refers to specialized requests such as legal
consultation, filing complaints, resolving neighborhood conflicts, tackling family
issues, household maintenance, voter registration, and miscellaneous other problems.
Across all survey respondents, 51 percent received at least one of these nonprogrammatic benefits. The distribution of material, social, and personal benefits is
summarized in Figure 1.
(Figure 1 about here)
Several variables pertaining to non-programmatic benefits are employed in the
multivariate analyses which follow. First are dichotomous variables with the value 1
if a respondent has consumed any social activity, material benefit or personal
assistance, respectively, and 0 otherwise. We also construct a summary variable that
takes a value 1 if any of the three types of benefits has been consumed and 0
otherwise. In addition, because it is likely that the quantity of goods consumed
affects voting behavior, we estimate how many benefits were received. If one person
has participated in day trips, banquets, and free health checks organized by his/her
District Councilor whereas another has only attended Chinese opera shows, we
assume that the first person has consumed more benefits than the second.
13
Some voters might enter an implicit contract only with a political machine that shares
similar ideological views. Under such circumstances it is unclear whether the effect of nonprogrammatic benefits, if any, is due to the benefits per se or ideological affinity.
14 Fieldwork was conducted via internet by YouGov Hong Kong. Extensive comparative
analyses indicate that internet and in-person surveys yield very similar results in multivariate
analyses of turnout and party choice (Sanders et al. 2007; Ansolabehere and Schaffner, 2014).
9
4.3 Results
4.3.1 Non-programmatic Benefits and Voting
We first examine whether receipt of benefits influences voting measured as turnout
and support for pro-democratic and pro-establishment candidates. The turnout
variable takes a value of 1 if a respondent voted in the 2015 District Council election
and 0 otherwise. Similarly, the two candidate choice variables have a value of 1 if a
respondent voted for a pro-establishment (pan-democratic) candidate and 0 otherwise.
Recognizing numerous factors affect turnout and party choice, we include
several other predictor variables in the multivariate models. First, we use two 0-1
dummy variables to measure pro-democratic and pro-establishment party
identifications. We also include predictors tapping perceptions of candidate and party
performance on valence issues including the economy, law and order, media freedom
and rule of law. In addition, since numerous studies (see Dalton, 2014) have shown
that perceptions of one's ability to influence the political process enhance electoral
participation, we include a measure of political efficacy. 15 Two other important
concepts in research on electoral turnout are interest in public affairs16 and sense of
civic duty17 both of which are included in the models as well. Finally, even if people
have a strong sense of political efficacy, they might conclude that it is not worthwhile
to participate in District Council elections because Councilors lack influence. Using a
question that asks respondents to compare District Councilors and Legislative
Councilors with respect to their ability to solve problems in their neighborhood, we
construct a variable that compares perceptions of the relative problem-solving
capacities of the District and Legislative Councilors.
Our model specifications also are informed by studies of ethnic identity and
voting behavior (e.g., Carlson 2015). Although ethnic-Chinese constitute 94 percent
of the Hong Kong population, ethnicity has become an important social cleavage
because of the large influx of immigrants from Mainland China since the retrocession.
One ethnicity variable assigns a value of 1 to those who were born in Hong Kong and
0 otherwise. Three other variables capture subjective perceptions of ethnic identity.
These are dummy variables constructed from a question that asks if respondents think
of themselves as a Hong Konger, a Chinese, a Hong Kong Chinese/Chinese Hong
Konger or other. Finally, we control for several socio-demographic characteristics
(age, education, gender and income) that previous research has shown to influence
turnout and electoral choice (Dalton, 2014).
Model parameters are estimated with logistic regression 18 and results are
presented in Table 1. To begin, it is noteworthy that the benefit variables are
significant predictors of turnout in these multivariate analyses. Regardless of how we
15
Respondents were asked to rate their influence on politics and public affairs on a 0 (no
influence) to 10 (great deal of influence) scale.
16
This variable is based on responses to the question “how interested are you in public
affairs?”
17
Respondents are asked to indicate agreement or disagreement with the following statement:
“I would be seriously neglecting my duty as a citizen if I didn't vote in elections.”
18
Analyses are performed using Stata 11's svy logistic regression procedure. Probabilities of
voting are computed using Stata's MARGINS command (Mitchell, 2012) with each predictor
variable varied across its range while other predictors are held at their mean values.
10
measure non-programmatic benefits, the benefit variables have sizable and
statistically significant effects on electoral participation. Converting the coefficient
on the benefit variable in model 1 into probability of voting shows that, ceteris
paribus, those who obtained non-programmatic benefits are 16 percent more likely to
go to the polls than are people who did not receive benefits. This finding is consistent
with Hypothesis 2.
(Table 1 about here)
Another noteworthy finding is that the most powerful benefits predictor of
turnout is personal benefits. Material benefits have the smallest effect, a finding
consistent with our qualitative fieldwork indicating that material benefits fail to
impress many Hong Kongers. Also, as demonstrated by model 3, the quantity of nonprogrammatic benefits obtained matters. All coefficients for the variables of interest
are positive, indicating that the more non-programmatic benefits provided to someone,
the more likely they are to go to the polls.
Regarding other attitudinal predictors, only civic duty has a positive and
statistically significant effect. Sense of political efficacy, interest in public affairs and
comparative performance judgments are not influential. In contrast, party
identification matters—people who identify themselves as pan-democratic supporters
are more likely to vote than are nonpartisans. However, the coefficient for proestablishment party identification is statistically significant in model 1 only. In
addition, there is no evidence that valence considerations drive electoral
engagement—judgments about government performance in various policy areas are
not significantly related to turnout. Regarding demographics, only age and income
are significant, with the relationships being positive in both cases.
Next, we consider the effects of non-programmatic benefits on party choice.
As observed, pro-establishment parties enjoy resource advantages over their pandemocratic counterpart. If pro-establishment parties provide more non-programmatic
benefits than other parties and those who receive such benefits are more likely to vote,
one would expect that there would be a significant relationship between receipt of
non-programmatic benefits and voting for pro-establishment parties. However, the
bivariate analysis displayed in Table 2 suggests that recipients of non-programmatic
benefits are about as likely to vote for pro-establishment candidates as for pandemocratic ones. To investigate this relationship more rigorously, we estimate a
multivariate model that controls for several other plausible predictors. Results are
presented in the last six columns of Table 1.
(Table 2 about here)
Models 4, 5 and 6 provide limited support for the idea that pro-establishment
candidates are advantaged by benefits more their pan-democratic counterpart—the
only predictor with a significant coefficient is personal assistance (model 5). When
we measure personal assistance as a polychotomous variable, its coefficient is no
longer significant (model 6). Other coefficients are not significant and some do not
have the expected (positive) sign. Thus, hypothesis 2 is only partially supported by
the data. In Models 7, 8, and 9, the dependent variable is pan-democratic voting.
Again, there is no consistent sign for the coefficients on the variables of interest and
their estimated effects are not statistically significant (p > .05).
11
These results should not be interpreted as evidence that material benefits and
social activities do not influence vote choice. Rather, they suggest that proestablishment parties do not monopolize the provision of non-programmatic benefits.
As both political camps distribute goods and organize social activities, receiving such
items does not distinguish pro-establishment supporters from pan-democratic ones.
The analyses indicate that the only non-programmatic benefit that may set these
groups apart is personal assistance. People who have received personal assistance
from District Councilors are more like to vote for pro-establishment parties. This
implies that pro-establishment parties outflank their pan-democratic counterpart in
giving personal assistance. Observers who attribute the electoral success of proestablishment parties solely to the distribution of material gifts overlook the
complexity of the effects of non-programmatic benefits.
For vote choice, party identification plays its predictably important role.
Table 1 shows that voters who identify themselves with the pan-democratic camp are
less likely to support pro-establishment parties, and vice versa (p < .01). Ethnicity
matters as well—native Hong Kongers are less likely to vote for pro-establishment
parties. Models 7, 8 and 9 also support hypotheses regarding the impact of valence
judgments and civic attitudes. People who support pan-democratic parties tend to see
voting as their civic duty and to be concerned about threats to media freedom. Age is
a significant predictor too, with younger people being more likely to vote for the pandemocrats.
4.3.2 Receipt of Non-programmatic Benefits
Existing theories suggest that education and income are strong predictors of the
receipt of non-programmatic benefits, partly because the votes of illiterate and poor
citizens are cheaper to buy, and partly because such citizens may depend on the
largess of political machines. Another factor closely related to income is home
ownership. The property market of Hong Kong is historically volatile. Low-skilled
workers can become rich by taking advantage of the boom-bust cycle in the property
market. Income-rich, asset poor people fare no better than the income-poor, asset-rich.
In Hong Kong, there is a popular impression that the political machines of
pro-establishment parties target senior citizens, who are more “gullible.” New
immigrants also are considered an easy catch for pro-establishment parties because
they face challenges in assimilating into society and many lack helpful social
connections.
To investigate these conjectures, we analyze receipt of nonprogrammatic benefits in terms of several predictors: age, income, gender, education,
home ownership, Hong Kong birth and support for pro-establishment parties. In
addition, we include a political knowledge index, constructed using responses to six
survey items.19 Table 3 presents results of the analysis.
(Table 3 about here)
When we use an aggregate measure for non-programmatic benefits (models 1
and 5), political knowledge is the only significant variable. Converting the coefficient
into a probability shows that a unit increase in political knowledge is associated with
19
To control for the possibility that respondents looked up answers to the knowledge
questions on the internet, we exclude respondents who spent more than 20 minutes on the
survey.
12
a change of 5.4 percentage points in the probability of seeking non-programmatic
benefits. In contrast, educational level is not statistically significant in any of the
analyses. Thus, the data do not support either hypothesis 4a or 4b. Juxtaposing this
result with the effect of political knowledge, one can see a point that has been underemphasized in previous research on machine politics in electoral autocracies. Making
use of a political machine to address one’s needs requires a degree of knowledge of
the political system, and such knowledge is not necessarily highly correlated with
level of formal education.20 Existing studies of electoral clientelism emphasize the
behavior of less educated people who are considered susceptible to the appeals of
machine politics because the machine can help them meet basic needs. Present
findings suggest an alternative consideration—politically knowledgeable people are
capable of taking advantage of a political machine to advance their own interests.
Because there are various non-programmatic benefits, it is instructive to
examine if there are systematic differences in the types of voters seeking different
benefits. Party identification, measured by support for pro-establishment parties, is
significant in analyses of each of the three types of non-programmatic benefits
(namely, material, social, and personal). However, partisanship is not significant
once we control for political knowledge, thereby suggesting that pro-establishment
supporters may not be more inclined to seek non-programmatic benefits.
Economic well-being is a strong predictor of receipt of personal nonprogrammatic benefits. Even after controlling for political knowledge, income is still
positively and significantly related to receipt of personal benefits (model 8). This
result, which supports Hypothesis 3b rather than Hypothesis 3a, is at odds with
exiting theories, but it is quite consistent with findings from our qualitative interviews.
In particular, low-income people often are able to address their needs through formal
institutional channels. However, middle-class persons may not have such
straightforward solutions because their problems typically do not involve basic needs
that can be addressed by government agencies. This is why they turn to District
Councilors who can provide one-stop personal assistance. Finally, Table 3 shows that
age, ethnic status, gender and home ownership fail to exhibit systematic relationships
with any of the three types of non-programmatic benefits.
4.3.3 Receipt of Non-programmatic Benefits and Major Policy Controversies
In concluding analyses, we focus on two major controversies in contemporary Hong
Kong. The first concerns Hong Kong's prolonged struggle for democratization which
culminated in the 2014 Occupy Movement protests. Although Occupy activists and
sympathizers fervently advocate immediate democratization, others continue to
prioritize social stability and economic prosperity. Our survey contained two agreedisagree questions designed to gauge attitudes toward democratization: (a) “Hong
Kong has a democratic political system;” (b) “Western-style democracy is not suitable
for Hong Kong.” We examine if there are systematic differences in answers to these
questions between recipients of non-programmatic benefits and other respondents.
A second, related controversy concerns increasing integration with Mainland
China. Although Hong Kong is now a city of China, mainlanders wishing to visit
Hong Kong are required to obtain a special travel document. This requirement reflects
Hong Kong’s status as a special administrative region, an institutional arrangement
20
The correlation between political knowledge and educational level is only .02.
13
intended to preserve Hong Kong’s free market economy and high degree of civil
liberties. However, economic and social integration between Hong Kong and the
mainland has accelerated in recent years. Although this has generated new economic
opportunities, it also has strained relationships between native Hong Kongers and
mainland tourists and immigrants (Wong 2015). Some Hong Kongers perceive
mainlanders as competitors for housing, public resources and daily necessities. In
addition, the government has imposed policies to accelerate social integration,
including promoting Putonghua in schools where Cantonese is the traditional spoken
language. The prospect of closer integration with the mainland has polarized Hong
Kong’s society.
Our survey contains three questions related to social tensions between Hong
Kong citizens and people from the mainland. Respondents were asked if they agree
or disagree: (a) “Hong Kong has too many immigrants coming from Mainland China;”
(b) “Closer integration with Mainland China will benefit Hong Kong;” (c) “Primary
schools should teach Chinese language in Cantonese, rather than in
Putonghua/Mandarin.” Analyses of answers to these questions reveal that there are
systematic differences in views of democracy between recipients of nonprogrammatic benefits and other respondents (Table 4). Recipients have a
significantly higher average score on the statement “Hong Kong has a democratic
political system” than do non-recipients. Recipients also tend to see Western-style
democracy is unsuitable for Hong Kong. These findings suggest that recipients of
non-programmatic benefits are more satisfied with the political status quo than the
non-recipients. In contrast, results concerning views on integration with the mainland
are mixed. On the one hand, there is a significant difference between the two groups
with respect to views about advantages of closer integration—benefit recipients are
more likely to see integration in a positive light. On the other hand, there is no
systematic difference between the two groups on immigration and language issues.
(Table 4 about here)
People willing to sell their votes for items such as liquor and food are
generally thought to be more pragmatic than ideological. Extending this logic, one
might wonder whether recipients of non-programmatic benefits in Hong Kong hold
more moderate or ambivalent views on politics than do non-recipients. Table 4 shows
that this is not the case. Benefit recipients are not less likely to hold strong opinions—
defined as “strongly agree” or “strongly disagree”—on these thorny issues in Hong
Kong politics. Nor do recipients express greater ambivalence about the issues. On
the questions about democracy, the percentage of recipients who answered “neither
agree nor disagree” is significantly lower than the percentage of non-recipients. In
other words, recipients seem to be clearer about their stance. Regarding the
probability of answering “don’t know,” there are systematic differences between the
two groups in responses to all of the questions. However, contrary to expectations,
recipients are significantly less likely to think that they are uninformed about an issue.
In sum, the data are not consistent with Hypotheses 5 and 6.
5. Conclusion: Machine Politics Hong Kong Style
Machine politics exists in many countries, democratic and authoritarian, developing
and developed. In recent years, studies of authoritarian politics have accorded
considerable attention to the operation of political machines. Authoritarian
14
governments have an incentive to build these organizations because many of these
governments hold regular, quasi-competitive elections which encourage ruling parties
to respond to voters’ needs. Political machines provide authoritarian regimes with an
effective means to distribute goods and services that enable them to purchase political
support from their electorates. A government's ability to engage in these distributive
practices is facilitated by monopoly access to valuable state resources.
Existing studies often portray voters in authoritarian regimes as passive,
pliable, and ready to sell their votes for perks and privileges, food and liquor.
Political machines are characterized as powerful organizations penetrating deep into
voters’ social networks, providing largess that voters would otherwise not have. At
times, these organizations may be able to punish disloyal supporters by withdrawing
benefits. This portrait of machine politics in electoral autocracies is largely based on
studies of impoverished developing countries where voters, especially those in rural
areas, tend to be poor and uneducated. How do political machines operate in more
affluent electoral authoritarian regimes? How do such regimes use political machines
to co-opt urban or middle-class voters? Above, we have attempted to answer these
questions using the District Councils in post-1997 Hong Kong as a case study to
illustrate the dynamics of machine politics in an affluent non-democracy.
A striking feature of the Hong Kong case is that the power relationship
between voters and political machines is quite different from that observed in
previous studies. Because non-programmatic benefits offered by political machines
are not vital to Hong Kongers' livelihoods, these grassroots organizations typically
cannot coerce people to vote, let alone require them to support particular candidates.
This is not an argument that political machines in Hong Kong are ineffective. Rather,
to attract sizable numbers of voters, Hong Kong parties organize a diverse range of
social activities and offer highly customized personal assistance to forge close
connections with their constituents.
An important finding is that levels of income and education have little
predictive power in analyses of receipt of non-programmatic benefits. This is also
related to the affluence of Hong Kong’s society, where basic welfare is accessible to
many low-income citizens such that they need not depend on the largess of political
machines. Government agencies provide regular and formal channels to help less
well-off persons make ends meet.21 In contrast, middle-class people often need to
turn to District Councilors to solve problems that lack simple, one-stop solutions.
Indeed, our survey data testify that the probability of seeking personal help from
District Councilors is positively related to income level.
In addition, we find that receipt of non-programmatic benefits is strongly
correlated with turnout, but not vote choice. Because the ballot is secret and voters’
livelihoods are not dependent on non-programmatic benefits, political machines face
challenges in securing compliance. It is almost impossible to ensure that someone
who receives goods from a party machine actually votes for that party. Recognizing
this, some earlier studies have argued that a rational strategy is to distribute goods to
groups containing large proportions of likely loyalists. In effect, non-programmatic
21
Unlike countries such as Argentina, where public employees also engage in the distribution
of patronage (Oliveros, 2016), government agencies in Hong Kong play no such role.
15
benefits are used to buy turnout among groups of probable supporters, rather than
individual-level votes. Our analyses are consistent with this argument.
Viewed generally, present findings shed light on important developments in
Hong Kong politics. The increasingly volatile pro-democracy struggle, highlighted by
the 2014 Occupy Movement protests and the violent 2016 Mongkok riot, has attracted
strong media and scholarly interest.22 However, little attention has been paid to the
growing strength of pro-establishment parties in local elections. Pro-democracy media
and activists alike tend to attribute their rival’s electoral success to the provision of
material benefits, with pro-establishment parties capitalizing on their overwhelming
resource advantage to distribute small giveaways to large numbers of politically
apathetic and gullible voters. This interpretation is problematic; recipients of nonprogrammatic benefits are no more apathetic than non-recipients. Recipients also
tend to be more politically knowledgeable and wealthier than the non-recipients.
Moreover, the most effective tool to increase electoral support is not material gifts,
but highly customized personal assistance. Taken together, these findings suggest that
members of Hong Kong's pan-democratic elite may need to confront an inconvenient
truth—their pro-Beijing rivals are developing solid, possibly increasing support
among politically influential segments of the electorate. Machine politics, Hong
Kong style, propels this trend.
22
See Cheng (2016), forthcoming.
16
References
Aidt, T. S. and P. S. Jensen. 2016. "From Open to Secret Ballot: Vote Buying and
Modernization." Comparative Political Studies Posted online Feb. 22nd.
Ansolabehere, S., and B. F. Schaffner. 2014. "Does Survey Mode Still Matter?
Findings from a 2010 Multi-mode Comparison." Political Analysis 22: 285-303.
Blaydes, L. 2006. "Who Votes in Authoritarian Elections and Why? Determinants of
Voter Turnout in Contemporary Egypt. Philadelphia, PA: Annual Meeting of the
American Political Science Association.
Blaydes, L. 2010. Elections and Distributive Politics in Mubarak’s Egypt.
Cambridge: Cambridge University Press.
Carlson, E. 2015. "Ethnic Voting and Accountability in Africa: A Choice Experiment
in Uganda. World Politics 67: 353-85.
Camp, E., D., A., and S. Stokes. 2014. "Catalyst or Cause? Legislation and the
Demise of Machine Politics in Britain and the United States." Legislative Studies
Quarterly 39: 559-92.
Cheng, Edmund W. 2016. “Street Politics in Hybrid Regime: The Diffusion of
Political Activism in Post-colonial Hong Kong.” The China Quarterly forthcoming.
Clarke, H. D., D. Sanders, M. C. Stewart and P. Whiteley. 2004. Political Choice in
Britain. Oxford: Oxford University Press.
Dalton, R. 2014. Citizen Politics: Public Opinion and Political Participation in
Advanced Industrial Democracies. Thousand Oaks, CA: Sage Publications.
Diaz-Cayeros, A., B. Magaloni, and B. R. Weingast. 2003. "Tragic Brilliance:
Equilibrium Hegemony and Democratization in Mexico." Stanford: Stanford
University, Hoover Institution.
Downs, A. 1957. An Economic Theory of Democracy. New York: Harper & Row.
Fong, B. C. 2014. "The Partnership between the Chinese Government and Hong
Kong's Capitalist Class: Implications for HKSAR Governance, 1997–2012." The
China Quarterly, 217: 195-220.
Geddes B. 2006. "Why Parties and Elections in Authoritarian Regimes?" Los
Angeles: University of California at Los Angeles, Department of Political Science
Unpublished manuscript
Hale, H. 2007. "Correlates of Clientelism: Political Economy, Politicized Ethnicity,
and post-Communist Transition." In H. Kitschelt and S. I. Wilkinson, eds., Patrons,
Clients, and Policies: Patterns of Democratic Accountability and Political
Competition, pp. 227-50.
17
Hidalgo, F. D. and S. Nichter. 2016. "Voter Buying: Shaping the Electorate through
Clientelism." American Journal of Political Science 60: 436-55.
Hwee, Y. L. 2002. "Electoral Politics in Singapore." In D. Tomsa and A. Ufen,
eds., Electoral Politics in Southeast and East Asia pp. 203, 217.
Kam, C. 2016. "The Secret Ballot and the Market for Votes at 19th-Cenury British
Elections. Comparative Political Studies. Posted online Feb 26th.
Kitschelt, H., & S. I. Wilkinson. 2007. Patrons, Clients and Policies: Patterns of
Democratic Accountability and Political Competition. Cambridge: Cambridge
University Press.
Kwong, B. K. K. 2009. Patron-Client Politics and Elections in Hong Kong. London:
Routledge.
Lawson, C. and K. Greene. 2014. "Making Clientelism Work: How Norms of
Reciprocity Increase Voter Compliance." Comparative Politics 47: 61-85.
Lust-Okar, E. 2008. Competitive Clientelism in Jordanian Elections. In E. lust-Okar
and S. Zerhouni, eds., Political Participation in the Middle East and North Africa, pp.
75–94.
Ma, N., and C. K. Choy. 2003. "The Impact of Electoral Rule Change on Party
Campaign Strategy Hong Kong as a Case Study." Party Politics 9: 347-67.
Magaloni, B. 2006. Voting for Autocracy: Hegemonic Party Survival and its Demise
in Mexico. Cambridge: Cambridge University Press.
Ming Pao Daily News. 2015. Available at http://news.mingpao.com/ins15070214
35832578021 (accessed on February 27, 2016).
Mitchell, M. N. 2012. Interpreting and Visualizing Regression Models Using Stata.
College Station, TX: Stata Press.
Nichter, S. 2008. "Vote Buying or Turnout Buying? Machine Politics and the Secret
Ballot." American Political Science Review 102: 19-31.
Oliveros, Virginia. 2016. "Making it Personal: Clientelism, Favors, and the
Personalization of Public Administration in Argentina." Comparative Politics 48:
373-391.
Sanders, D., Clarke, H. D., M. C. Stewart, M. C., and P, Whiteley. 2007. "Does Mode
Matter for Modeling Political Choice? Evidence from the 2005 British Election
Study." Political Analysis 15: 257-285.
Simpser, A. 2013. Why Governments and Parties Manipulate Elections: Theory,
Practice, and Implications. Cambridge: Cambridge University Press.
18
Stokes, D. E. 1963. "Spatial Models of Party Competition." American Political
Science Review 57: 368-77.
Stokes, D. 1992. "Valence Politics." In D. Kavanagh, ed. Electoral Politics. Oxford:
Clarendon Press, 141-64.
Stokes, S. C. 2005. "Perverse Accountability: A Formal Model of Machine Politics
with Evidence from Argentina." American Political Science Review 99: 315-25.
Stokes, S. C., T. Dunning, M. Nazareno and V. Brusco. 2013. Brokers, Voters, and
Clientelism: The Puzzle of Distributive Politics. Cambridge: Cambridge University
Press.
Svolik, M. W. 2012. The Politics of Authoritarian Rule. Cambridge: Cambridge
University Press.
Wang, C. S., and C. Kurzman. 2007. "The Logistics: How to Buy Votes." In F. C.
Schaffer, ed. Elections for Sale: The Causes and Consequences of Vote Buying, pp.
61-78.
Wedeen, L. 2009. Peripheral Visions: Publics, Power, and Performance in Yemen.
Chicago: University of Chicago Press.
Wong, S. H. W. 2014. "Resource Disparity and Multi-level Elections in Competitive
Authoritarian Regimes: Regression Discontinuity Evidence from Hong
Kong." Electoral Studies 33: 200-19.
Wong, S. H. W. 2015. Electoral Politics in Post-1997 Hong Kong: Protest,
Patronage, and the Media. New York: Springer.
Ufen, A. 2008. "The 2008 Elections in Malaysia: Uncertainties of Electoral
Authoritarianism." Taiwan Journal of Democracy, 4: 155-69.
19
Figure 1. Percentages of Various Kinds of Material, Personal and Social
Non-Programmatic Benefits Distributed to the Hong Kong Electorate
55.4
Total Personal + Social
33.3
Total Social
1.4
Other Social
9.6
Shows
13.2
Health Services
0.9
Foreign Travel
Domestic Travel
5.0
Banquets
4.7
20.4
Total Personal
0.8
Other Personal
6.5
Voter Registration
1.4
Neighborhood Conflicts
2.2
Legal Consultation
3.6
Household Maintenance
4.7
File Complaints
1.2
Family Problems
44.6
Total Material
3
Other Material
2.6
Food & Beverage
8.7
Detergents
1.2
Cash Coupons
30.2
Calendars
0
10
20
30
40
50
60
Percent
20
Table 1. Multivariate Logistic Regression Models of Turnout and Party Choice
21
Table 2. Non-Programmatic Benefits and Electoral Choice
Electoral Choice
Pro-establishment
Pan-democrat
Independent
Blank Vote
Total
Non-programmatic Benefits
Non-recipients
Recipients
28.5%
30.3%
35.3
30.6
32.9
35.1
3.3
4.0
100
100
Note: respondents who voted in the 2015 District Council election.
22
Table 3. Multivariate Logistic Regression Models of Receipt of
Various Kinds of Non-Programmatic Benefits
23
Table 4. Multivariate Model of How Recipients View Non-Programmatic Benefits
24
Appendix
Socio-Demographic Characteristics of Survey Sample and Hong Kong Population
Note: sample size N = 2160.
25