The 2013 Malaysian elections: ethnic politics or urban wave?
Ng, Jason Wei Jian ; Rangel, Gary John ; Vaithilingam, Santha 等
In this article we examine the electoral impact of urbanization
vis-a-vis ethnicity in Malaysia. We employ a robust econometric
technique, the fractional response logit model, on data from the
recently concluded thirteenth general election. The findings show that
there are both an ethnic effect and an urban effect in determining the
distribution of parliamentary seats among the political groups. Strong
support for the opposition coalition, Pakatan Rakyat, was evident in
urban constituencies, while the ruling coalition, Barisan Nasional,
continued to enjoy success in rural constituencies. Although Barisan
Nasional is still dependent on Bumiputera support, its success is also
dependent on non-Bumiputera support from rural constituencies. However,
with declining birthrates among the Chinese electorates, this support
may not be forthcoming in future elections. We also provide insights for
both coalitions to consider in developing strategies for the next
election. Keywords: Malaysia, thirteenth general election, ethnic
politics, fractional logit response model, urbanization
**********
There have been thirteen general elections in Malaysia since it
gained independence from the British in 1957. In all thirteen elections,
the same coalition, Barisan Nasional (BN) (1) or National Front, has
been returned to power. The thirteenth general election (GE13), held on
May 5, 2013, was the most fiercely contested election in Malaysia's
history. For the first time, there was a degree of uncertainty about
BN's ability to retain power. The opposition coalition, Pakatan
Rakyat (PR) or People's Alliance, which was formed in 2008,
appeared to have become credible enough to launch a genuine threat to
BN's stranglehold on power. However, in the end, BN retained power
with a simple but comfortable majority by winning 133 of the 222
parliamentary constituencies contested. This was a slight reduction from
the 138 seats that it won in the previous election held in 2008. This
outcome again denied BN the two-thirds majority it had enjoyed from
independence in 1957 until 2008. By polling only 47.4 percent of the
votes cast, it recorded its lowest ever popular support in history.
Peninsular Malaysia was the major battleground where BN won
eighty-five seats against PR's eighty. (2) While this result was
the same as in the 2008 election, there was, however, a notable change
in the distribution of the seats. BN regained some of the
Malay-dominated rural seats it had lost in 2008, but also lost a number
of urban seats in traditional BN strongholds, especially in the southern
part of the peninsula. Given that most Chinese Malaysians live in urban
areas, Prime Minister Najib Razak subsequently termed this phenomenon
the Chinese Tsunami, purportedly reflecting the huge exodus of Chinese
voters from BN to PR (Noh 2014).
However, political analysts have observed that while there was an
increase in Chinese support for PR, the electoral outcome also saw a
major swing in the urban electorate against BN, causing a further
widening of Malaysia's rural-urban rift (Boo 2013). One of the PR
leaders, Lim Kit Siang, described this as a "Malaysian and urban
tsunami" (The Star 2013). This debate of whether the electoral
outcome was a result of an urban or a Chinese swing against BN has been
inconclusive due to the high correlation between urban parliamentary
constituencies and the proportion of Chinese Malaysian voters in any
given constituency.
Against this backdrop, the aim of this study is to identify which
of the two factors, ethnicity or urbanization, provides a stronger
explanation for the erosion of BN's popular votes in GE13. We
measure the proportion of votes won by BN due to urbanization and ethnic
factors using a robust econometric technique, the fractional response
logit model proposed by Papke and Wooldridge (1996). This model extends
the generalized linear model by accounting for the bounded nature of the
data, which is a proportion quantity.
The article is organized as follows. We begin by providing a
description of the parliamentary system and development of politics in
Malaysia since its independence. The next section describes the sources
of the data and explains the methodology used to model the proportion of
votes won by BN, and how the ethnicity and urbanization impacts can be
quantified. We then present and discuss the estimation results,
concluding with the implications of the results. This article, however,
for reasons explained in the next section, only includes election data
from Peninsular Malaysia and excludes the two Borneo states of Sabah and
Sarawak.
Malaysian Politics and the Electoral System
Historical Background
Peninsular Malaysia, or Malaya as it was then known, became
independent from British colonial rule in 1957. In 1963, Singapore and
the Borneo states of Sabah and Sarawak merged with Malaya to form a new
nation called Malaysia. However, two years later, Singapore separated
from Malaysia due to political differences. Since then, Malaysia
consists of thirteen states: eleven in Peninsular Malaysia (i.e., the
old Malaya) and the two states of Sabah and Sarawak. (3)
In our study we focus on electoral data only from Peninsular
Malaysia for several reasons. First, the main source of economic and
political power lies in Peninsular Malaysia due to historical reasons,
the level of development, and concentration of population (Khoo 2013).
Second, the political parties operating in the Borneo states are to a
certain extent autonomous from those in the peninsula (Noor 2013). Thus,
there is a distinct possibility that whichever party or coalition wins
in the peninsula can persuade politicians who have won on opposition
party tickets in Sabah and Sarawak to join their coalition (Yusoff
2001). Finally, the ethnic composition of Sabah and Sarawak is quite
different from that of Peninsular Malaysia. In particular, the ethnic
makeup of Sabah and Sarawak is much more diverse (Sim 2010). Therefore,
ethnic factors exercise a much stronger influence on the electoral
outcomes in the peninsula relative to Sabah and Sarawak. Of the 222
federal constituencies, 165 (or nearly 75 percent) are from Peninsular
Malaysia. It is for these reasons that this study is confined to the
peninsula for the remainder of this article.
Malaysia is ethnically a very diverse nation. The indigenous ethnic
groups are classified as Bumiputera (which can be transliterated as
"son of the soil") and this classification includes the
Malays, Orang Asli (who are the aboriginal people in Peninsular
Malaysia), and the various indigenous ethnic groups in the Borneo states
of Sabah and Sarawak. According to the 2010 census, the Bumiputera
community made up 67.4 percent of the total Malaysian population. The
other two major ethnic groups are the Chinese and Indians, who made up
24.6 and 7.3 percent of the population, respectively.
Table 1 presents the population distribution by ethnic groups from
1957 to 2010 in Peninsular Malaysia. As can be seen from the table, at
the time of independence in 1957, there was almost equal representation
of Malays (4) and non-Malays. Therefore, there was considerable fear
among the Malays that they would be swamped by the Chinese and Indian
immigrants, and their descendants (Abdullah 1997). (5) To partly assuage
these fears, the British colonial authorities, together with the local
political elites, ensured that the special position of the Malays and
other indigenous people was enshrined in the proposed constitution for
the new nation by designating Malay as the national language, Islam as
the official religion, and a quota for Bumiputeras in public sector jobs
and scholarships for higher education, among other items.
The Electoral System and General Elections
Malaysia has gone through thirteen general elections since
independence. Table 2 provides a summary of the election results for all
thirteen general elections. The last column measures the distortion
effect, which is defined as the difference between the percentage share
of seats that BN won and the percentage of popular votes it received.
(6) The Alliance/ BN benefited significantly from the distortion effect
during the first eleven general elections. (7)
There were several reasons for this success. First, the rapid
economic growth that Malaya, and later Malaysia, enjoyed during the
first forty years of independence uplifted the livelihood of most
Malaysians. In particular, universal education and health care were
provided at very low cost to all citizens. Physical infrastructure
improved tremendously. This track record of economic prosperity and the
ensuing political and ethnic stability enabled BN to claim credit for
this growth (Mutalib 2000).
Another reason for BN's success was the divisions among the
opposition parties. From 1957 to 1998, the opposition was made up of
fragmented individual parties choosing to go to the polls without any
formal electoral pact (Ufen 2009). By operating as a multiethnic
coalition, Alliance/BN was able to occupy the middle ground in the
ethnic spectrum and not portray itself to be an ethnic-based party.
However, opposition politics in Peninsular Malaysia has been dominated
by two ethnic-based parties: Parti Islam Se-Malaysia (PAS), an Islamist
party drawing its support almost exclusively from the Malay-Muslim
community, and the Democratic Action Party (DAP), which gets most of its
support from the non-Malays. This fragmentation, together with the
first-past-the-post system, enabled BN to win a large number of seats
comfortably, even though its share of votes was between 50 and 65
percent for each of the first eleven elections. In fact, the distortion
effect mentioned earlier was amplified mainly due to the ability of BN
to hold the middle ground and win convincingly in ethnically mixed
seats.
A third reason for BN's long string of victories was the
increasing weightage given to rural and ethnically mixed constituencies
over several delineation exercises (Lim 2002). In Malaysia, the rural
areas of the country are mostly populated by Malays. Over the years,
there has been constant gerrymandering by the Election Commission (EC)
(Lim 2002), which is tasked to delineate the apportioned constituencies
based on Article 46 of the Malaysian Federal Constitution. The
constitution, however, has left it to the EC to interpret and apply
important but vague and undefined terms such as "a measure of
weightage," "rural," and "urban" when
delineating electoral constituencies. The EC's application of rural
weightage has been a source of controversy as it takes precedence over
the fundamental principle of equal size (Lim 2002).
However, the situation changed dramatically since the twelfth
general election (GE12) held in 2008. BN's share of popular votes
in Peninsular Malaysia, which has been more than 50 percent since 1974,
has declined significantly in the last two general elections. A
plausible reason for this shift was the emergence of a multiethnic
opposition coalition that provided an alternative vision for the
country's future that all communities could accept (Leong 2012). In
GE12, the former deputy prime minister, Anwar Ibrahim, who had been
expelled from the government and jailed in 1998, was able to forge an
electoral pact between his newly formed party, Parti Keadilan Rakyat
(PKR), and the two leading opposition parties (i.e., DAP and PAS) in the
peninsula. This informal group then went on to produce a common election
manifesto and campaigned jointly in many parts of the country. The
election result was stunning. BN suffered a severe setback in terms of
both the number of seats as well as the share of popular votes. More
damaging, BN lost control of five of the eleven state governments in the
peninsula. Among them were the most industrialized and urbanized states
of Selangor, Penang, and Perak. BN was also routed in the federal
territory of Kuala Lumpur, the commercial capital and largest city in
Malaysia. However, within a year, the PR state government in Perak was
toppled when three of its state assembly members crossed over to BN.
Following this somewhat unexpected success in 2008, the electoral
pact was transformed into a coalition called the Pakatan Rakyat (PR) or
People's Alliance, made up of Anwar Ibrahim's PKR, DAP, and
PAS. During the past five years, PR's component parties have been
able to attract a number of young leaders from the different ethnic
groups in the country. State financial coffers have been substantially
improved through good governance in the four PR-governed states of
Kelantan, Kedah, Penang, and Selangor (Auditor General of Malaysia
2011a, 2011b). (8) In the latter two states, there was a marked
improvement in administrative efficiency in the state government
machinery. In spite of the BN-controlled federal government's
continuous attempt to undermine this coalition (Liow and Pasuni 2010),
PR has been able to maintain its unity up to the GE13, held in May 2013.
While BN still had the advantage of using the extensive federal
government machinery to help its election campaign, as well as had
control over the mainstream media, the emergence of Internet-based news
portals and blogs leveled the playing field to some extent (Gomez and
Chang 2013; Pepinsky 2009; Rajaratnam 2009).
The election campaign was very intense, because for the first time
in Malaysia's electoral history there was no assurance that BN
would be able to retain its parliamentary majority. Ultimately, the
results were disappointing to Pakatan Rakyat. As mentioned above, while
there was no change in the number of seats won by both sides in
Peninsular Malaysia, there was a shift in the voting pattern. In
particular, BN made gains in some rural areas while it lost few urban
seats.
Ethnicity and Urbanization
The foundations of participatory political systems arose from rapid
industrialization. Weber (1978) argued that modern political upheaval
can be traced back to economic changes and shifting populations.
Nevertheless, Lemer (1964) hypothesized that a modernizing society
needed to achieve a certain minimalist threshold of urbanization before
a participant society can be seen to emerge. Urbanization also results
in the growth of modern interest groups whose demands and opposition to
political elites lead to expansion of political communities. In summary,
rapid urbanization results in a decentralization of political power
(Lipset 1959).
Urbanism does not just involve growth of cities. It also leads to
broader social transformation. Specifically, it leads to the
mobilization of both the rural and urban electorate by politicians.
Thompson (2013) argues that when there is a lack of an electoral
process, rural identities and the cultural discrimination felt by people
of rural origin may be a source of social unrest. A clear example of
this is the prolonged political impasse ongoing in Thailand. However,
when a proper electoral process is in place, a flourishing two-party
system may develop. An early study by Cutright (1963) in the United
States affirmed that increasing urbanization seems to be conducive for
the development of a competitive party system. There have also been
documented cases of changes in voting patterns due to urbanization in
other developed countries such as Turkey (Shmuelevitz 1996) and Italy
(Fried 1967).
Urbanization has also been linked to political instability. Given
the large influx of migrants into urban areas, the lack of planning of
amenities by urban authorities often leads to an expectation gap between
the dream of city life and the harsh realities that these migrants face.
This in turn causes relative deprivation and social-psychological
maladjustment, which leads to political radicalization and support for
protest movements where grievances are aired through demonstrations.
Such scenarios are prevalent not only in Latin America (Cornelius 1969)
but also in Africa as well. Resnick (2012) classifies this marginalized
group as the urban poor and found that this group tends to be
politically mobilized through the use of populist strategies by
opposition parties in a number of African countries.
Compared to the literature on urbanization, there has been more
research on the role of ethnicity in the determination of electoral
outcomes. Ethnicity is a frequently used tool to galvanize political
support. Its use is especially prevalent when there is a sudden
democratic transition (Ottaway 1999). In such situations, voters may
resort to voting along ethnic lines as they may feel likelier to receive
greater benefits by voting for a politician of the same ethnicity,
rather than for someone outside of the ethnic group (van de Walle 2007).
This so-called cognitive shortcut is more prevalent among those with
lower education attainment and as such it may be difficult for those
individuals to distinguish the differences between political parties
(Noris and Mattes 2003).
The question of whether ethnicity remains a significant factor in
the urban context has been a subject of debate among political analysts.
One school of thought maintains that urbanization actually leads to
greater interethnic competition over scarce, but highly visible,
resources and opportunities (Bates 1983; Melson and Wolpe 1970).
Urbanization, however, could contribute to more cosmopolitan worldviews
that may nullify the ethnicity effect (Lipset 1959; Parsons 1975). This
could lead to what Thompson (2013) refers to as an "urban
cosmopolitan chauvinism" bias whereby urbanites perceive themselves
as far superior to their rural peers. Also, prolonged interaction with
other ethnic groups increases awareness of commonly shared
characteristics, whereas infrequent contact can reinforce hostile ethnic
stereotyping (Allport 1979).
Prior studies (Feagin 1972; Mohd Fuad et al. 2011) have analyzed
the effects of ethnicity and urbanization independently and researchers
have tended to treat both factors as additive rather than interactive.
In contrast, Whitby (1985) introduced an interaction term consisting of
both these factors to analyze the voting patterns of politicians elected
to the US Congress. He concluded that the interaction between
urbanization and an increasing proportion of the black population led to
more liberal voting behavior on legislation passed by the US Congress.
Urbanization therefore conditions the effect of race.
To understand the role of ethnicity and urbanization in determining
the electoral outcome in the context of a developing country, one has to
examine the demographic development of Malaysia since its independence
in 1957. During British colonial rule and immediately after
independence, agriculture was the dominant sector of the economy. During
that period, most urban centers in the country had large Chinese
majorities. Given that most Malays were employed in the agricultural
sector, they lived predominantly in rural areas. Since independence,
Malaysia has seen rapid modernization and industrialization with
manufacturing taking over from agriculture as the main contributor to
the economy. Rapid industrialization gave impetus to the movement of
population from rural villages to towns and cities. According to Yaakob,
Masron, and Fujimaki (2012), Malaysia has experienced rapid growth in
urbanization during the last five decades, but urbanization was more
pronounced during the 1980s and 1990s. Levels of urbanization have grown
from 28.4 percent in 1970 to 71 percent in 2010 (Department of
Statistics 2011). Malays now form the majority in most urban centers in
Peninsular Malaysia because of the rural-urban migration that has taken
place (McGee 2011).
Following a racial riot that occurred after the 1969 general
election, the government began the implementation of the New Economic
Policy (NEP), a legislated affirmative action policy meant to rectify
the economic imbalance between the Malays and the non-Malays (Chin
2001). The implementation of the NEP aided by rapid industrialization
and economic development has resulted in the rise of a Malay middle
class that is increasingly less dependent on affirmative action but more
interested in universal issues like participatory democracy, justice,
and human rights (Saravanamuttu 2001).
Middle-class Malays are now better educated, better informed, with
better access to alternative media (Pepinsky 2009). They are now more
economically independent and located in urban areas and are at ease with
interethnic economic and social relationships (O'Shannassy 2009).
They, together with their non-Malay counterparts, have been pressing the
cause for democracy more fervently under a nascent multiracial platform
(Liow 1999). This cohort opted for change and is located in urban areas
(Welsh 2013). The rise of the middle-class Malays culminated in the
significant loss of BN's two-thirds majority in the 2008 and 2013
elections, respectively, as seen in Table 2 above (Khoo 2013).
Several factors have also contributed to the rising discontent
among Malaysians across racial divisions. These include rising crime,
corruption scandals, weakness of the judicial system, and increasing
cost of living in general (Moten 2009). Other studies using a
sociological approach argue that economic growth and the pressures of
materialism and urbanization have strengthened rationalism and weakened
ethnic considerations in deciding whom to vote for (Mansor 1992, 1999).
Indeed, as a country moves into the middle-income range during
economic transition, it leads to changes in the social structures,
beliefs, and culture that foster democracy. Huntington (1991) refers to
this as a "Third Wave" of democratization. The recent election
results have also triggered debate of whether we have indeed seen the
end of ethnic politics in Malaysia. It is without doubt that parties
canvassing on a pure ethnic political platform have been severely
weakened, but to say that ethnic politics have been totally extinguished
in Malaysia would be premature (Arakaki 2009; Balasubramaniam 2006; Lian
and Appudurai 2011; Moten 2009; Noh 2014; Pepinsky 2009).
Nevertheless, the uncoupling of ethnic interests from ethnic
identity of their advocates is steadily progressing as exemplified by
the actions undertaken by opposition parties in power at the state
level. Politicians from both sides of the political divide are beginning
to see the benefits of addressing sensitive or racially charged issues
as an across-the-board responsibility (Mohamad 2008). This is evidenced
in the last two elections with the emergence of a two-party coalition
system with the charismatic Anwar Ibrahim leading a credible opposition
coalition that has presented itself as an alternative to the ruling BN
coalition (Mohamad 2008; Moten 2009). The emergence of a two-party
coalition system may be linked to the rapid urbanization of the country.
Thompson (2013), however, asserted that although the current
discursive argument points to a sharp urban-rural divide between city
and village dwellers, it ignores the fact that there is substantial
movement between rural and urban voters. He points out that although PAS
used the pervasive urban cosmopolitan chauvinism tactic in mobilizing
the rural dwellers against its arch-enemy BN, it has not been an
outright success. In contrast to the Mahathir administration, which has
been accused of neglecting rural development (Thompson 2013), the former
prime minister Abdullah Badawi and the current prime minister Najib
Razak have made the interests of rural Malaysia a key cornerstone in
their policymaking decisions. They realize that there is much to lose if
rural Malaysia continues to be ignored in the name of economic progress.
Increasing urbanization has affected the ethnic composition of
urban areas in Malaysia. Thus, the question is, which of these two
factors is leading to the changes in the electoral dynamics of
Peninsular Malaysia? An econometric methodology to separate and quantify
the marginal effect of the urbanization-ethnicity interaction is
presented in the next section.
Data and Methodology
Data
The dependent variable in the model is defined as the proportion of
votes won by BN in each of the 165 parliamentary constituencies
considered in this study, calculated as the number of votes that BN
garnered for a parliamentary constituency, divided by the total number
of valid votes for that corresponding parliamentary constituency. These
data were obtained from the electoral results published online by the
Malaysian Election Commission. (9)
The explanatory variables in the model specification include (1)
the total number of eligible voters from each of the four ethnic groups
in each parliamentary constituency, classified as Bumiputera, Chinese,
Indians, and Others; and (2) Area--the physical area of the
parliamentary constituencies that is used as a continuous measure of
urban development. The ethnic group classification of Bumiputera
encompasses the Malays, Orang Asli (which refers to the aborigines of
Peninsular Malaysia), Bumiputera Sabah, and Bumiputera Sarawak.
Using The Star newspaper (10) (April 21, 2013), which had obtained
data from the Malaysian Election Commission on the percentage
composition of the four ethnic groups in each of the 165 parliamentary
constituencies, the total number of eligible voters from each of the
four ethnic groups was computed by multiplying the respective ethnic
percentage composition with the total number of eligible voters in that
particular constituency. This study does not directly use the percentage
composition of the four ethnic groups provided by The Star as the
explanatory variables primarily because the use of such compositional
explanatory variables (with a unit sum constraint) in a regression
analysis, even for a linear regression model, may lead to biased and
incorrect estimates, resulting in misleading inferences (Hron,
Filzmoser, and Thompson 2012). (11)
Data on the physical area of the parliamentary constituencies were
obtained from Greenberg and Pepinsky (2013), and have also been used by
Ostwald (2013) in his analysis of the relationship between district size
and voter density across the Malaysian parliamentary constituencies.
In addition to using Area as the continuous measure of
urbanization, we also use data from Politweet that categorize the 165
parliamentary constituencies as either urban, semi-urban, or rural to
aid our analysis. (12) According to Politweet, a constituency is defined
as rural if there are villages, small towns, or farmlands distributed
within the constituency. Semi-urban constituencies are those that
contain larger towns and/or numerous small towns as compared to a rural
constituency, and may also contain small villages. Urban constituencies
refer to cities where urban development occurs in a majority of the
constituency. The Politweet classification was done based on Google Maps
satellite imagery and the Election Commission of Malaysia maps. The
three-step methodology in determining the level of urban development of
a constituency can be found in Appendix 1. Based on this classification,
the number of rural, semi-urban, and urban constituencies is eighty-one,
forty-four, and forty, respectively.
By combining the data on the physical area of the constituencies
with the Politweet urbanization classification, Figure 1 plots the areas
of the 165 parliamentary constituencies in their respective urban
development classification, with the areas ranging from 0.0005 to
0.66433. (13)
It can be seen from Figure 1 that urban constituencies are
associated with smaller areas, while rural constituencies are associated
with larger areas. In particular, the median areas for the rural,
semi-urban, and urban constituencies are 0.060, 0.037, and 0.003,
respectively. Therefore, it is reasonable to conclude that parliamentary
constituencies that are larger in physical size are also more rural in
nature, while parliamentary constituencies that are smaller in physical
size are more urban in nature. In other words, along the area continuum,
constituencies with small areas refer to urban constituencies while
constituencies with large areas refer to rural constituencies.
[FIGURE 1 OMITTED]
Methodology
The dependent variable, proportion of votes garnered by BN, is a
proportion quantity restricted to an interval of 0 and 1 (or referred to
as a unit interval). Therefore, the model specification for a fractional
dependent variable, [y.sub.i], as described by Papke and Wooldridge
(1996), is assumed as follows:
[y.sub.i] = F([x'.sub.i] [beta]) + [u.sub.i], i = 1,2, ..., n
(1a)
E([y.sub.i]|[x.sub.i]) = F([x'.sub.i], [beta]), i = 1,2, ...,
n (lb)
where [x.sub.i] are the explanatory variables for the ith
observation and F(.) is typically chosen to be a known cumulative
distribution function satisfying 0 < F([x'.sub.i] [beta]) < 1
for all [x'.sub.i] [beta] [member of] R, ensuring that the
predicted values ofy will lie between the unit interval. The term
[u.sub.i] is a random error term with a conditional mean of 0. Equation
la decomposes the observation into two components--the deterministic (or
predictable) component given by F([x'.sub.i] [beta]) and a random
(or unpredictable) component given by [u.sub.i]. Equation lb is derived
by taking the conditional expectations of Equation la, and it models the
conditional expectation of [y.sub.i] (i.e., the average value of y,
given a set of x; values) as a deterministic function given by
F([x'.sub.i] [beta]). It is the conditional expectation function in
Equation lb that is estimated.
The vector of parameters in Equation la, [beta], is estimated using
the Bernoulli quasi-maximum log-likelihood estimator (QMLLE), with the
Bernoulli log-likelihood function given by
[l.sub.i] ([beta]) = [y.sub.i] log [F([x'.sub.i] [beta]] + (1
- y) log [1 - F([x'.sub.i] [beta])] (2)
The QMLLE of [beta] in this model is consistent and asymptotically
normally distributed irrespective of the conditional distribution of
[y.sub.i]. In this article, F(*) is chosen to be a logistic function
that is strictly monotonic and defined as
F(z) [equivalent to] exp(z) / [1 + exp(z)]' (3)
where z is defined as [x'.sub.i] [beta] from Equation la. The
model specification in Equation la, coupled with this logistic function
in Equation 3, is hence referred to as the fractional response logit
model. (14)
The nonlinear model specification in Equation 1 a is chosen over
the linear specification because linear regression models are not
appropriate for bounded values. Therefore, using the linear regression
model will result in model misspecification with inefficient estimators,
thus leading to incorrect inferences (Hron, Filzmoser, and Thompson
2012). Moreover, predicted values from a linear specification are not
guaranteed to lie in the unit interval. Linear regression techniques
also assume that the effect of any explanatory variable is constant
throughout the range of values of the explanatory variable.
Further, by specifying the fractional response model in Equation
la, no a priori assumption on the range of values that the dependent
variable can take is made, except that it must be within the unit
interval. (15) In doing so, the model specified in Equation la accounts
for all possibilities, including the possibility of a constituency
observing a value of either 0 or 1 for the dependent variable. It should
also be noted that the fractional response model is used not only when
the dependent variable records extreme values of 0 or 1, but that it has
an added benefit of being able to deal with such values, avoiding the
previous practice of ad hoc transformations.
Model Specification
Using Equation lb, the expected proportion of votes garnered by BN
is modeled as a function of the number of voters in each of the four
different ethnic groups, and the area of the parliamentary
constituencies that is used as a continuous measure of urban
development. This model is further augmented by including interaction
terms between the variables. The model is subsequently represented as
such:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)
where Bumiputera, Chinese, Indians, and Others represent the number
of voters (in 10,000s) from the respective ethnic groups of the
parliamentary constituencies, and Area is the proxy to measure the level
of urban development for the parliamentary constituencies.
The present study improves earlier studies (Feagin 1972; Mohd Fuad
et al. 2011) by including the interaction effects of urbanization on
ethnicity in determining the proportion of votes garnered by BN. The
inclusion of interaction terms accounts for the synergy effect of both
ethnicity and urbanization, thereby allowing for the complementary
effect of urbanization on ethnicity to influence the proportion of votes
garnered by BN. Moreover, Brambor, Clark, and Golder (2006, 64) note
that when analyzing political studies, interaction terms should be
included whenever the analysis involves a conditional hypothesis,
defined as when "a relationship between two or more variables
depends on the value of one or more other variables."
Results
Estimation Results
The model specified by Equation 4 is estimated by quasi-maximum
likelihood methods and the results are presented in Table 3.
Before interpreting the results, a few points are in order with
regard to the coefficients. First, for a fractional response logit
model, unlike linear regression models, [[beta].sub.j] alone does not
represent the marginal effect of the explanatory variable [x.sub.j].
Appendix 2 details how the marginal effects of an explanatory variable
of interest can be computed. Second, the sign of the coefficient
[[beta].sub.j] for this class of models indicates the direction of the
marginal effects of the explanatory variable on the dependent variable.
Taking the above two points together, only the signs of the coefficients
in Table 3, and not the magnitudes, are interpretable.
For the purpose of this article, the discussion of results will be
centered on the three main ethnic groups that are the most important in
Peninsular Malaysia politics: the Bumiputera, Chinese, and Indians.
Referring to Table 3, the results show that of the three main ethnic
groups, only the Chinese ethnic group is a statistically significant
variable, bearing a negative coefficient. Area was found to be
statistically insignificant at the conventional level of significance.
This result implies that the degree of urban development of a particular
constituency, by itself, does not influence the level of support for the
government. However, the positive and statistical significance of the
interaction term between Area and Chinese reveals that there is a
complementary effect between these two variables. The results imply that
there is a significant extra effect of urbanization (as proxied by Area)
on the Chinese voters to influence the proportion of votes garnered by
BN, ceteris paribus. More specifically, the positive coefficient of this
interaction term implies that for every additional Chinese voter in a
parliamentary constituency, ceteris paribus, the increase in the
proportion of votes to BN is higher in larger (rural) parliamentary
constituencies than smaller (urban) parliamentary constituencies. This
positive effect negates the decrease in the proportion of votes to BN
arising from the negative coefficient of Chinese.
However, the other ethnic group variables (Bumiputera and Indians),
and their interaction with Area, are all statistically insignificant.
This implies that an increase in either Bumiputera or Indian voters in a
parliamentary constituency, ceteris paribus, will result in no change in
the proportion of votes to BN. There is also no added effect that comes
from the level of urbanization in that constituency. Therefore, it seems
that the voting intentions of the Bumiputeras and Indians are constant
irrespective of the level of urban development of the parliamentary
constituency that they are in.
The results suggest that the Chinese-Urbanization factor is having
the most dominant influence on the proportion of votes garnered by BN.
The next subsection provides plots to visualize the underlying patterns
in the ethnicity and urbanization effects on the proportion of votes to
BN. These plots provide interesting insights into understanding the
underlying relation between these variables and reveal deviations from
patterns that may not be detected easily.
Ethnicity and Urbanization Effects
Using the estimation results in Table 3, Figure 2 provides a
visualization of the ethnicity and urbanization effects by plotting the
marginal effects of the number of voters from each ethnic group on the
predicted proportion of votes to BN, across the range of areas of the
parliamentary constituencies. Specifically, Figure 2 plots the change in
the predicted proportion of votes to BN that arises from an increase in
the number of voters from each respective ethnic group, for the range of
areas of the parliamentary constituencies. The horizontal axis in Figure
2 represents the range of areas of the 165 parliamentary constituencies
and is used as a continuous measure of urban-rural development, where a
constituency with a smaller area is associated with being an urban
constituency and vice-versa. The vertical axis represents the change in
the predicted proportion of votes to BN that arises from an increase in
the number of voters from each of the ethnic groups.
[FIGURE 2 OMITTED]
Referring to Figure 2, it is evident that regardless of the area of
the constituency, an increase in the number of Bumiputera voters in a
constituency, ceteris paribus, results in no change in the predicted
proportion of votes to BN. Therefore, whether the constituency is an
urban or rural region, an increase in the number of Bumiputera voters in
that constituency, ceteris paribus, does not alter the level of support
for the ruling coalition, BN.
Furthermore, the marginal effects of the Chinese voters are
significantly larger than those of the Bumiputera voters. Given an
increase in the number of Chinese voters relative to the other ethnic
groups, the change in the predicted proportion of votes to BN is
negative in smaller (urban) constituencies while this change is positive
in larger (rural) constituencies. Although the support for BN is
positive in large (rural) constituencies with a higher relative number
of Chinese voters, this level of rural Chinese support dwindles with
larger constituencies, as shown by the tapering of the solid line in
Figure 2. This implies that the level of urban development of a
constituency influences the level of Chinese support for BN.
The predicted change in proportion of votes to BN from an increase
in the number of Indian voters, ceteris paribus, is negative for all
areas of constituencies in Figure 2. However, these marginal effects are
statistically insignificant from zero and therefore an increase in the
relative number of Indian voters in either an urban or rural
constituency will not influence the level of support for BN.
Ethnicity and Urbanization Effects: An Extended Analysis Using the
Categorical Measure of Urban Development
The earlier analysis used area of constituencies as the continuous
measure of urban development and subsequently plotted marginal effect
graphs to illustrate the ethnicity and urbanization effects on the
proportion of votes to BN. To extend and complement the former analysis,
we also show the ethnicity and urbanization effects on the proportion of
votes to BN by using the categorical measure of urban development where,
as discussed earlier, the parliamentary constituencies are classified as
urban, semi-urban, or rural. Specifically, for each ethnic group, we
plot and observe the predicted proportion of votes to BN in urban,
semi-urban, and rural constituencies, over the range of number of voters
from that ethnic group. To do this, we substitute, in turn, the median
areas representing each of the region classifications (0.003 for urban,
0.037 for semi-urban, and 0.060 for rural) into Equation lb and obtain
the predicted proportion of votes to BN for the range of voter numbers
in each ethnic group. Figures 3-5 subsequently display the results.
In Figure 3, the predicted vote share for BN across the three
region classifications declines very slightly throughout the range of
Bumiputera voter numbers. Therefore, since an increase in the number of
Bumiputera voters, ceteris paribus, changes the predicted proportion of
votes to BN only marginally in either an urban, semi-urban, or rural
constituency, it can be inferred that for the Bumiputeras, ethnicity has
little influence on the level of support for BN. It is notable that in
all three constituency types, the Bumiputera predicted vote to BN is
less than 50 percent. However, it is evident from Figure 3 that it is
the level of urban development that influences the level of support for
BN among the Bumiputeras. Regardless of the number of Bumiputera voters
in a given constituency, the model predicts that the proportion of votes
to BN will be the lowest in an urban constituency, and it will be the
highest in a rural constituency.
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
Figure 4 plots the predicted vote share to BN across the three
region classifications over the range of Chinese voter numbers,
revealing the presence of both ethnicity and urbanization effects. At
small Chinese voter numbers (around 5,000) in a given constituency, the
predicted proportion of Chinese votes to BN is approximately 55 percent,
regardless of whether the constituency is urban, semi-urban, or rural.
However, the negative trend lines for all three region classifications
show that the level of support for BN decreases quite rapidly as the
number of Chinese voters increases, thus capturing the ethnicity effect.
This ethnicity effect is further fueled by the urbanization factor, as
seen by the divergence in the predicted proportion of votes to BN across
the three region classifications, with urban constituencies reflecting a
sharper drop in the proportion of votes as compared to rural
constituencies. In particular, for the same increase in the number of
Chinese voters from 5,000 to 75,000, the predicted proportion of votes
to BN in an urban constituency drops from 55 percent to approximately 20
percent, as compared to the drop to approximately 33 percent for a rural
constituency.
In contrast to the divergent trend lines in Figure 4 for the
Chinese voters, Figure 5 shows the trend lines across the three region
classifications for the Indian voters converging. In a constituency with
a low number of Indian voters, the predicted proportion of votes to BN
is the highest if it is a rural district, and is the lowest if it is an
urban district. This disparity in the vote share to BN across the three
region classifications diminishes in constituencies with a higher number
of Indian voters, with the predicted vote share to BN decreasing for
both the rural and semi-urban constituencies. The predicted vote share
to BN for an urban constituency, however, remains constant over the
range of Indian voters. It is also noted that for constituencies with
fewer than approximately 25,000 Indian voters, the predicted proportion
of votes to BN is the lowest in urban areas, reflecting the urbanization
effect on the support for BN.
However, the findings from Figure 5 for Indian voters should be
treated with some caution. In most constituencies, the Indian share of
votes is relatively small. The Kota Raja parliamentary constituency has
the highest number of Indian voters with a 29.5 percent share. Another
nine parliamentary constituencies have 20-24 percent of Indian voters,
while in fifteen other parliamentary constituencies, they form 10-20
percent of the voters. Thus out of 165 parliamentary seats in Peninsular
Malaysia, Indian vote share exceeds 10 percent only in twenty-five
parliamentary seats, thus reducing their impact on the electoral outcome
in most parliamentary seats.
[FIGURE 5 OMITTED]
A few additional points are also observed from Figures 3, 4, and 5.
First, across all the ethnic groups, the support for BN is always the
lowest in urban constituencies and the highest in rural constituencies.
This is a clear indication that the level of urban development
influences the level of support for BN, with more urbanized
constituencies being pro-opposition. Second, as shown in Figure 3, the
Bumiputera vote is evenly split with a slim majority for the opposition
Pakatan Rakyat. In the aggregate, Bumiputera support for BN ranges from
46 percent in urban areas to 49 percent in rural areas. Third, the
ethnic effect is the strongest among the Chinese, with support for BN
declining sharply in constituencies that have a larger number of Chinese
voters relative to the other ethnic groups.
Discussion
GE13 was arguably the most closely contested election in Malaysian
history. In terms of popular votes, GE13 was the worst outcome for the
ruling coalition in BN's history. For Malaysia as a whole, PR
obtained more votes than BN. In Peninsular Malaysia, BN obtained only
45.7 percent of the votes cast, while PR obtained more than 54 percent
of the votes. It is also intriguing to note from Figure 3 that even
among Bumiputera voters, BN was unable to obtain majority support. Among
the non-Bumiputera, a clear majority voted for PR. Yet, BN won
eighty-five seats to PR's eighty seats. A large proportion of
BN's electoral success, however, can be attributed to victories in
rural parliamentary seats as classified by Politweet. Of the eighty-five
seats BN won in Peninsular Malaysia, sixty-six were classified as rural.
The key to understanding this outcome is to examine Chinese voter
behavior in constituencies where their numbers are small. As can be seen
from Figure 4, when the Chinese voters make up a small number in
absolute terms in urban, semi-urban, and rural constituencies, their
support for BN is more than 50 percent. The higher than 50 percent
support for BN by Chinese voters appears to offset the slightly less
than 50 percent Malay support for BN in rural areas, thus helping BN
over the finish line in rural parliamentary constituencies. So,
ironically, given that most of BN's eighty-five electoral seats are
from the rural constituencies, BN's victory in the thirteenth
general election can be partly attributed to the support of Chinese
voters, and to a lesser extent, Indian voters, in constituencies where
they are small in number. BN won sixty-one of the sixty-six rural seats
where Chinese voters numbered 20,000 or fewer. This rural effect is
further amplified by the malapportionment of voters between rural and
urban constituencies. The variance between rural and urban seats has
been increasing over several rounds of electoral delineations (Lim 2002;
Ostwald 2013).
It is not clear why the Chinese support for BN increases when their
numbers decrease in electoral constituencies (see Figure 4). One
possible explanation is that where Chinese voters make up a small
percentage of the electorate, they may tend to feel less secure. As
their numbers increase, they may lose their sense of insecurity, and
thus their support for BN declines sharply to as much as only 20
percent, especially for urban seats. Second, in rural constituencies,
the role of government development spending has a bigger impact on the
economic well-being of voters. This may influence their preference for
BN among Chinese voters in rural areas. The motivation for the change in
Chinese voter behavior as their numbers change needs to be explored in
greater depth.
What are the implications of these findings for BN and PR? As the
country transitions from a developing nation to a developed nation,
Malaysia will become increasingly urbanized. Rural and semi-urban areas
will become increasingly more urban in character as development moves
beyond urban areas. This may result in BN's vote share decreasing
across all ethnic groups. It is also compounded by the fact that younger
voters tend to read news on the alternate media over which the BN
government has very little control. Future delineations that skew
rural-urban constituency sizes are becoming more difficult due to a more
aware electorate. (16) Thus, BN has to adopt a strategy that accounts
for these factors.
BN has to refocus on urban areas for success in future elections.
BN needs to reach out to urbanites of all ethnic groups and not treat
urban seats as a lost cause. Less practice of divisive politics would be
one step in the right direction. They should also woo both the Malay and
the non-Malay votes in rural and semi-urban areas where they have some
advantage in these areas. This should be especially targeted toward the
Chinese voters since their voting patterns in rural and semi-urban areas
play a decisive role.
Pakatan Rakyat has a clear decisive advantage in urban
parliamentary seats. Support from all ethnic groups is above 50 percent.
Urban parliamentary seats are therefore the easiest for PR to retain
with popular support. Therefore, the strategy PR should employ to
maintain urban support would be the provision of good governance,
especially efficient municipal services in states where PR is the ruling
coalition (Penang, Selangor, and Kelantan). This will ensure that there
will be no swing back to BN in the coming general election. Contrary to
BN's less than decisive advantage in rural and semi-urban seats, PR
has opportunities to make inroads in these two seat classifications.
Early identification of candidates for each rural and semi-urban
constituency is imperative and it would be especially advantageous if
the identified candidates were local politicians. Early identification
of candidates would reduce political bickering among would-be aspirants
and allow identified candidates to work in their respective
constituencies in building up their support base.
Conclusion
More recently, the debate of ethnicity's having an effect on
voter sentiment has attracted considerable interest in a multiethnic
society such as Malaysia. However, the findings in this article suggest
that there is more to it than ethnic undertones. This article has made
two contributions. First, although the results of Malaysia's GE13
displayed an ethnic effect, the rapid urbanization of the country has
also played a role in determining the outcome of the elections.
Malaysians across ethnic lines voted overwhelmingly for PR in urban
areas, whereas support for BN remains reasonably strong in rural areas.
This article also sheds new light on the long-standing explanation that
BN's continued dominance in Malaysian general elections is due to
the solid rural Bumiputera vote. Our analyses have shown that
non-Bumiputera votes in rural and semi-urban areas were the key to
BN's holding on to power despite having slightly less than 50
percent support from Bumiputera voters, even in rural constituencies.
However, reliance on the non-Bumiputera voter group is highly precarious
due to the shrinking number of Chinese voters in rural areas. This is
caused by rural-urban migration and the decline in birthrates among the
Chinese population (Department of Statistics 2012).
Second, the fact that Malaysia is becoming increasingly urbanized
also needs to be taken into account. Political parties from both sides
of the political divide need to take heed of this development. As the
country develops further toward becoming a high income nation, voting
patterns for race-based parties may decline significantly. Instead,
political parties that can identify with the rising demands of voters on
bread and butter issues like the increasing cost of living, social
justice, corruption, and basic human rights may gain the upper hand in
future general elections.
A cautionary remark, however, is in order where the main point of
contention would lie in the categorization of seats into urban,
semi-urban, and rural constituencies done by Politweet. They have
acknowledged that a better method can be used for the categorization of
parliamentary constituencies through a gridding process (Balk et al.
2006). This study looked at the impact of urbanization and ethnicity on
electoral support at the parliamentary level in aggregate. Local
micro-factors that may have influences on the results of individual
seats were not considered in our analysis. Therefore, future studies may
look at the impact of urbanization and ethnicity at the state
legislative level, which may provide further insight to the results
obtained at the federal level.
Jason Wei Jian Ng is lecturer with the Department of Econometrics
and Business Statistics at Monash University Malaysia and can be
contacted at jason.ng.wj@monash.edu. He received his PhD in econometrics
from Monash University Australia. He is an early career researcher whose
research focuses on applied econometrics in a wide range of policy
issues such as financial development and economic growth, subjective
well-being, and residential house pricing.
Gary John Rangel is lecturer of finance at Monash University
Malaysia and can be contacted at gary.rangel@monash.edu. He received his
PhD in finance from University Science Malaysia. His research focuses on
behavioral finance, real estate, risk perception, and Malaysian politics
and his work has appeared in International Finance Review and
International Journal of Banking and Finance.
Santha Vaithilingam is associate professor at Monash University
Malaysia and is head of the Econometrics and Business Statistics
Department. She can be contacted at santha.vaithilingam@monash.edu. Her
research interests are in computable general equilibrium (CGE),
financial econometrics, and behavioral economics. More recently, her
work has been focused on modeling human behavior pertaining to
technology adoption using advanced econometric techniques.
Subramaniam S. Pillay is associate professor of business and
finance at Taylor's University, Malaysia, and can be contacted at
subramaniam.pillay@taylors.edu.my. He received his PhD in policy
analysis from the University of British Columbia. His research focuses
on economic and currency crisis, and politics in Southeast Asia.
Appendix 1
The following three-step methodology was employed by Politweet in
identifying the level of urban development of the parliamentary
constituencies.
1. Locate the constituency on Google Maps (and Bing Maps, when the
image was not clear).
2. Identify the area covered by urban development, and the degree
of development.
3. Define the seat as either rural, semi-urban, or urban based on
the definitions of the classifications described in the text.
Appendix 2
Using the general model specification in Equation 1, the marginal
effect of an explanatory variable of interest, [x.sub.j], on the
conditional mean E(y|x) is given as
[partial derivative]E(y|x) / [partial derivative] =
f([x'.sub.i][beta]).[[beta].sub.j] (A2.1)
where f[[x'.sub.i] [beta]) is the derivative of the cumulative
logistic distribution function defined as
f(z) [equivalent to] dF(z)/dz = exp(z) / [[1 + exp (z)].sup.2]
(A2.2)
and [[beta].sub.j] is the coefficient attached to the jth
explanatory variable. Therefore, to compute the value of the marginal
effect of [x.sub.j] on the conditional expected value of y using
Equation A2.1, appropriate values of all the explanatory variables, such
as the mean or median value of the explanatory variables, need to be
substituted into the function f[[x'.sub.i] [beta]) before
multiplying it with [[beta].sub.j].
Notes
The authors would like to thank Kai Ostwald for facilitating the
acquisition of the data on the areas of the electoral districts; Tom
Pepinsky for providing the details behind the calculation of these
areas; Peter Filzmoser for his assistance pertaining to our queries on
the isometric log-ratio transformation; Mahendhiran Nair for his
comments on this article; and the editor and the two referees for their
constructive feedback in the reviewing process of this article.
(1.) From 1957until 1973, the coalition was known as the Alliance
Party and was made up of three parties--the United Malays National
Organisation (UMNO), the Malaysian Chinese Association (MCA), and
Malaysian Indian Congress (MIC)--representing the three major ethnic
groups in Peninsular Malaysia (or Malaya, as it was then known). Its
name was changed to National Front or Barisan Nasional in 1971 after it
expanded to become a thirteen-member coalition encompassing political
parties from the Borneo states of Sabah and Sarawak, as well as those
from Peninsular Malaysia.
(2.) The terms constituency and seat are used interchangeably.
(3.) In addition to the thirteen states, there are three federal
territories that are directly administered by the central government:
Kuala Lumpur, the main commercial center of the country; Putrajaya, the
administrative capital; and Labuan Island off the coast of Sabah.
(4.) In Peninsular Malaysia, Malays form nearly 98 percent of the
Bumiputera population; so for all intents and purposes, Bumiputera and
Malay can be used interchangeably.
(5.) These fears did not abate with the granting of independence by
the British. Singapore's founding prime minister Mr. Lee Kuan Yew
in his official autobiography (Lee 1998) indicated this fear still
persisted during the formation of Malaysia in 1963. There was fear that
the inclusion of Singapore in Malaysia would "swamp the
Malays" (Lee 1998, 363).
(6.) If we include Sabah and Sarawak, BN's advantage is even
bigger as the rural overweightage is more pronounced in those two
states.
(7.) The distortion effect calculation is based on Brown (2005).
(8.) Penang recorded a budget surplus of RM13 8 million in 2011, an
increase of 312 percent compared to RM33 million in 2010 when there was
a 95 percent reduction in debt. Selangor also recorded an increase in
revenue from RM1.57 billion in 2010 to RM 1.634 billion in 2011 (Auditor
General of Malaysia 2011a, 2011b).
(9.) The results of the thirteenth general elections for each
partiamentary seat can be obtained from the Malaysian Election
Commission website (http://result
prul3.spr.gov.my/module/keputusan/paparan/paparan_Laporan.php#).
(10.) The Star is a daily newspaper owned by MCA, a component party
of BN.
(11.) In particular, compositional data do not follow the usual
Euclidean geometry where most statistical methods rely on it. Therefore,
Hron, Filzmoser, and Thompson (2012) recommend transforming
compositional variables to the Euclidean geometry via an isometric
log-ratio transformation. The interested reader can refer to that paper
for further details. In this study, we do not attempt to use the
isometric transformed variables because (1) there are interpretability
issues (even in a linear regression setting) and (2) there is currently
no research on how to estimate regressions with interaction variables
that contain both compositional and noncompositional variables, as will
be shown later in our model specification.
(12.) Politweet is a nonpartisan research firm that has been
monitoring Malaysian politics and activism on Twitter since 2009, and on
Facebook since late 2012. See
http://politweet.wordpress.com/2013/05/21/the-rural-urban
-divide-in-malaysias-general-election/.
(13.) The area figures do not contain any units of measurement.
According to Pepinsky, the physical areas of the electoral districts
were created from the Geographic Information Systems (GIS) maps and were
subsequently imported into R whereby the "maptools" library
was used to calculate the area of each polygon representing an electoral
district. The procedure in R does not return any unit of measurement.
Nevertheless, a larger number represents a larger physical area of a
constituency.
(14.) Using either the probit or logit model is an alternative
strategy, but it involves transforming the dependent variable into a
binary form, and will not provide meaningful results in quantifying the
ethnicity and urbanization effects on BN's electoral performance.
(15.) Based on the data collected, the range of the dependent
variable (proportion of votes garnered by BN) is between 0.13 and 0.84.
Therefore, one could argue that given that the dependent variable does
not lie near the unit interval bounds, a linear regression could
suffice. However, doing so will be making an a priori assumption that
the range of values the dependent variable can take is a subset of the
unit interval.
(16.) An increasingly aware electorate has culminated into
nongovernmental organizations (NGOs) seeking to pressure the Election
Commission to ensure a clean and fair electoral process in Malaysia. The
most well-known NGO is the Coalition for Clean and Fair Elections,
popularly known as BERSIH. It has held mass rallies in 2007,
2011,and2013, demanding reform of the postal balloting system, free and
fair access to mass media for all parties, cleaning up of the electoral
roll, elimination of dirty money politics, as well as elimination of
corruption.
References
Abdullah, Firdaus Hj. 1997. "Affirmative Action Policy in
Malaysia: To Restructure Society, to Eradicate Poverty." Ethnic
Studies Report 15, 2: 189-221.
Allport, Gordon W. 1979. The Nature of Prejudice. 25th anniversary
ed. Reading, MA: Addison-Wesley Publishing Company.
Arakaki, Robert K. 2009. "2008 Malaysian Election: The End of
Malaysia's Ethnic Nationalism?" Asian Politics and Policy 1,1:
79-96.
Auditor General of Malaysia. 2011a. "Auditor General's
Report 2011 on the Penang State Government Financial Statement and the
Financial Management of the State Department and Agencies." Kuala
Lumpur, Malaysia: National Audit Department.
--. 2011b. "Auditor General's Report 2011 on the Selangor
State Government Financial Statement and the Financial Management of the
State Department and Agencies." Kuala Lumpur, Malaysia: National
Audit Department.
Balasubramaniam, Vejai. 2006. "Embedding Ethnic Politics in
Malaysia: Economic Growth, Its Ramifications and Political
Popularity." Asian Journal of Political Science 14, 1: 23-39.
Balk, D. L., U. Deichmann, G. Yetman, F. Pozzi, S. I. Flay, and A.
Nelson. 2006. "Determining Global Population Distribution: Methods,
Applications and Data." In Advances in Parasitology, ed. Alastair
Graham, Simon I. Flay, and David J. Rogers, 119-156. Amsterdam, Holland:
Academic Press.
Bates, Robert H. 1983. "Modernization, Ethnic Competition and
the Rationality of Politics in Contemporary Africa." In State
Versus Ethnic Claims: African Policy Dilemmas, ed. Donald Rothchild and
Victor A. Olorunsola, 152-171. Boulder, CO: Westview.
Boo, Su Lyn. 2013. "GE13 an Urban, Not Chinese Swing, Says
Analysts." The Malaysian Insider,
www.themalaysianinsider.com/malaysia/article/ge 13
-an-urban-not-chinese-swing-say-analysts (accessed October 10, 2013).
Brambor, Thomas, William Roberts Clark, and Matt Golder. 2006.
"Understanding Interaction Models: Improving Empirical
Analyses." Political Analysis 14, 1: 63-82.
Brown, Graham K. 2005. "Playing the (Non)Ethnic Card: The
Electoral System and Ethnic Voting Patterns in Malaysia."
Ethnopolitics 4, 4: 429-M45.
Chin, James. 2001. "Malaysian Chinese Politics in the 21st
Century: Fear, Service and Marginalisation." Asian Journal of
Political Science 9, 2: 78-94.
Cornelius, Wayne A., Jr. 1969. "Urbanization as an Agent in
Latin American Political Instability: The Case of Mexico." The
American Political Science Review 63, 3: 833-857.
Cutright, Phillips. 1963. "Urbanization and Competitive Party
Politics." The Journal of Politics 25, 3: 552-564.
Department of Statistics (DOS). 2011. "Population and Housing
Census of Malaysia." Kuala Lumpur, Malaysia: Department of
Statistics.
--. 2012. "Population Projections, Malaysia 2010-2040."
Kuala Lumpur, Malaysia: Department of Statistics.
Feagin, Joe R. 1972. "Civil Rights Voting by Southern
Congressmen." The Journal of Politics 34, 2: 484 499.
Fried, Robert C. 1967. "Urbanization and Italian
Politics." The Journal of Politics 29, 3: 505-534.
Gomez, James, and Han Leong Chang. 2013. "New Media and
General Elections: Online Citizen Journalism in Malaysia and
Singapore." In Democracy, Media, and Law in Malaysia and Singapore:
A Space for Speech, ed. Andrew T. Kenyon, Tim Majoribanks, and Amanda
Whiting. Abingdon, UK: Routledge.
Greenberg, Sarah, and Thomas B. Pepinsky. 2013. "Data and Maps
for the 2013 Malaysian General Elections." Department of Government
Working Paper. Ithaca, NY: Cornell University.
Hron, K., P. Filzmoser, and K. Thompson. 2012. "Linear
Regression with Compositional Explanatory Variables." Journal of
Applied Statistics 39, 5: 1115-1128.
Huntington, Samuel P. 1991. The Third Wave: Democratization in the
Late Twentieth Century. The Julian J. Rothbaum Distinguished Lecture
Series. Norman: University of Oklahoma Press.
Khoo, Boo Teik. 2013. "The 13th General Elections (GE13) in
Peninsular Malaysia: An Analysis of Issues, Outcomes, and
Implications." In 13th General Election in Malaysia: Issues,
Outcomes, and Implications, ed. Boo Teik Khoo. Chiba-shi, Japan:
Institute of Developing Economies.
Lee, Kuan Yew. 1998. The Singapore Story: Memoirs of Lee Kuan Yew.
Singapore: Marshall Cavendish Editions.
Leete, Richard. 1996. Malaysia s Demographic Transition: Rapid
Development, Culture, and Politics. Kuala Lumpur: Oxford University
Press.
Leong, Keith Yu Keen. 2012. The Future of Pakatan Raykat: Lessons
from Selangor. Kuala Lumpur: Institute for Democracy and Economic
Affairs.
Lemer, Daniel. 1964. The Passing of Traditional Society:
Modernizing the Middle East. London: The Free Press of Glencoe.
Lian, Kwen Fee, and Jayanath Appudurai. 2011. "Race, Class and
Politics in Peninsular Malaysia: The General Election of2008."
Asian Studies Review 35, 1: 63-82.
Lim, Hong Hai. 2002. "Electoral Politics in Malaysia:
'Managing' Elections in a Plural Society." In Electoral
Politics in Southeast and East Asia, ed. Aurel Croissant and Marei John.
Singapore: Friedrich-Ebert-Stiftung, Office for Regional Co-operation in
Southeast Asia.
Liow, Joseph. 1999. "Crisis, Choice and Change: Malaysian
Electoral Politics at the End of the 20th Century." Asian Journal
of Political Science 7, 2: 45-74.
Liow, Joseph, and Afif Pasuni. 2010. "Debating the Conduct and
Nature of Malaysian Politics: Communalism and New Media Post-March
2008." Journal of Current Southeast Asian Affairs 4, 4: 39-65.
Lipset, Seymour M. 1959. "Some Social Requisites of Democracy:
Economic Development and Political Legitimacy." The American
Political Science Review 53, 1: 69-105.
Mansor, Mohd Noor. 1992. "The Determinants of a Malay Ethnic
Alignment." PhD diss., University of Bristol.
--. 1999. "Crossing Ethnic Borders in Malaysia: Measuring the
Fluidity of Ethnic Identity and Group Formation." Akademika 55, 1:
61-82.
McGee, Terry. 2011. "The Urbanisation Transition in Malaysia
in the Twenty-First Century." Akademika 81, 2: 109-121.
Melson, Robert, and Howard Wolpe. 1970. "Modernization and the
Politics of Communalism: A Theoretical Perspective." American
Political Science Review 64, 4: 1112-1130.
Mohamad, Maznah. 2008. "Malaysia--Democracy and the End of
Ethnic Politics?" Australian Journal of International Affairs 62,
4: 441-459.
Mohd Fuad, M. J., A. B. Junaidi, A. Buang, S. Selvadurai, A. C. Er,
and N. Lyndon. 2011. "Ethnic Attitudes, Political Preferences, and
the Politics of Stability." World Applied Sciences Journal 13
(Special Issue on Human Dimensions of Development): 34-38.
Moten, Abdul Rashid. 2009. "2004 and 2008 General Elections in
Malaysia: Towards a Multicultural, Bi-Party Political System?"
Asian Journal of Political Science 17, 2: 173-194.
Mutalib, Hussin. 2000. "Malaysia's 1999 General Election:
Signposts to Future Politics." Asian Journal of Political Science
8, 1: 65-89.
Noh, Abdillah. 2014. "Malaysia 13th General Election: A Short
Note on Malaysia's Continuing Battle with Ethnic Politics."
Electoral Studies 34: 266-269.
Noor, Farish A. 2013. "The 13th Malaysian General Elections
from a Sabah Perspective." The Round Table 102, 6: 541-548.
Noris, Pippa, and Robert Mattes. 2003. "Does Ethnicity
Determine Support for the Governing Party? The Structural and
Attitudinal Basis of Partisan Identification in 12 African
Nations." Kennedy School of Government Research Working Papers
Series. Cambridge: Harvard University.
O'Shannassy, Michael. 2009. "Beyond the Barisan Nasional?
A Gramscian Perspective of the 2008 Malaysian General Election."
Contemporary Southeast Asia 31,1: 88-109.
Ostwald, Kai. 2013. "How to Win a Lost Election:
Malapportionment and Malaysia's 2013 General Election." The
Round Table 102, 6: 521-532.
Ottaway, Marina. 1999. "Ethnic Politics in Africa: Change and
Continuity." In State, Conflict, and Democracy in Africa, ed.
Richard Joseph, 299-318. Boulder, CO: Lynne Rienner.
Papke, Leslie E., and Jeffrey M. Wooldridge. 1996.
"Econometric Methods for Fractional Response Variables with an
Application to 401(k) Plan Participation Rates." Journal of Applied
Econometrics 11, 6: 619-632.
Parsons, Talcott. 1975. "Some Theoretical Considerations on
the Nature and Trends of Change of Ethnicity." In Ethnicity, Theory
and Experience, ed. Nathan Glazer and Daniel Patrick Moynihan, 53-83.
Cambridge: Harvard University Press.
Pepinsky, Thomas B. 2009. "The 2008 Malaysian Elections: An
End to Ethnic Politics?" Journal of East Asian Studies 9, 1:
87-120.
Rajaratnam, Usha Devi. 2009. "Role of Traditional and Online
Media in the 12th General Election, Malaysia." Journal of the South
East Asia Research Centre for Communications and Humanities 1, 1: 35-58.
Resnick, Danielle. 2012. "Opposition Parties and the Urban
Poor in African Democracies." Comparative Political Studies 45, 11:
1351-1378.
Saravanamuttu, Johan. 2001. "Is There a Politics of the
Malaysian Middle Class?" In Southeast Asian Middle Classes:
Prospects for Social Change and Democratisation, ed. A. Rahman Embong,
103-118. Bangi: Penerbit Universiti Kebangsaaan Malaysia.
Saw, Swee Hock. 2007. The Population of Malaysia. Singapore: ISEAS
Publishing.
Shmuelevitz, Aryeh. 1996. "Urbanization and Voting for the
Turkish Parliament." Middle Eastern Studies 32, 2: 162-176.
Sim, Kwang Yang. 2010. "The Cesspool of Sarawak
Politics." Hornbill Unleashed, February 10.
http://hombillunleashed.wordpress.com/2010/12/13/12707/ (accessed
October 13, 2013).
The Star. 2013. "More of a 'Malaysian and Urban'
Tsunami, Says Kit Siang." May 20.
www.thestar.com.my/News/Community/2013/05/20/More-of-a
-Malaysian-and-urban-tsunami-says-Kit-Siang.aspx (accessed October 20,
2013).
Thompson, Eric C. 2013. "Urban Cosmopolitan Chauvinism and the
Politics of Rural Identity." In Cleavage, Connection and Conflict
in Rural, Urban and Contemporary Asia, ed. Tim Bunnell, D.
Parthasarathy, and Eric C. Thompson, 161-179. Dordrecht, Netherlands:
Springer.
Ufen, Andreas. 2009. "The Transformation of Political Party
Opposition in Malaysia and Its Implications for the Electoral
Authoritarian Regime." Democratization 16, 3: 604-627.
van de Walle, Nicolas. 2007. "Meet the New Boss, Same as the
Old Boss? The Evolution of Political Clientelism in Africa." In
Patrons, Clients, and Policies: Patterns of Democratic Accountability
and Political Competition, ed. H. Kitschelt and S. Wilkinson, 50-67. New
York: Cambridge University Press.
Weber, Max. 1978. "Urbanisation and Social Structure in the
Ancient World." In Weber: Selections in Translation, ed. W. G.
Runciman, trans. Eric Matthews. Cambridge, UK: Cambridge University
Press.
Welsh, Bridget. 2013. "Malaysia's Election and
Najib's Challenged Mandate." Asia Pacific Bulletin, 216: 1-2.
Whitby, Kenny J. 1985. "Effects of the Interaction Between
Race and Urbanization on Votes of Southern Congressmen."
Legislative Studies Quarterly 10, 4: 505-517.
Yaakob, Usman, Tarmiji Masron, and Masami Fujimaki. 2012.
"Population Growth and Urbanization in Peninsular Malaysia
1911-2000." Journal of Ritsumeikan Social Sciences and Humanities
4: 79-101.
Yusoff, Muhammad Agus. 2001. "Sabah Politics Under
Pairin." Jati 6: 29-48.
Table 1 Population Distribution by Ethnic Group in Peninsular
Malaysia, 1957-2010
1957 1970 1980 1991 2000 2010
Malays/Bumiputera 49.8 52.7 55.3 58.3 62.4 64.6
Chinese 37.2 35.8 33.8 29.4 27.4 25.9
Indians 11.1 10.7 10.2 9.5 9.5 8.9
Others 2.0 0.8 0.8 2.7 0.7 .06
Total (a) 100.1 100.0 100.1 99.9 100.0 100.0
Sources: Leete (1996) for 1957 to 1991 data; Saw (2007) for 2000
data; and Department of Statistics (2011) for 2010 data.
Note: a. Totals do not all add up to 100 due to rounding.
Table 2 Summary of Parliamentary Election Results for Peninsular
Malaysia
General No. of Seats Won Share of Seats Won (%)
Election
Number Year Total Alliance/BN Opp. Alliance/BN Opp.
1 1959 104 74 30 71.2 28.8
2 1964 104 89 15 85.6 14.4
3 1969 104 67 37 64.4 35.6
4 1974 114 104 10 91.2 8.8
5 1978 114 94 20 82.5 17.5
6 1982 114 103 11 90.4 9.6
7 1986 132 112 20 84.8 15.2
8 1990 132 99 33 75.0 25.0
9 1995 144 123 21 85.4 14.6
10 1999 144 102 42 70.8 29.2
11 2004 165 147 18 89.1 10.9
12 2008 165 85 80 51.5 48.5
13 2013 165 85 80 51.5 48.5
General Share of Votes Won (%)
Election Distortion
Number Alliance/BN Opp. Effect (%)
1 51.8 48.2 19.4
2 58.5 41.5 27.1
3 48.6 51.4 15.8
4 61.5 38.5 29.7
5 57.1 42.9 25.4
6 61.3 38.7 29.1
7 58.1 41.9 26.7
8 55.5 44.5 19.5
9 66.3 33.7 19.1
10 55.4 44.6 15.4
11 63.7 36.3 25.4
12 49.6 50.4 1.9
13 45.7 54.3 5.8
Table 3 Fractional Response Logit Regression Results
Variable Regression Results
Bumiputera -0.0014
(0.0237)
Chinese -0.2279 ***
(0.0284
Indians 0.0034
(0.0727)
Others -1.2622 ***
(0.4859)
Area -0.0789
(1.0256)
Bumiputera x Area 0.0107
(0.2349)
Chinese x Area 1.5269 **
(0.6691)
Indians x Area -1.3102
(1.3504)
Others x Area 6.0453
(3.9607)
Constant 0.3518 ***
(0.1163)
Notes: As discussed in the text, only the sign and not the magnitude
of the coefficients presented in Table 3 can be interpreted. Standard
errors are in parentheses. * p [less than or equal to] 0.10; ** p
[less than or equal to] 0.05; *** p [less than or equal to] 0.01
