. 18
( 35 .)


1976 7.0 4.1 4
1979a 11.6 8.2 7
1983a 5.9 5.6 4
1986 8.0 6.1 4
1988 9.8 7.7 6
1992a 6.7 6.0 5
1992b 6.6 6.1 5
1995 6.8 6.4 7
1996 4.6 4.3 6
Average 7.7 6.2 5.5
The ¬rst election under the 1978 constitutions was held in 1979 but political parties were
not formally legalized until the 1981 Political Party Act. However, they did exist informally
and were allowed to organize and campaign. In the 1979 and 1983 elections, candidates
were allowed to run as independents. In 1979, 619 independents ran and captured 31.5%
of the vote and 20.9% of the seats. Their numbers were much smaller in 1983 with 417
independents capturing 7.5% of the votes and 7.4% of the seats. The question then arises
how to count independents. The alternatives are to count all independents together as a
single “independent” party or to count each independent separately as a party of one. The
former may understate the number of parties, while the latter may overstate. Since I believe
the reality of the Thai situation was closer to many parties of one rather than a single
independent party I have reported ENP where each independent is counted separately. In
calculating ENP by vote share, there was an additional challenge “ the lack of readily
accessible data on individual independent candidate vote shares. Where those data were
lacking I calculated the average vote share for independent candidates (total independent
vote share/total number of independent candidates) recognizing that this will in¬‚ate ENP.
If, rather than counting each independent separately, I group all independents together in
one “party” ENP by vote share would be 5.4 in 1979, 5.7 in 1983 for an average of 7.1 over
the period. ENP by seat share would be 6.1 in 1979, 5.4 in 1983 for an average of 6.0.
Sources: Ministry of Interior, election reports (1986, 1988, 1992a, 1992b, 1995, 1997);
Manut (1986).

block vote electoral system and the coordination of voters, candidates,
and parties within districts? To answer these questions I analyze how
coordination, as well as failures to coordinate, contributed to the
number of political parties in Thailand.
As discussed in Chapters 1 and 2, the in¬‚ation of the party system
can arise as the product of either of two separate types of coordination:
intra-district or cross-district coordination (aggregation). Intra-district,
or district-level coordination produces a large party system where
Building Party Systems in Developing Democracies

either (a) permissive electoral rules do not generate incentives for
coordination around a small number of parties or candidates within a
given district or (b) impediments exist that prevent coordination on a
small number of competitors, even where there are apparent electoral
incentives for coordination (Duverger 1954; Taagepera and Shugart
1989; Cox 1997). I demonstrate that intra-constituency coordination
has not been the primary source of multiple parties “ the average
number of parties in each constituency is much lower than the number
of parties nationally, and about what one would expect given Thailand™s
electoral system. This is an important contribution. Because of the
unusual nature of the block vote system, the application of tools such as
Duverger™s law (Duverger 1954) or Cox™s M þ 1 rule (Cox 1997) to the
Thai case is not immediately obvious. Nevertheless, I show that it is
possible to predict the number of parties at the district level from the
electoral system.
Cross-district coordination has been much more problematic in
Thailand. Cross-district coordination or aggregation failures occur
when different parties run in various districts across the country. I
argue that even though intra-district coordination in Thailand is fairly
good, aggregation has been very poor resulting in a large number of
parties and poor nationalization. I will begin with a discussion of the
different features of the Thai electoral system and their impact on the
effective number of parties at the district level.

4.3.1 The District-Level Party System in Thailand (1978“1996)

One of the major determinants of the number of parties locally is the
type of electoral system within which parties, candidates, and voters
must work.26 The electoral system Thailand used for most of its
history “ the block vote “ is unusual and warrants some description.27

The second major determinant is whether or not societal cleavages are present (e.g.,
ethnic, religious, or linguistic cleavages). With the possible exception of an urban-
rural cleavage, Thailand largely lacks the deep social cleavages found in many of its
neighbors (e.g., Malaysia, Indonesia, Burma). At the district level, the urban-rural
cleavage has not played a role in determining the number of parties as districts tend to
be either urban or rural. The extent to which the urban-rural cleavage has affected
aggregation is discussed in Chapter 5.
Other uses of the block vote include elections in Mauritius and nineteenth-century
Great Britain and elections for the Philippines Senate.
Thailand: Aggregation, Nationalization, Number of Parties 101

table 4.6. Basic Electoral System Data

1983 1986 1988 1992a 1992b 1995 1996
Total Districts 134 138 142 142 142 155 156
Total Seats 324 347 357 360 360 391 393
Three-Seat Districts 65 80 82 85 85 88 88
Two-Seat Districts 60 49 51 48 48 60 61
One-Seat Districts 9 9 9 9 9 7 7
Sources: Ministry of Interior, election reports (1986, 1988, 1992a, 1992b, 1995, 1997),
Manut (1986), Law (1987).

Under the 1978 and 1992 constitutions, Thailand was broken down
into between 142 and 156 electoral districts (depending on the election
year), which together were responsible for ¬lling between 360 and 393
seats in the House of Representatives.28 Electoral districts were broken
down into one-, two-, and three-seat districts. Most Thai districts had a
district magnitude of three (M ¼ 3) or two (M ¼ 2), while a few were
single seat districts (M ¼ 1).29 Seats were allocated by province
(changwat), with each province receiving the number of seats com-
mensurate with its population (one seat for every 150,000 people). See
Table 4.6 for a summary of these data for the last six elections prior to
the new 1997 Constitution. If a province had a large enough population
for more than three seats, the province was divided into more than one
district, and the seats were distributed so as to avoid single-seat dis-
tricts. For example, if the population of a province warranted four
seats, the province would be divided into two districts, each with two
seats. Seven seats would be divided into three districts of three, two,
and two seats. Single-seat districts occurred only in provinces with a
population less than 225,000.30
Under the block vote, voters were allowed to vote for as many
candidates as there were seats in a district, and seats were awarded to
the top vote-getters on the basis of the plurality rule. Voters could not
group their votes on one candidate (cumulation was forbidden) but

Thailand has a bicameral legislature consisting of an elected House and (until the
1997 constitutional reforms) an appointed Senate.
In Thailand, electoral units are the termed “constituencies” rather than “districts. ”
Each additional 75,000 people above 150,000 was counted as an additional 150,000.
A province with 200,000 people would receive one seat while a 225,000-person
province would receive two seats.
Building Party Systems in Developing Democracies

panachage was allowed (i.e., voters could split their votes between
candidates from different parties). Finally, voters were not required to
cast all of their votes “ they could partially abstain (plumping).31
Parties were required to ¬eld a full team of candidates for any district
they wished to contest (e.g., three candidates in a three-seat district).
Given Thailand™s electoral system, how many parties would we
expect? The answer is not immediately obvious. The block vote elec-
toral system has not been studied much by students of electoral systems.
However, it is possible to make some predictions. The expectations for
the effective number of parties in each Thai constituency should vary
according to the district magnitude and according to one™s expectations
about the importance of Thai party labels.
To begin with, the M þ 1 rule (Cox 1997), a generalization of
Duverger™s law, is a useful predictor of the effect of an electoral system
on the number of parties at the district level.32 The M þ 1 rule states
that no more than M þ 1 candidates/parties are viable in any single seat
districts (M ¼ 1) and no more than M þ 1 parties are viable in multiseat
districts (M > 1). In other words, in single-seat districts, the expected
number of parties would be two; in two-seat districts, three parties, etc.
The expectations of the M þ 1 rule rest on an assumption of strategic
coordination by candidates, parties, and voters. Candidates and parties
decide whether to enter a race partly on the basis of their chance of
winning a seat or seats (strategic entry) (Cox 1997). Using the example
of a single-seat district, third-place candidates or parties have an
incentive to withdraw from the race or not to enter at all, or in the case
of political parties, to join with one of the two front-running parties. If
coordination among candidates or parties over strategic entry fails,
then strategic voting can reduce the number of viable contenders in a
given constituency. Voters, realizing that their votes are wasted if they
cast them for third-place contenders, have an incentive to transfer their
votes to their most preferred of the two strongest contenders. In a
single-seat constituency, voters who under normal circumstances
would prefer the third-place contenders will instead vote for their most
preferred of the top two contenders so as not to waste their votes.

See Cox (1997, 42“3) for a general discussion of cumulation, panachage, and
Duverger™s law states that electoral systems with single-member districts and plurality
voting rules will produce a two-party system (Duverger 1954).
Thailand: Aggregation, Nationalization, Number of Parties 103

Strategic Entry Assumptions: Strategic Voting Assumptions:
1. The identity of the frontrunners must 1. Voters must be short term
be common knowledge. instrumentally rational.
2. The primary goal of candidates or 2. Voters must have access to “reasonably
parties must be victory in the current accurate and publicly available
election information” on candidate or party
3. There must be myopic (price-taking)
adjustment on the part of voters.

Source: Cox, Gary. 1997. Making Votes Count. Cambridge: Cambridge University Press.
¬gure 4.1. M þ 1 Rule Assumptions

However, certain assumptions must be met in order for the M þ 1 rule
to completely hold (Cox 1997).33 These are listed in Figure 4.1.
For the most part, these assumptions are met in Thai elections with two
caveats. First, it is important to note that vote buying is common in Thai
elections (Hicken 2007b). To the extent that vote buying and selling dic-
tates how voters cast their votes, strategic voting assumptions one and
three may not hold. If voters base their vote on which candidates give them
the most money rather than which candidates have the best chance at
winning seats, then they are not “short term instrumentally rational” in
the manner described in Figure 4.1. This is not to say such voters are not
rational; rather, they have different goals than those of the instrumentally
rational voter. Likewise, vote-selling voters may not adjust their votes even
after receiving information that their vote buyer is a trailing candidate.
Vote buying, in the form of votes for cash or goods, certainly occurs in
Thailand (Sombat 1993; Arghiros 1995; Anek 1996; Surin and McCargo
1997; Nelson 1998; Callahan 2000; Hicken 2007b) but there is some
question about how this affects election results. If one candidate engages
in vote buying within a district, other candidates will have an incentive to
do the same. Some scholars report that the result is that Thai voters often
accept money from many candidates and parties and then vote for their
preferred candidate(s) anyway (Sombat 1999). If this is the case, then
vote buying might not have a large effect on strategic voting. In addition,
campaign managers and vote canvassers/vote buyers buy votes in a
strategic manner.

These are the necessary assumptions to generate a tendency to bipartism in a single-
member district. For the assumptions necessary to produce pure bipartism, see Cox
(1997, 76“9).
Building Party Systems in Developing Democracies

During the ¬nal days prior to voting, many campaign managers opted to
abandon weaker members of their team who seemed to have little prospect of
being elected. Financial resources could then be directed towards buying votes
on behalf of the one or two candidates with the best electoral prospects.
(Callahan 2000, 35)

A second caveat is that the informational assumptions behind stra-
tegic voting and strategic entry may sometimes be problematic. First,
accurate polling data can be hard to come by in Thailand, especially in
rural districts. Thus some candidates and voters may lack the infor-
mation necessary to distinguish the frontrunners from the rest of the
candidates. Second, party labels often help communicate information
on the viable candidates and parties in any given district and thus help
candidates and voters coordinate their behavior. However, when party
labels mean little to voters or candidates, and party support in a given
district varies widely from election to election, then it will be more
dif¬cult for voters to obtain “reasonable and accurate” information on
party and candidate standings. Even if party labels are weak, though,
there are other cues to which voters and candidates can look to assess
candidate/party viability. These would include a candidate™s personal
reputation and electoral history; the family, group, or faction to which
a candidate belongs; or the amount of money the candidate spends
campaigning (or buying votes) (Napisa 2005).
In summary, for the most part, the assumptions that underlie the
M þ 1 rule are met in the context of Thai elections, although with the
possible exceptions noted previously. Given this, for single-seat dis-
tricts, a straightforward application of Duverger™s law and the Cox™s
M þ 1 rule is possible (Duverger 1954; Cox 1999). In Thailand™s single-
seat districts, one would expect the average effective number of parties
to be around two “ slightly more where the assumptions behind the
M þ 1 rule are not met.
What about Thailand™s two- and three-seat districts? There the
application of the M þ 1 rule is less obvious. The M þ 1 rule has not often
been applied to cases where voters have multiple votes, as Thai voters do
in two- and three-seat districts. When voters have multiple votes, does the
M þ 1 rule still apply? Will the effective number of parties still vary by
district magnitude? The answer depends on what we assume about the
value of party labels to Thai voters. If voters truly cast their votes
according to party labels, the M þ 1 rule will not apply “ the effective
Thailand: Aggregation, Nationalization, Number of Parties 105

number of parties should be around two across all districts, regardless of
district magnitude.
To see why, consider a single-seat district (M ¼ 1) where the NAP


. 18
( 35 .)