. 22
( 35 .)


different constituencies, this argument goes, Bangkok-based candidates
and provincially based candidates have a hard time forming alliances.
In short, Thailand is a country with “two separate political cultures
and two competing agendas” (Pasuk and Baker 1998, 245). The gap
between these urban and rural interests, it is argued, is too great for any
single party to bridge (Anek 1996).
I do not dispute the existence of a division between Bangkok and
the provinces; however, I do question whether this gap is insurmountable.
There are parties that have been able to simultaneously draw support from
both Bangkok and the provinces, most notably the Democrat Party which
has, at times, done well in Bangkok and the more rural Southern region.
More recently, in 2001 the Thai Rak Thai Party was able to win the
support of both provincial and urban voters. Its share of the Bangkok vote
(42%) is nearly identical to what it received in the rest of the country,
outside of the South (44%). In addition, urban- and rural-based parties
have been able to form alliances with each other once in power. An
example is the 1995 grand coalition of the provincially based Chart Thai
and New Aspiration Party with the Bangkok-based Palang Dharma
and Nam Thai Parties. There is no a priori reason that similar cross-
constituency alliances could not be formed prior to elections under the
umbrella of a single party, as indeed they have been in recent elections. In
addition, Chapter 4 demonstrated that coordination failures are
pronounced even within regions where there is no signi¬cant urban“rural
This brings me to a second, but related, alternative explanation. Some
scholars suggest that Thailand has evolved into a party system made up
of several major parties that dominate different regions of the country (see
Surin and McCargo 1997). If this is true, then this could suggest that
Building Party Systems in Developing Democracies

region is an important cleavage in Thailand and a hindrance to better
aggregation. Indeed, there is a good deal of evidence in support of the
regionalization argument. Most parties do draw the bulk of their support
from a particular region and, as I argued in Chapter 4, the Democrat
Party has traditionally dominated Southern elections. If different parties
dominated other regions of the country in the same way, aggregation
across regions could certainly be dif¬cult. Indeed, the parties do have
dif¬culty linking across regions as the cross-regional in¬‚ation score of 30
in Figure 5.1 suggests. Nonetheless, two points of caution are worth
noting regarding claims that Thailand is developing or has developed a
regionalized party system. First, the fact remains that aggregation is
poorer between the district and regional levels (I ¼ 34) than it is between
the regional and national levels (I ¼ 30) (see Figure 5.1). In other words,
the evidence suggests that even candidates from within the same region
have dif¬culty joining together under a common party label.
A second concern with the regionalization claim is whether regions
outside of the South are really dominated by one or two political parties.
To examine this question I calculate the effective number of parties for
each region of Thailand (ENPreg) by aggregating party vote shares by
region. I then compare the number of regional parties (ENPreg) with the
average effective number of parties (ENPavg) at the district level in each
region. ENPavg tells us how many parties there are on average in each
district in a given region. ENPreg tells us how many parties there are
region wide “ aggregating across the various districts in a region. ENPreg,
then, is similar to ENPnat but on a regional scale. By comparing these

I = 34

I = 16 I = 21 I = 30
3.2 effective 3.9 effective 4.9 effective 7.0 effective
parties parties parties parties

I = 54
¬gure 5.1. Where the Thai Party System Gets In¬‚ated (Multidistrict
Provinces Only)
Explaining Aggregation in Thailand 125

two numbers I can determine whether there is a difference between the
average number of parties found in a region™s districts and the total
effective number of parties for that region. In a region with one or two
dominant parties, one would expect the effective number of parties at
the district level (ENPavg) to be low “ the dominant party (or parties)
should be the clear frontrunner in any given constituency. In addition,
aggregation between districts should be good “ truly dominant parties
should be the frontrunners in most of the districts region-wide, and thus
ENPreg should be near ENPavg. Finally, if one or two political parties
truly dominate regions to a degree similar to which the Democrats
dominate the South, then ENPavg and ENPreg for non-South regions
should compare favorably to ENPavg and ENPreg for the South.
Figure 5.2 compares the effective number of regional parties (ENPreg)
for each Thai region with the average effective number of parties in each
region™s districts (ENPavg) (Figure 5.2a). It is clear from this ¬gure that
even at the regional level aggregation is poor. In nearly every region there
is a large gap between the effective number of parties at the district and
regional levels. Even when this gap narrows in 1996, 33% of the regional
party system can still be explained by poor aggregation (I ¼ 33). Figure
5.2 also shows that, as expected, aggregation in the Southern region is
very good. There is very little difference between the number of parties at
the regional level (ENPreg; Figure 5.2b) and the number of parties at the
constituency level (ENPavg; Figure 5.2a) “ except for the year of the split
in the Democrat Party (1988). In addition, both ENPavg and ENPreg
are generally lower in the South than in other regions, again with the
exception of 1988. These results suggest that outside the South, no party
has been able to dominate an entire region. Regionally based parties do
draw the majority of their support from one particular region, and they
may indeed have dif¬culty forming alliances with candidates or parties in
other regions, but as Figure 5.2 demonstrates, these parties have
plenty of dif¬culty forging alliances across districts within their own

Bangkok is an exception. While Bangkok™s ENPavg is comparable in most elections to
ENPavg for the Central and Northern regions, ENPreg is much lower in Bangkok. In
short, there is better aggregation in Bangkok than in any other region outside of
the South. Even though Bangkok has more parties running in a given constituency
than the South, those parties tend to be the same from constituency to constituency.
Thus, the difference between ENPavg and ENPreg is very small.
Building Party Systems in Developing Democracies

(a) Average Effective Number of Parties per District by Region

9 Bangkok
Effective Number of Parties

6 South
1986 1988 1992b 1995 1996
Election Year

Effective Number of Parties by Region

Effective Number of Parties

7 North






1986 1988 1992b 1995 1996
Election Year
¬gure 5.2 Average Effective Number of Parties per District by Region
(a) and (b) Effective Number of Parties by Region

If social or regional cleavages cannot adequately account for
the lack of cross-district coordination in Thailand, then other
factors must be coming into play to prevent aggregation. Why hasn™t
there been greater coordination between urban and rural areas, or
Explaining Aggregation in Thailand 127

across regions by candidates and parties? A major reason is certainly
the lack of incentives to do so. Given stronger aggregation incentives
it seems likely that candidates/parties would ¬nd a way to bridge
these divides. As we will see in the next section, this is indeed the case.

5.4 aggregation and the 1997 constitutional
In 1997 Thailand adopted a new constitution. The ¬rst House of
Representatives election under this constitution was held in 2001.8 For
the ¬rst time since democratic elections were restored in 1979, a single
party, the newly formed Thai Rak Thai Party, nearly captured a majority
of the seats in the House. (Shortly after the election, two smaller parties
decided to merge with Thai Rak Thai, giving the party a legislative
majority.) As a result, the effective number of parliamentary parties fell
quite dramatically to 3.1 from an average of 6.1 during the previous six
elections (Table 5.1). A similar decline is evident in the effective number
of electoral parties (as measured by vote shares) “ from 7.1 to 3.8. This
trend continued in 2005 with the effective number of electoral parties
falling further to 2.4. I argue that this change in the effective number of
parties nationally is a direct result of the constitutional changes Thailand
adopted in 1997. Speci¬cally, I argue that reducing district magnitude
(M) led to improved strategic coordination within districts. However, it
is the stronger incentives for coordination by parties and candidates
across districts that are primarily responsible for reducing the number of
parties. In short, the constitutional changes increased aggregation
incentives, resulting in a decline in the number of parties nationally.
Whenever one is interested in the effects of institutional changes
endogeneity and direction of causation issues are a concern. Speci¬cally
one must consider whether the constitutional changes and subsequent
changes to the party system are both simply re¬‚ections of the interests
and capabilities of major political actors. In the Thai case, however,
treating the constitutional changes as relatively exogenous seems
reasonable. First, the constitutional drafting process was generally free

In 2000, Thailand™s ¬rst ever Senate elections were held. The new Senate had only
delaying power, and Senators could not belong to a political party. See more on this
later in this chapter.
Building Party Systems in Developing Democracies

table 5.1. Effective Number of Parties in Thailand

Effective Number of Parties Effective Number of Parties
Election Year (by seat shares) (by vote shares)
1986 6.1 8.0
1988 7.7 9.8
1992a 6.0 6.7
1992b 6.1 6.6
1995 6.4 6.8
1996 4.3 4.6
Average: 1986“96 6.1 7.1
2001a 3.1 3.8
2005a 1.6 2.4
The 2001 and 2005 election used a mixed-member system with 400 seats elected from
single-seat districts and 100 seats elected using national party lists with proportional
representation. The effective number of parties for these elections is calculated using
all House votes and all House seats.
Sources: Ministry of Interior, election reports (1986, 1988, 1992a, 1992b, 1995, 1997);
Nelson (2002). Electoral Commission of Thailand (2005).

from partisan political in¬‚uence (Prudhisan 1998). The drafting process
was carried out by a body entirely separate from the legislature and
specially created for the purpose of writing a new constitution (the
Constitutional Drafting Assembly “ CDA). The CDA consisted of indi-
rectly elected representatives from each of the country™s provinces
alongside appointed experts in the ¬elds of public law, political science,
and public administration. By statute, the CDA draft was subject only to
an up or down vote in the legislature and could not be amended.9 Within
the CDA itself, the majority of the work of constitutional design and
drafting was delegated to a select committee of academics and techno-
crats with no clear partisan af¬liations.
Second, consistent with the fact that the drafting was outside of the
control of Thailand™s existing political elite, the constitutional reforms
threatened the interests of many of Thailand™s traditional power
centers.10 It is not surprising then that support for the CDA draft con-
stitution was greeted in some quarters with wariness and even outright


. 22
( 35 .)