. 23
( 35 .)


In the event of a no vote, the draft would then go before the people in the form of a
The nature of the reforms was very much the re¬‚ection of middle class (Bangkok)
preferences (Connors 2002). See also McCargo 2002.
Explaining Aggregation in Thailand 129

opposition. The political elite of many of the major political parties and
factions, including the ruling NAP, expressed strong reservations about
the draft. The fact that most ultimately voted to adopt the draft con-
stitution, despite their very serious misgivings, is a function of two fac-
tors. First, the Constitutional Amendment Bill required an up or down
vote of the draft by parliament without amendment. This made it
impossible for legislators to pick apart the draft or delay it via the
amendment process. Second, the coincident occurrence of the Asian
economic crisis, a chain reaction that began in Thailand in late June/early
July 1997, effectively raised the stakes connected with passage or
rejection of the draft. The crisis struck just as the drafting process was
wrapping up. It shone a spotlight on some of the shortcomings in the
Thai political system (MacIntyre 2002). In the minds of many voters and
investors, the constitutional draft became a symbol of the government™s
commitment to dif¬cult but needed political and economic reforms.
Constitutional reform and the broader reform agenda became so linked,
in fact, that the stock market and currency markets reacted quickly and
noticeably to expressions of opposition or support by leading govern-
ment of¬cials. In the end, the potential economic and political costs of a
no vote outweighed the risks of reform, and the draft was adopted by a
vote of 518 to 16 (with 17 abstentions).
Finally, it is dif¬cult to draw a clear link between existing political
interests and reform processes and outcomes. As mentioned previously,
the leaders of the largest party during the drafting period were not
enthusiastic supporters of the proposed constitution and only came
around once the crisis-related implications were apparent. The party
that many believed stood to gain the most from the new rules, the
Democrat Party, was in the political opposition throughout the draft-
ing and passage process. Finally, the party that ultimately bene¬ted the
most from the reforms, the Thai Rak Thai Party, did not exist when the
constitution was being drafted and adopted, nor were its future leaders,
most notably Thaksin Shinawatra, involved in the drafting process. In
short, in this case it does not seem unreasonable to treat the constitutional
reforms as exogenous to the subsequent changes in the party system.
In the rest of the chapter, I ¬rst review the 1997 constitutional reforms
and use the aggregation incentives theory to generate hypotheses on the
effect of these reforms on the number of parties. I then test these
hypotheses using data from the 2001 and 2005 elections.
Building Party Systems in Developing Democracies

table 5.2. Constitutional Reforms

1978/1991 Constitutions 1997 Constitution
 1“3 seat constituencies  Mixed-member system
House of
 Block vote  400 single-seat constit.
 100 national party list
 Appointed  Elected using SNTV,
 Allowed  90-day membership
Party Switching
 Limited  Mandated

5.4.1 The Effect of Political Reform on Aggregation
Incentives and the Number of Parties

The 1997 constitution and subsequent party and electoral laws drasti-
cally revamped Thailand™s electoral and political landscape. Reforms
included changes in the way elections are administered, the establishment
of several semiautonomous oversight agencies, and the creation of an
elected Senate “ the ¬rst ever in Thailand. Here, however, I will focus on
the reforms that might be expected to bear on aggregation and the
number of political parties. These reforms are summarized in Table 5.2.11

Decreased District Magnitude
One of the most striking changes in the 1997 Constitution was the
move to 400 single-seat districts in place of the multiseat districts that
were previously the norm. As discussed in the last chapter, electoral
theory suggests that lowering district magnitude should also lower
the effective number of parties at the district level.
Whether the move to single-seat districts actually leads to an average
effective number of parties of two, corresponding to the M þ 1 rule,
depends on how well the assumptions behind the M þ 1 rule hold.12
There are several reasons to expect ENPavg to be slightly larger than
2 for the 2001 election. First, it may take time for candidates and voters
to divine and respond to the incentives generated by a new electoral
system. The new electoral system necessitated a redrawing of district

For discussion of other important reforms and their effects, see Hicken (2006).
See the last chapter for a discussion of these assumptions.
Explaining Aggregation in Thailand 131

boundaries, in some cases pitting incumbent against incumbent and
in others leaving districts without incumbents. Party allegiances also
shifted as party leaders and potential candidates attempted to antici-
pate what reforms would mean for various parties™ electoral prospects.
In short, these changes meant that some of the cues that facilitate
strategic coordination by voters and candidates (e.g., electoral histo-
ries, party labels) were lacking.
Second, to the extent that vote buying and selling (long features of Thai
elections) continue to dictate how some voters cast their votes, some of
the strategic coordination assumptions may not hold. Vote buying cer-
tainly occurred in the 2001 election, as it did in past elections, but as
discussed in Chapter 4, there is some question about how this impacts
election results.13 If one candidate engages in vote buying within a dis-
trict, most other candidates will have an incentive to do the same. The
result is that Thai voters often accept money from many candidates and
parties and then vote for their preferred candidate(s) anyway. If this is the
case, then vote buying might not have a large effect on strategic voting.
To summarize, the uncertainty connected with a new electoral system
combined with the occurrence of vote buying may keep the effective
number of parties above 2. Over time, one would expect ENPavg to fall as
voters and candidates adjust to the new rules and as vote buying
diminishes as a result of development and increased enforcement of anti-
vote-buying laws.14

One of the most striking features of the 1997 Constitution was its call for
greater decentralization. Political and economic power has traditionally
been highly centralized in Thailand. To the extent decentralization
actually lead to greater political and economic power at the subnational
level, aggregation incentives should have decreased. However, the
decentralization provisions of the constitution had not been imple-
mented at the time of the 2001 elections. The process of decentralization

See Callahan (2002) and Hicken (2007b) for discussions of the effect of the consti-
tutional reforms on vote buying.
During the 2001 election, the newly created Electoral Commission penalized and/or
disquali¬ed several candidates found guilty of vote buying and other illegal practices.
Elections were re-run in many districts where electoral law violations were found. See
Nelson (2002).
Building Party Systems in Developing Democracies

did commence under Thaksin (post-2001), but although local elections
were held and some budgetary power was decentralized, progress
towards meaningful decentralization remained slow (Painter 2005).15
Eventually vertical decentralization should reduce the size of the
aggregation payoff, but not enough time has passed to assess whether
this is indeed the case.

Greater Power for the Prime Minister
The change in the 1997 Constitution with the biggest bearing on
aggregation incentives was increased powers for the prime minister
relative to factions within his own party. Two changes are particularly
worth noting. First, cabinet members were now required to give up
their seats in parliament if they chose to join the cabinet. Since parties
or ministers that chose to leave the cabinet, or were expelled by the
prime minister, could no longer return to parliament, the stakes asso-
ciated with breaking with the prime minister were much higher.
Second, the 1997 Constitution placed new restrictions on party
switching. In order to compete in future elections, candidates had to be
members of a political party for at least 90 days. The rule was designed
to curb the 11th hour party switching by individuals and factions that
traditionally occurred in the run-up to Thai elections. Once the House
was dissolved, elections had to be held within 45 days (if the House™s
term expired) or 60 days (if parliament was dissolved) “ not enough
time for would-be party switchers to meet the membership require-
ment. The prime minister, with the power to dissolve the House and
call new elections, gained the most from this change.16 The prime
minister could credibly threaten to call new elections if party factions
tried to bolt, thus forcing the members of the faction to sit out one
election.17 According to the theory, enhanced power for the prime
minister over intra-party factions increases the payoff to being the
largest party in government. If the theory is correct, this should result in
better aggregation and fewer parties.

In fact, the reforms carried out under the banner of decentralization have actually
recentralized authority under the prime minister™s of¬ce (Painter 2005).
Formally it was the king who dissolved the House and called for new elections upon
the advice of the prime minister.
For this reason, some prominent Thai factions were in favor of amending the con-
stitution to allow for easier party switching.
Explaining Aggregation in Thailand 133

table 5.3. Summary of Constitutional Changes

National Party System
District Party System (Aggregation
(Reduce or in¬‚ate
Constitutional Incentives Stronger or
Change Weaker)
Decrease district Reduce “
Greater power for PM “ Stronger
Decentralization (not “ (Weaker)
fully implemented)

Table 5.3 summarizes the reforms just discussed along with the
expected effects of these changes on the district level and national party
On balance then, Thailand™s 1997 constitutional reforms pushed in
the same direction “ toward a reduction in the number of parties at the
national level (ENPnat). A portion of this reduction should re¬‚ect fewer
parties at the district level (ENPavg) due to the move to single-seat dis-
tricts. However, since the average effective number of parties in pre-
reform Thailand was already quite modest (3.2), any large decline in the
number of parties nationally should be the result of better aggregation
between districts. I summarize these expectations in hypothesis form

Hypothesis 1: The move from multiseat to single-seat districts will be
associated with a fall in ENPavg.
Hypothesis 2: Aggregation in the post-reform elections will improve
relative to pre-reform elections as measured by the in¬‚ation score (I).
Hypothesis 3: The post-1997 elections should have a smaller ENPnat
post-reform than previous elections.
Hypothesis 4: Better aggregation should play a larger role in
lowering ENPnat than the decline in ENPavg.

In the appendix, I include a discussion of two other reforms “ the addition of an
elected Senate and a national party list tier for House elections “ and explain why
these reforms do not generate strong predictions about changes in aggregation
incentives, though they do have important bearing on other dimensions of the party
Building Party Systems in Developing Democracies

5.4.2 Empirical Results

To test these hypotheses I compiled district- and national-level
electoral returns for the 2001 and 2005 elections to the Thai House
of Representatives. (2001 was the ¬rst election to be held under the
1997 Constitution.) The data from these two elections were then
compared to data from ¬ve pre-reform elections to determine
whether the local party and national party systems have changed in
the hypothesized manner.19 All of the hypotheses are supported
by the data.
At the district level, a move to single-seat districts was accompanied
by a decline in ENPavg as hypothesized (H1). ENPavg for the elections
prior to 1997 was 3.2. During the 2001 election, ENPavg fell by nearly
16% to 2.7. It fell a further 35% to 2.0 in 2005 (see Table 5.5). An OLS
regression with robust standard errors reveals that the average effective
number of parties in each district is signi¬cantly lower in post-reform
elections (see Table 5.4). Voters and candidates clearly responded
strategically to the change in the district electoral system “ speci¬cally
the reduction in district magnitude. Also as expected, the average
effective number of parties did not immediately fall to 2 in 2001, but
by the 2005 election the contest in each district was, on average, a
two-party affair.
If voters and candidates at the district level were able to coordinate
on a smaller number of parties than they had in past elections, were
they able to do the same across districts? Did aggregation improve as
hypothesized (H2)? Figure 5.3 compares the effective number of parties
at the district and national levels before and after constitutional
reform.20 Note the narrowing of the gap between the effective number
of parties nationally and the average effective number of parties locally
in 2001 and 2005. This is evidence of improved aggregation. Better

These were the 1986, 1988, September 1992, 1995, and 1996 elections. The March
1992 election is once again excluded due to incomplete district-level electoral data.


. 23
( 35 .)