Party Politics and Preferences for CAFTA in Costa Rica

Globalization and Domestic Politics:
Party Politics and Preferences for
CAFTA in Costa Rica
Raymond Hicks
Helen V. Milner
Dustin Tingley
Princeton University
Background
• Many developing countries have chosen to
liberalize their trade regimes.
• If citizens in a developing country could vote on
trade liberalization, what would influence their
vote?
• Costa Rican referendum on CAFTA in 2007
gives us an opportunity to ask this question.
• We focus on the role of ‘bottom-up’ economic
preferences and ‘top-down’ preferences
generated by organized political parties.
Party Politics and Costa Rica
• PLN and PAC both left parties, but are
contrasts
• PLN established party, PAC very new
– PLN more organized than PAC, larger
geographic reach
• PLN pro-CAFTA, PAC wanted
renegotiation of treaty, and advocated for
voting No in referendum
Economic Effects of CAFTA-DR
• Negative for low-skilled
– Low-skilled industries see larger tariff cuts
– Greater competition from US agriculture
– Increased telecommunications costs in rural
areas
• Positive for high-skilled
– Little tariff changes in high-skilled industries
– Increased foreign direct investment in highskilled industries
– Decreased cost of telecommunications in
urban areas
Individual preferences over CAFTA
Bottom-up
• individuals form preferences based on perceived
economic consequences
Top-down
• Parties can provide cues on complicated policy
areas and frame referendum proposals as being
in the interests of particular groups
• Parties with greater organizational capacity
should be better able to influence voters through
cueing and framing, and influence local political
actors (mayors) to lobby polities, provide
transportation to polls
Hypotheses
• Cantons with more voters employed in low skill
occupations should be most likely to oppose
CAFTA.
Controlling for economic factors that affect
preferences:
• differences in individual support across
members of pro- and anti-CAFTA parties.
• PLN will have a larger influence on individual
voter preferences than the PAC.
• cantons where the PLN was especially
successful at winning votes and organizing prior
to 2007 are the ones most likely to support
CAFTA.
Analysis Road Map
• Collected range of qualitative evidence on
party involvement in CAFTA: the PLN
cued, framed, and used its organizational
muscle (see paper for details)
• Collected public opinion data
• Collected canton level referendum data
Public opinion data
• Universidad de Costa Rica survey, July
and August 2007.
• DV: Preferences for CAFTA, IV’s: political
(vote in ’06 Prez election), economic
demographics (income/education), gender.
• Party variables highly significant, large
effect, and in predicted direction
• Positive effect of income but relatively
small influence
Referendum Analysis
• Collected canton level (81) referendum
results, electoral return data from 2002
(earliest available) and 2006 legislative
elections, and most recent census data.
• Constructed variables measuring canton
level skill level using detailed (4 digit)
industry level employment codes.
• Also collected educational levels (highly
correlated with skill level measure) and
unemployment.
Results 1
• Lower skill=less support for CAFTA
• PLN vote share in 2006 positive and highly
significant, PAC negative but generally not
significant
• PLN vote share in 2002 insignificant
• Straight difference between 2002 and
2006 PLN vote shares, and % (’02)
difference significant and +.
• Still hard to say this identifies the influence
of the PLN
Results 2
• Dichotomize party change variables at
median (high change=treated, low
change=control)
• Nearest neighbor matching on skill level
• Post matching, we interpret treatment as
indicative of high effort by party
• Difference in means between matched
pairs using only common support
• Average difference in vote between
cantons with high/low PLN change: 6%
• Also conducted formal sensitivity analysis
Conclusions
• First paper to investigate sources of citizen
preferences on a particular trade
agreement in the developing world
• Studied top down and bottom up sources
of preferences
• Opposite finding from SS predictionsdetails of trade treaty matter
• Political parties can play an important rolebut levels of organization appear to matter
• Hard to establish causal influence of
parties with our data
Future Research
• Collect additional data on party activities in
CR during the referendum campaign and
afterwards (where did they focus
resources, more information on how they
tailored frames to particular populations,
etc.).
• Interview party leaders, local party
operatives on role in referendum
• GIVEN CAFTA, will PLN engage in
compensation programs to smooth effects
of liberalization?
SUPPLEMENTARY SLIDES
PLN and PUSC Changes
• The decline of the PUSC (right party) between 1998 and
2006 may have led those with pro-CAFTA-DR preferences
to simply migrate to the PLN.
• More PUSC supporters switched to the PAC than switched
to the PLN.
• In January, prior to the election, three former PUSC party
leaders defected to the PAC, whose leaders hoped the
move would help garner the leftist party votes from the
PUSC (Murillo, 2006a).
• PUSC encouraged voters to vote for the PAC candidate in
the Presidential election and the PUSC candidates in the
legislative elections (Wilson, 2007).
• No significant correlations between our changes in PUSC
vote share (discussed below) and changes in PLN vote
share.
Year
Primary
Labor
Technology
20
06
20
04
20
02
20
00
19
98
19
96
19
94
0
.2
.4
.6
.8
1
Changing Costa Rican Exports to US
Effect of Income on DR-CAFTA Preferences
.1
0
-.1
-.2
-.2
-.1
0
.1
.2
PositiveNegBalanceCAFTA
.2
FavorOpposeCAFTA
0
1
2
Category
3
4
0
1
2
Category
3
4
3
4
Y-axis represents change in predicted probability of being in category if moving from low to high income
Gender held at male and other variables held at median
PLN vs PAC Cleavage on DR-CAFTA Preferences
.2
0
-.2
-.4
-.4
-.2
0
.2
.4
PositiveNegBalanceCAFTA
.4
FavorOpposeCAFTA
0
1
2
Category
3
4
0
1
Y-axis represents change in predicted probability of being in category if moving from PAC to PLN
Gender held at male and other variables held at median
2
Category
0
0
.05
2
.1
Density
4
.15
6
.2
Distribution of Tariff Changes Across Cantons
2
4
6
8
Tariff Change
10
12
Estimates weighted using Canton level employment data
.2
.4
.6
.8
TariffChange%PreTariff
1
.7
.6
.5
.4
.3
.3
.4
.5
.6
Percent Pro-CAFTA
.7
Estimated Canton Tariff Changes
0
6
8
Tariff Change
10
12
.2
.4
.6
TariffChange%PreTariff
0
.2
.4
Percent Low Socio-Economic
.8
.2
.4
.6
.8
4
CR Tariff Change as % of Pre Tariff
2
4
6
8
10
12
2
.2
.4
Percent Low Socio-Economic
.6
Estimates weighted using Canton level employment data
.6
0
20
40
60
80
100
Influence of Various Sources on Vote Position
Post-referendum poll
fl
fl
fl
fl
fl
n
n
n
n
n
I
I
I
I
I
pin
pin
pin
pin
pin
O
O
O
O
O
z
ty
m
lis
e
r
ert
a
r
o
a
p
F
P
S
P
Ex
NA
None
Categories: experts, family, party, President, Solis, TV
A little
T
fl
n
I
pin
O
V
A lot
Quartile plot For Treatment and Control Group
LowSocEcon%
Common Support; Replacement
.15
.1
0
.05
.2
qqtreat
.4
.2
.6
.25
All Cantons
0
.2
.4
.6
qqcontrol
.05
.1
.15
.2
.25
Common Support; No Replacement
.05
.1
.15
qqcontrol
.2
.25
.05
.1
.15
qqcontrol
.2
.25