Transcript manger_F930_2_pres

The Political Economy of
Discrimination: Modelling the
Spread of Preferential Trade
Agreements
Mark S. Manger
Assistant Professor
McGill University
[email protected]
Research Question
What explains the geographic
variation in the spread of PTAs?
Current Knowledge:
Bottom-up Explanations
• Milner (1997), Chase (2003, 2005): Multinational firms
seek greater economies of scale through PTAs.
Problem: Suggest no PTAs in Asia Pacific
• Chase (2005): PTAs to facilitate regional production
sharing. Problem: Predicts PTAs with China, not ASEAN
• Mansfield, Milner, Rosendorff (2002): Democratic
Countries are more likely to sign PTAs. Problem:
Everybody’s doing it now.
Current Knowledge:
Top-down Explanations
•
•
Grieco (1997): Capabilities shift hypothesis;
disadvantaged countries will shun PTAs with rising
powers. Problem: Opposite of what we observe.
Mansfield and Reinhardt (2003): Growing WTO
membership creates friction; more PTAs induce states
to likewise sign agreements. Problem: Leaves
geographic pattern of PTAs unexplained.
Current Knowledge: Domino
theory of regionalism
•
•
Baldwin’s (1996) predicts regional spread of PTAs.
Countries join agreements because they fear trade
diversion.
Problems:
trade diversion is not evident
complex network of PTAs does not conform to
expectations. Countries rarely join existing PTAs.
Spatially Dependent PTAs
• Countries will sign PTAs when their neighbours are
doing so.
• Baldwin hypothesis: trade diversion leads countries to
join existing PTAs with neighbours
• Existing PTA not open to expansion or unattractive:
Excluded countries form alternative PTAs with other
proximate countries
• Competition for export markets or FDI: Developing
countries sign PTAs when their neighbours are gaining
preferential access to major markets
Model
W= row-standardized spatial weight matrix of the
dependent variable at t-1 divided by distance
First-cut at only modelling geographic proximity
Undirected dyad-year framework with binary dependent
variable
DV: All reciprocal PTAs 1960-2004, not counting partial
scope agreements
Estimators: Familiar logit with cubic splines; Bayesian
MCMC
Variables
Controls:
Economic model:
GDP per capita for i and j
GDP per capita j
Difference in GDP
Distance between i and j
Bilateral trade
Political economy variables:
Spatial weights in Wy
Trade dependence i on j
Trade dependence j on i
Alliance
Democracy i and j
Multilateral trade round underway
New dispute with 3rd party
Dispute loss with 3rd party
Hegemony
PTA density and PTA density squared
Number of WTO members
Trade partner PTA coverage i
Trade partner PTA coverage j
Future research avenues
• Space is not just geography: explore
similarity measures like GDP/cap, export
profiles, trade links.
• Develop appropriate spatial lag estimator
for binary DV panel data
• Computational challenges: R is sloooow