preferential radet agreements as dynamic farsighted networks
TRANSCRIPT
Preferential Trade Agreements as DynamicFarsighted Networks
James Lake
Southern Methodist University
1st Annual CIRANO Workshop on Networks in Trade and Finance
November 9, 2012
Motivation � WTO and trade liberalization
I Multilateral liberalization
I Non-discrimination: Most Favored Nation (MFN) principle
I Rounds: Kennedy (60s), Tokyo (70s), Uruguay ('94)
I Preferential Trade Agreements (PTAs)
I Members only liberalize trade between themselves
I Discrimination stands in contrast to MFN principle
I Since Uruguay round in 1994, huge increase in PTAs
I Doha round: completely stalled in 2008
Question
I How do PTAs a�ect possibility of global free trade?
I Compare equilibrium of two games (e.g. Saggi and Yildiz(2010))
I Multilateralism: MFN principleI Bilateralism: MFN principle + PTAs
I Possible e�ects of PTAs
I Undermine global free trade (Strong Stumbling Bloc)
I Global free trade only attained under multilateralism
I Necessary for global free trade (Strong Building Bloc)
I Global free trade only attained under bilateralism
Question
I How do PTAs a�ect possibility of global free trade?
I Compare equilibrium of two games (e.g. Saggi and Yildiz(2010))
I Multilateralism: MFN principleI Bilateralism: MFN principle + PTAs
I Possible e�ects of PTAs
I Undermine global free trade (Strong Stumbling Bloc)
I Global free trade only attained under multilateralism
I Necessary for global free trade (Strong Building Bloc)
I Global free trade only attained under bilateralism
Outline of network formation model
I Dynamic farsighted model of network formation
I Only dynamic model to ask Strong Building-Stumbling Blocquestion
I Novel feature: endogenous order of negotiations
I Exogenous order is a well known problem (e.g. Jackson (2008))
I Especially important due to asymmetric market size
I Simultaneous move link announcement game each periodI Equilibrium path of networks is a Farsighted Dynamic
Network Equilibrium
Key intuition behind results
I Focus on Bilateral Agreements (BAs)
I Two country PTAs
I Preference erosion
I BA gives members preferential access
I But eroded if your partner forms future BAsI Fears of preference erosion can undermine BA formationI Fears stronger when BA members larger
Theoretical results
Question: equilibrium of multilateralism (MFN) vs bilateralism
(MFN+BAs) games
1. BAs undermine global free trade (Strong Stumbling Blocs)
I 2 larger countries, 1 smaller country
I Global free trade only attained under multilateralismI Intuition: big enough world market but strong fears of
preference erosion
2. BAs necessary for global free trade (Strong Building Blocs)
I 1 larger country, 2 smaller countries
I Global free trade only attained under bilateralismI Intuition: small world market but weak fears of preference
erosion
Theoretical results
Question: equilibrium of multilateralism (MFN) vs bilateralism
(MFN+BAs) games
1. BAs undermine global free trade (Strong Stumbling Blocs)
I 2 larger countries, 1 smaller country
I Global free trade only attained under multilateralismI Intuition: big enough world market but strong fears of
preference erosion
2. BAs necessary for global free trade (Strong Building Blocs)
I 1 larger country, 2 smaller countries
I Global free trade only attained under bilateralismI Intuition: small world market but weak fears of preference
erosion
Theoretical results
Question: equilibrium of multilateralism (MFN) vs bilateralism
(MFN+BAs) games
1. BAs undermine global free trade (Strong Stumbling Blocs)
I 2 larger countries, 1 smaller country
I Global free trade only attained under multilateralismI Intuition: big enough world market but strong fears of
preference erosion
2. BAs necessary for global free trade (Strong Building Blocs)
I 1 larger country, 2 smaller countries
I Global free trade only attained under bilateralismI Intuition: small world market but weak fears of preference
erosion
Observable implications
1. Probability of BA non-formation depends on strength of
preference erosion
I Empirical evidence: Chen and Joshi (2010, JIE)
2. How order of BA formation relates to market size asymmetry
I Anecdotal evidence from following negotiations:
I US-Canada-ColombiaI US-Canada-KoreaI US-Australia-Korea
Outline
Introduction
Network formation game
Underlying trade model
Farsighted Dynamic Network Equilibrium
FDNE of bilateralism game
FDNE of multilateralism game
Role of BAs in equilibrium
Implications for real world negotiations
Conclusion
Network positions and actions
I Actions: link announcement game
I At each network player i can announce
1. Want to form BA with player j (if doesn't exist yet)2. Want to move to Free Trade (if not there yet)3. Don't want to form any agreement
Network positions and actions
I Actions: link announcement game
I At each network player i can announce
1. Want to form BA with player j (if doesn't exist yet)2. Want to move to Free Trade (if not there yet)3. Don't want to form any agreement
Equilibrium concept � dynamic dimension
I Think of networks as nodes of an extensive form game
I Rules of game
I Start from empty networkI One agreement per period (endogenous)I Mutual consent required to form agreementI Previous agreements can't be severed
I Global free trade is absorbing state
I Solve by backward induction, but...
I At each network: solve link announcement game
I Equilibrium Binding Agreement (Ray and Vohra, JET, 1997)
I Payo�s are discounted sum of one period payo�s
Equilibrium � backward induction
Equilibrium � backward induction
Equilibrium � backward induction
Equilibrium concept � simultaneous dimension
Equilibrium Binding Agreement (EBA; Ray & Vohra (JET, 1997))
I Deviating coalition S doesn't take actions of N \ S as given
I S anticipates �equilibrium reactions� of N \ SI Resulting action pro�le
I Nash across S and N \ SI Cooperation within but not across coalitions
I Deviation self enforcing if no subcoalition of S or N \ S wantsto deviate
I EBA if no self enforcing deviations
I Network induced by EBA is an EBA network
I What if no unique equilibrium reaction of N \ S?I Ray and Vohra (1997): S anticipates �optimistically�I Diamantoudi (2003, ET): S anticipates �pessimistically�
Why EBA over CPNE?
Equilibrium concept � simultaneous dimension
Equilibrium Binding Agreement (EBA; Ray & Vohra (JET, 1997))
I Deviating coalition S doesn't take actions of N \ S as given
I S anticipates �equilibrium reactions� of N \ SI Resulting action pro�le
I Nash across S and N \ SI Cooperation within but not across coalitions
I Deviation self enforcing if no subcoalition of S or N \ S wantsto deviate
I EBA if no self enforcing deviations
I Network induced by EBA is an EBA network
I What if no unique equilibrium reaction of N \ S?I Ray and Vohra (1997): S anticipates �optimistically�I Diamantoudi (2003, ET): S anticipates �pessimistically�
Why EBA over CPNE?
Outline
Introduction
Network formation game
Underlying trade model
Farsighted Dynamic Network Equilibrium
FDNE of bilateralism game
FDNE of multilateralism game
Role of BAs in equilibrium
Implications for real world negotiations
Conclusion
Political economy oligopolistic trade model
I Cournot competition in single homogeneous good
I Single �rm from each countryI Common (nonprohibitive) tari�I Constant and common marginal costI Segmented markets
I Linear inverse demand function in country j : P j = αj − Q j
I Market size asymmetry in the αj termsI s (small), m (medium) and ` (large)
I Government only cares about �rm pro�ts
I Results robust with national welfare and optimal tari�s
�Myopic� BA formation incentives
I Non-members
I Discrimination embodied in BAs hurt non-members
I Members
I Bene�t: get preferential access to partner market
I Bigger if partner has larger market, fewer BAs
I Cost: give preferential access to own market
I Bigger if have larger market, fewer BAs
I Threshold: one shot net bene�t positive for any BA i�
α`
αs≡ α`s ≤ α`s
�Myopic� BA formation incentives
I Non-members
I Discrimination embodied in BAs hurt non-members
I Members
I Bene�t: get preferential access to partner market
I Bigger if partner has larger market, fewer BAs
I Cost: give preferential access to own market
I Bigger if have larger market, fewer BAs
I Threshold: one shot net bene�t positive for any BA i�
α`
αs≡ α`s ≤ α`s
Outline
Introduction
Network formation game
Underlying trade model
Farsighted Dynamic Network Equilibrium
FDNE of bilateralism game
FDNE of multilateralism game
Role of BAs in equilibrium
Implications for real world negotiations
Conclusion
EBA at hub�spoke network
I Assume α`s ≤ α`s . Then, spokes form own BA
EBA at insider�outsider network?
EBA at insider�outsider network?
I Hub bene�ts exceed cost of preference erosion i�
πHubi +
β
1− βπFreeTradei >
1
1− βπInsider ,ji ⇒ β < βFT−I
i ,j (.)
I i wants to become hub when β < βFT−Ii,j (.)
I Small enough weight on preference erosion
EBA at insider�outsider network
I EBA network conditional on (ij)
I Smaller insider j has greater incentive to become hub
I β > βFT−Ij,i (.): (ij) Pareto dominant for i , j
I β ∈(βFT−Ii,j (.) , βFT−I
j,i (.)): j wants to become hub
I Induces i to become hub so avoid spoke discriminationI Even though β > βFT−I
i,j (.)
I β < βFT−Ii,j (.): i wants to become hub
Farsighted Dynamic Network Equilibrium
I FDNE: equilibrium path of EBA networks
I Key point
I β > βFT−Im,` (.): m and ` remain insiders
I β < βFT−Im,` (.): reach global free trade
E�ect of greater asymmetry on FDNE
I Greater asymmetry: ↑ αms ≡ αm
αs or ↑ α`s ≡ α`
αs ⇒↓ βFT−Im,` (.)
I Greater asymmetry reduces scope for global free trade
I m and ` protect more rents as insidersI Preference erosion more costly
Outline
Introduction
Network formation game
Underlying trade model
Farsighted Dynamic Network Equilibrium
FDNE of bilateralism game
FDNE of multilateralism game
Role of BAs in equilibrium
Implications for real world negotiations
Conclusion
FDNE of multilateralism game
I Only two possible outcomes
I Empty networkI Direct move to global free trade
I Blocked by ` when ` too large relative to m and s
I ` blocks global free trade when views world market as
�too small�
Outline
Introduction
Network formation game
Underlying trade model
Farsighted Dynamic Network Equilibrium
FDNE of bilateralism game
FDNE of multilateralism game
Role of BAs in equilibrium
Implications for real world negotiations
Conclusion
Bilateralism (MFN+BAs) vs Multilateralism (MFN)
I Multilateralism FDNE
Bilateralism (MFN+BAs) vs Multilateralism (MFN)
I Bilateralism FDNE
Bilateralism (MFN+BAs) vs Multilateralism (MFN)
I Bilateralism FDNE
Bilateralism (MFN+BAs) vs Multilateralism (MFN)
I Bilateralism FDNE
Bilateralism (MFN+BAs) vs Multilateralism (MFN)
I Strong Stumbling Blocs (SSB)
I Global free trade only under multilateralism
Bilateralism (MFN+BAs) vs Multilateralism (MFN)
I Strong Stumbling Blocs (SSB)
I Global free trade only under multilateralism
I BAs SSBs when �2 larger countries, 1 smaller country�
Bilateralism (MFN+BAs) vs Multilateralism (MFN)
I Strong Building Blocs (SBB)
I Global free trade only under bilateralism
Bilateralism (MFN+BAs) vs Multilateralism (MFN)
I Strong Building Blocs (SBB)
I Global free trade only under bilateralism
I BAs SBBs when �2 smaller countries, 1 larger country�
Outline
Introduction
Network formation game
Underlying trade model
Farsighted Dynamic Network Equilibrium
FDNE of bilateralism game
FDNE of multilateralism game
Role of BAs in equilibrium
Implications for real world negotiations
Conclusion
Observable implications for BA non-formation
I BA formation probabilities should depend on network structure
I Key characteristic: fear of future preference erosion
I Chen and Joshi (2010, JIE)
I Pr (X − Y BA | Hub-Spoke) ≈4Pr (X − Y BA | Insider-Outsider)
Order of negotiations vs. order agreements concluded
I Pre 1990: Colombia small relative to US, Canada
I β > βFT−ICan,US (.): US, Canada remain insiders
I 1990s and 2000s: Colombia grows relative to US, Canada
I β < βFT−ICan,US (.): Canada wants to form BA with Colombia
I Canada-Colombia negotiations induce US-Colombia BA
I Similar interpretation for other negotiationsI Interpretation depends on endogenous order of negotiations
Implications for ambiguity of Article XXIV
I PTAs permitted by GATT Article XXIV
I Literature treats PTA as immediate and complete removal of
trade barriers
I In practice, two ambiguities
I Tari�s must be removed on �substantially all trade�I Tari�s can be �phased out� over time
I Ambiguities generally seen as counter-productive
I But could be productiveI Increased hub bene�t ⇒ mitigates fear of preference erosion
Conclusion
I How does presence of BAs a�ect possibility of global free
trade?
I 2 large countries, 1 small country
I Strong Stumbling Blocs: undermine global free tradeI Big world market, strong preference erosion fears
I 2 small countries, 1 large country
I Strong Building Blocs: necessary for global free tradeI Small world market, weak preference erosion fears
I Observable implications
I Order that negotiations commence should di�er from orderthat agreements concluded
I BA formation probabilities should depend of strength ofpreference erosion fears
Multiple equilibria
I Consider EBA network resulting after (ij)
I For β ∈(βFT−Ii ,j (.) , βFT−I
j ,i (.)): larger insider became hub on
path to free trade
I Interpretation: US-Canada-Colombia negotiations
I But there is another equilibrium here: (ij ,FT )
I No pro�table joint deviationI i prefers (ij). But fear of (ij , jk) deters deviation.I Interpretation: CUSFTA became NAFTA
What happens when no preference erosion?
Why CPNE rather than EBA?
I Existence
I CPNE non-existence well known problem
I E.g: Condorcet paradoxI Arises in this model
I Deviation coalition S takes actions of N \ S as �xed
I If i wants to deviate from (ij , ik) then not CPNE
I But i should consider what j and k will doI EBA captures such considerations
EBA