Learning by Ruling: A Dynamic Model of Trade Disputes

Giovanni Maggi and Robert W. Staiger

Yale and Dartmouth

June 2016

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 1 / 27

Introduction

There have been over 500 disputes in the WTO since 1995.Sometimes govs settle early, sometimes they “fight it out” to a court(DSB) ruling. In GATT+WTO, about 50% of disputes settle early.

Stakes of trade disputes can be large, so important to understandwhat determines dispute initiation and resolution.

Some interesting dynamic patterns: countries fight less as time goesby. Plots 1+2.

The judicial system is being used less and less... Is this bad news?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 2 / 27

Introduction

There have been over 500 disputes in the WTO since 1995.Sometimes govs settle early, sometimes they “fight it out” to a court(DSB) ruling. In GATT+WTO, about 50% of disputes settle early.

Stakes of trade disputes can be large, so important to understandwhat determines dispute initiation and resolution.

Some interesting dynamic patterns: countries fight less as time goesby. Plots 1+2.

The judicial system is being used less and less... Is this bad news?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 2 / 27

Introduction

There have been over 500 disputes in the WTO since 1995.Sometimes govs settle early, sometimes they “fight it out” to a court(DSB) ruling. In GATT+WTO, about 50% of disputes settle early.

Stakes of trade disputes can be large, so important to understandwhat determines dispute initiation and resolution.

Some interesting dynamic patterns: countries fight less as time goesby. Plots 1+2.

The judicial system is being used less and less... Is this bad news?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 2 / 27

1995 1999 2004 20095

10

15

20

25

30

35

40

45

50

Num

ber

(wto

dis

pute

s)

Plot 1

RULINGDISPUTE

0 2 4 6 8 10 12 14

Age of Dyad in Years

0

1

2

3

4

5

6#10-6 Plot 2

RULINGDISPUTE

Note: The vertical axis records the ratio X/Z, where X is the average number of disputes or rulings involving dyads of a given age, and Z is the average trade volume of dyads in this age group.

Introduction

There have been over 500 disputes in the WTO since 1995.Sometimes govs settle early, sometimes they “fight it out” to a court(DSB) ruling. In GATT+WTO, about 50% of disputes settle early.

Stakes of trade disputes can be large, so important to understandwhat determines dispute initiation and resolution.

Some interesting dynamic patterns: countries fight less as time goesby. Plots 1+2.

The judicial system is being used less and less... Is this bad news?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 2 / 27

Introduction

We propose a theory that can attribute this trend to the effects ofjudicial learning (good news, not bad news).

A key model prediction: If there is court learning by ruling, thefrequency of disputes and rulings should decline with court experience.

We attempt to gauge empirically the importance, scope and form ofjudicial learning in the WTO dispute settlement system.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 3 / 27

Introduction

In the empirical literature on learning by doing, econometricians (tryto) measure productivity directly.

We can’t measure court accuracy directly, but we can try to inferlearning effects indirectly:

We check if WTO data exhibits imprint of court learning in accordancewith our model predictions.

We seek to shed light on the scope and form of court learning asviewed through the lens of our model.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 4 / 27

Introduction: the model

Key ingredients of our model:

Importing gov chooses a policy, then exporting gov chooses whether todispute;

If a dispute is initiated, govs bargain “in the shadow of the law,”subject to negotiation costs;

If invoked, DSB issues a ruling to maximize govs’joint payoff based onnoisy information;

Learning by ruling: court accuracy increases with experience, but atdiminishing rate;

Govs are “large”players that repeatedly engage in disputes (so theyinternalize benefits of court learning).

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 5 / 27

Introduction: preview of theoretical results

In a static setting, there is never a ruling. But a dispute can arise,and is more likely if DSB less accurate.

In a dynamic setting, the presence of court learning can give rise toequilibrium rulings.

Likelihood of rulings and disputes decreases w/ cumulative rulings, atleast if govs are patient enough.

Results above largely extend to a multi-country setting with learningspillovers;

Also to setting where disputed policy is discrete and transfers are costly(and baseline rate of rulings exists even in static setting).

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 6 / 27

Introduction: preview of empirical results

Focus on the prediction that likelihood of rulings/disputes decreaseswith cumulative rulings.

Allow scope of court learning to be general, specific to the disputingcountries, or specific to the issue area (e.g. GATT/WTO Article).

We find evidence of article-specific and disputant-specific learning,but only weak evidence of general-scope learning.

Once learning by ruling taken into account, estimated time trend ofdisputes and rulings turns positive.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 7 / 27

Introduction: related literature

Maggi and Staiger (2011): does not allow for bargaining/settlement,does not consider learning effects (and focuses on institutionaldesign). Maggi and Staiger (2015a): govs can settle, but model isstatic, with no dispute-initiation stage. Focuses only on how contractform (“property” vs “liability”) impacts prob of settlement.

Other models that generate trade disputes in equilibrium: Park(2011), Beshkar (2016), Staiger and Sykes (forthcoming). Main focusnot on determinants of dispute outcome.

Judicial learning: Baker and Mezzetti (2012), Beim (2014).

Empirical work on trade dispute outcomes: Guzman and Simmons(2002), Busch and Reinhardt (2001, 2006), Bown and Reynolds(2014), Maggi and Staiger (2015b).

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 8 / 27

The basic model

A single industry; importing gov (Home) chooses tariff T ; exportinggov (Foreign) passive in this industry.

Home’s payoff: ω(T , θ) = CS(T ) + R(T ) + θ · PS(T ).

Foreign’s payoff: ω∗(T ) = CS∗(T ) + PS∗(T ).

Joint-payoff-maximizing policy: T fb = argmaxT [ω(T , θ) +ω∗(T )].

Veil of ignorance: ex-ante each gov equally likely to be importer orexporter (i.e. claimant or defendant).

Absent international transfers, Pareto frontier is concave. Withtransfers, Pareto frontier is linear with slope -1. Figure 1a.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 9 / 27

Figure 1: Continuous policy (static setting)

ω*

ω

slope=-1

FB

N

The basic model

Political shock θ (“state of the world”) not verifiable.

Incomplete contract: does not specify policy T , but the court (DSB)can “fill gaps” ex-post.

If invoked, DSB observes a noisy signal of T fb , denotedT dsb = T fb + ε, with Var(ε) = σ2, and issues a ruling T dsb tomaximize govs’joint payoff.

Timing: (0) θ realized and observed by govs; (1) Importer chooses T ;(2) Exporter acquiesces or initiates dispute; (3) If dispute initiated,govs negotiate over policy T and a transfer; (4) If govs disagree, DSBruling is triggered.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 10 / 27

The dispute subgame

Bargaining protocol: each gov makes take-or-leave offer with prob1/2.

Govs automatically incur symmetric litigation cost C L if court istriggered.

Gov i’s net disagreement payoff: ωNDi = ωD

i − C L, whereωDi = E [ωi (T dsb , θ)|θ].

Negotiation costs: part of bargaining surplus melts away.

If ωBi is player i’s bargaining payoff absent negotiation costs, her payoff

gain from bargaining is κ(

ωBi −ωNDi

)with κ ∈ (0, 1).

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 11 / 27

Static benchmark

Backward induction: start with the dispute subgame, where govsbargain in the shadow of the law.

Given concave frontier, disagreement point D is below frontier, andSoutheast of FB point (b/c uncertainty about court ruling hurtsimporter, benefits exporter).

Litigation costs further worsen threat point for both govs (ND).Figure 1b.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 12 / 27

Figure 1: Continuous policy (static setting)

ω*

ω

slope=-1

FB

N

ND D

Static benchmark

If govs negotiate, given the negotiation cost κ they move part-waytoward negotiation frontier. Net bargaining payoff point Bnet.

Moving backwards (Figure 1c):

(i) if Bnet point below no-transfer frontier, Home chooses T s.t.Foreign is indifferent b/w complaining and not, hence no dispute (B0).

(ii) If Bnet above no-transfer frontier, Home triggers a dispute (bychoosing a bad T ).

Clearly there is never a ruling in equilibrium.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 13 / 27

Figure 1: Continuous policy (static setting)

ω*

ω

slope=-1

FB

N

ND

Bnet

B0

D

Static benchmark

Proposition 1: In the static setting: (i) there is never a DSB ruling;(ii) the likelihood of a dispute is increasing in σ.

Graphical intuition for part (ii): as σ ↑, point Bnet moves Southeaststarting below no-transfer frontier, and crosses this frontier from theleft. Figure 2.

Remark 1: In the static setting, the equilibrium joint payoff isdecreasing and piecewise concave in the DSB noise σ.

Note, decreasing σ increases joint surplus through off-equilibriumeffects, because there is no ruling in equilibrium:

Lower σ improves disagreement point in case of dispute.

If there is no dispute, lower σ improves the would-be negotiationoutcome, thus inducing Home to choose a better T (off-off-equilibriumeffect).

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 14 / 27

Figure 2: Continuous policy (static setting)

ω*

ω

FB

N

ND(σ) Bnet(σ)

Learning by ruling

Basic idea: DSB becomes more accurate with experience.

Two periods, t = 1, 2.

Discount factor δ ∈ (0,∞).

The same game takes place in each period, and θ is iid, solearning-by-ruling is the only source of dynamics.

We model learning-by-ruling similarly as in standard models oflearning-by-doing for firms:

ability of DSB increases with cumulative past rulings.

If there has been a ruling at t = 1, DSB is more accurate at t = 2, soσ is lower.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 15 / 27

Learning by ruling

At t = 2 of course there is no ruling. But learning can give rise toequilibrium rulings at t = 1.

Recall that at t = 2, joint payoff (Ω) is decreasing in σ.

Given veil of ignorance, going to court at t = 1 implies common futurepayoff gain ∆ > 0.Learning is beneficial (∆ > 0) even though there is no ruling inequilibrium at t = 2, because it improves disagreement point fortomorrow’s negotiation.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 16 / 27

Learning by ruling

At t = 1, disagreement payoffs are (ωND + δ∆,ω∗ND + δ∆).Graphically, label this point ND + δ∆.

If ND + δ∆ is above negotiation frontier, then a dispute ends inruling.

And going backwards, Home chooses a T that triggers a dispute.Figure 3.

Thus the presence of court learning (together with “large”players andnegotiation costs) can explain equilibrium rulings.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 17 / 27

Figure 3: Continuous policy (two period setting)

ω*

ω

FB

N

ND

D

ND+ Δ δ

Impact of court experience on current outcomes

How does court experience affect the likelihood of rulings anddisputes?

Keep two periods, t = 1, 2, but suppose there is an initial stock ofrulings x , inherited from a “past”period t = 0.

Will examine how likelihood of rulings and disputes at t = 1 dependson x .

Learning-by-ruling curve σ(x) decreasing and convex, withlimx→∞ σ(x) > 0.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 18 / 27

Impact of court experience on current rulings

Static effect: x ↑ reduces today’s ineffi ciency from going to court,because as σ ↓ the disagreement point ND gets closer to negotiationfrontier.

Dynamic effect: x ↑ reduces future gain from going to court (∆).This follows because EΩt=2 is concave in σ (Remark 1) and σ

′′> 0.

If δ suffi ciently large, dynamic effect dominates static effect andPr(Ruling) decreases with x .

Even if δ is small, Pr(Ruling) decreasing in x for x suffi ciently large,because as learning gets exhausted ∆→ 0, so Pr(Ruling) must hitzero at some point.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 19 / 27

Figure 3: Continuous policy (two period setting)

ω*

ω

FB

N

ND

D

ND+ Δ δ

Impact of court experience on current rulings

Proposition 2a: At t = 1, Pr(ruling) is decreasing in x for xsuffi ciently large, and is globally decreasing in x if δ is high enough.

Note, frequency of DSB use not a good measure of the institution’seffectiveness.

A declining ruling frequency is not bad news, in fact it’s a symptom ofbeneficial learning.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 20 / 27

Impact of court experience on current disputes

Next consider the impact of x on the probability of a dispute at t = 1.

As in static setting, there is dispute iff net bargaining payoff point(Bnet∆ ) above no-transfer frontier, except that now disagreement pointis ND + δ∆.

Proposition 2b: At t = 1, Pr(dispute) is decreasing in x for xsuffi ciently large, and is globally decreasing in x if δ is high enough.

Intuition for high δ case:

The “dynamic” effect (increasing x reduces ∆) dominates.This worsens disagreement point, and hence (due to negotiation costs)worsens net bargaining payoffs, making dispute less appealing.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 21 / 27

Impact of court experience on current settlement rate

Note, the model does not yield sharp predictions on the conditionallikelihood of settlement:

Remark 2: At t = 1 the likelihood of settlement conditional on adispute may go up or down with x , even if δ is high.

Intuition:

Effect of x ↑ on ruling margin (i.e. when ND + δ∆ is on negotiationfrontier) may be stronger or weaker than effect of x on dispute margin(i.e. when Bnet∆ is on no-transfer frontier), depending on probdistribution of θ.

So Pr(ruling)/Pr(dispute) can go either way.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 22 / 27

Empirical evidence

We focus on a key implication of the model: the likelihood of currentrulings and disputes should tend to decrease with cumulative rulings.

We show this implication generalizes to a many-country extension ofour model, where learning could be specific to an issue area, or to thedisputant countries, or it could be general.

So we explore the effect of cumulative rulings at various levels(country dyad, article, general).

Dual objective: (1) Check if key model prediction is consistent withdata; (2) If so, gauge the strength and scope of learning by ruling.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 23 / 27

Regressions

Focus first on the impact of court experience on the likelihood ofrulings. Start with undirected-dyads, then use directed dyads. Table 2.

Results are consistent with:

article-specific learning;

disputant-specific (directed dyad- and claimaint-specific) learning;

only weak evidence of general-scope learning.

Note positive coeffi cient of t trend: suggests that court learning canexplain declining trend in rulings (Plots 1+2).

Results of dispute regressions are similar, except for the positivedefendant-article specific effect: a “bandwagon”mechanism?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 24 / 27

Regressions

Focus first on the impact of court experience on the likelihood ofrulings. Start with undirected-dyads, then use directed dyads. Table 2.

Results are consistent with:

article-specific learning;

disputant-specific (directed dyad- and claimaint-specific) learning;

only weak evidence of general-scope learning.

Note positive coeffi cient of t trend: suggests that court learning canexplain declining trend in rulings (Plots 1+2).

Results of dispute regressions are similar, except for the positivedefendant-article specific effect: a “bandwagon”mechanism?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 24 / 27

Regressions

Focus first on the impact of court experience on the likelihood ofrulings. Start with undirected-dyads, then use directed dyads. Table 2.

Results are consistent with:

article-specific learning;

disputant-specific (directed dyad- and claimaint-specific) learning;

only weak evidence of general-scope learning.

Note positive coeffi cient of t trend: suggests that court learning canexplain declining trend in rulings (Plots 1+2).

Results of dispute regressions are similar, except for the positivedefendant-article specific effect: a “bandwagon”mechanism?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 24 / 27

(1) (2) (3) (4)

0.186*** 0.119 0.209*** 0.235**

(0.0646) (0.0913) (0.0786) (0.0999)

-0.0127* -0.160*** 0.117*** -0.0479

(0.00761) (0.0168) (0.0186) (0.0299)

-0.0272*** -0.0644*** 0.0293 -0.0799

(0.00256) (0.00540) (0.0277) (0.0502)

-0.00190** -0.00663*** -0.0128 -0.416**

(0.000841) (0.00199) (0.0941) (0.184)

-0.0374*** -0.196***

(0.00844) (0.0193)

-0.0509*** -0.102***

(0.00410) (0.00775)

0.000918 0.00122

(0.00109) (0.00253)

-0.0107*** -0.0212***

(0.00188) (0.00448)

-0.000515 -0.0248***

(0.00362) (0.00912)

-0.00130 -0.00691***

(0.000834) (0.00203)

t 0.198*** 1.277*** t 0.235*** 1.409***

(0.0260) (0.111) (0.0258) (0.117)

t2 -0.00726*** -0.0313*** t2 -0.0104*** -0.0373***

(0.00225) (0.00416) (0.00225) (0.00433)

Constant -9.201*** -10.90*** Constant -9.701*** -12.87***

(1.076) (0.723) (1.077) (0.792)

Observations 439,584 112,560 Observations 545,142 149,520Pseudo R2 0.277 0.229 Pseudo R2 0.242 0.218

Y Y Y Y kFE Y Y kFE Y Y

are undirected (directed) dyad fixed effects. kFE are article fixed effects.*** p<0.01, ** p<0.05, * p<0.1

Table 2: Logit

VARIABLESUndirected Dyad

VARIABLESDirected Dyad

Standard errors in parentheses

CR_𝑖𝑖𝑖𝑖kt CR_𝑖𝑖𝑖𝑖kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖kt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖kt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜kt

CR_𝑖𝑖𝑖𝑖𝑛𝑛kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖nkt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)nkt

CR_𝑖𝑖𝑖𝑖nkt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜nkt

CR_n(𝑖𝑖𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖nkt

CR_n(𝑖𝑖𝑖𝑖)nkt

DLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt DLogit_𝑖𝑖𝑖𝑖kt

𝑖𝑖𝑖𝑖FE 𝑖𝑖𝑖𝑖FE

𝑖𝑖𝑖𝑖FE (𝑖𝑖𝑖𝑖FE)

Regressions

Focus first on the impact of court experience on the likelihood ofrulings. Start with undirected-dyads, then use directed dyads. Table 2.

Results are consistent with:

article-specific learning;

disputant-specific (directed dyad- and claimaint-specific) learning;

only weak evidence of general-scope learning.

Note positive coeffi cient of t trend: suggests that court learning canexplain declining trend in rulings (Plots 1+2).

Results of dispute regressions are similar, except for the positivedefendant-article specific effect: a “bandwagon”mechanism?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 24 / 27

(1) (2) (3) (4)

0.186*** 0.119 0.209*** 0.235**

(0.0646) (0.0913) (0.0786) (0.0999)

-0.0127* -0.160*** 0.117*** -0.0479

(0.00761) (0.0168) (0.0186) (0.0299)

-0.0272*** -0.0644*** 0.0293 -0.0799

(0.00256) (0.00540) (0.0277) (0.0502)

-0.00190** -0.00663*** -0.0128 -0.416**

(0.000841) (0.00199) (0.0941) (0.184)

-0.0374*** -0.196***

(0.00844) (0.0193)

-0.0509*** -0.102***

(0.00410) (0.00775)

0.000918 0.00122

(0.00109) (0.00253)

-0.0107*** -0.0212***

(0.00188) (0.00448)

-0.000515 -0.0248***

(0.00362) (0.00912)

-0.00130 -0.00691***

(0.000834) (0.00203)

t 0.198*** 1.277*** t 0.235*** 1.409***

(0.0260) (0.111) (0.0258) (0.117)

t2 -0.00726*** -0.0313*** t2 -0.0104*** -0.0373***

(0.00225) (0.00416) (0.00225) (0.00433)

Constant -9.201*** -10.90*** Constant -9.701*** -12.87***

(1.076) (0.723) (1.077) (0.792)

Observations 439,584 112,560 Observations 545,142 149,520Pseudo R2 0.277 0.229 Pseudo R2 0.242 0.218

Y Y Y Y kFE Y Y kFE Y Y

are undirected (directed) dyad fixed effects. kFE are article fixed effects.*** p<0.01, ** p<0.05, * p<0.1

Table 2: Logit

VARIABLESUndirected Dyad

VARIABLESDirected Dyad

Standard errors in parentheses

CR_𝑖𝑖𝑖𝑖kt CR_𝑖𝑖𝑖𝑖kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖kt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖kt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜kt

CR_𝑖𝑖𝑖𝑖𝑛𝑛kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖nkt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)nkt

CR_𝑖𝑖𝑖𝑖nkt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜nkt

CR_n(𝑖𝑖𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖nkt

CR_n(𝑖𝑖𝑖𝑖)nkt

DLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt DLogit_𝑖𝑖𝑖𝑖kt

𝑖𝑖𝑖𝑖FE 𝑖𝑖𝑖𝑖FE

𝑖𝑖𝑖𝑖FE (𝑖𝑖𝑖𝑖FE)

Regressions

Focus first on the impact of court experience on the likelihood ofrulings. Start with undirected-dyads, then use directed dyads. Table 2.

Results are consistent with:

article-specific learning;

disputant-specific (directed dyad- and claimaint-specific) learning;

only weak evidence of general-scope learning.

Note positive coeffi cient of t trend: suggests that court learning canexplain declining trend in rulings (Plots 1+2).

Results of dispute regressions are similar, except for the positivedefendant-article specific effect: a “bandwagon”mechanism?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 24 / 27

(1) (2) (3) (4)

0.186*** 0.119 0.209*** 0.235**

(0.0646) (0.0913) (0.0786) (0.0999)

-0.0127* -0.160*** 0.117*** -0.0479

(0.00761) (0.0168) (0.0186) (0.0299)

-0.0272*** -0.0644*** 0.0293 -0.0799

(0.00256) (0.00540) (0.0277) (0.0502)

-0.00190** -0.00663*** -0.0128 -0.416**

(0.000841) (0.00199) (0.0941) (0.184)

-0.0374*** -0.196***

(0.00844) (0.0193)

-0.0509*** -0.102***

(0.00410) (0.00775)

0.000918 0.00122

(0.00109) (0.00253)

-0.0107*** -0.0212***

(0.00188) (0.00448)

-0.000515 -0.0248***

(0.00362) (0.00912)

-0.00130 -0.00691***

(0.000834) (0.00203)

t 0.198*** 1.277*** t 0.235*** 1.409***

(0.0260) (0.111) (0.0258) (0.117)

t2 -0.00726*** -0.0313*** t2 -0.0104*** -0.0373***

(0.00225) (0.00416) (0.00225) (0.00433)

Constant -9.201*** -10.90*** Constant -9.701*** -12.87***

(1.076) (0.723) (1.077) (0.792)

Observations 439,584 112,560 Observations 545,142 149,520Pseudo R2 0.277 0.229 Pseudo R2 0.242 0.218

Y Y Y Y kFE Y Y kFE Y Y

are undirected (directed) dyad fixed effects. kFE are article fixed effects.*** p<0.01, ** p<0.05, * p<0.1

Table 2: Logit

VARIABLESUndirected Dyad

VARIABLESDirected Dyad

Standard errors in parentheses

CR_𝑖𝑖𝑖𝑖kt CR_𝑖𝑖𝑖𝑖kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖kt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖kt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜kt

CR_𝑖𝑖𝑖𝑖𝑛𝑛kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖nkt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)nkt

CR_𝑖𝑖𝑖𝑖nkt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜nkt

CR_n(𝑖𝑖𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖nkt

CR_n(𝑖𝑖𝑖𝑖)nkt

DLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt DLogit_𝑖𝑖𝑖𝑖kt

𝑖𝑖𝑖𝑖FE 𝑖𝑖𝑖𝑖FE

𝑖𝑖𝑖𝑖FE (𝑖𝑖𝑖𝑖FE)

(1) (2) (3) (4)

0.0345*** 0.0138*** 0.0150*** 0.0119**

(0.00734) (0.00470) (0.00525) (0.00549)

-0.00140*** -0.000849** 0.00288** 0.000884

(0.000474) (0.000354) (0.00127) (0.000797)

-0.00135*** -0.000525*** 0.00100 0.000407

(0.000362) (0.000126) (0.00139) (0.000927)

-1.09e-05 -8.13e-06 0.0128** -0.00122

(1.32e-05) (1.76e-05) (0.00493) (0.00250)

-0.00160*** -0.000957***

(0.000374) (0.000254)

-0.00117*** -0.000713***

(0.000188) (0.000187)

-5.47e-06 6.72e-06

(1.28e-05) (2.23e-05)

-5.33e-05** -5.12e-05*

(2.13e-05) (2.82e-05)

-0.000199 8.92e-05

(0.000261) (0.000103)

-6.42e-06 -3.11e-06

(1.01e-05) (1.55e-05)

t 0.00130** 0.00317*** t 0.00112** 0.00249***

(0.000603) (0.000699) (0.000435) (0.000521)

t2 -1.84e-05 -0.000106 t2 -1.90e-05 -8.47e-05*

(3.17e-05) (6.77e-05) (2.44e-05) (5.00e-05)

Constant -0.00806** 0.00311 Constant -0.00820*** -6.12e-06

(0.00329) (0.00377) (0.00256) (0.00289)

Observations 439,584 112,560 Observations 545,142 149,520R2 0.044 0.021 R2 0.029 0.017

Y Y Y Y kFE Y Y kFE Y YCE CE

are undirected (directed) dyad fixed effects. kFE are article fixed effects.CE are standard errors clustered by undirected ( ) or directed ( ) dyads

*** p<0.01, ** p<0.05, * p<0.1

Table 3: OLS

VARIABLESUndirected Dyad

VARIABLESDirected Dyad

Standard errors in parentheses

CR_𝑖𝑖𝑖𝑖kt CR_𝑖𝑖𝑖𝑖kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖kt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖kt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜kt

CR_𝑖𝑖𝑖𝑖𝑛𝑛kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖nkt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)nkt

CR_𝑖𝑖𝑖𝑖nkt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜nkt

CR_n(𝑖𝑖𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖nkt

CR_n(𝑖𝑖𝑖𝑖)nkt

D_𝑖𝑖𝑖𝑖kt R_𝑖𝑖𝑖𝑖kt R_𝑖𝑖𝑖𝑖kt D_𝑖𝑖𝑖𝑖kt

𝑖𝑖𝑖𝑖FE 𝑖𝑖𝑖𝑖FE

𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖

𝑖𝑖𝑖𝑖FE (𝑖𝑖𝑖𝑖FE) 𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖

Regressions

Focus first on the impact of court experience on the likelihood ofrulings. Start with undirected-dyads, then use directed dyads. Table 2.

Results are consistent with:

article-specific learning;

disputant-specific (directed dyad- and claimaint-specific) learning;

only weak evidence of general-scope learning.

Note positive coeffi cient of t trend: suggests that court learning canexplain declining trend in rulings (Plots 1+2).

Results of dispute regressions are similar, except for the positivedefendant-article specific effect: a “bandwagon”mechanism?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 24 / 27

(1) (2) (3) (4)

0.186*** 0.119 0.209*** 0.235**

(0.0646) (0.0913) (0.0786) (0.0999)

-0.0127* -0.160*** 0.117*** -0.0479

(0.00761) (0.0168) (0.0186) (0.0299)

-0.0272*** -0.0644*** 0.0293 -0.0799

(0.00256) (0.00540) (0.0277) (0.0502)

-0.00190** -0.00663*** -0.0128 -0.416**

(0.000841) (0.00199) (0.0941) (0.184)

-0.0374*** -0.196***

(0.00844) (0.0193)

-0.0509*** -0.102***

(0.00410) (0.00775)

0.000918 0.00122

(0.00109) (0.00253)

-0.0107*** -0.0212***

(0.00188) (0.00448)

-0.000515 -0.0248***

(0.00362) (0.00912)

-0.00130 -0.00691***

(0.000834) (0.00203)

t 0.198*** 1.277*** t 0.235*** 1.409***

(0.0260) (0.111) (0.0258) (0.117)

t2 -0.00726*** -0.0313*** t2 -0.0104*** -0.0373***

(0.00225) (0.00416) (0.00225) (0.00433)

Constant -9.201*** -10.90*** Constant -9.701*** -12.87***

(1.076) (0.723) (1.077) (0.792)

Observations 439,584 112,560 Observations 545,142 149,520Pseudo R2 0.277 0.229 Pseudo R2 0.242 0.218

Y Y Y Y kFE Y Y kFE Y Y

are undirected (directed) dyad fixed effects. kFE are article fixed effects.*** p<0.01, ** p<0.05, * p<0.1

Table 2: Logit

VARIABLESUndirected Dyad

VARIABLESDirected Dyad

Standard errors in parentheses

CR_𝑖𝑖𝑖𝑖kt CR_𝑖𝑖𝑖𝑖kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖kt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖kt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜kt

CR_𝑖𝑖𝑖𝑖𝑛𝑛kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖nkt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)nkt

CR_𝑖𝑖𝑖𝑖nkt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜nkt

CR_n(𝑖𝑖𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖nkt

CR_n(𝑖𝑖𝑖𝑖)nkt

DLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt DLogit_𝑖𝑖𝑖𝑖kt

𝑖𝑖𝑖𝑖FE 𝑖𝑖𝑖𝑖FE

𝑖𝑖𝑖𝑖FE (𝑖𝑖𝑖𝑖FE)

1995 1999 2004 20095

10

15

20

25

30

35

40

45

50

Num

ber

(wto

dis

pute

s)

Plot 1

RULINGDISPUTE

0 2 4 6 8 10 12 14

Age of Dyad in Years

0

1

2

3

4

5

6#10-6 Plot 2

RULINGDISPUTE

Note: The vertical axis records the ratio X/Z, where X is the average number of disputes or rulings involving dyads of a given age, and Z is the average trade volume of dyads in this age group.

Regressions

Focus first on the impact of court experience on the likelihood ofrulings. Start with undirected-dyads, then use directed dyads. Table 2.

Results are consistent with:

article-specific learning;

disputant-specific (directed dyad- and claimaint-specific) learning;

only weak evidence of general-scope learning.

Note positive coeffi cient of t trend: suggests that court learning canexplain declining trend in rulings (Plots 1+2).

Results of dispute regressions are similar, except for the positivedefendant-article specific effect: a “bandwagon”mechanism?

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 24 / 27

(1) (2) (3) (4)

0.186*** 0.119 0.209*** 0.235**

(0.0646) (0.0913) (0.0786) (0.0999)

-0.0127* -0.160*** 0.117*** -0.0479

(0.00761) (0.0168) (0.0186) (0.0299)

-0.0272*** -0.0644*** 0.0293 -0.0799

(0.00256) (0.00540) (0.0277) (0.0502)

-0.00190** -0.00663*** -0.0128 -0.416**

(0.000841) (0.00199) (0.0941) (0.184)

-0.0374*** -0.196***

(0.00844) (0.0193)

-0.0509*** -0.102***

(0.00410) (0.00775)

0.000918 0.00122

(0.00109) (0.00253)

-0.0107*** -0.0212***

(0.00188) (0.00448)

-0.000515 -0.0248***

(0.00362) (0.00912)

-0.00130 -0.00691***

(0.000834) (0.00203)

t 0.198*** 1.277*** t 0.235*** 1.409***

(0.0260) (0.111) (0.0258) (0.117)

t2 -0.00726*** -0.0313*** t2 -0.0104*** -0.0373***

(0.00225) (0.00416) (0.00225) (0.00433)

Constant -9.201*** -10.90*** Constant -9.701*** -12.87***

(1.076) (0.723) (1.077) (0.792)

Observations 439,584 112,560 Observations 545,142 149,520Pseudo R2 0.277 0.229 Pseudo R2 0.242 0.218

Y Y Y Y kFE Y Y kFE Y Y

are undirected (directed) dyad fixed effects. kFE are article fixed effects.*** p<0.01, ** p<0.05, * p<0.1

Table 2: Logit

VARIABLESUndirected Dyad

VARIABLESDirected Dyad

Standard errors in parentheses

CR_𝑖𝑖𝑖𝑖kt CR_𝑖𝑖𝑖𝑖kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖kt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖kt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜kt

CR_𝑖𝑖𝑖𝑖𝑛𝑛kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖nkt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)nkt

CR_𝑖𝑖𝑖𝑖nkt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜nkt

CR_n(𝑖𝑖𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖nkt

CR_n(𝑖𝑖𝑖𝑖)nkt

DLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt DLogit_𝑖𝑖𝑖𝑖kt

𝑖𝑖𝑖𝑖FE 𝑖𝑖𝑖𝑖FE

𝑖𝑖𝑖𝑖FE (𝑖𝑖𝑖𝑖FE)

Regressions

Note lack of evidence for most narrow form of court learning.

Could reflect upward bias from unobserved heterogeneity at the(undirected or directed) dyad-and-article level.

The countries in dyad−→ij might have a particularly litigious relationship

over article k.

We can check this interpretation by including−→ij k fixed effect.

Identification only from within−→ij k variation over time; diminished

ability to assess impact of variables with little within-−→ij k variation.

Introduces incidental parameter problem for rulings regression; focus ondispute regression where this problem does not arise.

Table 4

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 25 / 27

(1) (2) (3) (4)

0.186*** 0.119 0.209*** 0.235**

(0.0646) (0.0913) (0.0786) (0.0999)

-0.0127* -0.160*** 0.117*** -0.0479

(0.00761) (0.0168) (0.0186) (0.0299)

-0.0272*** -0.0644*** 0.0293 -0.0799

(0.00256) (0.00540) (0.0277) (0.0502)

-0.00190** -0.00663*** -0.0128 -0.416**

(0.000841) (0.00199) (0.0941) (0.184)

-0.0374*** -0.196***

(0.00844) (0.0193)

-0.0509*** -0.102***

(0.00410) (0.00775)

0.000918 0.00122

(0.00109) (0.00253)

-0.0107*** -0.0212***

(0.00188) (0.00448)

-0.000515 -0.0248***

(0.00362) (0.00912)

-0.00130 -0.00691***

(0.000834) (0.00203)

t 0.198*** 1.277*** t 0.235*** 1.409***

(0.0260) (0.111) (0.0258) (0.117)

t2 -0.00726*** -0.0313*** t2 -0.0104*** -0.0373***

(0.00225) (0.00416) (0.00225) (0.00433)

Constant -9.201*** -10.90*** Constant -9.701*** -12.87***

(1.076) (0.723) (1.077) (0.792)

Observations 439,584 112,560 Observations 545,142 149,520Pseudo R2 0.277 0.229 Pseudo R2 0.242 0.218

Y Y Y Y kFE Y Y kFE Y Y

are undirected (directed) dyad fixed effects. kFE are article fixed effects.*** p<0.01, ** p<0.05, * p<0.1

Table 2: Logit

VARIABLESUndirected Dyad

VARIABLESDirected Dyad

Standard errors in parentheses

CR_𝑖𝑖𝑖𝑖kt CR_𝑖𝑖𝑖𝑖kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖kt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖kt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜kt

CR_𝑖𝑖𝑖𝑖𝑛𝑛kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖nkt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)nkt

CR_𝑖𝑖𝑖𝑖nkt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜nkt

CR_n(𝑖𝑖𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖nkt

CR_n(𝑖𝑖𝑖𝑖)nkt

DLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt RLogit_𝑖𝑖𝑖𝑖kt DLogit_𝑖𝑖𝑖𝑖kt

𝑖𝑖𝑖𝑖FE 𝑖𝑖𝑖𝑖FE

𝑖𝑖𝑖𝑖FE (𝑖𝑖𝑖𝑖FE)

Regressions

Note lack of evidence for most narrow form of court learning.

Could reflect upward bias from unobserved heterogeneity at the(undirected or directed) dyad-and-article level.

The countries in dyad−→ij might have a particularly litigious relationship

over article k.

We can check this interpretation by including−→ij k fixed effect.

Identification only from within−→ij k variation over time; diminished

ability to assess impact of variables with little within-−→ij k variation.

Introduces incidental parameter problem for rulings regression; focus ondispute regression where this problem does not arise.

Table 4

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 25 / 27

Regressions

Note lack of evidence for most narrow form of court learning.

Could reflect upward bias from unobserved heterogeneity at the(undirected or directed) dyad-and-article level.

The countries in dyad−→ij might have a particularly litigious relationship

over article k.

We can check this interpretation by including−→ij k fixed effect.

Identification only from within−→ij k variation over time; diminished

ability to assess impact of variables with little within-−→ij k variation.

Introduces incidental parameter problem for rulings regression; focus ondispute regression where this problem does not arise.

Table 4

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 25 / 27

Regressions

Note lack of evidence for most narrow form of court learning.

Could reflect upward bias from unobserved heterogeneity at the(undirected or directed) dyad-and-article level.

The countries in dyad−→ij might have a particularly litigious relationship

over article k.

We can check this interpretation by including−→ij k fixed effect.

Identification only from within−→ij k variation over time; diminished

ability to assess impact of variables with little within-−→ij k variation.

Introduces incidental parameter problem for rulings regression; focus ondispute regression where this problem does not arise.

Table 4

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 25 / 27

Regressions

Note lack of evidence for most narrow form of court learning.

Could reflect upward bias from unobserved heterogeneity at the(undirected or directed) dyad-and-article level.

The countries in dyad−→ij might have a particularly litigious relationship

over article k.

We can check this interpretation by including−→ij k fixed effect.

Identification only from within−→ij k variation over time; diminished

ability to assess impact of variables with little within-−→ij k variation.

Introduces incidental parameter problem for rulings regression; focus ondispute regression where this problem does not arise.

Table 4

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 25 / 27

Regressions

Note lack of evidence for most narrow form of court learning.

Could reflect upward bias from unobserved heterogeneity at the(undirected or directed) dyad-and-article level.

The countries in dyad−→ij might have a particularly litigious relationship

over article k.

We can check this interpretation by including−→ij k fixed effect.

Identification only from within−→ij k variation over time; diminished

ability to assess impact of variables with little within-−→ij k variation.

Introduces incidental parameter problem for rulings regression; focus ondispute regression where this problem does not arise.

Table 4

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 25 / 27

Regressions

Note lack of evidence for most narrow form of court learning.

Could reflect upward bias from unobserved heterogeneity at the(undirected or directed) dyad-and-article level.

The countries in dyad−→ij might have a particularly litigious relationship

over article k.

We can check this interpretation by including−→ij k fixed effect.

Identification only from within−→ij k variation over time; diminished

ability to assess impact of variables with little within-−→ij k variation.

Introduces incidental parameter problem for rulings regression; focus ondispute regression where this problem does not arise.

Table 4

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 25 / 27

VARIABLES VARIABLES(1) (2) (3) (4)

LOGIT OLS LOGIT OLS

-0.932*** -0.148*** -1.262*** -0.152***

(0.0853) (0.0128) (0.102) (0.0115)

0.0236*** 3.07e-06 0.0582** 0.00207*

(0.00835) (0.000268) (0.0276) (0.00113)

-0.0134*** -0.000593* -0.0101 -0.00392**

(0.00254) (0.000304) (0.0441) (0.00189)

-0.00219*** -1.68e-05 0.182 0.00414

(0.000845) (1.38e-05) (0.133) (0.00773)

0.00778 5.52e-05

(0.0108) (0.000210)

-0.0276*** -0.000476***

(0.00403) (0.000144)

0.00177 -2.09e-06

(0.00117) (1.28e-05)

-0.0101*** -3.27e-05*

(0.00198) (1.96e-05)

-0.00204 -0.000163

(0.00371) (0.000245)

-0.00178** -1.33e-05

(0.000846) (1.05e-05)

t 0.204*** 0.00130** t 0.246*** 0.00112**

(0.0262) (0.000603) (0.0262) (0.000435)

t2 -0.00743*** -1.84e-05 t2 -0.0106*** -1.90e-05

(0.00225) (3.16e-05) (0.00226) (2.44e-05)

Constant 0.00428** Constant 0.00332**(0.00181) (0.00136)

Observations 26,253 439,584 Observations 29,193 545,142(Pseudo) R2 0.0392 0.024 (Pseudo) R2 0.0516 0.023

Y Y Y YCE N CE N

CE are standard errors clustered by undirected ( ) or directed ( ) dyads are undirected (directed) dyad-and-article fixed effects.*** p<0.01, ** p<0.05, * p<0.1Standard errors in parentheses

Table 4: Dispute Regressions with Dyad-and-Article Fixed Effects

Undirected Dyad Directed Dyad

CR_𝑖𝑖𝑖𝑖kt CR_𝑖𝑖𝑖𝑖kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖kt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖kt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜kt

CR_𝑖𝑖𝑖𝑖𝑛𝑛kt

CR_ 𝑛𝑛𝑖𝑖 𝑖𝑖nkt

CR_𝑖𝑖(𝑛𝑛𝑖𝑖)nkt

CR_𝑖𝑖𝑖𝑖nkt

CR_𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜nkt

CR_n(𝑖𝑖𝑖𝑖)kt

CR_𝑖𝑖𝑖𝑖nkt

CR_n(𝑖𝑖𝑖𝑖)nkt

DLogit_𝑖𝑖𝑖𝑖kt D_𝑖𝑖𝑖𝑖kt D_𝑖𝑖𝑖𝑖kt DLogit_𝑖𝑖𝑖𝑖kt

𝑖𝑖𝑖𝑖kFE 𝑖𝑖𝑖𝑖kFE 𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖

𝑖𝑖𝑖𝑖kFE (𝑖𝑖𝑖𝑖kFE) 𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖

Alternative interpretations of empirical evidence

Through the lens of our model:

court learning of the particular scope and form we emphasize is theonly way to account for our empirical findings.

Mechanisms outside our model:

In principle legal precedent could explain article-specific effects, but itis unlikely to explain disputant-specific effects.

Govs learning about each other? Our data does not support this story:likelihood of rulings does not decrease with cumulative settlements.

A backlog of cases coming out of GATT, hence the flurry of disputes inearly WTO years? If so, we would expect cumulative settlements tohave similar impact as cumulative rulings, but this is not the case.

The WTO court hit a resource constraint? Hard to square that withthe pattern of cumulative ruling effects we find.

“Bad news” story? Not obvious this would predict that both rulingsand disputes decline with cumulative rulings as our findings indicate.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 26 / 27

Conclusion

We explore the idea that judicial learning may contribute to explainthe declining trend in WTO disputes and rulings.

At the theoretical level, we study the implications of learning-by-rulingfor the dynamics of trade disputes.

At the empirical level, we seek to gauge the scope and strength ofjudicial learning.

Road ahead: estimate learning-by-ruling curve and spillovers, andideally get at welfare implications.

Maggi and Staiger (Yale and Dartmouth) Trade disputes June 2016 27 / 27