The Economic Structure of International
Trade-in-Services Agreements
Robert W. Staiger and Alan O. Sykes
Dartmouth and Stanford
December 2016
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
1 / 28
Introduction
There is now an established literature on the economics of
international trade agreements
successful in illuminated many features of real-world trade agreements
focused almost entirely on trade in goods
a focus that made sense when most services were non-traded
But the importance of trade in services has grown rapidly over the
past several decades
services now at the top of the trade liberalization agenda
The need for the literature to consider trade-in-services agreements
has become more pressing
In this paper we take a …rst step in …lling this lacuna
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
2 / 28
Introduction
WTO agreements cover both goods trade and trade in services
GATT is the central WTO agreement covering goods trade
GATS is the WTO agreement covering services trade
There are striking di¤erences between GATT and GATS
The broad structure of GATT can be understood from the perspective
of the ToT theory
We show that the broad structure of GATS can also be understood
through the lens of the ToT theory
but only if this theory is augmented with a set of restrictions, re‡ecting
salient features of services trade, on the policy instruments and trade
data available to govs
This is the main positive message of our paper
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
3 / 28
Introduction
The GATT/WTO has been highly successful in liberalizing goods
trade; GATS has been largely unsuccessful in liberalizing services trade
What explains this di¤erence in success?
A potential reason: the distinct nature of integration that each
agreement has attempted
Both agreements seek to expand market access,
but GATT was designed for “shallow integration,” expanding market
access primarily through negotiated reductions in tari¤s
while GATS re‡ects an orientation towards “deep integration,”
expanding market access primarily through negotiated change or
removal of domestic regulations in the service sector
=) GATS raises signi…cant challenges for negotiations seeking to
expand market access that do not arise with GATT
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
4 / 28
Introduction
Our augmented ToT model
can help interpret the deep-integration focus of GATS
and clarify the underlying problems that a trade-in-services agreement
must solve
An understanding of the underlying problems can inform the
consideration of alternative design approaches to solve the problems
We …nd that a shallow-integration approach more in line with that
taken by GATT might be possible in a trade-in-services agreement
thereby sidestepping some of the most contentious issues that may
have stymied negotiation progress thus far
This is the main normative message of our paper
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
5 / 28
Institutional Background: GATT vs GATS
GATT market access liberalization: tari¢ cation & shallow integration
=) Concentrate protective measures in the form of tari¤s by
agreeing to certain across-the-board prohibitions
in addition to MFN obligation, which prohibits tari¤ discrimination
across trading partners
use of quantitative restrictions prohibited
use of domestic taxation/regulation that discriminates against imported
goods prohibited by national treatment (NT) obligation
further elaboration of NT obligations in WTO TBT/SPS Agreements
=) Negotiate tari¤ reductions
detailed product-by-product tari¤ commitments
=) Market access implications of agreed tari¤ commitments secured
by MFN/NT/TBT/SPS and nonviolation (NV) clause
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
6 / 28
Institutional Background: GATT vs GATS
GATS market access liberalization: deep integration
Primarily “Mode 3” services
commercial presence in importing nation by a foreign service provider
=) No concentration of protective measures into any particular form
=) Other than MFN, no across-the-board prohibitions of any kind
=) Sector-by-sector negotiations over behind-the-border measures
relaxation/removal of quantitative restrictions, ownership restrictions,
licensing restrictions
even NT obligations
=) Market access implications of agreed commitments secured by
MFN and NV clause, and NT where NT agreed
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
7 / 28
A Benchmark Model of Services Trade
A simple partial equilibrium model of trade between two countries
“Mode 3” service demanded only in the domestic country
D=α
P, P the consumer price of the service in the domestic market
The service must be produced where it is consumed (Mode 3)
A market imperfection: its consumption generates a local “eye sore”
negative externality
noise/dust levels associated with construction services
health/safety risks imposed on general public by sub-standard service
suppliers in the construction, food or health services industries
general risks to the domestic …nancial system associated with
imprudent banking practices
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
8 / 28
Benchmark Model
The domestic gov imposes a regulatory standard as a condition of
entry, r for domestic and ρ for foreign service providers
per-unit externality levels θ (r ) and θ (ρ), θ decreasing and convex
The domestic gov can invest (Id and If ) in reducing
compliance/conformity-assessment costs
per-unit compliance cost for standard s is φ(s, I ) = κ (s )
κ increasing/convex in s and λ increasing/concave in I
λ(I ) where
Supply of domestic and foreign service providers given respectively by
Sd
Sf
= qd
= qf
φ(r , Id ) for qd
φ ( r , Id )
φ(ρ, If ) for qf
φ(ρ, If )
with qd , qf the producer prices of the service in the domestic market
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
9 / 28
Benchmark Model
By de…nition import tari¤s/export taxes collected at the border
unavailable for Mode 3 service trade
But in Benchmark Model we allow an expansive list of non-tari¤ …scal
instruments (in speci…c terms, tax if positive, subsidy if negative)
a nondiscriminatory sales tax t imposed by the domestic gov
a discriminatory sales tax or surcharge tf levied on foreign service
providers by the domestic gov
a discriminatory sales tax or surcharge tf levied on foreign service
providers by the foreign gov
With all taxes set at non-prohibitive levels, the pricing relationships:
qd + t = P = qf + tf + t + tf
Note: one price P for the service in the domestic market regardless of
the standard to which it is supplied (eye sore externality)
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
10 / 28
Benchmark Model
De…ne the “world” price of the foreign service qw
qw = qd
qf + tf , note:
tf
Market clearing D = Sd + Sf determines the market-clearing world
price as a function of the tax and regulatory policies:
1
[α 2tf + tf t + φ(r , Id ) + φ(ρ, If )]
3
Market-clearing levels of each of the other prices as functions of the
tax and regulatory policies:
q̃w =
P̃
=
q̃d
=
q̃f
=
1
[α + (tf + tf ) + 2t + φ(r , Id ) + φ(ρ, If )]
3
1
[α + (tf + tf ) t + φ(r , Id ) + φ(ρ, If )]
3
1
[α 2(tf + tf ) t + φ(r , Id ) + φ(ρ, If )]
3
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
11 / 28
Benchmark Model
De…ne the market-clearing foreign producer price and world price of
the “raw” unregulated service:
q̃f0
q̃f
q̃w0
q̃w
1
[α
3
1
φ(ρ, If ) = [α
3
φ(ρ, If ) =
2(tf + tf )
2tf + tf
t + φ(r , Id )
t + φ(r , Id )
2φ(ρ, If )]
2φ(ρ, If )]
For any ρ and If , a one-to-one correspondence between q̃w and q̃w0 ,
but we refer to q̃w0 rather than q̃w as “the terms of trade” in services
Note
q̃f0 is also the trade volume in Mode 3 services
neither gov can impose a tari¤ on this trade volume, but each of the
non-tari¤ policies can act as an NTB to trade in Mode 3 services
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
12 / 28
Benchmark Model
Domestic welfare: CS plus PS plus TR, minus disutility of the
consumption-generated externality
Consumer surplus and producer surplus:
CS
=
Z α
P̃
PS
=
[α
Z q̃d
P ]dP
[q
φ(r ,Id )
CS (P̃ )
φ(r , Id )]dq
PS (r , Id , q̃d )
Trade and domestic tax revenue TR (r , ρ, Id , If , P̃, q̃d , q̃w0 ):
[P̃
q̃d ] [q̃d
φ(r , Id )] + [P̃
q̃w0
φ(ρ, If )] [(α
P̃ )
(q̃d
φ(r , Id ))
Utility cost of domestic externality (Z ):
Z
= θ (r ) [q̃d φ(r , Id )] + θ (ρ) [(α
Z (r , ρ, Id , P̃, q̃d )
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
P̃ )
(q̃d
φ(r , Id ))]
December 2016
13 / 28
Benchmark Model
Domestic welfare (wlog abstract from political economy):
W
= CS (P̃ ) + PS (r , Id , q̃d ) + TR (r , ρ, Id , If , P̃, q̃d , q̃w0 )
Z (r , ρ, Id , P̃, q̃d ) c [Id + If ]
W (r , ρ, Id , If , P̃, q̃d , q̃w0 )
Foreign welfare: CS plus politically-weighted PS plus TR
W (q̃f0 , q̃w0 ) = γ
PS (q̃f0 ) + TR (q̃f0 , q̃w0 )
where
PS (q̃f0 ) =
TR (q̃f0 , q̃w0 ) =
Z q̃ 0 +φ(ρ,If )
f
φ(ρ,If )
[q̃w0 q̃f0 ]
[ qf
φ(ρ, If )]dqf =
q̃f0
Z q̃ 0
f
0
qf dqf
Note: foreign producer surplus impacted by ρ only through foreign
supply decisions for the unregulated service (through q̃0 )
Note: Wq̃0w = q̃f0 < 0 and Wq̃ 0 = q̃f0 > 0, hence Wq̃0w + Wq̃ 0 = 0
w
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
w
December 2016
14 / 28
E¢ cient Policies
E¢ cient policies maximize joint welfare
World price q̃0w cancels from sum of domestic and foreign tax
revenue, so W + W = G (r , ρ, Id , If , P̃, q̃d , q̃f0 )
E¢ cient policies
∂θ (r E )
∂r
∂κ (r E )
= 0;
∂r
∂λ(IdE ) E
Sd
∂Id
∂θ (ρE )
∂ρ
∂λ(IfE ) E
Sf
∂If
c = 0;
∂κ (ρE )
=0
∂ρ
c =0
t E = θ (r E )
tfE + tf E =
Staiger and Sykes (Dartmouth and Stanford)
(γ
1) SfE
Trade-in-Services Agreements
December 2016
15 / 28
Nash Policies
Absent a trade-in-services agreement, Nash policies
∂θ (r N )
∂r
∂κ (r N )
∂r
∂λ(IdN ) N
Sd
∂Id
∂θ (ρN )
∂ρ
= 0;
∂λ(IfN ) N
Sf
∂If
c
= 0;
tN
= θ (r N )
tfN + tf N
=
(γ
∂κ (ρN )
=0
∂ρ
c =0
3
1) SfN + SfN
2
Comparing Nash and E¢ cient, only discriminatory sales taxes
distorted in Nash (ToT manipulation)
Benchmark Model suggests a “shallow” focus on liberalizing tf and tf
might have been natural for GATS
but as with GATT, an e¢ cient agreement would need additional rules
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
16 / 28
A Benchmark Trade-in-Services Agreement
To see this, suppose agreement binds t̄f = 0 and
t̄f = (γ
1) SfE , leaving all other policies unconstrained
∂θ (r R )
∂r
∂κ (r R )
∂r
=
∂θ (ρR )
∂ρ
∂κ (ρR )
∂ρ
=
∂λ(IdR ) R
Sd
∂Id
∂λ(IfR ) R
Sf
∂If
c
=
c
=
tR
=
1
[S R
2SdR f
1
[S R
2SfR f
( θ (r R )
θ (ρR ))]
( θ (r R )
∂κ (r R )
∂r
θ (ρR ))]
∂κ (ρR )
∂ρ
∂λ(IdR )
∂Id
∂λ(IfR )
1 R
[ Sf
(θ (r R ) θ (ρR ))]
2
∂If
1
1
[θ (r R ) + θ (ρR )] + SfR
2
2
1 R
[S
2 f
( θ (r R )
θ (ρR ))]
=) r R < r E < ρR , IdR too big, IfR too small, t R too high
Now consider adding some additional across-the-board rules
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
17 / 28
A Benchmark Trade-in-Services Agreement
First, a national treatment (NT) rule applied to regulation – but not
taxation – implying the restriction r
ρ
Suppose agreement binds t̄f = 0 and t̄f = (γ
1) SfE , leaving all
other policies unconstrained beyond NT
∂θ (r R )
∂r
∂κ (r R )
∂r
∂λ(IdR ) R
Sd
∂Id
∂λ(IfR ) R
Sf
∂If
c
c
tR
= 0;
∂θ (ρR )
∂ρ
∂κ (ρR )
=0
∂ρ
1 R ∂λ(IdR )
S
2 f ∂Id
1 R ∂λ(IfR )
=
S
2 f ∂If
1
= θ (r R ) + SfR
2
=
=) NT su¢ cient to prevent distortions of regulatory standards,
independent of foreign service provider market share
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
18 / 28
A Benchmark Trade-in-Services Agreement
Next, a Technical Barriers to Trade (TBT) rule: govs are obligated to
adopt regulations that are no more trade restrictive than necessary to
achieve their objectives
=)
∂λ(IdR ) R
∂ I d Sd
= c and
∂λ(IfR ) R
∂ I f Sf
=c
Suppose agreement binds t̄f = 0 and t̄f = (γ
1) SfE , leaving all
other policies unconstrained beyond NT and TBT rules
∂θ (r R )
∂r
∂κ (r R )
∂r
∂λ(IdR ) R
Sd
∂Id
= 0;
∂θ (ρR )
∂ρ
∂κ (ρR )
=0
∂ρ
∂λ(IfR ) R
Sf
c =0
∂If
1
t R = θ (r R ) + SfR
2
=) NT and TBT su¢ cient to prevent distortions of regulatory
standards and investments in reducing costs of compliance and
conformity assessment
Staiger and Sykes (Dartmouth and Stanford)
c
= 0;
Trade-in-Services Agreements
December 2016
19 / 28
A Benchmark Trade-in-Services Agreement
Finally, a non-violation (NV) rule whose primary purpose is to
dissuade govs from introducing new commercial measures subsequent
to negotiations that undercut market access commitments
Begin at Nash policies where tfN + tf N =
remaining policy choices satisfy
∂θ (r N )
∂r
∂κ (r N )
∂r
∂λ(IdN ) N
Sd
∂Id
= 0;
tN
= θ (r N )
1) SfN + 32 SfN and
∂θ (ρN )
∂ρ
= 0;
c
(γ
∂λ(IfN ) N
Sf
∂If
∂κ (ρN )
=0
∂ρ
c =0
Suppose agreement binds t̄f = 0 and t̄f = (γ
1) SfE , leaving all
other policies unconstrained beyond NT and TBT rules
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
20 / 28
A Benchmark Trade-in-Services Agreement
Remaining policy choices satisfy
∂θ (r R )
∂r
∂κ (r R )
∂r
∂λ(IdR ) R
Sd
∂Id
c
tR
= 0;
∂θ (ρR )
∂ρ
∂λ(IfR ) R
Sf
∂If
1
= θ (r R ) + SfR
2
= 0;
∂κ (ρR )
=0
∂ρ
c =0
NV could prevent change from t N = θ (r N ) to t R = θ (r R ) + 12 SfR ,
and allow the agreement to achieve e¢ cient policies
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
21 / 28
A Benchmark Trade-in-Services Agreement
) NT, TBT and NV work in tandem to facilitate shallow integration
based on negotiated market access commitments over tf and tf
NT addresses incentives to distort regulatory standards r and ρ that
arise once market access commitments over tf and tf are made
TBT addresses incentives to distort compliance-cost-reducing
investments Id and If that arise once market access commitments over
tf and tf are made
NV prevents the introduction of new “commercial measures”/…scal
instruments (t) from frustrating these market access commitments
(plus secondary role wrt changes in r and ρ)
And with only tf and tf distorted in Nash, Benchmark Model
suggests that a GATT-like shallow integration approach to services
trade would have been very natural for govs to pursue
Why is GATS so di¤erent?
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
22 / 28
The Implications of Limited Service-Sector Policy Options
Discriminatory domestic sales tax instrument tf
with goods trade, absent NT a discriminatory sales tax could be
imposed at the border (tari¤ by another name)
for Mode 3 services trade, a discriminatory sales tax must be imposed
at the point of production/consumption of the service
Limited evidence that imposing higher taxes on foreigners can in some
circumstances be feasible in service sector (Hendrix & Zodrow, 2003):
“Almost all states tax rentals of tangible personal property...re‡ecting
the popularity of taxes than may be exported to nonresidents...”
But for the most part, such taxes probably best thought of as
unavailable (perhaps for reasons of high transaction costs)
introduce this policy constraint into the Benchmark Model
tf
Staiger and Sykes (Dartmouth and Stanford)
0
Trade-in-Services Agreements
(Assumption 1)
December 2016
23 / 28
The Implications of Limited Service-Sector Policy Options
Note: tf = 0 and tf =
But Nash now
1) SfE still consistent with e¢ ciency
(γ
∂θ (r N )
∂r
∂κ (r N )
∂r
=
∂θ (ρN )
∂ρ
∂κ (ρN )
∂ρ
=
∂λ(IdN ) N
Sd
∂Id
∂λ(IfN ) N
Sf
∂If
Staiger and Sykes (Dartmouth and Stanford)
c
=
c
=
tN
=
tf N
=
1
[S N
2SdN f
1
[S N
2SfN f
( θ (r N )
θ (ρN ))]
( θ (r N )
∂κ (r N )
∂r
θ (ρN ))]
∂κ (ρN )
∂ρ
∂λ(IdN )
∂Id
∂λ(IfN )
1 N
[ Sf
(θ (r N ) θ (ρN ))]
2
∂If
1
1
[θ (r N ) + θ (ρN )] + SfN
2
2
1 N
N
(γ
1 ) Sf + S f
2
1 N
[S
2 f
Trade-in-Services Agreements
( θ (r N )
θ (ρN ))]
December 2016
24 / 28
The Implications of Limited Service-Sector Policy Options
Distortions widespread in Nash, shallow approach to service trade
liberalization no longer obvious
But still possible to avoid direct negotiations over standards, if
across-the-board NT, TBT and NV rules introduced …rst
Remaining policy choices satisfy
∂θ (r R )
∂r
∂κ (r R )
∂r
∂λ(IdR ) R
Sd
∂Id
c
∂θ (ρR )
∂ρ
= 0;
= 0;
∂λ(IfR ) R
Sf
∂If
1
t R = θ (r R ) + SfR ; tf N = (γ
2
E
Bind t̄ = θ (r ) and t̄f = (γ
1) SfE
∂κ (ρR )
=0
∂ρ
c =0
1
1)SfN + SfN
2
NV prevents subsequent changes in r and ρ from undercutting market
access commitments
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
25 / 28
The Implications of Limited Service-Sector Policy Options
Discriminatory foreign sales tax/subsidy instrument tf
foreign gov must be able to administer program of sales tax/subsidies
to its service …rms within jurisdiction of domestic gov
Perhaps even less reason to think this policy instrument is available
introduce this policy constraint into the Benchmark Model
tf
0
(Assumption 2)
For simplicity, relax Assumption 1 so that original e¢ ciency frontier
still attainable with tf = (γ
1) SfE (and tf
0)
Nash choices for domestic gov unchanged by Assumption 2, so still
possible to avoid direct negotiations over standards
bind t̄f =
(γ
1) SfE (and tf
0) and add NT/TBT/NV
Note: critical role for market power as source of international
ine¢ ciency is diminished under Assumption 2
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
26 / 28
Discussion
Can sales taxes be as …nely tuned to individual service industries as
regulatory standards?
if not, the capacity of the NT rule applied to standards for channeling
distortions into nondiscriminatory sales taxes is quali…ed
Can the concept of “like product” central to the NT rule be reliably
applied as a legal matter in the service sector?
if not, the utility of a shallow integration approach for services trade
will be undermined
Can govs measure and monitor with reasonable accuracy the changes
in import volumes and prices that would be required for the reliable
application of the NV rule in the service sector?
the fragmentary data on Mode 3 service trade currently available could
pose a roadblock to shallow integration for services trade
Are world prices determined by bargaining between Mode 3 service
providers and purchasers rather than market clearing conditions?
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
27 / 28
Conclusion
There are striking di¤erences between GATT and GATS
We show that the broad structure of GATS can be understood
through the lens of the ToT theory
but only if this theory is augmented with a set of restrictions, re‡ecting
salient features of services trade, on the policy instruments and trade
data available to govs
The GATT/WTO has been highly successful in liberalizing goods
trade; GATS has been largely unsuccessful in liberalizing services trade
A potential reason: the distinct nature of integration that each
agreement has attempted
We …nd that a shallow-integration approach more in line with that
taken by GATT might be possible in a trade-in-services agreement
thereby sidestepping some of the most contentious issues that may
have stymied negotiation progress thus far
Staiger and Sykes (Dartmouth and Stanford)
Trade-in-Services Agreements
December 2016
28 / 28