Trade and industrial policy
Table of Contents
1 General principles ............................................................................................................................. 2
1.1 The evolution of the multilateral trading system ....................................................................... 2
1.2 Most-favored nation and national treatment principles ............................................................. 2
Non discrimination....................................................................................................................... 2
Minimisation of distorsions (disciplines on specific instruments) .............................................. 3
2. The instruments of trade policy ....................................................................................................... 4
2.1 Tariffs and non-tariff measures .................................................................................................. 4
2.2 Why are quantitative restrictions (QR) bad ? ............................................................................ 4
2.3 Trade-defense instruments and anti-dumping regulations ......................................................... 6
2.4 Measuring the restrictiveness of trade policy ............................................................................ 7
Tarifs moyens............................................................................. Error! Bookmark not defined.
Le TTRI et l’OTRI ....................................................................................................................... 8
Mesurer la restrictivité des mesures non tarifaires ...................................................................... 9
3. La politique industrielle ................................................................................................................. 14
4. Le régionalisme .............................................................................................................................. 18
4.1 Les déterminants du commerce bilatéral ................................................................................. 18
4.2 Les accords régionaux.............................................................................................................. 20
4.3 Création et détournement de commerce .................................................................................. 22
4.4 Gains non traditionnels ............................................................................................................ 26
Références .......................................................................................................................................... 27
1
1 General principles
1.1 The evolution of the multilateral trading system
The WTO is the guardian of the treaties that form the basis of the multilateral trading system
(MTS). Its growing importance relates to its capacity to settle disputes between sovereign states and
to its sanction system. A few things to remember:
o
o
o
o
From Bretton Woods to the WTO
Negotiation « rounds »
« Mission creep » : from tariffs to NTBs and regulations
Dispute settlement: From the first panels to the Appellate body
1.2 Most-favored nation and national treatment principles
The MTS’s overarching principles are non discrimination and the minimization of policy-induced
distortions.
Non discrimination
Article I of the GATT (most-favored nation principle) states that all partners should be treated alike:
PART I
Article I
General Most-Favoured-Nation Treatment
1. With respect to customs duties and charges of any kind imposed on or in connection with
importation or exportation or imposed on the international transfer of payments for imports or
exports, and with respect to the method of levying such duties and charges, and with respect to all
rules and formalities in connection with importation and exportation, and with respect to all matters
referred to in paragraphs 2 and 4 of Article III,* any advantage, favour, privilege or immunity
granted by any contracting party to any product originating in or destined for any other country shall
be accorded immediately and unconditionally to the like product originating in or destined for the
territories of all other contracting parties.
Article III (national treatment) states that “like products” should be treated alike (whether imported
or produced domestically)
PART II
Article III*
National Treatment on Internal Taxation and Regulation
1. The contracting parties recognize that internal taxes and other internal charges, and laws,
regulations and requirements affecting the internal sale, offering for sale, purchase, transportation,
2
distribution or use of products, and internal quantitative regulations requiring the mixture, processing
or use of products in specified amounts or proportions, should not be applied to imported or
domestic products so as to afford protection to domestic production.
2. The products of the territory of any contracting party imported into the territory of any other
contracting party shall not be subject, directly or indirectly, to internal taxes or other internal charges
of any kind in excess of those applied, directly or indirectly, to like domestic products. Moreover,
no contracting party shall otherwise apply internal taxes or other internal charges to imported or
domestic products in a manner contrary to the principles set forth in paragraph 1.
However there are exceptions. The architecture of WTO agreements with respect to national
treatment and acceptable exceptions is shown in Figure 1:
Figure 1: The architecture of WTO agreements on national treatment
Article III (national
treatment)
Overarching
principle
Article XX (exceptions)
SPS
agreement
TBT
agreement
Licensing
agreement
Trade-facilitation
agreement
General
exceptions
Disciplines on
specific
instruments
Procedures
Minimisation of distorsions (disciplines on specific instruments)
Regulations that can affect international trade are covered by the disciplines of the TBT (Technical
Barriers to Trade) and SPS (Sanitary and Phyto-Sanitary measures) agreements. The SPS
agreement imposes that measures be based on science :
5.2. In the assessment of risks, Members shall take into account available scientific evidence;
relevant processes and production methods; relevant inspection, sampling and testing methods;
prevalence of specific diseases or pests; existence of pest — or disease — free areas; relevant
ecological and environmental conditions; and quarantine or other treatment.
5.4. Members should, when determining the appropriate level of sanitary or phytosanitary protection,
take into account the objective of minimizing negative trade effects.
5.7. In cases where relevant scientific evidence is insufficient, a Member may provisionally adopt
sanitary or phytosanitary measures on the basis of available pertinent information, including that
from the relevant international organizations as well as from sanitary or phytosanitary measures
applied by other Members. In such circumstances, Members shall seek to obtain the additional
information necessary for a more objective assessment of risk and review the sanitary or
phytosanitary measure accordingly within a reasonable period of time.
3
2. The instruments of trade policy
2.1 Tariffs and non-tariff measures
The rise in importance of NTMs reflects partly the proliferation of regulation and partly the fact that
tariffs (the other major trade-policy instrument) have come down substantially over the last thirty
years, as shown in Figure 2.
Figure 2: Average tariffs at the country level against GDP/capita, 1980-2010
30
20
0
10
Tariffs(%)
40
50
MFN Applied Tariff Rates versus GDP per Capita
0
2000
4000
GDP per capita(US dollars)
1980s
2000s
6000
8000
1990s
Source : World Bank, 2011
The term « Non-tariff measures » (NTMs) covers all regulations other than tariffs that can affect
international trade :
o
o
o
o
o
Taxes
Quantitative restrictions (prohibitions, quotas)
Technical and SPS regulations
Anti-dumping regulations
Restrictive commercial practices
2.2 Why are quantitative restrictions (QR) bad ?
Article XI of the GATT specifically prohibits QRs:
Article XI*
General Elimination of Quantitative Restrictions
1.No prohibitions or restrictions other than duties, taxes or other charges, whether made effective
through quotas, import or export licences or other measures, shall be instituted or maintained by any
contracting party on the importation of any product of the territory of any other contracting party or
on the exportation or sale for export of any product destined for the territory of any other contracting
party.
4
Why is this? Under perfect competition at home, there is a basic tariff-quota equivalence theorem:
For every tariff, one can find a quota that has exactly the same effect on welfare.
Figure 3: Tariff-quota equivalence under perfect competition
(a) Tarif
(b) Quota
Prices
Prices
Domestic demand
Domestic demand
Residual (after-quota) demand
Marginal cost
Marginal cost
World price plus tariff
Tariff
A
B
C
D
Domestic price
A
World price
B
C
Quantities
D
World price
Quantities
Quantity imported
Quota
However, when the assumption of perfect competition on the home market is relaxed, a quota is
much worse than a tariff. To see this, let us consider the case of a monopoly on the home
(importing) market. The way we will proceed is in four steps
1.
2.
3.
4.
Monopoly in a closed economy
Monopoly under free trade
Monopoly with a tariff
Monopoly with a quota.
The argument is essentially this: In the closed economy, the home firm in a monopoly situation
holds back output in order to raise price (standard story). Free trade strips the home firm of its
monopoly power. Imposing a tariff allows the home firm to raise its price somehow, but not all the
way back to its closed-economy level. That is, a tariff does not hand back monopoly power to the
home firm. By contrast, a quota hands it back monopoly power on a silver plate. As a result, the
home firm again holds back output (that is, does not create jobs) and jacks up prices, the worst of
both worlds (high prices but no jobs!)
Figure 4: Non-equivalence in the presence of a monopoly on the importer market
(a) Tariff
(b) Quota
5
Prices
Prices
Domestic demand
Domestic demand
Residual (after-quota) domestic demand
Marginal revenue (irrelevant)
Marginal cost
Domestic price
Marginal cost
World price plus tariff
Tariff
A
B
C
D
C
A
D
World price
World price
Quantities
Quantities
Quota
Quantity imported
B
Upshot of the discussion : When there is market power in the importing country (not necessarily a
pure monopoly, but anything that deviates from competition), a quota is the worst way of protecting
the home producer from foreign competition.
2.3 Trade-defense instruments and anti-dumping regulations
Anti-dumping (AD) regulations allow a country’s producers to complaint to their government about
unfair pricing practices by their foreign competitors. The WTO AD agreement allows the importing
government to conduct an investigation in order to determine
1. Whether there was dumping in the form of either pricing below cost or pricing exports below
the domestic price (see below)
2. Whether there was injury in the importing country (for its domestic producers)
3. Whether the dumping caused the injury.
If the answer is yes to all three questions, the government is entitled to impose AD duties on all
firms in the exporting country (not just those that actually dumped). After five years, the duties
must be re-examined and can be extended only upon justification
In reality, AD regulations are a mess. Dumping is defined by the AD agreement as follows:
Article 2
Determination of Dumping
2.1 For the purpose of this Agreement, a product is to be considered as being dumped, i.e.
introduced into the commerce of another country at less than its normal value, if the export price of
the product exported from one country to another is less than the comparable price, in the ordinary
course of trade, for the like product when destined for consumption in the exporting country.
2.2 When there are no sales of the like product in the ordinary course of trade in the domestic market
of the exporting country or when, because of the particular market situation or the low volume of the
6
sales in the domestic market of the exporting country, such sales do not permit a proper comparison,
the margin of dumping shall be determined by comparison with a comparable price of the like
product when exported to an appropriate third country, provided that this price is representative, or
with the cost of production in the country of origin plus a reasonable amount for administrative,
selling and general costs and for profits.
Thus, standard business practices, like introducing new products on export markets at a discount,
can easily qualify as dumping. Moreover, the calculation of dumping margins is fraught with weird
practices. For instance, « zeroing » is a cute way of biasing the result. Here is an example:
Exporter’s domestic market
(Turkey)
Exporter’s foreign market
(EU)
Prices
Prices
Dumping
margins
Transaction 1
120
n/a
Transaction 2
95
5
Transaction 3
85
15
______________________________
Average dumping margin
10
Transaction 1
120
Transaction 2
95
Transaction 3
85
__________________
Normal value
100
In this case, under zeroing the firm is dumping even though it charges the exact same prices on its
domestic and export markets.
2.4 Measuring the restrictiveness of trade policy
Measuring the overall restrictiveness of a trade-policy stance involves two challenges :
o Combining the effects of several types of measures
o Aggregating their effects on all goods.
This is however very important, because assessments of how restrictive is a country’s policy enter
scoring systems used by donors to determine their aid commitments.
Average tariffs
Overall tariff restrictiveness can be measured by either simple or import-weighted average tariffs.
Unfortunately both are biased:
o The simple average over-weighs small items
o The weighted average under-weighs items with the most restrictive tariffs
In addition, neither takes into account the effect of tariff variability, which is a factor of inefficiency
in itself from a welfare perspective. To see this, consider the following approximation of the
deadweight loss:
7
The TTRI and OTRI
The tarif trade restrictiveness index (TTRI) is the rate of a (fictitious) uniform tariff that would give
the same welfare level (i.e. that would generate the same deadweight loss) as the current array of
tariffs at different rates. It is calculated as follows:
The Overall Trade Restrictiveness Index (OTRI) is the rate of a (fictitious) uniform tariff that would
give the same aggregate import level (all goods together) as the current array of tariffs at different
rates:
8
The MA-OTRI is an export-weighted average of the OTRIs a country faces on its destination
markets :
It is an inverse measure of a country’s market access; i.e. the higher it is, the more protection a
country faces on its export markets.
Figure 5 shows that both OTRI and MA-OTRI correlate negatively with income levels ; thus
o Poor countries are more protectionist
o They face higher barriers on their export markets.
Figure 5: Correlation of OTRI and MA-OTRI with income level
Source: Kee, Nicita and Olarreaga 2009
However, this is partly a statistical illusion created by the way the OTRI and MA-OTRI are
constructed. A country can have a high OTRI and yet grant deep tariff preferences to LDCs. For
instance, most industrial countries grant tariff-free, quota-free access to products from sub-Saharan
Africa. Thus, E.U. and U.S. MFN tariffs do not concern Africa, even though they enter in the
calculation of the E.U. and U.S. OTRI. When calculating the MA-OTRI of an African country, the
use of E.U. and U.S. OTRIs is thus misleading.
Measuring the restrictiveness of non-tariff measures
One can measure the restrictiveness of non-tariff measures (NTMs) in an « artisanal » way by
calculating the price gap, i.e. the difference of a given good’s price in an NTM-ridden market vs. a
market where no NTM is imposed. For instance, Table 1 shows a price gap calculation for goods
9
affected by import prohibitions in Nigeria. The market affected by the NTM (the import
prohibition) is Lagos, Nigeria’s port city, and the comparator market is Nairobi. Cotonou, Benin’s
capital, would have been a much better comparator, but there is no price data for Cotonou.
Table 1: Price gap calculation for Nigerian import bans
Table 1
Price-gap calculations, Lagos vs. Nairobi (percent)
Staples
Protein
Beverages
Household
supplies
178
61
30
-24
-7
-26
67
-12
Banned products
Other
Personal
care
products
194
-17
Total
92
15
Source : EIU, PRMTR calculations
On average, goods not affected by import prohibitions are 15% more expensive in Lagos than
Nairobi (second line of the table). However, goods affected by prohibitions are 92% more
expensive. Thus, the price gap is 92% - 15% = 77%. That is, import prohibitions raise the price of
affected products by 77%.
This price increase can be fed into household survey data in order to assess the effect on the cost of
living at each level of income. Formally, let k be a good and pk the estimated price gap:
pk  pk  pk*
(1)
Let i be a household and ik the weights of producst k = 1,…, K that this household consumes. The
change in real income (here the increase in the cost of living) generated by the NTM for household i
is :
I i   k ik pk
(2)
In the next step, we calculate the average I i by income bracket; say for each income quartile
(bottom 25%, next 25%, and so on). This gives a picture like Figure 6, which shows a regressive
effect (a stronger increase in the cost of living, i.e. a deeper cut in real income, for the bottom
quartile of the income distribution).
Figure 6: Effect of import bans by income quartile
10.0%
9.0%
8.0%
7.0%
6.0%
5.0%
4.0%
3.0%
2.0%
1.0%
0.0%
Total
1st Quartile
2nd Quartile
10
3rd Quartile
4th Quartile
Source : Treichel et al. 2012
Figure 7 shows the effect of a cash grant equivalent fully compensating the prohibitions’ effect on
real income on the distribution of income. The effect looks small on the picture, but given the size
of Nigeria’s population, is sufficient to lift three million out of poverty (out of a population of 140
million people).
0
.00001
.00002
.00003
Figure 7: Actual and simulated income distribution
0
20000
40000
60000
80000
100000
x
Baseline
Simulated
Note : Actual income distribution is while import prohibitions are in force. Simulated distribution is as if each
household received a cash grant equivalent to the real-income loss generated by the prohibitions; so the first quartile
receives a cash grant of 9.1%, the next one 8.9%, and so on.
More generally, one can analyze the effect of NTMs by using the variation of prices (or of
quantities traded) as the identification mechanism. Using prices, the estimation equation is as
follows. Let pij be the price of a good exported from country i to country j and t j the tariff imposed
by j. Let n be a type of NTM (n = A for SPS measures, n = B for a TBT measure, etc.). Let x ij be a
vector of characteristics of country pair (i,j) including, inter alia, their distance. Let
1 if country j imposes measure n on the product in question
otherwise
0
 jn  
And  i and  j country fixed effects by exporter and importer. The estimation equation is
ln pij   ln 1  t j    n  n jn 
NTMs

j
n
 n j jn
NTMs interagies avec importateur
 xij γ   i   j  uij
bilateral
The ad-valorem equivalent (AVE) of NTM n is1
an  e
n jn  n j jn
 1.
This expression is obtained by exponentiating the equation above, evaluating it at 
 1 and at   0
jn
jn
respectively, dividing the first by the second and subtracting one to get the rate of change of the price. Try it!
1
11
An example of regression results is shown in Table 2.
Table 2: Estimated price effect (AVE) of NTMs in sub-Saharan Africa
Estimator
ln (price
gap)
(1)
Dep. Var.:
Within-product
Percent
Tariff-corrected
price gap
price gap
(2)
(3)
(4)
-0.111
-11.495
-7.999
(0.99)
(1.31)
(0.60)
0.002
0.210
(0.14)
(0.17)
1.272
1.073
(15.54)***(13.48)***
SSA dummy
Ln (1 + tariff)
COL adjustment
Frequency ratios
SPS
NTM A × SSA
TBT
NTM B × SSA
PSI & Formalities
NTM C × SSA
Price measures
NTM D × SSA
QRs
NTM E × SSA
Observations
Number of products
Fixed effects
Products
Countries
R-squared (within)
0.113
(2.53)**
-0.021
(0.37)
0.018
(0.29)
-0.028
(0.29)
0.060
(0.43)
8.619
(2.22)**
-0.623
(0.13)
0.466
(0.09)
-2.304
(0.24)
7.061
(0.53)
14.002
(2.98)***
-6.934
(1.24)
1.963
(0.38)
4.128
(0.37)
17.794
(1.17)
12.951
(2.16)**
2.596
(0.50)
6.989
(0.97)
-1.393
(0.12)
17.937
(1.01)
1'260
42
1'260
42
1'260
42
1'260
42
yes
no
yes
no
yes
no
yes
yes
0.34
0.31
0.09
0.12
However, restrictiveness is not the whole story : NTMs address a need for regulatory intervention in
order to correct market failures.
Figure 8: Effect of a regulation on consumption with a negative externality
(a) Avant règlementation
(b) Après règlementation
12
Prices & monetary
valuations
Prices & monetary
valuations
Domestic demand
Domestic demand
Consumer surplus
NTM compliance cost
Foreign supply
Foreign supply
Consumer surplus
reduced
Quantities
consumed
Negative
externality
reduced (directly
and indirectly)
Negative
externality
One should therefore strive to improve them through a logical process rather than eliminate them
(Figure 9) :
Figure 9: Logical workflow of an NTM review
PRIVATE-SECTOR INTERVIEW
TRIGGER QUESTIONNAIRE
Issue is substantial?
No
Yes
WTO tests
o
Drop case
REGULATOR INTERVIEW
FACT-FINDING QUESTIONNAIRE
Necessity
Market failure?
o Propose regulation
phase-out
No
Yes
Regulation correctly targets
market failure?
Proportionality
No
Yes
o Propose regulation
redesign
Regulatory benefits clearly
outweigh its costs?
Yes
Tradefacilitation team
o Debate reglation
justification
No
o Regulation should be kept
o Look for ways to redesign it in a
cost-minimizing way
13
3. Industrial policy
Many governments regularly sollicit advice on how to put in place an industrial policy. How should
one justify using taxpayer resources to affect the allocation of resources, given the basic premise
that the “invisible hand” does it right? Clearly, like in the case of regulatory interventions in the
previous section, the justification must be grounded in the identification of a market failure. The
nature of the market failure, however, is likely to be different, having more to do with economies of
scale, learning etc. (Figure 10).
Figure 10: Non-concave production possibility frontier with economies of scale
Sector 1 (traditional)
Home relative producer prices at E1 (sector 2 inefficient)
International relative prices
E1
PPF
E2
Sector 2,
subject to
IRS (to be
targeted by
IP)
Region where sector 2
dispays increasing
returns to scale
Most governments do it through special economic zones (SEZs) designed to attract FDI (there are
over 130 in the world). SEZs typically offer « tax holidays » by which is meant a grace period
(usually ten years) during which companies established in the SEZ do not pay any profit or
corporate income tax. They also typically have exemptions from duties on their intermediate inputs,
have access to simplified administrative procedures, and a special, dedicated infrastructure (energy
at less than market rates, special transportation infrastructure like port installations, dedicated
telecom infrastructure, and so on).
What do governments get out of these costly concessions? 2 They expect to get “good” (highpaying) jobs, export revenue, a better visibility with foreign investors, and a “growth engine” with
backward linkages in the local economy. However, apart from big successes (Mauritius, China’s
Shenzhen, or Masan in Korea) the evidence so far is not terribly encouraging
2
The tax concessions and dedicated infrastructure can be very costly. Some countries forsake close to half their
potential fiscal revenue, which cripples their ability to provide public goods (health, education) to the population at
large. Electricity is sometimes sold at half price to large multinational companies which therefore waste it, while the
population at large is denied electricity access for very basic needs and hospitals put up with multiple power outages
each day.
14
For instance, backward linkages are, in many cases, virtually nonexistent (Figure 11).
Figure 11: Backward linkages
45
Percent sourced locally
40
35
30
25
20
15
10
5
0
Source : World Bank 2013
Salaries do not seem much higher than those offered in the domestic economy (Figure 12)
Figure 12: SEZ wage premium relative to average manufacturing wage
350
300
250
200
150
100
50
0
-50
Source : World Bank 2013
And the jobs do not seem to leverage much the local skills (Figure 13).
15
Figure 13: Percentage of SEZ employees trained in local vocational schools
16.0
14.0
12.0
Percent of workers trained in local vocational training
programs
10.0
8.0
6.0
4.0
2.0
-
Source : World Bank 2013
Worse, the tax incentives have been driving down average tax rates in developing countries, in
particular in sub-Saharan Africa, where they are heading toward zero (Figure 14).
Figure 14: Average corporate income tax rates in the world
Source: Abbas, Ali; A. Klemm, S. Bedi and J. Park (2012), “A Partial Race to the Bottom: Corporate tax Developments
in Emerging and Developing Countries; IMF working paper WP/12/28
This is particularly bad given that FDI seems to be only weakly sensitive to tax incentives. There
are spectacular examples of reactivity like when Antigua, a Caribbean island, eliminated the tax on
hotels (Figure 15), but this was largely a « beggar-thy-neighbor » policy as it diverted investment
away from other Caribeean islands.
16
Figure 15: Tourism investiment in Antigua vs. Windward Islands
In general, tax incentives do not compensate for poor business environments (Africa’s problem).
Figure 16 distinguishes between recipient countries with poor investment climates (IC), with the
orange regression line, and countries with good investment climates, those with the dark blue line.
The first line is almost flat and indeed its slope is not significantly different from zero.
Figure 16: FDI and the marginal tax rate on corporations
Source: James, Sebastian, 2007, “The Effect of Tax Rates on Declared Income: An Analysis of Indian Taxpayer
Response to Changes in Income Tax Rates,” Ph.D. diss., Harvard University, quoted in James (2009)
Figure 16’s results are confirmed by a panel regression on 47 countries (Table 3: FDI and effective
tax rates: Panel regression results).
Table 3: FDI and effective tax rates: Panel regression results
17
(lagged)
Source: Abbas et al. (2011)
Notes: METR: Marginal effective tax rate; EATR: Average effective tax rate. EATR for special regime: régimes
miniers ou zones franches.
Thus, while the idea that an SEZ could provided a sheltered environment in which the country’s
overall governance problems (poor infrastructure, inefficient and corrupt administration, heavy
bureaucratic procedures, uncertain investment climate) could be overcome on a limited scale, while
appealing, does not seem to work out in practice. If the country is a mess, so will be the SEZ. The
only argument that seems to have some reality is that for a government that wants to implement
reforms (e.g. lightening labor regulations when they are too heavy, liberalizing imports of
intermediates, etc.) but not attack lobbies frontally, an SEZ can provide a way to do it in a limited
way and avoid a big political battle. This is what happened in Mauritius and in China.
4. Regionalism
4.1 Determinants of bilateral trade : The gravity equation
The gravity equation is the basic workhorse in terms of empirical trade analysis. The analogy with
gravity comes from the fact that gravity can be expressed as F  gm1m2 / r 2 where F is “the Force”,
g is the gravitational constant, m1 and m2 are masses of the two bodies, and r is their bilateral
distance . Economic gravity is very similar: Trade between two countries depends on the product of
18
their GDPs (their economic mass) divided by the cost of trade, itself roughly proportional to
distance (inter alia). The key difference is that in economics the equivalent of g is not a constant.
The gravity equation can be derived in a simple framework where consumer preferences are CES
(constant elasticity of substitution) in each country. 3 Let i be the exporting country and j the
importing one. Consider a variety produced in country i. Let xij be the quantity exported from i to j
(in tons), pij the CIF price of variety i, and E j total expenditure in country j. The share of the variety
exported by i in the expenditure of country j is sij , so we can write
pij xij  sij E j
(3)
Because of CES preferences, it can be shown (I spare you the proof) that if Pj is a composite price
index for country j and  is the elasticity of substitution between varieties,
1
p 
sij   ij 
P 
 j
(4)
The variable (ad valorem) trade cost is  ij and the producer price in i is pi . Suppose that this
producer price is the same irrespective of the destination of exports (this is actually a property of
the model, not an assumption; it is called « mill pricing »). Then we have
pij   ij pi
(5)
Let Vij be the total value of export from i to j, and suppose that i exports ni varieties. Bilateral trade
between i and j, all varieties together, is thus
1
p 
Vij  ni pij xij  ni sij E j  ni  ij 
P 
 j
Ej
1
 p 
 ni  ij i 
 P 
 j 
(6)
Ej
Country i’s GDP is the sum of all that it sells or exports, including to itself:
1
 p 
Yi  Vij   ni  ij i 
 P 
j 1
j 1
 j 
m
m
Ej
(7)
Let us now invert this to solve as a function of ni pi1 :
ni pi1 
Yi
i
where
3
Le traitement dans cette section suit Baldwin et Taglioni (2006).
19
(8)
m
 E
i   ij1  1j
P
j 1
 j



(9)
where the term in the summation is a weighted average of the trade costs i faces. Substituting (9)
into (6) gives
 YE
Vij   ij1  i 1j
 P
 i j



(10)
Finally, noting that E j  Y j (income equals expenditure) we get an expression that looks very much
like gravity :
Vij   ij1
1
YY
i j
i Pj1
(11)
or, in logs and adding control variables and an error term,
lnVij  1
ln ij
coût du commerce
  2 ln Yi  3 ln Y j  xi γ1  x j γ 2  uij
PIB export.
PIB import.
(12)
controles
With importer and exporter fixed effects, the equation drastically simplifies to
ln Vij  1
ln  ij

coût du commerce
i   j
 uij
(13)
EF export. & import.
When the gravity equation is estimated on a panel of countries (i.e when several years of data are
available), the term 1 /  i Pj1  —equivalent to the gravitational constant, called in a trade context
the « multilateral resistance term » and measuring a country’s difficulty to trade with all its trade
partners—varies over time, so the fixed effects in equation (12) should also be made timedependent (i.e. they should be country×year fixed effects). If the bilateral trade cost  ij also varies
over time (e.g. with investments in transport infrastructure), the correct estimation equation is:
ln Vijt  1 ln  ijt   it   jt  uijt
(14)
Typically, the bilateral trade cost is itself expressed in terms of fundamentals including the possible
presence of a common land border, a common language or religion, a common colonial past, and so
on.
The gravity equation predicts bilateral trade flows in a very reliable way and is used both in order to
predict them and to identify the gap between predicted and actual trade flows (« under-trading » ).
It can also be used to assess the effect of various trade policy instruments, in particular regional
trade agreements, to which we now turn.
4.2 Regional trade agreements
« Regionalism » can facilitate trade negotiations within a club of neighboring countries that share
other policy objectives (like common security etc.) and have more to share and to gain from an
20
agreement than all countries together in a multilateral round. It thus offers a limited alternative to
multilateral (WTO) trade negotiations. It is also very much in fashion, as shown by the growth in
the number of trade agreements around the world (Figure 17).
Figure 17: Number of trade agreements in force
The result of this proliferation is to create very complicated preference structures (a « spaghetti
bowl » of agreements, as shown for Bangladesh in Figure 18).
Figure 18: Bangladesh’s preferential trade agreements
Germany
Poland
Ukraine
Hungary
Romania
Belarus
Former Yugoslavia
SAFTA
Uzbekistan
BIMSTEC
Former Czekoslovakia
Bulgaria
Albania
Nepal
Afghanistan
TPS-PRETAS
Uganda
Jordan
Cameroon Lebanon
Guinea
Syria
Tunisia
Lybia
UAE
Senegal
Algeria
Sudan
Iran
Philippines
Bhutan
Malaysia
Pakistan
Nigeria
Indonesia
Turkey
Egypt
Bangladesh
D-8
Lao PDR
Korea APTA
India
Kuweit
Sri Lanka
Zimbabwe
Kenya
Mali
Maldives
Brazil
Zambia
Source : World Bank (2012)
These agreements typically include
o Trade liberalization within the bloc
21
Myanmar
Cambodia
Thailand
Vietnam
o Cooperation in limited other areas (harmonization of regulations, security, management of
water resources, …)
It is customary to distinguish four broad type of agreements, in increasing order of deep
integration :
o
o
o
o
Free-trade agreements (FTAs ; example : NAFTA)
Customs unions (Mercosur, Russia-Kazakhstan, E.U.-Turkey)
Common markets (European Community before Maastricht Treaty)
Monetary unions (Eurozone, West Africa).
While preferential trade agreements are likely to generate more trade because they entail withinbloc trade liberalization, this trade expansion is not necessarily welfare enhancing. This is a
fundamental result derived by Jacob Viner in 1951, to which we now turn.
4.3 Trade creation and trade diversion
Trade creation in an FTA generates welfare gains, while trade diversion generates welfare losses. In
order to understand the mechanics, let us assume the following 3-country trade structure:
Figure 19: Basic trade structure for Vinerian effects
Home production
competes with
imports from B
and C
imports
Country A
Country B
FTA
imports
Country C
Table 4: Trade creation and trade diversion
Case 1: Neither creation nor diversion
Producers
Domestic price
Duty-paid price in A
With 20% MFN tariff
With tariff only on C
A
15.0
B
18.0
C
20.0
15.0
15.0
21.6
18.0
24.0
24.0
A
17.0
B
15.0
C
16.0
17.0
17.0
18.0
15.0
19.2
19.2
A
13.0
B
11.0
C
10.0
13.0
13.0
13.2
11.0
12.0
12.0
Case 2: Trade creation
Imports from partner displace
inefficient domestic production
Producers
Domestic price
Duty-paid price in A
With 20% MFN tariff
With tariff only on C
Case 3: Trade diversion
Imports from partner displace
efficient imports from rest of the
world
Producers
Domestic price
Duty-paid price in A
With 20% MFN tariff
With tariff only on C
22
Empirically, the presence of trade creation or diversion can be tested using the gravity equation
(which is why I introduced it in this chapter). The first step is to create two dummy variables
o The first is equal to one if both importer and exporter are in the preferential bloc, zero
otherwise;
o The second is equal to one if the importer is in the bloc but not the exporter.
If the first has a positive and significant coefficient, case 1 in Table 4 can be ruled out, but one
cannot discriminate between cases 2 and 3. If the second comes out with a negative (and
significant) coefficient, case 2 can be ruled out and we are in a case of trade diversion.
Table 5 gives an illustration using trade data from the East African Community (EAC), which was
an FTA until 2004 and a CU from 2005 on. The first dummy is positive and significant, so intrabloc trade was encouraged (see also Figure 20: EAC’s intra-bloc trade rose with out-of-bloc
exports), while the second is negative and significant, suggesting Case 3 (trade diversion).
Figure 20: EAC’s intra-bloc trade rose with out-of-bloc exports
Table 5: Trade creation and diversion in the EAC
23
Dep. Var.
Estimator
Exporter GDP
Importer GDP
Distance
Common border
Common language
Common colonial past
Exporter real ER
Importer real ER
Both in EAC, 2001-4
Both in EAC, 2005Importer in EAC, exporter out, 2001-4
Importer in EAC, exporter out, 2005Inverse Mills ratio
Constant
Observations
R-squared
Log
(trade)
Log
(trade)
OLS
(1)
Heckman PPML
(2)
(3)
0.865)***
(15.13)
1.176***
(23.96)
-1.582***
(92.09)
0.750***
(8.13)
0.793***
(22.00)
0.829***
(9.33)
0.000**
(2.26
-0.000***
(5.37)
1.356***
(3.78)
1.478***
(4.22)
-0.211***
(3.47)
-0.167**
(2.32)
0.72***
(0.07)
1.24***
(0.06)
-1.60***
(0.02)
0.75***
(0.10)
0.81***
(0.04)
0.78***
(0.09)
0.00
(0.00)
-0.00***
(0.00)
n.a.
Trade
Trade
ZIP
(4)
1.31***
(0.10)
1.29***
(0.10)
-0.86***
(0.03)
0.48***
(0.07)
0.31***
(0.06)
-0.03
(0.10)
-0.00
(0.00)
-0.00***
(0.00)
2.14***
(0.38)
n.a.
1.99***
(0.31)
-0.20*** -0.17***
(0.06)
(0.06)
-0.16** 0.00
(0.08)
(0.08)
0.03
(0.03)
11.200
11.41*** 6.15***
(37.70)***(0.35)
(0.65)
1.30***
(0.10)
1.28***
(0.10)
-0.87***
(0.03)
0.48***
(0.07)
0.31***
(0.06)
-0.03
(0.10)
-0.00**
(0.00)
0.00***
(0.00)
2.09***
(0.38)
1.94***
(0.32)
-0.19***
(0.07)
-0.01
(0.08)
336588
0.76
520,596
283,338
0.758
520,596
6.35***
(0.66)
If one wants to go beyond a mere gravity analysis, firm-level data suggests that firms in the region
are heavily specialized between those that sell on the regional market and those that sell outside
(Figure 21).
Figure 21: Distribution of regional shares in firm exports, average 2009-2011
(a) EAC
(b) EAC + DRC
24
40
50
40
Percent
30
30
0
0
10
10
20
20
Percent
0
.2
.4
.6
EAC share in firm exports
.8
1
0
.2
.4
.6
Region's share in firm exports
.8
1
Note: The horizontal axis measures the proportion of sales going to the EAC (panel a) or to the region inclusive of the
DRC (panel b), in “bins” of 5% each. The vertical axis measures the percentage of firms in each bin. That is, for
instance, almost 40% of Tanzania’s exporters were sending over 95% of their exports to the region on average over
2009-2011.
Source: Mission calculations using TRA data.
Moreover, products exported to regional markets are slightly more intensive in capital and human
capital (Figure 22).
Figure 22
Factor content of Tanzania’s regional and out-of-region exports, average 2009-2011
(b) Capital intensity
0
0
.1
Frequency
.2
Frequency
5.000e-06
.00001
.3
.000015
(a) Human-capital intensity
2
4
6
8
Human-capital intensity
Regional exports
10
12
0
Out-of-region exports
50000
100000
150000
200000
Capital intensity (dollars per worker)
Regional exports
250000
Out-of-region exports
Source: Mission calculations using TRA data and UNCTAD revealed factor intensity database.
All this suggests that the regional market acts like a protected market for « infant exporters ».
However, the learning does not seem to translate to a migration flow from the group of regional
exporters to the group of out-of-region exporters, as transitions are very small (Table 6)
25
Table 6: Annual transition probabilities between exporter categories
Non-OECD
In between
OECD
0.941
0.057
0.014
0.044
0.904
0.039
0.015
0.039
0.947
Non-OECD
In between
OECD
Source: Mission calculations using TRA data.
4.4 Non traditional gains
Beyond trade gains, regionalism encourages
o More predictability and stability of trade policies
o More independence from lobbies, especially in the presence of supranational institutions like
the EU Commission.
It also encourages cooperation in non-trade areas.
26
References (complete)
Abbas, Ali, A. Klemm, S. Bedi and J. Park (2012), “A Partial Race to the Bottom: Corporate Tax
Developments in Emerging and Developing Economies”; IMF working paper WP/12/28;
Washington, DC: International Monetary Fund.
Baldwin, Richard, and D. Taglioni (2006), ‘Gravity for Dummies and Dummies for Gravity; NBER
Working Paper 12516.
Carrère, Céline (2006), “Revisiting the effects of regional trade agreements on trade flows with
proper specification of the gravity model”; European Economic Review 50, 223-247.
Farole, Thomas (2011), Special Economic Zones in Africa: Comparing performance and learning
from experience; Washington, DC: The World Bank.
Kee, Hiau Looi, A. Nicita, and M. Olarreaga (2009), “Estimating Trade Restrictiveness Indices”;
Economic Journal 119, 172-199.
Klemm, Alexander, and S. Van Parys (2009), “Empirical Evidence on the Effect of Tax
Incentives”; IMF Working Paper 09/136; Washington, DC: IMF.
27