Transcript trade-diversion

Estimating the Effects of
Regional Integration
Nigel Grimwade (LSBU)
Regional economic integration
Definition: the process whereby separate national
economies are combined into larger economic regions
 Two basic conditions:1. Free movement of goods and factors of production
(labour and capital)
2. Absence of discrimination
 Brought about in 2 ways:1. Negative integration – removal of discrimination (e.g.
tariffs) and restrictions on movement
2. Positive integration – creation of common policies and
institutions

Forms of regional economic integration
Types of regional economic integration:1.
Preferential trading agreements (e.g. EU’s Lomé and
Cotonou Agreements)
2.
Free trade agreements – elimination of tariffs on all trade
between the member states (e.g. EFTA, NAFTA, AFTA)
3.
Customs unions – internal free trade plus a common
external tariff (e.g. EC/EU, Mercosur)
4.
Common Markets – free trade in goods and services and
free movement of capital and labour (e.g. SEM)
5.
Monetary unions – adoption of a common currency, single
central bank and common monetary policy (e.g. Euro zone)
Economic effects of regional integration
Static effects - short-run, once-and-for all
increase/decrease economic welfare due to:1. Reallocation of resources due to trade-creation and
trade-diversion
2. Terms of trade effects – change in the ratio of average
export prices to average import prices
 Dynamic effects – long run, on-going increase in national
income due to:1. Economies of scale (static and dynamic)
2. Increased competition
3. Increased investment (domestic and foreign)
4. More rapid rate of technological innovation

Problems in estimating the effects
of regional integration
Which effects to measure – effects on trade,
national income, rate of economic growth
 What time period to consider – short- or
long-run
 What types of integration to cover – tariffs,
NTBs, services, factor mobility, monetary
integration
 How to estimate the counterfactual – what
would have happened if integration had not
taken place

Simple extrapolation





Most early studies were concerned with the static effects –
trade-creation and trade-diversion
Focused on trade (exports and imports) between the
members (intra-area trade) and trade with non-members
(extra-area trade)
Made the simple assumption that intra- and extra-area
exports/imports would have continued growing at the same
rate had integration not happened (anti-monde)
Difference between actual trade and the anti-monde (the
residual) is the integration effect
One major problem – this assumes that conditions in the
post-integration period were the same as in the preintegration
Intra- and extra-area trade share





A better approach is to extrapolate intra-area and extraarea exports/imports as shares of total imports (see
diagram) e.g. Xij/Xi, Mij/Mi
Increase in actual intra-area export/import share
compared with the anti-monde is evidence for an
integration effect
But we have no way of knowing whether this is due to
trade-creation or trade-diversion
Is greatly affected by (a) the number of countries involved
and (b) the region’s share in total trade
Share may also be affected by an increase in the number
of members over time
Residual imputation
IntraRegional
Export
Share
%
0
Actual
Trend
T1
Time
Problems with simple residual models
Different factors may have affected intra- and extra area
exports/imports in the pre- and post-integration periods,
such as:1. Rate of economic growth – import demand
2. Phases of the business cycle
3. Changes in relative prices
4. Changes in the exchange rate
5. Structural changes
6. Reductions in multilaterally negotiated tariffs
 So, we cannot assume that what happened in the preintegration period would have happened in the postintegration period had integration not happened

Shares of intra- and extra-area trade in
national income/consumption





Extrapolate share of extra- and intra-area imports in
GNP/GDP or….
Extrapolate share of extra- and intra-area imports in
apparent consumption (domestic production less exports
plus imports)
An increase in share of AC coming from domestic
sources is evidence for gross trade-creation
An increase in share of AC coming from partner
countries is evidence for net trade-creation
A decrease in share of AC coming from rest of the world
is evidence for trade-diversion
Balassa’s income elasticity of demand for
imports approach





Income elasticity of demand for imports (∆M/∆Y) is
calculated for pre- and post-integration period
Rise in income elasticity of demand for imports for intraarea imports is defined as gross trade-creation
Increase in income elasticity of demand for imports from
member states is net trade-creation
Decrease in demand for imports from non-member
states is trade-diversion
Found evidence for net trade-creation following the
formation of the EC – but some trade diversion in
agricultural goods
Use of a control group




An alternative approach was to use a third country as a
normaliser or control group
Assumption was that intra- and extra-area imports would
have grown at the same rate as imports of the normaliser
country
Necessary that normaliser country was a similar country
not affected by any special factors
But such an approach is heavily dependent on the country
chosen
Intensity of Trade Approach






Based on the idea that there exists some “natural” amount of
trade that will take place between any two countries
If actual trade exceeds this natural amount, this suggests trade
is biased by other factors
Such factors include membership of a regional integration
scheme
Then, any increase in the degree of bias over time provides a
measure of the effects of integration
Several studies have used an intra-regional trade intensity
index or trade concentration ratio:Iij = Xij /Mj ÷ Mj /M
Where I is greater than one, this implies geographical bias in
trade due to natural and institutional factors
Intensity of trade approach
Increase in the index over time measures the trade
integration effect
 But trend over time gives a confusing picture (see Table
2):1. No change in some cases e.g. EU, EFTA
2. Upward trend in others e.g. Mercosur and Andean Pact
3. Downward trend in others
 Major weakness – lacks any proper measure of natural
trade and fails to take account of possible changes in
factors affecting natural trade

Gravity Models





Draws on Newton’s law of gravity in physics and the idea that
there is some “natural” amount of trade taking place between
any two countries, i and j (intensity of trade)
In the original gravity model, bilateral trade flows are affected
by two forces or masses - trade potential and trade resistance
Trade potential is shaped by size of each country’s GDP/GNP
and population – and thus by per capita GDP/GNP
Trade resistance is negatively related to distance between two
countries
But newer versions of the gravity models have added other
variables to increase the explanatory power of the model
Gravity Models
Newer versions of the gravity model have added other variables:1.
Geographical size – negative relationship
2.
Adjacency
3.
Common language (or cultural affinity)
4.
RTA membership
5.
Landlocked or island economies
6.
Common currency
7.
Former colonies
using a dummy variable for 2-7
 A simple form of the gravity equation might be:
log Xij = log A + β log Yi + β log Yj + μ log Hi + μ log Hj + γ log
Ni + γ log Nj + α log Dij + log εij
Gravity estimation





Simple approaches is to use the gravity equation to
estimate the amount of trade expected had integration
not occurred
Actual trade is then compared with predicted trade – the
residual measures the integration effect,
But this approach cannot distinguish between tradecreation and trade-diversion
Better approach is to estimate trade flows for integration
period only, using dummy variables for membership/nonmembership of the same trading bloc
Coefficient of the dummy variable will measure how
much extra trade was trade-creation and trade-diversion
An example: Frankel, 1997






Estimated the integration effects of 6 trading blocs
between 1965 to 1994
Found positive integration effect for all cases, with
strongest effects for ASEAN and ANZCERTA
Strong effects were also found for Mercosur and the
Andean Community
But there was no statistically significant effect for the EU
until after 1985
The EC bloc effect was stronger for the EC, but not
significant until after 1980
But the two EC enlargements of 1973 and 1985 had
major effects
Computable General Equilibrium Models





Involve construction of complex computer models
covering all the sectors in a country’s economy –
including goods, services, firms, households, government,
labour, etc.
Shows all the interconnections between different parts of
the economy using a series of equations
Model is then calibrated for a particular year and the
model is then subject to a shock to estimate the effects
Models may be single-country models (e.g. PRCGEM) or
multi-regional (e.g. GTAP)
Models can be used to study the effects of regional
integration
Uses of CGE Modelling





Were first used to analyse the effects of NAFTA and its
enlargement to Central and Southern America
Were also used extensively to study the effects of APEC
liberalisation
More recently, several models have been used to
simulate the effects of the EU and its eastern
enlargement
CGE models have also been used extensively to estimate
the effects of multilateral trade liberalisation e.g. effects of
the Uruguay Round
Can be used to predict what will happen (ex ante) as well
as estimate what has happened (ex post)
Conclusions






Important to have some idea of the size of the effects
from regional integration
Wide variety of different approaches have been used, but
they give wildly different results
Methodology used in this research has progressed from
the early days
No studies can tell us exactly what has happened or will
happen because of the counterfactual problem
But the use of econometric methods can give us a fair
estimate of the broad magnitude of these effects
CGE models are especially useful for estimating the long
run effects on economic growth