Transcript Assessing Barriers to International Trade: Measurement

Great Expectations: Ex Ante
Assessment of the Effects of
Trade Reform
Joe Francois and Will Martin
15 June 2006
We’ve made enormous progress


GTAP has given easy access to the data
needed for global trade analysis
MAcMap 6-digit data for policy analysis


Able to gain key insights into policy reforms


With preferences included
Help highlight problems, identify opportunities
Contrast with the Uruguay Round

Hard to get data even after the Round concluded
Are we there yet?

Paul Krugman’s “dirty little secret” remains:
the welfare measures from trade reform are
very small


0.7% of GDP in recent World Bank analysis
What sensible policy maker would stake her
career on gains of 0.7% of GDP?

A month’s growth in China or India
But perhaps it’s OK?


Our standard measures are based on a
rigorous theoretical framework
And perhaps the unmeasured gains scale up
these measures proportionately?


So we might rank policies, without providing a
“number”?
Alas, this is not always the case
Key questions

What needs to be measured?

What should we do better?

How might we do it?
What needs to be measured?
Some indicator of the potential for the
gainers to compensate the losers?
Where might these gains come from?






Allocative efficiency changes
Process productivity changes
Product variety & quality effects
Factor market & investment effects
Efficiency and Productivity
X
w
Q
a
U0
d
U1
c
e
d
b
U0
w
U1
Q
M
A disturbing disconnect

Traditional trade theory, and most empirical
modeling, focuses on allocative efficiency




The difference between b and e in the figure
With the PPF unchanged, except via input price
Generally small
Macro-growth literature focuses mainly on shifts in
the Production Possibility Frontier



Mainly due to process productivity
Price distortions, variety & quality effects on inputs
included, but no consumer gain
Sometimes large
What should we do better?







Welfare measures
Measures of distortions
Aggregation of distortions
Revenue replacement
Process productivity
Changes in product variety & quality
Investment, factor markets & growth
Welfare measures

The balance-of-trade function captures
production, expenditure & tariff revenues



generalizes measures based on expenditure fn
B = e(p,u) – r (p,v) – zp(p,u,v)´(p-p*)
Where e(p,u) is the expenditure fn; r(p,v) is
GDP; zp=ep – rp; p* is world price
2nd order approximation with fixed # of
products yields Harberger triangles
Can augment B to capture





Process productivity gains
Increases in product variety
Improvements in product quality
Declines in price-cost margins & output
increases at firm level
Investment and growth
The ubiquitous EV’s


The compensated measure of EV
ΔBc = B(p1,u0) - B(p0,u0)
In applied work, usually use uncompensated
EV = e(p0, u1) – e(p0, u0)

Includes income effects on distorted goods



Can differ a lot if revenues used for public goods
But quite close to a potential real GDP measure
Compensated measures internationally
comparable & additive
Measuring distortions

Doing much better with tariff measures


Ad valorem equivalents, preferences included
Key issue now is non-tariff measures (NTMs)

Kee, Nicita and Olarreaga infer these measures
from a finely disaggregated trade model


Suggest NTMs are twice tariffs in industrial-country
agriculture, four times as high in non-agriculture
Would scale up the costs of protection by a factor of
9 in agriculture, 25 in non-agriculture
Measuring distortions (2)



Francois and Woerz find ATC quotas were far higher
than we thought when rents captured by importers
are included
From price comparisons, Bradford finds NTMs may
be 10 times tariff barriers in the OECD
Not likely that the unmeasured benefits proportional
to our measured gains from tariff cuts
Better data on distortions

Detailed price comparisons and surveys of
exporters potentially very important




Kym Anderson’s project on agric distortions
Will provide better information on NTMs
May reduce measured protection– water in tariffs
Also some insights into the counterfactual

Is protection stochastic?


If so, benefits even if bound rate exceeds applied
Is mean protection stable, or increasing?
Services trade barriers



Current measures are extremely poor
Widely suspected that average barriers to
services trade are larger than in goods
And many of these barriers likely create
higher costs per unit than tariffs


Through rent seeking; reduction in competition;
or x-inefficiency
A priority, albeit a difficult one
Aggregating distortions

The ubiquitous weighted-average has major
problems



Averaging problem– understates costs
Weighting problem-- low weights on highly
protected goods
The differences can be huge


Manole and Martin found costs with optimal
aggregators 15 times those with weighted average
Some with higher average tariffs had lower costs
Estimated costs of protection
3.5
3
Optimal
2.5
2
1.5
1
0.5
0
0
0.05
0.1
0.15
Trade weighted
0.2
0.25
Aggregation

We have all the information we need to do
better



Currently, we just throw much of it away
There are papers at this conference looking
at different approaches
This is surely one area where we can do
better?
Revenue replacement

Most trade liberalization studies assume that
tariff revenues are redistributed costlessly to
consumers


Alas, such lump-sum transfers rarely exist
Tariff revenues must be replaced via an
alternative tax, or expenditure changes

Harrison, Rutherford & Tarr found that tariff
replacement via a VAT cut the benefits of
liberalization by 40%
Process productivity



Evidence from the firm level that liberalization
increases productivity by expanding the availability
and quality of intermediate inputs
Also an extensive literature on transmission of
knowledge through trade
Pavcnik found productivity gains in import-competing
firms through competition, exit and reallocation

The adjustment that governments often fear is a major
source of gain
Process productivity: exports
Exporting firms have higher productivity


But Clerides, Lach & Tybout, & Bernard & Jensen,
cast doubt on this explanation




Traditional explanations --Arrow’s learning-by-doing
Seemed to be little increase in productivity post-export
More efficient firms self-selected into exporting
Also highlighted the fact that firms are very
heterogeneous in their productivity
And that price-cost margins decline with
liberalization
Process productivity:
Heterogeneous firms
Melitz (2003) provided a framework where
trade liberalization with heterogeneous firms
yields productivity gains



High productivity firms self-select into exports
Competitive pressure following liberalization
causes exit of less-productive firms, reallocation
of resources to more efficient
Some recent studies question the rejection
of gains from learning-by-doing
Variety & quality of exports

Our traditional models assume that all export
growth is at the intensive margin




Expanded exports involve more of the same
products going to the same markets
Hummels & Klenow (HK 2005) find that 2/3 of
export expansion is in new products
Evenett & Venables find that 1/3 of export growth
in developing countries is from new markets
HK find quality upgrading on intensive margin
Export variety and quality

Assuming preference for variety, increases in
export variety augment export demand




Quality upgrading augments these gains


Shift out the export demand curve
Offset terms of trade fall implied by usual models
May help resolve old puzzles on elasticities
Hummels-Klenow find rising intensive-margin
prices in growing economies
These gains are additional to the gains from
process productivity
Factor Markets: Labor

Desirable to include labor supply


Relevant supply elasticities are compensated, so
even response by prime-working-age males
nontrivial
Labor market performance is important


Simple closures like fixed consumer real wage
increase the gains from libn, but are they realistic?
Doing better seems to require knowledge of
countries’ institutions
Investment climate

Investment climate always important


In Melitz-type models, firms “die” and need to be
replaced
Recent work suggests that the full gains from
trade are much greater if the investment
climate is supportive


Rutherford and Tarr (2002) find an 11% gain from
10% point tariff cut, with variety effects on
intermediates and with domestic reinvestment
But 37% with foreign investment allowed
To conclude:
Some key potential improvements



Improved estimates of distortions
Better aggregation of distortions
Better representation of process productivity




Heterogeneous firms & reallocation
Learning-by-doing?
Capturing changes in product variety & quality
Capturing gains associated with investment
Implementation

New data needs are difficult but important


Many of the new approaches involve
distributions of firm productivity, costs of
exporting, elasticities of substitution



Aggregation looks more straightforward
Might we be able to obtain adequate, agreed
measures of these parameters
To get agreed consequences of liberalization?
An ambitious agenda, and we can surely learn
a lot trying