Transcript Import Demand Elasticities

Estimating
Trade Restrictiveness Indices
DECRG-Trade
The World Bank
October 2004
Motivation:
Input to Global Monitoring Report.
Request by GMR:
• Comparison of trade restrictiveness across
countries
– Any patterns across regions/time?
• Two indicators:
– restrictiveness of own trade policy
– restrictiveness vis à vis Low Income Countries (LIC)
Two classic problems:
1) Trade policy takes many different forms
•
How can one aggregate tariffs, quotas, agricultural
subsidies, etc..?
2) Trade policy is determined at the tariff line level
•
How can one summarize 5000 different tariffs in one
aggregate measure?
The OTRI addresses these two problems with
sound theoretical approaches, and not ad-hoc
criteria.
1) How to measure the restrictiveness of
different trade policy instruments?
1) Focus on tariff data
–
And hope that all other instruments are (perfectly)
correlated.
2) Focus on trade data (import, export shares)
–
–
It summarizes the impact of all trade policies…
But also taste, macroeconomic shocks, rainfalls….
3) Measure all instruments with a common metric
- IMF’s TRI: ad-hoc criteria.
- AVE of NTBs: theoretically sound.
.
2) Which aggregation procedure?
1) Simple averages?
2) Import-weighted averages?
3) Frequency ratios?
3) IMF’s TRI?
4) GDP weighted for regional groups?
No. All are a-theoretical
- We follow Anderson and Neary (IER, 2003)
-
Import-volume Equivalent index:
-
What is the equivalent uniform tariff that would keep aggregate
imports at their observed levels?
Inputs to the GMR
• Three background papers:
– Estimating OTRIs (for 94 countries)
– Estimating of AVE of NTBs (for 94 countries)
– Estimating import demand elasticities (117 countries)
• Output includes three datasets:
– OTRIs (aggregate and bilateral) for 94 reporters.
– AVEs of Core NTBs and Agricultural domestic support for at
the 6 digit of the HS for 94 countries.
– Import demand elasticities for more than 117 countries at the
six digit level of the HS.
Plan
1. Methodologies
•
•
•
Estimating MTRIs
Estimating AVEs
Estimating Elasticities
2. Data sources
3. Results
•
•
•
Elasticities
AVEs
MTRIs
4. Extensions
1. Methodologies
Estimating OTRI
• What is the uniform tariff (equivalent) that would keep
aggregate imports at their current level?
 n s 0n,cn,cTn,c
Tc 
 n s 0n,cn,c
• Tc is a weighted-average of protection at the tariff line
level (Tn,c). Weigths depend on:
– Import shares:
– Import demand elasticities
Estimating AVEs of NTBs
• First we estimate the impact of NTBs on imports
(Leamer, 1990, Harrigan, 1993, Trefler, 1993).
– By country and by HS 6 digit tariff line (Leamer’s
comparative advantage approach).
• Then, we convert the import-quantity impact into
price or tariff equivalents (AVEs):
– Moving along import demand curves using estimates
of import demand elasticities.
Estimating Import Demand Elasticities
• GDP function approach (Kohli, 1991) and
Harrigan, 1997).
• Imports are inputs into domestic production,
given exogenous world prices, productivity
and endowments.
• Close links to trade theories -- general
equilibrium effects of the reallocation of
resources as prices or endowments change
• Kohli – aggregate level; Harrigan – industry
level
• Here: tariff line level (HS 6-digit)
2. Data Sources
Import Demand Elasticities:
Trade data (COMTRADE) - HS6 1988-2002 - 117 Countries.
WDI for factor endowments
Between 1000 and 4500 HS products for each country
Total number of import demand elasticities estimated is about 320,000
AVEs of NTB:
Core NTB (Price control, Quantity control, etc..)
Domestic Subsidies (WTO notifications)
UNCTAD Data (TRAINS) – HS6 (mostly around 2000) 94 Countries (EU
counts as one)
OTRI
WTO’s IDB
OECD Tariff and Trade CD-ROM. Unctad’s AVEs of specific tariffs.
94 Countries (EU counts as one).
MTRI estimated at the bilateral level (10,000 different MTRI)
3. Results
Import Demand Elasticities:
-Very precisely estimated.
-Larger for homogenous goods (i.e., larger for metal than machinery)
-Smaller as we aggregate the level at which we estimate them
- Larger in large and poor countries
AVEs of NTBs:
- Average increases with GDP per capita
- Contribution to overall level of protection increases with GDP per capita
- When present they are more restrictive than tariffs:
- in 77 percent of tariff lines where core NTBs are present
- in 45 percent of tariff lines where agricultural domestic support is present
OTRI:
- Peaks in middle income countries
- Peaks against middle income countries
- Higher in agriculture
- LIC face higher OTRI in some markets in spite of preferential schemes.
4. Extensions
•
•
•
•
Include cross-price effects in OTRI calculation.
Estimate standard errors of AVEs.
Include broader set of preferences.
Include AVEs of technical regulations.
• Regular updates?
• Others?