Transcript OECD Experts Meeting on the Services Trade Restrictiveness Index

Comments at:
OECD Experts Meeting on the Services Trade
Restrictiveness Index (STRI)
J. Peter Neary
Oxford University and CEPR
OECD
Paris, 3 July 2009
Outline
•
•
•
The OECD STRI
Theory
Uses and Abuses
2
The OECD STRI
• Inputs: Expert assessments
–
… Experts? Users? (i.e., potential exporters)
• Clever aggregation of qualitative data
–
… Uniform weights?
• Market structure
– e.g., telecoms: highly concentrated
• Measuring versus evaluating policy?
– Optimal level of regulation not zero?
– Ability to explain trade & FDI an independent criterion of a good
index?
• Policy versus natural barriers
3
Uses and Abuses
•
•
•
•
•
Negotiations?
Disseminating best practice?
Industry users?
Political Economy? [Opportunism]
Academics?
4
Measuring Trade Restrictiveness in Theory
Theory well developed for tariffs on goods imports
… Extended to allow for non-tariff barriers
… Though only by reducing them to tariff equivalents
Anderson-Neary (MIT Press 2007)
Key idea:
A true tariff index equals the tariff rate which, if applied uniformly to all
imports, would yield the same outcome as the existing (typically very
non-uniform) tariffs
Different outcomes imply different indexes:
–
–
“The Trade Restrictiveness Index (TRI)”: Welfare (real national income)
“The Mercantilist Trade Restrictiveness Index (MTRI)”: Total imports
(market access)
Can this approach be applied to services?
“The True Services Trade Restrictiveness Index”
•
“should ideally be a single trade cost estimate that, when replacing all existing
fixed and variable trade costs, will yield the same trade volume and the same
5
number of trading partners as the existing trade regime”
Measuring Trade Restrictiveness in Theory (cont.)
Problems with applying approach to services:
1. Price equivalents of barriers not available
2. Fixed costs matter as well as variable costs
3. Perfect competition definitely inappropriate
Problem: Theoretical approach in A-N [“GDP function”] assumes perfect
competition?
No: Feenstra-Kee (JIE 2008) show it extends to a widely-used alternative:
monopolistic competition, with heterogeneous firms producing
differentiated goods [Melitz (Econometrica 2003) + Pareto]
• So, this model has a well-defined GDP function
• Hence the A-N approach can be applied directly
• This resolves both problems 2 and 3
• Data set already collected can lead to solution to 1?
6
5. An Alternative Approach: Integrating Extensive and Intensive Margins (cont.)
How to implement this empirically?
• Feenstra-Kee, in their empirical application, assume GDP function is
translog
–
•
More promising: Take a linear approximation
–
–
•
BUT: This requires a lot of data
Theory: Feenstra (1994), Anderson-Neary (2005, p. 19)
Applications: Lloyd (AEP 2008), Kee-Nicita-Olarreaga (REStats 2008, EJ 2009),
Irwin (NBER 2007)
Estimates of import demand elasticities for services plus data on trade
flow zeroes and OECD data on services trade barriers could then be
used to estimate a theory-consistent STRI
7
Conclusion
• A lot done
•
… A lot more to do!
8
9