OFFICE OF ECONOMICS WORKING PAPER
U.S. INTERNATIONAL TRADE COMMISSION
Alternative Approaches in Estimating the Economic
Effects of Non-Tariff Measures: Results from Newly
The authors are with the Office of Economics of the U.S. International Trade
Commission. Office of Economics working papers are the result of the ongoing
professional research of USITC Staff and are solely meant to represent the
opinions and professional research of individual authors. These papers are not
meant to represent in any way the views of the U.S. International Trade
Commission or any of its individual Commissioners. Working papers are
circulated to promote the active exchange of ideas between USITC Staff and
recognized experts outside the USITC, and to promote professional
development of Office staff by encouraging outside professional critique of
Address correspondence to:
Office of Economics
U.S. International Trade Commission
Washington, DC 20436 USA
Alternative Approaches in Estimating the Economic
Effects of Non-Tariff Measures
Results from Newly Quantified Measures*
Soamiely Andriamananjara, Michael Ferrantino, and Marinos Tsigas**
U.S. International Trade Commission
This paper introduces a set of new estimates of NTM price gaps in a standard simulation model
and studies the economic effects of their removal. The economic impact of removing the NTMs
on footwear, wearing apparel, and processed foods are estimated and discussed using three
different techniques (tariff equivalent, export tax equivalent and sand-in-the-wheels). For all of
the considered sectors, NTM liberalization leads to a large increase in world trade, and an
improved global welfare. Most of the gains from the elimination of NTMs accrue to the
Draft prepared for the APEC Capacity-Building Workshop on Quantitative Methods for
Assessing NTMs and Trade Facilitation, October 8-10, 2003, Bangkok, Thailand.
* The views and conclusions expressed in this paper are those of the authors alone, and should not be in any
way attributed to the U.S. International Trade Commission as a whole or to any individual Commissioner.
** Research Division, Office of Economics, U.S. International Trade Commission, 500 E Street SW,
Washington, DC 20436, USA. E-mail addresses: Andriamananjara (firstname.lastname@example.org), Ferrantino
(email@example.com), Tsigas (firstname.lastname@example.org).
Alternative Approaches in Estimating the Economic Effects of Non-Tariff Measures
Results from Newly Quantified Measures
Through successive multilateral as well as bilateral trade negotiations, the general level of tariffs
has declined significantly during the past few decades. Concurrently, non-tariff measures (NTM)
have become more visible and their relative importance has considerably grown. Indeed, it has
been argued that the use of tari? s by governments has gradually been replaced by the use of
NTMs in order to attain the policy goals formerly achieved with tari? s (see e.g., Baldwin, 1984).
A large literature has now emerged that aims at studying the different existing types of NTMs.
Generally, one can distinguish three main types of contributions. The first type attempts to define
and to provide an organized classification of the different non-tariff measures affecting
international trade.1 Another substantial part of this literature concerns itself with the
quantification of the degree of restrictiveness of NTMs.2 A final branch consists of the use of
economic simulation models to estimate the economic effects of the removal of NTMs, based on
quantitative estimates of their economic effects. This paper is a part of a larger research program
that is currently being undertaken by economists at the U.S. International Trade Commission and
which attempts to cover all the three branches of research. It falls into the last category—
introducing newly estimated measures of NTM restrictiveness in a simulation model.
An important feature of the current research is that it attempts to assess the effects of NTMs
globally, combining data at a product-specific level with more aggregated data in the simulation
model in a manner which permits in principle comparisons across product sectors and regions.3
This approach differs from that of much previous work on NTMs. For many purposes, the
heterogeneous nature of both NTM policies and the products they are applied to indicates a
“handcrafted” approach in which the effects of policies are estimated on narrow product
categories bringing a large amount of specific institutional information to bear (Deardorff and
Stern, 1997). The present work represents an attempt to “mass-produce” estimates of NTM
effects which have previously been “handcrafted”, a process which inevitably introduces a certain
1 See for instance, Laird and Vossenaar (1991).
2 For a thorough review of the main contributions in this literature, see Bora, Kuwahara, and Laird (2002).
3 The most comparable work in this respect is that of Lawrence and Bradford (2003).
amount of noise into the estimates. It is hoped that the ability of the mass-produced estimates to
provide a survey of the landscape of NTM effects compensates at least partly for the loss of
handicraft precision in estimating the effects of particular policies in particular economies.4
Section 2 provides a conceptual framework and discusses different techniques regarding the
implementation of NTM price wedges in a model. The techniques discussed in this section
attempt to restore at least some of the “handicraft” tradition of NTM policy estimation by giving
consideration to the manner in which policies in particular sectors are usually implemented.
Section 3 characterizes a new set of estimated NTM price wedges as well as the computable
general equilibrium (CGE) as well as that is used to simulate the likely economic effects of their
removal. Section 4 presents the results of the simulation exercises for three sectors – footwear,
apparel and miscellaneous processed foods. The fourth section concludes.
2. Conceptual and Analytical Framework
To the extent that they are designed to limit trade, NTMs create an artificial scarcity and an
artificially high price. In general, the degree of restrictiveness of an NTM is measured by the
price differential that it drives between the price of imported goods and the producer price of the
domestic substitutes, or alternatively, between the domestic and the world price.5 The “wedge”
between the distorted and the non-distorted prices is the key input used in studying the potential
economic effects of the removal of a given NTM. This section discusses alternative ways to
implement a given price wedge into standard simulation models.
Because NTMs create a wedge between the world price and the domestic one, the most
straightforward way to model them is as a “tariff equivalent” above and beyond the actual tariffs.
This is generally appropriate, especially when the studied policy is implemented to directly affect
the domestic price of the imported good. For this type of policy, economic rents that results from
the higher import prices are captured by the importing economy. From the viewpoint of the
liberalizing country, the NTM removal is in this case expected to deteriorate the terms of trade
4 In the historical merchandise economy, consumers have frequently rejected mass-produced merchandise
products such as cake mixes and cigarettes on their first introduction, because of concerns regarding quality.
Subsequent improvements in quality caused the products to enter into widespread household use. It is to be
hoped that a similar learning curve operated with respect to mass-produced estimates of NTM effects.
5 Note that when foreign and domestic goods are not perfectly substitutes for each other, their price may
diverge even in the absence of any trade restraints. The introduction of a NTM will further increase
(i.e., pre-tariff prices of the imported good increase as demand for it increases) but to improve
resource allocation. Estimates of the effects NTMs for footwear and for apparel (except for
apparel importers imposing quotas under the Agreement on Textiles and Clothing) have been
implemented as tariff equivalents in this exercise.
Alternatively, NTMs can be modeled as export tax equivalents, since they restrict the ability of
exporters to ship their products. This approach hade been widely adopted in the study of
“voluntary export restraints” (VERs), which are administered by means of the exporting economy
granting licenses to particular firms to sell in the importing economy. For this type of policy, the
exporter earns the economic (quota) rents that result from being granted the right to export. In
contrast to the tariff-equivalent approach, the liberalizing country is in this case expected to
experience an improvement in its terms of trade (i.e., availability of cheaper untaxed imports) as
well as a better allocation of resources. Estimates of the effects of NTMs for apparel importers
whose policies fall under the Agreement on Textiles and Clothing have been implemented as
export tax equivalents in this exercise.
Another way to model NTMs is to introduce them as institutional frictions or “sand in the
wheels” of trade – i.e., policies that do not really create economic rents, only efficiency losses.
For instance, burdensome customs and administrative procedures, technic al regulations, sanitary
and phytosanitary (SPS) regulations, or other red tapes tend to produce an harassment effect and
to discourage imports into an economy. Removing this type of NTMs can be modeled as an
import-enhancing technological shock. The liberalizing country in this case is expected
experience deterioration in its terms of trade (i.e., world price of the imported good increases as
demand for it increases) combined with an improved resource allocation. The estimated effects
of NTMs affecting the miscellaneous food processing sectors have been modeled in this manner.
For the study of any given NTM, the choice of the most appropriate approach should be made on
a case by case basis. In the next section, we provide an illustration for each of those three
approaches using a widely used general equilibrium model, in order to determine the potential
economic effects of liberalizing newly estimated NTM price wedges.
3. Estimating the Effects of NTM Price Wedges - Methodology
As part of a large project on the quantification of NTMs, Dean, Feinberg and Ferrantino (2003)
provide ranges of new estimates of the NTM price-wedge in three selected sectors (footwear,
wearing apparel, and processed food)6 for a number of economies or regional aggregates. They
report different estimates for different model specifications (depending on which database or
combination of database they use). In this exercise, we pick the estimates from the specification
labeled “Composite”.7 These estimates are presented in Table 1.8 The absence of an estimated
wedge means one of three things : (a) the region had no NTMs on these products, (b) the policy
data contained no information on NTMs, or (c) the policy data did contain such information, but
the NTMs were not statistically associated with above-average prices given the characteristics of
the economy in question.
6 “Processed food” here refers to GTAP sector 25, “Food products nec.” This sector refers to
miscellaneous processed foods – in particular, it excludes meat and dairy products, processed rice and sugar,
and vegetable oils and fats. See Dean, Feinberg and Ferrantino (2003) for a list of the products used. to
estimate the wedges.
7 This specification introduces a composite dummy which takes a value of 1 if either the TRAINS or ITC
database records the presence of an NTM.
8 At the time of writing, work is underway to provide similar estimates for approximately 15-18 additional
GTAP sectors, which exhaust the available data and span the set of traded goods, though they exclude some
for which price data are not at present available.
Table 1. Estimated NTM price wedges for three selected sectors (percent)
Australia and New Zealand
Mexico, Central America
Rest of Latin America
Eastern Europe and Former
Middle East and North
Sub Saharan Africa
Rest of the World
Source: Dean, Feinberg and Ferrantino (2003)
The caveats presented in Dean et al. (2003) regarding these estimates should be borne in mind
when looking at the simulation results. For instance, these wedges in general were estimated for
relatively specific products but have been assigned to broader product categories for the purposes
of CGE modeling. Similarly, in some cases the existence of the measures analyzed may have
only been documented for one member of a regional grouping, but are applied to the import
policies of the entire regions. These mappings in principle mean that the estimated effects are
upper bounds. A computationally more expensive procedure, which would have provided lower
bounds, would have been to weight the measures so that they applied only to the narrow product
definitions of the price data used in the econometrics and only for the economies for which NTMs
have been documented. The choice to present upper-bound estimates reflects the judgment that
missing data for both product prices and NTM policies are extensive, and that the error involved
in treating the missing data like the available data may be smaller than that involved in treating
the missing data as if it represented situations that were completely free of NTM distortions.
In general, greater weight should be placed on the global effects and on the differences among
sectors than on the differences among economies at this stage of research. Changes in the
functional form, underlying data, or other details of the econometric exercise might redistribute
the estimated price-increasing effects of NTMs across economies, but are less likely to change the
estimated global amount of distortion by a substantial amount.
The estimates presented here are in the nature of sectoral liberalization initiatives – it is assumed
that all NTMs in a given sector are abolished worldwide on an MFN or “open regionalism” basis.
Estimating effects for three sectors on a simultaneous basis would not add much additional
information to that already presented. This method of presenting the results not only allows a
(small) computational savings, it can be considered to be in the broader tradition of APEC
initiatives. The Information Technology Agreement, which was a sectoral tariff initiative, began
through discussions in APEC which were generalized to the WTO, and the APEC Automotive
Dialogue and Chemicals Dialogue can be considered as examples of sectoral initiatives which
cover a wide variety of topics.
To estimate the economic impact of removing the NTMs, we use the Global Trade Analysis
Project (GTAP) framework which allows for the assessment and the decomposition of the welfare
effects of various trade agreements.9 GTAP has been widely used to study the likely effects of
different trade agreements and other trade policy issues, it is readily available to the public and,
the results reported in this paper can be easily replicated. 10
The GTAP modeling framework consists of a comparative static CGE model and a global
database. The CGE model is based on commonly applied assumptions of constant returns to
scale, perfect competition and product differentiation by economy of origin (i.e., the Armington
assumption). The database contains information on international and domestic markets and
primary factors, as well as tariffs and other taxes. An additional component of the data is the set
of parameters which, in the context of the model=s equations, determines responses to changes in
relative prices, among other things. The latest version of the standard GTAP database (base year
1997) is used to study the likely effects of removing the estimated price wedges.
The welfare impact of the removal of the studied NTMs is measured using the money metric
equivalent variation (EV), which can be broken down into component parts in order to enable us
to decompose the liberalization. The equivalent variation measures the welfare impact of a policy
change in monetary terms and it is defined as the amount of income that would have to be given
9 For additional information about the GTAP model and data, see Hertel and Tsigas (1997).
10 Several analytical works conducted using GTAP can be accessed at http://www.gtap.agecon.purdue.edu/.
to (or taken away from) the economy before the policy change to leave the economy as well off
as the economy would be after the policy change. A positive figure for equivalent variation
implies that the policy change would improve economic welfare.11 The equivalent variation of a
policy change consists mainly of two components: allocative efficiency and terms-of-trade.
Allocative efficiency contributions arise when the allocation of productive resources changes
relative to pre-existing policies; terms-of-trade contributions arise from changes in the prices
received from an economy’s exports relative to the prices paid for its imports.12
In this section, we introduce the estimated NTM policy measures into the GTAP modeling
framework and discuss the effects of their removal on trade, production, and welfare of different
a. Overall Characteristics
Four general equilibrium experiments are presented here – liberalizing respectively footwear,
apparel among the economies applying ATC policies, apparel among all economies applying
NTM policies, and miscellaneous processed foods. Of these, three of the experiments are similar
in that the estimated NTMs are concentrated in only two or three regions. These three
experiments share some common features. All of the liberalizing economies experience welfare
gains, which represent the gains to consumers from lower prices. All of the liberalizing
economies experience increases in both gross and net imports and decreases in production of the
products previously covered by NTMs. While most of the global welfare gains accrue to the
liberalizing economies, most other regions in the world economy experience at least some welfare
gains due to increased market access, with estimated welfare losses unusual geographically and
negligible in value when they do occur. Global production of the covered product falls,
indicating that the NTMs led to overproduction in general.
The case of generalized apparel liberalization, in which 10 of the 17 regions are assumed to
change policies, is more complex. In this case, at least some of the liberalizing regions
11 For more on the concept, see Varian (1999, pp. 252-253).
12 The standard GTAP simulations conducted here represent only the static impacts of a policy change,
while dynamic effects due to increased investment, increased competition, and economies of scale might be
important. It should also be pointed out that, under one of the central assumptions of the GTAP model,
each region has large enough market power to be able to affect world price by changing its policies.
experience increases in apparel production and net exports in the context of a more general
liberalization. Total global production increases, and the distributional effects of the policy are
more problematic. While aggregate global welfare as measured on an equivalent-variation basis
increases, welfare declines by a non-trivial amount in some liberalizing economies and some non-
liberalizing economies, due to adverse terms-of-trade effects associated with increased global
Dean, Feinberg and Ferrantino (2003) report price gaps for the footwear sector in Mexico, Central
America, and Caribbean (38 percent) and in Mercosur (95 percent). An inspection of the
underlying data reveals that the policy measures behind these wedges are mainly in the form of
quantitative import restrictions. In the GTAP model, these are treated as equivalent to ad valorem
tariffs, i.e., the quota rents are captured by the importing region in the form of government
revenues.13 Using a model closure which holds trade shares constant, the wedges are introduced
on top of the existing GTAP protection data. Thus if the initial GTAP price wedge (consisting
entirely of ad valorem tariffs) for Mexico, Central America, and Caribbean is around 20 percent,
the adjusted wedge is will be 58 percent (38 percent plus 20 percent) once the NTMs are included.
The policy experiment conducted is the removal of the part of the price wedge which relates to
the NTMs. The results are reported in Table 2. According to our simulations, shoes imports in
Mexico, Central America, and Caribbean and in Mercosur would jump by 118 percent ($1.7
billion) and 258 percent ($2.6 billion), respectively. Footwear exports would increase in many
regions, especially those in the Western Hemisphere (including those that are liberalizing) and in
Asia . Global trade in shoes is estimated to increase by almost 6 percent ($5 billion), while global
shoes output decreases by 0.6 percent (1.3 billion).
The removal of footwear NTMs in Mexico, Central America, and Caribbean and in Mercosur
would lead to deterioration in those regions’ terms of trade, in the sense that their increased
demand for foreign shoes leads to an increase in the pre-tariff import prices. The welfare losses
from the decline in the terms of trade ($227 million and $265 million, respectively), however, are
more than offset by a large improvement in resource allocation ($425 million and $1.4 billion,
13 The GTAP database does not have a broken out “footwear” sector. In our analysis, it is assumed that the
quantified NTMs apply uniformly to the much more aggregated “leather products” sector, which contains
footwear and other products.