WPS8570
Policy Research Working Paper 8570
Comparison of Welfare Gains
in the Armington, Krugman and Melitz Models
Insights from a Structural Gravity Approach
Edward J. Balistreri
David G. Tarr
Macroeconomics, Trade and Investment Global Practice
August 2018
Policy Research Working Paper 8570
Abstract
How large are the estimated gains from trade from a reduc- are estimated by the Melitz model. However, for individual
tion in trade costs in the heterogeneous firms Melitz (M) regions, there are numerous cases of reversed welfare rank-
model compared with the Armington (A) and Krugman (K) ings. i.e., Melitz < Krugman < Armington. For unilateral
models? Surprisingly little is known beyond the one-sector increases in tariffs, welfare gains are typically estimated with
model. This paper analyzes this question using a global the Armington model, but welfare losses with monopolistic
trade model that contains ten regions and various numbers competition models. The paper constructs a multi-sector
of sectors (1-10). Following Arkolakis et al. (2012), the Feenstra ratio for the Dixit-Stiglitz variety externality and
analysis holds the local trade response constant across the calculates changes in the terms-of-trade. These parameters
model comparisons based on a structural gravity estimate. provide economically intuitive explanations of the general
Various model features and scenarios are introduced that are pattern of results and exceptions. The paper concludes that
important to real economies, almost none of which has been gains from the reduction of trade costs for the world are:
examined across the three market structures with a constant Melitz > Krugman > Armington. For individual regions,
trade response. In response to global reductions in iceberg however, the welfare ranking of the Armington, Krugman
trade costs, in all the multi-sector models, the ranking of and Melitz market structures is model, data, parameter and
global welfare gains is Melitz > Krugman > Armington; and scenario dependent. The results highlight the need for data
the Krugman model captures between 75 and 95 percent and structural considerations in policy analysis.
on the additional gains above the Armington model that
This paper is a product of the Macroeconomics, Trade and Investment Global Practice. It is part of a larger effort by the
World Bank to provide open access to its research and make a contribution to development policy discussions around the
world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/research. The authors
may be contacted at dgtarr@gmail.com.
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.
Produced by the Research Support Team
Comparison of Welfare Gains in the Armington, Krugman and Melitz Models: Insights
from a Structural Gravity Approach *
Edward J. Balistreri, Associate Professor, Iowa State University
and
David G. Tarr, Former Lead Economist, The World Bank
Keywords: heterogeneous firms, gains from trade, structural gravity, monopolistic competition, variety
measure.
JEL classification: F12, F18
*The authors gratefully acknowledge support from the World Bank Research Committee under grant RF-P159745-
RESE-BBRSB for the project entitled "How Large are the Welfare Impacts of the Heterogeneous Firms Model:
Results from an assessment of the Trans-Pacific Partnership and from a Stylized Model." We thank Maryla
Maliszewska, Russell Hillberry, Aaditya Mattoo, Will Martin, Christophe Gouel and seminar participants at the
World Bank for comments. The views expressed are those of the authors and do not necessarily reflect those of the
World Bank, its Executive Directors or those acknowledged.
Contents
1. Introduction ........................................................................................................................................... 1
2. Literature Review.................................................................................................................................. 5
2.1 Rationales for the Gains from Trade ................................................................................................... 5
2.2 Literature on Welfare Comparisons between Armington, Krugman and Melitz ................................ 5
2.3 Applied Numerical literature on the welfare gains from trade liberalization...................................... 8
3. Trade Response Calibration, Intermediate Demand Structures and Labor-Leisure Choice ..................... 9
3.1 Calibration of the Trade Responses from a Structural Gravity Estimate ............................................ 9
3.2 Intermediate Demand Structures ....................................................................................................... 12
3.2.1 Cobb-Douglas Demand with data on intermediate use. ............................................................. 12
3.2.2 Cobb-Douglas Demand with an aggregate intermediate good. ................................................. 13
3.2.3 CES Demand for Intermediates with actual data on intermediate use. ...................................... 13
3.3 Labor-Leisure Choice ....................................................................................................................... 14
4. A Multi-Sector Feenstra Ratio and Terms-of-Trade Parameter for Interpreting Welfare Results .......... 15
4.1 The Multi-sector Feenstra Ratio........................................................................................................ 16
4.2 Terms-of-trade Parameter ................................................................................................................. 17
5. Model Results ......................................................................................................................................... 18
5.1 Welfare Equivalence in the Model with One Sector, One Factor and No Intermediates .................. 18
5.2 Breaking the Equivalence: Labor-Leisure Choice, Intermediate Goods and Multiple Sectors ........ 19
5.2.1 Labor-Leisure Choice (One Sector Model). .............................................................................. 19
5.2.2 Intermediates (One Sector). ....................................................................................................... 20
5.2.3 Multiple Sectors with Intermediates. ......................................................................................... 20
5.2.4 Relative Gains over Armington: Krugman compared to Melitz? .............................................. 22
5.3 Impact of Intermediate Demand based on Real Data and of a Lower Elasticity of Substitution for
Intermediate Inputs ................................................................................................................................. 22
5.4 Multiple Primary Factors, Sector Specific Factors and Labor-Leisure Choice ................................ 25
5.5. Tariff and Iceberg Trade Costs Reduction in a Nine-sector, Ten Region Model............................. 26
5.5.1 Global Free Trade in Tariffs ...................................................................................................... 26
5.5.2 Iceberg Trade Cost Reduction in the Nine-Sector Model .......................................................... 28
6. Sensitivity of the Welfare Results to the Trade Response and to Unilateral versus Global Policy
Scenarios ..................................................................................................................................................... 29
6.1 Sensitivity to the Trade Response ..................................................................................................... 29
6.2 Impact of a Unilateral Increase of Tariffs to Forty Percent .............................................................. 30
7. Conclusion .............................................................................................................................................. 31
References ................................................................................................................................................... 32
Table 1: Literature Results of the Ranking of the Armington, Krugman and Melitz Models Welfare
Gains from the Reduction in Trade Costs ............................................................................................... 36
Table 2: Equivalence of Welfare Results Across Market Structures in the Simplest One Sector Model,
with and without balanced trade* ............................................................................................................ 37
Table 3: Breaking the Equivalence: Labor-Leisure Choice, Intermediates and Multiple Sectors .......... 38
Table 4: Impacts of Intermediate Modeling Structure with Four-sectors with a Global Ten Percent
Reduction in Iceberg Trade Costs* ......................................................................................................... 39
Table 4a: --the Feenstra ratio and Proportional Changes in the Terms-of-trade* ................................... 39
Table 5: Impacts of Global Ten Percent Reduction in Iceberg Trade Costs with Multiple Primary
Factors, Sector-Specific Factors and Labor-Leisure Choice................................................................... 40
Table 5a: the Feenstra ratio and Proportional Changes in the Terms-of-trade* ..................................... 40
Table 6: Impact of Global Free Trade, Uniform Tariffs and Iceberg Cost Reduction in a More Realistic
Nine-Sector Model. ................................................................................................................................. 41
2
Table 6a: the Feenstra ratio and Proportional Changes in the Terms-of-trade* ..................................... 41
Table 7: Sensitivity of the Global Free Trade Welfare Results to a Common Change in the Trade
Response* ............................................................................................................................................... 42
Table 8: Impact of a Unilateral Increase in the Tariff Rates to a Uniform Forty Percent, Starting from
Initial Tariffs. .......................................................................................................................................... 43
Table 9: Benchmark Tariff Rates, by Product and Region of the Model................................................ 44
Table 10: Value Added by Region and Sector in the Benchmark Data .................................................. 45
Appendix A: Measurement of the Variety Impact in a Multi-Sector Model—A Multi-Sector Feenstra
Ratio ............................................................................................................................................................ 46
The Sato-Vartia Theorem........................................................................................................................ 46
The Sato-Vartia Index at the Sector Level .............................................................................................. 47
Feenstra’s Theorem ................................................................................................................................. 49
The Multi-Sector Feenstra Ratio ............................................................................................................. 49
Appendix B: Equations of the Model .............................................................................................................
3
1. Introduction
How large are the estimated gains from trade in the heterogeneous firms model of Melitz (2003)
compared with the homogeneous firms, monopolistic competition model of Krugman (1980) and the
perfectly competitive model of Armington (1969). This has been an important question since the
publication of the Melitz (2003) paper. In their well-known paper, entitled “New Theories, Same Old
Gains,” Arkolakis, Costinot and Rodriguez-Clare (2012) showed that in their highly stylized one-sector
model, 1 the welfare gains were identical. Key to their result was that they adjusted trade elasticities such
that the trade response was the same across the Armington, Krugman and Melitz models consistent with a
structural gravity estimate.
It is well known that the above result of Arkolakis et al. (2012) is fragile. Balistreri, Hillberry and
Rutherford (2010) found that with a labor-leisure choice in the Arkolakis et al. (2012) one-sector model
and a positive (negative) elasticity of labor supply with respect to the real wage, the Melitz model
produced larger (smaller) welfare gains than the Armington model. 2 Arkolakis et al. (2012) showed
analytically that if there is a unique aggregate intermediate good, the gains are larger in the monopolistic
competition models. Further, they showed that multiple sectors break the welfare equivalence (with
ambiguous impacts). Costinot and Rodriguez-Clare (2014) used numerical methods in multi-region,
multi-sector models. Based on their results for the average for the world, they also found that the welfare
gains are larger in the monopolistic competition models with a single aggregate intermediate good; there
were, however, several exceptions for individual regions in their results. They also found that multiple
sectors break the welfare equivalence. Melitz and Redding (2015) introduced a finite upper bound on the
Pareto distribution of productivity in a model otherwise identical to the stylized model of Arkolakis et al.
(2012). They showed that, compared with the Krugman model, there are larger welfare gains in the
heterogeneous firms model from reductions in trade costs and smaller welfare losses from increases in
trade costs. We summarize the known literature results regarding the relative welfare gains in table 1 and
in more detail in section 2.2.
There is a need to further examine the sensitivity of the results. There are a wide range of
modeling variations that have not been assessed for their relative welfare impacts. We develop numerous
new results for the relative gains from trade in the Armington, Krugman and Melitz models. For a fair
1
They assumed one sector, one factor of production; no labor-leisure choice; balanced trade in all regions; no initial
tariffs, iceberg trade costs, no intermediates and a global change in iceberg trade costs.
2
Balistreri et al., (2010), however, did not hold the trade response constant between the Armington and Melitz
models.
1
comparison, following Arkolakis et al. (2012), we hold the trade response in each scenario constant across
the three market structures based on a structural gravity estimate. Like Costinot and Rodriguez-Clare
(2014), our results are numerical. This allows us to assess not just the qualitative differences between the
estimated gains from trade from the market structure, but also the quantitative differences. In particular,
we answer the question of whether the Krugman model captures most of the gains from trade above the
Armington model. That is, what is the relative importance of the variety externality in the Krugman
model compared to the selection effect in the heterogeneous firms model in producing larger estimated
gains above the perfectly competitive Armington model? We employ a ten-region model of global trade
built on the GTAP 9 dataset. All of our exercises are comparative static without moving to autarchy.
To stay close to the earlier literature, we first replicate the Arkolakis et al. (2012) equivalence
result from a global ten percent reduction in iceberg trade costs, in a one-sector model. We then
progressively introduce real features of the data. We find that trade imbalances alone are insufficient to
break the equivalence result of Arkolakis et al. (2012) in their one-sector model. We then introduce three
features that break the welfare equivalence: (i) labor-leisure choice in a one-sector model;3 (ii)
intermediate demand in a single sector; and (iii) multiple (four) sectors with intermediate demand based
on the data.4 Next, we show that intermediate demand based on data as opposed to a single aggregate and
a lower elasticity of substitution both make a significant difference in the comparative results across the
models.5 In table 5, we investigate variations in primary factor assumptions: single primary factor; three
primary factors with and without a specific factor and labor-leisure choice. 6 Next, we expand our model
3
In addition to Balistreri et al., (2011), see also Adao, Arkolakis and Esposito (2017).
4
Arkolakis, Costinot and Rodriguez-Clare (2012) first showed that both multiple sectors and tradable intermediates
broke the welfare equivalence between monopolistic competition and the Armington model, but only tradable
intermediates led to unambiguously larger welfare estimates.
5
Costinot and Rodriguez-Clare (2014) incorporated intermediates in a multi-sector model, but they assumed an
aggregate intermediate good. So all their sectors use intermediates in the same proportion in that model. Further,
they did not test for sensitivity of the results to the elasticity of substitution for intermediates. We find that these
assumptions have a strong impact: the former increases the relative gains of the monopolistic competition models
and the latter reduces them.
6
Costinot and Rodriguez-Clare (2014, 200-223) evaluated the impact two factors of production, but only in their
autarchy exercises and with common cost shares across countries for the two factors. They report that there are
virtually no Heckscher-Ohlin effects in their Armington model, but one could question how general that result is
when factor cost shares are identical across all regions. Further, they do not report results in the monopolistic
competition models with multiple primary factors of production; so we do not have comparable results for the
Krugman or Melitz models.
Costinot and Rodriguez-Clare (2014, 231) indicate that we cannot generalize from autarchy exercises to
trade policy exercises. In an Armington model, they find opposite results regarding the impact of multiple sectors
between their autarchy exercise and a forty percent tariff increase. They conclude that “one should be careful when
extrapolating from the autarky exercises …to richer comparative static exercises. Models that point towards larger
2
to nine sectors and introduce both initial tariffs based on the data and also initial uniform tariffs that yield
the same tariff revenue as in the data; this allows us to evaluate global free trade with and without initial
uniform tariffs. 7 Finally, we use this model to investigate unilateral increases in tariffs by individual
regions; then we typically obtain estimated welfare gains in the Armington model but welfare losses in
the monopolistic competition models. Beyond the single sector model, with the exception the impact of
multiple sectors and a single aggregate intermediate good, we are the first to solve for the relative welfare
impacts of the Armington, Krugman and Melitz models in any of these model, data, parameter and
scenarios variations.
We take advantage of the calibrated share form of CES technologies and preferences suggested
by Rutherford (2002). We choose units such that initial prices are unity. Analogous to the “exact hat”
approach of authors such as Dekle et al. (2008) and Costinot and Rodriguez-Clare (2014), 8 we recover
the proportional changes in prices and quantities as the equilibrium solution to the numerical problem.
Our key results are the following.
If we consider the global welfare gains from the reduction in trade costs, in all model variants
beyond the simple one-sector model without intermediates, we find that the Melitz structure produces
larger welfare gains than the Krugman structure; and Krugman model produces larger welfare gains than
the Armington model, hereafter, M K A. If we consider the estimate of the Melitz model for the
increase in the global welfare gains above the Armington model, we find that the Krugman model
produces between 75 and 95 percent of the increase in the welfare gains above the Armington model,
depending on the model variant and data. This suggests that although the selection effect of the Melitz
model adds to the welfare gains above the Krugman model, the variety effect of the Krugman model is
quantitatively more important in explaining differences above the Armington model from global policy
changes.
gains from trade liberalization from one counterfactual scenario may very well lead to smaller gains from trade
liberalization for another.” In this paper, all exercises are comparative static.
7
Costinot and Rodriguez-Clare (2014, section 4.3) assumed uniform tariffs and zero initial tariffs when they
assessed 40 percent tariff changes in the Melitz and Krugman models; so they did not evaluate global free trade.
They assessed the impact of heterogeneous tariffs, but only in an Armington model without intermediates. Impacts
in the Armington model, however, may be misleading for the effects in a Krugman or Melitz model. For example, in
the cases of China and the United States in table 4 below that even the sign of the impact of a modeling variation in
the Armington model may be opposite of the sign of the impact in the Krugman or Melitz models.
8
The term "exact-hat" refers to the characterization of equilibrium impacts in proportional changes. That is, if ˆ
v is
the change in a variable denoted in the benchmark and counterfactual as v and v′, respectively, then ˆ
v can be
summarized as ˆ = v '/ v . See Dekle, Eaton and Kortum (2008) for an earlier application of the exact hat
v
methodology.
3
Regarding unilateral increases in tariffs without retaliation, we find welfare gains from terms-of-
trade gains in the Armington model. But due to variety losses that typically dominate the terms-of-trade
gains in monopolistic competition, we typically estimate welfare losses in the monopolistic competition
models.
There are numerous cases for specific regions of a reversed welfare ranking than the welfare
ranking for the average for the world, i.e., we have M K A for one or more of our ten regions in
many of our scenarios. These results show that for individual regions the welfare ranking of the
Armington, Krugman and Melitz market structures is model, data, parameter and scenario
dependent. We cannot conclude that one of these market structures will always produce larger or smaller
welfare gains than another without additional restrictions on the model.
We develop two parameters that are crucial to understanding the reasons for the welfare impacts
and, in particular, to understand reversed welfare rankings. These parameters are: (i) the percentage
change in the terms-of-trade; and (ii) our multi-sector extension of the Feenstra ratio, which provides a
measure of the variety impact on welfare. Without these parameters, interpreting the reasons for the
welfare rankings sometimes involves a lot of educated guesswork. In almost all cases, when the
Armington model provides more welfare gains than the monopolistic competition models, we find that
our parameter shows larger terms-of-trade gains in the Armington model. Regarding Krugman versus
Melitz, there are several cases where the Krugman model produces larger welfare gains than the Melitz
model and the Feenstra ratio shows larger variety gains in the Krugman model. We interpret the larger
welfare gains in these cases as due to larger expenditures in the Melitz model in sectors with smaller
variety increases or fewer varieties in the Melitz model.
In section 2, we summarize the literature, with a focus in section 2.2 on what we know about the
comparative welfare gains of the Armington, Krugman and Melitz models. In section 3, we discuss how
we calibrate the trade response based on gravity models. We also describe the three alternate intermediate
demand structures we model and our labor-leisure choice model. In section 4 we present our generalized
Feenstra ratio to measure variety impacts and the terms-of-trade parameters; these parameters prove
extremely useful explaining welfare results across market structures. We present the results of the
simulations in section 5 and sensitivity to the gravity estimate and the estimate of the Dixit-Stiglitz
elasticity in section 6. We conclude in section 7. In the two appendices we present the equations of the
models and our development of our multi-sector, multi-region measure of variety impacts.
4
2. Literature Review
2.1 Rationales for the Gains from Trade
One of the oldest propositions in economics is that there are gains from international trade. Ricardo
(1817) elucidated the principle of comparative advantage as the source of gains from international trade
and Samuelson (1939) established it rigorously. Krueger (1974) and Bhagwati (1982) showed that in the
presence of "rent-seeking" the gains from trade liberalization would be significantly larger than from
specialization gains from comparative advantage.9 Since 1979, numerous authors showed that under
conditions of increasing returns to scale and imperfect competition, the gains from trade liberalization
could be larger than under perfect competition for multiple reasons. The reasons included: (i) increased
competition from international trade could lower markups, which would lead to rationalization gains as
firms slide down their average cost curves, Krugman (1979); (ii) additional varieties in monopolistically
competitive markets are a source of gains from trade, Krugman (1980); (iii) international trade could add
additional varieties of intermediate inputs, Ethier (1982); (iv) foreign direct investment of multinationals
could be a source of significant gains from trade in imperfectly competitive markets, especially in
producer services markets, Markusen(1989; 2002), Markusen and Venables (1998); and Ethier and
Markusen (1996). Beginning with Melitz (2003), many theoretical papers have emphasized the
heterogeneous nature of firms in a monopolistic competition framework.10 These models of
heterogeneous firms provide a further rationale for the gains from international trade, as the endogenous
decisions of firms to enter or exit could lead to an increase in output by the more efficient firms and an
increase in the gains from trade. Following the Melitz (2003) paper, there has been a substantial increase
in research in international economics based on firm level data sets. Several new stylized facts about
international trade have been identified, including that only the most productive firms export and trade
liberalization induces an intra-industry reallocation of resources that may provide an additional gain from
trade.
2.2 Literature on Welfare Comparisons between Armington, Krugman and Melitz
In their influential paper, Arkolakis et al. (2012) construct a one-sector stylized version of the
Armington, Krugman, and Melitz models. The surprising result they found in their simplest model was
that the welfare impacts of reducing trade costs were equivalent across the three models. The simplifying
assumptions of their simplest model include: it contains one sector per country; has one factor of
production; does not contain intermediates; there is no labor-leisure choice; trade of each country is
9
Bhagwati used the term "directly unproductive profit-seeking" activities.
10
See, for example, Arkolakis et al. (2008) and Bernard, Redding and Schott (2007).
5
balanced and the reduction in trade costs is implemented as a global reduction in iceberg trade costs.
Crucial to their result is the argument that given the wide acceptance of gravity models in international
trade and the importance of the trade response to the welfare results, the structural parameters of the three
models must be adjusted to yield trade responses consistent with gravity models. Estimation of the gravity
model yields an estimate of the trade elasticity, which determines the trade response which, in turn,
determines the welfare impact.
Balistreri, Hillberry and Rutherford (2010) were the first to point out that the Arkolakis et al.
(2012) result in their highly stylized model was very fragile. Although they did not hold the trade
response constant, they introduced a labor-leisure choice in an otherwise identical model to the stylized
model of Arkolakis et al. (2012). They showed that, with a positive (negative) elasticity of labor supply,
the Melitz model produces larger (smaller) welfare gains than the Armington model. They noted the
equivalence in their model to a second sector. The intuition is that if there is a second non-tradable goods
sector, then changes in trade costs can lead to changes in the measure of goods that can be produced, and
models with perfect and monopolistic competition do not have the same welfare impacts. They noted that,
in addition to multiple sectors, intermediates would break the equivalence in the welfare results.
Arkolakis, Costinot and Rodriguez-Clare (2012) formally established that with intermediates, the
monopolistic competition models unambiguously produce larger gains from trade liberalization than the
perfect competition model; and there is non-equivalence of the welfare results with multiple sectors.
Melitz and Redding (2015) show that the Arkolakis et al. (2012) equivalence result in their one sector
model fails to hold if the Pareto distribution of productivity has a finite upper bound. Melitz and Redding
find that the endogenous decisions of heterogeneous firms to enter and exit the market provide "an extra
adjustment margin" that augments the gains from international trade compared to a Krugman style
homogeneous firms model. In their model, there are larger welfare gains from reductions in trade costs
and smaller welfare losses from increases in trade costs. The generality of the Melitz and Redding result,
regarding the welfare dominance of the model with heterogeneous firms and a truncated Pareto
distribution, has not yet been examined in models with more than one sector and other realistic features.
Costinot and Rodriguez-Clare (2014) use computer simulation models to investigate the relative
welfare impacts of Armington, Krugman and Melitz models. They begin in their section 3 with autarky
exercises, where they examine the impact of relaxing some of the simplifying assumptions of the one
sector model of Arkolakis et al. (2012). With multiple sectors, but no intermediates and zero trade
imbalances, they show that there are no selection effects in these autarky exercises, so the Melitz and
Krugman model results are identical. In their 34 regions and 31 sectors numerical model, without
intermediates, they find that the relationship between the monopolistic competition models and the
6
Armington model is ambiguous (Costinot and Rodriguez-Clare, 2014, 214-215).11 They also evaluate a
movement to autarchy with a single aggregate intermediate good without real data (where all sectors use
intermediates in the same proportions). Regarding the average for the world and most regions, they find
that M K A; but there are several exceptions to this pattern that are not discussed.
Costinot and Rodriguez-Clare (2014, table 4.3) also assess forty percent uniform tariff increases
across market structures, rather than movements to autarky, in what they refer to as their richer
comparative static exercises. Their models contain 10 regions and 16 sectors (with services always
perfectly competitive), a single primary factor of production, zero trade balances and zero initial tariffs. In
this model without intermediate goods, Costinot and Rodriguez-Clare find that aggregate losses of the
uniform tariff tend to be larger in the Krugman model compared with Armington (with three regions
being exceptions), i.e., A K , but are slightly lower in the Melitz model compared with Krugman, i.e.,
M K . They also consider a single aggregate intermediate good without real data shares (where all
sectors use intermediates in the same proportions). With a single aggregate intermediate, the aggregate
losses for the world are greatest in the Melitz model, next highest with Krugman and lowest with
Armington, i.e., M K A ; again there are exceptions to this ranking for four regions which are not
explained.
Costinot and Rodriguez-Clare succeed in defining equilibrium conditions for a general
equilibrium trade model with realistic features; but they have only numerically solved their model in the
Melitz or Krugman framework for limited special cases. When they solve a version of this model for the
Melitz and Krugman cases, however, they make the following simplifying assumptions: (i) there is an
aggregate intermediate good where all sectors consume intermediate factors in the same proportion (see
section 4.2.2 below for details); (ii) all tariffs are zero in the initial equilibrium; (iii) only a change in
uniform tariffs is considered; (iv) all trade balances are zero; (v) there is a single factor of production; (vi)
there are no sector specific factors; and (vii) there is no labor-leisure choice.
Costinot and Rodriguez-Clare subsequently assess the impact of allowing trade balances different
from zero initially; heterogeneous tariff changes; and multiple factors of production. But they only
analyze these assumptions in their Armington model. These evaluations in the Armington model are
useful for understanding impacts in the Armington model, but do not generalize to the Krugman or Melitz
models. As discussed in section 5.4, there are nine cases in the results in our table 4 below where the
impact of a model assumption in the Armington model has an opposite impact in the Krugman or Melitz
model, i.e., the sign of the impact is opposite. Clearly, knowledge of the impact of a modeling assumption
11
See Costinot and Rodriguez-Clare (2014, 219-220). In all of their models, they assume that all services are
perfectly competitive; then in their Krugman and Meltiz model, all other sectors are monopolistically competitive.
7
in the Armington model is insufficient to infer the impact of the same modeling assumption in the
Krugman or Melitz models in general. Consequently, in our summary table of known comparative
welfare results across the three market structures, we have rather limited results.
In summary, the structural gravity literature to date indicates that an aggregate intermediate good
results in larger welfare estimates in the monopolistic competition models and multiple sectors break the
welfare equivalence of the three market structures, but the welfare ranking of Melitz to Krugman varies
with the model the region and the nature of the counterfactual experiment. As we explained in the
footnotes of the Introduction section, the structural gravity literature to date has not tested the sensitivity
of the results to a wide range of important model variations. That is, what are the relative welfare impacts
of the Armington, Krugman or Melitz models with labor-leisure choice, heterogeneous tariff changes
across sectors, intermediates with real data on intermediate factor proportions, the impact of CES demand
for intermediates with an elasticity of substitution of intermediates less than one, multiple primary factors
of production (or specific factors) with different factor intensities across regions, or tariff exercises with
actual tariffs by region in the data. We also do not have results for the impacts of unilateral policy
changes as opposed to global policy changes. We address all of these issues in this paper. In table 1, we
summarize the literature results on comparing the three market structures.
2.3 Applied Numerical literature on the welfare gains from trade liberalization
The early literature was based on constant returns to scale models, where the gains were based on
comparative advantage and calculated from "Harberger triangles." The estimated gains from trade
liberalization were sometimes characterized by the "Harberger constant," i.e., the gains were generally
less than one percent of GDP from trade liberalization. Among others, de Melo and Tarr (1990) and
Jensen and Tarr (2003) showed that, even in a perfect competition constant returns to scale model, if there
were rents involved, the gains could be many multiples of the gains from the "Harberger triangles."
Regarding imperfect competition models with homogeneous firms, the path breaking article was
by Harris (1984), who showed that the gains might be much larger if the behavioral interaction of
oligopolists is altered by the trade policy. Harrison, Rutherford and Tarr (1997) estimated that the impact
of rationalization gains (sliding down the average cost curve) were small in a quantity adjusting model of
oligopoly. Rutherford and Tarr (2002) showed that in a fully dynamic model based on Paul Romer style
endogenous growth with gains from variety, the gains from trade liberalization would be many multiples
of the gains in a model with constant returns to scale. Markusen, Rutherford and Tarr (2005) and
Rutherford and Tarr (2008) showed that introducing foreign direct investment in services with Dixit-
Stiglitz endogenous productivity effects would substantially increase the welfare gains. Francois,
Manchin and Martin (2013) have summarized many approaches to modeling market structure in CGE
models and suggested ways that the alternate model structures could be tested.
8
Regarding heterogeneous firms, the first effort to numerically assess the welfare gains was by by
Zhai (2008). His model was developed into an application to the Trans Pacific Partnership in Petri,
Plummer and Zhai (2012). Unlike the Melitz model, however, neither Zhai’s model, nor the model of
Petri, Plummer and Zhai, allow entry or exit of firms, nor do these models allow uncertainly about the
productivity (Zhai, 2008, pp. 7, 8). But their models do allow existing firms to enter new markets and that
type of entry creates a new variety and a welfare gain. Since domestic firms face increased competition
from foreign entry, some would be expected to exit. The model of Zhai, however, does not allow firm
exit; consequently, the model exaggerates the variety externality. This explains why Petri, Plummer and
Zhai obtain such large increases in the variety externality in their model. Recently other authors have
adopted a computational setting for welfare analysis in frameworks with heterogeneous firms. The paper
by Caliendo, Feenstra, Romalis and Taylor (2015) is a good example. Their approach is in the tradition of
the exact hat approach.
Regarding papers that compare the welfare results across market structures, several papers have
compared the welfare impacts in the Krugman and Armington structures. These include Balistreri,
Rutherford and Tarr (2009), Rutherford and Tarr (2008) and Balistreri, Olekseyuk and Tarr (2017). These
papers, however, did not hold the trade response constant. The first numerical model of real economies
consistent with the full Melitz structure is Balistreri, Hillberry and Rutherford (2011). In their multi-
region model they find that the welfare gains from tariff reductions are several times larger than with an
Armington trade model. Jafari and Britz (2017) follow the model of Balistreri, Hillberry and Rutherford
to assess the Transatlantic Trade and Investment Partnership (TTIP); they also find that the Melitz model
produces considerably larger welfare gains than the Armington model. Neither Balistreri, Hillberry and
Rutherford (2011) nor Jafari and Britz (2017), however, hold the trade response constant across the
market structures.12
3. Trade Response Calibration, Intermediate Demand Structures
and Labor-Leisure Choice
3.1 Calibration of the Trade Responses from a Structural Gravity Estimate
Our intent is to hold the trade responses constant in the Armington, Krugman and Melitz models
based on a structural-gravity estimate. We take parameters based on econometric estimates for the Melitz
model and adjust elasticities in the Krugman and Armington model so that the trade response is the same.
12
Dixon, Jerie and Rimmer (2018) explain how to add Melitz equations to the GTAP general equilibrium modeling
system. In their application, they develop a 10 region, 57 sector model with 56 Armington sectors and one Melitz
sector to assess a unilateral tariff increase in the Melitz sector by the North American region.
9
To be precise, following Costinot and Rodriguez Clare, let W indicate the share of global
expenditures that are spent on goods that are produced in their respective home region:
r X rr
W ,
r , s X sr
where X sr is the total value of country r's expenditures on goods from country s. If W is the domestic
trade share in the benchmark data and
W is the domestic trade share in the in the counterfactual
ˆ is defined by:
equilibrium, then 1 − W is the proportional change in the global trade share where W
ˆ W .
W
W
Note that in the special case of a movement to autarky, in the counterfactual, W
' ˆ = 1
= 1 . Then and
W
W
the change in the trade response may be calculated from initial data and the trade elasticity without the
need of a model. We adjust the Armington and Krugman elasticities to match the observed percentage
increase in global trade intensity from the Melitz model. This gives us a fair comparison across structures
where they are parameterized to generate an equivalent aggregate trade response.
The calculation of the trade response is nontrivial in a multi-sector model. Although the trade
response is unique in a one sector model, in a multi-sector model there are trade responses in each sector.
Further, depending on model assumptions, the trade elasticity may not be constant.13 In this paper, we
hold the aggregate trade response constant across the Armington, Krugman and Melitz models, which
means that trade responses at the sector level are not necessarily constant across the models.
For the Pareto distribution shape parameter in the Melitz model, we choose the value 4.58 as the
mean from the structural estimation of Balistreri, Hillberry and Rutherford (2011). Arkolakis et al.
(2012) have shown that in the one sector Melitz model, the Pareto shape parameter is equal to the trade
elasticity; in particular, the trade elasticity is independent of the Dixit-Stiglitz demand elasticity. The
Dixit-Stiglitz elasticity remains relevant, however. In heterogeneous firms models with realistic features,
it affects the value to consumers of additional varieties, and in the Krugman model it is central to the trade
response as well. For the value of the Dixit-Stiglitz trade elasticity in the Melitz model, we take the value
13
Melitz and Redding (2015) showed that the "existence of a single constant trade elasticity and its sufficiency
property for welfare are highly sensitive to small departures from those Arkolakis, Costinot and Rodriguez-Clare
parameter restrictions."
10
of 5.0 that Hillberry and Hummels (2014, p.1240) report as the median estimate from cross section and
panel estimates.
Given these parameter values, for each scenario we calculate the trade response in the Melitz
model and adjust elasticities in the Krugman and Armington models to obtain the same trade response.
For the Krugman model, we adjust the Dixit-Stiglitz elasticities of substitution. In our nine-sector models,
we have some CRTS sectors in our Krugman model; to get the same trade response, we also adjust the
Armington elasticities for the CRTS sectors by the same multiple that we use for the adjustment of the
Dixit-Stiglitz elasticities. The selection effect magnifies the trade response in the Melitz model, but this
effect is absent in the Krugman model. To obtain the same trade response in the Krugman model as in the
Melitz model, we must impose larger Dixit-Stiglitz elasticities. This means that the value of an additional
variety will be greater in the Melitz model compared to the Krugman model.
Since the Krugman model has entry of firms, to hold the trade response constant, we find that we
typically must choose the Armington elasticities of substitution to be larger than the Dixit-Stiglitz
elasticities of our multi-sector Krugman models. In the Melitz model, the value of 4.58 for the shape
parameter and 5.0 for the Dixit-Stiglitz elasticities are invariant across the scenarios. In the case of the
Armington and Krugman models, the Armington and Dixit-Stiglitz elasticities vary with the scenario, and
those values are in the tables of results.
In an effort to keep the love of variety effect constant in the Krugman and Melitz models, one
might start with the Krugman model. That would require that we calibrate the Dixit-Stiglitz elasticities of
substitution to be consistent with an estimate of the trade elasticity from a gravity model. Then we would
attempt to impose those same Dixit-Stiglitz elasticities of substitution in the Melitz model. To hold the
trade response in the Melitz model consistent with the gravity estimate, we would then attempt to adjust
the Pareto shape parameter to achieve a trade response in the Melitz model consistent with the trade
response in the Krugman model. The problem with this calibration strategy is that it requires a value of
the Pareto shape parameter, a , very close to − 1 , where is the Dixit-Stiglitz elasticity of
substitution. At a − 1 it is well known that the Melitz equilibrium is ill defined, as the productivity of
the representative firm, and all the aggregates dependent on it, are unbounded. As a matter of
computation, the Melitz models fail to solve reliability as the value of the Pareto shape parameter
approaches − 1 from above. The economic intuition is that if the selection effect in the Melitz model is
positive, then the entry effect in the Krugman model will have to be larger for both models to be
consistent with the same trade response. To get larger entry effects in the Krugman model we need larger
values of the Dixit-Stiglitz elasticity, which will give us lower gains for a single variety in the Krugman
model (other things equal). Despite a variety being valued more highly in the Melitz model, our results
11
show that it is possible for shifts in expenditure shares in multi-sector models to sometimes lead to larger
variety gains in the Krugman model (a result explained by Feenstra, 1994).
3.2 Intermediate Demand Structures
We examine the impact of market structure with three intermediate demand structures: (i) Cobb-
Douglas demand using the data on intermediate use in the input-output tables at the sector level; (ii)
Cobb-Douglas demand with an aggregate intermediate which assumes that all sectors use intermediates in
the same proportions; and (iii) CES demand for intermediates with elasticity of demand for intermediates
of 0.5, using the data on intermediate use in the input-output tables at the sector level. Given that the
structure is the same across regions, we suppress subscripts for the region in the equations of this section
describing these intermediate use and demand structures.
3.2.1 Cobb-Douglas Demand with data on intermediate use.
Define:
Pis as the composite price of the i-th good in region s. This is the dual price variable that we define as the
left-hand side variable in appendix B, equations 2a, 2b, and 2c. The definition depends on the three
intermediate demand structures discussed below. Pis incorporates the technology and the optimization,
and its depends on the Armington, Krugman or Melitz market structure.
Define xij as the intermediate inputs from sector i used in sector j (in region s);
ij = Px
i ij / i Px
i ij = the share of intermediate expenditures of sector j on intermediates of sector i;
Total intermediates use of sector j are a Cobb-Douglas aggregate of its intermediates uses from the
sectors.
Z j = i xijij
When we model intermediates as Cobb-Douglas, we also assume the production function in sector j is
Cobb-Douglas; then output in sector j is:
1− j − j − j
Qj = K j j Lj j Rj j Z j where j + j + j 1 ; j 0, j 0, j 0,
where K j , Ll and R j are the primary factors capital, labor and a primary resource factor that is sector
specific, respectively, used in sector j. We investigate the impact of primary factors by assuming
variously that: (i) labor is the only primary factor; (ii) labor and capital are the only primary factors; and
(iii) also, in some models, there is a sector specific factor along with capital and labor.
12
3.2.2 Cobb-Douglas Demand with an aggregate intermediate good.
Costinot and Rodriguez-Clare (2014) and Balistreri et al. (2011) employ a simplifying
assumption regarding intermediate goods—they assume an aggregate intermediate good. Use of the
aggregate good implies that, given an aggregate amount of intermediate purchases, factor proportions
among intermediate goods are identical across sectors. For example, if autos and accounting services are
two sectors in the model, the share of steel as an intermediate good is the same in the two sectors. To be
more precise, define the following:
e jh as expenditure by household h on consumption goods in sector j;
Aj = h e jh + i Pj x ji as the total expenditure on goods in sector j for both consumption and intermediate
use, where Pj and xij were defined above;
j as the share of total expenditure on both intermediate and final goods that is spent on good j
j = Aj / i Ai .
Finally, define the aggregate good Z as a Cobb-Douglas aggregate good of all intermediate and final
purchases:
Z = i Aii
In this model, all firms use only the single composite commodity Z as its intermediates. That is, in all
sectors, output is a Cobb-Douglas aggregate of labor, capital, the primary resource factor and the
composite commodity Z:
Qi = K ii L i 1− i − i − i
i Ri Z
i
Note that intermediate use in the production function in sector i is not indexed by sector. It means that all
firms allocate expenditure on intermediates from different sectors in the same proportions, not different
intermediates based on the proportions in which they use the intermediates based on the data. Further, in
this model, utility of consumers is a function of the composite commodity Z (which includes
intermediates), rather than the final goods alone. We investigate the impact of these simplifying
assumptions below.
3.2.3 CES Demand for Intermediates with actual data on intermediate use.
In this formulation, total intermediates are a CES aggregate of intermediate uses from the sectors:
1
Zj =
j
j
i ij xij
j 1
13
Where ij and xij were defined in section 3.2.1. The production function is a CES aggregate of value-
added and total intermediate demand:
1
j j
Qj = j
VA( j )VA j + Z ( j ) Z j j 1
In our central formulation, we take the elasticity of substitution as j = = 0.5 for all j. And value-
added is a Cobb-Douglas aggregate of primary inputs:
1− j − j
VAj = K j j L j j R j ,
where the primary factors of production are defined above in section 3.2.1.
3.3 Labor-Leisure Choice
In our models with labor-leisure choice, we assume that utility in each region is a CES function of
leisure and an aggregate of all goods and services consumption. Utility of goods and services
consumption is a CES function of the various goods and services, so by two-stage budgeting, we may
consider an aggregate non-leisure consumption good C, with a dual price P. Suppressing subscripts for
the regions, we have that utility is:
1
1− 1− 1
U =
+ (1 − ) C
1 =
1−
Where = leisure and C = consumption of the aggregate good/service.
Let E = the total time endowment of the consumer/worker; W = the wage rate; L = labor supply; and P =
the price index of goods and services. The demand for leisure is:
(WE + Y )
= where k = W 1− + (1 − ) P1− and
Wk
the uncompensated elasticity of leisure demand with respect to the wage rate is:
E *W (1 − )W
= − −
W *k Wk
and the uncompensated elasticity of labor supply with respect to the wage rate is14:
E
L = = − L − 1
−E
We evaluate the labor supply elasticity in the neighborhood of the initial equilibrium, where we choose
units such that W = P = k = 1. Then we have:
14
See Ballard (2000) for details of the derivation.
14
E E
L = − − 1 − (1 − ) −
L E + Y
In the special case of a single primary factor, zero trade balance and zero non-labor factor income, the
elasticity of labor supply reduces to:
E
L = − 1 ( − 1)(1 − ) .
L
Then the elasticity of labor supply is positive if an only if the elasticity of substitution exceeds unity.
.
The compensated elasticity of leisure with respect to the real wage is:
W 1+ 2 −1
* = W k
−2 1−
− W − −1k 1−
k 1+
and the compensated elasticity of labor supply with respect to the real wage is:
E E
L
*
= − − 1 * = − 1 (1 − )
L L
It is evident that the compensated elasticity of labor supply is always positive, where we have used our
choice of units such that W= P = k = 1, to derive the right-hand side equality.
To calibrate our model, we need three parameters: σ, μ and E. The compensated and
uncompensated elasticities of labor supply, both depend on three parameters: σ, μ and E, or E/L. Based on
empirical estimates, we take the uncompensated and compensated elasticities of labor supply with respect
to the real wages as: 0.2 and 0.7, respectively. In addition, given our choice of units, we have the relation
E
that −1 = Then with our equations for the compensated and uncompensated elasticity of labor
L 1−
supply we have three equations in these three parameters.
4. A Multi-Sector Feenstra Ratio and Terms-of-Trade Parameter
for Interpreting Welfare Results
In general, the monopolistic competition model can produce larger gains than the Armington
model due to: (i) more varieties; (ii) lower markups over marginal costs that lead to rationalization gains;
and (iii) with heterogeneous firms, an increase in average firm productivity from the selection effect. But
in a multi-sector, multi-region model, there are also (iv) terms-of-trade effects that can impact differences
in welfare results across regions and market structures. Since we assume, as in Krugman (1980), that the
ratio of fixed to marginal costs for a firm is fixed with respect to the quantity, rationalization gains are
15
ruled out as an explanation of our results. In order to help with the interpretation of the variety, selection
and terms-of-trade effects, we have developed two parameters: a multi-sector version of the Feenstra ratio
and the proportional change in the terms-of-trade.15 We have found these parameters very insightful in
explaining results and we report the values of both of these parameters for all scenarios, for all ten regions
in this paper.
4.1 The Multi-sector Feenstra Ratio
Feenstra derived a measure we call the Feenstra ratio that precisely measures the welfare impact
of the variety externality in a one-sector monopolistic competition model. We extend that measure to a
multi-sector, multi-region Krugman or Melitz model. Details are in appendix A; here we provide the key
results.
Despite the fact that all varieties are consumed everywhere in the Krugman model, Feenstra
(1994; 2010) has shown that in a one-sector model, the welfare impact of a variety depends on its
expenditure share. From Feenstra (2010, Theorem 2), the Feenstra ratio is:
1/( −1)
(I ) e( pt −1 , I t −1 ) SV
F = t −1 = P ( pt , pt −1 , qt , qt −1 , I )
t ( I ) e( pt , I t )
where e is the unit expenditure function, pt and qt are the vectors of prices and quantities in period t, I t
is the set of goods available in period t at prices pt , I is the set of goods available in both periods t and t-
1 and P SV (...) is the Sato-Vartia index that shows the ratio of the unit expenditure function in the two
periods if the sets of available goods in the two periods are identical. Feenstra’s theorem tells us that the
proportional change in the unit expenditure function is the product of two terms that are: (i) a measure of
the change in the prices charged by firms on goods common to the two periods (the Sato-Vartia index);
and (ii) a ratio that is a measure of the value of the variety gain, the Feenstra ratio. The methodology
requires distinguishing changes in the price charged by firms on goods available in both periods from the
variety externality. Feenstra shows that 1 − t ( I ) is the expenditure on new goods relative to total
expenditure in period t, so a larger number of new goods in period t will tend to lower t ( I ) . A value of
1.01 for our Feenstra ratio means that the cost of a unit of utility declined by one percent due to new
varieties.
In a multi-sector, multi-region model, we define the Feenstra ratio for region s as:
15
Quantifying an isolated selection effect in an open economy is nontrivial. Melitz (2003, p. 1721) notes that his
aggregate productivity measure in an open economy could be less than his aggregate productivity measure in
autarchy due to the output loss in transit of exports.
16
Fs = is Fis
i
where is is the economy-wide expenditure share (absorption share) on goods or services of sector i in
region s; and Fis is the Feenstra ratio for sector i in region s defined as:
PDSis (0) SV
Fis = Pis
PDSis (1)
where PDSis (t ) is the Dixit-Stiglitz price index in sector i in region s, and PisSV is the Sato-Vartia index
for sector i in region s. Given that the Sato-Vartia index remains an “ideal” index of firm level prices at
the sub-sector level, and the Dixit-Stiglitz index is the unit cost of goods in sector i taking variety into
account, our index Fis is precise at the sector level. If we have a one-sector model, our measure reduces
to the ratio of Feenstra (2010). Given the existence of intermediates in our model, the aggregation across
sectors is an approximation, but we have calculated and presented this aggregate variety measure for more
than 40 scenarios in this paper, and the measure has consistently proved insightful and intuitive in
explaining the welfare results.
4.2 Terms-of-trade Parameter
We define the terms-of-trade for region r as the weighted average of export prices divided by the
weighted average of import prices:
exp P ir ir
TOTr = i
imp P
i sr
isr is
Pir is the dual price which we defined in section 3.2; it is fully specified in appendix B, equations 2. As in
section 3.1, define X isr as the total value of country r's expenditures on good i from country s (excluding
tariffs). The export weights are:
X
sr
irs
expir =
X
j sr
jrs
The import weights are:
(1 + tarisr ) X isr
impisr = for r s
(1 + tarjtr ) X jtr
j t r
17
Where tarisr is the tariff rate on imports from region s shipped to region r. Note that for exports, there is a
unique weight associated with any good exported from a particular region; but the import weights impisr
are bilateral by product.
5. Model Results
In order to clearly examine the impact of model assumptions and stay close to the results in the
literature, we begin with single sector and four-sector models and examine a ten percent reduction in
iceberg trade costs. For our tariff policy scenarios, we employ a more realistic nine-sector model. All our
models contain ten regions and an untruncated Pareto distribution in the Melitz model. To be consistent
with the literature, we rebalance the GTAP data so that all final demand is consumption demand. For our
welfare results, we report Hicksian equivalent variation as a percent of the benchmark value of real
consumption.
5.1 Welfare Equivalence in the Model with One Sector, One Factor and No
Intermediates
We have two sets of simulations in table 2 with welfare equivalence between three market
structures. First, we consider the simplest model of Arkolakis, Costinot and Rodriguez-Clare (2012), i.e.,
one-sector, one primary factor of production, no intermediate inputs, trade costs are iceberg, balanced
trade, no labor-leisure choice, untruncated Pareto probability distributions of productivity. The
counterfactual shock that we consider is a global uniform reduction in iceberg trade costs of ten percent.
In column 2, we calculate the gains from trade using the formula of Arkolakis, Costinot and Rodriguez-
Clare (2012). In columns 3-5, we calculate the welfare change from the reduction in trade costs using our
model. To be consistent with the model of Arkolakis et al. (2012), we rebalanced the data from the GTAP
dataset such that all demand is final demand in table 2 (i.e., no intermediates), and there are zero trade
balances in all regions of the model. Our simulations replicate the result of Arkolakis et al. (2012) that
the alternative market structures generate identical welfare impacts. Further, the table shows that welfare
can be derived from a very simple calculation based only on the domestic expenditure share trade
response and the trade elasticity. These are precisely the points made by Arkolakis et al. (2012).
In columns 6-8, we allow the initial trade balances based on the data. Note that the domestic
expenditure shares rr change as a result of the incorporation of the benchmark trade imbalances. In all
the subsequent simulations in this paper, we assume trace imbalances based on the data and that the
observed current-account imbalances are held fixed (in units of US consumption). Costinot and
18
Rodriguez-Clare (2014) explain how their welfare formula must be modified to incorporate benchmark
trade imbalances and the ambiguity this creates in terms of setting up the counterfactual.16 In our case, the
numeraire is chosen to be the aggregate price index in the US, and so we simply hold the value of each
region's current account fixed in these units. We find that in this class of one sector model, when rounded
to a single decimal point of percentage change, we retain the welfare equivalence of the three market
structures, i.e., the trade imbalances are not sufficient to separate the welfare results across the three
market structures.
5.2 Breaking the Equivalence: Labor-Leisure Choice, Intermediate Goods and
Multiple Sectors
In table 3, we introduce three model features that break the equivalence of the welfare results
between the market structures: (i) labor-leisure choice; (ii) intermediates in a single sector (without labor-
leisure choice); and (iii) multiple sectors with intermediates (without labor-leisure choice). In these
scenarios we again consider a ten percent reduction in iceberg trade costs. We retain the single primary
mobile factor assumption and allow trade imbalances in all scenarios based on the data. In our Krugman
and Melitz multi-sector models in these scenarios, all the sectors are monopolistically competitive. The
values of the elasticities in the Armington, Krugman and Melitz models to hold the trade response
constant (based on a gravity estimate) are shown at the top of columns 1-9.
5.2.1 Labor-Leisure Choice (One Sector Model).
We add labor-leisure choice to the model in the previous set of simulations and present results in
columns 1-3 of table 3. To facilitate comparison across the model structures, we continue to report
Hicksian equivalent variation as a percent of consumption of goods and services in the benchmark
equilibrium even in the models with labor-leisure choice.17 We see that labor-leisure choice breaks the
equivalence. For all regions, we have that M K A , i.e., the Krugman model produces larger welfare
gains than the Armington model and the Melitz model produces slightly larger welfare gains than the
Krugman model in all regions; in some cases (such as the OECD NEC and Australia-New Zealand),
however, the difference between the Krugman and Melitz model results is only in the second decimal.
The larger gains in the monopolistic competition models are explained by the fact that the decline in
trade costs increases real wages, which, with our positive elasticity of labor supply, makes more labor
available for production. More labor available, leads to more varieties and an increase in welfare due to
the Dixit-Stiglitz externality. Although they did not hold the trade response constant, this was the
16
Their welfare formula and the data needed to calculate welfare becomes progressively more complex as they
incorporate real features of the data.
17
That is, we do not include the imputed value of leisure in the denominator.
19
argument of Balistreri et al. (2010). This intuition is verified by noting that all Feenstra ratios in columns
3 and 5 of table 3a exceed unity.
Regarding the larger gains in the Melitz model compared to the Krugman model, as explained in
section 3.1, to calibrate to the common trade elasticity from gravity, we must select a higher value of the
Dixit-Stiglitz elasticity in the Krugman model. This means that an additional variety is valued more
highly in the Melitz model. Holding expenditure shares and the number of varieties constant, there will be
larger gains from variety in the Melitz model. We see that the Feenstra ratio values are higher for all
regions in the model other than the United States. For the United States, the expenditure shares (or fewer
varieties in the Melitz model) induce a very slight reversal of the Feenstra ratio values. Since the terms-
of-trade effects are the same in the Krugman and Melitz models, this suggests that selection effects in the
Melitz model explain the larger welfare gains in the Melitz model.
5.2.2 Intermediates (One Sector).
The impact of adding intermediates in a single sector model is shown in columns 4-6 of table 3. To
isolate the impact of intermediates, we hold labor supply fixed. We assume Cobb-Douglas production in
the single primary factor (labor) and the single intermediate good.
In all regions of the model, we see that M K A . As shown by Feenstra ratios greater than one in
all cases, the monopolistic competition models provide larger gains than Armington due to variety gains.
The Melitz model provides additional gains relative to Krugman in these simulations. Given the larger
Dixit-Stiglitz elasticity of substitution in the Krugman model, the value of a variety is greater in the
Melitz model. Except for the United States, the Feenstra ratio is slightly larger in the case of the Melitz
model (table 4a, columns 9 and 11); along with selection effects, this explains the larger gains from the
reduction in trade costs in the Melitz model. This model departs in multiple ways from the model of
Melitz and Redding (2015) in that it contains intermediates, a trade imbalance and untruncated Pareto
distributions. Nonetheless our results in columns 1-6 of table 3 are consistent with the conclusion of
Melitz and Redding (2015) that the heterogeneous model provides larger welfare gains than the Krugman
model in these single sector models.
5.2.3 Multiple Sectors with Intermediates.
In columns 7-9 of table 3, we present results with four monopolistically competitive sectors, labor as
a single primary factor of production and Cobb-Douglas demand for intermediates and labor (see section
4.2.1 for details of the intermediate demand structure). For the average welfare impact for the world and
for seven of the ten regions, we again have the ranking M K A . Additional varieties contribute to
larger welfare results and partly explain the larger net gains in the Krugman and Melitz models over
Armington for most of the regions and the average for the world (see the Feenstra ratio values in table
3a). Again, given our calibration to a common trade response, an additional variety is valued more highly
20
in the Melitz model; for six of our regions, the Feenstra ratio shows a larger welfare gain in the Melitz
model than the Krugman model, which (along with possible gains from the selection effect) explains the
larger gains in the Melitz model compared with the Krugman model.
With the introduction of multiple sectors, however, we have a reversal of the welfare ranking in
the case of the Low Income region, i.e., we have M K A . The larger welfare gains for the Low
Income region in the Armington model derive from larger terms-of-trade gains than in the monopolistic
competition cases (columns 12, 13 and 15 of table 3a). So a smaller terms-of-trade gain in the
monopolistic competition models outweighs their variety gains, leading to larger welfare gains in the
Armington model for the Low Income region. This shows that the significant terms-of-trade effects that
may be present in multiple sector, multiple region trade models, can dominate the variety and selection
effects for some regions.
For three regions, we have the reversed welfare ranking of M K : the Low Income region;
Middle-Income region; and Australia-New Zealand (in the second decimal). Despite the fact that a
variety is valued more highly in the Melitz model, in all three of these regions (and only these three
regions), we see that the Feenstra ratio shows lower variety gains in the Melitz model. From the Feenstra
ratio, we know that if expenditures are small on products that experience variety gains, the welfare impact
of the variety gain will be reduced. In these three regions, the lower Feenstra ratios in the Melitz case
show that, compared with the Krugman model, either expenditure shares shifted to sectors with fewer
increases in varieties or there was a smaller increase in varieties in the Melitz model.
The fact that not all regions follow the pattern of the average for the world is consistent with the
results of Costinot and Rodriguez-Clare (2014, table 4.3, p. 232). In their ten region model with 16
sectors and intermediates, they show reversed welfare rankings between Krugman and Armington in three
of their regions (Eastern Europe, North America and Rest of World) and reversed welfare results for
Krugman versus Melitz in the case of their Pacific Ocean region. They interpret the results for the average
for the world based on the variety externality in the monopolistic competition models and the value of
Dixit-Stiglitz elasticity for difference between Melitz and Krugman. They do not explain, however, the
cases where they find reversed relative welfare rankings.18 They do not use either expenditure shares
18
In their ten region, sixteen sector model with intermediates, they conduct comparative static exercises of forty
percent tariff increases. For seven of their regions and the average for the world, they find that the absolute value of
the welfare losses is largest under Melitz and smallest under Armington. They explain that the larger losses in the
Melitz case reflect the larger losses from a single variety in the Melitz case (and implicitly, consistent with their
earlier interpretations, the losses are larger under Krugman than under Armington due to the loss of varieties). In
their 34 region model, Costinot and Rodriguez-Clare (2014, table 4.1, p. 206) evaluate a movement to autarchy.
They find that on average for the world: M K A . But they find four regions (Australia, Greece, Romania and
the Russian Federation) with the reversed welfare ranking M K A .
21
across sectors nor terms-of-trade as part of their interpretation discussion. Our Feenstra ratio and terms of
trade parameters, however, show that these are the keys to understanding reversed welfare rankings.
5.2.4 Relative Gains over Armington: Krugman compared to Melitz?
In all model variations in tables 3-6, for the average for the world, we find that M K A.
Regarding the larger welfare gains of the monopolistic competition models, we ask how much of an
additional welfare gain does the Melitz model provide over the Krugman model?
Let K A = the ratio of the welfare gains in the Krugman model to the welfare gains in the
Armington model; M A = the ratio of the welfare gains in the Melitz model to the welfare gains in the
Armington model. We define the Krugman model’s share of the larger welfare gains in the Melitz model
over the Armington as: RK / M =
K A − 1 . For all scenarios in tables 3-6, we calculate and report this
A − 1
M
parameter in the text. To illustrate, consider the scenario with intermediate goods in a single sector model
in table 3, columns 4-6. Regarding the average for the world, the Krugman (Melitz) model results in gains
that are equal to 135 =3.8/2.8 (143 = 4.0/2.8) percent of the gains in the Armington model. Then the
Krugman model provides an additional 35 percent of welfare gains over the Armington model and the
Melitz model provides an additional 43 percent of welfare gains over the Armington model. Then, with
intermediates in a single sector model, the Krugman model’s share of the larger welfare gains in the
Melitz model over the Armington model as: 35/43 = .82. In the case of labor-leisure choice in table 3, for
the average for the world, the RK / M ratio is .87; and in the case of four-sectors in table 3, the ratio is .76.
We conclude from this that, regarding the gains to the world, compared to a Melitz model, a Krugman
model would capture the majority of the gains of the Melitz model above the Armington model.
5.3 Impact of Intermediate Demand based on Real Data and of a Lower Elasticity of
Substitution for Intermediate Inputs
In this subsection, we assess the impact of intermediate demand with (a) real data on intermediate
demand by sector; and (b) the elasticity of substitution for intermediates from different sectors.19
19
As explained in section 3.2.2, the structural gravity literature to date has produced results for the relative welfare
gains for Armington, Krugman or Melitz models only for the case of a single aggregate intermediate good with
Cobb-Douglas demand, which ignores the data that different sectors use particular intermediates in different
proportions. Costinot and Rodriguez-Clare (2014, p. 219) acknowledge that it would be better to use the data on the
intermediate shares, but their algorithm was unable to find a solution with the actual data (in the monopolistic
competition cases).
Costinot and Rodriguez-Clare (2014, table 4.1, p. 206) investigate the impact of using real data on intermediates in
their Armington model with Cobb-Douglas demand. Those results are useful for understanding impacts in the
22
While holding the trade response constant across market structure models, we evaluate a ten
percent reduction in iceberg trade costs in the Armington, Krugman and Melitz models for three
structures of intermediate demand discussed in section 3.2: (i) a single aggregate intermediate good
(columns 1-3); (ii) Cobb-Douglas demand for intermediates based on data on intermediate shares
(columns 4-6); and (iii) CES demand for intermediates based on data on intermediate shares, with
elasticity of substation equal to 0.5 (columns 7-9). To isolate the intermediate demand structure effect, in
all models in this subsection 5.3, we assume four sectors that are all monopolistically competitive in the
Krugman or Melitz cases, a single primary factor, zero tariffs, no labor-leisure choice, but we incorporate
trade imbalances from the data. Table 4 shows several interesting results.
First, allowing intermediate demand to be disaggregated based on data increases the gains in the
monopolistic competition models relative to the Armington model. With Cobb-Douglas demand for
intermediates and a single aggregate intermediate good, equivalent variation in the Krugman (Melitz)
model is 138 (147) percent of the equivalent variation in the Armington case; and these ratios rise to 155
(172) percent with intermediate shares based on data.
Second, the assumption of Cobb-Douglas demand for intermediates is very powerful. With the
lower elasticity of substitution, the quantitative differences between the welfare gains in the perfect
competition model and the monopolistic competition models is very significantly reduced. With CES
demand and an elasticity of substitution for intermediates of 0.5, equivalent variation in the Krugman
(Melitz) model falls to 107 (108) percent of the equivalent variation in the Armington case. With the
lower elasticity of substitution, firms are less able to substitute goods from sectors whose Dixit-Stiglitz
quality adjusted price falls the most; thus, the monopolistic competition models benefit less from the
variety increases.
Third, focusing on the average for the world (and for seven of our ten regions), we have non-
equivalence of the welfare results with M K A in all three intermediate model structures. Similar to
results in section 5.2.3, there are variety gains, verified by the Feenstra ratios in table 4a, that explain the
larger gains in the monopolistic competition models on average. There are possible productivity gains in
the Melitz model from the selection effect that would contribute to the larger welfare gains in the Melitz
model. Since an additional variety is valued more highly in the Melitz model, for six of the seven regions
(excluding the United States) where M K , we can see from the Feenstra ratios that the variety impact
is larger in the Melitz model for all three intermediate demand structures.
Armington model. But regarding the relative impacts of the Armington, Krugman and Melitz models, prior to this
paper, the literature does not provide any results for the welfare impacts of an intermediate demand structure with
real data in either the Cobb-Douglas case or for a lower elasticity of substitution.
23
Fourth, in multi-sector, multi-region models, terms-of-trade effects can result in reversed welfare
rankings for some regions, with larger gains in the Armington model compared to either or both of the
monopolistic competition models, i.e., we can have A K or A M . For the Low Income region, we
have A K M with all three intermediate demand structures and for the Middle-Income region and
Australia-New Zealand, in the case of the low elasticity of substitution of 0.5 we have A K M . Our
terms-of-trade parameter reveals that, for these three regions, the change in the terms-of-trade is better in
the Armington model than in both monopolistic competition models for all three intermediate demand
structures. Although our Feenstra ratio parameter shows that these regions gain from variety, these
results show that it is possible for a relatively better terms-of-trade impact in the Armington model to
dominate the welfare ranking over the monopolistic competition models.
Fifth, a smaller variety gain that can contribute to a reversed ranking of K M . For the three
regions Lower-Income, Middle-Income and Australia-New Zealand, and the intermediate demand
structures with real data, we have K M . In the cases of these regions and intermediate demand
structures, the Feenstra ratio shows a lower variety gain in the Melitz model than the Krugman model.
Since an additional variety is valued more highly in the Melitz model due to the lower Dixit-Stiglitz
elasticity, the larger Feenstra ratio in the Krugman model shows the importance of expenditure shares in
these cases or possibly a smaller increase in varieties.
Sixth, regarding the relative quantitative importance of the Krugman or Melitz model over the
Armington model, we see that, with these global changes, the Krugman model captures the majority of
the differences in these cases. Using the results or the average welfare change for the world, the values of
the parameter RK / M for the Krugman model’s share of the larger welfare gains in the Melitz model over
the Armington model are: .80 for Cobb-Douglas demand with a single aggregate intermediate; .76 for
Cobb-Douglas demand with data on intermediate shares; and .87 for CES demand with data on
intermediate shares.
Seventh, table 4 also provides concrete examples to the logical point that knowledge of the
qualitative impact of a modeling assumption in the Armington model does not imply that the same
modeling assumption will have the same qualitative impact in the Krugman or Melitz models. Consider
the impact of a Cobb-Douglas intermediate structure with real data compared to Cobb-Douglas demand
with an aggregate intermediate good. There are nine cases in table 4 where the sign of the impact in the
Armington model has the opposite sign of the impact in either the Krugman or Melitz model.20 Take
China in particular. If we compare the estimated welfare gains for China using the Armington model
(shown in table 4, columns 1 and 4), we see that they are smaller with the real data (3.5% – 3.6% < 0).
20
See the cases of Canada, China, Japan, Low Income and Middle Income.
24
But with the Krugman model, the welfare change for China of using real data is positive (9.7% - 7.8% >
0).
5.4 Multiple Primary Factors, Sector Specific Factors and Labor-Leisure Choice
Heretofore, we have assumed a single primary factor of production. In this section, we investigate
the impact of allowing primary factors of production to include labor, capital and a resource factor which
may be sector specific. Demand for primary factors is Cobb-Douglas. We also consider a case with labor-
leisure choice, using the structure explained in section 3.3. All scenarios assume: four monopolistically
competitive sectors; CES demand for intermediates based on the data on intermediate shares by sector,
and an elasticity of substitution equal to 0.5; initial trade imbalances; and initial tariffs. We take primary
factor share intensities from the GTAP data. We simulate a global ten percent reduction in iceberg trade
costs.
For the Armington, Krugman and Melitz market structures, we produce estimates for the welfare
gains with a single primary factor in columns 1-3 and with three primary factors columns 4-6. Holding
market structure constant, at a single decimal point of percentage welfare change, we see no change for
the average for the world and almost no change for any of the ten regions. This extends the Costinot and
Rodriguez-Clare result, of very little impact of multiple factors of production, to the Krugman and Melitz
models. Our models in this section and the model of Costinot and Rodriguez-Clare incorporate
differences in factor intensities across counties. We believe, however, that further research is warranted
into the question of whether differences in factor endowments are adequately incorporated before we
conclude that Heckscher-Ohlin forces are unimportant.
The results without labor-leisure choice (columns 1-9) show the same pattern as in table 4. That
is, regarding the welfare impacts, we have M K A for seven of the ten regions and for the average
for the world. For the regions Australia-New Zealand, Low Income and Middle Income, we have that
A K M . The additional factors of production do not change the qualitative result.
One quantitative difference, however, is that we find larger welfare gains in table 5. Comparing
columns 1-3 of table 5 with columns 7-9 of table 4, the only difference is the presence of initial tariffs in
table 5. The initial tariffs imply there is less trade than is optimal for the world, so the reduction of iceberg
trade costs leads to larger gains from a second-best effect.
Labor-leisure choice, however, does reverse the qualitative result for Australia-New Zealand and
the Low Income regions regarding the ranking of Armington compared to the monopolistic competition
models. Without labor-leisure choice, these regions obtain larger estimated gain in the Armington model,
but prefer the outcomes of the monopolistic competition models with labor-leisure choice. This result is
25
consistent with the earlier result we saw in table 3, where with labor-leisure choice the availability of
more resources led to more varieties and more gains in the monopolistic competition models.
The values of our parameter RK / M for the Krugman model’s share of the larger welfare gains in
the Melitz model over the Armington model in the four sets of scenarios are: .83; .83; .83; .82. Again, we
observe that, for these global changes in iceberg costs, the selection effect contributes to additional
welfare gains, but the variety impact of the Krugman model is quantitatively more important than the
selection effect of the Melitz model.
5.5. Tariff and Iceberg Trade Costs Reduction in a Nine-sector, Ten Region Model
In this subsection, we compare the welfare effects in the Armington, Krugman and Melitz models
of global free trade in a more realistic model structure. We incorporate initial tariffs based on the data for
the first time, and, as in the models of tables 3-5, we have initial trade imbalances. We expand our models
to nine-sectors, four of which are always modeled as perfectly competitive Armington sectors, and five of
which are modeled as monopolistically competitive when we evaluate the Krugman and Melitz models.21
All versions of these models contain Cobb-Douglas demand for three primary factors of production:
labor, capital and a resource factor. We allow some of the resource factor to be sector specific factor in all
of these models. We assume CES demand for intermediates based on the data on intermediate shares by
sector with an elasticity of substitution equal to 0.5. We also evaluate the impact of uniform tariffs and
labor-leisure choice. For sensitivity purposes with the earlier simulations, we evaluate and compare a
reduction in iceberg trade costs in this model. We continue to hold the trade response constant across the
three market structures.
5.5.1 Global Free Trade in Tariffs
We evaluate a movement to global free trade in three model variants: (i) initial tariffs based on
the data; (ii) initial uniform tariffs in each country where the uniform tariff rate equalizes the revenue for
initial tariffs based on the data; and (iii) initial tariffs based on the data with labor-leisure choice.
Focusing first on the results for the average for the world, in all three model variants we again have the
ranking of welfare results as M K A.
When we start with initial uniform tariffs and move to global free trade, we see that the gains in
the Armington model drop to very small gains—to about .05% of consumption from .24% of
consumption with initial tariffs based on the data. The lower gains with uniform tariffs in the Armington
21
We aggregate the 57 sectors of the GTAP database into the following sectors (the first four of which are always
perfectly competitive sectors in this subsection): (i) agriculture; (ii) natural resources extraction; (iii) utilities; (iv) all
other services, not included elsewhere; (v) food; (vi) textiles; (vii) refined resource products; (viii) other
manufacturing, not elsewhere classified; and (ix) business services.
26
model are explained by the property that the Harberger triangles of distortion costs of the tariff increase
with the square of the tariff. Since the Harberger distortion costs of the tariffs are smaller with uniform
tariffs, the gains in the monopolistic competition models are proportionately much larger than the
Armington model, rising to 406 (467) percent of the gains in the Armington model in the Krugman
(Melitz) model.
The values of our parameter RK / M for the relative importance of the Krugman model to the Melitz
model are: (i) .95 for initial tariffs based on data; (ii) .83 for initial uniform tariffs; and (iii) .90 for initial
tariffs with labor-leisure choice. Again, we observe that the variety impact of the Krugman model is
quantitatively more important than the selection effect of the Melitz model, in this case for global changes
in tariffs.
Turning to individual regions, we see that in the Armington model, the Low Income and Middle-
Income regions experience a welfare loss from global free trade (columns 7, 10 and 13). Our terms-of-
trade parameter shows that these are the regions that experience the strongest terms-of-trade loss due to
global free trade.22 In the Armington model, the terms-of-trade loss dominates the efficiency gains for
these regions.
In the case of China (with global free trade and initial non-uniform tariffs), we have a welfare
ranking reversal compared with the aggregate, i.e., we have A K M . This occurs both with and
without labor-leisure choice (see table 6, columns 7-9, and columns 13-15). The reasons are very
interesting. First, unlike the usual case when Armington produces larger welfare gains than the
monopolistic competition models, our terms-of-trade parameter shows that there is a more favorable
change in the terms-of-trade in the monopolistic competition models. Thus, the larger welfare gains in the
Armington model cannot be explained by the terms-of-trade. Our Feenstra ratio parameter for China in
these cases, however, shows that there is a loss of the Dixit-Stiglitz variety externality in the monopolistic
competition models. In our previous models, all sectors were monopolistically competitive and all our
estimated Feenstra ratios were all greater than one.23 In this model, however, we have perfectly
competitive sectors as well as monopolistically competitive sectors, and it is possible for regions to shift
expenditures away from monopolistically competitive sectors toward perfectly competitive sectors. Any
such shift would entail a loss of the variety externality. Our Feenstra ratio shows that when the model has
perfectly competitive sectors as well as monopolistically competitive sectors, it is possible to experience a
22
The Mexico-Peru-Chile region also experiences a small loss from global free trade in the Armington model
without labor-leisure choice, and has a small negative terms-of-trade loss in these cases.
23
Although we do not observe it in any of simulations, it is an open question whether in a multi-region monopolistic
competition model a region could shift expenditures toward sectors that experience a variety loss, resulting a net loss
of the variety externality.
27
welfare loss due to a reduction of the Dixit-Stiglitz variety externality. To augment the intuition for this
interpretation, we note that value-added in monopolistically competitive sectors in China declines in these
scenarios, both in absolute amounts and in percentage terms relative to the Armington sectors in these
models.24
Then why does Krugman provide more welfare gains than Melitz in the case of China in these
model variants? We observe that the terms-of-trade parameter for China has the same value in the
Krugman and Melitz cases (columns 13 and 15; and columns 23 and 25 in table 6a). And the Feenstra
ratio is larger by .002 in the Melitz case (column 14 versus 16 and column 24 versus 26 in table 6a). Then
the ranking K M cannot be explained by either by the terms-of-trade effect or by the variety
externality. Given the possibilities for differences in these models, that leaves only a larger productivity
gain in the Krugman model compared with the Melitz model as a possible explanation. Melitz (2003)
noted the possibility of a reduction in the productivity of the economy after a reduction in trade costs due
to the selection effect, and our China case may be an example. But the model in Melitz (2003) is a one-
sector monopolistically competitive model. Here we have multiple sectors and we include some perfectly
competitive sectors. The intersectoral shift in resources would also entail some productivity gain from the
more efficient intersectoral allocation of resources, and the intersectoral shifts are not identical across the
models.
5.5.2 Iceberg Trade Cost Reduction in the Nine-Sector Model
For sensitivity, we compare the impact of a ten percent reduction in iceberg trade costs in the
four-sector models (columns 7-9 of table 5) with the nine-sector models (columns 1-3 of table 6). The
only differences between the models in these scenarios is: (i) the number of sectors; and (ii) in our nine-
sector models, we assume four of the sectors are perfectly competitive, even when we allow five sectors
to be monopolistically competitive. We retain the welfare ranking for the average for the world that
M K A . The main difference in the results is that, while results for the Armington model are the
same in the four and nine-sector models, the welfare gains in the monopolistic competition models are
smaller in the nine-sector model. The intuition is that a smaller share of the economy is monopolistically
competitive in the nine-sector models, so the variety gains are smaller. Inspection of the Feenstra ratios
verifies the smaller variety gains. Comparable results and interpretations apply when we allow labor-
leisure choice, i.e., compare columns 4-6 of table 6, with columns 10-12 of table 5. The values of the
24
In the case of labor-leisure choice, in our Krugman model, we find that value-added in the CRTS sectors increases
by 0.2% while value-added in the IRTS sectors falls by 0.1%. In the Melitz model, value added in the CRTS sectors
falls by 0.4%, but value-added in the Melitz sectors of the model falls by 1.3%. Without labor-leisure choice, value-
added declines in both CRTS and IRTS sectors, but the percentage decline is larger in absolute value in the IRTS
cases.
28
parameter RK / M for the relative importance of the Krugman model to the Melitz model in our iceberg cost
reduction scenarios are: .88 without labor-leisure choice and .87 with labor-leisure choice.
However, the welfare ranking of for the Low Income region switches depending on the number of
sectors. For the Low Income region, compare columns 1-3 table 6, where we have M K A , with
columns 7-9 of table 5, where in the four sector model we have A K M . The only difference is the
number of sectors in the model. We see that for the Low Income region, in the nine-sector model there are
larger terms-of-trade gains in the monopolistic competition models; we had the opposite terms-of-trade
ranking in the four-sector model. This shows that in multi-sector, multi-region models, how the data are
aggregated can impact the qualitative ranking of the welfare gains.
6. Sensitivity of the Welfare Results to the Trade Response and to
Unilateral versus Global Policy Scenarios
Given the acknowledged importance of the trade response, in this section we examine the
sensitivity of the results to a change in the trade response. We also examine a unilateral increase in tariffs
to forty percent by each of our ten regions (in ten separate simulations). As one of our most realistic
models, we examine sensitivity in our nine-sector, ten-region model with three primary factors of
production, labor-leisure choice and actual tariffs and trade imbalances based on the data in the
benchmark equilibrium.
6.1 Sensitivity to the Trade Response
We continue to determine the trade response in the Melitz model and adjust the Dixit-Stiglitz
elasticities in the Krugman model and Armington elasticities to match the trade response. The key
parameters that impact the trade elasticity in the Melitz model are the Pareto shape parameter and the
Dixit-Stiglitz elasticity. We conduct sensitivity on these two parameters separately.
For the high and low values of the Pareto shape parameter, we take plus and minus two standard
deviations from the mean of the preferred probability distribution estimated by Balistreri et al. (2011),
while we hold the Dixit-Stiglitz elasticity of substitution constant at our central value of 5.0. Regarding
sensitivity to the Dixit-Stiglitz elasticity, for a solution to the Melitz model, we must have a − 1 ,
where a is the Pareto shape parameter and is the Dixit-Stiglitz elasticity. Given our central value of
4.58 for the Pareto shape parameter and 5.0 for the Dixit-Stiglitz elasticity, we take plus and minus 0.5 for
the Dixit-Stiglitz elasticity in our sensitivity analysis for . Our results for sensitivity to the Pareto shape
parameter are in table 7A and for the Dixit-Stiglitz elasticity in table 7B. The values of the Pareto shape
parameter in the Melitz model, Dixit-Stiglitz elasticities in the Krugman model and Armington elasticities
are shown in the first two rows of table 7A and table 7B. The percent change in the trade response due to
29
a movement to global free trade is shown in the third row of table 7A and 7B. The results with central
trade response are the same as in table 6, columns 13-15.
We see that a larger trade response leads to larger welfare gains for the world, but the differences
from the estimates of the central elasticities are not large. Further, we do not see any cases of a sign
change in the welfare impacts: those regions that lose with the central trade response, lose with the high
and low trade responses. In the case of China, its losses increase in the Melitz model with a high trade
response.
6.2 Impact of a Unilateral Increase of Tariffs to Forty Percent
For a unilateral change in tariffs, in the absence of retaliation from trade partners, there are terms-
of-trade effects that lead to gains from the imposition of tariffs up to the optimal tariff for the region. In
all of our previous policy scenarios, we have imposed a simultaneous change of the policy on a global
basis. In a movement to global free trade, however, the terms of trade effects of tariffs could be positive
or negative for any particular region (our results in section 5.6 are an example). In this subsection, we
investigate the ranking of the impacts across the three market structures of the unilateral imposition of a
forty percent uniform tariff by each region of our model, starting from initial tariffs in the data. These
results illustrate two points: (i) with unilateral tariff increases, it is possible to have positive estimated
welfare gains in the Armington model and negative estimated welfare gains in the monopolistic
competition models. In fact, opposite signs are our typical result; and (ii) for losses from trade policy
changes, it is possible for the absolute value of the losses to be larger in the Melitz model than the
Krugman model.
For each of our ten regions we simulate a unilateral increase in tariffs to 40 percent and compare
results for the three market structures, were we adjust elasticities in the Krugman and Armington models
to match the trade response of the Melitz model. The results of the 30 simulations are presented in table
8, where we only present the results for the region that is imposing the increase it its tariffs to a uniform
40 percent. In the Armington model, we see that all countries gain from the unilateral imposition of a 40
percent uniform tariff. Both a theoretical derivation25 and our numerical simulations are consistent with
25
Intuition into the value of the optimal tariff in a multi-region Armington model can be obtained from the
following. Given product differentiation at the level of the region in the Armington model, to achieve marginal
revenue equal to marginal costs, a representative region’s optimal export tax for good i in destination region r is
− 1 where ir is the elasticity of demand for exports of the representative country of good i in
* =
tir ir ( ir − 1)
destination region r. It has been shown by Harrison, Rutherford and Tarr (1997a, appendix C) that
ir
mm
ir = ir
mm
− ir k , where ir
mm is country r’s elasticity of substitution for imports of good i from different import
regions, ir is the representative country’s share of the market of good i in region r and k is a positive const ant. The
= ir ir − 1 . When the market share is small or
mm mm
( ir − ir k ) ( ir − ir k − 1)
* mm mm
optimal export tax is then: tir ir
mm
is
30
the optimal tariff for these countries in the range of about 25 percent. Although the countries experience
a welfare loss by further increasing the tariff above 25 percent, the gains from increasing tariffs from
initial levels to the optimal level dominate the losses for the increase above the optimal level to 40
percent.
The welfare results in the models with monopolistically competitive sectors are strikingly
different—seven of the ten regions have the opposite sign for the welfare impact as in the Armington
model. Except for the Unites States, Australia-New Zealand and the Middle-Income regions, the other
seven regions lose from the imposition of 40 percent uniform tariffs. The intuition is that even small
increases in tariffs typically result in a loss of varieties and a welfare loss from the loss of the Dixit-
Stiglitz externality. And for these seven regions, the losses are larger in absolute value in the Melitz
model compared with the Krugman model. The Feenstra ratios in table 8 show there is a welfare loss due
to the loss of varieties in all of these seven regions, and the variety loss is larger in the Melitz model than
in the Krugman model.
In the cases of the United States and the Middle-Income region, the Feenstra ratios show that they
experience a loss of welfare from the variety loss, but the terms of trade gain dominates the welfare loss
from the loss of varieties. The interesting case to explain is the Australia-New Zealand region in the
model with Krugman sectors; we observe a gain in welfare from the varieties (Feenstra ratio larger than
100%), despite the increase in tariffs. Feenstra has explained that it is not the number of varieties alone
that is important to the variety externality, but also the expenditure share. We have four perfectly
competitive sectors in our model; a shift in expenditures toward the monopolistically competitive sectors
would increase the value of varieties; a higher expenditure share can offset the loss in the number of
varieties. To provide support for our result regarding the Feenstra ratio in the Krugman model, we
calculate the change in value added for the Australia-New Zealand region; we find a 5.7 percent decline
in the value-added of the perfectly competitive sectors and a 1.6 percent increase in the value added of the
monopolistically competitive sectors.
7. Conclusion
We have produced numerous new results for the relative welfare impacts of the Armington,
Krugman and Melitz models for a substantially expanded range of modeling features that have not
large, then the optimal tax may be approximated by
* mm ( mm − 1) − 1 = mm ( mm − 1) − 1 = (5 / 4) − 1 = .25
tir ir ir , where we have employed our assumption of a
value of 5.0 for the central value of the Armington elasticity of substitution for all sectors and regions. The optimal
export tax will increase with the market share. From Lerner symmetry, this market power can be exploited by an
import tariff.
31
previously been examined in the literature. Since our results are numerical, we quantify the relative
importance of the three market structures.
In response to global changes in trade costs, in all of our multi-sector models, we find that the
ranking of global welfare gains is M K A . While the Melitz model produces more welfare gains than
the Krugman model in response to global changes in trade costs, we find that the Krugman model
captures the majority of the increase in the gains above the Armington model. Regarding individual
regions, however, we find several cases of reversed welfare rankings. i.e., M K A . With unilateral
increases in tariffs, we typically find that the Armington model produces gains, while the monopolistic
competition models produce losses. We develop two parameters for each scenario (the terms-of-trade and
the Feenstra ratio) which are crucial for understanding the reasons for the relative welfare rankings. When
the welfare gains are larger in the Armington model, it is usually because we find that the terms-of-trade
effects for that region are greater in the Armington model. In those cases where the Krugman model
produces larger welfare gains than the Melitz model, we usually find that the variety externality is greater
in the Krugman model, reflecting a shift in expenditure shares toward sectors with smaller variety gains
or less of a variety increase in the Melitz model. If the models with monopolistic competition sectors also
contain some perfectly competitive sectors, the Feenstra ratio identifies cases where there is a variety loss
from global free trade due to expenditure shifts to the perfectly competitive sectors or a loss of varieties.
We conclude that for individual regions, the welfare ranking of the Armington, Krugman and Melitz
market structures is model, data, scenario and parameter dependent.
References
Adao, Rodrigo, Costas Arkolakis and Federico Esposito (2017), “Endogenous Labor Supply and the Gains
from International Trade,” 2017 Meeting Papers 1044, Society for Economic Dynamics. Available
at: https://ideas.repec.org/p/red/sed017/1044.html.
Arkolakis, Costas, Arnaud Costinot and Andres Rodriguez-Clare (2012), “New Trade Models, Same Old
Gains?” American Economic Review, 102(1), 94–130.
Armington, Paul (1969), “A Theory of Demand for Products Distinguished by Place of Production,”
International Monetary Fund Staff Papers, Vol. 16(1), 159-178.
Balistreri, Edward J., Russell H. Hillberry and Thomas F. Rutherford (2011), “Structural Estimation and
Solution of International Trade Models with Heterogeneous Firms,” Journal of International
Economics, Vol. 83(2), 95-108.
Balistreri, Edward J., Russell H. Hillberry and Thomas F. Rutherford (2010), “Trade and Welfare: Does
Industrial Organization Matter,” Economics Letters, 109(2), 85–87.
32
Balistreri, Edward J. and Thomas F. Rutherford (2013), “Computing general equilibrium theories of
monopolistic competition and heterogeneous ﬁrms,”’ in Handbook of Computable General
Equilibrium Modeling Vol. 1, Peter B. Dixon and Dale W. Jorgenson eds., Amsterdam, North
Holland, Elsevier, Chapter 23, 1513–1570.
Balistreri, Edward J., Thomas F. Rutherford and David G. Tarr (2009), “Modeling Services
Liberalization: The Case of Kenya,” Economic Modeling, Vol. 26 (3), May, 668-679.
Balistreri, Edward J., Zoryana Olekseyuk and David G. Tarr (2017), “Privatization and the Unusual Case
of Belarusian Accession to the WTO,” The World Economy, Vol. 40, 2564-2591.
Balistreri, Edward J., David G. Tarr and Hidemichi Yonezawa (2015), “Deep Integration in Eastern and
Southern Africa: What are the Stakes?” Journal of African Economies. Article first published
online July 28, 2015, doi:10.1093/jae/ejv012
Ballard, Charles (2000), “How Many Hours are in a Simulated Day? The Effects of Time Endowment on
the Results of Tax-Policy Simulation Models.” Available at:
https://pdfs.semanticscholar.org/59b3/4cbb9cc1c02f2c0735dec3e3bcb9ad632507.pdf
Bernard, A.B., S. J. Redding, and P. Schott (2007), “Comparative Advantage and Heterogeneous Firms,”
Review of Economic Studies, 74(1), 31–66.
Bhagwati, Jagdish (1082), “Directly Unproductive Profit-Seeking Activities,” Journal of Political
Economy, Vol. 90, No. 5, 988-1002.
Caliendo, Lorenzo, Robert C. Feenstra, John Romalis and Alan M. Taylor (2015), “Tariff Reductions, Entry
and Welfare: Theory and evidence for the last two decades,” Working Paper 21768, National
Bureau of Economic Research, December.
Chaney, T. (2008), “Distorted Gravity: The Intensive and Extensive Margins of International Trade,” The
American Economic Review, Vol. 98(4), 1707-1721.
Costinot, Arnaud and Andres Rodriguez-Clare (2014), “Trade Theory with Numbers: Quantifying the
Consequences of Globalization,” in Handbook of International Economics, Elhanan Helpman,
Kenneth Rogoff and Gita Gopinath (eds.), Vol. 4, 197-262, Amsterdam: Elsevier.
Costinot, Arnaud and Andres Rodriguez-Clare (2013), “Online Appendix to Trade Theory with Numbers:
Quantifying the Consequences of Globalization,” March. Available at:
https://economics.mit.edu/files/9215
Dekle, Robert, Jonathan Eaton and Samuel Kortum (2008), “Global Rebalancing with Gravity: Measuring
the Burden of Adjustment,” IMF Staff Papers, Vol. 55(3), 511-540.
Devarajan, Shantayanan and Sherman Robinson (2013), “Contribution of Computable General Equilibrium
Modeling to Policy Formulation in Developing Countries,” in Handbook of Computable General
Equilibrium Modeling,” Peter Dixon and Dale Jorgenson (eds.), Vol. I, chapter 5, 277-301,
Elsevier.
Dixon, Peter, Michael Jerie and Maureen Rimmer (2018), Trade Theory in Computable General
Equilibrium Models: Armington, Krugman and Melitz, Singapore: Springer.
33
Eaton, Jonathan and Samuel Kortum (2002), “Technology, Geography and Trade,” Econometrica, Vol.
70(5), 1741-1779.
Ethier, Wilfred (1982), “National and International Returns to Scale in the Modem Theory of International
Trade," American Economic Review, Vol. 72(2):389-405.
Ethier, Wilfred and James R. Markusen, (1996), “Multinationals, Technical Diffusion, and Trade,” Journal
of International Economics, Vol. 41, 1-28.
Feenstra, Robert C. (2010), “Measuring the Gains from Trade Under Monopolistic Competition,”
Canadian Journal of Economics,” Vol. 43(1), 1-28.
Feenstra, Robert C. (1994), “New Product Varieties and the Measurement of International Prices,”
American Economic Review, Vol. 84, 157-177.
Francois, Joseph, M. Manchin and Will Martin (2013), “Market Structure in CGE Models of International
Trade,” in Peter Dixon and Dale Jorgenson (eds.), Handbook of Computable General Equilibrium
Models, Amsterdam: North Holland, Elsevier.
Harris, Richard, 1984. "Applied General Equilibrium Analysis of Small Open Economies with Scale
Economies and Imperfect Competition," American Economic Review, Vol. 74(5), pages 1016-32.
Harrison, Glenn W., Thomas F. Rutherford and David G. Tarr (1997), "Quantifying the Uruguay Round,"
Economic Journal, Vol. 107, No. 444, September, 1405-1430.
Harrison, Glenn W., Thomas F. Rutherford and David G. Tarr (1997a), "Trade Policy Options for Chile: A
Quantitative Evaluation,” World Bank Working Paper No. 1783. Available at:
http://documents.worldbank.org/curated/en/538631468015839517/pdf/WP1783.pdf,
Hillberry, Russell and David Hummels (2013), “Trade Elasticity Parameters for a Computable General
Equilibrium Model,” in Handbook of Computable General Equilibrium Modeling, Vol. 1B, Peter
B. Dixon and Dale W. Jorgenson (eds.), Amsterdam: Elsevier.
Jafari, Yaghoob and Wolfgang Britz (2017), “Modelling Heterogeneous Firms and Non-Tariff Measures in
Free Trade Agreements Using Computable General Equilibrium,” Economic Modelling,
https://doi.org/10.1016/j.econmod.2018.04.004, in press.
Jensen, Jesper, Thomas F. Rutherford and David G. Tarr (2010), “Modeling Services Liberalization: The
Case of Tanzania,” Journal of Economic Integration, Vol. 25 (4), December, 644-675.
http://dx.doi.org/10.11130/jei.2010.25.4.644
Jensen, Jesper and David G. Tarr (2003), “Trade, Exchange Rate and Energy Pricing Reform in Iran:
Potentially Large Efficiency Effects and Gains to the Poor,” Review of Development Economics,
Vol. 7, Number 4, November, 543-562.
Krueger, Anne O. (1974), “The Political Economy of the Rent -Seeking Society,” American Economic
Review, Vol. 3, 291-303.
Krugman, Paul (1979), “Increasing Returns, Monopolistic Competition and International Trade,”
Journal of International Economics, 9(4), 469–479.
34
Krugman, Paul. (1980), “Scale Economies, Product Differentiation, and the Pattern of Trade,” The
American Economic Review, 70(5), 950–959.
Markusen, James R. (1989), "Trade in Producer Services and in Other Specialized Intermediate Inputs,"
American Economic Review, Vol. 79, 85-95.
Markusen, James R. (2002), Multinational Firms and the Theory of International Trade, Cambridge, MA.:
MIT Press.
Markusen, James R., Thomas F. Rutherford and David G. Tarr (2005), “Tra de and Direct Investment in
Producer Services and the Domestic Market for Expertise,” Canadian Journal of Economics, Vol
38 (3), 758-777.
Markusen, James R. and Anthony Venables (1998), “Multinational Firms and the New Trade Theory,”
Journal of International Economics Vol. 46, 183-204.
Melitz, Marc J. (2003): “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry
Productivity,” Econometrica, Vol. 71(6), 1695–1725
Melitz. Marc J. and Stephen J. Redding (2015), “New Trade Models, New Welfare Implications,” American
Economic Review, Vol. 105(3), 1105-1146.
Melo, Jaime de, and David G. Tarr (1990), "Welfare Costs of US Quotas in Textiles, Steel and Autos,"
Review of Economics and Statistics, August, 489-497.
Petri, Peter A., Michael G. Plummer, and Fan Zhai (2012), “Trans-Pacific Partnership and Asia-Pacific
Integration: A Quantitative Assessment,” Washington, DC: Peterson Institute of International
Economics.
Ricardo, David (1817), On the Principles of Political Economy and Taxation. Available at:
http://www.econlib.org/library/Ricardo/ricP.html.
Rutherford, Thomas F. (2002), “Lecture Notes on Constant Elasticity Functions.” Avaiable at:
http://www.gamsworld.org/mpsge/debreu/ces.pdf
Rutherford, Thomas F. and David G. Tarr (2002), “Trade Liberalization, product variety and growth in a
small open economy: a quantitative assessment,” Journal of International Economics, Vol. 56 (2),
March, 247-272.
Rutherford, Thomas F. and David G. Tarr (2008), “Poverty Effects of Russia’s WTO Accession: modeling
‘real households’ with endogenous productivity effects,” Journal of International Economics, Vol
75 (1), pp 131-150.
Samuelson, Paul A. (1939), “The Gains from International Trade,” Canadian Journal of Economics, Vol.
5(2), 195–205.
Sato, Kazuo (1976), “The Ideal Log-Change Index Number,” The Review of Economics and Statistics, Vol.
58(2), 223-228.
Vartia, Y.O. (1976), “Ideal log-change index numbers,” Scandinavian Journal of Statistics,
Vol. 3, 121–6
35
Zhai, Fan (2008), “Armington Meets Melitz: Introducing Firm Heterogeneity in a Global CGE Model of
Trade,” Journal of Economic Integration, Vol. 23(3). September, 575-604.
Table 1: Literature Results of the Ranking of the Armington, Krugman and Melitz
Models Welfare Gains from the Reduction in Trade Costs
Ranking of
Welfare Gains
Model Assumptions and Methodology Key: A = Armington; K = Krugman; M = Melitz*
1. Analytic Solutions
One Sector, No Intermediates, One Factor, Zero Trade Balance, Iceberg Trade costs A=K=M Arkolakis, Costinot and Rodriguez-Clare (2012)
with Labor-Leisure choice and positive elasticity of labor supply **
M A Balistreri, Hillberry and Rutherford (2010)
with upper bound on the Pareto Probablity Distribution of Productivity M K Melitz and Redding (2015)
***
2. Computer Simulations, Autarky Exercises, Numerical Point Estimates without parameter sensitivity
33 Regions, 31 Sectors with One Primaty Factor, Zero Trade Balance, Iceberg Trade Costs
Muliple sectors, no intermediates A K M Costinot and Rodriguez-Clare (2014, table 4.1)
multiple sectors, one aggregate intermediate good K A
M M>K>A Costinot and Rodriguez-Clare (2014, table 4.1)
3. Computer Simulations, Comparative Static Exercises, Numerical Point Estimates without parameter sensitivity
10 regions, 16 sectors, Zero Initial Tariffs, Zero Trade Balances, uniform tariffs, one Primary Factor
Losses from imposing a Global Uniform 40% Tariff in a 10 Region, 16 Sector Model ***
Multiple sectors, no intermediates K M A Costinot and Rodriguez-Clare (2014, table 4.3)
multiple sectors, one aggregate intermediate good M K A Costinot and Rodriguez-Clare (2014, table 4.3)
*
If and only if the Melitz model produces welfare gains preferred to (larger than) the Krugman model, then we
write: M K and similarly for other rankings. We write K M for the same welfare gains.
**
Computer simulations: If the elasticity of labor supply to the traded goods sector is negative (zero), the estimated
welfare gain under Melitz is less than (equal to) Armington.
***
We report the Costinot and Rodriguez-Clare (2014, p. 204) results based on their average for the world, although
their results for individual regions in their model differ in several cases. We also rank the models based on their
definition that the "gains" from international trade as the absolute value of the real income change associated with
the increase in trade costs...
36
Table 2: Equivalence of Welfare Results Across Market Structures in the Simplest
One Sector Model, with and without balanced trade*
Impacts of Global Ten Percent Reduction in Iceberg Trade Costs
Full Model Results are Hicksian Equivalent Variation as a Percent of Consumption
1 2 3 4 5 6 7 8 9
domestic Zero Trade Balance for all Countries Data Based Trade Balances for all Countries domestic
trade Arkolakis et al. Full Model Calculations Full Model Calculations trade
share formula*** Armington Krugman Melitz Armington Krugman Melitz share
Region rr ε = 4.58** σ(D,M) = 5.58** σ = 5.58** σ=5.0; a=4.58** σ(D,M) = 5.58** σ = 5.58** σ=5.0; a=4.58** rr
Australia-New Zealand 77.2% 3.0% 3.0% 3.0% 3.0% 3.1% 3.1% 3.1% 77.8%
Canada 70.1% 4.1% 4.1% 4.1% 4.1% 4.1% 4.1% 4.1% 70.1%
China 72.1% 3.7% 3.7% 3.7% 3.7% 4.0% 4.0% 4.0% 72.8%
Japan 83.0% 2.1% 2.1% 2.1% 2.1% 2.1% 2.1% 2.1% 83.0%
Mexico-Chile-Peru 67.0% 4.6% 4.6% 4.6% 4.6% 4.8% 4.8% 4.8% 67.3%
Low Income NEC 55.5% 6.4% 6.4% 6.4% 6.4% 5.8% 5.8% 5.8% 54.4%
Middle Income NEC 75.6% 3.2% 3.2% 3.2% 3.2% 3.3% 3.3% 3.3% 76.4%
OECD NEC 81.4% 2.3% 2.3% 2.3% 2.3% 2.3% 2.3% 2.3% 81.6%
Philippines 62.9% 5.1% 5.1% 5.1% 5.1% 4.6% 4.6% 4.6% 61.2%
United States 84.6% 1.9% 1.9% 1.9% 1.9% 1.7% 1.7% 1.7% 82.9%
average for the World 79.0% 2.7% 2.7% 2.7% 2.7% 2.7% 2.7%
* Model with One Sector, One Factor, No Intermediates, No Labor-Leisure Choice, Untruncated Pareto distribution
of productivity.
** ε is the elasticity of demand for imports with respect to variable trade costs, taken from gravity (here the absolute
value of the gravity estimate is used); σ(D, M) is the Armington elasticity of substitution between imports and
domestic goods; σ is the Dixit-Stiglitz elasticity of substitution between varieties in a sector; and "a" is the shape
parameter in the Pareto distribution..
*** Arkolakis et al. (2012) derive the proportional change in welfare in this simple model with zero trade balances
−1/
as: ˆ , and and are country j’s shares of global expenditure on home goods in
W = jj where jj jj jj jj jj
the initial and counterfactual equilibria, respectively (ε = the absolute value of the elasticity of demand for imports
−1/
with respect to variable trade costs from gravity). Since this is a decline in trade costs, jj 1 , and the
−1/
percentage gains from trade using the Arkolakis et al. (2012) formula are 100* G j where: G j = jj − 1 .
Source: Authors’ calculations.
37
Table 3: Breaking the Equivalence: Labor-Leisure Choice, Intermediates and
Multiple Sectors
Impacts of Global Ten Percent Reduction in Iceberg Trade Costs
Results are Hicksian Equivalent Variation as a Percent of Consumption
All scenarios include trade imbalances, but have zero tariffs and a single primary factor
1 2 3 4 5 6 7 8 9
Labor-Leisure Choice Cobb-Douglas Demand for Intermediates
Single Sector-No intermediates Single Sector Four Sectors
Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Region σ(D,M) = 5.58* σ = 5.58* σ=5.0; a=4.58* σ(D,M) = 5.58* σ = 5.58* σ=5.0; a=4.58* σ(D,M) = 6.12* σ = 5.64* σ=5.0; a=4.58*
Australia-New Zealand 3.1% 3.23% 3.26% 3.2% 4.2% 4.4% 3.59% 3.71% 3.66%
Canada 4.1% 4.32% 4.36% 4.1% 5.1% 5.3% 4.2% 6.3% 6.9%
China 4.0% 4.19% 4.23% 3.9% 6.7% 7.5% 3.5% 9.8% 12.5%
Japan 2.1% 2.25% 2.27% 2.3% 2.9% 3.1% 2.1% 4.3% 4.9%
Mexico-Chile-Peru 4.8% 5.02% 5.07% 4.7% 5.9% 6.1% 4.9% 6.4% 6.8%
Low Income NEC 5.7% 6.10% 6.15% 5.7% 7.4% 7.7% 6.5% 5.8% 5.5%
Middle Income NEC 3.3% 3.49% 3.52% 3.4% 4.4% 4.6% 3.7% 3.7% 3.6%
OECD NEC 2.3% 2.42% 2.44% 2.4% 3.2% 3.4% 2.3% 3.8% 4.2%
Philippines 4.6% 4.91% 4.95% 4.7% 5.9% 6.1% 4.8% 7.2% 7.8%
United States 1.7% 1.85% 1.86% 1.9% 2.3% 2.4% 1.9% 2.8% 3.0%
average for the World 2.69% 2.85% 2.87% 2.79% 3.76% 3.98% 2.80% 4.34% 4.83%
* σ(D,M) is the Armington elasticity of substitution between imports and domestic goods; σ is the Dixit -Stiglitz
elasticity of substitution between varieties in a sector; and "a" is the shape parameter in the Pareto distribution.
Source: Authors’ estimates.
Table 3a: Parameters for Table 3--the Feenstra ratio and Proportional Changes in the Terms-of-trade
(TOT).*
1 2 3 4 5 7 8 9 10 11 12 13 14 15 16
Labor-Leisure Choice Cobb-Douglas Demand for Intermediates Cobb-Douglas Demand for Intermediates
Single Sector-No intermediates Single Sector Four Sectors
Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Region TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra
Australia-New Zealand 1.004 1.004 1.030 1.004 1.031 1.005 1.005 1.019 1.005 1.020 1.021 1.013 1.016 1.013 1.014
Canada 1.007 1.008 1.040 1.007 1.042 1.003 1.004 1.025 1.004 1.027 1.009 1.020 1.029 1.021 1.037
China 1.008 1.009 1.037 1.009 1.039 1.003 1.006 1.020 1.007 1.024 0.993 1.003 1.030 1.004 1.042
Japan 0.996 0.996 1.023 0.996 1.023 0.999 0.998 1.015 0.998 1.015 0.990 0.999 1.021 0.999 1.026
Mexico-Chile-Peru 1.011 1.012 1.044 1.012 1.046 1.007 1.009 1.028 1.008 1.030 1.012 1.013 1.030 1.012 1.033
Low Income NEC 1.007 1.009 1.059 1.009 1.061 0.998 1.001 1.037 1.001 1.039 1.016 1.009 1.028 1.009 1.025
Middle Income NEC 1.006 1.006 1.032 1.006 1.033 1.005 1.006 1.020 1.005 1.021 1.014 1.003 1.018 1.002 1.013
OECD NEC 0.996 0.996 1.025 0.996 1.025 0.998 0.997 1.015 0.997 1.016 0.993 0.997 1.018 0.997 1.020
Philippines 1.001 1.002 1.051 1.002 1.051 0.996 0.997 1.033 0.997 1.034 0.997 1.006 1.037 1.006 1.042
United States 0.983 0.983 1.023 0.983 1.022 0.987 0.984 1.015 0.985 1.014 0.987 0.986 1.017 0.987 1.017
average for the World
*See section 5 for a definition of these parameters. We present the inverse of the Feenstra ratio, so an increase is a
decline in the cost of a unit of utility due to variety increases. In the Armington model, the Feenstra ration = 1 in all
cases, so is not presented.
Source: Authors estimates
38
Table 4: Impacts of Intermediate Modeling Structure with Four-sectors with a
Global Ten Percent Reduction in Iceberg Trade Costs*
Results are Hicksian Equivalent Variation as a Percent of Consumption
1 2 3 4 5 6 7 8 9
domestic Cobb-Douglas Demand Cobb-Douglas Demand for Elasticity of Substition for
trade Single Composite Intermediate Intermediates ; real data Intermediates = 0.5; real data
share Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Region rr σ(D,M) = 5.86* σ = 5.59* σ=5.0; a=4.58* σ(D,M) = 6.13* σ = 5.65*σ=5.0; a=4.58*σ(D,M) = 5.96* σ = 5.61* σ=5.0; a=4.58*
Australia-New Zealand 89.3% 3.5% 3.7% 3.7% 3.6% 3.71% 3.66% 3.6% 2.9% 2.8%
Canada 84.4% 4.3% 5.3% 5.5% 4.2% 6.3% 6.9% 4.2% 4.5% 4.6%
China 91.1% 3.6% 7.8% 9.1% 3.5% 9.7% 12.5% 3.4% 4.9% 5.2%
Japan 91.6% 2.1% 3.4% 3.6% 2.1% 4.3% 4.9% 2.0% 2.8% 3.0%
Mexico-Chile-Peru 82.9% 4.8% 6.0% 6.2% 4.9% 6.4% 6.8% 4.8% 5.0% 5.0%
Low Income NEC 76.9% 6.3% 6.2% 6.1% 6.5% 5.8% 5.5% 6.5% 4.7% 4.4%
Middle Income NEC 88.3% 3.6% 3.8% 3.9% 3.7% 3.7% 3.6% 3.6% 2.9% 2.7%
OECD NEC 91.4% 2.3% 3.3% 3.6% 2.3% 3.8% 4.2% 2.3% 2.7% 2.7%
Philippines 79.2% 4.7% 6.5% 6.8% 4.8% 7.2% 7.8% 4.7% 5.3% 5.4%
United States 90.6% 1.9% 2.4% 2.5% 1.9% 2.8% 3.0% 1.9% 2.0% 2.1%
average for the World 90.0% 2.77% 3.83% 4.09% 2.80% 4.34% 4.83% 2.75% 2.93% 2.96%
*All scenarios with four IRTS sectors and trade imbalances included; but single primary factor and zero tariffs.
Source: Authors’ calculations.
Table 4a: --the Feenstra ratio and Proportional Changes in the Terms-of-trade*
1 2 3 4 5 7 8 9 10 11 12 13 14 15 16
Cobb-Douglas Intermediate Cobb-Douglas Demand for Intermediates Elasticity of Substition for Intermediates = 0.5
Single Composite Intermediate real data real data
Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Region TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra
Australia-New Zealand 1.0186 1.0110 1.0160 1.0112 1.0157 1.0209 1.0129 1.0159 1.0134 1.0139 1.0220 1.0113 1.0134 1.0112 1.0119
Canada 1.0103 1.0117 1.0251 1.0116 1.0272 1.0091 1.0199 1.0290 1.0207 1.0366 1.0108 1.0160 1.0236 1.0160 1.0270
China 0.9940 1.0020 1.0242 1.0024 1.0307 0.9928 1.0025 1.0303 1.0036 1.0424 0.9925 1.0000 1.0195 1.0000 1.0229
Japan 0.9927 0.9984 1.0166 0.9981 1.0188 0.9897 0.9991 1.0209 0.9986 1.0265 0.9892 0.9990 1.0160 0.9989 1.0191
Mexico-Chile-Peru 1.0117 1.0122 1.0279 1.0120 1.0302 1.0120 1.0126 1.0304 1.0123 1.0329 1.0129 1.0120 1.0256 1.0117 1.0266
Low Income NEC 1.0141 1.0089 1.0297 1.0090 1.0297 1.0163 1.0092 1.0275 1.0092 1.0255 1.0181 1.0081 1.0240 1.0078 1.0230
Middle Income NEC 1.0125 1.0047 1.0179 1.0046 1.0163 1.0144 1.0030 1.0176 1.0025 1.0134 1.0149 1.0039 1.0147 1.0037 1.0116
OECD NEC 0.9937 0.9967 1.0159 0.9967 1.0172 0.9931 0.9968 1.0181 0.9968 1.0204 0.9925 0.9976 1.0143 0.9978 1.0155
Philippines 0.9971 1.0049 1.0342 1.0049 1.0382 0.9973 1.0056 1.0372 1.0056 1.0424 0.9982 1.0046 1.0309 1.0047 1.0339
United States 0.9871 0.9865 1.0151 0.9867 1.0148 0.9869 0.9865 1.0172 0.9866 1.0174 0.9866 0.9875 1.0141 0.9879 1.0140
average for the World
*See section 5 for a definition of these parameters. In the Armington model, the Feenstra ration = 1 in all cases.
Source: Authors estimates
39
Table 5: Impacts of Global Ten Percent Reduction in Iceberg Trade Costs with Multiple
Primary Factors, Sector-Specific Factors and Labor-Leisure Choice
Results are Hicksian Equivalent Variation as a Percent of Consumption
All scenarios with four IRTS sectors; trade imbalances and initial tariffs incorporated; intermediates with CES demand**
1 2 3 4 5 6 7 8 9 10 11 12
Labor as the single factor Three mobile factors Three mobile factors Three mobile factors
Sector Specific Factor: No Sector Specific Factor: No Sector Specific Factor: Yes Sector Specific Factor: Yes
Labor-Leisure Choice: No Labor-Leisure Choice: No Labor-Leisure Choice: No Labor-Leisure Choice: Yes
Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Region σ(D,M) = 5.98* σ = 5.60* σ=5.0; a=4.58* σ(D,M) = 5.97* σ = 5.60* σ=5.0; a=4.58* σ = 5.81* σ = 5.59* σ=5.0; a=4.58* σ = 5.81* σ = 5.59* σ=5.0; a=4.58*
Australia-New Zealand 3.8% 3.4% 3.3% 3.8% 3.4% 3.3% 3.8% 3.6% 3.5% 3.74% 3.87% 3.88%
Canada 4.4% 5.1% 5.3% 4.4% 5.1% 5.3% 4.3% 5.0% 5.2% 4.3% 5.5% 5.7%
China 4.0% 6.5% 7.2% 4.0% 6.5% 7.1% 3.9% 6.1% 6.5% 3.9% 6.7% 7.4%
Japan 2.2% 3.4% 3.7% 2.2% 3.4% 3.6% 2.2% 3.2% 3.4% 2.2% 3.4% 3.7%
Mexico-Chile-Peru 5.0% 5.5% 5.7% 5.0% 5.5% 5.6% 5.0% 5.4% 5.5% 4.9% 5.7% 5.9%
Low Income NEC 7.9% 6.7% 6.5% 7.9% 6.7% 6.5% 7.8% 7.4% 7.4% 7.8% 8.0% 8.1%
Middle Income NEC 4.3% 3.8% 3.6% 4.3% 3.8% 3.6% 4.2% 4.0% 3.9% 4.2% 4.2% 4.2%
OECD NEC 2.5% 3.2% 3.4% 2.5% 3.2% 3.4% 2.5% 3.0% 3.1% 2.5% 3.3% 3.4%
Philippines 5.0% 6.1% 6.3% 5.0% 6.0% 6.3% 5.0% 5.8% 6.0% 5.0% 6.2% 6.5%
United States 2.0% 2.3% 2.4% 2.0% 2.3% 2.4% 1.9% 2.3% 2.3% 2.0% 2.5% 2.6%
average for the World 3.08% 3.63% 3.74% 3.07% 3.62% 3.73% 3.03% 3.53% 3.63% 3.03% 3.82% 3.99%
* σ(D,M) is the Armington elasticity of substitution between imports and domestic goods; σ is the Dixit-Stiglitz
elasticity of substitution between varieties in a sector; “a” is the shape parameter in the Pareto distribution.
**CES demand for intermediates with elasticity of substitution = 0.5.
Source: Authors’ estimates.
Table 5a: the Feenstra ratio and Proportional Changes in the Terms-of-trade*
1 2 3 4 5 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Labor as the Single Factor Three Mobile Factors Three Mobile Factors Three Mobile Factors
Sector Specific Factor: No Sector Specific Factor: No Sector Specific Factor: Yes Sector Specific Factor: Yes
Labor-Leisure Choice: No Labor-Leisure Choice: No Labor-Leisure Choice: No Labor-Leisure Choice: Yes
Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Region TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra
Australia-New Zealand 1.022 1.011 1.013 1.011 1.012 1.022 1.011 1.013 1.011 1.012 1.023 1.014 1.014 1.014 1.013 1.022 1.014 1.015 1.014 1.015
Canada 1.011 1.016 1.024 1.016 1.027 1.011 1.016 1.024 1.016 1.027 1.011 1.015 1.023 1.015 1.026 1.011 1.015 1.025 1.015 1.029
China 0.992 1.000 1.019 1.000 1.023 0.992 1.000 1.019 1.000 1.023 0.992 0.998 1.018 0.998 1.021 0.992 0.998 1.020 0.998 1.024
Japan 0.988 0.998 1.016 0.998 1.019 0.988 0.998 1.016 0.998 1.019 0.989 0.997 1.015 0.996 1.018 0.989 0.996 1.016 0.996 1.019
Mexico-Chile-Peru 1.013 1.012 1.025 1.012 1.027 1.013 1.012 1.025 1.012 1.026 1.013 1.013 1.025 1.012 1.026 1.013 1.013 1.026 1.012 1.028
Low Income NEC 1.020 1.010 1.024 1.010 1.025 1.020 1.010 1.024 1.010 1.025 1.021 1.017 1.027 1.017 1.029 1.020 1.018 1.029 1.018 1.032
Middle Income NEC 1.016 1.005 1.014 1.005 1.012 1.016 1.005 1.014 1.005 1.012 1.016 1.008 1.015 1.008 1.014 1.016 1.008 1.016 1.008 1.015
OECD NEC 0.992 0.997 1.014 0.998 1.015 0.992 0.997 1.014 0.998 1.015 0.992 0.996 1.014 0.996 1.015 0.992 0.995 1.015 0.995 1.016
Philippines 0.998 1.005 1.031 1.005 1.034 0.998 1.005 1.031 1.005 1.034 0.998 1.003 1.030 1.003 1.032 0.998 1.003 1.032 1.003 1.035
United States 0.986 0.987 1.014 0.987 1.014 0.986 0.987 1.014 0.987 1.014 0.986 0.986 1.014 0.986 1.013 0.986 0.986 1.015 0.986 1.015
average for the World
*See section 5 for a definition of these parameters. In the Armington model, the Feenstra ration = 1 in all cases.
Source: Authors estimates
40
Table 6: Impact of Global Free Trade, Uniform Tariffs and Iceberg Cost Reduction
in a More Realistic Nine-Sector Model.
All scenarios with: 4 CRTS and 5 IRTS sectors; Ten Regions; initial tariffs and trade
imbalances; three factors of production where part of the third factor is sector specific; and
intermediates with CES demand and elasticity of substitution = 0.5.
Results are Hicksian equivalent variation as a percent of consumption.**
1 2 3 4 5 6 8 9 10 7 11 12 13 14 15
Global Free Trade
10% reduction in iceberg costs 10% reduction in iceberg costs Global Free Trade Global Free Trade
Uniform Tariff: No Uniform tariff: Yes Uniform Tariff: No
Labor-Leisure Choice: No Labor-Leisure Choice: Yes Labor-Leisure Choice: No Labor-Leisure Choice: No Labor-Leisure Choice: Yes
Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Region σ(D,M) = 6.03* σ = 5.55* σ=5.0; a=4.58*σ(D,M) = 6.04* σ = 5.55* σ=5.0; a=4.58* σ = 8.00* σ = 5.81* σ=5.0; a=4.58* σ = 6.88* σ = 5.88* σ=5.0; a=4.58* σ = 8.03* σ = 5.83* σ=5.0; a=4.58*
Australia-New Zealand 3.3% 2.8% 2.9% 3.3% 2.9% 3.0% 0.6% 0.8% 0.9% 0.2% 0.1% 0.1% 0.6% 0.9% 1.0%
Canada 4.2% 4.8% 4.9% 4.2% 5.0% 5.1% 0.2% 0.4% 0.5% 0.1% 0.3% 0.3% 0.2% 0.5% 0.6%
China 4.0% 4.8% 4.6% 4.0% 5.2% 5.2% 0.3% -1.4% -2.4% 0.0% 0.5% 0.4% 0.3% -1.2% -2.2%
Japan 2.3% 3.0% 3.2% 2.3% 3.1% 3.3% 0.4% 1.2% 1.5% 0.1% 0.3% 0.4% 0.4% 1.3% 1.6%
Mexico-Chile-Peru 4.7% 4.9% 4.9% 4.6% 5.0% 5.1% -0.1% 0.0% 0.0% 0.2% 0.3% 0.3% -0.1% 0.0% 0.1%
Low Income NEC 7.1% 10.4% 11.0% 7.1% 11.6% 12.7% -0.2% 0.1% 0.2% -0.1% 1.4% 2.4% -0.2% 0.6% 0.9%
Middle Income NEC 3.7% 3.7% 3.7% 3.7% 3.7% 3.8% -0.2% 0.0% 0.1% -0.2% -0.1% -0.1% -0.2% 0.0% 0.2%
OECD NEC 2.6% 3.0% 3.1% 2.6% 3.2% 3.3% 0.4% 1.1% 1.2% 0.1% 0.4% 0.4% 0.4% 1.2% 1.4%
Philippines 5.2% 6.2% 6.4% 5.2% 6.3% 6.6% 0.1% 0.5% 0.7% 0.3% 0.6% 0.7% 0.1% 0.5% 0.7%
United States 2.0% 2.2% 2.2% 2.1% 2.3% 2.4% 0.4% 0.4% 0.5% 0.1% 0.1% 0.1% 0.4% 0.4% 0.5%
average for the World 2.96% 3.31% 3.36% 2.96% 3.47% 3.55% 0.24% 0.40% 0.41% 0.05% 0.22% 0.25% 0.24% 0.48% 0.51%
* See table 3 for definitions of elasticities. Source: Authors’ estimates
Table 6a: the Feenstra ratio and Proportional Changes in the Terms-of-trade*
1 2 3 4 5 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
10% Reduction in Iceberg Costs 10% Reduction in Iceberg Costs Global Free Trade Global Free Trade Global Free Trade
Uniform Tariff: No Uniform Tariff: Yes Uniform Tariff: No
Labor-Leisure Choice: No Labor-Leisure Choice: Yes Labor-Leisure Choice: No Labor-Leisure Choice: No Labor-Leisure Choice: Yes
Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Region TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra TOT TOT Feenstra TOT Feenstra
Australia-New Zealand 1.014 1.008 1.010 1.007 1.012 1.015 1.010 1.010 1.009 1.013 1.015 1.020 1.000 1.020 1.002 1.007 1.007 1.000 1.006 1.000 1.016 1.021 1.000 1.021 1.002
Canada 1.006 1.014 1.019 1.014 1.022 1.006 1.015 1.020 1.015 1.023 0.999 1.000 1.001 1.000 1.001 1.003 1.004 1.001 1.003 1.001 0.999 1.000 1.001 1.000 1.001
China 0.997 1.001 1.010 1.001 1.011 0.996 1.000 1.012 1.000 1.013 1.003 1.011 0.993 1.011 0.995 0.998 0.999 1.001 0.999 1.001 1.003 1.010 0.994 1.010 0.996
Japan 0.998 1.007 1.010 1.008 1.011 0.998 1.006 1.010 1.007 1.012 1.011 1.023 1.002 1.025 1.004 1.005 1.006 1.001 1.006 1.001 1.010 1.022 1.003 1.024 1.005
Mexico-Chile-Peru 1.009 1.017 1.018 1.017 1.023 1.009 1.018 1.019 1.017 1.024 0.995 0.995 0.999 0.995 1.000 1.005 1.005 1.000 1.005 1.001 0.995 0.995 1.000 0.995 1.000
Low Income NEC 1.011 1.044 1.021 1.045 1.050 1.012 1.050 1.023 1.053 1.058 0.977 0.983 1.000 0.978 1.008 0.986 1.017 0.995 1.027 1.021 0.977 0.987 1.001 0.984 1.014
Middle Income NEC 1.005 1.002 1.013 1.001 1.013 1.006 1.003 1.014 1.002 1.014 0.985 0.979 1.001 0.978 1.000 0.991 0.991 1.000 0.991 1.000 0.985 0.979 1.001 0.979 1.000
OECD NEC 0.998 0.997 1.010 0.997 1.010 0.998 0.996 1.011 0.996 1.011 1.003 1.005 1.003 1.006 1.003 1.005 1.004 1.001 1.003 1.001 1.003 1.005 1.003 1.006 1.003
Philippines 1.002 1.019 1.023 1.019 1.031 1.002 1.018 1.024 1.019 1.032 0.996 0.996 1.002 0.997 1.002 1.006 1.010 1.001 1.010 1.003 0.996 0.996 1.002 0.997 1.002
United States 0.991 0.983 1.011 0.984 1.010 0.990 0.983 1.012 0.983 1.010 1.016 1.010 1.000 1.010 1.000 1.006 1.003 1.000 1.002 1.000 1.016 1.010 1.000 1.009 1.000
*See section 5 for a definition of these parameters. In the Armington model, the Feenstra ration = 1 in all cases.
Source: Authors estimates
41
Table 7: Sensitivity of the Global Free Trade Welfare Results to a Common Change
in the Trade Response*
All results in the Model with Nine-Sectors, Three Primary Factors with Labor-Leisure Choice and
Trade Imbalances. Results are Hicksian Equivalent Variation as a Percent of Consumption
7A: Change in the Pareto Shape Parameter**
Low trade response Central trade response High trade response
Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Pareto shape parameter (a) 4.13 4.58 5.03
Armington or Dixit-Stiglitz elasticity 7.53 5.2 5 8.03 5.83 5 8.51 6.38 5
change in global trade share 1.82% 1.82% 1.82% 1.99% 1.99% 1.99% 2.15% 2.15% 2.15%
Region
Australia-New Zealand 0.5% 0.9% 1.0% 0.6% 0.9% 1.0% 0.6% 0.9% 1.1%
Canada 0.2% 0.5% 0.5% 0.2% 0.5% 0.6% 0.2% 0.4% 0.6%
China 0.3% -1.6% -1.9% 0.3% -1.2% -2.2% 0.3% -0.9% -2.4%
Japan 0.4% 1.4% 1.5% 0.4% 1.3% 1.6% 0.5% 1.3% 1.7%
Mexico-Chile-Peru -0.06% 0.0% 0.0% -0.05% 0.0% 0.1% -0.05% 0.0% 0.1%
Low Income NEC -0.2% 0.9% 0.9% -0.2% 0.6% 0.9% -0.1% 0.5% 0.8%
Middle Income NEC -0.2% 0.1% 0.1% -0.2% 0.0% 0.2% -0.2% 0.0% 0.2%
OECD NEC 0.4% 1.3% 1.3% 0.4% 1.2% 1.4% 0.5% 1.2% 1.4%
Philippines 0.1% 0.6% 0.7% 0.1% 0.5% 0.7% 0.1% 0.4% 0.7%
United States 0.4% 0.4% 0.4% 0.4% 0.4% 0.5% 0.4% 0.4% 0.5%
average for the World 0.224% 0.477% 0.484% 0.242% 0.478% 0.505% 0.260% 0.482% 0.526%
*The trade response is the change in the global trade share, defined above as 1 − W . It is calculated in the Melitz
model; then the Dixit-Stiglitz and Armington elasticities are adjusted in the Krugman and Armington models such
that the trade responses are equal in the three market structures.
**For the high and low value of the Pareto shape parameter, we take plus and minus two standard deviations from
the mean of the preferred distribution estimated by Balistreri et al. (2011).
7B. Change in the Dixit-Stiglitz Elasticity of Substitution
High trade response Central trade response Low trade response
Armington Krugman Melitz Armington Krugman Melitz Armington Krugman Melitz
Pareto shape parameter (a) 4.58 4.58 4.58
Armington or Dixit-Stiglitz elasticity 8.25 6.08 4.5 8.03 5.83 5 7.86 5.63 5.5
change in global trade share 2.06% 2.06% 2.06% 1.99% 1.99% 1.99% 1.93% 1.93% 1.93%
Region
Australia-New Zealand 0.6% 0.9% 1.2% 0.6% 0.9% 1.0% 0.6% 0.9% 0.9%
Canada 0.2% 0.5% 0.7% 0.2% 0.5% 0.6% 0.2% 0.5% 0.5%
China 0.3% -1.0% -3.8% 0.3% -1.2% -2.2% 0.3% -1.3% -1.4%
Japan 0.4% 1.3% 1.9% 0.4% 1.3% 1.6% 0.4% 1.4% 1.4%
Mexico-Chile-Peru -0.05% 0.0% 0.1% -0.05% 0.0% 0.1% -0.06% 0.0% 0.0%
Low Income NEC -0.2% 0.6% 0.2% -0.2% 0.6% 0.9% -0.2% 0.7% 0.7%
Middle Income NEC -0.2% 0.0% 0.4% -0.2% 0.0% 0.2% -0.2% 0.0% 0.1%
OECD NEC 0.4% 1.2% 1.6% 0.4% 1.2% 1.4% 0.4% 1.2% 1.2%
Philippines 0.1% 0.5% 1.0% 0.1% 0.5% 0.7% 0.1% 0.5% 0.5%
United States 0.4% 0.4% 0.6% 0.4% 0.4% 0.5% 0.4% 0.4% 0.4%
average for the World 0.251% 0.479% 0.510% 0.242% 0.478% 0.505% 0.236% 0.477% 0.481%
Source: Authors’ Estimates
42
Table 8: Impact of a Unilateral Increase in the Tariff Rates to a Uniform Forty
Percent, Starting from Initial Tariffs.
All results are for the region raising the tariffs*
Model is the Nine Sector, Three Factor Model with Labor-Leisure Choice and Trade Imbalances
Hicksian Equivalent Variation
as a percent of consumption Terms of Trade Changes Feenstra ratios
Region Armington Krugman Melitz Armington Krugman Melitz Krugman Melitz
Australia-New Zealand 0.34% 1.03% 1.03% 113.4% 114.4% 114.9% 100.2% 99.7%
Canada 0.80% -0.71% -1.01% 116.5% 114.5% 114.6% 99.4% 98.3%
China 0.91% -4.12% -4.72% 116.2% 113.5% 113.3% 98.6% 98.3%
Japan 0.41% -1.52% -1.79% 115.5% 113.6% 113.6% 99.2% 98.9%
Mexico-Chile-Peru 1.17% -0.46% -0.74% 115.7% 112.5% 112.8% 99.6% 98.2%
Low Income NEC 0.16% -0.21% -0.33% 111.6% 110.7% 111.0% 99.9% 98.6%
Middle Income NEC 0.82% 0.61% 0.55% 114.8% 114.7% 114.9% 99.9% 99.6%
OECD NEC 0.66% -0.79% -1.03% 117.4% 116.3% 116.3% 99.4% 99.3%
Philippines 0.86% -1.97% -2.30% 119.6% 114.9% 115.3% 99.0% 97.6%
United States 1.88% 0.51% 0.40% 125.8% 124.1% 124.1% 99.4% 99.3%
*Ten separate simulations for each of the three market structures, with only the results for the region
raising the tariffs shown.
Source: Authors’ estimates.
43
Table 9: Benchmark Tariff Rates, by Product and Region of the Model
Exports From:
Australia- Canada China Japan Mexico- Low Middle OECD Philippines USA
New Peru- Income Income nec
Zealand Chile nec nec
Imports into*:
Australia-New Zealand
Agriculture NA 0.78% 15.61% 10.06% 5.79% 9.31% 4.04% 11.73% 0.11% 0.62%
Food NA 15.34% 6.22% 31.80% 15.18% 16.83% 11.89% 39.24% 0.91% 3.78%
Natural Resources NA 0.00% 0.02% 0.01% 2.08% 1.64% 1.93% 0.16% 2.93% 0.02%
Manufactured Goods NA 0.44% 4.08% 0.27% 4.47% 6.52% 4.88% 1.48% 0.96% 0.08%
Refined Resource Products NA 0.33% 1.31% 0.64% 2.42% 5.98% 3.70% 1.50% 0.53% 0.20%
Textiles and Apparel NA 6.48% 4.56% 5.33% 6.04% 13.44% 6.37% 3.39% 2.23% 3.26%
Canada
Agriculture 0.15% NA 4.20% 9.93% 0.39% 8.33% 12.32% 6.36% 3.16% 0.00%
Food 0.99% NA 8.57% 40.55% 5.74% 16.04% 16.56% 21.26% 23.02% 1.69%
Natural Resources NA 0.37% 0.07% 0.02% 0.06% 1.91% 0.10% 2.97%
Manufactured Goods 2.11% NA 2.97% 0.03% 0.05% 9.25% 5.10% 1.53% 4.00%
Refined Resource Products 1.21% NA 2.48% 1.21% 0.08% 8.42% 3.88% 0.89% 4.05%
Textiles and Apparel 5.52% NA 7.34% 9.87% 0.47% 16.92% 11.50% 7.69% 8.52% 0.00%
China
Agriculture 0.78% 1.58% NA 14.79% 9.20% 15.39% 8.47% 43.69% 1.10% 1.00%
Food 1.84% 4.36% NA 11.38% 14.25% 15.60% 13.35% 18.56% 3.18% 2.83%
Natural Resources 0.08% 0.07% NA 1.36% 4.42% 11.27% 1.40% 2.49% 0.01% 0.29%
Manufactured Goods 2.18% 0.92% NA 0.03% 3.14% 11.30% 5.94% 1.75% 1.16% 0.85%
Refined Resource Products 3.55% 1.75% NA 0.36% 5.43% 12.89% 6.86% 2.97% 0.48% 2.17%
Textiles and Apparel 7.85% 13.82% NA 10.08% 10.09% 17.21% 14.53% 9.90% 1.57% 11.52%
Japan
Agriculture 0.31% 0.78% 3.05% NA 0.04% 7.75% 12.91% 5.88% 1.42% 2.72%
Food 1.84% 6.59% 5.12% NA 7.74% 14.77% 15.01% 25.13% 3.32% 3.47%
Natural Resources 0.31% 0.15% 2.57% NA 0.74% 8.29% 2.00% 6.73% 1.71% 0.36%
Manufactured Goods 14.34% 2.86% 6.85% NA 2.54% 13.42% 7.60% 3.76% 1.99% 1.00%
Refined Resource Products 1.60% 1.90% 5.26% NA 2.96% 8.09% 5.31% 3.52% 2.00% 1.96%
Textiles and Apparel 4.26% 3.76% 7.85% NA 3.44% 12.23% 7.90% 6.59% 0.17% 5.72%
Mexio-Peru-Chile
Agriculture 0.11% 0.00% 3.36% 1.02% NA 9.41% 4.93% 3.33% 7.02% 0.07%
Food 0.50% 5.91% 3.36% 16.72% NA 14.07% 6.80% 6.07% 4.19% 0.35%
Natural Resources 0.06% 0.02% NA 1.83% 0.87% 0.09% 2.97%
Manufactured Goods 6.06% 0.00% 7.63% 0.00% NA 6.35% 2.60% 0.33% 2.63%
Refined Resource Products 0.36% 0.40% 0.49% NA 7.43% 2.12% 0.17% 3.64% 0.06%
Textiles and Apparel 5.60% 0.00% 5.68% 5.18% NA 9.90% 3.10% 0.51% 2.04% 0.00%
Low Income Countries, nec
Agriculture 0.01% 0.11% 5.50% 0.62% 12.79% NA 5.79% 1.94% 6.72% 1.91%
Food 0.22% 1.84% 2.85% 2.79% 6.46% NA 4.56% 2.09% 2.60% 1.63%
Natural Resources 0.00% 0.00% 0.73% 0.00% 1.96% NA 0.61% 0.09% 0.86% 0.00%
Manufactured Goods 1.05% 0.43% 7.17% 0.08% 4.99% NA 2.90% 0.27% 2.26% 0.22%
Refined Resource Products 0.81% 0.02% 0.75% 0.35% 2.72% NA 2.24% 0.09% 1.27% 0.02%
Textiles and Apparel 0.15% 0.24% 6.38% 0.32% 25.52% NA 13.09% 0.13% 6.59% 9.67%
Middle Income Countries, nec
Agriculture 0.18% 0.66% 2.90% 2.79% 4.53% 7.30% NA 12.18% 7.49% 0.78%
Food 0.51% 8.58% 9.69% 10.42% 3.85% 13.44% NA 11.33% 9.60% 2.30%
Natural Resources 0.00% 0.00% 0.05% 0.01% 0.87% 1.98% NA 0.55% 0.26% 0.07%
Manufactured Goods 1.67% 0.92% 2.29% 0.02% 4.11% 8.71% NA 0.76% 1.20% 0.36%
Refined Resource Products 0.83% 0.58% 3.44% 0.59% 2.34% 7.55% NA 0.76% 1.43% 0.84%
Textiles and Apparel 4.70% 13.00% 4.07% 2.71% 13.24% 15.11% NA 4.51% 4.74% 8.82%
OECD, nec
Agriculture 1.25% 1.46% 9.01% 4.05% 1.01% 13.60% 7.91% NA 6.66% 4.15%
Food 3.18% 22.40% 10.03% 22.47% 7.34% 21.03% 15.52% NA 7.85% 2.55%
Natural Resources 0.16% 0.01% 0.77% 3.30% 0.27% 5.82% 5.49% NA 1.27% 0.11%
Manufactured Goods 8.02% 1.75% 7.54% 0.13% 1.80% 9.52% 6.38% NA 2.80% 0.73%
Refined Resource Products 1.72% 0.84% 4.48% 1.01% 0.85% 9.99% 5.48% NA 1.83% 1.33%
Textiles and Apparel 5.50% 8.97% 7.35% 12.01% 2.37% 15.22% 7.69% NA 4.39% 7.37%
Philippines
Agriculture 0.00% 0.03% 0.00% 8.23% 10.75% 39.10% 12.13% 22.00% NA 0.80%
Food 0.50% 6.09% 4.66% 5.99% 6.28% 25.03% 10.68% 10.28% NA 3.18%
Natural Resources 0.00% 0.22% 0.04% 9.83% 8.46% 2.82% 0.58% NA 0.01%
Manufactured Goods 0.19% 0.17% 0.13% 0.04% 2.67% 8.50% 2.83% 0.24% NA 0.22%
Refined Resource Products 0.29% 3.51% 0.30% 0.07% 7.95% 8.48% 2.20% 0.85% NA 0.22%
Textiles and Apparel 3.40% 15.01% 1.50% 24.42% 11.81% 10.69% 7.52% NA 12.74%
USA
Agriculture 0.00% 0.95% 3.21% 8.09% 0.82% 5.63% 4.87% 81.83% 3.84% NA
Food 0.61% 16.83% 8.31% 35.49% 0.92% 15.67% 13.99% 18.87% 5.97% NA
Natural Resources 0.00% 0.50% 0.14% 0.05% 5.08% 1.55% 0.26% 2.19% NA
Manufactured Goods 1.31% 5.27% 0.14% 0.14% 9.11% 4.51% 1.77% 2.45% NA
Refined Resource Products 0.19% 4.83% 1.07% 0.09% 8.91% 4.98% 2.02% 4.66% NA
Textiles and Apparel 3.77% 6.24% 12.72% 0.27% 14.44% 5.60% 6.82% 6.43% NA
*A blank cell means a zero tariff rate. Three sectors are not reported in the table: business services, utilities and
other services. Tariffs on business services and other services are zero. With the exception of five cases, tariffs on
utilities are zero, so are not reported. The only three above 0.04% are: 0.84% (0.17%) on imports from Low Income
countries into China (Middle-Income countries); and 0.28 percent on imports from Middle-Income countries into
OECD, nec.
44
Source: GTAP 9 dataset.
Table 10: Value Added by Region and Sector in the Benchmark Data
(In billions of US dollars)
Region
Australia- Canada China Japan Mexico- Low Middle OECD Philippines USA
New Peru- Income Income nec
Zealand Chile nec nec
Sector Sector
Classification
aggregate CRTS 880 1005 3377 3297 842 442 9527 9421 123 8984
aggregate IRTS 571 664 3640 2414 676 214 5653 9687 93 6247
Agriculture CRTS 35 20 541 50 50 93 1153 316 24 148
Business Services IRTS 420 402 1335 1352 310 106 2761 5037 38 3776
Utilities CRTS 29 38 110 103 20 15 391 397 5 299
Food IRTS 34 40 177 154 79 36 582 612 15 283
Natural Resources CRTS 127 108 384 18 99 107 2159 287 10 273
Other Manufactuing IRTS 54 107 958 481 137 19 901 1969 23 1158
Other Services CRTS 689 839 2342 3125 672 226 5824 8421 84 8264
Refined Resource Products IRTS 57 107 933 405 130 35 1177 1803 13 930
Textiles and Apparel IRTS 5 8 236 21 19 17 234 267 4 100
Source: GTAP 9 dataset.
45
Appendix A: Measurement of the Variety Impact in a Multi-Sector
Model—A Multi-Sector Feenstra Ratio
Feenstra (2010, equation 4) derived his index for the welfare impact of the variety externality in a
one-sector monopolistic competition model. In this note, we extend that measure to a multi-sector, multi-
region Krugman or Melitz model. We begin with a review of the key theorems. Since Feenstra’s theorem
depends on the Sato-Vartia index, we start with the Sato-Vartia index and then summarize Feenstra’s
theorem. We then generalize the Sato-Vartia index and the Feenstra ratio to our multi-sector, multi-region
model.
The Sato-Vartia Theorem
Define the log change price and quantity indices between time 0 and 1 as:
pi1
ln P = i =1 wi ln
n
(1)
pi 0
qi1
ln Q = i =1 wi ln
n
(2)
qi 0
Where ( pit , qit ) are the prices and quantities of i th good in time t and wi is its non-negative weight in
n
the indices such that i =1
wi = 1 . The expenditure index E is given by
n
pi1qi1
E= i =1
. (3)
n
p q
i =1 i 0 i 0
These indices are defined as “ideal” log change indices, if we can find weights wi (which are identical
for both the price and quantity indices) that result in
PQ E .26 (4)
For many years finding the weights that would yield an ideal log change index eluded researchers, and the
literature produced indices with weights that were approximations to the ideal index. The problem was
solved in 1976. Vartia (1976) produced two weighting formulas, known as Vartia I and Vartia II, both of
26
If PQ E with different weights in the price and quantity indices, the indices are said to be dual to
each other and they satisfy the weak factor reversal test. If the weights are identical in the price and
quantity indices, they are said to satisfy the strong factor reversal test and they are ideal. See Sato (1976).
.
46
which yield ideal log change indices. Sato (1976) independently discovered the Vartia I index. Moreover,
Sato showed that if the utility function is homothetic, then his price and quantity indices are dual to the
CES direct and indirect utility function. Given homothetic preferences, he shows that solving the partial
differential equation of the log change quantity index, yields the CES direct utility function.
pit qit
The solution for the weights is the following. Define sti = as the expenditure share
n
p q
k =1 kt kt
of good i at time t, t = 0,1. Then the weights of the Vartia I or the Sato-Vartia index are:
s1i − s 0i
ln( s1i ) − ln( s0i )
wi = . (5)
n s1 − s 0
k =1 ln(s1 k) − ln(ks0 )
k k
Rewriting equation (1), the Sato-Vartia price index may be written as:
wi
p
= i =1 i1
SV n
P (6)
pi 0
The Sato-Vartia Index at the Sector Level
We must extend the Sato-Vartia index to our model with multiple sectors and multiple regions.
Importantly, Sato (1976, footnote 11) notes that while the formula (6) applies for an economy-wide index
where each good i is a unique good, the index would remain an ideal sub-index for subgroups where i is a
sector composed of multiple goods. That is, this log-change formula can be applied to the sub-groups to
yield ideal sub-indices. We employ this ideal sub-group property to obtain Sato-Vartia indices at the
sector level.
Notation. Using 0 and 1 to denote the value of a variable in the benchmark and counterfactual
equilibrium, respectively, define:
(i) s0irs and s1irs as the expenditure shares in the benchmark and counterfactual equilibria, respectively,
for all varieties of goods in sector i shipped from region r into region s;
(ii) pf 0irs and pf 1irs as the gross price (including tariffs) of a firm’s variety of good i shipped form
region r to region s;
(iii) qf 0irs and qf 1irs as the quantity of a firm’s variety of good i shipped from region r to region s;
(iv) N 0irs and N1irs the number of firms (varieties) in sector i that are shipped from region r to region s.
47
In the Krugman model, all varieties in a given sector have the same price and are shipped to all
destinations, so we could drop the destination index. In the Melitz model, however, the measure of the
number of firms is bilateral, and firm level prices and quantities in (ii) and (iii) above are for the
representative firm as defined by Melitz (2003). Bilateral expenditure shares are given by the product of
the gross price of the representative firm times its quantity times the number of firms, i.e.,
pf 0irs qf 0irs N 0irs
s0irs =
pf 0its qf 0its N 0its
t
pf 1irs qf 1irs N1irs
s1irs =
pf 1its qf 1its N1its
t
Given the expenditure shares, the Sato-Vartia weights for our sub-indices are:
s1irs − s 0irs
ln( s1irs ) − ln( s 0irs )
wirs =
s1its − s 0its
t
ln( s1 ) − ln( s 0 )
its its
Our matrix of bilateral trade data at the country level does not contain any zeros. That is, we have
stirs 0 , for all i, r, s and t. Since we track expenditures based on sales of the average firm, we have
wirs 0 for all i, r, s, d, and we do not need an exception for zero expenditures. 27
We have w
t
its =1.
Analogous to the Sato-Vartia index in a single sector model, we define a Sato-Vartia index for the
sector i in region s of the change in prices charged by the firm. Our ideal price index for the change in the
prices charged by firms for the goods available in both periods is:
wirs
pf
= irs
SV
Pis (7)
r pf 0irs
27
In Feenstra’s approach, he allowed for the possibility of some good to be sold in only one of the periods by
defining the Sato-Vartia index over the set I = I t −1 where I t is the set of goods with expenditures in period t;
It
so I is the set of goods with expenditures in both periods. If there were no exports to some destinations for some
good i, in either the benchmark or counterfactual scenario, noting that lim wirs = 0 and similarly for s0irs , we
s1irs →0+
could define the weight as zero if the share is zero in either the benchmark of counterfactual. Then if a new variety
appears (or an old variety disappears) in the counterfactual equilibrium, this variety would not impact our Sato-
Vartia index.
48
Feenstra’s Theorem
1/( −1)
(I )
In his one-sector model, Feenstra shows that there is a ratio t −1 that precisely
t ( I )
measures the value of the Dixit-Stiglitz variety externality. In particular, Feenstra (2010, Theorem 4)
shows that:
1/( −1)
e( pt , I t ) (I )
= P SV ( pt , pt −1 , qt , qt −1 , I ) t (8)
e( pt −1 , I t −1 ) t −1 ( I )
where e is the unit expenditure function and P SV (...) is the Sato-Vartia price index, σ is the elasticity of
substitution between varieties, I t is the set of goods available in period t and I = I t −1 It is the set of
goods available in both periods . Intuitively, since the Sato-Vartia index is based on prices charged by
firms, it does not account for the Dixit-Stiglitz variety externality. Feenstra’s theorem tells us that the
proportional change in the unit expenditure function is the product of two terms that are: (i) a measure of
the change in the prices charged by firms on goods common to the two periods (the Sato-Vartia index);
and (ii) a ratio that is a measure of the value of the variety gain, the Feenstra ratio. The methodology
requires distinguishing changes in the price charged by firms on goods available in both periods from the
variety externality.
The Multi-Sector Feenstra Ratio
For what follows, it is convenient to rewrite equation 8 as:
1/( −1)
(I ) e( pt −1 , I t −1 ) SV
F = t −1 = P ( pt , pt −1 , qt , qt −1 , I ) (9)
t ( I ) e( pt , I t )
where we call the left hand side the Feenstra ratio.
Feenstra Ratio at the Sector Level. Define PDSis (1) and PDSis (0) as the Dixit-Stiglitz price
index in the sector i in region s, in the counterfactual and benchmark equilibrium, respectively. We use
these unit cost indices to adjust for the variety externality at the sector level in a manner analogous to the
Feenstra theorem. That is, we define the Feenstra ratio at the sector level as:
PDSis (0) SV
Fis = Pis (10)
PDSis (1)
where in equation 10 we use our ideal sector level Sato-Vartia index in region s. Since the Dixit-Stiglitz
price index is the unit cost function for goods in sector i taking variety into account, and the Sato-Vartia
index is the ideal index for the prices charged by firms, we believe we have an ideal measure of the
49
variety impacts at the sector level. In particular, in a one-sector model our measure is the measure of
Feenstra in equation 8, i.e., we have that Fis = F if the number of sectors equals 1.
Multi-Sector Feenstra Ratio. For an ideal aggregate measure of the variety impacts across sectors,
independent of the price impacts,28 we would have to account for the differential variety impacts that
accrue in terms of their use as both intermediates and final goods. This, in turn, depends on the production
technology for intermediates and utility across different goods. We take the simplest tact and calculate a
weighted average of the sector-level variety impacts. That is, we define:
Fs = i Fis (11)
i
where i is the economy-wide expenditure share (absorption share) on goods or services in sector i. Our
aggregate measure is an approximation to an ideal multi-sector measure of the variety impacts; but we
have calculated and presented this aggregate variety measure for more than 40 scenarios in this paper, and
the measure has proved useful and intuitive in explaining the welfare results.
28
Sato (1976) notes that if the utility function is multi-level CES, then it is possible to aggregate from the sub-
indices to get an ideal aggregate index.
50
Appendix B: Equations of the Model
1 Notation:
Sets:
Regions are indexed by r ∈ R or s ∈ R. Goods and services are indexed by i ∈ I or j ∈ I ,
with subsets k ∈ K ⊆ I for Krugman treatments, and m ∈ M ⊆ I for Melitz treatments.
˜ ⊆ F indicating a sector-speciﬁc factor.
Factors are indexed by f ∈ F , with the subset F
Variables:
Dr Full consumption index1
Yir Output index2
Air Supply index3
er Unit-expenditure index (true-cost-of-living index)
cir Price of domestic output (marginal cost)
Pir Price of goods and services
wf r Price of factors
˜f ir Price of sector-speciﬁc factors
w
Mir Number of entered ﬁrms (∀i ∈ K ∪ M )
pkrs Gross ﬁrm-level price (Krugman ﬁrms)
qkrs Firm-level quantity (Krugman ﬁrms)
Nmrs Number of operating Melitz ﬁrms
˜krs Gross ﬁrm-level price (Melitz representative ﬁrms)
p
˜krs Firm-level quantity (Melitz representative ﬁrms)
q
˜krs Firm-level productivity (Melitz representative ﬁrms)
φ
1
In the model variation where we have an aggregate intermediate good this real quantity is gross of
intermediates inputs. Consumption includes all ﬁnal demand: household consumption and investment, and
government spending.
2
Under monopolistic competition treatments Yir indicates the composite (of value added and intermedi-
ate) input that ﬁrms in industry i use for real ﬁxed and variables costs.
3
Under an Armington treatment this is the Armington aggregate, and under monopolistic competition
this is the Dixit-Stiglitz aggregate.
1
RAr Nominal Income
Instruments:
τirs Iceberg trade cost factor
tirs Tariﬀ
Parameters:
d0r Benchmark value of full consumption1
y 0ir Benchmark value of gross output of sector i
a0ir Benchmark value of domestic and imported supply of good i
αjir Benchmark share of intermediate input j in gross output for sector i
βf ir Benchmark share of factor f in value added for sector i
µr Benchmark share of leisure in full consumption
θir Benchmark share of i in total consumption of goods and services1
σ Elasticity of substitution between Armington, Krugman, or Melitz varieties
λA Preference weight in Armington aggregation of varieties
λK Preference weight in Krugman aggregation of varieties
λM Preference weight in Melitz aggregation of varieties
σ L Elasticity of substitution between leisure and consumption of goods and services
σ M Elasticity of substitution between intermediate inputs and value added
K
fkr Fixed cost associated with a Krugman ﬁrm
M
fmrs Fixed cost associated with operation of a Melitz ﬁrm on the r to s trade link
E
fmr Sunk cost associated with entry of a Melitz ﬁrm
a Pareto shape parameter
b Pareto lower support
δ Annual probability of ﬁrm death
¯f r Endowment of mobile factor f
F
SF f ir Endowment of factor f speciﬁc to industry i
BOP r Benchmark capital account surplus
2
2 Model Equations
2.1 Representing the basic technologies and preferences using du-
ality
With the option to include a labor-leisure choice the unit expenditure function is given by
( )1−σL 1/(1−σL )
∏
er = µr wLr
1−σ
L
+ (1 − µr ) θir
Pir . (1)
i
If µr = 0 labor supply is inelastically supplied, and preferences over goods and services
indicated by the Cobb-Douglas nest alone.
The production function is given by one of the following alternative equations depending
on the assumed treatment of intermediates. We always assume that value added inputs
combine in a Cobb-Douglas nest, but there are diﬀerent treatments of intermediates. If we
assume intermediates entering the top-level nest with an elasticity of substitution σ M ̸= 1
we have
1−σM 1/(1−σM )
∑ ∑ ∏ ∏
cir = 1−σ M
αjir Pjr + (1 − αjir ) (wf r )βf ir ˜f ir )βf ir
(w . (2a)
j j f∈
/F˜ f ∈F
˜
If σ M = 1 the production function is fully Cobb-Douglas:
∏ ∏ ∑ ∏ ∑
cir = (Pjr )αjir (wf r )(1− j αjir )βf ir ˜f ir )(1− j αjir )βf ir ;
(w (2b)
j f∈
/F˜ f ∈F
˜
and if we adopt the single-composite intermediate treatment the price of intermediates is
the unit expenditure index:
αjir ∏ ∏
∑ ∑ ∑
cir = er j (wf r )(1− j αjir )βf ir ˜f ir )(1− j αjir )βf ir .
(w (2c)
f∈
/F˜ f ∈F
˜
In the case that we do not have intermediates the α share parameters in the above equations
would be zero.
The next set of equations indicate the good or service supply price, which is a compos-
ite of domestic and imported goods. For the Armington structure we have a simple CES
aggregation of regional varieties:
[ ]1/(1−σ)
∑
1−σ
Pis = λA
irs [(1 + tirs )τirs cir ] . (3a)
r
3
If we assume monopolistic competition the supply prices are indicated by the Dixit-Stiglitz
aggregation of either Krugman ﬁrm-level varieties,
[ ]1/(1−σ)
∑
1−σ
Pks = λKkrs Mkr pkrs ; (3b)
r
or Melitz representative-ﬁrm varieties,
[ ]1/(1−σ)
∑
1−σ
Pms = λM ˜mrs
mrs Nmrs p . (3c)
r
2.2 Market clearance
Equations 1 through 3 indicate a dual representation of technologies and utility. We proceed
with the general equilibrium market-clearance conditions prior to establishing the bilateral
equilibrium conditions that are speciﬁc to the monopolistically competitive Krugman and
Melitz ﬁrms. First we establish the market clearance conditions for goods and services
available for domestic use. These goods trade at the price Pir , and are potentially used
in consumption and production. Supply is simply the quantity a0ir Air and demand is de-
rived by applying Shephard’s Lemma to the established dual consumption and production
technologies. The market clearance conditions are given by
∂er ∑ ∂cjr
a0ir Air = d0r Dr + y 0jr Yjr . (4)
∂Pir j
∂Pir
where er and cjr indicate the unit-cost (unit-expenditure) functions indicated in (1) and (2).
Now let us establish the market clearance conditions for output which has cir as the
associated price. Under the Armington structure output is exported directly and so market
clearance is given by
∑ ∂Pis
y 0ir Yir = τirs a0is Ais , (5a)
s
∂ [(1 + tirs )τirs cir ]
where the price index Pis is given by unit cost function deﬁned in (3a) above. For the
monopolistic competition formulations the market clearance conditions must account for the
various uses of output as it is allocated to ﬁxed and variable costs. Under Krugman we have
( )
∑
K
y 0kr Ykr = Mkr fkr + qkrs ; (5b)
s
and under Melitz we have
∑ ( )
E M ˜mrs
q
y 0mr Ymr = δfmr Mmr + Nmrs fmrs + . (5c)
s
˜mrs
φ
4
We proceed by specifying the market clearance conditions for primary factors. For sector-
speciﬁc factors (f ∈ F
˜ ) the rents are indicated by the following market clearance condition:
∂cjr
SF f ir = y 0ir Yir . (6)
∂w˜f ir
For mobile factors (f ∈/ F˜ ), that are not labor, we have to account for demand across the
diﬀerent sectors, and for labor we have to account for leisure demand:
∑ ∂cjr ∂er
¯f r =
F y 0ir Yir + d0r Dr . (7)
i
∂wf r ∂wf r
Leisure demand is given by the ﬁnal term on the right-hand side, and this is not zero only
if the factor is labor and we have chosen an elastic-labor-supply structure.
Normally the quantity associated with the consumption activity d0r Dr would be fully
exhausted in ﬁnal demand, but under our treatment of a single-composite intermediate input
some of this quantity is used in production. Market clearance for the consumption good is
given by
RAr ∑ ∂cjr
d0r Dr = + y 0jr Yjr . (8)
er j
∂er
Notice that changes in the ﬁrst term on the right-hand side indicate changes in money-metric
utility (Equivalent Variation). Also note that the second term drops out in all cases except
when we assume a single-composite intermediate input, because er only enters production if
we have equation (2c).
2.3 Krugman speciﬁc equations
Consistent with the Dixit-Stiglitz aggregation bilateral demand for a variety produced by a
small ﬁrm shipping from r to s is given by
( )σ
K Pks
qkrs = λkrs a0ks Aks . (9)
pkrs
Faced with this demand the small ﬁrm maximizes proﬁt by setting the gross price according
the standard markup:
(1 + tkrs )τkrs ckr
pkrs = . (10)
1 − 1/σ
Accumulated proﬁts earned across all markets will just cover ﬁxed costs, which indicates the
free entry condition: ∑ pkrs qkrs
K
= fkr ckr . (11)
s
σ (1 + tkrs )
Associated with this equilibrium condition is Mkr the number of entered ﬁrms in region r.
5
2.4 Melitz speciﬁc equations
The Melitz model is slightly more complex because there is both entry and selection. We
proceed along the same path as the Krugman formulation, but we note that the equilibrium is
deﬁned on a representative variety (ﬁrm) for each trade link. Demand for the representative
variety is ( )σ
M Pms
˜mrs = λmrs a0ms Ams
q . (12)
˜mrs
p
Optimal pricing is given by
(1 + tmrs )τmrs cmr
˜mrs =
p . (13)
˜mrs (1 − 1/σ )
φ
We now need a condition that indicates selection into each bilateral market. The marginal
ﬁrm will be earning zero proﬁts, and Melitz shows how the revenues of the marginal and
average ﬁrms are related using the distribution of productivity draws. With a Pareto distribu-
tion (with shape parameter a) the zero-cutoﬀ proﬁt condition speciﬁed on the representative
ﬁrm’s proﬁt is given by
p ˜mrs a + 1 − σ
˜mrs q M
= fmrs cmr . (14)
(1 + tmrs ) aσ
We still need a zero proﬁt condition that determines how many ﬁrms enter (take a pro-
ductivity draw). A potential entrant has positive expected proﬁts from potentially multiple
markets. Equilibrium requires these expected proﬁts just equal the annualized sunk cost
associated with establishing a variety. With δ as the rate of ﬁrm death we have the following
free-entry condition:
∑ Nmrs p ˜mrs (σ − 1)
˜mrs q E
= δfmr cmr . (15)
s
M mr (1 + tmrs )aσ
Finally, we must establish the productivity of the representative ﬁrm, which depends on the
density of operating to entered ﬁrms and the parameters of the distribution:
( )−1/a ( )1/(1−σ)
Nmrs a+1−σ
˜mrs = b
φ . (16)
Mmr a
2.5 Income balance
The ﬁnal task is to establish income balance, where the nominal expenditures of the rep-
resentative agent must equal the value of factor endowments, any tariﬀ revenue, plus the
capital account surplus. The model is based on real quantities (it is linearly homogeneous in
prices), and thus the transfer associated with the capital account must be speciﬁed in terms
of a real unit. We use the aggregate consumption commodity for the United States as the
transfer commodity. Of course, the sum across all regional capital account surpluses is zero
(there is no global surplus or deﬁcit), so there is no net impact on equation (8). The regional
6
budget is given by
∑
RAr = ¯f r
wf r F
f∈
/F˜
∑∑
+ ˜f r SF f ir
w
f ∈F
˜ i
∑ ∂Pir
+ tisr cis a0ir Air
∂ [(1 + tirs )τirs cir ] (17)
/ K ∪M
i∈
∑
+ tksr Mks pksr qksr /(1 + tksr )
k
∑
+ ˜msr q
tmsr Nksr p ˜msr /(1 + tmsr )
m
+ eusa BOP r .
3 Computation
The plain algebraic formulation of the model presented above hides a few details that facil-
itate an eﬃcient computational model. The model is solved as a Mixed Complementarity
Problem (MCP) within the General Algebraic Modeling System (GAMS) software using the
Mathematical Programming System for General Equilibrium (MPSGE) as a subsystem.4
The base Armington model, formulated using MPSGE, is augmented with the Krugman
and Melitz modules as outlined in the decomposition methods suggested by Balistreri and
Rutherford (2013).
An MCP takes advantage of variational-inequality (or sometimes complementary-slack )
equilibrium conditions whereby positive variables (prices and activity indexes) are associate
with speciﬁc inequality constraints. For example, the market clearance condition (4) is
actually computed using the following variational-inequality condition:
( )
∂er ∑ ∂cjr
a0ir Air − d0r Dr + y 0jr Yjr ≥ 0;
∂Pir j
∂P ir
Pir ≥ 0; and (18)
[ ( )]
∂er ∑ ∂cjr
Pir a0ir Air − d0r Dr + y 0jr Yjr = 0.
∂Pir j
∂Pir
In words we would say supply is greater than or equal to demand and if the associated price
is positive supply equals demand. If supply is greater than demand the associated price
4
GAMS Development Corporation. General Algebraic Modeling System (GAMS) Release 24.9.1. Wash-
ington, DC, USA, 2017. MPSGE is a subsystem written by Thomas F. Rutherford (Rutherford, 1999) that
automatically generates and checks consistency of the nonlinear constant-returns system.
7
in that market must be zero. Competitive production activities have a similar structure—
marginal cost may exceed price for slack activities. While in this application the CES
technologies and preferences keep the equilibrium away from corners, the MCP formulation
is computationally eﬃcient and extensible to equilibrium problems that include non-scarce
commodities and slack activities.
One additional note on computation is that we take advantage of the calibrated-share form
of the CES technologies and preferences as suggested by Rutherford (2002). We choose units
such that the central prices are one in the benchmark, and the activity indexes are naturally
one. This yields an equilibrium problem that is naturally well scaled for computation (the
central variables are unity in the known equilibrium). One may relate our scaling strategy to
the “exact-hat” methods popular with other authors (Dekle et al., 2008; and Costinot and
Rodr´ıguez-Clare, 2014). Like those methods we directly recover the proportional changes in
prices and quantity indexes as the equilibrium solution to the numeric problem.
8