Endogenous Market Structures and International Trade: Theory and Evidence

AuthorFederico Etro
Publication Date01 Jul 2015
Scand. J. of Economics 117(3), 918–956, 2015
DOI: 10.1111/sjoe.12084
Endogenous Market Structures and
International Trade: Theory and Evidence
Federico Etro
Ca̍Foscari, University of Venice, 30123 Venice, Italy
Under constant elasticity of substitution (CES) preferences and Cournot (or Bertrand) com-
petition, a larger market induces exits of domestic firms, lower prices, and larger production
of surviving firms because of competition from more foreign f irms, even without resort-
ing to the selection effects of Melitz. The elasticity of the number of firms to population
decreases with substitutability between goods, and it reaches 0.5 under Cournot competi-
tion with homogeneous goods: empirical evidence supports this structural relation against
the unitary elasticity of monopolistic competition. The results hold also in a 2 ×2×2
Heckscher–Ohlin model with imperfect competition generating inter- and intra-industry trade
due to comparative advantage or comparative preferences.
Keywords: Comparative advantage; comparative preferences; gains from trade; Krugman
JEL classification:F11; F12
I. Introduction
In traditional theories of trade, such as the neoclassical theory based on
perfect competition and the Krugman theory based on monopolistic com-
petition, the impact of trade and the gains from trade are associated with
international specialization or with the consumption of new varieties of
goods. These two approaches (integrated in Helpman and Krugman, 1985)
neglect the fact that trade usually affects the structure of markets, and, in
particular, the degree of competition, the number of firms active in each
country, the production level of each firm, and the mark-ups, with conse-
quences for welfare as well. A separate line of research started by Brander
(1981) and Markusen (1981) has shown that opening up to trade in imper-
fectly competitive markets generates price reductions. However, the main
I am grateful to Beverly Lapham, Jonathan Eaton, Jacques Thisse, and two anonymous
referees for insightful comments, and to Elena Stepanova for excellent research assistance.
Dirk Czarnitzki has worked jointly with me on a preliminary version of the empirical
results of Section IV and I am highly indebted with him. Finally, I am grateful to Aditya
Bhattacharjea, Parishit Ghosh, Debasis Mondal, Mausumi Das, and other seminar participants
at the Delhi School of Economics (India).
CThe editors of The Scandinavian Journal of Economics 2014.
F. E t r o 919
implications of this endogenous-market-structure approach have not been
investigated in a systematic way; it is not by chance that leading interna-
tional trade economists (such as Neary, 2010) talk about “two and a half
theories of trade”.
In this paper, first we develop a rather general one-sector partial equi-
librium model of endogenous market structures (EMSs) with Cournot and
Bertrand competition. We show that gains from trade emerge under general
conditions and can be associated with both the increase in the number of
goods consumed and the reduction of their prices.1Then, we focus on a
microfounded demand based on the constant elasticity of substitution (CES)
preferences, introduced by Dixit and Stiglitz (1977) and applied to trade in
the models of Krugman (1980) and Melitz (2003) under monopolistic com-
petition with homogeneous and heterogeneous firms, respectively. In both
models, the mark-up was constant for all the firms and the number of fir ms
was linearly increasing with respect to the size of the market under free
trade.2We depart from monopolistic competition and examine Cournot and
Bertrand competition. With competition in quantities, an expansion of the
market associated with trade increases less than proportionally the number
of firms, because stronger competition reduces the mark-ups and forces
the firms to produce more in order to cover the fixed costs. For instance,
in the case of homogeneous goods, the elasticity of the number of firms
with respect to the size of the market is β=0.5. This means that doubling
the size of a market increases the number of firms by about 40 percent,3
expands their production, and reduces their average cost and mark-up, with
relevant welfare gains even if there are zero gains from variety. When
the substitutability between goods decreases, the implied elasticity of the
number of firms increases (but remains below unity), and trade creates
gains from both the consumption of new varieties and the reduction of
their prices. Similar results emerge in the case of competition in prices;
expanding the market size attracts more firms (in this case, approximately
in a linear way), which strengthens competition, reduces the prices, and
requires a larger production for each firm.
The model is then extended to a general equilibrium 2 ×2×2 set-
up with identical preferences and different factor endowments between
countries. A trading equilibrium with diversification induces factor price
1This is in line with the estimates of Feenstra and Weinstein (2010) on the US, for which
the gains from variety are quantitatively similar to the gains from lower mark-ups.
2For an alternative microfoundation of monopolistic competition, see Bertoletti and Etro
(2013), who assume additive indirect utility, as opposed to additive direct utility (as in Dixit
and Stiglitz, 1977). For a generalization, see Bertoletti and Etro (2014).
3If the relation between the number of firms Nand the size of the market Sis ln N=βln S,
a doubling in the size of the market attracts 2βfirms. The model of Krugman (1980) with
monopolistic competition implies β=1.
CThe editors of The Scandinavian Journal of Economics 2014.
920 Endogenous market structures and international trade
equalization, and equal output and mark-ups for all firms (under both
price and quantity competition). Assuming EMSs in the capital-intensive
sector and perfect competition in the labor-intensive sector, the trading
equilibrium induces intra-industry trade in the capital-intensive sector with
positive net exports of the capital-abundant country.4Assuming Cobb–
Douglas production functions in both sectors, we can fully solve the model
in closed form and we can verify that the mentioned relation between the
total number of firms and the size of the integrated market is robust to this
extension; the relation between the market size and the number of domestic
firms depends on multiple factors, but follows a similar logic in the case
of symmetric countries.
We employ the model to consider the case of different homothetic pref-
erences in the two countries, analyzing the realistic situation in which the
imperfectly competitive sector is the capital-intensive sector and one coun-
try has a relative preference for the goods of this capital-intensive sector.
In the extreme case of identical factor endowments (no technological com-
parative advantage), this country produces more capital-intensive goods in
autarky, but reduces their production under free trade and starts importing
them. Therefore, the process of local business destruction can be much
more drastic for one of the two countries. As far as we know, this is the
first formalization in this set-up of the implications of trade due to com-
parative preferences rather than comparative advantage. Finally, we extend
the model to differences in productivities between firms, as in the model
of Melitz (2003), which is based on monopolistic competition and, under
free trade, preserves the linearity of the number of firms in the size of the
market (because the average profit of the firms is invariant in market size).
Introducing strategic interactions restores the role of trade in inducing local
business destruction, larger production of each surviving firm, and lower
prices without resorting to trade costs.
We also provide a preliminary test of the structural relation between the
number of firms in manufacturing sectors and the market size measured
by industry sales. Because our model of market integration abstracts from
transport costs and trade barriers, we focus on market structures emerging
in a large integrated market. Our investigation is based on a panel of
industry-level data for German manufacturing. We reject a linear relation
between the number of firms and the market size (against the Dixit–Stiglitz
monopolistic competition case) and the absence of a relation (against the
perfectly competitive case), and we estimate an elasticity βbetween 0.5
and 0.65, in support of the case of Cournot competition. This is in line
with the elasticity of 0.65 found by Eaton et al. (2011) in a French dataset
4Our generalized 2 ×2×2 model nests traditional models: the Heckscher–Ohlin model, the
Brander–Markusen model, and the Helpman–Krugman model.
CThe editors of The Scandinavian Journal of Economics 2014.

To continue reading

Request your trial