scispace - formally typeset
Search or ask a question
Journal ArticleDOI

Comparing alternative methods to estimate gravity models of bilateral trade

01 Jun 2013-Empirical Economics (Springer-Verlag)-Vol. 44, Iss: 3, pp 1087-1111
TL;DR: A survey of the most recent literature concerning the specification and estimation methods of the gravity equation can be found in this paper, where the Heckman sample selection model performs better overall for the specification of gravity equation selected.
Abstract: The gravity equation has been traditionally used to predict trade flows across countries. However, several problems related with its empirical application still remain unsolved. The unobserved heterogeneity, the presence of heteroskedasticity in trade data or the existence of zero flows, which make the estimation of the logarithm unfeasible, are some of them. This paper provides a survey of the most recent literature concerning the specification and estimation methods of this equation. For a dataset covering 80% of world trade, the most widely extended estimators are compared, showing that the Heckman sample selection model performs better overall for the specification of gravity equation selected.

Summary (2 min read)

1. Introduction

  • The gravity model of trade, which was originally inspired by Newton’s gravity equation, is based on the idea that trade volumes between two countries depend on their sizes in relation to the distance between them.
  • After some additional discussions concerning its specification in the nineties, the debate has now turned to the performance of different estimation techniques.
  • Helpman et al. (2008) propose a theoretical foundation based on a model with heterogeneity of firms à la Melitz (2003) and an adapted Heckman procedure to predict trade taking into account these features.
  • The next section briefly reviews the different theoretical foundations of the gravity equation to justify the election of the empirical specification of the gravity equation chosen.

2. The gravity equation

  • The theoretical foundation of the gravity equation appeared seventeen years after its empirical specification.
  • Regarding the specification, Anderson and van Wincoop (2003) propose an augmented version of the Anderson (1979) model based on the assumption of differentiation of goods according to place of origin.
  • Finally, the variables Pi and Pj are the multilateral resistance terms and are defined as a function of each country’s full set of bilateral trade resistance terms.
  • Anderson and van Wincoop (2003) use the observed variables in their model (distances, borders, and income shares) to obtain the multilateral trade resistance terms.
  • Their sample contains the same 30 US states and 10 Canadian provinces that McCallum (1995) includes.

3.1. Linear methods

  • Since the logarithm of zero is not defined, truncation and censoring methods have been proposed in the literature to treat the problem of zero flows in data.
  • In addition, a panel framework permits recognising how the relevant variables evolve through time and identifying the specific time or country effects.
  • Two main techniques are employed to fit data depending on the a priori assumptions.
  • By contrast, the random effects model imposes no correlation between the individual effects and the regressors, implicitly assuming that the unobserved heterogeneous 2 See Wei (1996), Wolf (1997), and Head and Mayer (2000) for further information.

3. 2. Nonlinear methods

  • As Santos Silva and Tenreyro (2006) points out, the log-linearisation of the gravity equation changes the property of the error term, thus leading to inefficient estimations in the presence of heteroskedasticity.
  • Among nonlinear estimation methods, the most frequently used are Nonlinear Least Squares (NLS), Feasible Generalised Least Squares (FGLS), Heckman sample selection model and Gamma and Poisson Pseudo Maximum Likelihood (GPML and PPML).
  • The model allows for some positive correlation between both error terms to better reflect the real decision process.
  • Not robust to heteroskedasticity - Sample selection bias Tenreyro (2006) FGLS (Feasible Generalised Least Squares) -.
  • It provides a rationale for zero trade flows - Unbiased estimates - Difficult to estimate - Additional data is required (exclusion variables) Helpman et al. (2008); Santos Silva and Tenreyro (2008).

4. Comparing estimation methods for a baseline gravity

  • The new workhorse in the estimation of the gravity equation is still unclear.
  • Econometric estimation presents some challenges that remain unsolved as of yet.
  • First, the exclusion of the multilateral trade resistance terms leads to biased estimates due to the omission of variables.
  • Since the logarithm of zero is unfeasible, some information would be lost.
  • This problem is becoming more important due to the use of disaggregated data, in which over 50% of values is zero.

4.1. Data and model

  • The sample covers bilateral exports of 80 countries over the 1980-2008 period.
  • All the countries of the EU15, the CEE new European members, and 6 Middle East and North African (MENA) countries (Morocco, Tunisia, Egypt, Turkey, Israel and Algeria) as well as most OECD countries are included.
  • For the sake of comparison, a gravity equation based on Anderson and van Wincoop’s (2003) theoretical model will be used: ijtjtitijij ijijijjtitijt εγγγdα smctryαcomlaαcontigαyαyαX ln++++ln+ +++ln+ln=ln 6 54321 (3) The dependent variable is the logarithm of the volume of exports in current dollars from country j to i, obtained from the CHELEM-CEPII database.
  • Dij is a variable representing the geodesic distance between i and j and is obtained from the CEPII database.
  • Due to the inclusion of these dummies, GDP terms are dropped from the estimation.

4.2 Results

  • Before estimating equation (3), some specification tests were conducted.
  • As expected, both the exporter and importer GDP increases exports regardless of the estimation method used, while the distance reduces exports.
  • The main differences among estimators are revealed in the magnitude of coefficients.
  • PPML notably reduces the magnitude of the coefficients as well as the standard errors.
  • While other methods treat zero flows as inexistent, Heckman considers them to be unobserved.

5. Concluding remarks

  • The gravity model is considered one of the most successful empirical frameworks in international economics.
  • Second, the logarithm of zero is unfeasible.
  • After applying several criteria to test goodness of fit, it is argued that ad hoc methods are not appropriate for estimating the gravity equation since they provide biased and inefficient estimates.
  • This paper suggests that the Heckman sample selection model is the preferred estimation method within nonlinear techniques when data are heteroskedasticity and contain a significant proportion of zero observations.
  • The author is indebted to J. Milgram for her helpful comments.

7. References

  • The gravity equation in international trade: some microeconomic foundations and empirical evidence.
  • Some econometric considerations, also known as The gravity model.

Did you find this useful? Give us your feedback

Content maybe subject to copyright    Report

1
Comparing alternative methods to estimate
gravity models of bilateral trade
Estrella Gómez Herrera. Department of Economic Theory, University of Granada,
Granada, Spain (e-mail: estrellagh@ugr.es)
Abstract. The gravity equation has been traditionally used to predict trade flows across countries.
However, several problems related with its empirical application still remain unsolved. The unobserved
heterogeneity, the presence of heteroskedasticity in trade data or the existence of zero flows, which make
the estimation of the logarithm unfeasible, are some of them. This paper provides a survey of the most
recent literature concerning the specification and estimation methods of this equation. For a dataset
covering 80% of world trade, the most widely extended estimators are compared, showing that the
Heckman sample selection model performs better overall for the specification of gravity equation
selected.
Keywords International trade, gravity model, estimation methods
JEL Classification C13, C33, F10
1. Introduction
The gravity model of trade, which was originally inspired by Newton’s gravity
equation, is based on the idea that trade volumes between two countries depend on their
sizes in relation to the distance between them. In the last fifty years, this model has been
widely used to predict trade flows.
The gravity equation appears to be highly effective at this point as proven at a
very early date by the works of Linnemann (1966) and Leamer and Stern (1971).
However, several controversies have arisen regarding the model. The theoretical
framework was put into doubt and afterwards justified by Bergstrand (1989) for the
factorial model, Deardorff (1998) for the Hecksher-Ohlin model, Anderson (1979) for
goods differentiated according to their origin, and Helpman et al. (2008) in the context
of firm heterogeneity. After some additional discussions concerning its specification in
the nineties, the debate has now turned to the performance of different estimation
techniques. New estimation problems concerning the validity of the log linearisation
process of the gravity equation in the presence of heteroskedasticity and the loss of
information due to the existence of zero trade flows have been recently explored.
Traditionally, the multiplicative gravity model has been linearised and estimated
using OLS assuming that the variance of the error is constant across observations
(homoskedasticity), or using panel techniques assuming that the error is constant across
countries or country-pairs. However, as pointed out by Santos Silva and Tenreyro
(2006), in the presence of heteroskedasticity, OLS estimation may not be consistent and
nonlinear estimators should be used. Another challenge described in the literature
concerns the zero values. Helpman et al. (2008) propose a theoretical foundation based
on a model with heterogeneity of firms à la Melitz (2003) and an adapted Heckman
procedure to predict trade taking into account these features. Recently, the works of
Burger et al. (2009), Martin and Pham (2008), Martínez-Zarzoso et al. (2007),

2
Siliverstovs and Schumacher (2009) and Westerlund and Wilhelmsson (2009) have
obtained divergent results when comparing alternative estimation methods.
This paper reviews most estimation methods and problems and provides a survey
of the literature related to this topic. The performance of several linear and nonlinear
estimators is compared using a three-dimensional (i, j, t) dataset, analysing the most
relevant properties of each one. To this end, a gravity equation based on Anderson and
van Wincoop’s (2003) theoretical model is used. Using this equation, the fit of different
estimation procedures applied to a large dataset of bilateral exports for 80 countries
(80% of world trade) over the 1980-2008 period is discussed. The fit of each method is
compared through different measures, revealing the main advantages and disadvantages
of each one. It is shown that methods that do not properly treat the presence of zero
flows on data exhibit noticeably worse performance than others. On the other hand,
nonlinear estimators show more accurate results and are robust to the presence of
heteroskedasticity in data. Overall, the Heckman sample selection model is revealed to
be the estimator with the most desirable properties, confirming the existence of sample
selection bias and the need to take into account the first step (probability of exporting)
to avoid the inconsistent estimation of gravity parameters.
The rest of the paper is organised as follows. The next section briefly reviews the
different theoretical foundations of the gravity equation to justify the election of the
empirical specification of the gravity equation chosen. Section 3 compares in detail the
different estimation methods available in the gravity literature. In Section 4, data are
presented and the results of different estimations methods are discussed and compared
to different criteria. Conclusions are drawn in Section 5. The figures and tables are
provided in the Appendix.
2. The gravity equation
The theoretical foundation of the gravity equation appeared seventeen years after its
empirical specification. The first article providing a microfoundation of this equation
was Anderson (1979) and was based on the Armington assumption of specialisation of
each nation in the production of only one good. Bergstrand (1985) initially supported
this hypothesis, completing the theoretical foundation with a more detailed explanation
of the supply side of economies and the inclusion of prices in the equation.
A few years later, a new wave of developments came with what has been called “the
new trade theory”. The main improvement was the replacement of the assumption of
product differentiation by country of origin by the assumption of product differentiation
among producing firms. In this line, Bergstrand (1990) provided a foundation based on
Dixit and Stiglitz’s monopolistic competition assumption. In addition, he generalised
the model by introducing prices and incorporating the Linder hypothesis. Helpman
(1987) also derived a foundation relying on the assumption of increasing returns to scale
where products were differentiated by firms, not only by country, and firms were
monopolistically competitive. However, some years later Deardoff (1998) asserted that
the gravity equation could be derived from standard trade theories, conciliating both the
old and the new theories.
Later on, the “new new trade theory” insisted on the heterogeneity of firms regarding
their exporting behaviour (Melitz 2003), thereby giving a theoretical foundation for the
presence of zero trade flows in data. In this line, Helpman et al. (2008) generalised the
empirical gravity equation by developing a two-stage estimation procedure that takes
into account extensive and intensive margins of trade. They showed that the incorrect

3
treatment of zero flows may lead to biased estimates and developed a complete
framework to provide a rationale for the existence of these flows.
Regarding the specification, Anderson and van Wincoop (2003) propose an augmented
version of the Anderson (1979) model based on the assumption of differentiation of
goods according to place of origin. Their main contribution is the inclusion of
multilateral resistance terms for the importer and the exporter that proxy for the
existence of unobserved trade barriers. This model is interesting overall to the extent
that the discussion of the multilateral resistance may matter for heteroskedasticity
considerations. In this model, countries are representative agents that export and import
goods. Goods are differentiated by place of origin and each country is specialised in the
production of only one good. Preferences are identical, homothetic and approximated by
a constant elasticity of substitution (CES) function.
The linear gravity equation estimated by Anderson and van Wincoop is as follows:
(
)
(
)
(
)
(
)
ijjiijijjiij
εPσPσbσdρσyykX +ln-1+ln-1+ln-1+ln-1+ln+ln+=ln
(1)
where X
ij
is the nominal value of exports from i to j; k is a positive constant, y
i
and y
j
are
the nominal income of each country, generally proxied by its GDP, and d
ij
is a measure
of the bilateral distance between i and j, which are introduced to proxy for transport
costs. b
ij
is a dummy variable that takes value one if two countries share a border.
Finally, the variables P
i
and P
j
are the multilateral resistance terms and are defined as a
function of each country’s full set of bilateral trade resistance terms. The variable of
interest for Anderson and van Wincoop is b
ij
since their objective is to estimate the trade
effect of national borders. They apply their equation to regional data.
The multilateral price indices (P
i
and P
j
) are not observed and should be estimated.
Anderson and van Wincoop (2003) use the observed variables in their model (distances,
borders, and income shares) to obtain the multilateral trade resistance terms. Assuming
symmetric trade costs, using 41 goods market-equilibrium conditions
1
and a trade cost
function defined in terms of
observables, they obtain the P
i
and P
j
terms. Although they
argue that this method is more efficient than any other, it is highly data consuming and
has not been frequently used by other authors.
An alternative solution is to include a remoteness variable to proxy for these multilateral
trade resistance indexes:
=
j
ROWj
ij
i
yy
d
Rm
)(
(2)
where the numerator would be the bilateral distance between two countries, and the
denominator would be the share of each country’s GDP in the rest of the world’s GDP.
Head and Mayer's (2000) remoteness variable describes the full range of potential
suppliers to a given importer, taking into account their size, distance and relevant costs
of crossing the border. Wei (1996), Wolf (1997), and Helliwell (1996) provide other
examples of regressions including a remoteness variable. Alternatively, Feenstra (2002)
proposes introducing importer and exporter fixed effects to account for the specific
country multilateral resistance term. The coefficient of the dummies for the importer
1
Their sample contains the same 30 US states and 10 Canadian provinces that McCallum
(1995) includes. There are 20 additional states, plus Columbia, which they aggregate into one.
Hence, they finally have 41 equations.

4
and the exporter should reflect the multilateral resistance for each country. Several
studies using this approach are described in the Appendix (Table A1). Finally, Baier and
Bergstrand (2009) suggest generating a linear approximation of the P
i
and P
j
terms by
means of a first-order Taylor series expansion.
Concerning the proxy for supply and demand sizes, the common practice is to use
importer’s and exporter’s GDP correspondingly. In some cases GDP per capita is also
introduced as a proxy for capital-labour intensities.
Transaction costs are frequently proxied by geographical distance. However, it is
commonly accepted that geographical distance may be a poor approximation
2
. Thus,
this variable is often completed with other proxies for trade barriers specified as
indicator variables. For instance, adjacency takes value one if trade partners share a
common border, common language takes value one if both countries share a language,
colonial links captures the effect of having had a common coloniser or having been
colonised by another country in the past; religion takes value one when both countries
have the same religion; access to water takes value one if a country has access to water,
or Regional Trade Agreement (RTA) which assess the effect of RTAs on trade. All
these factors affect international trade via transaction costs and complete the
geographical distance variable in order to reflect the economic distance.
3. Summary of estimation methods
As mentioned above, interest in the last years has focused on estimation methods to
accurately predict trade flows. In this section, a brief summary of some of the most
important estimation methods as well as a revision of related empirical literature (Table
1) are presented.
3.1. Linear methods
Since the logarithm of zero is not defined, truncation and censoring methods have been
proposed in the literature to treat the problem of zero flows in data. However, these
procedures reduce efficiency due to the loss of information and may lead to biased
estimates due to the omission of data. Furthermore, as Westerlund and Wilhelmsson
(2009) point out, the elimination of trade flows when zeros are not randomly distributed
leads to sample selection bias.
In addition, a panel framework permits recognising how the relevant variables evolve
through time and identifying the specific time or country effects. Over the last years,
researchers such as Egger (2000), Rose and van Wincoop (2001), Mátyás (1998), Egger
and Pfaffermayr (2003, 2004), Glick and Rose (2002), Brun et al. (2002), and Melitz
(2007) have turned towards panel data
3
. Two main techniques are employed to fit data
depending on the a priori assumptions. The fixed effects estimator assumes the
existence of an unobserved heterogeneous component that is constant over time and
which affects each individual (pair of countries) of the panel in a different way. By
contrast, the random effects model imposes no correlation between the individual
effects and the regressors, implicitly assuming that the unobserved heterogeneous
2
In addition, there is no single opinion about how distance should be measured. The most
common measures are the great circle formula and the distance between the two principal cities.
See Wei (1996), Wolf (1997), and Head and Mayer (2000) for further information.
3
See Appendix A for further information.

5
component is strictly exogenous. Under the null hypothesis of zero correlation, the
random effects model is more efficient. However, if the null is rejected, only the fixed
effects model provides consistent estimators
4
.
3. 2. Nonlinear methods
As Santos Silva and Tenreyro (2006) points out, the log-linearisation of the gravity
equation changes the property of the error term, thus leading to inefficient estimations
in the presence of heteroskedasticity. If the data are homoskedastic, the variance and the
expected value of the error term are constant but if they are not -as usually happens with
trade data-, the expected value of the error term is a function of the regressors. The
conditional distribution of the dependent variable is then altered and OLS estimation is
inconsistent. Heteroskedasticity does not affect the parameter estimates; the coefficients
should still be unbiased, but it biases the variance of the estimated parameters and,
consequently, the t-values cannot be trusted. Hence, the recent literature concerning
estimation techniques have opted to use nonlinear methods as well as two parts models
for estimating the gravity equation.
Among nonlinear estimation methods, the most frequently used are Nonlinear Least
Squares (NLS), Feasible Generalised Least Squares (FGLS), Heckman sample selection
model and Gamma and Poisson Pseudo Maximum Likelihood (GPML and PPML).
Santos Silva and Tenreyro (2006) claim that NLS is inefficient since it gives more
weight to observations with larger variance and is not robust to heteroskedasticity.
Martínez-Zarzoso et al. (2007) propose Feasible Generalised Least Squares (FGLS) as
the most appropriate model if the exact form of heteroskedasticity in data is ignored
since it weighs the observations according to the square root of their variances and is
robust to any form of heteroskedasticity. Manning and Mullahy (2001) propose Gamma
Pseudo Maximum Likelihood (GPML). In this case the conditional variance of the
dependent variable is assumed to be proportional to its conditional mean. This estimator
therefore assigns less weight to observations with a larger conditional mean. Martínez-
Zarzoso et al. (2007) computes the performance of this estimator, finding it to be
adequate in the presence of heteroskedasticity, although it shows less accuracy when
zero trade values are present. Finally, Poisson Pseudo Maximum Likelihood (PPML) is
similar to GPML, but assigns the same weight to all observations. Santos Silva and
Tenreyro (2006) point out that this is the most natural procedure without any further
information on the pattern of heteroskedasticity.
In addition, two-step estimation methods have also been proposed to estimate the
gravity equation. This is the case of Heckman sample selection model. In the first step,
a Probit equation is estimated to define whether two countries trade or not and in a
second step, the expected values of the trade flows, conditional on that country trading,
are estimated using OLS. In order to identify the parameters on both equations, a
selection variable is required. This exclusion variable should affect only the decision
process; hence, it should be correlated with a country’s propensity to export but not with
its current levels of exports. Some examples in the literature are the common language
and common religion variable (Helpman et al. 2008), governance indicators of
regulatory quality (Shepotylo 2009), or the historical frequency of positive trade
4
The Hausman test provides a method for testing the adequacy of the random effect model. If
the null is rejected, the random effects model is not consistent. However, it is important to note
that this result does not imply that the fixed effect model is adequate.

Citations
More filters
Posted Content
TL;DR: The authors fit a gravity model to the trade of 76 market economies and applied the model to data on East European economies to estimate what their trading potential might have been, had behaved like market economies in the mid-1980s.
Abstract: This paper fits a gravity model to the trade of 76 market economies. It then applies the model to data on East European economies to estimate what their trading potential might have been, had behaved like market economies in the mid-1980s. At existing levels of national income, the liberalization of Eastern Europe and the Soviet Union is unlikely to affect their mutual trade and trade with developing countries, but it will increase trade with industrial counties by factors of three to thirty. West Germany and the USA are the principal beneficiaries of this new trade, increasing their exports and imports by over 20%. Trade must flow both ways, however: the West cannot increase its exports to the East without correspondingly increasing its imports.

281 citations

Posted Content
TL;DR: In this article, the authors extend McCallum's (1995) result, based on a gravity model of 1988 trade flows, that a typical Canadian province trades 22 times more with other provinces than with US states of similar size and distance.
Abstract: Extending McCallum's (1995) result, based on a gravity model of 1988 trade flows, that a typical Canadian province trades 22 times more with other provinces than with US states of similar size and distance, this paper asks how Quebec trade patterns compare with those of other provinces The results, based on revised data for 1988, 1989 and 1990, show that while the typical province trades more than 20 times as much with other provinces as with comparable US states, for Quebec the multiple is even greater Thus trade between Quebec and the United States appears to be an even less viable alternative to interprovincial trade for Quebec than it is for the rest of Canada The implications of these results for international economics are considerable, as they show that trade linkages within a national economy are far greater than previously imagined If these results are confirmed, they imply that the fabric of national economies is far tighter than that of the global trading system, even for countries operating without substantial trade barriers

275 citations

Journal ArticleDOI
01 Apr 2019
TL;DR: In this article, the authors focus on the global shipping network, which is central to invasive species spread worldwide, and predict invasion risk to surge in middle-income countries, particularly in Northeast Asia.
Abstract: Socioeconomic factors, including population growth, global trade and the worldwide transport of materials, interact with environmental drivers to determine the sustainability of natural systems. We focus on the global shipping network, which is central to invasive species spread worldwide. We explain 90% of the variation in global shipping traffic and a twofold increase in shipping using basic socioeconomic indicators and a temporal validation set. Combining our model with global economic development scenarios, we project global maritime traffic to increase by 240–1,209% by 2050. Integrating our predictions with global climate change projections and shipping-mediated invasion models, we forecast invasion risk to surge in middle-income countries, particularly in Northeast Asia. Shipping growth will have a far greater effect on marine invasions than climate-driven environmental changes: while climate change might actually decrease the average probability of invasion, the emerging global shipping network could yield a 3- to 20-fold increase in global invasion risk. Shipments of natural resources and goods connect distant regions but sometimes move more than their intended cargo. This study models the growth of the global shipping network and the implications for spreading invasive species in a changing climate, forecasting substantial increases in ship movements and a 3- to 20-fold increase in invasion risk in coming decades.

201 citations

Journal ArticleDOI
TL;DR: In this paper, an empirical model that takes into account both that preferential trade agreement membership is endogenous and that the world matrix of bilateral trade flows contains numerous zero entries is presented by means of (possibly two-part) Poisson pseudo-maximum likelihood estimation with endogenous binary indicator variables in the empirical model.
Abstract: Recent work by Anderson and van Wincoop (2003) establishes an empirical modeling strategy which takes full account of the structural, non-(log-)linear impact of trade barriers on trade in new trade theory models. This framework has never been used to evaluate and quantify the role of endogenous preferential trade agreement (PTA) membership for trade. Apart from paying attention to structural modeling of the impact of trade policy on trade, this paper aims at delivering an empirical model which takes into account both that preferential trade agreement membership is endogenous and that the world matrix of bilateral trade flows contains numerous zero entries. These features are treated in an encompassing way by means of (possibly two-part) Poisson pseudo-maximum likelihood estimation with endogenous binary indicator variables in the empirical model.

197 citations

Journal ArticleDOI
TL;DR: In this article, the authors analyzed different market channels for sweet pepper in Thailand and found that there is a general preference for marketing options that do not involve a contract and that additional provision of inputs and credit can increase the attractiveness of contracts.

174 citations

References
More filters
Journal ArticleDOI
TL;DR: In this article, the bias that results from using non-randomly selected samples to estimate behavioral relationships as an ordinary specification error or "omitted variables" bias is discussed, and the asymptotic distribution of the estimator is derived.
Abstract: Sample selection bias as a specification error This paper discusses the bias that results from using non-randomly selected samples to estimate behavioral relationships as an ordinary specification error or «omitted variables» bias. A simple consistent two stage estimator is considered that enables analysts to utilize simple regression methods to estimate behavioral functions by least squares methods. The asymptotic distribution of the estimator is derived.

23,995 citations

Journal ArticleDOI
TL;DR: This paper developed a dynamic industry model with heterogeneous firms to analyze the intra-industry effects of international trade and showed how the exposure to trade will induce only the more productive firms to enter the export market (while some less productive firms continue to produce only for the domestic market).
Abstract: This paper develops a dynamic industry model with heterogeneous firms to analyze the intra-industry effects of international trade. The model shows how the exposure to trade will induce only the more productive firms to enter the export market (while some less productive firms continue to produce only for the domestic market) and will simultaneously force the least productive firms to exit. It then shows how further increases in the industry's exposure to trade lead to additional inter-firm reallocations towards more productive firms. The paper also shows how the aggregate industry productivity growth generated by the reallocations contributes to a welfare gain, thus highlighting a benefit from trade that has not been examined theoretically before. The paper adapts Hopenhayn's (1992a) dynamic industry model to monopolistic competition in a general equilibrium setting. In so doing, the paper provides an extension of Krugman's (1980) trade model that incorporates firm level productivity differences. Firms with different productivity levels coexist in an industry because each firm faces initial uncertainty concerning its productivity before making an irreversible investment to enter the industry. Entry into the export market is also costly, but the firm's decision to export occurs after it gains knowledge of its productivity.

9,036 citations

Posted Content
TL;DR: This article showed that the gravity model usually estimated does not correspond to the theory behind it and showed that national borders reduce trade between the US and Canada by about 44% while reducing trade among other industrialized countries by about 30%.
Abstract: The gravity model has been widely used to infer substantial trade flow effects of institutions such as customs unions and exchange rate mechanisms. McCallum [1995] found that the US-Canada border led to trade between provinces that is a factor 22 (2,200%) times trade between states and provinces, a spectacular puzzle in light of the low formal barriers on this border. We show that the gravity model usually estimated does not correspond to the theory behind it. We solve the 'border puzzle' by applying the theory seriously. We find that national borders reduce trade between the US and Canada by about 44%, while reducing trade among other industrialized countries by about 30%. McCallum's spectacular headline number is the result of a combination of omitted variables bias and the small size of the Canadian economy. Within-Canada trade rises by a factor 6 due to the border. In contrast, within-US trade rises 25%.

6,043 citations

Journal ArticleDOI
TL;DR: In this article, a method that consistently and efficiently estimates a theoretical gravity equation and correctly calculates the comparative statics of trade frictions was developed to solve the famous McCallum border puzzle.
Abstract: Gravity equations have been widely used to infer trade flow effects of various institutional arrangements. We show that estimated gravity equations do not have a theoretical foundation. This implies both that estimation suffers from omitted variables bias and that comparative statics analysis is unfounded. We develop a method that (i) consistently and efficiently estimates a theoretical gravity equation and (ii) correctly calculates the comparative statics of trade frictions. We apply the method to solve the famous McCallum border puzzle. Applying our method, we find that national borders reduce trade between industrialized countries by moderate amounts of 20-50 percent.

4,997 citations

Journal ArticleDOI
TL;DR: In this paper, the gravity equation for trade was used to provide new estimates of this equation, and significant differences between the estimated estimator and those obtained with the traditional method were found.
Abstract: Although economists have long been aware of Jensen's inequality, many econometric applications have neglected an important implication of it: the standard practice of interpreting the parameters of log-linearized models estimated by ordinary least squares as elasticities can be highly misleading in the presence of heteroskedasticity. This paper explains why this problem arises and proposes an appropriate estimator. Our criticism to conventional practices and the solution we propose extends to a broad range of economic applications where the equation under study is log-linearized. We develop the argument using one particular illustration, the gravity equation for trade, and apply the proposed technique to provide new estimates of this equation. We find significant differences between estimates obtained with the proposed estimator and those obtained with the traditional method. These discrepancies persist even when the gravity equation takes into account multilateral resistance terms or fixed effects

4,492 citations


"Comparing alternative methods to es..." refers background or methods or result in this paper

  • ...On the other hand, PPML shows the lowest coeffi ients; a result that is in line with Santos Silva and Tenreyro (2006) and Siliverstov and Schumacher (2009)....

    [...]

  • ...As Santos Silva and Tenreyro (2006) points out, the log-linearisation of the gravity equation changes the property of the error term, thus leading to inefficient estimations in the presence of heteroskedasticity....

    [...]

  • ...…A2 Articles related to the problem of zero-flows and heteroskedasticity Article Data Estimation methods Dependent variable Simulation studies Santos Silva and Tenreyro (2006) 136 countries; 1990 PPML, NLS, GPML, OLS, ET-tobit, OLS (y > 0.5) OLS (y+1) Trade - PPML, NLS, GPML OLS; OLS(y + 1);…...

    [...]

  • ...Santos Silva and Tenreyro (2006) claim that this is the preferred estimation method in the presence of heteroskedasticity....

    [...]

  • ...Santos Silva and Tenreyro (2006) point out that this is the most natural procedure without any further information on the pattern of heteroskedasticity....

    [...]

Frequently Asked Questions (2)
Q1. What are the contributions mentioned in the paper "Comparing alternative methods to estimate gravity models of bilateral trade" ?

This paper provides a survey of the most recent literature concerning the specification and estimation methods of this equation. 

This paper has provided an in-depth review of recent developments in the literature on estimation methods for the gravity equation, finding that there are at least two problems related to the log linearisation of the gravity equation that require further research as there is no consensus about the optimal method to solve them. This paper suggests that the Heckman sample selection model is the preferred estimation method within nonlinear techniques when data are heteroskedasticity and contain a significant proportion of zero observations.