The Impact of Services Trade Restrictiveness on

Food Trade

Zongo Amara∗

April 2020

Abstract
International trade in goods requires service inputs such as transport, banking and
financial services for production and transportation. Trade in goods and services are now
closely linked and contribute to the growth of international trade. The goal of this study
is to investigate the effects of restrictions in the banking, accounting, transportation and
logistics sectors (cargo handling and custom brokerage) on food trade. We use a gravity
model with panel data from 2014 to 2018 for 36 OECD countries, the OECD indices of
individual country restrictions and regulatory difference by country pair to capture the
level of restrictions in these sectors. Our results suggest that importing and exporting
country restrictions have non-significant effects on aggregate food exports, but negative
and significant impacts on exports of agricultural raw materials and perishable products
(meat, dairy products, eggs, etc.). The regulatory disparity between countries in logistics
and banking sectors emerge as the main barrier for food exports. However, these results
can be mitigated through regulatory cooperation or harmonization of regulations.

Keywords: Food trade, service sector, heterogeneity of standards, gravity model.

JEL Classification: F13, F14, K23, Q17.

∗ University
of Brodeaux, Office R182, +33 768 908 623, LAREFI - Bâtiment Recherche Économie
[email protected], avenue Léon Dugit, 33608 Pessac

Electronic copy available at: https://ssrn.com/abstract=3644987

 I. Introduction
Following the GATT agreements in 1948 (General Agreement on Tariffs and Trade),
tariffs have been significantly decreased in world trade. For decades, high tariffs were
the main obstacle to trade. However, under many implemented trade agreements
(multilateral, bilateral and regional), tariffs have fallen to low levels (the simple average
world tariff rate declined from 10.13% in 2000 to less than 7% in 2015, cf. figure 1)1 . In
the meantime, we observe a huge increase of non-tariff barriers (NTBs), particularly
production standards, which represent the main obstacle to global economic growth,
(Kee et al., 2009; IMF, 2017).

11

10

5
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018

Figure 1: Tariff rate, applied, simple mean, all products (%)

Considered as rules that establish production characteristics and procedures, they

are intended to optimize product reliability. The most restrictive of these standards
remain the SPS and TBT measures, which considerably affect agricultural production.
Indeed, when each country imposes its own market access standards, exporters and
suppliers must comply with them. However, compliance with these technical, health and
quality regulations generates in most cases high fixed costs (cost of product adaptation),
WTO (2005). It is not just adapting the products, but also having necessary equipment,
technology and skills. Therefore, the existence of fixed costs may affect the decision to
export, Riker (2014). For some authors, the level or stringency of standards is not an
obstacle, but the regulatory heterogeneity between countries appears to be restrictive,
Kox and NordÅs (2007). However, it is not the level of regulation that discourages
foreign suppliers, but rather the difference in regulation between origin and destination
markets. Thus, additional compliance costs to establish a firm in a foreign country
would be minimal if standards and qualifications were recognized in the home country.

For services, the consequences of this disparity remain very significant, UNCTAD (2010).
Indeed, the regulatory environment in the services shows a sector highly regulated
1 Data provided by the World Bank through the World Integrated Trade Solution database.

Electronic copy available at: https://ssrn.com/abstract=3644987

 because faced of increased competition. As for goods sector, services are also affected
by national regulations and trade in services is more impacted by these restrictions
than trade in goods, Kox and NordÅs (2007). Moreover, for goods and commodities,
production and quality requirements only apply to goods. For services, this includes the
supplier, its foreign personnel and equipment. The regulatory restriction on services not
only prevents foreign suppliers to access domestic service markets, but can also deter
them from undertaking further investment after establishment in the market, so-called
regulation behind borders. Regulatory divergences between nations in this sector gen-
erate not only fixed costs, but also maintenance costs or operational restrictions2 . To
summarize, the disparity of standards between nations forces service providers to adapt
their business model in each export market and as a firm’s business model is often the
vector of its competitiveness, these regulatory barriers represent a high cost for firms.

This paper examines the different effects of services restrictions on food trade and
discusses how to mitigate these restrictive impacts 3 . Our study contributes to the
literature about the impacts of non-tariff barriers on international trade. It differs from
the existing literature because we focus on the impact of restrictions and regulatory
disparity between OECD countries in the banking, accounting, transports and logistics
sectors on food exports, a topic has not been studied in the literature. Indeed, the
literature has focused on the effects of services restrictions on trade in services (NordÅs
and Rouzet (2016); Ingo and al. (2012)), and no study has so far examined food trade.
These sectors, considered as service providers, have a strong and close link with food
export activities. In addition, Figure A.2 shows significant restrictions in the banking,
accounting and transport sectors.

International trade is a risky activity, importers may not pay after receiving the goods
and exporters may not deliver if they pay in advance. To reduce the risk of international
trade, banks offer specific trade finance products4 : the most important are letters of
credit (LC) and documentary collections, Amiti and Weinstein (2011)5 , Paravisini and al
(2014)6 . According Copeland and Mattoo (2007), the development of financial services
can promote the effectiveness of service sectors by competition and thus support eco-
nomic growth.

The global financial crisis is an example of a strong impact of credit on trade (see figure
A. 4). However, the credit crunch led to a significant reduction in the availability of
external financing, thus reducing firms’ production and export capacities. The weak
economic outlook resulted in slower global demand and imports. Using monthly US
import data, Chor and Manova (2012) find that countries with higher interbank rates
and stricter credit conditions less exported to United States during the crisis. These
effects have been particularly pronounced in sectors that require significant external
financing. Moreover, Bricongne et al (2010) find that French firms’ exports in sectors
2 Maintenance costs include: costs related to the tax burden, the social security system, limiting the variety
of services, imposing fixed prices for certain services.
3 We will focus on the effects of regulatory difference on food trade.
4 We use trade finance to refer to formal borrowing by firms from banks or other financial institutions to

facilitate international trade activities.

5 They showed that, in Japan, firms linked to under-performing banks reduced their exports.
6 They found that reduction of credit supply to firms led to decline of exports in Peru.

Electronic copy available at: https://ssrn.com/abstract=3644987

most dependent on external financing have been more affected by the recent global
crisis. Using time-varying bilateral data on foreign bank ownership for many countries,
Claessens, Hassib and Horen (2017) show that a greater domestic presence of foreign
banks, associated with an increase of exports in sectors more dependent on external
financing. Another alternative to letters of credit is trade credit insurance, Fabrice Morel
(2010). According to Moore (2000), a large number of factors such as good prudential
regulation, supervision, or appropriate legal and accounting systems can be a key
condition for maintaining open trading systems. However, restrictions in the banking
sector can constitute very significant fixed costs and can affect agri-food exports.

Transport and logistics are also major factor in international trade. Choosing the appro-
priate mode of transport and logistics is essential to ensure efficient and cost-effective
import or export operations, so-called "transit time". It provides vital distribution for
production, as well as essential personal mobility, by connecting firms to global markets.
Transport is a crucial element for economic growth and competitiveness, the most
important of which are maritime, road and air transport (cf. figure A.3). We use food
products because affected by SPS and TBT standards and require important services as
inputs.

To conduct this study, we use a gravity model based and OECD restrictiveness index,
which measures the recent level of restriction in services sector. The latter is a robust
measure compared to the World Bank index because it is a sectoral index (covering 22
sectors in 46 countries), while the World Bank index includes 5 service sectors in 103
countries. Our results suggest that individual country restrictions have non-significant
effects on aggregate food exports, but negative and significant impacts on exports of
agricultural raw materials and perishable products (meat, dairy products, eggs, etc.).
The restrictions implemented by the exporting country have significant negative effects
on exports higher than the importing country’s restrictions. Regulations in the logistics
and banking sectors are restrictive for food trade.

The remainder of this paper is structured as follows: in the first part, we review the
literature on the effects of these restrictive measures on trade flows. In the second part,
we describe our econometric model with data, sources, and the type of regression used.
The last section presents our different results.

II. Literature review

In this section, studies that examined the impact of restrictive policies on international
trade used two essential methods: the analysis on restriction index and analysis on
tariff equivalent. Indeed, the economic literature that studied the impact of standards
in services on trade, used the OECD/World Bank Trade Restrictiveness Index, and the
computation of tariff equivalent.

Electronic copy available at: https://ssrn.com/abstract=3644987

 i. Service Trade Restrictiveness Index (STRI) and service trade:
Gravity Analysis

The existing literature on service restrictions and trade is exclusively empirical. To

evaluate the effects of regulatory barriers in services on international trade measured by
sectoral STRI index, we use gravity model. Although specific to trade flows in goods
and commodities, some authors applied it to services and found it adapted to trade
in services, Head et al, 2009; Walsh, 2008. However, Kox and NordÅs (2009) using a
gravitational approach, examine trade flows in transport and business services, and
their interaction with an overall regulatory indicator. The analysis of Kox and NordÅs
(2007) considers financial services and other business services in their model.

The authors that apply the gravity model to analyze the impacts of STRI index and
the regulatory heterogeneity index on service trade are NordÅs and Rouzet (2016),
NordÅs (2016). Based on a gravity model with aggregate data and the PPML (Pseudo-
Maximum Likelihood Estimator) as the estimation method, they find that the most
restrictive countries in the service sector, import and export significantly less services.
In addition, the negative impact of restrictions in services on exports is about twice as
large as on imports. The most affected sectors are the banking, financial and transport
sectors, considered as service providers. Considering the regulatory disparity between
countries, they finds that regulatory heterogeneity in services has negative impacts
on cross-border trade in services. In this case, countries trade more with partners
with similar regulations. A low heterogeneity index (harmonization or convergence of
standards) is associated with a strong stimulation of services trade. According to their
study, if the STRIs of importer and exporter countries are low, harmonization stimulates
trade in services, but if the STRIs are high, harmonization attempts to limit trade.

Another approach that differs from the first is the analysis of Ingo and al. (2012). They
use the restrictiveness index developed by the World Bank, and not the OECD measure,
to capture the impact of regulatory policies on trade in services7 . Through the PPML
estimate, they find that higher levels of STRI discourage investment. Van der Marel and
Shepherd (2013) in his analysis (very similar to the previous one) also finds a negative
relationship between the World Bank’s bilateral restrictiveness index and cross-border
trade in transport and financial services. Riker, D (2014), in his study highlights the
impact of restrictions on foreign suppliers (import restrictiveness index) and cross-border
trade in services. He also finds negative effects of the latter on cross-border trade in
services. Further, it simulates the effect on U.S. financial services exports if its trading
partner eliminated restrictions on these imports from all countries. He notes that, while
China and India do not apply any barriers to market entry, the United States is recording
a significant rise of its financial services exports, both in dollars ($186.0 million and
$42.2 million) and rate change (10.14% and 3.76%). On the other hand, in a country
like Germany, US exports have increased slightly (7.7 million dollars or 0.23%). Indeed,
according to him, in the financial services sector, Germany is a relatively important
export market for the United States, after the United Kingdom, but the impact on trade
is less, because the level of restrictiveness in this country is relatively low compared to
7 Foreign investment inflows and access to financial services through the provision of bank lending.

Electronic copy available at: https://ssrn.com/abstract=3644987

 countries like China or India.

Another analysis that differs from those mentioned above and that is part of our paper
remains Ariu and al (2018). They explore the interaction between international goods
trade and restrictions on services. They consider data from Belgian firms from 1995 to
2005, data on PMR index (Product Market Regulation) and customs duties on goods and
services. They come to the following conclusion: when import barriers for goods and
commodities rise, we have a decrease of firms’import of services. Further, these authors
use their results to quantify the impacts of lowering barriers to goods and services on
trade between US and EU. They find that liberalization of services sector has direct and
significant effects on goods trade.

Our paper also contributes to recent literature on firm-level trade in services (Breinlich
and Criscuolo, 2011; Crozet et al, 2016; Ariu, 2016; Conti et al., 2010; Gaulier et al., 2011;
Walter and Dell’mour, 2010). These studies describe characteristics of firms exporting
services and find that very few firms are able to export services due to regulatory
barriers in the market.

ii. Analysis of the Tariff Equivalent of Restrictive Policies in Ser-

vices
The literature formerly discussed shows that the restrictions of services has negative
effects on international trade. However, we do not specify the transmission channels
through which these negative effects work. At this level, we assume that all restrictions
on trade in services can be converted into tariff equivalents, Whalley (2004).

In this section, two methodologies are used : the analysis by the residues obtained from
the gravity equation, Park (2000) and the fixed effects method of importing countries 8 ,
Fontagné, Guillin and Mitaritonna (2011); Guillin, A (2013) and Fontagné, Mitaritonna
and Signoret (2016). For them, the difference in trade flows between the presence of
market entry restrictions and so-called "Benchmark" restrictions explains these tariffs 9 .
In other words, tariff equivalents are deducted by comparing importing country fixed
effects or residues of model with restrictions against "benchmark" restrictions. They
come to the same conclusion that developed countries have low levels of restrictions
in their markets and developing and emerging countries apply high tariffs on services.
Although this methodology has data advantages (to fill data gaps), it also has important
weakness: this method implies that the unobserved importer-level demand factors, other
than GDP and price index measures, are captured by the trade barrier estimations. In
this case the fixed effects of importing countries do not only consider non-tariff barriers
but also the demand factors of the importing country, Deardorff and Stern (2007). The
computation of tariff equivalents using the importing country’s fixed effects not reflect
the correct tariff computation because the latter covers all unobservable events and not
just restrictions. Therefore, it is better to use the residual method.

In order to provide solutions to omitted variables bias in previous analyses, Gooris

8 extension of Park (2002).
9 The"benchmark" restrictions are compared to countries with the largest positive difference between the
average effective value of imports and the average expected value.

Electronic copy available at: https://ssrn.com/abstract=3644987

 and Mitaritonna (2015) add importing country price index to the classical variables.
They compute the tariff equivalents by assessing the ratio of service flows at a given
restrictiveness level over benchmark flows without restrictions.

The second-analysis of computation of tariff equivalents, which differs from the first, is
based on a standard econometric model. Authors such as Jafari and Tarr (2017); Rouzet
and Spinelli (2016) and Khachaturian (2015) compute the tariff equivalents of restrictive
policies through a price-cost margin of foreign firms. They use the price performance
index as a variable to capture tariffs for restrictive policies in services. According to
their interpretation, a positive impact increasing the price-cost margin is considered as
a consumption tax (paid by the consumer and seen as a rent for the producer) and a
negative coefficient of STRI index on price-cost margin is production tax, supported
by the producer and not imposed on the consumer. Moreover, after the econometric
analysis, they compute the tariff equivalent of regulation by differentiating the price
with restrictions from price without restrictions. Their results suggest that in all sectors,
high-income countries have lower tariff equivalents than emerging and least developed
countries. However, the analysis of tariff equivalent by prices is very limited, because it
is necessary to identify the appropriate prices to use and this is likely to be problematic.
While it is fairly easy to obtain information on the price paid by the importers of a
good, it might become difficult to obtain the domestic market price especially at a
disaggregated level 10 .

The review’s analysis shows a negative impact of STRI index on trade flows, and
significantly affects the banking and transport sectors. The divergence of regulations
between countries has a significant negative impact on trade. Our work is a extension of
previous studies that addressed the issue of standards in international trade.

III. Gravity Model of Bilateral Trade

In order to conduct an empirical analysis of effects of restrictive measures on trade flows,
we use gravity model developed by Anderson (1979). It is based on the assumption
that products are differentiated by the origin of country, i.e. by location (Armington’s
hypothesis), where consumers have defined preferences for these differentiated products.
In this approach, each country can import a good from another country at any market
price. In this case all goods are traded, all countries trade and in balance, national
income is the sum of domestic and foreign demand for the single good produced. In its
model, trade costs are considered as transport costs.
After Anderson’s theoretical approach, authors started from trade theories to find a
theoretical framework for the gravity equation. Bergstrand (1989) shows that a gravity
equation remains a direct implication of a trade model based on monopolistic compe-
tition by Paul Krugman (1980). Identical countries manage to exchange differentiated
goods because consumers prefer variety. According to concept of monopolistic competi-
tion, it is not the location of firms that determines differentiated goods trade but the
preference of consumers for variety. Eaton and Kortum (2002) derive a gravitational
equation from a Ricardian model, and Helpman and al (2008); Chaney, T (2008) obtained
it from a theoretical model of international trade by firm heterogeneity.
10 This becomes even more difficult if data collection has to be done for a large set of countries.

Electronic copy available at: https://ssrn.com/abstract=3644987

The general formulation of the gravity equation is as follows:

Xij =GSi M j Φij

Where Xij is the value of exports from country i to country j, M j represents demand of
importing country (importing country’s GDP), Si is the value of exporting country’s
GDP, G is a variable that does not depend on i or j and represents the level of global
liberalization, Φij represents the ease of access by exporter i to market j.

Anderson and van Wincoop’s (2003) show that the control of trade costs remains
crucial in order to properly specify the gravity equation. However, trade costs are very
important for the gravity equation. Two countries will trade less if they were separated
by an ocean or by vast stretches of deserts and mountains. Trade between two nations
for this purpose is determined by relative trade costs, i.e. trade costs between the two
nations (absolute costs) and trade costs between the country (importer, exporter) and
the rest of the world, which will be called the MTR (Multilateral Trade-Resistance).
However, the multilateral resistance can be controlled through the time fixed effects of
importing and exporting country, (Anderson and Yotov, 2012) or using a proxy.
To estimate this equation, we need to linearize it. Using the logarithm of each variable
in the model, the equation becomes :

LnXij = a0 + a1 lnYi + a2 lnYj + a3 lntij + a4 lnΠi + a5 lnPj + eij (1)

Where a0 is the constant, a3 = 1- σ, Xij is the value of exports from country i to country
j, Yi and Yj the GDP of exporting and importing countries, tij bilateral costs between
our pairs of countries, Πi terms measuring barriers to trade between each country and
the rest of the world, Pj the price index of importing country, eij is the error term. In
practice, the gravity equation links the logarithm of monetary value of trade between
two countries to logarithm of their respective GDPs, a composite term reflecting barriers
and trade incentives between these two countries, and terms measuring barriers to trade
between these countries and rest of the world.

IV. Individual STRIs and Food Exports : Augmented gravity

models

Using Anderson and Van Wincoop’s (2003), Anderson and Van Wincoop (2004), our
baseline regression equation is the following:

Xij,t = exp[ β 0 + β 1 tij + β 2 Zij,t + β 3 STRIi,kt + β 4 STRIj,kt + µit + γ jt + αt + eijt ] (2)

Where (i) and (j) denote trading partners, (t) to is year.

Xij,t denotes nominal food exports from exporter (i) to importer (j) in year (t). We use
nominal exports and not those deflated by U.S. aggregate price indexes to avoid bias

Electronic copy available at: https://ssrn.com/abstract=3644987

 problems. Since there are global trends in inflation rates, inclusion of this term probably
creates biases via spurious correlations, Baldwin and Taglioni (2006). tij is the vector of
time-invariant bilateral control variables (bilateral distance, common language, common
border) 11 , Zij,t is a time-variant vector of bilateral variables. We have Real Effective
Exchange Rate (REER), binary variable equals 1 if i and j share the same regional
trade agreement (RTA) and 0 otherwise 12 . STRIi,kt and STRIj,kt are the STRI indices
of respectively the exporter and the importer on a scale of 0 to 1, they represent our
explanatory variables of interest and capture the level of restrictions in exporter and
importer countries in the sector k (banking, accounting, transport and logistics sectors),
µit and γ jt are other variables that vary respectively of exporting and importing country
in year t, αt is a year fixed effect (capture the global macroeconomic cycle) and eijt is a
error term.

V. Data sources
As mentioned above, our paper attempts to analyse the effects of restrictive measures
in services on food trade flows. Indeed, we use panel data on trade in food products
between 36 OECD countries (bilateral trade between countries) from 2014 to 2018. In-
deed, panel data have the advantage of reducing the bias generated by heterogeneity
between countries. While in a cross-section, trade between pairs of countries can only
be controlled by the observed characteristics of pairs of countries (such as common lan-
guage, common border). In a panel the heterogeneity of country pairs can be controlled
using country pair fixed effects. The data for 2014-2018 are based on the implementation
and evolution of the STRI index. We focus on food products to examine the impacts of
services restrictions on this sector, so it is heavily affected by sanitary and phytosanitary
(SPS) standards and technical regulations (TBT).

To perform this study, we use informations about bilateral food exports (annual fre-
quency) from United Nations Conference on Trade and Development database (UNC-
TADstat) which uses the Classification Standard International Trade (SITC Rev.4)13 . The
choice of service sector is essential because it accounts for almost 2/3 of world GDP
and nearly 80% of total employment in OECD countries. We consider the banking,
accounting, transport and logistics sectors due to their significant role in the movement
of goods and commercial presence Ariu and al (2018). We also introduce an index
that captures the level of restrictions in services sector (STRI). The STRI Index provides
a database of regulations affecting trade in 22 service sectors in 46 countries14 . For
each sector the database covers 5 policy areas: restrictions on the entry of foreigners,
restrictions on the movement of persons, other discriminatory measures, barriers to
competition and regulatory transparency. The qualitative information on these 5 areas
has been converted into quantitative indices by sector ranging from 0 to 1 (where 0
corresponds to no restrictions and 1 to a sector completely closed to service providers).
11 Dummy variables equal 1 if countries share a common border, comman language and 0 otherwise.
12 We include dummy variables that represent European Economic Area (Intra EEA) to control the deeper
integration in services, and NAFTA, which is a major trade agreement on commodities.
13 Data can be accessed at: https://unctadstat.unctad.org/wds/TableViewer/dimView.aspx
14 36 OECD countries plus Brazil, People’ s Republic of China, Colombia, Costa Rica, India, Indonesia,

Malaysia, Russia, South Africa and Thailand

Electronic copy available at: https://ssrn.com/abstract=3644987

 Data are available on the OECD STRI database15 . The data of real effective exchange
rate are extracted of WORLD BANK database. Bilateral resistance variables such as the
bilateral distance between the two capitals, common border, and language from CEPII
database, binary variables that materialize regional trade agreements16 extracted from
WTO (Regional Trade Agreements Information System, RTA-IS).

VI. Econometric Issues

Several questions related to estimation of standard gravity equation are considered
in our study. Recently, researchers identified three problems inherent in gravitational
models: zero trade flows, terms of multilateral trade resistance and endogeneity issue.
The estimation of gravity equation will be conducted with an OLS estimator. However,
the results of this estimator may constitute a bias in the presence of "Zero trade" also in
the presence of heteroscedasticity, OLS estimation may not be consistent. Indeed, this
estimator not including countries not trading with each other, compromises our results,
because zero trade reveals crucial information (lack of information, high transport costs,
landlocked countries) so omitting it can constitute a considerable bias in our study17 . If
we look at our database, zero trade represents less than 5% of all trade. But the question
of heteroskedasticity leads us to consider better estimator.

One way to deal a zero trade issue consists of using a two-step estimation procedure.
The decision to export is evaluated in the first step, while the second step focuses
on the value of exports. Heckman’s model appears to be the best in the literature.
However, in the presence of fixed effects in the first stage, the Heckman model leads to
the incidental parameter problem. Helpman et al. (2008) also developed a two-stage
estimation procedure that focuses both on extensive estimation (export decision from i
to j) and intensive margins (export volume) of trade. While this approach offers a better
understanding of the determinants of trade flows, it provides biased estimates in the
presence of heteroscedasticity in the trade data, Santos Silva and Tenreyro, 2013. To
avoid biased estimation results, we use the Poisson estimator suggested by Silva and
Tenreyro (2006) 18 . The PPML is used in our case in order to face the constraints of zero
trade between States, it also estimates the non-linear shape of the gravity model in the
presence of heteroscedasticity. However, important assumption of PPMLs estimator is
equidispersion, which means the conditional variance of dependent variable and its
conditional mean are equal. PPML estimation can be assessed by solving the following
condition:
∑(X p -exp(Z p β)) = 0 (3)
p

where p is the country pairs, X p is the unilateral trade (i.e. exports or imports) between
country pairs in non-logarithmic levels and Z p is complete vector of gravity equation as
defined above.
15 https://stats.oecd.org/Index.aspx?DataSetCode=STRI.
16 We use trade agreements on both goods and services, as we study the effects of restrictions in services on
food products.
17 Indeed, zero commerce is associated with high bilateral fixed costs of trade.
18 Pseudo-Maximum Likelihood Estimator (PPML).

10

Electronic copy available at: https://ssrn.com/abstract=3644987

 To obtain a consistent estimate, the literature stipulates to include both exporter-year
and importer-year fixed effects in the estimations, to take account inward and outward
multilateral resistance, Feenstra, 2004; Baldwin and Taglioni 2006; Baier and Bergstrand
2007. This method is not feasible in our study because country fixed effects would
absorb the effect of STRI and preclude us from dissociating its impact from that of other
country-specific characteristics. Indeed, these fixed effects consider multilateral trade
resistance terms (MTRs) and size effects such as severity, but also other determinants
of origin and destination country that are not considered in the specific mode19 . This
method will be considered in section 8, where we will use the index of regulatory
heterogeneity between pairs of countries with importer and exporter year fixed effects.
To solve the endogeneity issue between our dependent variable and the interest variable
STRI20 , literature suggests using an instrument for service trade restrictiveness. Section
10 discusses the endogeneity issue.

VII. Empirical Results

We estimate the effects of restrictions in 6 service sectors on food trade using a gravity
model with PPML as estimate. The results are presented in Tables A.4 to A.7. Table
A.4 shows the results on overall food trade (columns 1 to 6). In other cases, we have
the results on disaggregated food trade according to the Standard International Trade
Classification (SITC) for disaggregated trade flows.
In the different specifications, we find the following effects: our dummy variables repre-
senting the common border, regional trade agreements, Intra-EEA, and and importer’s
and exporter’s GDP have positive and significant impacts on food trade between OECD
member countries. Bilateral distance have restrictive effects on trade flows (significant
result), Disdier and Head, 2008, Head and Mayer, 2014. The real effective exchange rate
has not significant effect on trade flows. We think that the bilateral exchange rate does
not have significant effects, but exchange rate volatility has significant negative effects
on trade flows, Jong Woo Kang and Suzette Dagli (2018).
If we look at our control variables, service restrictions in both importing and exporting
countries have no significant effects on aggregate food trade. These effects become
significant and negative when we consider bilateral trade in disaggregated food prod-
ucts. A careful analysis shows that the products most affected by its restrictions remain
agricultural raw materials and perishable products (meat, dairy products, eggs, etc.).
The restrictions of the exporting country have a greater limitative effect on disaggregated
food exports. A higher STRI of the exporting country is associated with significantly
lower exports. In services, restrictions in logistics customs brokerage, accounting, bank-
ing and road transport impact food exports (respectively significant at least at the 10, 5
and 1 per cent level in customs brokerage, banking, accounting and road transport on
perishable products and least at 1 per cent on agricultural raw materials).
Banking and accounting activities have a very key role in the export of food and man-
ufacturing products through commercial presence. Exporters have less incentive to
19 Wehave :GDPs, population, most favored nation (MFN), tariffs.
20 Moreopen economies are encouraged to implement more liberal policies, which creates reverse causality
between these policies and trade.

11

Electronic copy available at: https://ssrn.com/abstract=3644987

 establish in an area with huge barriers in the banking and accounting sector. One reason
is that large restrictions in the OECD countries’ banking sector hamper export financing.
In this case, the lending is costly and therefore does not allow the establishment of a
commercial presence abroad, Paravisini et al (2014). In road transport and logistics
sectors, restrictions in these sectors affect food trade than maritime sector. Food products
are sensitive and perishable over time. Road transport appears as a mode of transport
with a long transit time. More transport time is longer due to restrictions; more the
product will have a poor quality on arrival.
This study shows the detrimental impact of services restrictions on global value chains
and quality of products. The novel conclusion is that services are strongly linked to
commodities through their use as inputs in the production and trade of goods. It also
shows that services restrictions implemented by a country have negative effects on
export performance, NordÅs and Rouzet (2016). The negative correlation assumes a
competitiveness channel linking services regulation and exporter performance. These
restrictions particularly excludes small food exporters because export costs are very high.

VIII. Regulatory Difference in Services and Food Trade

In this section, we investigate the impact of the regulatory difference in services between
pairs of countries on food exports. Following Kox and NordÅs (2007), the regulatory
difference between pairs of countries emerges as the most restrictive trade policy com-
pared to individual country restrictions. Indeed, this regulatory disparity is considered
as bilateral trade costs and affects trade through gravity estimation.
Using the index of regulatory heterogeneity between country pairs of OECD countries
in services21 , we estimate its effects on aggregate food exports. The index is constructed
as follows: from the database of country specific STRIs, for each sector we create a
matrix where each cell contains country i and j for measure m 22 . If the pair of countries
has the same answer for measure m, the cell is scored as zero and 1 otherwise. For
each country pair and each measure, a heterogeneity index is created by computing a
weighted average of these scores (an average of the 5 measures).
We have two types of index, one based on qualitative responses on the presence or
absence of regulations and another on the score that highlights the restrictiveness of
regulations. Similar the individual STRI measures, they are scored on a scale of 0 to 1
(less restrictive to services closed to foreign suppliers).
To capture the effect of the regulatory difference on bilateral food flows, we construct an
interaction variable between individual STRI variable (exporter and importer) and the
regulatory heterogeneity variable. Indeed, the regulatory disparity has less important
effects on trade if the importer or exporter country is completely closed to service
providers and significant impacts if the countries are open to services, NordÅs and
Rouzet (2016).
Our gravity equation is as follows
21 STRIregulatory heterogeneity indices.
22 Barriers
to entry, competition, restrictions on movement of persons, other discrimination, regulatory
transparency.

12

Electronic copy available at: https://ssrn.com/abstract=3644987

 Xij,t = exp[α0 + α1 tij + α2 Zij,t + α3 STRIheterij,kt + α4 STRIheterij,kt ∗ STRIi,kt
(4)
+α5 STRIheterij,kt ∗ STRIj,kt + µit + γ jt + eijt ]

With STRIheterij,kt the regulatory difference between country pair in sector k at year
t 23 , STRIsi,kt , STRIj,kt are STRIs of exporting and importing country, µit , γ jt dummy
variables that represent the exporter-importer year fixed effects (inward and outward
multilateral resistance). The inclusion of country year fixed fixed effects in the gravity
equation accounts for the multilateral price terms at the same time as well as variation
in all time-varying country variables, Feenstra, 2004; Baldwin and Taglioni 2006; Baier
and Bergstrand 2007.
The results with PPML as estimate is presented in Table A.8 and Table A.9. Columns
1 to 6 represent the results of the regulatory disparity on trade flows and the rest the
effects of our interaction terms. The regulatory difference in cargo handling, customs
brokerage and banking sectors have negative and significant values in our regression. In
addition, our interaction variables have negative and significant signs in cargo-handling,
customs brokerage and banking sectors. The negative sign of the regulatory difference
and our interaction terms suggest that the regulatory difference has negative effects
on food exports although both exporting and importing countries are totally closed or
opened to service providers.
Regulatory disparity has important effects on services trade if countries are open to
services suppliers and less important effects if they are completely closed, NordÅs and
Rouzet (2016) . In this case individual country restrictions remain the main barrier to
entry for service suppliers. In food trade, disparity of regulation in services, sanitary
and technical standards are the main barrier to entry of products.

IX. Robustness check

Our main gravity estimates suggest negative and significant results for the impact
of regulatory heterogeneity on food exports. We performed four robustness tests to
evaluate the sensitivity of our results by focusing on the level of restrictions between
countries.
The first test considers the regulatory disparity between EEA countries. Using an
interaction variable between the binary variable EEA and the regulatory disparity, we
estimate the effects on aggregate trade in food products. The EEA reflects a fully
integrated internal market governed by the free movement of goods and services.
The results are presented in Table A. 10. They are positive and significant in all sectors.
The low regulatory disparity between EEA countries boosts food trade contrary to
countries with high disparity.
Second, we estimated the consequences of the regulatory difference between OECD
and non-OECD (emerging) countries on their food trade. We consider OECD countries
as exporting countries and emerging countries as importing countries. The results
presented in Table A.11 show negative and significant impacts at less than 10% in the
23 We use score index to compare it to individual country restrictions based on the score

13

Electronic copy available at: https://ssrn.com/abstract=3644987

 cargo handling sector, banking and very significant in the maritime sector. Compared
these results to the third test (Table A.12) where the exporting are the emerging countries
and the importing are the OECD countries, we find negative and significant impacts in
the logistics sector (cargo handling and customs services) and the maritime sector. The
regulatory disparity in services has greater negative effects on North-South or South-
North trade. The disparity in logistics and maritime services emerge as restrictive on
trade flows. Regulatory cooperation between North-South countries would be beneficial
for international food trade.
The third test investigates the impact of the regulatory difference on bilateral food
trade between net food exporting and importing countries as classified by UNCTADstat
24 . Table A.13 presents the results and shows the limiting and significant effects of

restrictions in logistics, banking and road transport services. This confirms our results
and strengthens evidence that the regulatory disparity in services affects food trade.
The last robustness test examines the consequences of the regulatory disparity on
bilateral trade in manufactured and intermediate goods25 . The aim is to compare the
effects of the regulatory disparity on food and on manufactured and intermediate
goods. The results presented in Table A.14 show negative and more significant effects
on manufactured and intermediate goods than food trade.

X. Endogeneity Issue
To solve the endogeneity issue between the regulatory heterogeneity in services and
the independent variable, the literature suggests using an instrument that is correlated
with the regulatory heterogeneity variable and uncorrelated with the error term. This
endogeneity occurs when we consider liberal policies in the EU. EU countries trade
more with each other because the restrictions are lower (so we have an reverse causality
between these restrictions and food trade). Despite the free movement of goods and
services, disparities in banking, accounting and legal services among EU Member States
negatively affect trade, NordÅs and Rouzet (2016).
Regulatory Disparity are composite measures that capture each country’s restrictive
policies to create a composite index by country pair. However, it is very difficult to
find an instrument that is closely related to our variable of interest. But to reduce the
risk of bias caused by endogeneity, we will consider the lagged variable of regulatory
disparity. Indeed, the index of regulatory disparity remains invariant in some sectors,
so the measure implemented this year explains the measure for the following year.
The results are reported in Table A.14 and show negative and significant effects of
regulatory disparity in the logistics and banking sector on trade in food commodities.

24 United Nations Conference on Trade and Development (UNCTAD).

25 Data extracted from UNCTADstat for manufacturing goods and OECDstat for intermediate goods

14

Electronic copy available at: https://ssrn.com/abstract=3644987

 XI. Conclusion
This study is an extension of previous studies based on the impacts of services re-
strictions on international trade. However, it differs from earlier studies because we
examine restrictions in 6 service sectors on trade of food products, topic not studied in
the literature. As a result of this analysis, harmonization policies have been highlighted.

Indeed, our paper aims to study the impacts of service standards on food trade, pre-
sented a tool for quantifying services restrictions and evaluated these effects on trade
flows. We started from a gravity modeling using panel data on bilateral trade in food
products between OECD countries from 2014-2018. We find that restrictions in services
have non-significant effects on overall food trade. However, there are negative and
significant effects on the disaggregated trade in food goods, particularly on agricultural
raw materials and perishable products. The restrictions implemented by the exporting
country have greater restrictive effects on exports than the importing country’s restric-
tions. Regulations in the logistics, banking, accounting and road transport sectors are
restrictive on food trade.
The regulatory difference between countries in logistics and banking negatively impact
overall food export.
The results seem robust to alternative specifications. Four different aspects were ana-
lyzed. First, we estimate the effects of the regulatory difference in EEA countries on
food trade, second, the effects of regulatory difference on trade between OECD and
emerging countries (North-South), and third, the impact on trade between emerging
and OECD countries (South-North). Finally, estimate the consequences of the regulatory
difference on trade between net food exporting and importing countries.
The results of our robustness checks show significant and positive effects of EEA coun-
tries regulations on trade. The regulatory disparity has negative and more significant
effects on South-North than North-South trade. It profoundly affects trade between
major food exporters and importers. The more restrictive rules of the emerging countries
have negative effects on their export performance.
Our study differed from existing literature, however, it has limitations based on the data.
Indeed, the restrictiveness indices in services are composite measures that invariant over
time for some sectors, so the data are very short (2014-2018). Regulations are adjustment
policies that require time if a firm or exporter does not have resources to conform.
Therefore, the data would not capture the effects on trade over the long term. However,
we can improve our study by considering subsidy policy coupled with regulation. It may
be useful to consider a binary variable reports the existence of a "Most favoured nation"
MFN and "National treatment" clause to take account treatment subject to foreign export
once on the domestic market.

This study of impacts of services restrictions on food products shows that regulatory
cooperation between countries has become an increasingly key element in international
food trade. Regulatory harmonization in economic integration areas or harmonization
of regulations between emerging and developed countries, net food exporters and
importers countries significantly stimulate trade flows.

15

Electronic copy available at: https://ssrn.com/abstract=3644987

 XII. appendix
Table A. 1:
OECD, non OECD, EEA and Major net food-exporting countries

OECD Non-OECD EEA countries Major net

countries Emerging countries food-exporting economies
Australia China Austria Australia
Austria ( People0 sRepublico f ) Belgium Belgium
Belgium Colombia CzechRepublic Brazil
Canada CostaRica Denmark Canada
Chile India Estonia Chile
CzechRepublic Indonesia Finland CostaRica
Denmark Malaysia France Denmark
Estonia Russia Germany Hungary
Finland SouthA f rica Greece Iceland
France Thailand Hungary Indonesia
Germany Iceland Ireland
Greece Ireland Mexico
Hungary Italy Netherlands
Iceland Latvia NewZealand
Ireland Lithuania Norway
Israel Luxembourg Poland
Italy Netherlands SouthA f rica
Japan Norway Spain
Korea Poland Thailand
Latvia Portugal Turkey
Lithuania SlovakRepublic UnitedStates
Luxembourg Slovenia o f America
Mexico Spain
Netherlands Sweden
NewZealand Switzerland
Norway UnitedKingdom
Poland
Portugal
SlovakRepublic
Slovenia
Spain
Sweden
Switzerland
Turkey
UnitedKingdom
UnitedStates

16

Electronic copy available at: https://ssrn.com/abstract=3644987

 Table A.2:
Descriptive statistics

Variables obs. Mean Std. Dev. Min Max

Aggregated f ood 6,298 4.06e+08 1.67e+09 0 2.63e+10
Manu f acturing goods 6,297 4270767 1.56e+07 4.034 3.01e+08
Intermediate goods 6,297 2763745 1.06e+07 0 2.30e+08

Disaggregated food products

Foods live animal 6,258 422482.8 1617761 0.006 2.50e+07
Beverages 6,051 56961.86 229428.3 0.002 5440184
Animal and vegetable oil 5,245 25735.59 122478.2 0.001 2191775
Oils seeds 3,987 20541.02 114354 0.002 2175221
Agricultural raw materials 6,158 79133.7 358488.1 0 9402706
Perishable products 6,300 148378.6 496587.5 0 8289848

Traditional variables of the gravity model

RTA 6,300 0.8009524 0.3993156 0 1
langij 6,300 0.0587302 0.2351376 0 1
borderij 6,300 0.0642857 0.2452807 0 1
Ln distij 6,300 3.469138 0.5056946 1.775372 4.29195
GDPi,t 6,300 1.39e+12 2.92e+12 1.50e+10 1.79e+13
GDPj,t 6,300 1.39e+12 2.92e+12 1.50e+10 1.79e+13
Intra EEA 6,300 0.4777778 0.4995456 0 1
N AFTA 6,300 0.0047619 0.0688475 0 1
Ln REER 5,599 1.989038 0.0534931 1.841338 2.194555

Heterogeneity STRI
Logis cargo handij,t 6,300 0.237079 0.0679761 0.063 0.472
Logis customs broij,t 6,300 0.2320959 0.147744 0.03 0.822
Accountingij,t 6,300 0.3021184 0.1321572 0.083 0.867
Bankingij,t 6,300 0.2225016 0.0617251 0.055 0.426
Roadij,t 6,300 0.2035533 0.1104277 0.024 0.646
Seaij,t 4,350 0.2569039 0.0770907 0.061 0.501

Exporter STRI
Logis cargo handi,t 6,300 0.2174444 0.0637243 0.12 0.41
Logis customs broi,t 6,300 0.2171111 0.1312967 0.11 1
Accountingi,t 6,300 0.2858889 0.1610676 0.1 1
Banking j,t 6,300 0.2031111 0.053766 0.13 0.37
Roadi,t 6,300 0.2002778 0.0852739 0.1 0.62
seai,t 5,250 0.236 0.0604927 0.15 0.35

Importer STRI
Logis cargo hand j,t 6,300 0.2174444 0.0637243 0.12 0.41
Logis customs broi,t 6,300 0.2230556 0.14338 0.13 1
Accountingi,t 6,300 0.286 0.1609945 0.1 1
Bankingi,t 6,300 0.2033889 0.053568 0.13 0.37
Roadi,t 6,300 0.2003333 0.0852639 0.1 0.62
seai,t 5,250 0.2360667 0.0604364 0.15 0.35

17

Electronic copy available at: https://ssrn.com/abstract=3644987

 STRI
Lo
gi
Lo st
ic
gi
st s
ic ca
s r

0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
st go
or -h
ag an
e dl
in
Lo an
d g
gi
st w
ic ar
s eh
o
fre us
Lo ig e
gi h tf
st
ic or
s w
cu ar
st di
om ng
s
br
ok
er

5
6
7
8
9
10
11

ag
e

2000
Ac
co
un
t in
g

2002
Ar
ch
ite
ct
ur
e
En

2004
gi
ne
er
in
g

2006
Le
M ga
ot l
io
n

2008
pi
ct

18
ur
es
Br
oa
dc
as 2010
So tin

Figure A. 2:
Figure A. 1:

un g
d
re
co
2012

rd
in
g
Te
le
2014

co
m
Ai
rt
ra
M ns
2016

ar po
iti r t
m
e
OECD countries’ sectoral STRIs in 2017

R tra
oa ns
d
2018

fre po
ig rt
ht
tra
R
Tariff rate, applied, simple mean, all products (%)

ai ns
lf po
re
ig r t
ht

Electronic copy available at: https://ssrn.com/abstract=3644987

tra
ns
po
rt

C
ou
rie
r
D
is
C tri
bu
om
m tio
n
er
ci
a lb
an
ki
ng
In
su
r an
c e
C
om
pu
te
C r
on
st
ru
ct
io
n
 Figure A. 3:
Extra European Union goods transported in thousands of tonnes

18
maritime transport
Road transport
16 Air transport

14

12

10

0
2010 2011 2012 2013 2014 2015 2016 2017

0.2
Food products
110 0.15 Manufactured goods
importation Services

100
exportation 0.1
volume of credit (% GDP)
0.05
90
0

80
-0.05

70 -0.1

-0.15
60
-0.2
50
-0.25
2006 2008 2010 2012 2014 2016 2018
40
2004 2006 2008 2010 2012 2014 2016 2018

Figure A. 5: OECD countries’

Figure A. 4: US import, export and annual export growth from 2005 to
credit volumes from 2005 to 2017 2017

19

Electronic copy available at: https://ssrn.com/abstract=3644987

 Table A.4:
STRI and Cross-Border Exports of Food Commodities: PPML Estimates

Total f ood Food and live

products animals
(SITC 0 + 1 + 22 + 4) SITC 0
Model Cargo Custom Accounting Banking Road Sea Cargo Customs Accounting Banking Road Sea
Handling Brokerage Handling Brokerage
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) 11 (12)
Ln GDPi,t 1.4921∗∗∗ 1.5048∗∗∗ 1.4950∗∗∗ 1.4751∗∗∗ 1.4892∗∗∗ 1.2601∗∗∗ 1.4774∗∗∗ 1.4830∗∗∗ 1.4721∗∗∗ 1.4521∗∗∗ 1.4620∗∗∗ 1.2075∗∗∗
(0.0966) (0.0961) (0.0937) (0.1013) (0.0975) (0.1202) (0.1086) (0.1077) (0.1013) (0.1136) (0.1083) (0.1365)
Ln GDPj,t 1.7891∗∗∗ 1.8198∗∗∗ 1.8574∗∗∗ 1.7973∗∗∗ 1.7929∗∗∗ 1.6943∗∗∗ 1.7255∗∗∗ 1.7470∗∗∗ 1.8185∗∗∗ 1.7217∗∗∗ 1.7168∗∗∗ 1.6234∗∗∗
(0.0984) (0.0977) (0.1083) (0.0982) (0.0988) (0.1206) (0.1078) (0.1080) (0.1201) (0.1084) (0.1079) (0.1290)
LnREERij,t 1.0946 1.2087 2.1564∗∗ 1.0468 1.0780 2.2260∗∗ 1.1306 1.2344 2.4630∗∗ 0.9843 1.0617 2.2066∗∗
(0.8407) (0.8604) (1.0259) (0.8316) (0.8549) (0.9244) (0.8410) (0.8614) (1.0958) (0.8318) (0.8617) (0.9703)
RTAij,t 0.3026 0.2416 0.5352∗∗∗ 0.3331 0.3249 0.3435 0.5668∗∗ 0.5396∗∗ 0.8297∗∗∗ 0.6084∗∗ 0.6389∗∗ 0.5898∗∗
(0.2074) (0.2297) (0.2046) (0.2174) (0.2335) (0.2103) (0.2405) (0.2594) (0.2301) (0.2480) (0.2576) (0.2358)
N AFTA 0.7355∗∗ 0.6116∗∗ 0.4304 0.7092∗∗ 0.7074∗∗ 0.6379 0.6765∗∗ 0.5841∗∗ 0.3283 0.6614∗∗ 0.6839∗∗ 0.5219
(0.2959) (0.2462) (0.3224) (0.3168) (0.2958) (0.4006) (0.2864) (0.2578) (0.3136) (0.3155) (0.3150) (0.3953)

20
Intra − EEA 0.8377∗∗∗ 0.8380∗∗∗ 0.7814∗∗∗ 0.7758∗∗∗ 0.7997∗∗∗ 0.4729∗∗ 0.7006∗∗∗ 0.6816∗∗∗ 0.6394∗∗∗ 0.6111∗∗∗ 0.6435∗∗∗ 0.2348
(0.1491) (0.1537) (0.1452) (0.1951) (0.1524) (0.2178) (0.1615) (0.1651) (0.1568) (0.1979) (0.1614) (0.2091)
Ln distij -0.9660∗∗∗ -1.0869∗∗∗ -0.9158∗∗∗ -0.9826∗∗∗ -0.9846∗∗∗ -1.0785∗∗∗ -0.9612∗∗∗ -1.0450∗∗∗ -0.9068∗∗∗ -0.9692∗∗∗ -0.9403∗∗∗ -1.1531∗∗∗
(0.2052) (0.2180) (0.2016) (0.2038) (0.2080) (0.2332) (0.2172) (0.2340) (0.2062) (0.2163) (0.2199) (0.2383)
langij 0.1167 0.1664 0.1195 0.1036 0.1079 0.1141 0.0662 0.0941 0.0729 0.0451 0.0185 0.0520
(0.2032) (0.1928) (0.1941) (0.2104) (0.2076) (0.2353) (0.2099) (0.2156) (0.1998) (0.2265) (0.2343) (0.2398)
borderij 0.8706∗∗∗ 0.8329∗∗∗ 0.8878∗∗∗ 0.8838∗∗∗ 0.8800∗∗∗ 0.9259∗∗∗ 0.9361∗∗∗ 0.9156∗∗∗ 0.9562∗∗∗ 0.9565∗∗∗ 0.9748∗∗∗ 1.0023∗∗∗
(0.2137) (0.2138) (0.1983) (0.2131) (0.2167) (0.2275) (0.2207) (0.2259) (0.2016) (0.2223) (0.2266) (0.2326)
STRI importer -1.6071 0.3360 -0.5589 -0.1405 -0.2704 -0.9301 -1.9092∗ 0.0683 -0.4298 -0.8263 -0.8637 -1.0900
(1.1393) (0.4105) (0.4886) (1.3449) (1.0372) (0.9915) (1.1587) (0.4657) (0.5025) (1.3317) (1.1613) (1.0624)
STRI exporter 1.4828 0.5939∗ -1.5281∗∗∗ -0.3725 0.1504 -1.4498 1.7268 0.4674 -2.0019∗∗∗ -0.4010 -0.1863 -1.4062

Electronic copy available at: https://ssrn.com/abstract=3644987

(1.2610) (0.3556) (0.4754) (1.5827) (0.7761) (1.3102) (1.3061) (0.3717) (0.5129) (1.5099) (0.8227) (1.4572)

R2 0.728 0.748 0.744 0.719 0.723 0.728 0.699 0.709 0.719 0.688 0.683 0.705
Observations 5597 5597 5597 5597 5597 3769 5568 5568 5568 5568 5568 3756
Year − FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6), (7), (8), (9), (10), (11), (12) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
∗, ∗∗, ∗ ∗ ∗ denote signicance respectively at the 10% 5%nd 1%Λ levels.
 Table A.5:
STRI and Cross-Border Exports of Food Commodities: Continued

Beverages Animal and vegetable

SITC11 oils, f ats and waxes SITC4
Model Cargo Custom Accounting Banking Road Sea Cargo Customs Accounting Banking Road Sea
Handling Brokerage Handling Brokerage
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) 11 (12)
Ln GDPi,t 1.5879∗∗∗ 1.6282∗∗∗ 1.6442∗∗∗ 1.6015∗∗∗ 1.6795∗∗∗ 1.4785∗∗∗ 1.3871∗∗∗ 1.3670∗∗∗ 1.4189∗∗∗ 1.2691∗∗∗ 1.3356∗∗∗ 1.1442∗∗∗
(0.1268) (0.1167) (0.1552) (0.1242) (0.1147) (0.1544) (0.1547) (0.1528) (0.1622) (0.1474) (0.1539) (0.1785)
Ln GDPj,t 2.2375∗∗∗ 2.2377∗∗∗ 2.0801∗∗∗ 2.2506∗∗∗ 2.1963∗∗∗ 2.0799∗∗∗ 1.6807∗∗∗ 1.7301∗∗∗ 1.7202∗∗∗ 1.6489∗∗∗ 1.7245∗∗∗ 1.4381∗∗∗
(0.1756) (0.1620) (0.1319) (0.1853) (0.1573) (0.1633) (0.2135) (0.2114) (0.2272) (0.1907) (0.2136) (0.2229)
Ln REERij,t 1.0304 1.7330 -0.8121 0.9529 1.4480 2.5624∗ -1.4583 -1.8051 -1.0095 0.0235 -1.5113 3.8543∗∗
(1.0796) (1.0956) (1.5300) (1.1798) (1.0261) (1.5122) (1.5041) (1.2304) (2.0611) (1.2888) (1.3448) (1.8845)
RTAij,t -0.1782 -0.3966∗∗ -0.0942 -0.1770 -0.4117∗ -0.2128 -0.3356 -0.4267 -0.2941 -0.2644 -0.2451 -0.3034
(0.1727) (0.1970) (0.2147) (0.1953) (0.2239) (0.2077) (0.4197) (0.4507) (0.4353) (0.4186) (0.4490) (0.3349)
N AFTA 0.1250 -0.0193 0.1642 0.1853 0.0225 0.4437 1.1960∗∗∗ 1.0684∗∗∗ 0.9880∗∗ 1.1779∗∗∗ 1.0841∗∗∗ 0.9496∗∗
(0.5936) (0.4082) (0.7093) (0.5966) (0.3816) (0.7290) (0.3475) (0.3408) (0.4892) (0.3973) (0.3746) (0.4047)

21
Intra − EEA 0.7770∗∗∗ 0.8534∗∗∗ 0.7611∗∗∗ 0.8174∗∗∗ 0.8456∗∗∗ 0.9369∗∗∗ 1.0669∗∗∗ 1.1654∗∗∗ 1.1667∗∗∗ 0.8449∗∗∗ 1.1160∗∗∗ 0.6206∗
(0.2160) (0.1885) (0.1935) (0.3136) (0.1876) (0.3385) (0.2706) (0.2783) (0.2660) (0.3054) (0.2723) (0.3594)
Ln distij -1.0741∗∗∗ -1.2708∗∗∗ -0.9889∗∗∗ -1.0368∗∗∗ -1.2306∗∗∗ -0.8847∗∗ -1.6022∗∗∗ -1.6911∗∗∗ -1.6068∗∗∗ -1.5771∗∗∗ -1.5778∗∗∗ -1.7748∗∗∗
(0.2533) (0.2193) (0.3255) (0.2595) (0.2124) (0.3802) (0.4041) (0.4200) (0.3508) (0.3773) (0.3897) (0.3758)
langij 0.4586∗ 0.6023∗∗∗ 0.4890 0.4812∗∗ 0.6232∗∗∗ 0.4317 -0.1557 -0.1363 -0.1267 -0.1976 -0.1782 -0.1640
(0.2787) (0.1699) (0.3130) (0.2379) (0.1639) (0.3444) (0.3484) (0.4159) (0.3418) (0.4258) (0.4314) (0.3416)
borderij 0.3251 0.2233 0.3308 0.3204∗ 0.2129 0.3458 0.8505∗∗ 0.8237∗∗ 0.8312∗∗ 0.8853∗∗ 0.8699∗∗ 0.9337∗∗
(0.2212) (0.1587) (0.2587) (0.1929) (0.1551) (0.2575) (0.3780) (0.4074) (0.3551) (0.3735) (0.3988) (0.3856)
STRI importer -0.0151 0.3352 -2.0702∗∗∗ 1.5056 -0.5760 1.9961 1.5736 1.1654∗ 0.1748 1.6752 1.3919 1.7851
(1.1651) (0.3886) (0.4139) (1.3181) (0.9047) (1.4254) (2.1506) (0.6863) (0.9846) (2.1173) (1.2313) (2.2325)
STRI exporter 2.7422 2.0581∗∗∗ 1.1427∗∗ 1.4282 3.6205∗∗∗ -1.8040 -4.1272∗ -1.6135 -1.4159 -11.2385∗∗∗ -5.2190∗∗ -5.6461∗∗

Electronic copy available at: https://ssrn.com/abstract=3644987

(2.9338) (0.4788) (0.5203) (4.0683) (0.8755) (1.6559) (2.1463) (0.9909) (1.1805) (3.3214) (2.5434) (2.7622)

R2 0.410 0.549 0.407 0.410 0.575 0.372 0.432 0.447 0.420 0.470 0.458 0.502
Observations 5411 5411 5411 5411 5411 3656 4759 4759 4759 4759 4759 3322
Year − FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6), (7), (8), (9), (10), (11), (12) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
∗, ∗∗, ∗ ∗ ∗ denote signicance respectively at the 10% 5%nd 1%Λ levels.
 Table A.6:
STRI and Cross-Border Exports of Food Commodities: Continued

Oil seeds and Agricultural raw materials

oleaginous f ruits STIC 22 (SITC 2 less 22, 27 and 28)
Model Cargo Custom Accounting Banking Road Sea Cargo Customs Accounting Banking Road Sea
Handling Brokerage Handling Brokerage
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) 11 (12)
Ln GDPi,t 1.6147∗∗∗ 1.5461∗∗∗ 1.5265∗∗∗ 1.5807∗∗∗ 1.4110∗∗∗ 1.6426∗∗∗ 1.1953∗∗∗ 1.1498∗∗∗ 1.1852∗∗∗ 1.1202∗∗∗ 1.1197∗∗∗ 0.9857∗∗∗
(0.2390) (0.2480) (0.1896) (0.2427) (0.2439) (0.2781) (0.1284) (0.1231) (0.1116) (0.1257) (0.1193) (0.1532)
Ln GDPj,t 1.3173∗∗∗ 1.4733∗∗∗ 1.4581∗∗∗ 1.3962∗∗∗ 1.5250∗∗∗ 1.4415∗∗∗ 1.6566∗∗∗ 1.6614∗∗∗ 1.7810∗∗∗ 1.6327∗∗∗ 1.6735∗∗∗ 1.5959∗∗∗
(0.2019) (0.1918) (0.2072) (0.1974) (0.1872) (0.2950) (0.1450) (0.1344) (0.1665) (0.1365) (0.1349) (0.1524)
Ln REERij,t 1.7379 0.7548 2.9837 1.3461 1.8988 1.8259 1.7134 1.4776 3.3994 1.9714 1.5915 2.8780∗∗
(2.2874) (2.4638) (2.8761) (2.4060) (2.5383) (1.6236) (1.5164) (1.1508) (2.4922) (1.2053) (1.2055) (1.3090)
RTAij,t -0.7279∗ -0.8488∗∗ -0.3948 -0.7550∗ -0.7473∗ -0.4741 -0.0255 0.0719 0.0929 0.0721 0.1681 0.0417
(0.4358) (0.4067) (0.4557) (0.4215) (0.3887) (0.4477) (0.2567) (0.2570) (0.2569) (0.2502) (0.2447) (0.2489)
N AFTA 1.3303∗∗ 0.6824 0.6555 1.0698∗ 0.8140 1.1936 1.1023∗∗∗ 1.1764∗∗∗ 0.7993∗∗ 1.0942∗∗∗ 1.1698∗∗∗ 0.9154∗∗∗
(0.5795) (0.5880) (0.6052) (0.5751) (0.6663) (0.8241) (0.2457) (0.2415) (0.3251) (0.2535) (0.2598) (0.3046)

22
Intra − EEA 0.3194 0.2726 0.3431 0.2722 0.3091 -0.2354 0.5731∗∗∗ 0.5211∗∗∗ 0.5656∗∗∗ 0.3636∗ 0.5176∗∗∗ 0.1896
(0.4021) (0.3559) (0.3522) (0.3717) (0.3726) (0.4855) (0.1886) (0.1775) (0.1758) (0.1891) (0.1789) (0.1953)
Ln distij -0.1556 -0.4764 -0.1876 -0.3014 -0.3765 -0.0675 -1.0392∗∗∗ -0.9487∗∗∗ -1.0723∗∗∗ -1.0382∗∗∗ -0.9334∗∗∗ -1.2307∗∗∗
(0.4003) (0.4690) (0.3590) (0.4099) (0.4665) (0.4497) (0.2644) (0.2749) (0.2484) (0.2558) (0.2630) (0.2548)
langij -0.7048∗∗ -0.4249 -0.6644∗∗ -0.6391∗ -0.5061 -0.7857∗∗ -0.0171 -0.1254 -0.0047 -0.0993 -0.1998 -0.0036
(0.3128) (0.3261) (0.2769) (0.3341) (0.3587) (0.3727) (0.1800) (0.2098) (0.1744) (0.1984) (0.2180) (0.1873)
borderij 1.5865∗∗∗ 1.4411∗∗∗ 1.5642∗∗∗ 1.5514∗∗∗ 1.5315∗∗∗ 1.7419∗∗∗ 0.9214∗∗∗ 0.9973∗∗∗ 0.9336∗∗∗ 0.9800∗∗∗ 1.0435∗∗∗ 0.8866∗∗∗
(0.4060) (0.4389) (0.3907) (0.4258) (0.4590) (0.4803) (0.2683) (0.2789) (0.2515) (0.2725) (0.2757) (0.3031)
STRI importer -1.5564 1.7719∗∗ -0.9930 3.0567 2.4561 -7.7812∗∗ -0.7488 -0.0028 0.6495 -0.4905 -0.4732 -0.7603
(2.5825) (0.7262) (1.0024) (2.4512) (1.6362) (3.0212) (1.3732) (0.4133) (0.4471) (1.2777) (0.6795) (1.4301)
STRI exporter -1.3964 -3.4664∗ -3.9924∗∗∗ -4.2406 -8.2135∗∗∗ -1.7754 -0.8397 -2.3388∗∗∗ -2.6566∗∗∗ -4.9270∗∗∗ -4.7867∗∗∗ -1.4197

Electronic copy available at: https://ssrn.com/abstract=3644987

(3.7898) (1.8097) (1.2337) (3.3949) (2.8551) (2.9366) (1.7967) (0.6177) (0.7608) (1.3595) (1.0241) (1.8458)

R2 0.330 0.389 0.362 0.382 0.393 0.338 0.645 0.707 0.613 0.698 0.711 0.676
Observations 3565 3565 3565 3565 3565 2446 5464 5464 5464 5464 5464 3678
Year − FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6), (7), (8), (9), (10), (11), (12) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
∗, ∗∗, ∗ ∗ ∗ denote signicance respectively at the 10% 5%nd 1%Λ levels.
 Table A.7:
STRI and Cross-Border Exports of Food Commodities: Continued

Perishable products
STIC 01 + 02 + 03
Model Cargo Custom Accounting Banking Road Sea
Handling Brokerage
(1) (2) (3) (4) (5) (6)
Ln GDPi,t 1.2535∗∗∗ 1.2377∗∗∗ 1.2856∗∗∗ 1.1982∗∗∗ 1.2270∗∗∗ 0.9480∗∗∗
(0.1257) (0.1243) (0.1108) (0.1312) (0.1213) (0.1545)

Ln GDPj,t 1.8040∗∗∗ 1.8381∗∗∗ 1.9253∗∗∗ 1.7840∗∗∗ 1.8173∗∗∗ 1.6911∗∗∗

(0.1220) (0.1129) (0.1217) (0.1142) (0.1101) (0.1400)

Ln REERij,t 0.1042 -0.2026 2.4621 -0.0612 -0.2143 1.9477∗

(1.3301) (1.2660) (1.5910) (1.3286) (1.2802) (1.0502)

RTAij,t 0.8493∗∗∗ 0.9149∗∗∗ 1.0952∗∗∗ 0.9535∗∗∗ 0.9980∗∗∗ 0.8862∗∗∗

(0.2777) (0.2869) (0.2373) (0.2760) (0.2738) (0.2678)
N AFTA 0.1120 0.0216 -0.3505 0.0195 0.0589 -0.1583
(0.2832) (0.2969) (0.2778) (0.3696) (0.3789) (0.3908)

Intra − EEA 0.5517∗∗∗ 0.5034∗∗ 0.5222∗∗∗ 0.3316 0.4769∗∗ -0.0786

(0.2006) (0.2024) (0.2009) (0.2204) (0.1991) (0.2024)

Ln distij -0.7195∗∗∗ -0.7646∗∗∗ -0.7492∗∗∗ -0.7785∗∗∗ -0.7290∗∗∗ -1.0963∗∗∗

(0.2254) (0.2395) (0.2101) (0.2274) (0.2272) (0.2188)

langij 0.1769 0.1262 0.1509 0.1340 0.0800 0.1940

(0.2078) (0.2266) (0.1861) (0.2332) (0.2380) (0.2097)

borderij 0.8327∗∗∗ 0.8634∗∗∗ 0.8676∗∗∗ 0.8689∗∗∗ 0.8988∗∗∗ 0.8605∗∗∗

(0.2161) (0.2238) (0.1966) (0.2227) (0.2254) (0.2104)

STRI importer -3.2135∗∗ 0.0213 -0.1585 -2.1635 -0.9171 -1.7108

(1.3971) (0.7126) (0.4386) (1.9617) (1.6597) (1.1202)

STRI exporter 0.7684 -0.6748∗ -2.9320∗∗∗ -2.7525∗∗ -1.8046∗∗∗ -3.1338∗∗

(1.3819) (0.3548) (0.5176) (1.0708) (0.6501) (1.5361)

R2 0.531 0.543 0.597 0.537 0.540 0.585

Observations 5599 5599 5599 5599 5599 3770
Year − FE Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
∗, ∗∗, ∗ ∗ ∗ denote signicance respectively at the 10% 5%nd 1%Λ levels.

23

Electronic copy available at: https://ssrn.com/abstract=3644987

 Table A.8:
Food Exports, Heterogeneity and Exporter STRI, PPML Regressions

Regulatory STRI and Regulatory

heterogeneity heterogeneity
Model Cargo Custom Accounting Banking Road Sea Cargo Customs Accounting Banking Road Sea
Handling Brokerage Handling Brokerage
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

RTAij,t 0.5146∗∗∗ 0.5991∗∗∗ 0.6640∗∗∗ 0.6386∗∗∗ 0.6588∗∗∗ 0.6801∗∗∗ 0.5324∗∗∗ 0.6162∗∗∗ 0.6594∗∗∗ 0.6469∗∗∗ 0.6737∗∗∗ 0.7137∗∗∗
(0.1471) (0.1481) (0.1493) (0.1540) (0.1452) (0.1433) (0.1464) (0.1467) (0.1481) (0.1530) (0.1467) (0.1417)

Ln distij -1.8137∗∗∗ -1.8532∗∗∗ -1.9676∗∗∗ -1.8496∗∗∗ -1.9614∗∗∗ -1.9570∗∗∗ -1.8322∗∗∗ -1.8965∗∗∗ -1.9664∗∗∗ -1.8707∗∗∗ -2.0092∗∗∗ -1.9714∗∗∗
(0.1355) (0.1285) (0.1204) (0.1134) (0.1350) (0.1300) (0.1347) (0.1235) (0.1205) (0.1142) (0.1297) (0.1279)

langij 0.0491 0.0604 0.1062 0.0866 0.1057 0.0341 0.0494 0.0777 0.1079 0.0968 0.1270 0.0636
(0.1238) (0.1226) (0.1235) (0.1197) (0.1254) (0.1476) (0.1253) (0.1218) (0.1234) (0.1199) (0.1254) (0.1512)

24
borderij 0.7305∗∗∗ 0.7396∗∗∗ 0.7319∗∗∗ 0.7506∗∗∗ 0.7312∗∗∗ 0.7365∗∗∗ 0.7329∗∗∗ 0.7269∗∗∗ 0.7291∗∗∗ 0.7325∗∗∗ 0.7154∗∗∗ 0.7359∗∗∗
(0.1143) (0.1126) (0.1121) (0.1050) (0.1140) (0.1218) (0.1146) (0.1125) (0.1121) (0.1069) (0.1119) (0.1200)

Heterogeneity scoreij,t -3.1090∗∗∗ -2.1816∗∗∗ 0.0704 -2.3665∗∗∗ -0.0865 0.2542

(0.7821) (0.7212) (0.4652) (0.8414) (0.8400) (0.7596)

Heterogeneity scoreij,t
∗STRIi,t -12.4783∗∗∗ -7.7586∗∗∗ -0.0941 -10.3077∗∗ 4.1275 4.1869
(3.2738) (2.4186) (1.2815) (4.0711) (3.6436) (3.4320)
R2 0.938 0.940 0.936 0.940 0.936 0.940 0.938 0.940 0.936 0.940 0.937 0.941
Observations 6298 6298 6298 6298 6298 4349 6298 6298 6298 6298 6298 4349
Exporter − Year − FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Importer − Year − FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Electronic copy available at: https://ssrn.com/abstract=3644987

Notes : The dependent variable is bilateral f ood product.
Columns (1), (2), (3), (4), (5), (6), (7), (8), (9), (10), (11), (12) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
*, **, *** denote signicance respectively at the 10% 5% and 1% levels.
 Table A.9:
Food Exports, Heterogeneity and Importer STRI, PPML Regressions

STRI and Regulatory

heterogeneity
Model Cargo Custom Accounting Banking Road Sea
Handling Brokerage
(1) (2) (3) (4) (5) (6)
RTAij,t 0.5190∗∗∗ 0.6419∗∗∗ 0.6702∗∗∗ 0.6446∗∗∗ 0.6632∗∗∗ 0.7048∗∗∗
(0.1471) (0.1471) (0.1493) (0.1546) (0.1458) (0.1390)

Ln distij -1.8289∗∗∗ -1.9079∗∗∗ -1.9699∗∗∗ -1.8581∗∗∗ -1.9740∗∗∗ -1.9698∗∗∗

(0.1337) (0.1262) (0.1197) (0.1141) (0.1301) (0.1277)

langij 0.0429 0.0760 0.1048 0.0939 0.1114 0.0578

(0.1249) (0.1233) (0.1244) (0.1195) (0.1262) (0.1470)

borderij 0.7357∗∗∗ 0.7380∗∗∗ 0.7354∗∗∗ 0.7387∗∗∗ 0.7269∗∗∗ 0.7353∗∗∗

(0.1148) (0.1142) (0.1116) (0.1063) (0.1137) (0.1203)
Heterogeneity scoreij,t
∗STRIj,t -13.5829∗∗∗ -5.2164∗∗ 0.7586 -11.3970∗∗∗ 0.7075 3.4072
(3.3629) (2.5431) (1.1795) (4.2192) (3.8162) (3.7144)
R2 0.938 0.937 0.936 0.939 0.936 0.941
Observations 6298 6298 6298 6298 6298 4349
Exporter − Year − FE Yes Yes Yes Yes Yes Yes
Importer − Year − FE Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
*, **, *** denote signicance respectively at the 10% 5% and 1% levels.

Table A.10:
Food Exports, Heterogeneity STRI: EEA restrictions

EEA
Restrictions
Model Cargo Custom Accounting Banking Road Sea
Handling Brokerage
(1) (2) (3) (4) (5) (6)
RTAij 0.4201∗∗∗ 0.4950∗∗∗ 0.5379∗∗∗ 0.5018∗∗∗ 0.4158∗∗∗ 0.3666∗∗
(0.1484) (0.1503) (0.1525) (0.1547) (0.1460) (0.1449)

Ln distij -1.9193∗∗∗ -1.9417∗∗∗ -1.9058∗∗∗ -1.9424∗∗∗ -1.9583∗∗∗ -1.8378∗∗∗

(0.1232) (0.1233) (0.1236) (0.1262) (0.1187) (0.1304)

langij 0.1615 0.1501 0.1010 0.1473 0.1545 0.0505

(0.1194) (0.1225) (0.1246) (0.1182) (0.1214) (0.1370)

borderij 0.7525∗∗∗ 0.7341∗∗∗ 0.7727∗∗∗ 0.7010∗∗∗ 0.7244∗∗∗ 0.7799∗∗∗

(0.1142) (0.1152) (0.1121) (0.1153) (0.1056) (0.1215)

Heterogeneity scoreij,t ∗ Intra − EEA 2.7235∗∗∗ 2.2666∗∗∗ 1.0892∗∗ 1.8766∗∗ 3.5253∗∗∗ 3.6350∗∗∗
(0.7001) (0.7961) (0.5384) (0.7717) (0.7976) (0.6767)
R2 0.940 0.937 0.935 0.937 0.943 0.946
Observations 6298 6298 6298 6298 6298 4349
Exporter − Year − FE Yes Yes Yes Yes Yes Yes
Importer − Year − FE Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
*, **, *** denote signicance respectively at the 10% 5% and 1% levels.

25

Electronic copy available at: https://ssrn.com/abstract=3644987

 Table A.11:
Food Exports, Heterogeneity STRI: OECD-Emerging Countries

OECD-Emerging
Countries
Model Cargo Custom Accounting Banking Road Sea
Handling Brokerage
(1) (2) (3) (4) (5) (6)
RTAij,t 0.6367∗∗∗ 0.5876∗∗∗ 0.5476∗∗∗ 0.6001∗∗∗ 0.6052∗∗∗ 0.6779∗∗∗
(0.2004) (0.1980) (0.1978) (0.1918) (0.1977) (0.2016)

borderij 1.4364∗∗∗ 1.4220∗∗∗ 1.4329∗∗∗ 1.4159∗∗∗ 1.3799∗∗∗ 1.4589∗∗∗

(0.2605) (0.2485) (0.2365) (0.2316) (0.2417) (0.2267)

borderij 0.5655∗ 0.5723∗ 0.6064∗ 0.4728 0.5426∗ 0.5704∗

(0.3047) (0.3087) (0.3244) (0.3197) (0.3045) (0.2979)

Ln distij -2.5765∗∗∗ -2.6338∗∗∗ -2.8474∗∗∗ -2.5792∗∗∗ -2.5548∗∗∗ -2.4471∗∗∗

(0.4063) (0.4152) (0.4571) (0.3839) (0.4099) (0.3911)

Heterogeneity scoreij,t -2.6167∗∗ -1.1089 1.5922 -2.6879∗ -1.3847 -2.4576∗∗∗

(1.2687) (1.0577) (1.0423) (1.4957) (1.0388) (0.7775)
R2 0.958 0.958 0.958 0.958 0.958 0.963
Observations 1765 1765 1765 1765 1765 1471
Exporter − Year − FE Yes Yes Yes Yes Yes Yes
Importer − Year − FE Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
*, **, *** denote signicance respectively at the 10% 5% and 1% levels.

Table A.12:
Food Exports, Heterogeneity STRI: Emerging-OECD Countries

Emerging − OECD
Countries
Model Cargo Custom Accounting Banking Road Sea
Handling Brokerage
(1) (2) (3) (4) (5) (6)
RTAij,t 0.1014 0.0888 -0.0170 0.0951 0.0903 0.0901
(0.1574) (0.1597) (0.1697) (0.1580) (0.1621) (0.1637)

borderij 0.8693∗∗ 0.8453∗ 0.8306∗∗ 0.7887∗ 0.7890∗ 0.8734∗∗

(0.4323) (0.4315) (0.4226) (0.4253) (0.4419) (0.4447)

langij 0.4747∗ 0.5219∗∗ 0.4541∗∗ 0.4056∗ 0.4935∗∗ 0.4423∗∗

(0.2452) (0.2458) (0.2312) (0.2205) (0.2508) (0.2227)

Ln distij -1.3945∗∗∗ -1.3301∗∗∗ -1.5430∗∗∗ -1.3509∗∗∗ -1.3348∗∗∗ -1.2565∗∗∗

(0.2572) (0.2769) (0.2863) (0.2933) (0.2879) (0.3076)

Heterogeneity scoreij,t -4.2883∗∗∗ -2.6552∗∗ 1.1157∗∗ -2.4427 -1.5737 -2.3518∗∗

(1.4497) (1.2419) (0.4968) (1.6669) (1.1766) (1.1015)
R2 0.859 0.833 0.816 0.828 0.824 0.843
Observations 1775 1775 1775 1775 1775 1490
Exporter − Year − FE Yes Yes Yes Yes Yes Yes
Importer − Year − FE Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
*, **, *** denote signicance respectively at the 10% 5% and 1% levels.

26

Electronic copy available at: https://ssrn.com/abstract=3644987

 Table A.13:
Food Exports, Heterogeneity STRI: Major Exporter and Importer Food Countries

Regulatory
heterogeneity
Model Cargo Custom Accounting Banking Road Sea
Handling Brokerage
(1) (2) (3) (4) (5) (6)
RTAij,t 0.5749∗∗∗ 0.6860∗∗∗ 0.7945∗∗∗ 0.6782∗∗∗ 0.7696∗∗∗ 0.7490∗∗∗
(0.1939) (0.1656) (0.1772) (0.1666) (0.1742) (0.1832)

langij 0.0103 0.0026 0.0626 -0.1104 0.0315 0.0543

(0.1786) (0.1589) (0.1542) (0.1750) (0.1687) (0.1722)

borderij 0.7895∗∗∗ 0.7773∗∗∗ 0.8110∗∗∗ 0.8688∗∗∗ 0.8018∗∗∗ 0.7742∗∗∗

(0.1497) (0.1420) (0.1533) (0.1345) (0.1522) (0.1602)

Ln distij -1.7574∗∗∗ -1.8023∗∗∗ -1.9552∗∗∗ -1.8201∗∗∗ -1.7944∗∗∗ -1.8536∗∗∗

(0.2471) (0.2439) (0.2440) (0.2208) (0.2638) (0.2619)

Heterogeneity scoreij,t -3.4931∗∗∗ -3.7597∗∗∗ -0.0574 -5.1261∗∗∗ -2.0561∗ -0.8819

(1.0934) (1.0463) (0.6563) (1.1192) (1.1908) (1.1723)
R2 0.879 0.889 0.881 0.915 0.882 0.879
Observations 2618 2618 2618 2618 2618 1999
Exporter − Year − FE Yes Yes Yes Yes Yes Yes
Importer − Year − FE Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
*, **, *** denote signicance respectively at the 10% 5% and 1% levels.

27

Electronic copy available at: https://ssrn.com/abstract=3644987

 Table A.15:
Endogeneity Issue: OECD countries

Regulatory
heterogeneity
Model Cargo Custom Accounting Banking Road Sea
Handling Brokerage
(1) (2) (3) (4) (5) (6)
RTAij,t 0.5744∗∗∗ 0.6240∗∗∗ 0.6571∗∗∗ 0.6416∗∗∗ 0.6608∗∗∗ 0.6845∗∗∗
(0.1439) (0.1459) (0.1484) (0.1510) (0.1451) (0.1459)

langij 0.0742 0.0797 0.1092 0.0963 0.1075 0.0574

(0.1212) (0.1224) (0.1234) (0.1215) (0.1243) (0.1366)

borderij 0.7264∗∗∗ 0.7339∗∗∗ 0.7284∗∗∗ 0.7420∗∗∗ 0.7298∗∗∗ 0.7428∗∗∗

(0.1138) (0.1133) (0.1132) (0.1107) (0.1137) (0.1197)

Ln distij -1.9053∗∗∗ -1.9241∗∗∗ -1.9649∗∗∗ -1.9158∗∗∗ -1.9672∗∗∗ -1.9413∗∗∗

(0.1269) (0.1233) (0.1207) (0.1175) (0.1255) (0.1254)

Heterogeneity scoreij,t−1 -2.0146∗∗∗ -1.3081∗∗∗ -0.1018 -1.4831∗∗∗ 0.0142 0.0827

(0.5828) (0.4719) (0.3131) (0.5655) (0.5602) (0.5533)
R2 0.937 0.938 0.936 0.938 0.936 0.940
Observations 6297 6297 6297 6297 6297 4348
Exporter − Year − FE Yes Yes Yes Yes Yes Yes
Importer − Year − FE Yes Yes Yes Yes Yes Yes

Notes : The dependent variable is bilateral f ood product.

Columns (1), (2), (3), (4), (5), (6) are per f ormed using PPML estimation.
Standard errors are reported in parentheses and clustered by country − pair level.
*, **, *** denote signicance respectively at the 10% 5% and 1% levels.

28

Electronic copy available at: https://ssrn.com/abstract=3644987

 References

[1] Amiti, Mary and David E. Weinstein. (2011). "Exports and Financial Shocks". The
Quarterly Journal of Economics, 126 (4), 1841-1877.
[2] Anderson, J. E. and Van Wincoop, E. 2003. "Gravity with gravitas: A solution to
the border puzzle". The American Economic Review.
[3] Anderson, J. E. and Van Wincoop, E. 2004. "Trade costs". Journal of Economic
Literature.
[4] Anderson, J. E. and Yotov, Y. V. 2012. "Gold standard gravity". Technical report,
National Bureau of Economic Research.
[5] Ariu, A. 2016. "Services Versus Goods Trade: A Firm Level Comparison". Review of
World Economics 152 (1), 19-41.
[6] Ariu, A. Mayneris, F. and Parenti, M. 2018. "One way to the top: How services
boost the demand for goods". Technical report, CEPR Discussion Papers.
[7] Baier, Scott L., and Jeffrey H. Bergstrand. 2007. "Do Free Trade Agreements Ac-
tually Increase Members’ International Trade?" . Journal of International Economics,
71(1): 72-95.
[8] Baier, Scott L., and Jeffrey H. Bergstrand. 2009. "Estimating the effects of free trade
agreements on international trade flows using matching econometrics" . Journal of
International Economics, vol. 77, issue 1, 63-76 .
[9] Baldwin, R., Skudelny, F. and Taglioni, D. 2005. "Trade Effects of the Euro, Evi-
dence from Sectoral Data". Frankfurt am Main: European Central Bank.
[10] Baldwin, R. and Taglioni, D. 2006. "Gravity for dummies and dummies for gravity
equations". NBER Working Paper No 12516.
[11] Baltagi, B.H., Egger, P., Pfaffermayr, M. 2003. "A generalized design for bilateral
trade flow models". Econom. Lett. 80, 391-397.
[12] Ben Shepherd. 2010." Dealing with Reverse Causality". ARTNeT Capacity Building
Workshop for Trade Research: Gravity Modeling.
[13] Beverelli, Cosimo; Fiorini, Matteo ; Hoekman, Bernard. 2017. "Services trade pol-
icy and manufacturing productivity: The role of institutions". Journal of International
Economics, Elsevier, vol. 104(C), pages 166-182.
[14] Borchert, Ingo; Mattoo, Aaditya. 2016. "The trade reducing effects of restrictions
on liner shipping". Journal of Comparative Economics, Elsevier, vol. 44(2), pages
231-242.
[15] Borchert, I. Batshur, G. and Aaditya, M. 2012. "Guide to the Services Trade Re-
strictions Database". World Bank Policy Research Working Paper WPS6108.
[16] Borchert, Ingo; Gootiiz, Batshur and Mattoo, Aaditya. 2012. "Policy barriers to
international trade in services : evidence from a new database". Policy Research
Working Paper Series 6109, The World Bank.
[17] Breinlich, H. , Criscuolo, C. 2011. "International trade in services: a portrait of
importers and exporters". Journal of International Economics 84 (2), 188-206 .

29

Electronic copy available at: https://ssrn.com/abstract=3644987

[18] Bricongne, J.C., Fontagné, L., Gaulier, G., Taglioni, D., Vicard, V. 2010. "Firms
and the global crisis: French exports in the turmoil". Journal of International Eco-
nomics, 87(1),134-146.
[19] Carrere, Celine and de Melo, Jaime and Wilson, John S. 2013. "The Distance
Puzzle and Low-Income Countries": An Update". Journal of Economic Surveys, Vol.
27, Issue 4, pp. 717-742, 2013.
[20] Carsten Fink, Aaditya Mattoo, Ileana Cristina Neagu. 2002. "Trade in Interna-
tional Maritime Services: How Much Does Policy Matter?". The World Bank Economic
Review, Volume 16, Issue 1, June 2002, Pages 81-108.
[21] CEPII. 2005. "Geographic and Bilateral Distance Database". Paris: Centre d’Etudes
rospectives et d’Informations Internationales.
[22] Chaney, T. 2008. "Distorted Gravity: The Intensive and Extensive Margins of
International Trade" . American Economic Review 2008, 98:4,1707-1721.
[23] Chor, Davin and Manova, Kalina. 2012. "Off the cliff and back? Credit conditions
and international trade during the global financial crisis". Journal of International
Economics, Elsevier, vol. 87(1), pages 117-133.
[24] Claessens, S., O. Hassib, and N. Van Horen. 2017. "The role of foreign banks in
trade". CEPR Discussion Papers 11821.
[25] Conti, G. , Turco, A.L. , Maggioni, D. 2010. "Exporters in services: new evidence
from Italian firms". Applied Economics Quarterly 56 (1), 73-98 .
[26] Copeland, B. and A. Mattoo. 2007. "The Basic Economics of Services Trade in the
Handbook of International Trade in Services". Oxford University Press
[27] Crozet, Matthieu; Milet, Emmanuel and Mirza, Daniel. 2016. "The impact of
domestic regulations on international trade in services: Evidence from firm-level
data". Journal of Comparative Economics, Elsevier, vol. 44(3), pages 585-607.
[28] Deardorff, Alan V. and Robert M, Stern. 2007. "Empirical Analysis of Barriers
to International Services Transactions and the Consequences of Liberalization".
Chapters, in: John R. Bryson and Peter W. Daniels (ed.), The Handbook of Service
Industries, chapter 13, Edward Elgar Publishing.
[29] Disdier, A.-C. and Head, K. 2008. "The Puzzling Persistence of the Distance Effect
on Bilateral Trade". Review of Economics and Statistics 90[1], 37-48.
[30] Eaton, J and Kortum, S. 2002. "Technology, Geography, and Trade". Econometrica
,vol.70, no 5, 1741-1779.
[31] Fabrice Morel. 2010. "Credit insurance in support of international trade Observa-
tions throughout the crisis". Export Credit Insurance Report.
[32] Feenstra, Robert, James Markusen, and Andrew Rose. 2001." Using the Gravity
Equation to Differentiate Among Alternaive Theories of Trade". Canadian Journal of
Economics, vol.34, p.430-47.
[33] Feenstra, R. 2004. "Advanced International Trade: Theory and Evidence". Princeton
University Press, Princeton, N.J.
[34] Fontagné Lionel, Amelie Guillin and Cristina Mitaritonna. 2011. "Estimation of
Tariff Equivalents for the Services Sectors" . CEPII working paper, no 2011-24.

30

Electronic copy available at: https://ssrn.com/abstract=3644987

[35] Fontagné, L. Mitaritonna, C. and Signoret, J. 2016. "Estimated Tariff Equivalents
of Services NTMs". CEPII Working Paper.
[36] Fukunari KimuraHyun-Hoon Lee. 2006. "The Gravity Equation in International
Trade in Services". Review of World Economics, Volume 142, pp 92 - 121.
[37] Gary, H and Sherry S. 2007. "Services Trade: Past Liberalization and Future
Challenges". Journal of International Economic Law, Volume 10, Issue 3, September
2007, Pages 605-630.
[38] Gaulier, G. , Milet, E. , Mirza, D. 2011. "Les Firmes Françaises dans le Commerce
International de Services". Economie et Statistique 435-436, 125-147.
[39] Geloso Grosso M. and al. 2015. " Services Trade Restrictiveness Index (STRI):
Scoring and Weighting Methodology ". OECD Working Papers on Trade Policy,
no 177, OECD Publishing, Paris.
[40] Gooris, J and Mitaritonna, C .2015. "Which import restrictions matter for trade in
services ?". CEPII Working Paper, no 2015-33, december 2015.
[41] Guillin, A. 2013. "Assessment of Tariff Equivalents for Services Considering the
Zero Flows". World Trade Review, 12, 3, 549-75.
[42] Head, K. Mayer, T. and Ries, J. 2009. "How remote is the offshoring threat?".
European Economic Review 53(4), 429-444.
[43] Head, K. and Mayer, T. 2014. "Gravity equations: Workhorse, toolkit, and cook-
book". Handbook of International Economics, Vol. 4, eds.
[44] Heckman J. 1979: "Sample Selection Bias as a Specification Error ". Econometrica,
47(1): 153-161.
[45] Helpman, E., Melitz, M., and Rubinstein, Y. 2008. "Estimating trade flows: Trad-
ing partners and trading volumes". The Quarterly Journal of Economics, 123(2):441487.
[46] Hoekman, B. and Shepherd, B. 2015. "Services, firm performance, and exports.
The case of the East African Community". International Growth Centre, Working
paper.
[47] Hofmann Claudia, Alberto Osnago, and Michele Ruta. 2017. "Horizontal Depth:
A New Database on the Content of Preferential Trade Agreements". Policy Research
Working Paper Series 7981, World Bank, Washington, DC
[48] IMF. 2017. "Promoting Inclusive Growth". IMF Annual Report 2017.
[49] Jacks, D. S., Meissner, C. M. and Novy, D. 2008. "Trade costs, 1870-2000". American
Economic Review: Papers and Proceedings 98(2): 529-34.
[50] Jafari, Y and Tarr, D. 2015. "Estimates of ad valorem equivalents of barriers against
foreign suppliers of services in eleven services sectors and 103 countries ". The
World Economy, Vol. 40, Issue 3, pp. 544-573.
[51] James E. Anderson. 1979. "A Theoretical Foundation for Gravity Equation". Ameri-
can Economic Review, 69(1):106-16.
[52] Jeffrey Bergstrand. 1989. "The Generalized Gravity Equation, Monopolistic Com-
petition, and the Factor-Proportions Theory in International Trade". The Review of
Economics and Statistics, vol. 71, issue 1, 143-53.

31

Electronic copy available at: https://ssrn.com/abstract=3644987

[53] Jong Woo Kang and Suzette Dagli. 2018. "International trade and exchange rates".
Journal of Applied Economics, 21:1, 84-105.
[54] Joseph Francois and Bernard Hoekman. 2010. "Services Trade and Policy". Journal
of Economic Literature.
[55] Kee, Hiau Looi, Alessandro Nicita and Marcelo Olarreaga. 2009. "Estimating
Trade Restrictiveness Indices". The Economic Journal, 119 (534), pp. 172-199.
[56] Kox, Henk L.M. and NordÅs, Hildegunn Kyvik. 2007. "Services trade and do-
mestic regulation". OECD Trade Policy Papers 49, OECD Publishing.
[57] Kox, H. and H.k. Nordàs. 2009. "Regulatory Harmonization and Trade in Services:
Volume and choice of mode". FREIT Trade Working Paper No. 040, 2009.
[58] Krugman, P. 1980. "Scale Economies, Product Differentiation, and the Pattern of
Trade". The American Economic Review, Vol. 70, pp. 950-959.
[59] Lennon, C., D. Mirza and G. Nicoletti. 2009. "Complementarity of Inputs across
Countries in Services Trade". Annals of Economics and Statistics, 93-94, 1-23.
[60] Manning W. and J., Mullahy. 2001. "Estimating log models: to transform or not to
transform?". Journal of Health Economics, 20 (4), 461-494.
[61] Martìnez-Zarzoso I, Nowak-Lehmann F, Vollmer S. (2007). "The log of gravity
revisited (in revision)". CEGE Discussion Paper 64, University of Gottingen.
[62] Mayer, T and Zignago, S. 2011. "Notes on CEPII’s distances measures: The GeoDist
database". CEPII, Working Paper, 2011-25.
[63] Melise Jaud and Madina Kukenova. 2011. "Financial development and survival of
African agri-food exports". World Bank Policy Research Working Paper No. 5649.
[64] Moore, M. 2000. "Financial services and the WTO". Speech Delivered at the Bank
for International Settlements. World Trade Organisation (WTO) (September 11,
Base l).
[65] Niepmann, Friederike and Tim Schmidt-Eisenlohr. 2013. "No Guarantees, No
Trade: How Banks Affect Export Patterns". Staff Report 659, Federal Reserve Bank
of New York 2013.
[66] NordÅs, H. K. 2010. "Trade in goods and services: Two sides of the same coin?".
Economic Modelling, vol. 27, no 2, pp. 496-506.
[67] NordÅs, HK and Rouzet, D. 2016. "The Impact of Services Trade Restrictiveness
on Trade Flows". The World Economy, Volume 40, Issue 6, 1155-1183.
[68] NordÅs, HK. 2016. "Services Trade Restrictiveness Index (STRI): The Trade Effect
of Regulatory Differences". OECD Trade Policy Papers, no 189, OECD Publishing,
Paris.
[69] Novy, D. 2009. "Gravity redux: measuring international trade costs with panel
data". University of Warwick- Department of Economics.
[70] Paravisini, Daniel, Veronica Rappoport, Philipp Schnabl, and Daniel Wolfen-
zon. 2014. "Dissecting the Effect of Credit Supply on Trade: Evidence from Matched
Credit-Export Data". Review of Economic Studies.
[71] Park, S.C. 2002. "Measuring Tariff Equivalents in Cross-Border Trade in Services".
Trade Working Papers no 353. East Asian Bureau of Economic Research.

32

Electronic copy available at: https://ssrn.com/abstract=3644987

[72] Piermartini, Roberta, and Yoto Yotov. 2016. "Estimating Trade Policy Effects with
Structural Gravity". WTO Working Paper ERSD-2016-10
[73] Riker, D. 2015. "Domestic Regulations and U.S. Exports". U.S. International Trade
Commission, Office of Economics, Research Division.
[74] Rouzet, D. and F. Spinelli. 2016. "Services Trade Restrictiveness, Mark-Ups and
Competition". OECD Trade Policy Papers, no 194. OECD Publishing, Paris.
[75] Rouzet, D and al. 2014. "Services Trade Restrictiveness Index (STRI)": Financial
Services. OECD Publishing.
[76] Santos Silva, J and Silvana Tenreyro. 2006. "The Log of Gravity". The Review of
Economics and Statistics, 88(4): 6416-658.
[77] Santos Silva, J.M.C.and Tenreyro, S. 2009. "Trading Partners and Trading Volumes:
Implementing the Helpman-Melitz-Rubinstein Model Empirically". Oxford bulletin
of economics and statistics.
[78] Santos Silva, J and Silvana Tenreyro. 2011. "Further simulation evidence on the
performance of the Poisson pseudo-maximum likelihood estimator". Economics
Letters, vol. 112, issue 2, 220-222.
[79] Tamar Khachaturian. 2015. " Services Trade Restrictions and Company Profits:
Telecommunications". United States International Trade Commission: Office of
Industries Working Paper ID-042 November 2015.
[80] Tamar Khachaturian. 2018. "Trade Restrictions and Modes of Supply in Services
Trade". Office of Industries Working Paper ID-05.
[81] Tinbergen, J. 1962. "Shaping the World Economy: Suggestions for an International
Economic Policy". American Journal of Agricultural Economics, Volume 46, Issue 1,
February 1964, Pages 271-273.
[82] UNCTAD. 2010. "Employment, globalization and development". Trade and devel-
opment report, United Nations.
[83] UNCTAD. 2013. "Non-Tariff Measures to Trade: Economic and Policy Issues for
Developing Countries". New York and Geneva: United Nations.
[84] UNCTAD-WTO. 2014. "A Practical Guide to Trade Policy Analysis". World Trade
Organization.
[85] Van der Marel, Erik and Shepherd Ben. 2013. "Services Trade, Regulation and
Regional Integration: Evidence from Sectoral Data". The World Economy36, no 11
(2013): 1393-1405.
[86] Walsh, K. 2008. "Trade in Services: Does Gravity Hold?". Journal of World Trade 42
(2), 315-334 .
[87] Walter, P., Dell’mour, R. 2010. "Firm Level Analysis of International Trade in
Services". Irving Fischer Committe on Central Bank Statistics Working Papers 4.
[88] Whalley, J. 2004. "Assessing the Benefits to Developing Countries of Liberalisation
in Services Trade". The World Economy.
[89] Westerlund, J. and F. Wilhelmsson. 2009. "Estimating the gravity model without
gravity using panel data". Applied Economics 43(6): pp. 641-649.

33

Electronic copy available at: https://ssrn.com/abstract=3644987

[90] World Trade Organization . 2005. "Analysis of the links between trade, trade
standards and the WTO". World Trade Report, WTO.
[91] Yildirim B. 2009. "A Key Element of Economic Growth. Transport for a Global
Economy. Challenges Opportunities in the Downturn". International Transport
Forum, OECD. pp. 8-10.
[92] Yotov, Y.V., Piermartini, R., Monteiro, J-A., Larch, M. 2016. "An advanced guide
to trade policy analysis: The structural gravity model". United Nations and World
Trade Organization. Geneva.

34

Electronic copy available at: https://ssrn.com/abstract=3644987