# ECONOMICS WORKING PAPER SERIES

ESTIMATED TARIFF EQUIVALENTS OF SERVICES NTMS

Lionel Fontagné
Cristina Mitaritonna
José E. Signoret

Working Paper 2016-07-A

500 E Street SW

Washington, DC 20436

## July 2016

Office of Economics working papers are the result of ongoing professional research of USITC Staff and are solely meant to represent the opinions and professional research of individual authors. These papers are not meant to represent in any way the views of the U.S. International Trade Commission or any of its individual Commissioners. Working papers are circulated to promote the active exchange of ideas between USITC Staff and recognized experts outside the USITC and to promote professional development of Office Staff by encouraging outside professional critique of staff research.

Estimated Tariff Equivalents of Services NTMs
Lionel Fontagné, Cristina Mitaritonna, and José E. Signoret
Office of Economics Working Paper 2016-07-A
July 2016

Lionel Fontagné

Paris School of Economics $— MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaacbaqcLbyaqa aaaaaaaaWdbiaa=rbiaaa@37C4@$ University Paris 1
Centre d’Études Prospectives et d’Informations Internationales (CEPII)

Cristina Mitaritonna

Centre d’Études Prospectives et d’Informations Internationales (CEPII)

José E. Signoret

U.S. International Trade Commission, Office of Economics

1. Introduction

As previous trade liberalizations have lowered tariffs worldwide, nontariff measures, or NTMs, remain as the most visible issues to be addressed. This is especially true in the case of services, where measures likely to affect services trade do not involve tariffs. In fact, significant trade agreements currently under negotiation, such as a number of mega-regional free trade agreements and an ambitious multilateral Trade in Services Agreement (TISA), particularly aim at reducing services NTMs. The result is that there is wide interest from both policy makers and economists in understanding the impact of these measures.

Unfortunately, quantifying the restrictiveness of NTMs for services has proven to be difficult. The main limitation has been the lack of comprehensive data, whether trade, policy, or microeconomic data.

Gathering data on actual policies has been especially challenging. Up until recently, information on the regulatory environment affecting services trade was not available for a wide range of countries, nor was it necessarily collected in a consistent fashion. Efforts by the World Bank (Borchert, Gootiiz, and Mattoo, 2012) and the Organisation for Economic Co-operation and Development (OECD) (Geloso Grosso et al., 2015) to collect this information and to develop services trade restrictiveness indexes as “summary statistics” of the level of restraints have been an important advance. By themselves, however, these indexes (often called STRIs) have no obvious economic meaning, and they would require some empirical analysis to associate them with trade costs.

One strand of the literature has used the STRI information to explain the observed international variation in price-cost margins, conditional on other firms and country factors. The pioneering work of Dee (2004) and Dihel and Sheperd (2007) provide early examples of this work. Fontagné and Mitaritonna (2013) conduct a similar analysis and stress the limitations of this approach. First, qualitative information must be arbitrarily transformed into a quantitative measure of the restrictiveness of each measure considered individually. Second, this approach is very data-consuming with respect to both microeconomic evidence and the determinants of price-cost margins in services sectors. As a consequence, estimated coefficients on restrictiveness indexes have to be used out of sample. This is a huge restriction, because the distribution of firms or markups within a given sector necessarily diverges across countries$— MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaqefmuySLMyYL gaiuaajugGbabaaaaaaaaapeGaa8hfGaaa@3A94@$even more so when countries at different levels of development are considered.

Jafari and Tarr (2015) nevertheless employ this method, using the World Bank STRI, to compute ad valorem equivalents (AVEs) for a large set of countries. Other recent research using STRI information has relied on gravity modeling. The analyses by van der Marel and Shepherd (2013) and Riker (2014) using the World Bank index, or by Nordås and Rouzet (2015) using the OECD STRI, are examples of gravity models that incorporate STRI values in their econometric specifications. Finally, Gooris and Mitaritonna (2015) exploit the ordinal nature of restrictiveness levels to assess the trade impact of a measure in place in a given country for a given sector. They conclude that the trade-impeding effects of restrictions may be nonlinear.

Although available for a large set of countries, data on services trade flows are much more aggregated than data on merchandise trade. But even if information on trade in services is limited, it is possible to use it to compute AVEs of the effects of barriers to trade in services on the quantities (instead of the prices) of traded services. Of course, bilateral trade data are not available for all possible trading partners, and the existing gaps have to be filled via estimation methods with reconstructed data. Sources of such data may include the services trade data in the Global Trade Analysis Project (GTAP) and other global databases.

Fontagné, Guillin, and Mitaritonna (2011) illustrate how to carry out such an analysis and then compare the results obtained with different identification strategies and different datasets. Their analysis uses a reduced form of the gravity approach to estimate services trade without relying on STRI information. The tariff equivalents are then inferred by comparing the inward multilateral resistance term for each country with that of a benchmark country. In this note, we build on this approach and provide AVEs of restrictions on trade in services for 117 countries in 2011. This estimation uses GTAP data, making it consistent with modeling efforts by several trade economists.

The econometric estimation used is this note is a practical approach that can be applied to obtain AVEs even for services sectors that do not line up with STRI sectors and for years for which there is no STRI information. It also bypasses the need for data on price-cost margins within sectors and countries. In fact, as in this note, estimates across many countries and sectors can be updated as more recent trade data become available. Further, this approach is not susceptible to how regulations are scored and how these scores are then weighted in forming the STRI.

Our results compare with those of Fontagné, Guillin, and Mitaritonna (2011), who used GTAP data for the year 2004. We use the same method, although the set of countries for which we have estimated the equivalents is larger. For the sake of comparison, we also provide results obtained with Fontagné, Guillin, and Mitaritonna’s original sample of countries, but for 2007 and 2011. In the latter case, we estimate and compute AVEs on the reduced sample of 64 countries.

The reader, however, must be aware that the estimated AVEs are only rough estimates of the restrictiveness of regulations on services. Our approach is likely to pick up trade cost factors beyond policy restraints, so that the tariff equivalent results may reflect a broad range of “frictions.” And although these estimates can easily be incorporated into large-scale trade models, they are not likely to be fully actionable. Berden et al. (2009), for instance, consider that, overall, 50 percent of NTM AVEs in goods and services in transatlantic trade may be actionable$— MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaqefmuySLMyYL gaiuaajugGbabaaaaaaaaapeGaa8hfGaaa@3A94@$and even that 50 percent is probably an ambitious goal. Evidence on this question has to be taken from trade negotiators or specialists in the concerned sectors, and it is beyond the scope of this note. Notice, however, that we systematically benchmark our estimations against a country with a zero AVE, although regulations on services may be present in this country. We are not measuring the cost of regulations, but rather their restrictiveness for trade flows.

The rest of this note is organized as follows. Section 2 presents the empirical approach used. Section 3 describes the data and econometric identification. Results are presented in section 4. The last section concludes the discussion. The full set of updated results is included with this note as an Excel file.

## 2. Empirical Framework

Many methodological issues can arise when using the gravity approach to construct AVEs of barriers to trade in services. First, the distribution of residuals of the estimated equation is sensitive to problems of specification and omitted variables, which affects the estimation of tariff equivalents. Hence it may be preferable to rely on a strategy based on country fixed effects. Second, an assumption must be made on the elasticity of substitution used to transform the parameter estimates to AVEs. The value of the equivalents is highly sensitive to this assumption, although the ranking of the restrictiveness of countries’ regulations in a given sector is not. These constraints led us to choose a simple identification strategy in tackling unobserved dimensions (like restrictiveness) in the importing country. Fontagné, Guillin, and Mitaritonna (2011) show that identifying the quantity effect of barriers to trade in services by relying on the GTAP database provides highly usable measurements of AVEs. We have built on this approach.

Our first step is to estimate a gravity equation in a cross-section relying on partially reconstructed data. In the GTAP database on trade in services for 2011, based on OECD data, the gaps for missing data are filled and then the database reconciled. There are accordingly two issues here: we rely on a very specific set of data as regards trade in services, namely the GTAP dataset of trade in services, while deriving estimates only with cross-sectional methods. These issues are not major obstacles, however, as illustrated by Fontagné, Guillin, and Mitaritonna (2011): while the latter study used alternative data sources and a panel database, the results were shown to be not too different.

## 3. Data and identification

We make use of a relative small set of data sources. The main source of data used in this paper remains the GTAP database, which provides bilateral trade in services for 15 services sectors for the year 2011: air transport (atp), communications (cmn), construction (cns), dwellings (dwe), electricity (ely), gas distribution (gdt), insurance (isr), other business services (obs), other financial intermediation (ofi), other government services (osg), other transport (otp), recreational services (ros), trade (trd), water (wtr), and water transport (wtp). This dataset has more countries than the GTAP database for the year 2004 that was used in Fontagné, Guillin, and Mitaritonna (2011). For sake of comparison, we provide tariff equivalents computed using alternatively the (reduced) sample of countries found in Fontagné, Guillin, and Mitaritonna (2011) and the enlarged sample of countries available in the GTAP database for the year 2011. For sake of comparison with Fontagné, Guillin, and Mitaritonna (2011), we keep the decomposition into the following services sectors: cmn, cns, obs, trn, trd, isr, ofi, and osg. In this listing, trn stands for transport, which includes the three transport sectors (atp, wtp, and otp). As in the reference article, wtp is also kept separately.

The gravity variables were obtained from the Centre d’Études Prospectives et d’Informations Internationales (CEPII) and updated for as necessary using data obtained from the World Bank.

For each services sector, the estimating equation used in the current analysis takes the form:[1]

$x ij =c+0.8 y j + α ij dis t ij + β ij D ij + γ i + γ j + ε ij MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadIhapaWaaSbaaSqaa8qacaWGPbGaamOAaaWd aeqaaOWdbiabg2da9iaadogacqGHRaWkcaaIWaGaaiOlaiaaiIdaca WG5bWdamaaBaaaleaapeGaamOAaaWdaeqaaOWdbiabgUcaRiabeg7a H9aadaWgaaWcbaWdbiaadMgacaWGQbaapaqabaGcpeGaamizaiaadM gacaWGZbGaamiDa8aadaWgaaWcbaWdbiaadMgacaWGQbaapaqabaGc peGaey4kaSIaeqOSdi2damaaBaaaleaapeGaamyAaiaadQgaa8aabe aak8qacaWGebWdamaaBaaaleaapeGaamyAaiaadQgaa8aabeaak8qa cqGHRaWkcqaHZoWzpaWaaSbaaSqaa8qacaWGPbaapaqabaGcpeGaey 4kaSIaeq4SdC2damaaBaaaleaapeGaamOAaaWdaeqaaOWdbiabgUca Riabew7aL9aadaWgaaWcbaWdbiaadMgacaWGQbaapaqabaaaaa@6304@$(1)

Where $x ij MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadIhapaWaaSbaaSqaa8qacaWGPbGaamOAaaWd aeqaaaaa@3C1B@$ represents the log of exports from country $i MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadMgaaaa@39D5@$ to partner $j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadQgaaaa@39D6@$. Trade costs other than regulations between $i MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadMgaaaa@39D5@$ and $j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadQgaaaa@39D6@$ are proxied by $dis t ij MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadsgacaWGPbGaam4CaiaadshapaWaaSbaaSqa a8qacaWGPbGaamOAaaWdaeqaaaaa@3EE6@$, the log of their bilateral distance. The vector $D ij MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadseapaWaaSbaaSqaa8qacaWGPbGaamOAaaWd aeqaaaaa@3BE7@$ contains bilateral trade determinants common in the gravity literature, including distance, contiguity, a colonial relationship, a common language, engagement in a free trade agreement, etc., here controlled by dummy variables. We include exporter $γ i MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeo7aN9aadaWgaaWcbaWdbiaadMgaa8aabeaa aaa@3BD6@$ and importer $γ j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeo7aN9aadaWgaaWcbaWdbiaadQgaa8aabeaa aaa@3BD7@$ fixed effects to estimate the usual multilateral resistance terms. Since we do not have time series, gross domestic products (GDPs) should collapse in these fixed effects. However, we would like to clean the importer fixed effect from the importer GDP, in order to capture only the degree of restrictiveness of trade in this importing country. We constrain the coefficient for expenditures in the importing country to 0.8, to normalize exports by the potential size of the market.[2] The last term represents an error term.

The last step involves the derivation of the tariff equivalents. There are two ways to compute the average protection applied by each importer: either from the estimated residuals (as in Park 2002), or else from the importer fixed effect coefficients. Although the importer fixed effect coefficient captures something larger than the protection itself, it is preferable to residuals. Not only do the latter contain mixed information other than protection, but their magnitude and goodness largely depend on the fit of the equation performed.

We need to examine the underlying theory in more depth to understand the assumptions involved in relying on the canonical gravity equation derived by Anderson and van Wincoop (2003) in order to compute the revealed trade barriers of country $j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadQgaaaa@39D6@$.

Exports from country $i MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadMgaaaa@39D5@$ to country $j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadQgaaaa@39D6@$ accounting for a share $S j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadofapaWaaSbaaSqaa8qacaWGQbaapaqabaaa aa@3B08@$ of world income can be expressed as a simple function of the product of their GDP and of trade costs.[3] Noting $y MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadMhaaaa@39E5@$ as GDP (subscript $w MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadEhaaaa@39E3@$ for world), $τ MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabes8a0baa@3AAC@$ trade costs (the power of the AVE) and $σ MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeo8aZbaa@3AAA@$ the elasticity of substitution, we have the following expression for the bilateral trade flows $x ij MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadIhapaWaaSbaaSqaa8qacaWGPbGaamOAaaWd aeqaaaaa@3C1B@$:

$x ij = y i y j y w ( τ ij Π i P j ) (1−σ) MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadIhapaWaaSbaaSqaa8qacaWGPbGaamOAaaWd aeqaaOWdbiabg2da9maalaaapaqaa8qacaWG5bWdamaaBaaaleaape GaamyAaaWdaeqaaOWdbiaadMhapaWaaSbaaSqaa8qacaWGQbaapaqa baaakeaapeGaamyEa8aadaWgaaWcbaWdbiaadEhaa8aabeaaaaGcpe WaaeWaa8aabaWdbmaalaaapaqaa8qacqaHepaDpaWaaSbaaSqaa8qa caWGPbGaamOAaaWdaeqaaaGcbaWdbiabfc6aq9aadaWgaaWcbaWdbi aadMgaa8aabeaak8qacaWGqbWdamaaBaaaleaapeGaamOAaaWdaeqa aaaaaOWdbiaawIcacaGLPaaapaWaaWbaaSqabeaapeGaaiikaiaaig dacqGHsislcqaHdpWCcaGGPaaaaaaa@54AE@$(2)

Where $P j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadcfapaWaaSbaaSqaa8qacaWGQbaapaqabaaa aa@3B05@$ refers to the price index of the demand function in $j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadQgaaaa@39D6@$ and $Π i MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabfc6aq9aadaWgaaWcbaWdbiaadMgaa8aabeaa aaa@3BAD@$ to a term which transmits the effect of an increase in the level of import restrictiveness for non-j services importers that stimulates exports from i to j:

$Π i = ( ∑ j S j ( τ ij / P j ) ) 1/(1−σ) MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabfc6aq9aadaWgaaWcbaWdbiaadMgaa8aabeaa k8qacqGH9aqpdaqadaWdaeaapeWaaabua8aabaWdbiaadofapaWaaS baaSqaa8qacaWGQbaapaqabaaabaWdbiaadQgaaeqaniabggHiLdGc caGGOaGaeqiXdq3damaaBaaaleaapeGaamyAaiaadQgaa8aabeaak8 qacaGGVaGaamiua8aadaWgaaWcbaWdbiaadQgaa8aabeaak8qacaGG PaaacaGLOaGaayzkaaWdamaaCaaaleqabaWdbiaaigdacaGGVaGaai ikaiaaigdacqGHsislcqaHdpWCcaGGPaaaaaaa@52AE@$(3)

How the estimated equation fits in is simple. In our model (equation 1), $y j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadMhapaWaaSbaaSqaa8qacaWGQbaapaqabaaa aa@3B2E@$ is observed, $y i Π i (σ−1) MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadMhapaWaaSbaaSqaa8qacaWGPbaapaqabaGc peGaeuiOda1damaaBaaaleaapeGaamyAaaWdaeqaaOWaaWbaaSqabe aapeGaaiikaiabeo8aZjabgkHiTiaaigdacaGGPaaaaaaa@4318@$ is captured by the exporter fixed effect, $P j (σ−1) τ ij (1−σ) MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadcfapaWaa0baaSqaa8qacaWGQbaapaqaa8qa caGGOaGaeq4WdmNaeyOeI0IaaGymaiaacMcaaaGccqaHepaDpaWaa0 baaSqaa8qacaWGPbGaamOAaaWdaeaapeGaaiikaiaaigdacqGHsisl cqaHdpWCcaGGPaaaaaaa@48B5@$ by the importer fixed effect, and $y w MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadMhapaWaaSbaaSqaa8qacaWG3baapaqabaaa aa@3B3B@$ is in the constant.

We must now assume that the regulation on a services sector in j has the same impact on exports for all affected partners i, hence $τ ij = τ j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabes8a09aadaWgaaWcbaWdbiaadMgacaWGQbaa paqabaGcpeGaeyypa0JaeqiXdq3damaaBaaaleaapeGaamOAaaWdae qaaaaa@4111@$ and that the impact on $Π i MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabfc6aq9aadaWgaaWcbaWdbiaadMgaa8aabeaa aaa@3BAD@$ of changes in $τ j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabes8a09aadaWgaaWcbaWdbiaadQgaa8aabeaa aaa@3BF5@$ is small enough to be neglected, as a result of a small enough $S j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadofapaWaaSbaaSqaa8qacaWGQbaapaqabaaa aa@3B08@$ (the weight of the destination country expenditure in j over the world production).

As regulations cannot be directly observed, one should compare the actual trade with a theoretical situation (subscript free), excluding any trade costs associated with restrictive regulations on trade in services in the destination j,[4] hence computing the theoretical ratio of $x ij / x ij free MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadIhapaWaaSbaaSqaa8qacaWGPbGaamOAaaWd aeqaaOWdbiaac+cacaWG4bWdamaaDaaaleaapeGaamyAaiaadQgaa8 aabaWdbiaadAgacaWGYbGaamyzaiaadwgaaaaaaa@43E3@$, which after simplification is equal to $α j σ−1 τ ij 1−σ MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeg7aH9aadaqhaaWcbaWdbiaadQgaa8aabaWd biabeo8aZjabgkHiTiaaigdaaaGccqaHepaDpaWaa0baaSqaa8qaca WGPbGaamOAaaWdaeaapeGaaGymaiabgkHiTiabeo8aZbaaaaa@46CD@$, with $α≡ P j / P j free MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeg7aHjabggMi6kaadcfapaWaaSbaaSqaa8qa caWGQbaapaqabaGcpeGaai4laiaadcfapaWaa0baaSqaa8qacaWGQb aapaqaa8qacaWGMbGaamOCaiaadwgacaWGLbaaaaaa@451F@$. If we assume that $P j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadcfapaWaaSbaaSqaa8qacaWGQbaapaqabaaa aa@3B05@$ and $P j free MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadcfapaWaa0baaSqaa8qacaWGQbaapaqaa8qa caWGMbGaamOCaiaadwgacaWGLbaaaaaa@3ECC@$ are similar, then the $α MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeg7aHbaa@3A86@$ term can be omitted.

As the theoretical value $x ij free MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadIhapaWaa0baaSqaa8qacaWGPbGaamOAaaWd aeaapeGaamOzaiaadkhacaWGLbGaamyzaaaaaaa@3FE2@$ cannot be observed either, there is a need to define a benchmark country, supposed to be the free trader of the sample. The benchmark is the country with the highest positive difference between the actual and predicted average import values. All calculations must be made relative to this benchmark, and one must normalize the ratio of actual to predicted trade for each importing country by the same ratio computed for the benchmark. Under the assumptions made above and after summation over j’s partners, the log of this double ratio therefore becomes the difference in the logs, as follows:

$ln (1+ t j ) 1−σ = γ j − γ 0 MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiGacYgacaGGUbGaaiikaiaaigdacqGHRaWkcaWG 0bWdamaaBaaaleaapeGaamOAaaWdaeqaaOWdbiaacMcapaWaaWbaaS qabeaapeGaaGymaiabgkHiTiabeo8aZbaakiabg2da9iabeo7aN9aa daWgaaWcbaWdbiaadQgaa8aabeaak8qacqGHsislcqaHZoWzpaWaaS baaSqaa8qacaaIWaaapaqabaaaaa@4B96@$(4)

Where $γ 0 MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeo7aN9aadaWgaaWcbaWdbiaaicdaa8aabeaa aaa@3BA2@$ is the highest importer fixed effect coefficient.

From equation 1 we get the estimation of $ln (1+ t j ) 1−σ MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiGacYgacaGGUbGaaiikaiaaigdacqGHRaWkcaWG 0bWdamaaBaaaleaapeGaamOAaaWdaeqaaOWdbiaacMcapaWaaWbaaS qabeaapeGaaGymaiabgkHiTiabeo8aZbaaaaa@43D4@$, however the specific values of the tariff equivalents ( $t j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiaadshapaWaaSbaaSqaa8qacaWGQbaapaqabaaa aa@3B29@$ ) will also depend on the elasticity of substitution $σ MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeo8aZbaa@3AAA@$ that is not estimated in the model, but needs to be assumed.

Following Fontagné, Guillin, and Mitaritonna (2011) and for sake of comparison, we use the assumption that $σ=5.6 MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeo8aZjabg2da9iaaiwdacaGGUaGaaGOnaaaa @3DE1@$. A higher $σ MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbdfgBPj MCPbqedmvETj2BSbqefm0B1jxALjhiov2DaerbuLwBLnhiov2DGi1B TfMBaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaabaqaciGacaGaaeqabaWaaqaafaaake aaqaaaaaaaaaWdbiabeo8aZbaa@3AAA@$ provides lower AVEs, and vice versa. The relative ranking among the different countries, however, is not sensitive to the assumed value of the elasticity of substitution.

## 4. Results

We now show the results of the estimated equations using data for 2011 and the largest set of countries (117) and for nine sectors of services, disregarding water, electricity, gas manufacture and distribution, recreational and other services, and dwellings. As noted, air transport and other transport are not considered separately, but included in all transports (trn), a category that also includes water transport. Distance has a negative impact on trade in services, because distance is a proxy of many other dimensions besides transport costs (e.g., communication costs). Contiguity and a colonial relationship facilitate trade in services, because of the existence of business networks and the fact that certain services (such as construction) are provided cross border. The liberalization of trade in services in Europe shows up in the positive coefficient for Europe, and the same conclusion holds for NAFTA, notwithstanding the lower integration in North America. Again, business services and the presence of affiliates of foreign companies may play a role here. What is observed in Europe or North America contrasts sharply with the observed parameter estimate for Latin America.

Most important for us, however, are the importing country fixed effects, as explained above. We now rely on these fixed effects to compute the AVEs, using an elasticity of substitution of 5.6. Results are shown in table 2 for the top five more open countries and the top five more protective ones, for each category of services. We report in parentheses in table 2 also the fixed effect of the best performer, so that the estimated AVEs in 2011 can be recalculated using different elasticities of substitutions.

An Excel file with the complete set of results for the 9 sectors and the 117 countries is available with this note. This file does not report results for a limited number of pairs (sector, importer), as the parameter estimates were not significant. This set of unreported pairs corresponds to 7.7 percent of the total number of pairs. In figure 1 we show the distribution of estimated AVEs by services sector for the remaining 92.3 percent of the sector-importer pairs.

For the sake of comparison with Fontagné, Guillin, and Mitaritonna (2011), who computed AVEs for 64 countries in 2004, we perform the same calculation for 2007 and 2011, restricting our estimation of parameters and derivation of tariff equivalents to these 64 countries. Results are provided in the Excel file.

Table 1 Results of estimations for whole GTAP sample of countries, 2011

 cmn cns isr obs ofi trd trn osg wtp Log distance -0.0398*** -0.0902*** -0.0206** -0.0672*** -0.0739*** -0.0600*** -0.0318*** -0.0509*** -0.0102 (0.00820) (0.0171) (0.00948) (0.0100) (0.0118) (0.0105) (0.00909) (0.00977) (0.0150) Contiguity 0.389*** 0.292*** 0.444*** 0.325*** 0.370*** 0.411*** 0.368*** 0.407*** 0.465*** (0.0328) (0.0661) (0.0378) (0.0401) (0.0469) (0.0416) (0.0361) (0.0388) (0.0607) Common language -0.0100 0.0172 -0.00119 -0.0209 -0.0409* -0.0307 -0.0569*** -0.0135 -0.00993 (0.0169) (0.0348) (0.0196) (0.0207) (0.0242) (0.0217) (0.0187) (0.0202) (0.0313) Colonial relationship 0.257*** 0.177** 0.197*** 0.272*** 0.266*** 0.385*** 0.318*** 0.295*** 0.222*** (0.0380) (0.0730) (0.0436) (0.0465) (0.0532) (0.0482) (0.0423) (0.0456) (0.0681) ASEAN -0.0851 0.183 -0.00758 -0.0768 -0.0973 0.00422 -0.115 -0.117 -0.0942 (0.0818) (0.157) (0.0939) (0.100) (0.114) (0.104) (0.0912) (0.0982) (0.146) Latin America -0.325*** 0.0879 -0.346*** -0.324*** -0.353*** -0.309*** -0.264*** -0.275*** -0.224** (0.0508) (0.113) (0.0591) (0.0622) (0.0730) (0.0645) (0.0565) (0.0609) (0.0929) Europe 0.0852*** 0.334*** 0.218*** 0.325*** 0.501*** 0.502*** 0.178*** 0.379*** 0.317*** (0.0268) (0.0523) (0.0308) (0.0328) (0.0377) (0.0341) (0.0299) (0.0322) (0.0483) NAFTA 0.809*** -0.758* 0.295 0.533** 0.926*** 1.222*** 0.703*** 0.726*** 0.773** (0.207) (0.397) (0.238) (0.254) (0.290) (0.263) (0.231) (0.249) (0.371) ANZCERTA 0.143 0.819 0.397 0.0191 0.219 0.370 0.177 0.416 0.472 (0.354) (0.678) (0.407) (0.434) (0.495) (0.449) (0.395) (0.426) (0.633) Constant -3.102*** -4.929*** -2.099*** -0.977*** -1.479*** -2.946*** -1.236*** -0.127 -6.659*** (0.102) (0.204) (0.117) (0.124) (0.144) (0.129) (0.113) (0.121) (0.184) Obs 13,555 12,195 13,380 13,468 12,987 13,428 13,651 13,746 12,877 Importer fixed effects yes yes yes yes yes yes yes yes yes Exporter fixed effects yes yes yes yes yes yes yes yes yes

Notes: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1.

Sectors include communication (cmn); construction (cns); insurance (isr); business services (obs); finance (ofi); trade (trd); all transport (trn); other services (osg), comprising education, health, defense and public administration; and water transport (wtp).

Constrained parameter on GDPj not reported.

Definition of regional dummies as in Park (2002) and Fontagné, Guillin, and Mitaritonna (2011), for the sake of comparison: ASEAN = KHM, MYS, PHL, SGP, THA, VNM, and IND; Latin America = BOL, BRA, CHL, COL, ECU, MEX, NIC, PER, PRY, URY, VEN, and ARG; Europe = AUT, BEL, BGR, CYP, CZE, DEU, DNK, ESP, EST, FIN, FRA, GBR, GRC, HUN, IRL, ITA, LTU, LUX, LVA, MLT, NLD, POL, PRT, ROM, SVK, SVN, and SWE.

Source: Authors’ calculations.

Table 2 Top five free traders and restrictive importers of services, ad valorem equivalent, percent, by sector, 2011

 Sector Top 5 more open countries, by sector Top 5 more restrictive countries, by sector cmn LUX DNK BEL SGP GBR BGD PRY NAM BRA TTO 0.0 (2.121) 15.1 17.3 19.0 19.5 140.6 143.4 147.5 148.9 172.3 obs IRL LUX SGP MLT BEL ECU MEX PRY BFA LAO 0.0 (1.769) 8.4 11.8 15.2 16.3 153.6 155.5 157.8 170.7 280.1 trn SGP HKG GRC DEU GBR BFA BGD NAM ZMB LAO 0.0(1.223) 0.4 6.6 7.3 9.1 79.3 84.2 85.0 85.5 148.0 wtp GRC NOR SGP HKG ISR BFA PRY ZMB BGD LAO 0.0(3.667) 7.6 11.3 16.8 20.1 186.3 192.2 194.4 228.2 275.1 cns AZE SAU ZMB LUX MYS BFA BRA PRY BGD TTO 0.0(2.815) 0.5 6.8 7.7 12.8 161.4 170.6 173.5 180.5 180.6 isr IRL LUX DNK SGP MEX MOZ RWA LAO BLR TTO 0.0(1.569) 10.8 23.0 23.9 26.5 124.8 130.0 131.8 135.6 170.8 ofi LUX IRL HKG TWN MLT ZMB GIN BGD TTO LAO 0.0(2.292) 24.3 31.9 33.6 34.3 165.8 170.9 174.7 196.4 290.6 trd IRL SGP LUX HKG NLD GIN TGO BGD ZMB LAO 0.0(2.454) 13.5 17.8 21.6 25.9 126.4 131.0 147.9 151.3 217.5 osg SAU KWT BRN MDG UKR ZMB IND ETH TGO LAO 0.0(1.212) 8.9 35.2 35.6 38.3 107.6 112.9 119.5 129.3 146.3

Note: Numbers in parentheses are estimated fixed effects for the benchmark countries.

Source: Authors’ calculations.

Figure 1 Kernel density of AVEs estimated for 117 countries and 9 services sectors, ad valorem equivalent, 2011

Source: Authors' calculations.

## 5. Conclusion

This note derives a set of AVEs of restrictions on cross-border trade in services for 118 countries and 9 sectors, using the GTAP database. The trade data refers to 2011. These equivalents are derived from a quantity method using a gravity model of trade. The econometric estimation is performed sector by sector and the reported AVEs in this note and in the accompanying dataset are based on an assumption of common elasticity of substitution across sectors. Based on the econometric estimation and the elasticity assumption, AVEs for frictions affecting cross-border services trade are large in many sectors. It should be noticed, however, that only a fraction of these estimated costs are reasonably subject to policy changes (or “actionable”) in most circumstances.

## References

Anderson, James E., and Eric van Wincoop. “Gravity with Gravitas: A Solution to the Border Puzzle.” American Economic Review 93, no. 1 (2003): 170$– MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaqefmuySLMyYL gaiuaajugGbabaaaaaaaaapeGaa83eGaaa@3A93@$192.

Berden, Koen, Joseph Francois, Martin Thelle, Paul Wymenga, and Saara Tamminen. “Non-Tariff Measures in EU-US Trade and Investment$— MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaqefmuySLMyYL gaiuaajugGbabaaaaaaaaapeGaa8hfGaaa@3A94@$An Economic Analysis.” Ecorys. Report for the European Commission, OJ 2007/S 180-219493, 2009. https://trade.ec.europa.eu/doclib/docs/2009/december/tradoc_145613.pdf.

Borchert, Ingo, Batshur Gootiiz, and Aaditya Mattoo. “Guide to the Services Trade Restrictions Database.” World Bank Policy Research Working Paper WPS6108, 2012. https://documents.worldbank.org/curated/en/2012/06/16441094/guide-services-trade-restrictions-database.

Dee, Philippa. “Services Trade Liberalisation in South East European Countries.” Background Paper. Paris: OECD Publishing, January 2004.

Dihel, Nora, and Ben Shepherd. "Modal Estimates of Services Barriers." OECD Trade Policy Papers No. 51. Paris: OECD Publishing, 2007.

Fontagné, Lionel, Amélie Guillin, and Cristina Mitaritonna. “Estimation of Tariff Equivalents for the Services Sectors.” CEPII working paper No 2011$– MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaqefmuySLMyYL gaiuaajugGbabaaaaaaaaapeGaa83eGaaa@3A93@$24, 2011.

Fontagné, Lionel, and Cristina Mitaritonna. “Assessing Barriers to Trade in the Distribution and Telecom Sectors in Emerging Countries.” World Trade Review 12, no. 1 (2013): 57$– MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaqefmuySLMyYL gaiuaajugGbabaaaaaaaaapeGaa83eGaaa@3A93@$78. DOI: 10.1017/S1474745612000456.

Geloso Grosso, Massimo, Frédéric Gonzales, Sébastien Miroudot, Hildegunn Kyvik Nordås, Dorothée Rouzet, and Asako Ueno. "Services Trade Restrictiveness Index (STRI): Scoring and Weighting Methodology." OECD Trade Policy Papers No. 177. Paris: OECD Publishing, 2015.

Gooris, Julien, and Cristina Mitaritonna. “Which Import Restrictions Matter for Trade in Services?” Working paper no. 2015-33. Paris: CEPII research center, 2015.

Jafari, Yaghoob, and David G. Tarr. “Estimates of Ad Valorem Equivalents of Barriers against Foreign Suppliers of Services in Eleven Services Sectors and 103 Countries.” World Economy, October 15, 2015. DOI: 10.1111/twec.12329.

Nordås, Hildegunn Kyvik, and Dorothée Rouzet. "The Impact of Services Trade Restrictiveness on Trade Flows: First Estimates." OECD Trade Policy Papers no. 178. Paris: OECD Publishing, 2015.

Park, S.C. “Measuring Tariff Equivalents in Cross-Border Trade in Services.” Trade Working Papers no. 353. East Asian Bureau of Economic Research, 2002.

Riker, David. “Estimates of the Impact of Restrictions on Cross-Border Trade in Services.” U.S. International Trade Commission. Office of Economics Research Note No. RN-2014-08A, 2014.

Van der Marel, Erik, and Ben Shepherd. “Services Trade, Regulation and Regional Integration: Evidence from Sectoral Data.” World Economy 36, no. 11 (2013): 1393$– MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaqefmuySLMyYL gaiuaajugGbabaaaaaaaaapeGaa83eGaaa@3A93@$1405.

## Appendix

Table A.1 Regional trade agreements: member listing

 Dummy Members of the agreement NAFTA Canada, Mexico, United States Europe Austria, Belgium, Bulgaria, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, United Kingdom ASEAN Brunei, Burma (Myanmar), Cambodia, Indonesia, Laos, Malaysia, Philippines, Singapore, Thailand, Vietnam ANZCERTA Australia, New Zealand

[1] This is model 2 in Fontagné, Guillin, and Mitaritonna (2011).

[2] Theory suggests an elasticity of 1, although it is often found to vary from that value. Based on past experience, we constrain this parameter to 0.8, but this choice does not affect the results.

[3] In the case of services under Mode 1 there are no transport costs and trade costs are simply the revealed protection exerted by the presence of regulation.

[4] Notice that τ equals 1 in absence of trade barriers.