Policy Research Working Paper 9700 Heterogeneous Impacts of SPS and TBT Regulations Firm-Level Evidence from Deep Trade Agreements Ana Margarida Fernandes Kevin Lefebvre Nadia Rocha Development Economics Development Research Group & Macroeconomics, Trade and Investment Global Practice June 2021 Policy Research Working Paper 9700 Abstract This paper estimates the impacts of regulating the use of harmonization of sanitary and phytosanitary regulations in sanitary and phytosanitary and technical barriers to trade preferential trade agreements also have greater benefits for measures through preferential trade agreements on exports the exports of smaller firms, and so do preferential trade of firms in Chile, Colombia, and Peru along the firm size agreements, including stronger transparency provisions spectrum. The analysis exploits novel data from the World for sanitary and phytosanitary and technical barriers to Bank Deep Trade Agreements database and customs cov- trade regulations. The results are robust to dropping larger ering the universe of exporting firms in each country over exporters and highly concentrated export sectors to address 1996–2015. The paper uses a firm-product gravity equation endogeneity. The benefits of sanitary and phytosanitary and with a stringent set of fixed effects and controls for the overall technical barriers to trade provisions are mainly driven by depth of the preferential trade agreements and product-spe- sectors with more heavily-regulated products. Entry into cific bilateral tariffs. The findings show that firms’ exports new product markets and increases in export quality partly increase significantly in destination markets with preferen- explain the rising exports of smaller firms. Finally, the esti- tial trade agreements, including a larger number of sanitary mated impacts are similar regardless of the income level of and phytosanitary and technical barriers to trade provisions, the preferential trade agreement partners. and the effect is stronger for smaller firms. Provisions for the This paper is a product of the Development Research Group, Development Economics and the Macroeconomics, Trade and Investment Global Practice.. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at afernandes@worldbank.org, klefebvre@worldbank.org, and nrocha@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Heterogeneous impacts of SPS and TBT regulations: Firm-Level Evidence from Deep Trade Agreements Ana Margarida Fernandes1 , Kevin Lefebvre2 , and Nadia Rocha∗ 1 1 World Bank 2 CEPII JEL Classification: F13, F14, F15, Q17, O13, L15. Keywords: non-tariff measures, product standards, preferential trade agreements, exporter dynamics, firm heterogeneity. ∗ World Bank, 1818 H Street, Washington DC, USA. Ana Margarida Fernandes, Email: afernandes@worldbank.org ; Kevin Lefebvre, Email: kevin.lefebvre@cepii; Nadia Rocha, Email: nrocha@worldbank.org. We thank Bishakha Barman and Natalia Camelo for excellent research assistance. We are also grateful to Scott Baier, José De Sousa, Anne-Célia Disdier, Lionel Fontagné, Roberta Piermartini and participants at the CESifo Area Conference on the Global Economy and the World Bank webinar on the economics of deep trade agreements for comments. This paper has benefited from support from the Umbrella Facility for Trade trust fund financed by the governments of the Netherlands, Norway, Sweden, Switzerland and the United Kingdom. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank of Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the countries they represent. All errors are our responsibility. 1 Introduction The benefits of trade liberalization are uneven across firms, partly due to the costs of enter- ing export markets (Bernard et al., 2007, 2012; Melitz and Redding, 2015). Fixed market entry costs imply that the largest and most productive firms are the most likely to enter and successfully grow in export markets (Melitz, 2003; Chaney, 2008; Bernard et al., 2011).1 An important component of fixed market entry costs is diversity across countries in regulations, standards, testing and certification procedures referred to as non-tariff measures (NTMs), and specifically Sanitary and Phyto-Sanitary (SPS) and Technical Barriers to Trade (TBT) regulations (Fontagné and Orefice, 2018).2 While SPS and TBT regulations are generally imposed with legitimate goals to address market failures and protect consumers, health, or the environment, they entail for firms additional investments in technology, costs of adjusting products or production processes, and costs of certifications in order to comply with divergent regulations imposed by destination countries. These compliance costs can be prohibitive for small firms (WTO, 2012). Preferential Trade Agreements (PTAs) have increasingly regulated the use of NTMs between member countries, with special attention given to SPS and TBT measures.3 Accordingly, different integration approaches have been followed across PTAs. Some PTAs include a large number of SPS and TBT provisions (e.g., the Comprehensive and Progressive Agreement for Trans-Pacific Partnership of 2018) while others include very few (e.g., the Canada-Chile PTA of 1997). Some PTAs mutually recognize members’ regulations as being equivalent to their domestic ones (e.g., the Peru-Chile PTA of 2009), other PTAs opt for a harmonization process to unify their domestic regulations with international or regional standards (e.g., the 1 The prevalence of export superstars that account for the bulk of exports across countries reported in (Freund and Pierola, 2015) is a consequence of this fact. 2 SPS and TBT are the most prevalent types of NTMs (UNCTAD, 2012; Espitia et al., 2020). 3 While the agreement on SPS (1995) and the agreement on TBT (1980) negotiated at the WTO provide some basic transparency provisions and encourage the use of international standards, PTAs go beyond the scope of those agreements by strengthening the regulatory cooperation between countries through specific patterns of integration such as harmonization or mutual recognition, and transparency commitments that increase the predictability of SPS and TBT measures (i.e., requirements on electronic publications in Chile - EU, 2003). 1 EU-Chile PTA of 2003).4 These distinct approaches could affect the fixed costs of exporting for firms in different ways. Mutual recognition of member countries’ regulations is expected to reduce the fixed costs for firms from having to comply with a multiplicity of regulations (Baldwin, 2000). Firms only need to comply with their domestic regulation to have their products accepted in PTA partners’ markets. The harmonization of member countries’ reg- ulations may initially increase the fixed costs of exporting for firms as they need to cope with the newly established standards. Then the use of a common standard (especially if it is an international standard) may increase the size of the market allowing firms to exploit economies of scale.5 But evidence on the impacts of including SPS and TBT provisions in PTAs on firm export performance is lacking. Do the SPS and TBT regulations incorporated in PTAs benefit small firms or are benefits mitigated since large firms influence the design of PTAs as suggested by Rodrik (2018)? Do mutual recognition and harmonization of PTA member countries’ regulations have a heterogeneous impact across firms? This paper addresses these questions by investigating the impact of PTAs regulating SPS and TBT on firms along the size spectrum combining two main novel sources of data. The first source of data we exploit is the World Bank’s Deep Trade Agreements database of Mattoo et al. (2020) that contains all provisions related to the policy areas of SPS and TBT (as well as many other policy areas) included in PTAs signed over the last 40 years. We combine this data with longitudinal administrative firm-level databases from customs in Chile, Colombia, and Peru covering the universe of exporters between 1996 and 2015 part of the Exporter Dynamics Database described in Fernandes et al. (2016). These three countries share many similarities in their economic and trade structures and constitute an excellent laboratory for studying the impacts of PTAs as they were signatories to close to 40 new PTAs with a variety of trading partners that entered into force during the sample period. Our empirical strategy isolates the change in firm export performance driven by the regula- 4 Common standards within a PTA can be existing international standards (e.g., the Codex Alimentarius Committee standards for food safety) or new regional standards developed by the PTA members. 5 Schmidt and Steingress (2019) argue that an expected benefit from harmonization is higher demand if harmonized rules are based on consumer preferences and on compatibility between products across countries. 2 tion of SPS and TBT in each PTA from tariff removal and trade integration in other pol- icy areas awarded by the PTA in a bilateral firm-product gravity equation estimated using Poisson Pseudo-Maximum Likelihood estimation (PPML) and controlling for firm-product- destination fixed effects. Specifically, we estimate the impact of SPS and TBT provisions in a PTA on exports within a firm-product-destination after the entry in force of the PTA. Impor- tantly, our gravity equation controls for unobserved time-varying firm-product-specific fixed effects that capture productivity and demand shocks. In order to investigate the heteroge- neous impacts across firms, we interact the measures capturing the SPS and TBT provisions in the PTA with size categories defined based on measures of firm export market shares in their initial year. Our specifications control for bilateral product-specific tariff rates and for a measure of the depth of the PTA in policy areas other than SPS and TBT to isolate the im- pact of special provisions such as SPS and TBT from tariff concessions and other integration provided by the PTAs, as well as for destination markets’ GDP to account for time-varying global demand. Preferential trade agreements contain varying patterns of SPS and TBT integration. We consider different types of SPS and TBT provisions and examine whether they have differ- ential impacts on exports across the firm size spectrum. First, we investigate the impact of SPS and TBT provisions flagged as essential in order to capture regulatory changes more likely to bring deeper integration of SPS and TBT among partner countries.6 Second, we consider provisions on the mutual recognition and harmonization of SPS and TBT to un- derstand what type of integration of SPS and TBT among PTA partner countries is most conducive to firm export growth. Finally, we consider the role of transparency of SPS and TBT provisions which could be important to reduce information asymmetries across firms regarding SPS and TBT regulations and prevent governments from introducing SPS or TBT 6 Essential provisions are those categorized by experts as (i) substantive, that is, those that require specific liberalization and integration commitments and obligations, or (ii) related to procedures that are crucial to achieve the substantive commitments (Mattoo et al., 2020). Essential provisions are key to make the commitments in PTAs effective in achieving integration across member countries. Details on essential SPS and TBT provisions are provided in Section 2. 3 regulations without scientific evidence. Our main findings are as follows. SPS and TBT provisions in PTAs have a clear impact on exports that is heterogeneous across firm size categories. We find that (i) firm exports increase significantly in destination markets with PTAs including a larger number of overall or essential SPS and TBT provisions and the effect is stronger for smaller firms, (ii) exports of smaller firms to destination markets with PTAs that include provisions for the harmoniza- tion of SPS regulations benefit significantly more than those of larger firms and (iii) exports of smaller firms to destination markets with PTAs with more transparency provisions of SPS and TBT regulations increase significantly more than those of larger firms. Our results are robust to alternative measures of firm size and different product aggregation levels. The evi- dence further shows that the benefits of SPS and TBT provisions in PTAs for the exports of smaller firms are mainly driven by sectors with higher de facto SPS and TBT trade barriers and more heavily-regulated products (chemical, wood, metals and machinery). Moreover, en- try into new product-markets partly explains the rising exports of smaller firms and so does an increase in export quality. Finally, the estimated impacts of SPS and TBT provisions in PTAs on firm exports are similar regardless of the income level of the PTA partners. One concern about our main findings is that the content of the PTAs could be endogenous, influenced by large firms as argued by Rodrik (2018) and shown to be the case for US firms by Blanga-Gubbay et al. (2020). While credible instruments for the SPS and TBT provisions in PTAs are not available, the (origin-)firm-product-destination fixed effects included in our specification play the same role (at a more granular level) as the country-pair fixed effects included in aggregate gravity equations to control for endogeneity of PTAs and their content (e.g., Baier and Bergstrand (2007); Disdier et al. (2014); Anderson and Yotov (2020)). It is possible that these fixed effects do not fully eliminate a bias due to large firms influencing the content of PTAs to their advantage, but our main finding is that SPS and TBT provisions in PTAs benefit less not more the exports of large firms. Still, we explicitly address endogeneity by excluding from the sample either the largest exporting firms in Chile, Colombia, and Peru 4 or the sectors with the highest degree of concentration of exports across firms, which may be better organized for lobbying. Either way, our main findings are robust. The fact that we estimate similar impacts of SPS and TBT provisions both in PTAs signed with other developing countries (South-South) for which Chilean, Colombian, and Peruvian firms may have some bargaining power in the negotiations as well as in PTAs signed with high-income countries (North-South) in which such bargaining power is weaker also supports the idea that our estimates do not suffer from an endogeneity bias. The integration of SPS and TBT in a PTA may affect different dimensions of trade costs. SPS and TBT regulations constitute to a large extent fixed costs of exporting if they require e.g., firms to change their production process. But they may also contribute to iceberg-type variable costs of exporting if they require firms to use higher quality (more costly) inputs to meet certain standards. And they can be an additive cost if they require firms to go through a certification procedure for each unit exported (Martin, 2012). While mutual recognition of PTA partners’ SPS and TBT regulations is expected to reduce the fixed costs of exporting, the harmonization of such regulations could increase costs for firms to comply with the newly established common standards, at least in the short run. However, if these common standards are more stringent, their implementation may increase consumer confidence in the products and expand market size. Predictions from standard trade models with heterogeneous firms imply that the integration of SPS or TBT in PTAs, by reducing fixed costs of entry, should have a positive effect on the extensive margin of trade, as small firms are able to enter the export market (Melitz, 2003; Chaney, 2008; Crozet and Koenig, 2010). The effect on the intensive margin of trade is ambiguous.7 Our finding of stronger impacts of SPS and TBT provisions in PTAs on exports of smaller firms is consistent with the theoretical predictions but it is unable to disentangle the precise contribution of the extensive versus the intensive margin. But the finding on firm entry into new product-markets is an explicit result on ex- 7 This ambiguity is due to the fact that lower variable costs increase firm exports, while lower fixed entry costs can actually decrease firm exports through the indirect effect of stronger competition in the destination country. 5 tensive margin impacts of SPS and TBT inclusion in PTAs being stronger for smaller firms. Our paper relates to three literatures. A first extensive literature studies the impacts of preferential trade agreements on trade using structural gravity equations. Some selected key papers are Baier and Bergstrand (2007); Baier et al. (2014); Mattoo et al. (2017); Larch et al. (2018); Baier et al. (2019); Correia et al. (2020); Disdier et al. (2020); Anderson and Yotov (2020). These papers provide evidence of a strong positive impact of PTAs on bilat- eral trade based on state-of-the-art estimates of gravity equations by PPML. Mattoo et al. (2017) use the same World Bank Deep Trade Agreements database that we exploit and show that deep agreements lead to more trade creation and less trade diversion than shallow (not deep) agreements.8 We innovate relative to previous studies by being the first to examine the impact of SPS and TBT provisions in PTAs on exports at the firm-product level. To our knowledge, only two papers examine the impact of SPS and TBT liberalization in the con- text of PTAs between high-income (mostly the European Union) and developing countries and focusing on trade at the product level and find that such liberalization fosters trade, especially when enforceable international regulatory mechanisms are included (Disdier et al., 2014, 2020). The second literature studies the role of harmonization and mutual recognition of stan- dards for trade. Conceptually, discussions on the contrast between mutual recognition and harmonization of standards for trade are provided in Baldwin (2000) and Costinot (2008). Empirically, harmonization and mutual recognition of standards among high-income coun- tries are shown to increase trade, with a larger effect for mutual recognition by Chen and Mattoo (2008). Disdier et al. (2014) find that the harmonization of TBT standards in PTAs between high-income countries and developing countries has a detrimental effect on their trade when the harmonization is done on the basis of regional standards. A few studies take a micro perspective and examine impacts on firms. Portugal-Perez et al. (2010) show that the harmonization of EU electronics regulations with international standards led to the en- 8 The paper uses several measures of the depth of PTAs based on the extent to which the different policy areas are covered and are legally enforceable. 6 try of new US exporters of electronics products into the EU. Schmidt and Steingress (2019) quantify the effect of the harmonization of standards on French firm exports. Exploiting har- monization events based on a novel cross-country standard equivalences database, they show the introduction of harmonized standards increases trade through entry and higher exports by existing exporters.9 Our contribution to this literature is to show the heterogeneous effect of provisions for the harmonization of SPS in PTAs on firm exports, namely the stronger benefit for exports of small firms relative to large firms. A third, smaller literature examines the effects of regulatory trade barriers on firm-level trade. Fontagné et al. (2015) estimate the impact of SPS concerns raised in the WTO Committee on SPS - which are SPS measures perceived as de facto trade barriers - on French firms’ exports using customs data. They show that SPS concerns reduce firms’ exports at the extensive margin (the probability of exporting) and at the intensive margin (export value) but increase firms’ export prices. The negative impact of SPS concerns is lower for larger firms exporting to multiple destinations. Fontagné and Orefice (2018) focus on TBT concerns raised in the WTO committee on TBT and show they have a negative effect on French firms’ export par- ticipation that is magnified for multi-destination firms able to divert their exports towards destinations with no TBT. Fernandes et al. (2019) estimate the effect of product standards (on pesticides) on firms’ export decisions using customs data for 42 developing countries. They show that an increase in the stringency of product standards in the destination coun- try, relative to the exporting country, lowers firms’ probability of exporting, deters exporters from entering new markets, and fosters exit from existing markets. Smaller exporters are more affected in their market entry and exit decisions by the relative stringency of standards than larger exporters. Macedoni and Weinberger (2019) use a large panel of countries to show that countries have fewer exporters selling to destinations with larger numbers of standards, 9 To rationalize their results, they propose a trade model with heterogeneous firms and endogenous adop- tion of product standards. Product standards reduce information asymmetries and ensure the compatibility of inputs and devices and this increases demand by consumers. This increases firms’ incentives to produce goods that meet the standard despite high investment costs and higher marginal costs to produce meeting the standard. Their model’s prediction is that a new standard (through harmonization) leads to both entry of exporters as well as increases in exports of existing exporters. 7 but the average value per exporter is not affected. They argue this is strong evidence that quality standards act as a fixed cost to trade. Relative to these studies, our novelty is to examine the integration of regulatory barriers in PTAs and to show that PTAs alleviate the regulatory burden with a particular benefit for small firms. The remainder of the paper is structured as follows. Section 2 describes the data sources. Section 3 describes our empirical approach. Section 4 describes our main results while 5 presents results from several robustness checks and extensions. Section 6 concludes. 2 Data and Descriptive Statistics 2.1 Deep Trade Agreements Database Our analysis is based on the novel database on the content of preferential trade agreements described in Mattoo et al. (2020).10 The database contains detailed information on the commitments included in 18 policy areas across 283 PTAs signed and notified to the WTO between 1958 and 2017.11 SPS and TBT provisions are covered in, respectively, 269 and 263 PTAs signed between 1960 and 2017.12 The templates used for the mapping of these provisions cover areas such as (a) the reference to the WTO rules; (b) the type of integration approach (harmonization versus mutual recognition) for standards, technical regulations, and conformity assessment procedures, (c) transparency; (d) institutions or mechanisms to ad- minister the agreement and solve disputes; and (e) cooperation among regional partners on standards-related issues.13 A set of essential provisions are identified which comprise substantive provisions, that is, 10 See https://datatopics.worldbank.org/dta/index.html. 11 The policy areas are: anti-dumping, countervailing duties, competition policy, environmental laws, ex- port taxes, investment, intellectual property rights, labor market, migration (visa and asylum), movement of capital, public procurement, SPS, services, state-owned enterprises, subsidies, TBT, rules of origin, and trade facilitation. 12 For more information on the mapping of the content of SPS and TBT provisions in PTAs, see Espitia et al. (2020) and Stone and Casalini (2020). 13 The templates used for the mapping of SPS and TBT provisions are provided in Appendixes B.1 and B.2. 8 those that require specific liberalization and integration commitments and obligations, and those related to procedures, transparency, enforcement or objectives, which are viewed as in- dispensable to achieving SPS and TBT integration across member countries (Mattoo et al., 2020).14 Examples of essential provisions aiming at further SPS and TBT integration in- clude the promotion of the use of international standards as in the Canada-Colombia PTA.15 Essential provisions focusing on procedures include, for instance, the right of the importing country to audit the exporting party’s competent authorities, inspection systems, or produc- tion procedures as in the EU-Colombia and EU-Peru PTAs.16 Globally, the number of PTAs including SPS and TBT provisions has increased, from 32 percent of the PTAs before 1990 to 85 percent between 2010 and 2017 (see Appendix A.1). Despite the increasing prevalence of SPS and TBT provisions in PTAs, Figure 1 shows the inclusion of essential provisions is less common. On the integration approach, Figure 2 shows that for SPS, harmonization provisions are more prevalent than mutual recognition provi- sions. In contrast, for TBT, very few PTAs include provisions on either harmonization or mutual recognition. For our focus countries, Chile, Colombia, and Peru, the increasing preva- lence of SPS and TBT provisions in PTAs is confirmed: as of 2017, more than 80 percent of the PTAs they signed include SPS and TBT chapters. Table 1 lists all the PTAs they signed during our sample period and provides information on the SPS and TBT provisions included. On the integration approach, harmonization of SPS is included in 22 of the 38 PTAs signed by the three countries, some examples include the Panama-Chile PTA of 2008 and the Chile-Colombia PTA of 2009. In contrast, mutual recognition of SPS is present in only five PTAs: the Andean Community, the Chile-Mexico PTA of 1999, the Chile-India 14 The identification of essential provisions is based on experts’ knowledge and, hence, is subjective. 15 Article 605 of the Canada-Colombia PTA states: Each Party shall use relevant international standards, guides and recommendations as a basis for its technical regulations and conformity assessment procedures in accordance with Articles 2.4 and 5.4 of the TBT Agreement. 16 Article 93 of the EU-Colombia and EU-Peru PTAs states that: In order to maintain confidence in the effective implementation of the provisions of this Chapter, each Party, within the scope of this Chapter, shall have the right to: (a) carry out, in accordance with the Guidelines set out under Appendix 3 of Annex VI (Sanitary and Phytosanitary Measures), verification of all or part of the control system of the authorities of another Party; the expenses of such verification shall be borne by the Party carrying out the verification. 9 PTA of 2007, the Colombia-Northern Triangle PTA of 2009, and the Peru-Chile PTA of 2009. With the exception of the EFTA-Peru PTA of 2011, none of the other PTAs includes harmonization or mutual recognition of TBT.17 2.2 Firm Customs Data Our analysis relies on transaction-level customs data for Chile, Colombia, and Peru over the 1997-2015, 1996-2015, and 2000-2015 periods, respectively. The data obtained from the coun- tries’ customs agencies was collected by the Trade and Integration Unit of the World Bank Research Department, as part of their efforts to expand the Exporter Dynamics Database de- scribed in Fernandes et al. (2016). Each country’s raw database covers the universe of exports and provides information at the exporting country-firm-destination-HS 6-digit product-year level on export values (in current USD) and weights (in kilograms but only for Colombia and Peru). Each country’s raw database was subjected to uniform reformatting and to a series of cleaning procedures detailed in Fernandes et al. (2016).18 Firms are identified by their actual names, their tax identification number, or an artificial unique code randomly created by the customs agency, which allows us to have a panel of firms for each country. Regarding product nomenclatures, we convert the HS 1996, 2002, 2007, and 2012 versions (used in the raw data) to the HS 1992 version (as this is convenient to merge with the tariff variable described in Section 2.3). While we have information for each firm on its exports in all sectors, we exclude from the analysis all observations pertaining to HS Chapters 25-27 to avoid potential biases related 17 In the content of preferential trade agreements database, if two countries sign more than one PTA with each other, the numbers of provisions in each policy chapter are cumulated across all PTAs. For example if country A and country B signed an agreement in 2000 with provisions 1 and 2 in the SPS chapter and signed another agreement in 2005 with provisions 1 and 3, between 2000 and 2004, 2 SPS provisions would be in force while from 2005 onward 3 SPS provisions would be in force between countries A and B. 18 Export values are Freight on Board (FOB) figures measured in USD converted from local currency to USD using exchange rates taken from the IMF’s International Financial Statistics. The quality of the customs data is assessed as in Fernandes et al. (2016) by a comparison of total exports obtained from aggregating the transaction-level customs data at the country level with the total exports obtained at the country level from COMTRADE/WITS (World Integrated Trade Solution). The ratios for Chile, Colombia and Peru average close to 100% over the sample period. 10 to commodity cycle price booms that were highly prevalent in Latin America during our sample period. The impact of SPS provisions is measured on firm exports of agricultural and agro-food HS 6-digit products (i.e., those belonging to HS Chapters 01-24) while the impact of TBT provisions is measured on firm exports of any HS 6-digit products (except HS 25-27). Our key outcome variable is firm export value at the destination-HS 6-digit product-year level. For the PPML estimation of our firm gravity equation described in Section 3, we need to expand our initial exporting country-firm-destination-HS 6-digit-year database to account for zeros. Specifically, we fill in the initial database so that each firm-destination-HS 6-digit has an observation (a row) in all of that exporting country’s sample years, with a 0 export value in a year when exports by the firm-destination-HS 6-digit do not take place.19 This expanded database combined with the type of fixed effects included in our firm gravity equa- tion allows us to exploit changes in firms’ exports over time as PTAs enter into force over the sample period to identify our coefficients of interest.20 Table 2 shows summary statistics for each exporting country on the number of exporting firms, their export values, destination and product diversification patterns. The statistics are averages across the sample period and are provided for all exporting firms (Panel A) and for firms exporting agricultural and agro-food products (Panel B). The numbers of exporting firms are quite similar across the three countries. The median number of destination markets and products per firm is low, 1 or 2 at most. Firms selling agricultural and agro-food prod- ucts have higher average and median export values and are more diversified than the average for all exporting firms. These patterns are not surprising given our countries’ comparative advantage in agricultural and agro-food products. 19 A similar fill-in procedure is used in the study of SPS standards and export decisions by Fernandes et al. (2016). 20 Appendix Table C.2 shows an example of our data fill-in procedure. We did not pursue the approach of filling in the initial database so as to make it a ‘square’ matrix where every firm in an exporting country would have an observation for every destination-HS 6-digit-year combination as is done in traditional gravity equation estimation based on more aggregate data since such database would be highly cluttered by zeros (as most firms export few HS 6-digit products to few destinations), would embed unfeasible economic choices, and would be computationally impossible to handle. Appendix Table C.3 shows the total numbers of observations in the initial database and the expanded database for each country. 11 2.3 Additional data All our firm gravity equations include a measure of bilateral tariffs that controls for tariff liberalization taking place between origin countries and destination markets over the period, whether unilaterally or in the context of PTAs. We rely on a global tariff database covering tariffs at the HS 6-digit product level for 197 importing countries and their trading partners from 1988 to 2017 from Teti (2020). The database addresses two major issues with tariff data from TRAINS, missing data and misreporting. Additionally, some of our specifications control for a measure of GDP in the destination market, taken from the World Development Indicators. Summary statistics on these variables are provided in Appendix Table C.4. 3 Empirical framework To examine the effects of SPS and TBT provisions in PTAs on firm exports, we consider the following two specifications: yf jkt = exp(αf jk + αf kt + β1 × ln(τijkt ) + β2 × ln(GDPjt ) + β3 × Depthijt + β4 × P rovijt ) + f jkt (1) and yf jkt = exp(αf jk + αf kt + γ1 × ln(τijkt ) + γ2 × ln(GDPjt ) 3 + γ3 × Depthijt + γs × P rovijt × T erciles f jk + f jkt (2) s=1 where y are exports, f is a firm (in origin country i), j is a destination market, k is an HS 6-digit product, t is a year, and epsilon is an independent and identically distributed error term. Our key variable of interest in Equation (1) is P rov which is either (i) the number of 12 SPS or TBT provisions in a PTA between origin country i and destination j (normalized as described in Appendix C.1), for total provisions, essential provisions, or transparency provi- sions or (ii) a dummy variable for at least one harmonization provision for SPS or TBT (on technical regulations, standards, or procedural regulations) being included in the PTA or (iii) a dummy variable for at least one mutual recognition (MR) provision for SPS or TBT (on technical regulations, standards, or procedural regulations) being included in the PTA. The use of normalized measures enables us to account for PTAs with no SPS and TBT provisions and makes the measures easier to interpret, the measure is equal to 1 for the PTA in the sample with the highest number of provisions. Our main estimates to test for heterogeneity in the impacts of SPS and TBT provisions in PTAs across firms are based on Equation (2) where P rov enters interacted with each firm size category T erciles . As a baseline, we consider three mutually exclusive size categories defined by the tercile of the market share that each firm has in a destination-HS 6-digit market in its initial year in the sample. This definition of firm size is guided by how specific the SPS and TBT measures are, assuming that most are applicable to very narrowly defined HS 6-digit products and thus that their elimination or harmonization through the inclusion of SPS and TBT provisions in a PTA could benefit differentially firms that are small from the point of view of that product’s market. In robustness checks (Appendix D.1), we consider alternative definitions of firm size. Equations (1) and (2) include firm-destination-HS 6-digit fixed effects that account for unob- served heterogeneity at a highly disaggregated level and allow us to identify our coefficients of interest β4 and γs based on within firm-destination-HS 6-digit changes in exports as PTAs with SPS and TBT provisions enter into force over the sample period.21 Importantly, Equa- tions (1) and (2) also control for very stringent firm-HS 6-digit product-year fixed effects that capture both productivity differences as well as product-specific supply or demand shocks experienced by firms. Hence, our firm gravity equation can be viewed as a generalized 21 The firm-destination-HS 6-digit product fixed effects also control for more aggregate unobserved country- sector factors such as domestic competition that might bias the effect of the PTAs. 13 difference-in-differences specification that estimates the impact of SPS and TBT provisions in PTAs on exports of a firm-HS 6-digit product before and after the entry in force of a PTA in a given destination (panel dimension) relative to exports of a firm-HS 6-digit product in destinations with no PTAs (cross-section dimension) While our empirical strategy captures the effect of the content of PTAs in terms of provisions included, we are not able to study the impact of their effective applicability, which is not observable. However, this concern is mitigated for two reasons. First, our fixed effects struc- ture controls for any systematic under-enforcement bias in PTAs signed by Chile, Colombia, and Peru. Second, the previous version of the Deep Trade Agreements database described in Hofmann et al. (2017) reveals that the SPS and TBT chapters included in 75 and 80 percent, respectively, of the PTAs signed by Chile, Colombia and Peru are considered as legally enforceable .22 The inclusion in our specifications of τ , the tariff rate imposed by destination j on exports of product k by firms from a given origin, and of the Depth variable capturing the normalized number of provisions in all 16 policy areas (other than SPS and TBT) discussed in Section 2.1 allows us to isolate the impact of SPS or TBT provisions from tariff concessions and other forms of integration provided by the PTA. Equations (1) and (2) are estimated by PPML introduced by Silva and Tenreyro (2006) with the recent command of Correia et al. (2020). Inference is based on Huber-White standard errors robust to heteroskedasticity clustered by origin-destination. Equations (1) and (2) include destination GDP to control for time-varying destination-specific demand. A variant of our equations includes destination-year fixed effects which are a more flexible way to account for demand and macroeconomic shocks in destination markets (Ap- pendix D.2).23 Due to computational limitations, we are only able to estimate this vari- ant using a more aggregated version of our data at the exporting country-firm-destination- 22 Legally enforceable means that (i) the language used is sufficiently precise and committing and (ii) the chapter has not been excluded from the PTA’s dispute settlement procedures. 23 These fixed effects account in particular for the Global Financial Crisis that is part of our sample period. 14 industry-year level, considering the 15 industries defined in Appendix C.5. In this variant we are still able to include a very stringent account for firm shocks through firm-industry-year fixed effects. One concern about the estimation of Equations (1) and (2) is that the design and content of PTAs is highly influenced by large firms in the PTA signatory countries as argued by Rodrik (2018) and shown for the US by Blanga-Gubbay et al. (2020). Identifying credible instru- ments for our variables of interest - the SPS and TBT provisions in PTAs - would be highly challenging, but we tackle potential endogeneity in three ways. First, we exclude from the sample the largest exporting firms in Chile, Colombia, and Peru, which may have influenced the design of the PTAs. Second, we exclude from the sample the sectors with the highest degree of concentration of exports across firms, which may be better organized in terms of lobbying. Third, we consider differences in the impacts of SPS and TBT provisions separately for PTAs signed by Chile, Colombia, and Peru with other developing countries (South-South) versus those with high-income countries (North-South). Firms in Chile, Colombia, and Peru may have much less bargaining power in PTA negotiations of the North-South type than of the South-South type. 4 Main Results 4.1 Baseline Results Table 3 presents the results from estimating Equation (1) for SPS provisions in columns (1)-(2) and for TBT provisions in columns (5)-(6) and from estimating Equation (2) for SPS provisions in columns (3)-(4) and for TBT provisions in columns (7)-(8). The positive but insignificant coefficient on the P rov variable in columns (1) and (5) suggests an ambiguous effect on firm exports to destinations with which the origin country has a PTA including SPS or TBT provisions as an average. This ambiguity may be due to underlying heterogeneity across both (i) the specific types of SPS and TBT provisions included in the PTA as well as 15 (ii) firm size. Acknowledging first that not all SPS and TBT provisions in PTAs are equal in their potential to achieve integration and thus boost firm exports, we focus on the SPS and TBT provisions considered as essential. Column (2) shows a significant increase in firm exports of agricul- tural and agro-food products to destinations with which their country signed a PTA with a larger number of essential SPS provisions. Specifically, the estimates in column (2) imply that upon entry into force of a PTA with the maximum number of essential SPS provisions (like the Andean Community), that is, when the normalized measure of SPS provisions moves from 0 to 1, firm exports of agricultural and agro-food products increase on average by 23 percent, after accounting for the depth of other provisions and potential tariff liberalization also brought by such PTA.24 Essential TBT provisions have a positive but insignificant effect on average firm exports in column (6). The average effects of SPS and TBT provisions across firms mask important differences in the magnitude of the effects across size categories. Columns (3) and (4) provide strong evidence of a substantially larger benefit for the exports of smaller firms than of larger firms from the entry into force of PTAs with more SPS provisions. All unreported t-tests comparing the coefficients across smaller and larger (and smaller and medium) size categories for SPS provisions indicate that they are significantly different (at the 95% confidence level). The es- timates in column (3) imply that upon entry into force of a PTA with the maximum number of SPS provisions (like the CPTPP), the exports of agricultural and agro-food products of smaller firms grow by 45 percent, whereas those of larger firms do not increase significantly, with all else constant. The magnitude is even higher, with an increase in firm exports by 93 percent when we consider essential SPS provisions in column (4). The exports of smaller firms also increase more strongly than those of larger firms from the entry into force of PTAs with more TBT provisions, as per columns (7) and (8). But the unreported t-tests show that while the coefficients for smaller firms are significantly different from those for medium firms, 24 The percent increase in firm exports is computed as exp(0.203)-1, as described in Appendix C.1. 16 they are not significantly different from those for larger firms. Table 4 presents the estimates of Equation (2) for SPS and TBT provisions on mutual recog- nition (MR) (column (1)), harmonization (columns (2) and (4)), and transparency (columns (3) and (5)).25 For SPS, the type of integration through MR versus harmonization provi- sions is an important factor behind the increase in smaller firms’ exports of agricultural and agro-food products.26 Column (2) shows a boost in exports of agricultural and agro-food products for smaller firms selling to destination markets with which their country harmo- nizes SPS measures of 44 percent, compared to close to 20 percent for larger firms selling in those markets.27 Chile, Colombia, and Peru have no TBT MR provisions in their PTAs and only one of their PTAs includes TBT harmonization provisions. Hence, the results in column (4) of Table 4 should be taken with caution as the identification reflects only the EFTA-Peru PTA of 2011.28 Transparency provisions in the areas of SPS and TBT seem to also be important drivers behind the relatively larger increase of smaller firms’ exports compared to larger firms’ ex- ports, as seen in columns (3) and (5) of Table 4. Specifically, the estimates in column (3) imply that, upon entry into force of a PTA with the maximum number of provisions for the transparency of SPS regulations (like the CPTPP or the Pacific Alliance), the agricultural and agro-food exports of smaller firms grow by 85 percent whereas those of larger firms do not increase significantly.29 This finding is consistent with the idea that transparency provi- sions can reduce the level of uncertainty related to SPS and TBT measures, which can affect disproportionally small firms (Ghosal and Loungani, 2000; Ghosal and Ye, 2015) particularly in developing countries (Bloom, 2014). 25 Since none of the PTAs signed by Chile, Colombia, or Peru during our sample period include MR of TBT, the specification with P rov being MR of TBT cannot be estimated. 26 However, the effect of SPS harmonization provisions is very difficult to disentangle from the effect of SPS MR provisions, as all the PTAs including some MR provisions also include some harmonization provisions. 27 The t-test indicates that the coefficients across smaller and larger size categories are significantly different (at a 95% confidence level). 28 Moreover, the t-test shows the coefficients across smaller and larger size categories are not significantly different (at a 95% confidence level). 29 The t-test indicates that the coefficients across smaller and larger size categories are significantly different (at a 95% confidence level). 17 An unexpected finding in the specifications for firm agricultural and agro-food exports in Tables 3 and 4 is the negative and significant impact of PTA depth in policy areas other than SPS and TBT. To explore this finding, we estimate a variant of Equation (1) where the P rov measure is dropped and a new Depth measure capturing depth of the PTA in all 18 policy areas (including SPS and TBT) is included. The estimated impact of PTA depth on firm agricultural and agro-food exports is now positive though insignificant, so the concern of an unexpected negative impact is mitigated.30 4.2 Robustness Results Our baseline results point to significant size heterogeneity in the impacts of SPS and TBT provisions in PTAs across firms. We conduct two robustness checks of the definition of firm size. First, we consider new dummy variables that split the sample into three mutually exclusive size categories defined by the market share that each firm has in a destination- industry market in its initial year in the sample. This definition of size informs on firm exports across different industries, rather than very narrowly defined HS 6-digit products. Second, we consider new dummy variables that split the sample into three mutually exclusive size categories defined by the market share that each firm has in total exports in a destination in its initial year in the sample. This definition of size informs on firm overall exports to a market, rather than more narrowly defined industries or products. The results from these robustness checks are provided in Appendix D.1. The estimates show that the integration of more SPS and TBT provisions in PTAs, especially essential SPS and TBT provisions, benefits significantly more the exports of smaller firms than larger firms.31 We also conduct a robustness check of the way destination demand effects are accounted for, replacing destination GDP by destination-year fixed effects in a specification using data at the exporting country-firm-destination-industry-year level, as described in Section 3. Given 30 These results are available from the authors upon request. 31 In Appendix Table 11 there are stronger positive effects on exports of firms in the medium size category but the greater benefit for smaller relative to larger firms holds. 18 the higher level of aggregation, Equation (2) can be estimated by maintaining very stringent firm-destination-industry (panel dimension here) and firm-industry-year fixed effects. The results are shown in Appendix D.2 and the findings are qualitatively similar to those in our baseline Tables 3 and 4. 4.3 Addressing Endogeneity Our results so far suggest a more beneficial impact of PTAs including SPS and TBT provisions on the exports of smaller firms than those of larger firms in Chile, Colombia, and Peru. This pattern is the opposite of what would be expected if large firms played a strong role in the design and content of PTAs signed by their countries. Such influence would presumably be used to design SPS and TBT provisions that might benefit them most. If the current estimated impacts across firm size categories suffered from an endogeneity bias due to the role of large firms, the differential benefit for smaller firms from such PTAs could be in reality smaller. Hence, we tackle these potential endogeneity concerns in the three ways. First, we exclude from the sample the largest exporting firms in Chile, Colombia, and Peru, which may have influenced the design of the PTAs. The largest exporting firms are identified for each country and year as those in the top 1 percent of the distribution of firm total exports.32 The estimates of Equations (1) and (2) based on the smaller sample are shown in Appendix Table 14. The resulting estimates are qualitatively similar to our baseline estimates but (not surprisingly) with substantially weaker impacts on larger firms. Second, we exclude from the sample the sectors with the highest degree of concentration of exports across firms. They are defined for each country as those for which the largest 1 percent of firms are responsible for more than 50 percent of sector exports (where sectors are HS 2-digit chapters) in at least one sample year.33 The estimates of Equations (1) and (2) 32 All firms that appear at least once in the top 1 percent of that distribution during the period are dropped from the sample. 33 The HS 2-digit chapters dropped are listed in Appendix C.6 for Chile, Colombia, and Peru. Note that since such highly concentrated exports exhibit very high total export values, their exclusion dramatically 19 based on the smaller sample are shown in Appendix Table 15. The results are qualitatively similar to our baseline results. Third, we consider differences in the impacts of SPS and TBT provisions separately for PTAs signed by Chile, Colombia, and Peru with other developing countries (South-South) versus those with high-income countries (North-South) and we discuss those results in Section 5.4. Firms in Chile, Colombia, and Peru may have much less bargaining power in PTA negotiations of the North-South type than of the South-South type. All in all, we argue that our findings of a stronger benefit for exports of small firms from deep PTAs with provisions on SPS and TBT are not the result of pressure by influential domestic firms or sectors to design the content of the PTA to their advantage. 5 Extensions 5.1 Impacts across Sectors and Products with Stronger SPS or TBT Requirements The impact of SPS and TBT provisions in PTAs may differ across sectors due to hetero- geneous fixed entry costs or differences in technology. We examine the impact of SPS and TBT provisions in PTAs across sectors by re-estimating Equation (2) separately for each sector. The results in Table 5 show that including SPS provisions in PTAs increases smaller firms’ exports relative to those of larger firms in a similar way across animal, vegetables, and foodstuffs sectors. However, TBT provisions in PTAs induce a more differentiated response of smaller firms’ exports across sectors. Smaller firms in the chemicals, wood, metals and machinery sectors experience a stronger export performance following the entry into force of PTAs with TBT provisions. The potentially heavy TBT requirements in some of those sectors may act as a fixed cost with more harmful consequences for smaller firms. The in- tegration of those TBT regulations with partner countries through provisions in PTAs may reduces the sample coverage of total exports. 20 thus reduce the magnitude of those fixed costs and stimulate the exports of smaller firms relatively more than in other sectors. Firm exports of products with heavier regulatory requirements in terms of SPS and TBT are likely to respond differently than those of products with lighter regulatory requirements. To address this possibility, we consider the specific trade concerns raised by WTO members in the WTO Committees on SPS and on TBT measures, which are viewed as capturing de facto trade barriers (Fontagné et al., 2015).34 We aggregate our firm-destination-HS 6-digit-year level data to the firm-destination-industry-year level in order to estimate Equation (2) sepa- rately for industries subject to at least one specific trade concern versus industries subject to no such concerns. Both panels in Table 6 shows that the effects of SPS and TBT provisions on smaller firms’ exports are entirely driven by industries subject to at least one specific trade concern. This constitutes useful evidence that the mechanism that our results identify is an integration and liberalization of SPS and TBT regulations across partner countries via the inclusion of related provisions in PTAs. 5.2 Impacts on Firm Export Market Entry Since SPS and TBT requirements may act as a fixed entry cost, regulating SPS and TBT by their inclusion in PTAs would decrease such cost and a boost to firm export market entry could be expected, especially the entry of smaller firms. Due to the fill-in of our database to account for zeros in trade, our baseline results capture a combination of impacts at the intensive and extensive margins. But we explicitly examine firm export market entry defining it as a dummy variable equal to one if a firm-destination-HS 6-digit triplet export a positive trade flow in year t, while exporting a zero trade flow in t -1. We estimate Equation (2) using firm entry as the dependent variable and provide the results in Table 7. The estimates show that including SPS provisions in PTAs significantly increases the entry of smaller firms into new product-markets. PTAs with mutual recognition or harmonization provisions for 34 We consider all specific trade concerns raised between 1995 and 2010 by all WTO member countries separately in the SPS or TBT committee. 21 SPS also foster firm export entry.35 For TBT, the inclusion of provisions in PTAs has an insignificant impact on firm export entry for smaller firms. 5.3 Impacts on Firm Export Quality By setting requirements on SPS and TBT in the destination market closer to those in the domestic market, the inclusion of SPS and TBT provisions in a PTA may allow firms to improve the quality of the products they export. To examine this possibility, we estimate Equation (2) using a measure of firm export quality at the HS-6 digit product-destination level as the dependent variable for Colombia and Peru.36 The measure is obtained following the methodology proposed by Amiti and Khandelwal (2013) and used at the firm-level by Bajgar and Javorcik (2020) as we describe in Appendix G. The results in Table 8 provide only weak evidence that the estimated increase in smaller firms’ exports is driven by an improvement in quality (rather it may be driven by increased volume). In particular, the inclusion of essential or transparency SPS provisions in PTAs leads to an improvement in the quality of smaller firms’ exports of agricultural and agro-food products. The inclusion of transparency TBT provisions also leads to increased export quality for small firms. As an alternative proxy for export quality, we consider export unit values, defined as the ratio of export value to export weight. The results in Appendix Table 18 are quite similar to those based on the estimated firm export quality measures, providing some evidence that the inclusion of essential SPS provisions in PTAs leads to an increase in unit values. 5.4 PTAs with Different Types of Trading Partners Table 1 shows that Chile, Colombia, and Peru have engaged in an integration strategy that involves the signature of PTAs with a very diverse set of trading partners, some regional and 35 All unreported t-tests comparing the coefficients across small and large size categories indicate that they are significantly different (at a 95% confidence level). 36 We are unable to compute firm export quality measures for Chile since its data does not include quantity information. 22 others more distant and at very different stages of development. Our baseline results in Table 3 and especially in Table 4 on the role of harmonization of SPS may depend on the type of trading partner. In particular, the strong role of the inclusion of provisions on harmonization of SPS could be driven by one or a few asymmetric PTAs in which high-income countries, for example the US or European countries dictate the regulations with which partner countries need to comply. To address this possibility, we estimate Equation (2) separately across the sample of high-income destination countries and the sample of medium-income and low- income destination countries. The results are shown in Tables 9 and 10. Surprisingly, we find that the patterns are very similar across destination countries at different stages of development. In particular, we find that the effect of an increase in essential SPS provisions benefits significantly more the exports of smaller firms whether the PTA is signed with a high-income country or not, but an increase in essential TBT provisions has a significant effect on exports of smaller firms only if the PTA is signed with a medium-income or low- income partner. The importance of SPS harmonization in fostering smaller firms’ exports is present regardless of the destination market. The role of transparency of SPS is also found to be positive for small firms’ exports to all destination markets. However transparency of TBT provisions is beneficial for smaller firms’ exports only to medium-income or low-income partners. One possible interpretation is that in those countries SPS and TBT regulations are more opaque whereas in high-income countries information may be easily available to any potential exporter, thus the inclusion of such transparency provisions in a PTA plays no role. We also separate the impacts of PTAs across those signed with the EU or US versus those signed with the rest of the world. The impacts of SPS and TBT provisions in PTAs on firms’ exports across firm sizes, provided in Appendix Tables 16 and 17 are again quite similar across the two groups of destination markets. 23 6 Conclusion This paper studies how rules on SPS and TBT in PTAs signed by Chile, Colombia, and Peru affect their firms’ export performance along the firm size spectrum. We combine novel data on the content of SPS and TBT provisions in trade agreements with customs data covering the universe of exporting firms in each country over the 1996-2015 period. After accounting for the overall depth of the PTA and product-specific bilateral tariffs, our firm gravity equations show that firms’ exports increase significantly to destination markets with PTAs including a larger number of SPS and TBT provisions and the effect is stronger for smaller firms. We further provide evidence that provisions for the harmonization of SPS regulations in PTAs benefit more the exports of smaller firms and so do PTAs that include stronger transparency provisions of SPS and TBT regulations. We find that the rising exports of smaller firms ensuing from PTAs with SPS and TBT provisions are partly explained by those smaller firms’ higher entry into new product-markets. This finding shows that PTAs, and especially the SPS and TBT provisions they include, are a useful tool for making the benefits of trade liberalization reachable by small firms. References Amiti, M. and Khandelwal, A. K. (2013). Import Competition and Quality Upgrading. The Review of Economics and Statistics, 95(2):476–490. Anderson, J. E. and Yotov, Y. V. (2020). Short run gravity. Journal of International Economics, 126:103341. Baier, S. L. and Bergstrand, J. H. (2007). Do free trade agreements actually increase mem- bers' international trade? Journal of International Economics, 71(1):72–95. Baier, S. L., Bergstrand, J. H., and Feng, M. (2014). Economic integration agreements and the margins of international trade. Journal of International Economics, 93(2):339–350. 24 Baier, S. L., Yotov, Y. V., and Zylkin, T. (2019). On the widely differing effects of free trade agreements: Lessons from twenty years of trade integration. Journal of International Economics, 116:206–226. Bajgar, M. and Javorcik, B. (2020). Climbing the Rungs of the Quality Ladder: FDI and Domestic Exporters in Romania. Economic Journal, 130(628):937–955. Baldwin, R. E. (2000). Regulatory protectionism, developing nations, and a two-tier world trade system. Brookings Trade Forum, 2000(1):237–280. Bernard, A. B., Jensen, J. B., Redding, S. J., and Schott, P. K. (2007). Firms in international trade. Journal of Economic Perspectives, 21(3):105–130. Bernard, A. B., Jensen, J. B., Redding, S. J., and Schott, P. K. (2012). The empirics of firm heterogeneity and international trade. Annual Review of Economics, 4(1):283–313. Bernard, A. B., Redding, S. J., and Schott, P. K. (2011). Multiproduct Firms and Trade Liberalization *. The Quarterly Journal of Economics, 126(3):1271–1318. Blanga-Gubbay, M., Conconi, P., and Parenti, M. (2020). Lobbying for Globalization. CEPR Discussion Papers 14597, C.E.P.R. Discussion Papers. Broda, C. and Weinstein, D. E. (2006). Globalization and the gains from variety. The Quarterly Journal of Economics, 121(2):541–585. Chaney, T. (2008). Distorted gravity: The intensive and extensive margins of international trade. American Economic Review, 98(4):1707–21. Chen, M. X. and Mattoo, A. (2008). Regionalism in standards: good or bad for trade? Canadian Journal of Economics/Revue canadienne d'économique, 41(3):838–863. Correia, S., Guimarães, P., and Zylkin, T. (2020). Fast poisson estimation with high- dimensional fixed effects. The Stata Journal: Promoting communications on statistics and Stata, 20(1):95–115. 25 Costinot, A. (2008). A comparative institutional analysis of agreements on product standards. Journal of International Economics, 75(1):197–213. Crozet, M. and Koenig, P. (2010). Structural gravity equations with intensive and extensive margins. Canadian Journal of Economics/Revue canadienne d'économique, 43(1):41–62. Disdier, A.-C., Fontagné, L., and Cadot, O. (2014). North-south standards harmonization and international trade. The World Bank Economic Review, 29(2):327–352. Disdier, A.-C., Stone, S., and van Tongeren, F. (2020). Trade and Economic Effects of IRC: Further Empirical Evidence from SPS and TBT provisions. PSE Working Papers hal-02953673, HAL. Espitia, A., Pardo, S., Piermartini, R., and Rocha, N. (2020). Technical barriers to trade. In Handbook of Deep Trade Agreements, pages 343–366. The World Bank. Fernandes, A. M., Ferro, E., and Wilson, J. S. (2019). Product standards and firms’ export decisions. The World Bank Economic Review, 33(2):353–374. Fernandes, A. M., Freund, C., and Pierola, M. D. (2016). Exporter behavior, country size and stage of development: Evidence from the exporter dynamics database. Journal of Development Economics, 119:121–137. Fontagné, L. and Orefice, G. (2018). Let’s try next door: Technical barriers to trade and multi-destination firms. European Economic Review, 101:643–663. Fontagné, L., Orefice, G., Piermartini, R., and Rocha, N. (2015). Product standards and margins of trade: Firm-level evidence. Journal of International Economics, 97(1):29–44. Freund, C. and Pierola, M. D. (2015). Export superstars. Review of Economics and Statistics, 97(5):1023–1032. 26 Larch, M., Wanner, J., Yotov, Y. V., and Zylkin, T. (2018). Currency unions and trade: A PPML re-assessment with high-dimensional fixed effects. Oxford Bulletin of Economics and Statistics, 81(3):487–510. Macedoni, L. and Weinberger, A. (2019). Quality heterogeneity and misallocation: The welfare benefits of raising your standards. SSRN Electronic Journal. Martin, J. (2012). Markups, quality, and transport costs. European Economic Review, 56(4):777–791. Mattoo, A., Mulabdic, A., Ruta, M., Mattoo, A., Mulabdic, A., and Ruta, M. (2017). Trade creation and trade diversion in deep agreements. Policy Research Working Paper Series 8206, The World Bank. Mattoo, A., Rocha, N., and Ruta, M. (2020). Handbook of Deep Trade Agreements. Wash- ington, DC: World Bank. Melitz, M. J. (2003). The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica, 71(6):1695–1725. Melitz, M. J. and Redding, S. J. (2015). New trade models, new welfare implications. Amer- ican Economic Review, 105(3):1105–46. Portugal-Perez, A., Reyes, J.-D., and Wilson, J. S. (2010). Beyond the information technology agreement: Harmonisation of standards and trade in electronics. The World Economy, 33(12):1870–1897. Rodrik, D. (2018). What do trade agreements really do? Journal of Economic Perspectives, 32(2):73–90. Schmidt, J. and Steingress, W. (2019). No double standards: Quantifying the impact of standard harmonization on trade. SSRN Electronic Journal. 27 Silva, J. M. C. S. and Tenreyro, S. (2006). The log of gravity. Review of Economics and Statistics, 88(4):641–658. Stone, S. and Casalini, F. (2020). Sanitary and phytosanitary measures. In Handbook of Deep Trade Agreements, pages 367–390. The World Bank. Teti, F. (2020). 30 Years of Trade Policy: Evidence from 5.7 Billion Tariffs. ifo Working Paper Series 334, ifo Institute - Leibniz Institute for Economic Research at the University of Munich. UNCTAD (2012). Non-Tariff Measures To Trade: Economic And Policy Issues For Devel- oping Countries Developing Countries In International Trade Studies.World Trade Orga- nization. Technical report, United Nations Conference On Trade And Development. WTO (2012). World trade report 2012: Trade and public policies: A closer look at non-tariff measures in the 21st century. Technical report, World Trade Organization. 28 Figure 1: Essential SPS and TBT Provisions in PTAs Essential SPS provisions Essential TBT provisions 300 300 20 20 Number of Agreements Number of Agreements Cumulative Cumulative 150 150 10 10 0 0 0 0 0 0 0 0 0 0 0 60 70 80 90 00 10 20 6 7 8 9 0 1 2 19 19 19 19 20 20 20 19 19 19 19 20 20 20 Total number of new PTAs Total number of new PTAs New PTAs with less than 25% of essential SPS prov. New PTAs with less than 25% of essential TBT prov. New PTAs with more than 25% of essential SPS prov. New PTAs with more than 25% of essential TBT prov. Note: the graph plots the number of new PTAs entering in force over the period 1958-2017. It disantangles PTAs including zero, less than 25% and more than 25% of the total number of essential SPS provisions (left) and PTAs including zero, less than 25% and more than 25% of the total number of essential TBT provisions (right). Deepest agreements include 54% of essential SPS prov. and 47% of essential TBT prov.The red line shows the cumulated number of PTAs. Source: Mattoo, Rocha, Ruta (2020). 29 Figure 2: SPS and TBT Mutual Recognition and Harmonization Provisions in PTAs MRA provision of SPS Harmonization provision of SPS 300 300 20 20 150 150 10 10 0 0 0 0 60 70 80 90 00 10 20 60 70 80 90 00 10 20 19 19 19 19 20 20 20 19 19 19 19 20 20 20 MRA provision of TBT Harmonization provision of TBT 300 300 20 20 150 150 10 10 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 6 7 8 9 0 1 2 6 7 8 9 0 1 2 19 19 19 19 20 20 20 19 19 19 19 20 20 20 Number of new agreements (left axis) Number of new agreements with specific provisions (see note) Cumulated number of agreements (right axis) Note: the graph plots the number of new RTAs entering in force over the period 1958-2017. It disantangles RTAs including at least one provision on mutual recognition of SPS (upper-left), RTAs including at least one provision on harmonization of SPS (upper-right), RTAs including at least one provision on mutual recognition of TBT (bottom-left) and RTAs including at least one provision on harmonization of TBT (bottom-right). The red line shows the cumulated number of RTAs. Source: Mattoo, Rocha, Ruta (2020). 30 Table 1: Selected PTAs signed by Chile, Colombia, or Peru Agreements Entry SPS TBT in force Total Essential MR Harm. Transp. Total Essential MR Harm. Transp. Canada - Chile 1997 1 0 NO NO 1 7 1 NO NO 1 Chile - Mexico 1999 21 9 YES YES 6 17 6 NO NO 2 Chile - Costa Rica (Chile - Central America) 2002 0 0 NO NO 0 12 2 NO NO 2 Chile - El Salvador (Chile - Central America) 2002 0 0 NO NO 0 12 2 NO NO 2 EU - Chile 2003 26 10 NO YES 4 8 2 NO NO 0 EFTA - Chile 2004 8 1 NO NO 3 5 0 NO NO 0 Korea, Republic of - Chile 2004 22 8 NO YES 5 8 2 NO NO 1 United States - Chile 2004 11 1 NO NO 1 11 2 NO NO 2 Trans-Pacific Strategic Economic Partnership 2006 15 4 NO NO 2 15 3 NO NO 2 Chile - China 2006 20 7 NO YES 4 12 2 NO NO 2 Chile - Japan 2007 10 2 NO NO 2 6 0 NO NO 0 Chile - India 2007 12 4 YES YES 3 6 0 NO NO 0 Panama - Chile 2008 19 6 NO YES 4 10 2 NO NO 2 Chile - Honduras (Chile - Central America) 2008 0 0 NO NO 0 12 2 NO NO 2 Colombia - Northern Triangle (El Salvador, Guatemala, Honduras) 2009 0 0 NO NO 0 14 3 NO NO 2 Canada - Peru 2009 15 5 NO YES 3 12 2 NO NO 2 Chile - Colombia 2009 18 7 NO YES 1 16 6 NO NO 2 United States - Peru 2009 12 2 NO NO 1 12 2 NO NO 2 Peru - Chile 2009 20 9 YES YES 2 18 6 NO NO 2 Peru - Singapore 2009 21 7 NO YES 5 13 2 NO NO 2 Australia - Chile 2009 13 2 NO YES 3 13 2 NO NO 2 Peru - China 2010 22 6 NO NO 6 11 1 NO NO 2 Chile - Guatemala (Chile - Central America) 2010 0 0 NO NO 0 12 2 NO NO 2 Turkey - Chile 2011 7 1 NO NO 1 10 0 NO NO 2 Canada - Colombia 2011 16 5 NO YES 2 14 2 NO NO 2 Peru - Korea, Republic of 2011 14 4 NO NO 3 12 2 NO NO 3 EFTA - Colombia 2011 12 2 NO NO 2 7 0 NO NO 1 EFTA - Peru 2011 9 2 NO NO 1 10 2 NO YES 2 Chile - Nicaragua (Chile - Central America) 2012 0 0 NO NO 0 12 2 NO NO 2 Japan - Peru 2012 10 1 NO NO 2 10 2 NO NO 2 Peru - Mexico 2012 17 6 NO NO 4 16 6 NO NO 2 United States - Colombia 2012 16 4 NO NO 3 12 2 NO NO 2 Chile - Malaysia 2012 15 2 NO YES 1 10 2 NO NO 2 Panama - Peru 2012 23 7 NO YES 4 11 2 NO NO 2 EU - Colombia and Peru 2013 19 5 NO NO 3 10 0 NO NO 3 Costa Rica - Peru 2013 22 7 NO YES 5 13 2 NO NO 2 Hong Kong, China - Chile 2014 14 2 NO NO 3 11 2 NO NO 2 Chile - Vietnam 2014 14 2 NO YES 2 10 2 NO NO 2 Source. Authors’ calculation using Mattoo et al. (2020) Notes. The table lists all PTAs signed by Chile, Colombia and Peru during our sample time window. Agreements containing mutual recognition or harmonization provisions are in bold. 31 Table 2: Summary statistics on exporting firms Panel A - All exporting firms Country Total Avg # Avg Avg exports Median # Median Median exports Max Max Max # firms HS6 products # markets per firm HS6 products # markets per firm # HS6 # markets exports per firm per firm (in ´ 000 USD) per firm per firm (in ´ 000 USD) products (in ´ 000 USD) Chile 37180 7.5 3.5 17384.2 2.0 1.0 30.1 1874.0 135.0 1.3e+08 Colombia 44336 8.3 3.3 6576.3 2.0 1.0 36.3 773.0 127.0 2.0e+07 Peru 37094 11.4 3.0 8564.1 3.0 1.0 42.7 1636.0 123.0 2.8e+07 Panel B - Firms exporting agricultural products Country Total Avg # Avg Avg exports Median # Median Median exports Max Max Max # firms HS6 products # markets per firm HS6 products # markets per firm # HS6 # markets exports per firm per firm (in ´ 000 USD) per firm per firm (in ´ 000 USD) products (in ´ 000 USD) Chile 9407.0 4.7 6.0 18722.5 2.0 2.0 250.1 294.0 135.0 3.8e+06 Colombia 9219.0 3.4 3.7 11787.8 1.0 1.0 46.8 158.0 126.0 2.0e+07 Peru 10048.0 4.9 4.0 9232.1 2.0 1.0 85.3 324.0 122.0 5.5e+06 32 Table 3: Baseline Results - Average and size heterogeneity in effects of SPS and TBT provisions Export value of agricultural products Export value of any product All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT ln(1 + τ ) -1.622∗∗∗ -1.616∗∗∗ -1.630∗∗∗ -1.620∗∗∗ -1.814∗∗∗ -1.823∗∗∗ -1.818∗∗∗ -1.834∗∗∗ (0.412) (0.412) (0.411) (0.412) (0.420) (0.420) (0.422) (0.422) ln(GDP ) 1.332∗∗∗ 1.346∗∗∗ 1.315∗∗∗ 1.329∗∗∗ 2.170∗∗∗ 2.164∗∗∗ 2.163∗∗∗ 2.159∗∗∗ (0.191) (0.192) (0.189) (0.190) (0.177) (0.177) (0.177) (0.175) Depth -0.363∗∗∗ -0.262∗ -0.382∗∗∗ -0.279∗ 0.163 0.133 0.166 0.159 (0.133) (0.147) (0.134) (0.147) (0.231) (0.262) (0.234) (0.265) Prov 0.0932 0.203∗∗ 0.145 0.278 (0.0596) (0.0945) (0.133) (0.423) 1st tercile × Prov 0.370∗∗∗ 0.657∗∗∗ 0.387∗∗∗ 0.842∗ (0.0779) (0.122) (0.150) (0.443) 2nd tercile × Prov -0.0123 0.0461 0.131 0.397 (0.0698) (0.104) (0.176) (0.512) 3rd tercile × Prov 0.0488 0.146 0.0871 0.0250 (0.126) (0.177) (0.176) (0.504) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 1532134 1532134 1532134 1532134 6283454 6283454 6283454 6283454 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Provision measures are normal- ized between 0 and 1. The specifications control for tariffs, GDP of the destination, and PTA depth. 33 Table 4: Baseline Results - Size heterogeneity in effects of different types of integration of SPS and TBT provisions Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT ln(1 + τ ) -1.602∗∗∗ -1.624∗∗∗ -1.621∗∗∗ -1.780∗∗∗ -1.808∗∗∗ (0.408) (0.412) (0.411) (0.415) (0.421) ln(GDP ) 1.350∗∗∗ 1.287∗∗∗ 1.304∗∗∗ 2.159∗∗∗ 2.160∗∗∗ (0.196) (0.195) (0.191) (0.182) (0.182) Depth -0.298∗∗ -0.110 -0.351∗∗∗ 0.0559 0.124 (0.133) (0.186) (0.134) (0.222) (0.210) 1st tercile × Prov 0.352∗∗∗ 0.444∗∗∗ 0.616∗∗∗ -0.585 0.212∗∗ (0.0658) (0.115) (0.169) (1.166) (0.0908) 2nd tercile × Prov 0.00527 0.0905 -0.0690 1.001∗∗ 0.0389 (0.0527) (0.0943) (0.140) (0.486) (0.110) 3rd tercile × Prov 0.143 0.203∗ -0.148 1.060∗∗∗ 0.164 (0.111) (0.119) (0.281) (0.248) (0.130) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 1532134 1532134 1532134 6283454 6283454 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Provision measures are normal- ized between 0 and 1. The specifications control for tariffs, GDP of the destination, and PTA depth. 34 Table 5: Heterogeneous effects of SPS provisions across sectors Panel A: SPS provisions Export value of agricultural products Sector Animal Vegetable Foodstuffs 1st tercile × Prov 0.520∗∗ 0.361∗∗∗ 0.424∗∗∗ (0.250) (0.101) (0.0921) 2nd tercile × Prov -0.424∗∗ -0.0295 0.0450 (0.167) (0.0921) (0.0785) 3rd tercile × Prov -0.168 -0.0335 0.177 (0.293) (0.234) (0.132) fixed-effects fkj fkt fkj fkt fkj fkt cluster ij ij ij N 239414 795806 496914 Panel B: TBT provisions Export value of any product Sectors: Animal Vegetable Foodstuffs Chemicals Plastics Raw Hides Wood Textile Footwear Stone Metals Machinery Transportation Miscellaneous 1st tercile × Prov 0.203 0.278∗∗∗ 0.557∗∗∗ 0.763∗∗ 0.212 0.0506 0.723∗∗ -0.756∗ -0.272 0.179 1.117∗∗ 1.016∗∗∗ 2.094 0.962∗∗∗ (0.352) (0.104) (0.130) (0.317) (0.481) (0.623) (0.366) (0.389) (0.603) (0.721) (0.483) (0.331) (1.789) (0.340) 2nd tercile × Prov -0.709∗∗∗ -0.0841 0.0658 -0.464∗ 0.0599 0.914∗∗∗ 0.00289 -0.668∗∗∗ 0.866 0.229 0.824∗ 0.461∗ 1.625∗∗ -0.157 (0.227) (0.107) (0.125) (0.238) (0.275) (0.322) (0.272) (0.228) (0.599) (0.614) (0.433) (0.259) (0.780) (0.251) 3rd tercile × Prov -0.203 -0.0377 0.0435 -0.400∗ -0.364 0.212 0.202 -0.0167 0.828∗∗ 1.143∗∗∗ 0.126 0.326 1.836 -0.115 (0.308) (0.295) (0.142) (0.238) (0.332) (0.477) (0.249) (0.257) (0.373) (0.428) (0.333) (0.282) (1.162) (0.256) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij ij ij ij ij ij ij N 239414 795806 496914 511177 423791 141723 477309 1292868 89365 210349 447294 700472 70289 386647 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Provision measures are normal- ized between 0 and 1. The specifications control for tariffs, GDP of the destination, and PTA depth. 35 Table 6: Heterogeneous effects of SPS and TBT provisions across sectors depending on trade concerns Panel A: SPS provisions Export value of agricultural products Export value of any product All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT No SPS STCs At least one SPS STC No TBT STCs At least one TBT STC 1st tercile × Prov 0.658 0.758 0.772∗∗∗ 1.190∗∗∗ -0.0646 -0.240 0.567∗ 1.022∗∗ (0.818) (1.276) (0.238) (0.262) (0.758) (1.056) (0.319) (0.505) 2nd tercile × Prov -0.191 -0.410 0.261 0.377 1.112 1.942 0.259 0.362 (0.735) (0.834) (0.215) (0.232) (0.888) (1.481) (0.296) (0.499) 3rd tercile × Prov -0.442 -0.670 -0.203 -0.282 -0.940 -2.824∗∗ -0.0443 -0.320 (0.527) (0.513) (0.209) (0.218) (0.868) (1.209) (0.290) (0.498) fixed-effects fhj fht jt fhj fht jt fhj fht jt fhj fht jt fhj fht jt fhj fht jt fhj fht jt fhj fht jt cluster ij ij ij ij ij ij ij ij N 41678 41678 1109606 1109606 209274 209274 4025557 4025557 Panel B: TBT provisions Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT Harmonization TBT Transparency TBT No SPS STCs At least one SPS STC No TBT STCs At least one TBT STC 1st tercile × Prov 0.712 -0.318 0.573 0.577∗∗∗ 0.745∗∗∗ 0.251∗ 0.0800 0.321 1.221 0.482∗∗∗ (0.742) (0.374) (0.465) (0.101) (0.182) (0.133) (1.448) (0.313) (0.924) (0.162) 2nd tercile × Prov 0.0684 0.0913 -0.371 0.325∗∗∗ 0.207 -0.00749 -0.458 0.554∗ -1.361 0.331∗∗ (0.458) (0.493) (0.420) (0.0842) (0.137) (0.118) (1.283) (0.309) (1.155) (0.151) 3rd tercile × Prov -0.255 -0.254 -0.339 -0.0249 -0.143 -0.292∗∗∗ -0.728 0.0789 0.359 0.238 (0.223) (0.227) (0.251) (0.0732) (0.120) (0.0991) (1.993) (0.273) (0.259) (0.148) fixed-effects fkj fkt jt fkj fkt jt fkj fkt jt fkj fkt jt fkj fkt jt fkj fkt jt fkj fkt jt fkj fkt jt fkj fkt jt fkj fkt jt cluster ij ij ij ij ij ij ij ij ij ij N 41678 41678 41678 1109606 1109606 1109606 209274 209274 4025557 4025557 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Provision measures are normal- ized between 0 and 1. The specifications control for tariffs, GDP of the destination, and PTA depth. 36 Table 7: Export entry – Average and size heterogeneity in effects of SPS and TBT provisions. Panel A: All and essential SPS and TBT provisions Entry in agricultural products Entry in any product Prov: All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT Prov 0.192 0.297∗∗ -0.245 0.142 (0.143) (0.129) (0.151) (0.152) 1st tercile × Prov 0.279∗ 0.420∗∗∗ -0.245 0.205 (0.150) (0.148) (0.157) (0.176) 2nd tercile × Prov 0.191 0.288∗∗ -0.207 0.179 (0.152) (0.141) (0.146) (0.161) 3rd tercile × Prov 0.0825 0.131 -0.276∗ 0.0588 (0.142) (0.132) (0.156) (0.156) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 648452 648452 648452 648452 2736535 2736535 2736535 2736535 Panel B: Different types of integration of SPS and TBT provisions Entry in agricultural products Entry in any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.228∗∗∗ 0.234∗∗ 0.0550∗ -0.116 -0.0993∗∗∗ (0.0864) (0.0953) (0.0302) (0.157) (0.0338) 2nd tercile × Prov 0.170∗∗ 0.0870 0.0191 -0.149∗∗ -0.0838∗∗∗ (0.0830) (0.0964) (0.0309) (0.0681) (0.0325) 3rd tercile × Prov 0.0676 0.0234 -0.00587 -0.336∗∗ -0.103∗∗∗ (0.0742) (0.0790) (0.0243) (0.167) (0.0321) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 648452 648452 648452 2736535 2736535 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Provision measures are normal- ized between 0 and 1. The specifications control for tariffs, GDP of the destination, and PTA depth. 37 Table 8: Export quality – Average and size heterogeneity in effects of SPS and TBT provisions Panel A: All and essential SPS and TBT provisions Quality of agricultural products Quality of any product Prov: All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT Prov 0.0758 0.171 0.372 0.0482 (0.415) (0.396) (0.912) (0.575) 1st tercile × Prov 0.625 0.965∗∗ 0.196 0.0621 (0.464) (0.444) (1.031) (1.100) 2nd tercile × Prov 0.135 0.0652 -0.161 0.101 (0.370) (0.382) (0.896) (0.441) 3rd tercile × Prov -0.0340 0.119 -0.697 -4.769∗∗ (0.469) (0.485) (1.260) (2.257) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 173140 173140 173140 173140 1021249 1021249 1021249 1021249 Panel B: Different types of integration of SPS and TBT provisions Quality of agricultural products Quality of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.198∗ -0.132 0.118∗ -0.475∗∗∗ 0.551∗∗∗ (0.107) (0.194) (0.0612) (0.117) (0.0979) 2nd tercile × Prov 0.00148 -0.606∗ 0.0187 -0.201 0.400∗∗∗ (0.128) (0.332) (0.0599) (0.136) (0.0954) 3rd tercile × Prov -0.0189 0.357 -0.0420 -0.220∗∗∗ 0.331∗∗∗ (0.222) (0.278) (0.0998) (0.0471) (0.123) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 173140 173140 173140 1021249 1021249 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Provision measures are normal- ized between 0 and 1. The specifications control for tariffs, GDP of the destination, and PTA depth. 38 Table 9: Country income group - Size heterogeneity in effects of all and essential SPS and TBT provisions Panel A: High-income countries Export value of agricultural products Export value of any product Prov: All SPS Essential SPS All TBT Essential TBT 1st tercile × Prov 0.859∗∗∗ 1.099∗∗∗ 0.848 1.029 (0.147) (0.164) (0.843) (0.691) 2nd tercile × Prov 0.342∗∗∗ 0.282∗∗ 0.593 0.709 (0.128) (0.120) (0.922) (0.813) 3rd tercile × Prov 0.302 0.235 0.517 0.121 (0.186) (0.227) (0.811) (0.824) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij N 888246 888246 2442665 2442665 Panel B: Low and middle-income countries Export value of agricultural products Export value of any product Prov: All SPS Essential SPS All TBT Essential TBT 1st tercile × Prov 0.689∗∗∗ 0.833∗∗ 0.845∗∗ 1.617∗∗ (0.260) (0.390) (0.332) (0.787) 2nd tercile × Prov -0.0434 -0.153 -0.0698 -0.380 (0.301) (0.387) (0.202) (0.471) 3rd tercile × Prov -0.125 -0.206 -0.233 -0.971 (0.303) (0.356) (0.241) (0.722) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij N 480787 480787 2797370 2797370 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Provision measures are normal- ized between 0 and 1. The specifications control for tariffs, GDP of the destination, and PTA depth. 39 Table 10: Country income group – Size heterogeneity in effects of different types of integration of SPS and TBT provisions Panel A: High-income countries Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.465∗∗∗ 0.751∗∗∗ 0.184∗∗∗ -0.518 0.0377 (0.0723) (0.140) (0.0316) (1.143) (0.124) 2nd tercile × Prov 0.0637 0.196∗∗ 0.0385∗ 1.099∗∗ -0.00577 (0.0528) (0.0841) (0.0233) (0.500) (0.118) 3rd tercile × Prov 0.145 0.162 -0.000539 1.045∗∗∗ 0.102 (0.155) (0.108) (0.0425) (0.233) (0.144) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 888246 888246 888246 2442665 2442665 Panel B: Low and middle-income countries Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.478∗ 0.483∗∗ 0.159∗∗∗ 0 0.260∗∗∗ (0.256) (0.205) (0.0387) (.) (0.0986) 2nd tercile × Prov 0.0122 -0.0686 0.0211 0 -0.0348 (0.193) (0.240) (0.0356) (.) (0.0723) 3rd tercile × Prov -0.0426 -0.144 0.0236 0 -0.0831 (0.198) (0.235) (0.0578) (.) (0.0806) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 480787 480787 480787 2797370 2797370 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Provision measures are normal- ized between 0 and 1. The specifications control for tariffs, GDP of the destination, and PTA depth. 40 Appendix A A.1 SPS and TBT provisions in PTAs 300 20 Number of Agreements Cumulative 150 10 0 0 60 70 80 90 00 10 20 19 19 19 19 20 20 20 Total number of new PTAs New PTAs with SPS and TBT provisions Source: Mattoo et al. (2020). Notes: the graph plots the number of new PTAs entering in force over the period 1958-2017 that include at least one provision on SPS or TBT. The red line shows the cumulated number of PTAs. B Templates of SPS and TBT provisions in PTAs The measures of SPS and TBT provisions in the database are based on the SPS and TBT chapters that are covered in, respectively, 269 and 263 PTAs signed between 1960 and 2017. 41 The other policy areas covered in PTAs are: anti-dumping, countervailing duties, competi- tion policy, environmental laws, export taxes, investment, intellectual property rights, labor market, migration (visa and asylum), movement of capital, public procurement, services, state-owned enterprises, subsidies, rules of origin, and trade facilitation. Espitia et al. (2020) and Stone and Casalini (2020) provide details on the mapping of the content of TBT and SPS provisions in PTAs. Below is a list of the entire set of SPS and TBT provisions in PTAs. The provisions that appear in bold are the essential provisions. B.1 Template of SPS provisions in PTAs I. Reference to the WTO SPS Agreement - Are any specific Annexes of the SPS Agreement adopted? - Does the Agreement refer to the WTO SPS Agreement? - Does the Agreement use the same definitions as the SPS Agreement? - Does the Agreement use the same rules as the SPS Agreement? II. Integration Approach A. Standards - Are Parties encouraged to take into account other parties’ standards when elab- orating new standards? - Are there specified existing standards to which countries shall harmonize? - Is equivalence recognized? - Is mutual recognition recognized? - Is the burden of justifying non-equivalence on the importing country? - Is the creation of concerted/regional standards referenced? - Is there a time schedule for achieving mutual recognition? - Do parties recognize the adaption to regional conditions (including regionaliza- tion, zoning and/or compartmentalization) - Do parties reference international standards? B. Risk Assessment - Is the burden of evaluating risk on the exporting country? - Is there reference to international standards/procedures? - Do the parties recognize that SPS measures are based on documented and sci- entific (if not available, objective) evidence? - Is the participation of interested parties referenced? C. Audits/Control Inspections - Are there specified existing standards to which countries shall harmonize? - Does the importing party have the right to audit the exporting party’s competent authorities, inspection systems, or production procedure? - Is mutual recognition in force? - Is the burden of justifying non-equivalence on the importing country? - Is the use of international standards promoted? - Is the use or creation of regional standards promoted? - Are there provisions for advance rulings? - Are there provisions for pre-certification processes for exporting firms? - Is the participation of interested parties referenced? 42 - Is there a provision on control and inspection? III. Transparency Requirements - Do parties have to notify each other prior to the entry into force of a new standard or regulation? - Does the Agreement allow the participation of interested parties of the other party in the development of standards? - Does the agreement specifically reference the participation of regulatory authorities of the other party in the development of standards? - Is there a derogation clause on notification period for emergency? - Is there a duty to translate the document in the language of the other party(ies)? - Is there a limitation to the obligation to notify, for reasons of law enforcement, public in- terest, or commercial interest? - Is there a provision on electronic publication? - Is there a provision on exchange of information? - Is there a specified minimum time period for comments? - Is there a transparency provision? IV. Institutions A. Administrative Bodies - Do parties establish a SPS committee? - Do parties establish a working group? - Do parties establish SPS contact/enquiry points? - Is a body for administering the Agreement established? - Is the recourse to the DS for SPS chapter disallowed? - Is the SPS Committee open? - Is the SPS Committee the designated first place for dispute resolution? - Is there a fixed periodical meeting for the committee? - Is there a mechanism mandated to issue administrative decisions? - Is there a mechanism to issue recommendations? V. Further Cooperation Among Members - Is there a general IRC clause/Common policy/standardization program (beyond trade- related objectives)? - Is there a provision on technical assistance? - Is there a provision for technical consultations/cooperation? VI. MRAs - Is there an MRA in force? VII. Others - Is there a provision on special/preferential treatment? - Is there a provision on certifications? - Is there a provision on labelling, marking, and packaging? - Is there a provision on traceability? - Is coordination for participating in international or regional accreditation agencies refer- enced? - Is testing data to be made available? 43 B.2 Template of TBT provisions in PTAs I. Reference to WTO-TBT Agreement - Does the agreement refer to the WTO TBT Agreement? - Does the agreement use the same definitions as the TBT Agreements? - Does the agreement go beyond the TBT Agreement? II. Integration Approach: A. Standards: (i) Mutual Recognition: - Is mutual recognition in force? - Is there a time schedule for achieving mutual recognition? - Is the burden of justifying non-equivalence on the importing country? (ii) Harmonization: - Are there specified existing standards to which countries shall harmonize? - Is the use or creation of regional standards promoted? - Is the burden of justifying non-equivalence on the importing country? B. Technical Regulation: (iii) Mutual Recognition: - Is mutual recognition in force? - Is there a time schedule for achieving mutual recognition? - Is the burden of justifying non-equivalence on the importing country? (iv) Harmonization: - Are there specified existing standards to which countries shall harmonize? - Is the use or creation of regional standards promoted? - Is the burden of justifying non-equivalence on the importing country? C. Conformity Assessment: (v) Mutual Recognition: - Is mutual recognition in force? - Is there a time schedule for achieving mutual recognition? - Is the burden of justifying non-equivalence on the importing country? (vi) Harmonization: - Are there specified existing standards to which countries shall harmonize? - Is the use or creation of regional standards promoted? - Is the burden of justifying non-equivalence on the importing country? III. Transparency Requirements: A. Notification: - Is the time period allowed for comments specified? - Is the time period allowed for comments longer than 60 days? B. Contact points/consultations for exchange of information: IV. Institutions: A. Administrative Bodies: - Is a regional body established? B. Dispute Settlement Mechanism: - Is there a regional dispute settlement body? - Are there regional consultations foreseen to resolve disputes? 44 - Is there a mechanism to issue recommendations? - Are recommendations mandatory? - Is the recourse to the DS for technical regulations disallowed? V. Further Cooperation Among Members: - Common policy/standardization program (beyond trade-related objectives) - Technical Assistance - Metrology VI. MRAs: - Is there an MRA in force? C Miscellaneous Data Issues C.1 Normalization of measures of provisions in PTAs To ease interpretation of estimated coefficients, we normalize our main variables of interest capturing numbers of SPS and TBT provisions (as well as number of other provisions to capture depth) between 0 and 1 based on the formula below. The formula is given for the example of essential SPS provisions: essential SP S essential − M IN (SP S essential ) SP Snormalized = M AX (SP S essential ) − M IN (SP S essential ) The maximum and minimum of essential provisions used in the formula above are computed based on the information in the Deep Trade Agreements database for all countries. The maximum numbers of provisions are as follows: • SPS provisions: 30 in the CPTPP • essential SPS provisions: 12 in the Andean Community • SPS transparency provisions: 7 in the CPTPP and the Pacific Alliance • TBT provisions: 10 in the Colombia-Mexico PTA of 1995 and the Peru-Chile PTA of 2009 • essential TBT provisions: 7 in the Colombia-Mexico PTA of 1995 • TBT transparency provisions: 3 in the Peru- Korea PTA of 2011 and the EU-Colombia and Peru PTA of 2013 For SPS and for TBT provisions the minimum across all countries is actually 0. This normalization of the number of provisions is easier to interpret than a simple count of the number of SPS provisions in each PTA as the latter would capture the marginal impact of including one additional SPS provision (all else equal). With the normalization, we capture the impact of increasing the number of SPS provisions (relative to the minimum of 0) as a fraction of the maximum number of SPS provisions. In contrast to an alternative that would 45 take the logarithm of the number of SPS provisions in each PTA, the normalization allows us to account for bilateral pairs of countries enforcing no SPS provisions. In the PPML estimation of the firm gravity equation, a coefficient b = 0.5 on the normalized numbers of essential SPS provisions indicates that an increase from 0 to the maximum number of essential SPS provisions (of 30 as shown in Appendix Table B1) would increase export values by 65 percent on average (exp(0.5)-1=0.65) with all else equal. C.2 Data fill-in procedure C.3 Observations in initial and expanded customs databases Country # obs. all years # obs. all years Before Fillin After Fillin Chile 1,138,721 7,064,050 Colombia 1,676,031 9,727,303 Peru 1,260,125 6,705,624 46 C.4 Summary statistics on tariffs 1995 2015 Country Average tariff Share of 0 tariff Average tariff Share of 0 tariff Chile 13.0 22.3 7.6 42.5 Colombia 12.7 22.8 7.7 40.7 Peru 12.6 23.8 7.6 41.1 RoW 13.1 23.2 7.8 39.0 C.5 Definition of industries For our analysis using data at the firm-destination-industry-year level, we aggregate data from the firm-destination-HS 6-digit-year level to the 15 industries listed that aggregate across HS 6-digit products in HS 2-digit chapters as follows: Animal & Animal Products (HS 2-digit 01-05), Vegetable Products (HS 2-digit 06-15), Foodstuffs (HS 2-digit 16-24), Chemicals & Allied Industries (HS 2-digit 28-38), Plastics & Rubbers (HS 2-digit 39-40), Raw Hides, Skins, Leather, & Furs (HS 2-digit 41-43), Wood & Wood Products (HS 2-digit 44-49), Textiles (HS 2-digit 50-63), Footwear & Headgear (HS 2-digit 64-67), Stone & Glass (HS 2-digit 68-71), Metals (HS 2-digit 72-83), Machinery & Electrical (HS 2-digit 84-85), Transportation (HS 2-digit 86-89), Miscellaneous 90-97). C.6 Sectors excluded to address endogeneity Some of our specifications drop the HS 2-digit chapters below that exhibit high sectoral export concentration in the top 1% of firms (they are responsible for more than 50 percent of sectoral exports) to address potential endogeneity in the content of PTAs: - Chile: 21, 29, 38, 39, 40, 42, 44, 48, 49, 62, 63, 69, 70, 72, 73, 74, 82, 83, 84, 85, 87, 94, 96 - Colombia: 19, 21, 28, 32, 34, 38, 39, 40, 42, 44, 48, 49, 56, 58, 61, 62, 63, 69, 70, 71, 72, 73, 74, 76, 82, 84, 85, 87, 90, 95, 96 - Peru: 11, 12, 17, 19, 34, 35, 39, 40, 42, 44, 46, 48, 49, 51, 55, 58, 61, 62, 63, 68, 69, 70, 71, 72, 73, 74, 82, 83, 84, 85, 87, 90, 94, 95, 96, 97. 47 D Robustness checks D.1 Alternative firm size definitions Table 11: Size terciles based on sectoral exports – Size heterogeneity in effects of SPS and TBT provisions Panel A: All and essential SPS and TBT provisions Export value of agricultural products Export value of any product All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT Prov 0.0932 0.203∗∗ 0.145 0.278 (0.0596) (0.0945) (0.133) (0.423) 1st tercile × Prov 0.943∗∗∗ 1.558∗∗∗ 1.236∗∗∗ 2.767∗∗∗ (0.116) (0.176) (0.129) (0.317) 2nd tercile × Prov 0.848∗∗∗ 1.288∗∗∗ 0.953∗∗∗ 1.983∗∗∗ (0.100) (0.188) (0.116) (0.310) 3rd tercile × Prov 0.0221 0.0946 0.103 0.190 (0.0595) (0.0936) (0.136) (0.436) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 1532134 1532134 1532134 1532134 6283454 6283454 6283454 6283454 Panel B: Different types of integration of SPS and TBT provisions Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.606∗∗∗ 1.020∗∗∗ 1.819∗∗∗ 0.846 0.826∗∗∗ (0.103) (0.157) (0.326) (0.662) (0.101) 2nd tercile × Prov 0.616∗∗∗ 0.814∗∗∗ 1.876∗∗∗ 0.179 0.678∗∗∗ (0.0971) (0.159) (0.219) (0.142) (0.0750) 3rd tercile × Prov 0.0658 0.125 -0.00627 0.899∗∗∗ 0.0911 (0.0488) (0.0920) (0.125) (0.285) (0.0930) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 1532134 1532134 1532134 6283454 6283454 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Depth and provision measures are normalized between 0 and 1. 48 Table 12: Size terciles based on total exports – Size heterogeneity in effects of SPS and TBT provisions Panel A: All and essential SPS and TBT provisions Export value of agricultural products Export value of any product All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT Prov 0.0932 0.203∗∗ 0.145 0.278 (0.0596) (0.0945) (0.133) (0.423) 1st tercile × Prov 0.852∗∗∗ 1.410∗∗∗ 1.322∗∗∗ 3.138∗∗∗ (0.151) (0.224) (0.147) (0.384) 2nd tercile × Prov 0.958∗∗∗ 1.489∗∗∗ 1.010∗∗∗ 2.130∗∗∗ (0.116) (0.185) (0.136) (0.350) 3rd tercile × Prov 0.0558 0.145 0.118 0.219 (0.0599) (0.0948) (0.135) (0.431) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 1532134 1532134 1532134 1532134 6283454 6283454 6283454 6283454 Panel B: Different types of integration of SPS and TBT provisions Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.556∗∗∗ 1.125∗∗∗ 1.558∗∗∗ 0.883 0.934∗∗∗ (0.138) (0.233) (0.379) (0.717) (0.139) 2nd tercile × Prov 0.649∗∗∗ 0.927∗∗∗ 2.096∗∗∗ 0.168∗∗∗ 0.710∗∗∗ (0.116) (0.175) (0.276) (0.0561) (0.0873) 3rd tercile × Prov 0.0905∗ 0.155∗ 0.0485 0.898∗∗∗ 0.101 (0.0496) (0.0924) (0.127) (0.285) (0.0917) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 1532134 1532134 1532134 6283454 6283454 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Depth and provision measures are normalized between 0 and 1. 49 D.2 Controlling for destination demand effects Table 13: Controlling for demand effects – Size heterogeneity in effects of SPS and TBT provisions Panel A: All and essential SPS and TBT provisions Export value of agricultural products Export value of any product Prov: MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.656∗∗∗ 0.778∗∗∗ 0.316∗∗ -0.198 0.523∗∗∗ (0.0869) (0.177) (0.139) (0.590) (0.160) 2nd tercile × Prov 0.326∗∗∗ 0.346∗∗ 0.0814 -1.414 0.328∗∗ (0.0911) (0.141) (0.108) (1.287) (0.156) 3rd tercile × Prov -0.00698 -0.108 -0.266∗∗∗ 0.365 0.231∗ (0.0709) (0.112) (0.0982) (0.259) (0.139) fixed-effects fsj fst jt fsj fst jt fsj fst jt fsj fst jt fsj fst jt cluster ij ij ij ij ij N 1055259 1055259 1055259 3664227 3664227 Panel B: Different types of integration of SPS and TBT provisions Export value of agricultural products Export value of any product Prov: MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.656∗∗∗ 0.778∗∗∗ 0.316∗∗ -0.198 0.523∗∗∗ (0.0869) (0.177) (0.139) (0.590) (0.160) 2nd tercile × Prov 0.326∗∗∗ 0.346∗∗ 0.0814 -1.414 0.328∗∗ (0.0911) (0.141) (0.108) (1.287) (0.156) 3rd tercile × Prov -0.00698 -0.108 -0.266∗∗∗ 0.365 0.231∗ (0.0709) (0.112) (0.0982) (0.259) (0.139) fixed-effects fsj fst jt fsj fst jt fsj fst jt fsj fst jt fsj fst jt cluster ij ij ij ij ij N 1055259 1055259 1055259 3664227 3664227 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Depth and provision measures are normalized between 0 and 1. 50 E Addressing endogeneity E.1 Excluding large firms Table 14: Excluding large firms – Size heterogeneity in effects of SPS and TBT provisions Panel A: All and essential SPS and TBT provisions Export value of agricultural products Export value of any product Prov: All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT Prov -0.00159 0.0890 0.145 0.278 (0.0755) (0.117) (0.133) (0.423) 1st tercile × Prov 0.192∗ 0.414∗∗∗ 0.228 0.501 (0.106) (0.159) (0.150) (0.358) 2nd tercile × Prov -0.143∗ -0.105 -0.186∗ -0.360 (0.0792) (0.122) (0.0976) (0.257) 3rd tercile × Prov -0.0370 -0.00709 0.0468 0.150 (0.104) (0.157) (0.108) (0.230) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 1227149 1227149 1227149 1227149 6283454 6283454 5292082 5292082 Panel B: Different types of integration of SPS and TBT provisions Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.270∗∗∗ 0.326∗∗ 0.292 1.079∗∗ 0.140 (0.0812) (0.134) (0.251) (0.514) (0.113) 2nd tercile × Prov -0.0336 -0.0594 -0.361∗ 0.371∗ -0.0902 (0.0697) (0.119) (0.193) (0.198) (0.0586) 3rd tercile × Prov -0.00852 0.0222 -0.130 0.151 -0.0224 (0.0868) (0.140) (0.237) (0.340) (0.0825) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 1227149 1227149 1227149 5292082 5292082 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Depth and provision measures are normalized between 0 and 1. 51 E.2 Excluding concentrated sectors Table 15: Excluding concentrated sectors – Size heterogeneity in effects of SPS and TBT provisions Panel A: All and essential SPS and TBT provisions Export value of agricultural products Export value of any product Prov: All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT Prov 0.530∗∗∗ 0.526∗∗∗ -0.0376 0.0372 (0.144) (0.139) (0.264) (0.497) 1st tercile × Prov 0.767∗∗∗ 0.917∗∗∗ 0.775∗∗∗ 1.359∗∗∗ (0.168) (0.199) (0.225) (0.338) 2nd tercile × Prov 0.405∗∗∗ 0.368∗∗∗ -0.0554 -0.988 (0.145) (0.139) (0.192) (0.603) 3rd tercile × Prov 0.560∗∗∗ 0.597∗∗∗ 0.181 -0.334 (0.171) (0.212) (0.209) (0.366) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 536547 536547 536547 536547 6283454 6283454 1442266 1442266 Panel B: Different types of integration of SPS and TBT provisions Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.460∗∗∗ 0.476∗∗∗ 0.102∗∗∗ -0.751∗∗ 0.218∗∗ (0.104) (0.116) (0.0273) (0.341) (0.0883) 2nd tercile × Prov 0.112 0.198∗∗ 0.0319 -0.916∗∗∗ 0.0805 (0.0882) (0.0837) (0.0244) (0.226) (0.0753) 3rd tercile × Prov 0.273∗∗ 0.360∗∗ 0.0641∗∗ -0.915 0.0815 (0.131) (0.150) (0.0302) (0.568) (0.0705) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 536547 536547 536547 1442266 1442266 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Depth and provision measures are normalized between 0 and 1. 52 F Heterogeneity in partner countries F.1 PTAs signed with EU or US Table 16: PTA partner - Size heterogeneity in effects of all and essential SPS and TBT provisions Panel A: EU and US PTAs Export value of agricultural products Export value of any product All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT Prov 0.277∗∗∗ 0.257∗∗∗ 0.500 -0.401 (0.0872) (0.0803) (1.510) (0.636) 1st tercile × Prov 0.784∗∗∗ 1.021∗∗∗ 1.487 1.348∗∗ (0.122) (0.144) (1.675) (0.600) 2nd tercile × Prov 0.154∗ 0.0568 0.886 0.220 (0.0860) (0.0856) (1.819) (0.895) 3rd tercile × Prov 0.114 -0.0575 0.154 -1.941∗∗ (0.194) (0.266) (1.479) (0.858) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 467880 467880 467880 467880 1049992 1049992 1049992 1049992 Panel B: Rest of the world PTAs Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.501∗∗∗ 0.719∗∗∗ 0.171∗∗∗ 0 0.279∗ (0.0809) (0.119) (0.0266) (.) (0.159) 2nd tercile × Prov 0.00435 0.0581 0.00114 0 0.144 (0.0456) (0.0640) (0.0187) (.) (0.148) 3rd tercile × Prov 0.123 -0.175 -0.0412 0 0.185 (0.218) (0.142) (0.0475) (.) (0.192) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 467880 467880 467880 1049992 1049992 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Depth and provision measures are normalized between 0 and 1. 53 F.2 PTAs signed with RoW Table 17: PTA partner – Size heterogeneity in effects of different types of integration of SPS and TBT provisions Panel A: EU and US PTAs Export value of agricultural products Export value of any product All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT Prov 0.332 0.665∗∗∗ 0.0827 0.146 (0.209) (0.230) (0.184) (0.387) 1st tercile × Prov 0.489∗∗ 0.943∗∗∗ 0.253 0.577 (0.218) (0.249) (0.188) (0.498) 2nd tercile × Prov 0.219 0.547∗∗ 0.0703 0.215 (0.218) (0.253) (0.162) (0.348) 3rd tercile × Prov 0.438∗ 0.724∗∗ 0.0512 -0.0517 (0.262) (0.295) (0.271) (0.556) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 869091 869091 869091 869091 4372704 4372704 4372704 4372704 Panel B: Rest of the world PTAs Export value of agricultural products Export value of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov 0.306∗ 0.453∗∗∗ 0.0535 -2.717∗∗∗ 0.0875 (0.169) (0.113) (0.0339) (0.721) (0.0692) 2nd tercile × Prov 0.167 0.198∗ -0.0195 0.514 0.0494 (0.136) (0.120) (0.0321) (0.613) (0.0475) 3rd tercile × Prov 0.309∗∗ 0.358∗∗ -0.00295 0.964 0.0270 (0.149) (0.155) (0.0487) (0.666) (0.0755) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 869091 869091 869091 4372704 4372704 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Depth and provision measures are normalized between 0 and 1. 54 G Firm export quality In order to infer the quality of products exported by firms in Colombia and Peru, we rely on the methodology of Amiti and Khandelwal (2013) used at the firm-level by Bajgar and Ja- vorcik (2020).Accordingly, we estimate quality as the residual of the following equation (ˆf kjt ): ln (quantityf kjt ) + σs ln (pricef kjt ) = αk + αjt + σs × αjt + f kjt (3) where quantityf kjt is the quantity in kilograms of HS 6-digit product k exported by firm f to destination j in year t, pricef kjt is the unit value of HS 6-digit product k charged by firm f in destination j in year t. σs is the elasticity of substitution varying across HS 2-digit industries taken from Broda and Weinstein (2006) and the specification accounts for product fixed effects αk and destination-year fixed effects αjt . Equation (3) is estimated by OLS and the residual gives quality for product k exported by firm f to destination j in year t. The underlying assumption behind this methodology is that the ability of an exporter to sell more quantity at a given price should come from higher quality. The quality measures are then used as the dependent variable in Equation (2). H Estimations on firm export unit values Table 18: Export unit values – Average and size heterogeneity in effects of SPS and TBT provisions Panel A: All and essential SPS and TBT provisions UV of agricultural products UV of any product Prov: All SPS Essential SPS All SPS Essential SPS All TBT Essential TBT All TBT Essential TBT Prov 0.297 0.562∗∗ -0.0456 -0.0415 (0.214) (0.223) (0.668) (0.525) 1st tercile × Prov 0.496∗∗ 0.696∗∗∗ 0.0661 1.095 (0.244) (0.242) (0.617) (0.899) 2nd tercile × Prov 0.365 0.458 -0.863∗ -0.800 (0.342) (0.454) (0.523) (0.505) 3rd tercile × Prov 0.221 0.614∗∗∗ -0.124 -5.158∗∗∗ (0.168) (0.222) (0.817) (1.840) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij ij ij ij N 173140 173140 173140 173140 1021470 1021470 1021470 1021470 55 Panel B: Different types of integration of SPS and TBT provisions UV of agricultural products UV of any product MR SPS Harmonization SPS Transparency SPS Harmonization TBT Transparency TBT 1st tercile × Prov -0.0621 0.0748 -0.0439 -0.0871 0.257∗ (0.0751) (0.114) (0.0401) (0.0719) (0.143) 2nd tercile × Prov -0.0135 -0.129 -0.0131 0.0362 0.00269 (0.150) (0.235) (0.0563) (0.0716) (0.131) 3rd tercile × Prov -0.0630 1.081∗∗∗ -0.0922 0.153∗ 0.263∗∗ (0.132) (0.320) (0.0653) (0.0881) (0.108) fixed-effects fkj fkt fkj fkt fkj fkt fkj fkt fkj fkt cluster ij ij ij ij ij N 173140 173140 173140 1021470 1021470 Notes : PPML estimation is used. Standard errors clustered by origin-destination presented in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. Depth and provision measures are normalized between 0 and 1. 56